Every one of the 6,000 entries was scored on four things: whether it names a test or measurement, whether it makes a directional prediction ("should"), whether it carries a quantity, and whether its logical form is a bound, threshold, count, scale or reversal. Hedged phrasing ("may be", "could be") was penalised, and so was any entry far from a workable length. The top twenty per set were kept.
This is a testability rule, not a truth rule. Nothing here is verified; the selection favours hypotheses that could be killed by an experiment.
| Test | Claim being tested | Result | Verdict |
|---|---|---|---|
| Counting bound | Conditioning degrades as parameters approach data points | K=3 → 1.35, K=5 → 1.5, K=8 → 1.93, K=12 → 2.26, K=20 → 2.79 (N=60) | supported, but slowly |
| Derived-variable inflation | A derived column inflates R² by an order of magnitude | direct R² ≈ 0.9919; derived R² ≈ 0.0023 at σ=0.1 | not reproduced |
| Rounding bias | Quantising the reported value shifts the mean | step 0.1 shifts mean by 0.00176; step 0.5 by -0.00304 (n=1000) | negligible |
| Two-clock race | A fitted time constant depends on the observation window | fitted τ = 1.0 at windows 0.2, 1.0 and 5.0 (true τ = 1.0) | not reproduced |
| Sampling mass | Required mass scales as the cube of particle size | 1 mm → 0.001 g; 20 mm → 8.0 g (factor 8,000) | definition, not a test |
Three of five failed, and the failures are informative. The counting bound is real but far gentler than the earlier work suggested when data points greatly exceed parameters. The derived-variable and two-clock claims were constructed too cleanly here — the derived column was a numerical gradient dominated by noise, and the exponential fit was exact by construction, so neither test could have shown the effect even if it exists. The rounding test is the honest negative: quantisation of a symmetric error shifts a mean by about one part in a thousand, which is not the failure mode the hypothesis imagined.
Grouped by area, with the logical form of each claim and its testability score.
| ID | Form | Claim | Score |
|---|---|---|---|
| H14 | Artefact | The "enhanced" thermal conductivity of nanofluids is partly a measurement artefact of convection in the hot wire's boundary layer, and the effect should disappear when the wire is horizontal below a critical Rayleigh number — a test that has not been done in either orientation in the same apparatus. | 6.91 |
| H39 | Artefact mimicking signal | A "1/f" spectrum measured below 1 Hz in a laboratory is usually the building's thermal time constant, not the device's; the test is a simultaneous measurement in a thermally isolated enclosure, which should remove the slope without changing the device. | 6.89 |
| H8 | Count | A structure with N bolted joints has at most 2N independent damping parameters, and any identification claiming more is fitting noise; the test is a coordinate-descent fit on a structure with a known joint count. | 6.78 |
| H5 | Artefact mimicking effect | The reported "slow light" in some resonant media is a group-delay measurement through a cavity whose finesse was inferred rather than measured; the delay is real but belongs to the cavity, not the medium, and scales with finesse in a way that separates them. | 4.98 |
| H70 | Hidden term | The measured stiffness of any thin sample includes the compliance of the load train, and for samples stiffer than the machine the reported modulus is the machine's, not the sample's — a bound that is testable by changing the sample's length at fixed cross-section. | 4.98 |
| H25 | Count | A polycrystal's yield strength depends on the grain-boundary character distribution through at most three independent parameters, so samples with the same average grain size but different character mixtures must differ measurably — a test requiring character mapping rather than grain counting. | 4.91 |
| H72 | Bound | A sensor's response time and its detection limit are bounded by the same diffusion length, so improving one by a factor of two must worsen the other by the same factor, and any claim of simultaneous improvement is measuring different quantities. | 4.91 |
| H19 | Misattribution | The "induction period" before nucleation in many solutions is the time required for the container's surface to saturate, not for the bulk to reach supersaturation — testable by doubling the surface area at fixed supersaturation, which should halve the period. | 4.89 |
| H52 | Bound | The sensitivity of any differential measurement is bounded by the matching of its two arms, not by the noise of either, so improving the detector alone cannot improve a mismatched pair — the bound is on the ratio of mismatches. | 4.89 |
| H64 | Inversion | A noisy measurement can be more useful than a quiet one when the noise is correlated with a nuisance variable, because the correlation allows the nuisance to be estimated and removed; deliberately adding correlated noise should improve the corrected result. | 4.89 |
| H90 | Identity | Thermal boundary conductance measured by thermoreflectance and by a steady-state method differ chiefly by the inclusion of the transducer's own interface, which is present in one and not the other — so the methods measure different interfaces, not different values. | 4.89 |
| H79 | Misattribution | The apparent higher reactivity of nanocatalysts is largely the higher density of low-coordination sites, which is a geometric count rather than a size effect, and it can be tested by comparing particles of the same size with different morphologies. | 4.87 |
| H6 | Reversal | Thermal lensing in a pumped crystal reverses sign at a pump intensity set by the competition between dn/dT and the end-face bulging, and the reversal intensity is a cleaner measurement of the photoelastic constants than either effect alone. | 4.84 |
| H75 | Identity | The two "different" degradation modes of a battery — capacity fade and impedance rise — are one process viewed at two current rates, and a single measurement at a third rate should place both on the same curve. | 4.84 |
| H100 | Threshold | Every porous material has a pore size above which gas flow is viscous and below which it is diffusive, and the transition radius is where the mean free path equals the radius — but the practical threshold for a measurement is a decade larger, because the instrument integrates over a pore-size distribution. | 4.84 |
| H76 | Artefact | Reported "negative capacitance" in some devices is the impedance analyser's own phase correction applied outside its calibrated range, and it should vanish when the measurement is repeated with a reference capacitor of similar magnitude in parallel. | 4.8 |
| H83 | Bound | No absorptive cooler can cool below the dew point of the air it exhausts, which bounds the practical temperature drop far above the thermodynamic limit and makes reported deep cooling a sign of a mis-measured humidity. | 4.8 |
| H67 | Bound | The throughput of any parallel system is bounded by its slowest shared resource, not its fastest stage, and this bound is why scaling up a laboratory process by duplicating vessels fails at a predictable count. | 4.78 |
| H97 | Artefact | "Anomalous" first-cycle behaviour in a battery is the formation of the measurement's own reference electrode, not a property of the active material, and it should disappear if the reference is formed before the first measurement. | 4.78 |
| H63 | Hidden variable | The yield of a crystallisation process depends on the vessel's thermal history for days before the batch, through the number of surviving nuclei on the wall; cleaning between batches should change yields measurably. | 4.73 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3555 | Hidden term | Single-lap shear strength contains a peel component from load-path eccentricity, and shimming the grips to align the paths removes part of what is reported as shear strength. The same joints tested aligned and misaligned should differ by an amount that scales with adherend thickness. | 6.98 |
| H3564 | Misattribution | Hot-wet strength loss is charged to hydrolysis of the adhesive, but in many joints water displaces the adhesive from the interface and the loss reverses on drying. The same joints tested wet, dried, and re-wet should recover if displacement dominates and not recover if hydrolysis does. | 6.98 |
| H3567 | Threshold | Below a stress intensity no subcritical crack growth occurs in the adhesive even at saturation, so a durability test should report a threshold rather than a time to failure. A load series at full saturation should show arrest below a load that the time-to-failure ranking does not reveal. | 6.93 |
| H3586 | Count | A bolted joint's preload has at most two independent loss mechanisms, embedment and thermal mismatch, once the fastener and stack are fixed, so a model with four loss terms is fitting the test. Preload measured at two temperatures and two surface finishes should account for all of it. | 6.93 |
| H3517 | Reversal | Faster heating can leave a diffusion-limited resin more fully cured than slow heating, because the glass transition lags the ramp and mobility persists longer than at equilibrium. Two ramps to the same hold, with Tg and conversion measured afterwards, should invert the usual expectation that slow is more complete. | 6.91 |
| H3579 | Bound | In a corrosive environment a joint's strength is bounded below by the corroded adherend rather than by the adhesive, once the corrosion layer is thick enough. An exposure series with adherend section loss measured should show failure moving into the metal. | 6.91 |
| H3562 | Identity | Cleavage and peel are the same loading at different adherend thickness ratios, so a joint weak in peel should be weak in cleavage once thickness is scaled. A thickness series tested in both modes should collapse onto one curve. | 6.87 |
| H3569 | Identity | Creep under load and stress relaxation under fixed displacement are one compliance seen at two boundary conditions, so a joint that creeps is a joint that relaxes. Measuring both on the same specimen should give relaxation times that agree. | 6.87 |
| H3573 | Trade-off | A toughened adhesive gains peel resistance and loses creep resistance, because the rubber particles that blunt cracks also lower the modulus. A series of rubber contents tested in peel and in creep should show the two rankings inverting. | 6.84 |
| H3515 | Bound | The residual stress a bondline can develop is bounded by the adhesive's shear modulus at gelation multiplied by its free shrinkage strain, so a stiff adhesive that gels late can develop less stress than a soft one that gels early. Measuring the curvature of thin adherend pairs against gel time should show the ranking by modulus alone failing. | 6.71 |
| H3528 | Misattribution | Off-ratio adhesive is blamed on the meter, but in a cartridge system the first bead is rich in one component because the two streams wet the nozzle differently. Discarding the first few centimetres and testing the next should remove the effect without touching the meter. | 5.0 |
| H3536 | Inversion of standard practice | Thicker anodic layers are conventionally better for durability, but under peel a thick brittle oxide fractures and hands the crack a sharp interface, so a thin oxide can outperform it. An anodising-time series tested in peel and in shear should rank the two modes oppositely. | 5.0 |
| H3505 | Inversion of standard practice | Coupling agents are laid down thicker for "full coverage", yet beyond a monolayer the silane builds a hydrolytically weak interphase, so dry strength rises while hot-wet strength falls. A thickness series from sub-monolayer to tens of nanometres (estimate), tested dry and after boiling, should cross over. | 4.98 |
| H3520 | Causal ordering | Loss of tack is used to infer conversion, but tack can vanish before gelation when a low-molecular-weight fraction segregates to the surface and crystallises there. Surface composition and bulk conversion measured on the same cure series should show tack loss preceding gelation by a measurable interval. | 4.98 |
| H3522 | Null claim | Post-cure beyond full conversion should change no mechanical property except through the moisture it drives off, so a post-cured and a fully cured but unpost-cured specimen tested dry ought to be identical. Any difference should vanish when both are dried to the same mass. | 4.98 |
| H3590 | Unit trap | Torque becomes preload through a nut factor that hides thread and bearing friction, so a preload quoted from torque is a friction measurement. A strain-gauged bolt recorded alongside torque and lubrication should show the nut factor moving with the lubricant rather than the preload. | 4.98 |
| H3514 | Hidden variable | Conversion at the centre of a bondline differs from the coupon because the adherends act as heat sinks, and the difference is governed by the Biot number rather than by the resin. Curing at two adherend thicknesses and measuring Tg at the mid-plane should separate the two. | 4.96 |
| H3537 | Identity | The water-break test and the contact angle measure the same contamination, and where they disagree the cause is roughness rather than chemistry. A set of surfaces of controlled roughness and controlled contamination should show the two tests diverging only on the rough ones. | 4.96 |
| H3561 | Conservation claim | In a tubular joint the torque is carried partly by adhesive shear and partly by friction of the press fit, and a torque-only test cannot separate them. Instrumenting the tube in torsion and again in tension should give two different apparent shear strengths. | 4.96 |
| H3566 | Bound | Strength loss from moisture cannot saturate before the bondline itself does, so a test shorter than the diffusion time reports a rate rather than a durability. Joints aged to different fractions of equilibrium should show the loss still climbing when plotted against square-root time, with no plateau. | 4.96 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3190 | Count | A separation system needs at most two independent inhibits, because a third adds failure probability faster than it removes it once the inhibit's own failure rate exceeds the release's. A fault-tree count with the inhibit's rate measured rather than assumed puts the crossover where the tree says it should be. | 6.89 |
| H3158 | Crossover | An inflatable decelerator's drag benefit crosses over with its mass penalty at a ballistic coefficient: below it the inflatable wins and above it the rigid aeroshell does, for the same entry mass. Two decelerators sized to one entry mass, tested in one facility at two ballistic coefficients, should reverse their ranking across the crossover. | 6.8 |
| H3188 | Limiting-case collapse | A Whipple shield's hole and its rear-surface spall are one shock process seen at two standoffs, so threshold thickness plotted against standoff should be one curve for both. Two damage modes would give two curves. A standoff series with hole diameter and spall area both measured puts them on one curve or does not. | 6.8 |
| H3109 | Path dependence | With clearance-fit bolts the load a joint carries is set by which bolt seats first, so measured stiffness depends on assembly sequence and not only on geometry. Two joints from identical parts in opposite bolt orders should differ by more than the part scatter, and re-torquing in reverse order should move the stiffness without changing anything else. | 6.73 |
| H3112 | Conservation audit | A pyroshock's high-frequency content is the strain energy stored in the preloaded bolt, released into the structure in a time set by the bolt's own wave transit. Spectrum amplitude above a few kilohertz should therefore scale with preload rather than with charge size. Firing one charge against three preloads tests it, and a preload-independent shock would falsify the account. | 6.69 |
| H3126 | Hidden variable | Vacuum thrust measured in a finite facility carries a term from the plume's back-flow pressurizing the nozzle's exterior, and that term changes sign when the flow separates because the separated plume expands into a different volume. Two facility pressures at one chamber condition should move measured thrust by an amount that reverses at separation onset, which is the test. | 6.69 |
| H3151 | Threshold | There is a surface temperature below which a reusable silica tile's coating does not devitrify and above which its emissivity changes permanently, and that threshold marks the boundary between the tile's reusable and recession regimes. A tile heated in steps, with emittance measured after each step at room temperature, should show a step change rather than a gradual drift. | 6.69 |
| H3121 | Reciprocity | Reciprocity should fail measurably across a slipping bolted interface, because the slip dissipates differently depending on which side is driven. The asymmetry of the transfer function is then a measure of the slip rather than an inference from damping. Driving the same structure from both sides at increasing force, and plotting asymmetry against force, puts the microslip threshold where the asymmetry leaves zero. | 6.6 |
| H3157 | Null claim | A spallation-resistant coating should make no difference to recession in a pure-shear environment, because it resists compressive char loading and a pure-shear environment supplies none. Coated and uncoated samples at four shear levels should converge at the high-shear end, and that convergence is the test. | 5.0 |
| H3143 | Sign argument | A lateral thrust component measured on a gimballed engine reverses when the gimbal's offset is reversed, so a misalignment trend across a lot that fails to reverse belongs to the adapter. A trend that does reverse belongs to the engine. Two engines re-mounted in swapped adapters test it. | 4.93 |
| H3173 | Sign argument | An apparent pitch misalignment measured on an air bearing reverses with tank fill, because propellant motion adds a delayed torque; a trend that tracks fill is a propellant effect, not a thruster one. Firing at three fill levels in one orientation separates them by the sign of the reversal. | 4.91 |
| H3200 | Energy budget | The energy a shaker delivers should equal what the article dissipates plus what the fixture dissipates. The residual is the fixture's loss, usually the largest term and almost never measured. Driving the fixture with a dummy mass of known damping and then with the article separates them by subtraction. | 4.91 |
| H3118 | Null claim | Reverberant acoustic tests in two chambers should agree above the chamber's Schroeder frequency and disagree below it, because below that frequency the field is a few modes rather than diffuse. Testing it needs one article in both chambers on the same suspension, and that comparison would also expose any facility-dependent margin. | 4.87 |
| H3162 | Dimensional argument | The recession-to-char-depth ratio is the only dimensionless group that should govern an ablator's mechanical loss, so two tests with that ratio matched but durations different must agree on it. A disagreement indicates a time-dependent mechanism the group cannot capture. | 4.87 |
| H3142 | Boundary-condition ownership | Above some chamber pressure the pressure measured at the nozzle exit belongs to the cell's diffuser rather than to the nozzle, so the altitude performance inferred from it is a facility property. One nozzle with two diffuser geometries, at one chamber condition, should move the inferred exit pressure while thrust stays put. | 4.84 |
| H3155 | Hysteresis | A ceramic tile's first heating produces a larger expansion than later cycles because microcracks close irreversibly, so a thermal-vacuum distortion test on a never-cycled article measures a transient flight will not see. Pre-cycling one article three times and comparing its distortion with a virgin twin's at one gradient isolates the one-time set. | 4.84 |
| H3133 | Crossover | Ignition of a torch igniter is most reliable away from stoichiometric, because a rich kernel starves and a lean one is quenched at the injector's cold wall before it reaches the main chamber. Ignition probability against mixture ratio should be a hump whose peak moves with wall temperature, and two wall temperatures test it. | 4.8 |
| H3199 | Irreversibility | Environmental test order changes the outcome: vibration before thermal vacuum opens connectors that thermal vacuum alone leaves closed. The reverse order produces fewer failures for the same levels, which is why test order is a variable and not a formality. Two identical units in the two orders at one level give the order's contribution directly. | 4.78 |
| H3140 | Ceiling | A regeneratively cooled engine's pressure is capped by the coolant's critical pressure rather than by the wall's temperature limit, because past critical the coolant cannot absorb the heat at the flow the wall allows. A hydrocarbon-cooled and a hydrogen-cooled chamber of one geometry should reach very different ceilings, and a pressure series to each ceiling tests it. | 4.73 |
| H3152 | Scale argument | Subscale ablator tests cannot reproduce flight recession because char thickness grows with the square root of time while mechanical erosion grows linearly, so a one-third-scale test over-weights erosion by about the square root of the scale factor (estimate). Two scales at matched enthalpy and duration, compared through the recession-to-char-depth ratio, test the scaling rather than the material. | 4.73 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4415 | Contact-geometry claim | Ink footprints are used to judge contact patch size, yet ink transfers only above a threshold pressure, so the printed area is smaller than the load-bearing area and the shortfall changes with load. The missing region is the low-pressure shoulder, which carries a growing share as the tyre deflects. Comparing an ink print with a pressure film at three loads on the same tyre and inflation would yield the threshold pressure and test whether it stays constant. | 6.29 |
| H4456 | Efficiency misattribution | Transmission efficiency measured at high load is used to predict cruise consumption, but at light load the oil's churning and the seals' drag dominate the mesh loss, and those terms do not scale with torque. A gearbox efficient at full throttle can be poor at a hundred newton-metres. A torque sweep at fixed speed with oil temperature held constant would separate the load-proportional loss from the offset. | 4.51 |
| H4483 | Vent-reversal claim | Exposed wheels are often fitted with arch vents, yet the vent's flow reverses at speed because the arch's pressure falls below the underbody's, so a vent that helps in the tunnel at low speed can hurt on the road. The reversal speed is where the two pressures cross. Vent mass flow and direction measured at four speeds, with the wheel's drag on a separate balance, would show the crossing. | 4.47 |
| H4421 | Compliance audit | Long pedal travel when the brakes are hot is attributed to fluid compressibility, but the fluid's bulk modulus falls by only about a fifth at a hundred and fifty degrees (estimate), while a rubber hose's expansion can account for several times that volume. The hose's contribution scales with pressure squared and the fluid's linearly, so the two are separable. A pressure-volume measurement with a rigid line substituted for the hose, at three temperatures, would apportion the travel. | 4.29 |
| H4432 | Calibration error | Wheel-speed sensors are calibrated against a nominal rolling radius and every anti-lock threshold is expressed in that unit, so a tyre worn by five millimetres changes the speed the controller believes it sees by about one and a half per cent (estimate). That is enough to move a threshold tuned at the edge. Fitting tyres in three wear states and logging the controller's computed speed against satellite speed exposes the offset and tests whether the thresholds move with it. | 4.24 |
| H4466 | Cooling-drag claim | Fan power is measured on a bench at fixed speed, but on the car the fan's operating point is set by ram pressure at the radiator face, and at speed the ram flow can drive the fan backward so that it draws power instead of moving air. Fan drag is therefore a function of road speed, not a constant. Fan shaft torque and radiator-face pressure logged across a speed sweep would place the point at which the fan stalls and reverses. | 4.2 |
| H4404 | Identity claim | Radial force variation measured on a uniformity machine and the rim's own runout are treated as separate quantities, yet once the tyre is mounted they are one: the bead conforms to the rim and a runout of two-tenths of a millimetre appears as a first-harmonic force at the spindle. The tyre's stiffness supplies the conversion factor, which means the same tyre measures differently on different rims. Mounting one tyre on three rims of measured runout and regressing the first harmonic against runout should place the points on a line through the origin if the identity holds. | 3.87 |
| H4406 | Scale argument | The speed at which a standing wave locks onto a rolling tyre should vary as the square root of inflation pressure and inversely as the square root of tyre radius, since both enter the wave speed along the deflected carcass. That makes the pressure sensitivity mild, about five per cent of critical speed for a ten per cent pressure change (estimate), which is why pressure checks alone seldom explain a failure. A speed ramp on a drum, deflection measured at the crown and repeated at three pressures and two tyre sizes, would test both exponents at once. | 3.84 |
| H4472 | Reynolds claim | Drag coefficients from a quarter-scale model are higher than the full-scale value, and the difference is charged to Reynolds number, but much of it is the model's boundary-layer transition position, which trips elsewhere on a smooth model. Transition strips placed at the full-scale location should remove most of the discrepancy without changing Reynolds number. Two scales, with and without strips, would separate the two causes. | 3.56 |
| H4419 | Wear-step attribution | Inner pads usually wear faster than outer ones, and the step is put down to caliper flex; on a sliding caliper the inner pad also runs hotter because it sits on the piston's heat path, and wear rises with temperature. The explanations differ in what a swap does. Exchanging the pads' positions and repeating the duty cycle should reverse the step if heat is the cause and leave it in place if flex is. | 3.36 |
| H4401 | Misattribution | Footprint pressure recorded on a pressure-sensitive plate is read as the pressure the tyre would put on a road, yet the plate's own bending stiffness keeps the carcass from sinking into its cells, so the edges come out flattened. The plate alters the boundary condition it is meant to observe, and the error grows as its stiffness approaches the tyre's. Two plates of different stiffness, one steel and one compliant, should give edge-to-centre ratios that differ by fifteen per cent or more at the same inflation and load (estimate). | 3.02 |
| H4490 | Coincidence claim | Mass per area is a poor way to choose a panel treatment, because coincidence sets the frequency above which the panel radiates efficiently, leaving a mass-only treatment ineffective below it and over-designed above. The crossing is set by thickness and stiffness. Transmission loss measured on one panel above and below its coincidence frequency, at equal mass, would show the transition. | 2.67 |
| H4493 | A-weighting claim | Improvements measured in A-weighted decibels can be insensitive to low-frequency changes, so a treatment that removes a thirty-hertz boom may show almost no dBA improvement while the car is plainly quieter. The weighting attenuates a change at that frequency by more than thirty decibels (estimate). The same measurement in unweighted third-octave bands would reveal the change the dBA number hides. | 2.67 |
| H4481 | Skin-friction claim | Skin friction on a fastback's rear glass is neglected because the flow is assumed separated, but the flow can reattach ahead of the spoiler and the friction there becomes a measurable share of drag (estimate). That share grows with speed while pressure drag saturates. Oil-film interferometry on the rear glass at three speeds would locate the reattachment line and give friction's contribution. | 2.62 |
| H4445 | Loading-state bias | Alignment is set at curb weight, but the rear subframe's bushings deflect under driver and fuel load, so the running thrust angle differs from the one on the rack. The bias is systematic and explains why some cars pull only when fully loaded. Toe measured optically at three load states, with pressures and ride heights held, would give the compliance steer and its direction. | 2.58 |
| H4457 | Thermal-lag claim | Turbo boost limit is set by the wastegate actuator, whose spring preload changes as the can heats, so the boost curve drifts over the first minutes of a pull. The drift is a spring-rate change, not a control error. Boost logged against time with the can's temperature measured, on a rig where the can is heated independently of the turbine, yields the preload's thermal coefficient. | 2.56 |
| H4438 | Fixture artefact | Damping measured on a damper dynamometer includes the rod seal's friction, which appears as a force offset roughly constant with velocity and therefore inflates the low-speed coefficient. The error is largest at the small amplitudes that matter for ride. Running the damper at zero velocity with the gas charge fitted, then with it released, would separate seal friction from gas force and leave the hydraulic coefficient. | 2.53 |
| H4468 | Torque-split claim | Torque split is quoted from a differential's geometry, but the split at the wheels is set by the tyres' rolling radii, and a pressure difference of two-tenths of a bar between sides is enough to bias it measurably (estimate). The bias then feeds back into differential temperature. Wheel torques measured on a split-roller rig at matched and mismatched pressures would give the bias and its sign. | 2.53 |
| H4459 | Vapour-lock claim | Injector dead time is calibrated on a bench at room temperature, but fuel vapour pressure at the tip rises with temperature and a bubble forming during the low-pressure phase delays the needle's opening. The error appears as a lean excursion at hot idle rather than as a constant offset. Dead time measured at three fuel temperatures, with the tip instrumented, would locate the step and its onset. | 2.51 |
| H4469 | Ventilation claim | Cooling drag is estimated from the radiator's ram flow, but much of the air entering the grille leaves through the wheel arches and the underbody, so drag is set by total inlet flow rather than cooling flow. Sealing the arch exits changes cooling drag without changing radiator flow. Radiator flow and total inlet flow measured together, with exits sealed one at a time, would apportion the two. | 2.51 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4248 | Causal chain — springback | The density of a calendered electrode continues to change after the roll because the binder relaxes and the particles rearrange, so a density measured immediately and one measured a day later can differ. A specification written on the short-time unloaded density will therefore drift, and the drift should scale with binder content rather than with calendering pressure. | 6.73 |
| H4253 | Probabilistic threshold — burr penetration | A burr causes a short only if its height and sharpness are sufficient to penetrate the separator under stack pressure, so there is a critical burr size below which the defect is benign. That threshold should depend on separator thickness and modulus rather than on burr height alone, which means an absolute micrometre specification is measuring the wrong variable. | 6.69 |
| H4260 | Symmetry — winding direction | Reversing the winding direction reverses the sign of the bending stress in the electrode, so a symmetric cell should perform identically in both directions and any measured difference is evidence of separator or electrode anisotropy. The test is simple, and the size of the asymmetry sets a limit on how much cell-to-cell variation can be attributed to the winding process itself. | 6.64 |
| H4284 | Scaling — formation time | Formation time scales with the square of electrode thickness if it is limited by electrolyte diffusion, but only linearly if it is limited by the interfacial reaction, so the scaling exponent identifies the limiting process. Measuring formation time at three thicknesses is enough to choose between the two, and the answer decides whether faster formation should come from thinner electrodes or from a better electrolyte. The exponent is a diagnostic rather than a fitting parameter. | 6.33 |
| H4257 | Selection — hi-pot alignment | A hi-pot test preferentially detects defects that happen to align with the applied field and are large enough to pass the current threshold, so the tested population is selected for a particular geometry of defect. Rotating the field or testing at two polarities should change which cells fail, and a cell that passes both is not necessarily defect-free but a cell whose defect is misaligned. The escape population is therefore geometric and can be estimated by rotating the field. | 6.24 |
| H4203 | Percolation threshold — drying-determined | Below a carbon content estimated near 2 wt% (estimate), electrode conductivity is decided less by how much carbon is present than by where drying leaves it, because a late redistribution can build or break the percolating cluster. Conductivity versus carbon content on freeze-dried and oven-dried films should share a high-content plateau and diverge below the threshold, and the divergence is the drying contribution made visible. | 5.56 |
| H4215 | Symmetry — addition order | Swapping the addition order of carbon and binder should be exactly equivalent if the two adsorb independently, so any measured difference is a direct measure of competition rather than of mixing history. The test is cheap, and the sign of the difference identifies which species wins the surface. | 4.93 |
| H4225 | Artefact — edge curl | Curl at a coated edge is attributed to surface tension, but a through-thickness drying stress gradient can produce the same sign and a similar magnitude. Coating the same formulation at two oven temperatures and measuring curl at fixed residual solvent separates them, because the stress gradient scales with drying rate while surface tension does not. | 4.78 |
| H4236 | Inversion — calendering rate | Calendering should improve rate capability by improving particle contact, but it also raises ionic tortuosity, and the contact benefit saturates while the tortuosity penalty keeps growing. A density series measured at two rates should cross, with the high-rate optimum at a lower density than the low-rate optimum, an inversion that a single-C-rate study cannot see. | 4.78 |
| H4270 | Selection — easy-fill cells | Cells that fill easily are those with larger or better-connected pores, so a study that uses only well-wetted cells is selected on the very transport property it intends to measure. Filling a batch and comparing the slowest quartile with the fastest should show a rate-capability difference that is a pore-geometry effect rather than an electrolyte effect. | 4.76 |
| H4294 | Conserved quantity — capacity ledger | End-of-line capacity must equal formation capacity minus the capacity consumed by SEI, self-discharge, and any irreversible structural change, and a discrepancy larger than measurement uncertainty means one term is missing. Closing the ledger on a sample of cells is a better audit of formation than a capacity specification, because it localizes the loss instead of only flagging it. | 4.71 |
| H4213 | Counterfactual — mixing order | If only the final adsorption equilibrium mattered, adding binder as a dry powder or as a pre-dissolved solution should give the same electrode at the same final adsorption. Any systematic difference between the two routes measures competitive adsorption, carbon blocking binder sites or binder coating carbon, and that difference rather than the order itself is the mechanism worth reporting. | 4.69 |
| H4246 | Path dependence — calendering temperature | Calendering warm and calendering cold to the same final density can leave different binder distributions, because binder mobility during the nip determines whether it is pushed aside or deformed with the particles. Two temperature series matched on final density, one measured for adhesion and one for impedance, should separate the mechanical from the transport consequence of the binder's relocation. | 4.69 |
| H4264 | Omitted variable — case stiffness | Two cells built from the same electrodes and separator but housed in cases of different stiffness experience different stack pressures, so their impedance and cycle life differ even though the electrode is nominally identical. Case gauge is therefore a hidden process variable, and a stack-pressure measurement on the finished cell should predict the performance difference better than any electrode measurement. | 4.67 |
| H4280 | Confounding — rack position | Formation is exothermic, and heat is removed through tabs and surfaces, so cells at different positions in a rack experience different temperature histories under a single nominal setpoint. Instrumenting the rack at the same positions used for electrical testing should show whether the formation protocol is actually a temperature protocol, and if it is, rack design is a formation variable. | 4.67 |
| H4282 | Selection — efficiency outliers | Cells with unusually high first-cycle efficiency are often kept as good, but a high efficiency can mean a thin or incomplete SEI that will grow later, so selecting on efficiency can enrich for unstable films. Following the high-efficiency outlier group through aging, rather than discarding it, would test whether the selection is inverted. The comparison is between outlier groups, not between means. | 4.62 |
| H4290 | Artefact — fixture pressure | The contact resistance measured at end of line includes the fixture's probe-to-tab interface, which depends on probe pressure and on the tab's oxide layer. A four-wire measurement at two pressures, or a fixture with a Kelvin probe, should show a pressure-dependent offset that belongs to the fixture, and without that control cell-to-cell comparisons contain a term larger than the effect being studied. | 4.62 |
| H4238 | Hidden term — elastic recovery | Gap-set and pressure-set calendering are not equivalent, because the electrode stores elastic strain and springs back after the nip, so the density measured offline is the unloaded density rather than the density the particles experienced. A gap-set and a pressure-set sample matched on final thickness should differ in particle contact and impedance if recovery is significant, and that difference is the hidden term. | 4.6 |
| H4229 | Symmetry — one-sided drying | Drying a coated foil from one side imposes a through-thickness solvent gradient that cannot be symmetric, so the two faces of a single-sided coating should differ in binder content and porosity. Flipping the web in the oven, or drying both sides simultaneously, should change the face-to-face difference only if the gradient is significant, so the size of the change measures how much the second side matters. | 4.53 |
| H4269 | Conserved quantity — electrolyte ledger | The electrolyte added to a cell must be found in the electrode pores, the separator, the headspace, and the dead volume of the fill port, and any claimed starvation must close that ledger before it can be believed. Weighing the cell before and after filling, and measuring the displaced volume, is enough to detect the common error of counting the fill amount as the pore-fill amount. | 4.53 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2875 | Threshold | The transition from adhesive to abrasive wear occurs when counterface asperity height becomes comparable to wear particle size, a ratio of two lengths rather than a material property. The transition should therefore move if particle size is changed at fixed counterface. Imposing a controlled particle size in the contact, and finding where the mechanism changes, tests it. | 6.73 |
| H2845 | Scale argument | Diffusion at the micrometre scale in tissue is not Fickian when obstacle spacing approaches the molecule's own size, and the controlling parameter is that ratio rather than porosity. Two matrices of equal porosity but different obstacle spacing should therefore differ in diffusivity by more than porosity predicts. Building such a pair and measuring with matched tracers tests the argument. | 6.69 |
| H2883 | Threshold | Toughness of a mineralized composite is matrix-dominated below the mineral percolation threshold and mineral-dominated above it, so a kink in toughness against mineral fraction should appear at the threshold rather than elsewhere. The prediction is that the kink's position does not move when the organic phase is changed. Testing two organic phases across the same mineral series checks it. | 6.69 |
| H2807 | Scale argument | Mechanical sensing of nanotopography is inferred from cell response, but a focal adhesion plaque is tens of nanometres thick and laterally much larger, so it averages any topography shorter than its own extent. The relevant scale for mechanical detection is therefore the plaque, and features below it should be invisible however sharp they are. Growing the plaque to different sizes on the same topography, and finding the size at which a response appears, tests this. | 6.33 |
| H2886 | Artefact | Low-frequency viscoelasticity measured on dentin includes the testing frame's compliance and the specimen's own fluid, and both scale with specimen size in ways the material does not. Testing three specimen sizes at matched aspect ratio, and checking whether loss tangent is size-invariant, separates them. | 4.98 |
| H2868 | Artefact | Anomalous case-two swelling kinetics is often the specimen's skin buckling rather than a transport regime, since a wrinkled skin changes apparent thickness without changing mass. Measuring mass and thickness simultaneously, and checking whether the volume change accounts for the mass change, identifies it. | 4.96 |
| H2872 | Hidden term | A pin-on-disc wear rate contains a term from counterface roughening during the test, and that term is not constant, so a steady-state rate is a moving average over a changing contact. Profiling the counterface at intervals, and computing wear rate against instantaneous rather than nominal contact pressure, removes it. | 4.91 |
| H2818 | Artefact | The induction period before measurable mass loss in a porous ceramic is frequently read as a chemical nucleation delay, but it is more likely the time for the pore network to fill. Vacuum infiltration before immersion should erase it while leaving any true nucleation delay intact. That single comparison discriminates. | 4.89 |
| H2887 | Unit trap | Contact-mechanics formulae assume a half-space, and a millimetre-scale specimen on a centimetre-scale holder violates that assumption; the resulting error depends on specimen thickness relative to contact radius, not on the material. Testing one material at three thicknesses, and plotting apparent modulus against that ratio, exposes the violation and yields the correction. | 4.87 |
| H2900 | Scale argument | The mechanical-matching literature compares moduli measured at different scales and calls the comparison a mismatch. The scale-free invariant is the strain energy density at the interface, computable from any consistent pair of measurements. Recomputing a set of reported pairs as energy densities, and checking whether the apparent mismatches persist, closes the argument. | 4.84 |
| H2884 | Reversal | Crosslinking collagen usually stiffens it, but it can lower toughness by suppressing the sacrificial sliding that dissipates energy, so toughness peaks at intermediate crosslink density. The peak sits where sliding length equals the distance between crosslinks. A crosslink series tested in tension, with sliding length inferred from the strain at yield, should show that peak. | 4.78 |
| H2865 | Threshold | A water-swollen polymer turns glassy above a strain rate at which water cannot redistribute within the deformation time. The threshold is the inverse poroelastic diffusion time, so it moves with the square of specimen thickness, a prediction no intrinsic viscoelastic transition would make. Testing one polymer at three thicknesses, and watching the transition rate shift, tests it. | 4.73 |
| H2897 | Bound | Interfacial shear strength cannot be driven above that of the weaker bulk material by any surface treatment, so past a certain treatment strength all treatments become equivalent. The bound belongs to the bulk rather than the interface, which is why plateaus appear in treatment series. Testing treatments of increasing strength until measured strength stops rising locates it. | 4.73 |
| H2831 | Count | A layered system carries as many independent stress-transfer modes as it has layers minus one, since each interface transmits one mode and the assembly carries one rigid-body reference. A three- layer system therefore has two modes, and reporting three double-counts one of them. Modal decomposition of a measured strain field across a three-layer specimen tests the count directly. | 4.71 |
| H2852 | Bound | A bending-dominated lattice cannot be softer than the limit set by its connectivity, given strut modulus and relative density. Scaffolds reported below that bound are either closed-cell, so that struts are not the load path, or were measured in a way that included the frame. Computing the bound for each reported geometry, and checking where reports fall, tests it. | 4.69 |
| H2859 | Unit trap | Porosity as a percentage hides the pore-size distribution, and two scaffolds of identical porosity can differ tenfold in permeability because permeability scales with the square of a characteristic pore radius. Reporting porosity alone is therefore an omission of a dimension, not a simplification. Measuring permeability on a pair matched on porosity but not on pore size is the demonstration. | 4.69 |
| H2892 | Threshold | Below a critical ratio of specimen thickness to indenter radius the substrate contributes to the measured modulus, and the threshold depends on the modulus ratio rather than on absolute dimensions. This is why one coating measures differently on different substrates. Testing a thickness series on two substrates of different modulus, and finding where the curves separate, locates the threshold. | 4.69 |
| H2822 | Selection effect | Degradation studies measure the specimens that stayed intact, and the ones that fragmented are precisely those in which the mechanism of interest ran fastest. The reported rate is therefore a lower bound on the true spread, and the bound tightens as the mechanism accelerates. Deliberately measuring the fragment size distribution instead of discarding fragments tests the size of the bias. | 4.67 |
| H2838 | Hidden term | Every tracer diffusivity measured in tissue contains a partition coefficient: what is reported as diffusivity is really diffusivity divided by one plus the bound fraction. A tenfold reduction can therefore be tenfold binding rather than tenfold hindrance. Measuring the equilibrium uptake of the same tracer on the same specimen supplies that coefficient, and dividing it out leaves the true diffusivity. | 4.67 |
| H2847 | Selection effect | Only solutes that reach the sampling site are counted, so measured transport in tissue is biased toward fast paths and away from dead ends. The bias can be estimated by running two tracers of different diffusivity together and comparing their ratio at the outlet with their ratio at the inlet. A ratio that drifts with path length is the signature. | 4.67 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4724 | Storage-delivery distinction | Legionella control is judged at the calorifier because the temperature there is easy to measure, but the risk lives in the branch, whose surface temperature is set by insulation, ambient and draw pattern. A cylinder held well above the disinfection threshold can coexist with a colonised dead leg, and only a sample taken at the branch after a stagnation period tests the thing that matters. | 4.56 |
| H4753 | Context transfer | Wind pressure coefficients measured on isolated models in a smooth boundary layer do not transfer to a city canyon, where the flow is channelled, the leeward pressures differ, and the canyon can even reverse the sign of the difference on some faces. A natural-ventilation design validated on an open site is not validated for its own site, and the error is not a constant factor. | 4.56 |
| H4745 | Boundary omission | Counting the evaporator alone cannot close a chiller plant’s energy balance: the condenser’s heat rejection, the tower’s evaporation and the pump work must be counted together, and a test that omits them reports a coefficient the balance cannot reproduce. The missing terms are the compressor’s shell losses and the pump heat, and a plant that looks compliant on the chiller’s own numbers can fail the building’s meter. | 4.51 |
| H4706 | Sublinear response | Cold jets do not drop in proportion to their supply velocity, because spreading dilutes momentum and buoyancy per unit volume together, so more air is a weaker remedy for a dumping diffuser than the throw curves suggest. The test is a two-flow series at fixed supply-to-room temperature difference: the drop point should move less than the flow ratio, and if it moves more, the diffuser’s own geometry is setting the separation. | 4.42 |
| H4705 | Non-scaling offset | Affinity laws assume that everything measured at the fan scales with speed squared, but the transmitter’s zero offset and the duct’s own stack pressure are additive terms that do not scale at all. Below about a third of design speed (an estimate) those terms become a large fraction of the reading, so the inferred pressure rise is too high and the inferred efficiency too low, and a trend fitted with a cube through the whole range will report a false decline at the bottom end. | 4.11 |
| H4767 | Saturation of a benefit | Moving air extends the comfort range through convection and evaporation, and in humid air the evaporative part saturates while the convective part continues, so the speed at which a fan stops helping is lower in a humid room. A design that specifies fan speed from a dry-climate test over-promises in the tropics. | 2.84 |
| H4727 | Set-point mismatch | Arrestor precharge is set at installation against fill pressure, but the pressure it sees varies with the expansion tank’s state and the relief valve’s setting. An arrestor correct at fill is undersized at the top of the range, and the peak it was meant to absorb reappears precisely when the system is hottest. | 2.82 |
| H4776 | Condition-dependent specification | Colour temperature is quoted at the rated condition, but the phosphor shifts as the junction warms, and the shift is larger for warm-white than for cool. A fixture measured cold is warmer in a plenum, so the colour consistency of an installation is a function of its thermal design rather than of binning alone. | 2.8 |
| H4756 | Path over rate | Hood capture is tested with make-up air supplied so as not to disturb the plume, but in a real kitchen the make-up enters at the nearest opening and can cross the hood’s face. The hood that passes can fail in service, and the make-up’s path, not the hood’s exhaust rate, is the variable that decides capture. | 2.76 |
| H4707 | Selection effect | Reported savings from duct sealing are drawn from buildings selected because a screening test found them leaky, so the published mean describes the screened population rather than the building stock. A randomised trial that seals half of an unscreened sample would give the expected saving. The gap between the two numbers is the cost of the screening. | 2.73 |
| H4732 | Symptom misreading | An undersized vent stack does not announce itself with a gurgle; it throttles the air behind the slug and the fixture drains slowly, which reads as a blockage. The tell is that the slow fixture is the one furthest from the stack and that rodding changes nothing. The test is a pressure measurement in the vent, not another rodding eye. | 2.67 |
| H4749 | Definition artefact | Temperature efficiency for a heat-recovery unit is computed from supply and exhaust temperatures, but the ratio is defined for balanced flow, and with unequal airflows the number depends on which stream is the capacity-rate minimum. The same core therefore reports different numbers at different flow ratios, and a commissioning sheet that compares them is comparing test conditions rather than cores. | 2.67 |
| H4769 | Wrong surface | Floor heating is limited by surface temperature, but the temperature the occupant feels is at the sole of the shoe, and shoe insulation and contact resistance set the difference. A barefoot test therefore sets a limit for a condition nobody occupies, and a floor at the limit can feel cold in a thin sole and warm in a thick one. | 2.67 |
| H4794 | Impossibility | One temperature sensor cannot distinguish a zone warmed by the sun from one warmed by people, because both raise the same variable while the correct responses differ: the first calls for cooling, the second for ventilation. The controller’s choice is a guess about an unmeasured cause, and no tuning removes the ambiguity, only biases it toward the more common cause. | 2.67 |
| H4725 | Steady-test blind spot | Certification tests a thermostatic mixing valve on steady flows, but its sensor has thermal mass, so a step in draw passes a temperature excursion before the valve responds. The fail-safe claim is a claim about the transient, and a test that steps the flow from full to a trickle, rather than holding each condition, is the only one that exercises it. | 2.64 |
| H4702 | Misidentified index | Sizing a run by equal friction equalises loss per metre, not total pressure at the terminals, and the index run is then often the branch with the most fittings rather than the longest one. The symptom is a starved terminal fed by generous ductwork, and the test is a total-pressure traverse at the terminals rather than a static reading at the fan. | 2.62 |
| H4717 | Stratification threshold | Thermoclines are usually credited to insulation, but the thickness of the transition layer is set by the inlet’s Richardson number, so a tall tank with a strong inlet jet mixes as thoroughly as a short one. Insulating the shell cannot restore a thermocline the inlet destroyed, and the test is to vary inlet velocity at fixed geometry and watch the outlet temperature’s decay. | 2.6 |
| H4719 | Degeneracy | Roughness and diameter are degenerate when only the total drop at one flow is known, because many pairs produce the same drop. A second flow separates them, since roughness enters the friction factor at high Reynolds number while diameter enters both the velocity and the factor, and a design that infers internal condition from a single test is fitting one equation to two unknowns. | 2.58 |
| H4748 | Metric's missing variable | Effectiveness measured by dosing the supply and reading the extract describes the room’s mixing, not the occupant’s exposure, because a contaminant released at the floor and one released at the ceiling have different histories in the same airflow. The same system can score well and still deliver a poor dose to a seated person. The source’s location is the variable the metric leaves out. | 2.58 |
| H4759 | Index insufficiency | Predicted mean vote is a whole-body index, and it says nothing about the vertical gradient or the radiant asymmetry a seated person feels, so a room can be neutral on the index and still be rejected. The index is a necessary condition for comfort and not a sufficient one, and using it as a compliance test requires the local terms to be checked separately. | 2.58 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H416 | Threshold | Step bunching sets in only above a critical miscut, below which the elastic repulsion between steps beats the attraction that would pair them, so the critical angle measures that repulsion directly. Vicinal cuts spanning 0.2° to 3° (estimate), annealed identically, should show the transition in the shape of the terrace-width distribution rather than in its mean. Broadening with anneal time would mean kinetics, not an elastic threshold. | 8.51 |
| H412 | Selection effect | Published adsorption-energy distributions are truncated by the pump-down that precedes every measurement, since species bound more weakly than roughly 40 kJ/mol (estimate) leave before the first spectrum is taken. Dosing at 10⁻⁶ mbar (estimate) and cooling before evacuation should restore the missing low-energy tail, and the tail's position should track how long the chamber was pumped. The restored tail's weight then bounds how much of the literature's "strong-site" population is evacuation. | 8.42 |
| H470 | Artefact mimics effect | Oxygen-evolution activity on a supposedly stable catalyst can be the metal dissolving and redepositing, with the corrosion current counted as catalysis. Measuring dissolved metal in the electrolyte alongside the current, over a series of catalysts, should correlate activity with dissolution if the current is corrosion. | 7.0 |
| H476 | Hysteresis asymmetry | Forward and reverse scans diverge when the surface reconstructs, and the size of the divergence measures the reconstruction's kinetics rather than the redox couple's. Triangular scans at one rate but different vertex dwell times should alter the asymmetry only if a slow surface process is behind it. | 6.96 |
| H482 | Reversal | Potassium promotes at low coverage and poisons at high, and the coverage where it turns is set by the alkali's own ordering transition rather than by a smooth electronic optimum. Sweeping coverage with the ordering measured by diffraction in the same chamber should put the reversal at that transition. | 6.91 |
| H456 | Identity | Contact resistance from a transfer-length structure and interface resistance from a four-point measurement are one quantity under two boundary conditions, and the difference between them is the current-crowding length rather than a second resistance. Structures of several contact widths, measured both ways, should collapse onto one curve against width over crowding length. | 6.84 |
| H480 | Bound | Ninety per cent of an XPS signal can come from a depth of a few nanometres (estimate) at the kinetic energies in common use, so a "surface" enrichment measured this way is an average over several atomic layers. Sweeping the take-off angle through its widest range should change the apparent enrichment, and that change is the bound made visible. | 6.69 |
| H422 | Boundary-condition ownership | Chlorine-free is not a property a surface owns; the bulk reservoir decides it, and at the temperatures where the surface looks clean the bulk is still delivering chlorine at a rate comparable to the experiment's own timescale. Two samples pre-treated to set two bulk activities, then quenched to one measurement temperature, should differ in surface coverage at equal nominal cleanliness. | 6.67 |
| H454 | Misattribution | Adhesion numbers from a pull test often describe the capillary force of the water layer between film and probe, and between 30% and 70% relative humidity (estimate) the number can move by a factor of two. Repeating the test in dry nitrogen and across a humidity series on one sample should give a straight line whose intercept is the adhesion. | 6.67 |
| H459 | Count | Modelled as a slab, a grain boundary has at most two independent channel densities however many dislocations it contains, because a measurement sees only the channels parallel and perpendicular to it. Boundaries of several misorientations, measured in both directions, should fall on a two-parameter family if that is all there is. A third parameter would appear as systematic scatter about that family, which the two-direction measurement would catch. | 6.49 |
| H481 | Threshold | No clean metal surface survives above a threshold oxygen chemical potential, and the threshold is a line in pressure and temperature rather than a pressure, crossing the usual working range near 10⁻⁶ mbar (estimate). Approaching one chemical potential along two paths, hot and low-pressure versus cool and high-pressure, tests whether the line is thermodynamic or kinetic. A line that shifts with heating rate is kinetic, and "clean surface" then becomes a statement about the ramp. | 6.33 |
| H491 | Misattribution | Activity lost over a hundred hours is chalked up to sintering, and the same loss appears when the support's surface hydroxyls have been consumed, because the support was co-catalysing the reaction all along. Regenerating with water below the sintering threshold should restore activity if the loss was chemical and leave it flat if it was morphological. A support whose hydroxyls cannot be restored should show the loss as irreversible, which is the cleaner half of the test. | 6.29 |
| H437 | Boundary-condition ownership | The hydrogen order a reactor reports can belong to the gas-side boundary layer rather than to the surface, because at high pressure a transport limit is first order in hydrogen whatever the catalyst does. Raising the linear velocity at fixed partial pressures should move the measured order toward the intrinsic one, and the direction of that move says which regime the old number came from. Were the order immovable with velocity, transport would be dead as an explanation. | 6.27 |
| H468 | Count | A single Tafel slope admits only a few mechanisms: at room temperature, 30, 40 and 60 mV/dec (estimates) each map onto specific rate-determining steps, and a slope between them is an average over a distribution rather than a new step. Current-interruption measurements at several currents, which recover the true surface overpotential, should collapse the distribution onto one of the discrete values. A slope that holds one intermediate value across a tenfold change in current is a mechanism, not an average. | 6.22 |
| H435 | Unit trap | Turnover frequencies per exposed atom assume one probe molecule per surface atom, and on many metals hydrogen titrates closer to two (estimate), so the number is high by that factor and a bimetallic "promotion" can be nothing but the change in stoichiometry. Titrating each sample with two adsorbates and counting both, instead of importing a handbook ratio, should make the promotion disappear or survive. Whatever the second adsorbate's stoichiometry turns out to be, it must be measured on the same surface. | 6.2 |
| H500 | Inversion of standard practice | The denominator decides the ranking: normalise one dataset by mass, by area, by site count and by turnover, and four catalysts can reorder three times. Publishing all four rankings for one benchmark reaction in one laboratory would turn a habit into a test, and the reversals would be the result. | 4.89 |
| H493 | Bound | Dispersion has a ceiling at a given loading, set by the chemical potential of a metal atom on the support against the bulk, and below roughly a nanometre (estimate) atoms dissolve into the support instead of forming particles. Loading series measured by chemisorption and by microscopy should diverge exactly where dissolution starts. | 4.84 |
| H407 | Scale argument | Below a cluster of roughly a hundred atoms (estimate) there is no facet for a binding energy to be a property of, and the measured size-dependence is really a dependence on edge-atom fraction. Mass-selected clusters of known atom count, measured in one apparatus, separate the two because count and edge fraction no longer vary together. | 4.78 |
| H448 | Count | One reflectivity curve at one wavelength carries a phase and an amplitude, so it can fix at most two independent thickness parameters however many layers the stack has. Measuring at three wavelengths, where the layers' indices differ, raises the count — and the count can be checked by fitting synthetic data whose thicknesses are known. | 4.78 |
| H498 | Boundary-condition ownership | Measured lifetimes can belong to the reactor wall, which catalyses a side reaction that eats the feed or the product, leaving the catalyst to take the blame for the wall's chemistry. A reactor lined with an inert material against one lined with the catalyst's own material should give lifetimes that differ by the wall's contribution. | 4.78 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5295 | Pore solution's evolution | The pore solution's composition is measured by expression from a young paste, but the composition changes as the C-S-H's Ca/Si falls, so a young sample's solution is not the solution the steel will see. A chloride threshold predicted from a young paste's pH should therefore be optimistic. | 6.96 |
| H5282 | Carbonation moves the baseline | Carbonation shrinkage is measured on a specimen that carbonates from the surface inward, but the carbonated skin is stiff and restrains the uncarbonated core, so the measured contraction is the composite's response and not the C-S-H's. A thin specimen, carbonated throughout, should shrink more per unit of CO2 than a thick one, which would show the restraint. | 6.73 |
| H5285 | Moisture's own creep | Drying creep is often attributed to the stress's effect on drying, but the drying itself generates pore-scale stresses that creep the C-S-H without any external load, so a specimen dried under a small load creeps more than the sum of the two tests. The extra creep should scale with the drying rate rather than with the load. | 6.73 |
| H5291 | Ice in the wrong pores | Freeze–thaw damage is attributed to the water that freezes in the capillary pores, but the gel pores' water freezes far below the test's range, so the damage is done by a small fraction of the water and the rest is a spectator. The damage should scale with the capillary water's volume rather than with the total freezable water, and the two diverge as the gel densifies. | 6.53 |
| H5284 | Recovery misread as damage | Creep recovery is small in concrete, which is read as evidence of damage, but a viscoelastic material with a broad relaxation spectrum recovers little on the test's timescale even without damage. The recovery measured over a longer interval should approach the predicted value, and the apparent damage should shrink as the observation window grows. A load held for a year and then released would show whether the missing recovery belongs to time or to damage. | 6.33 |
| H5243 | Autocatalysis by product | The conversion of CAH10 to C3AH6 releases water, and that water raises the local water content and speeds the conversion that produced it, so the process is autocatalytic in a way a simple rate law misses. A converted region should therefore spread rather than form uniformly, which is visible as a conversion front in a specimen kept wet and split open. The front's position would be measurable by mapping the hydrate assemblage across a split face. | 6.31 |
| H5233 | Rate versus amount | The sulfate's effect on the aluminate reaction is discussed in terms of how much gypsum a cement contains, but the reaction sees the sulfate's dissolution rate, so a coarse gypsum and a fine one at the same mass give different heat curves. The optimum SO3 is therefore a property of the gypsum's grind as much as of the clinker. Regrinding the gypsum at constant SO3 should move the optimum, which is a test no specification currently asks for. | 6.27 |
| H5209 | Scaling with the wrong variable | The sulfate optimum of a cement is indexed to the C3A percentage, but sulfate is consumed at the aluminate's surface, so the demand should scale with the aluminate's specific surface rather than with its mass. Two clinkers with the same C3A and different interstitial textures should need different gypsum, and the difference should follow the texture. | 5.76 |
| H5223 | Charge compensation decides the sign | Magnesium in alite is said to stabilise the M3 polymorph, but the stabilisation depends on how the substituted magnesium's charge is compensated, and sulfate and alkali offer different compensation routes. The same magnesia should push the polymorph in opposite directions in a low-sulfate and a sulfate-rich clinker. A pair of clinkers differing only in sulfate would show the reversal in the alite's cell parameters. | 5.58 |
| H5296 | Alkali's other source | Alkali–silica reaction is blamed on the cement's alkali, but some aggregates release alkali from their own feldspars, so a low-alkali cement with such an aggregate can still produce a high-alkali pore solution. The aggregate's release should be measurable as a rise in the expressed solution's sodium over time. | 4.93 |
| H5294 | Sulfate's own source | Sulfate attack is diagnosed by an external sulfate source, but a clinker with a high sulfate content can release sulfate from the C-S-H during drying or carbonation, so the attack can proceed with no external supply. The diagnosis should then include the paste's sulfate budget, which is rarely measured. | 4.91 |
| H5248 | Coupled properties reported as one | Fresh paste's thixotropy is attributed to the flocculation of cement particles, but the same flocculation is broken by the superplasticiser's adsorption, so a paste with a high dosage should be less thixotropic even at the same yield stress. The two properties are measured together and reported as a single workability. | 4.89 |
| H5250 | Reversible term inside a clock | The heat of hydration is used as a clock, but part of the measured heat is the water's adsorption on the C-S-H, which is released again on drying and is not part of the reaction's enthalpy. A degree of reaction from calorimetry therefore contains a reversible term that a bound-water measurement would not. | 4.82 |
| H5239 | Attribution across components | The chemically bound water of a silica-fume blend is used as a measure of reaction, but the fume's own surface adsorbs water that is not chemically bound and leaves at the same temperature as some of the C-S-H's water. The measured degree of reaction therefore contains a physical adsorption term that grows with the fume's fineness. | 4.76 |
| H5237 | Space bound | Below a water content of roughly 0.38 by mass of cement, which is an estimate rather than a measurement, there is not enough space for the products of complete hydration, so the reaction must stop before the anhydrous phase is consumed. The stopping point is set by the C-S-H's density rather than by the water, so a paste with a denser C-S-H can hydrate further at the same water content. | 4.44 |
| H5247 | Mixing energy as a chemical variable | Mixing energy is filed under handling, but it breaks the C-S-H's flocculates and exposes fresh surface, so a paste mixed harder has more nucleation sites and hydrates faster for reasons that are not chemical. Two pastes of identical composition mixed to different energies should differ in their heat curves for days. Two laboratories with different mixers are therefore not measuring the same paste, even when the mix design is identical. | 4.44 |
| H5266 | Competitive adsorption, non-additivity | Two admixtures in one mix compete for the same surface, and the stronger adsorbing one displaces the weaker, so a combination can perform worse than either alone at the same total dose. The effect should appear as a non-additive flow at fixed total dosage, and it is why a package cannot be optimised component by component. The displacement can be measured directly by comparing each polymer's adsorbed amount alone and in the pair. | 4.38 |
| H5238 | Parameter drift | The late hydration rate is fitted with a diffusion coefficient through the inner product, but that coefficient is not a constant: it falls as the inner product densifies, and the fit absorbs the fall into a single number. A rate law with a time-dependent coefficient should fit the same data with a coefficient that is a material property. The test is a hydration series at several temperatures, because a drifting coefficient and an activation energy leave different signatures. | 4.27 |
| H5251 | Degeneracy of two mechanisms | Two proposed mechanisms for the induction period, a metastable barrier and a nucleation lag, predict the same heat curve at a fixed water content and differ only in their response to a change in the amount of free surface. Seeding the paste with ground clinker should shorten a nucleation lag and leave a barrier intact, so one experiment separates them. The two mechanisms also differ in their temperature dependence, so a seeding test at two temperatures would close the argument. | 4.22 |
| H5300 | Bound that closes the set | Every durability mechanism in this list, chloride, sulfate, carbonation and alkali, is slowed by the same buffer: the C-S-H's capacity to hold and release the ion in question, so a material's durability is bounded by its buffer capacity per unit of transport. The bound is a ratio rather than a rate, and it explains why two concretes with the same diffusion coefficient can have very different lives. The ratio can be compared by measuring two ions that share a diffusion coefficient but differ in binding. | 4.11 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1813 | Derived quantity | Equivalent age at 20 °C is treated as a state variable for maturity, yet the activation energy it depends on shifts with water-to-binder ratio and with the presence of slag. Two specimens brought to the same equivalent age along different temperature paths, then tested for bound water and strength, should differ by more than the maturity rule allows, which would make equivalent age a bookkeeping device rather than a physical coordinate. | 8.42 |
| H1862 | Bound | Permeable pavement discharge is bounded by the subgrade's infiltration rate, not by the surface block's permeability, so a more permeable block over a low-permeability subgrade cannot increase system capacity. A full-scale test with two surface materials over the same subgrade, instrumented at the base, should show identical outflow once the subgrade is saturated. | 6.82 |
| H1876 | Bound | Envelope airtightness is bounded by its penetrations, not by wall area, because a few unsealed service entries can dominate leakage and membrane over the field of the wall cannot compensate. A blower-door test before and after sealing the penetrations, with the membrane unchanged, should show most of the improvement coming from the penetrations. | 6.82 |
| H1884 | Unit trap | Low-frequency noise complaints are assessed with A-weighted levels, which discount the frequencies that cause the complaint, so a plant room that passes the dB(A) criterion can still be the source of a disturbance. Re-measuring with C-weighting and a third-octave spectrum should show the low-frequency energy the A-weighting removed, and the complaint should track that energy. | 6.78 |
| H1847 | Scale argument | Small-scale footing tests in sand cannot reproduce the particle crushing under a prototype's confining stress, so the measured capacity is governed by dilation rather than crushing and the scaling law changes at a stress threshold. Three g-levels in the same sand should show the capacity factor departing from the constant-friction prediction above the stress where crushing begins. | 6.73 |
| H1900 | Threshold | Ground-penetrating radar will miss a void beneath a reinforced slab below a size threshold set by bar spacing, because the mesh reflection masks the deeper event within one range bin. Surveying test slabs with voids of decreasing size at fixed depth should show detection collapsing at a size that scales with bar spacing rather than with wavelength. | 6.73 |
| H1851 | Derived quantity | A liquefaction potential index is a weighted integral of factor of safety over depth, so two profiles with the same index can have opposite layering and the index cannot say which layer liquefies first. Constructing two profiles with equal index but reversed layer order and testing both in the centrifuge should produce different surface settlements and pore-pressure histories. | 6.71 |
| H1846 | Artefact mimics effect | Apparent liquefaction in a centrifuge model can be a boundary artefact, because pore pressure generated near the rigid container wall migrates inward and reaches the instruments at a delay that mimics liquefaction onset. Repeating the test with a flexible boundary or with instruments moved away from the wall should shift the apparent onset while the soil behaviour is unchanged. | 6.69 |
| H1860 | Hidden term | Culvert capacity computed from inlet control omits the outlet's backwater, which on a flat catchment can raise tailwater above the barrel invert for most of the design storm and convert the culvert to outlet control. Measuring headwater and tailwater together through a storm should show capacity falling short of the inlet prediction when tailwater rises, not when flow peaks. | 6.69 |
| H1881 | Scale argument | Small-chamber measurements of volatile emission are limited by boundary-layer resistance at the sample surface, so the emission rate per unit area depends on chamber airflow and not only on the material. One product at three chamber sizes and two air-change rates should show the apparent rate converging only when the boundary layer is thinner than the sample's diffusion length. | 6.69 |
| H1890 | Null result | Across retrofitted houses, added insulation thickness should show no correlation with measured gas consumption once pre-retrofit consumption and household size are controlled, because the rebound in internal temperature absorbs the theoretical saving. A before-and-after design with a control group is needed to show the null, and the null would make the theoretical saving a poor predictor of the bill. | 6.69 |
| H1841 | Identity | In a normally consolidated clay, secondary compression and primary consolidation are one process sampled at two time scales, with the end-of-primary criterion an arbitrary point on a continuous strain–log-time curve. Running a load step for ten times the conventional duration and looking for a break in slope should find none, which would make the separation an artefact of test duration. | 6.67 |
| H1877 | Threshold | Convective heat loss in a vertical cavity begins only above a critical Rayleigh number, so a cavity narrower than a threshold insulates better than a wider one even though the wider cavity holds more air. Measuring U-values of three cavity widths at the same temperature difference should show the narrowest performing best, with the ranking reversing above the critical width. | 6.67 |
| H1824 | Bound | Shot peening's gain in fatigue life is bounded by how much of the induced compressive stress survives the first few thousand cycles, not by the surface stress measured immediately after treatment. X-ray diffraction after interrupted cycling should show the benefit decaying with the same relaxation constant as the residual stress, so a treatment with a higher initial stress but faster relaxation loses its advantage. | 6.58 |
| H1829 | Scale argument | Fatigue tests on small specimens cannot be scaled to large members by stress alone, because the probability of finding an initiating inclusion in the highly stressed volume grows with that volume and the small specimen is effectively screened. Three sizes at identical stress, with life plotted against stressed volume, should give a slope steeper than the elastic prediction, which would identify initiation as the controlling step. | 6.53 |
| H1808 | Artefact mimics effect | Reports of autogenous healing in cracked concrete often measure a healed surface rather than a restored ligament, because carbonation precipitates calcite across the crack mouth within days and blocks the water that would reveal the interior. Curing identical specimens in CO2-free water and in air should separate the two, since only the air-cured set will show a load recovery that disappears when the mouth is ground back. | 6.51 |
| H1828 | Artefact mimics effect | Acoustic emission attributed to crack growth in a fatigue test can come from fretting in the loading fixture, and the two sources are not separated by amplitude alone. Running a dummy specimen with a saw-cut instead of a crack, at the same load and frequency, should produce a hit rate that accounts for a measurable share of the crack emissions and whose spectrum matches the fixture resonance. | 6.51 |
| H1867 | Scale argument | Clogging tests on permeable concrete use a sediment gradation chosen for convenience, and the clogging rate depends on the ratio of sediment size to pore size more than on sediment mass, so a lab result cannot be scaled to a pavement receiving wind-blown dust. The same pavement with three gradations spanning the field range should collapse the rate onto a curve in the size ratio rather than in the mass. | 6.44 |
| H1826 | Reversal | Under a broken passive film and a small anode-to-cathode area ratio, a stainless steel fastener can lose section faster than a galvanised carbon-steel one, because the stainless cathode is efficient while the zinc sacrifices itself over a large area. Immersing both types in chloride solution with a controlled anode area should reverse the usual ranking at a specific area ratio. That ratio is measurable and is not the ratio assumed in the standards. | 6.38 |
| H1802 | Hidden term | Chloride diffusion coefficients measured by bulk immersion carry a term for chloride bound to the C-S-H, and that term is not constant: it grows as the free concentration falls, so a profile fitted with one apparent D will underpredict ingress at low exposure concentrations. Refitting the same profile with binding switched off, using an isotherm measured separately, should move the fitted D by an amount that tracks the isotherm's slope rather than the paste's porosity. | 6.33 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H636 | Scale argument | The optimal stopping index in a semi-convergent iteration scales inversely with the square of the noise level, a consequence of stopping when the residual matches the noise rather than when it stops falling. Halving the measurement noise should therefore move the optimum by a factor of four, not two. A noise-injection sweep on a single dataset finds the optimum without any new measurement. | 6.6 |
| H642 | Bound | A model trained on one instrument's data cannot beat that instrument's calibration accuracy on absolute quantities, but it can beat it on ratios, because calibration error is common-mode and cancels. The bound predicts where transfer learning will and will not work, and it needs no new measurements to check. Test one ratio quantity and one single-channel quantity; the ratio should transfer and the absolute should not. | 6.53 |
| H697 | Bound | The number of correct digits surviving an inverse problem is set by the forward operator's condition number, so precision quoted after inversion is illusory past that count: a condition number of 1e6 destroys six digits regardless of the arithmetic's word length. The bound applies to the recovered parameter rather than to the residual. Perturb the data at the tenth digit and measure how far the parameter moves. | 6.51 |
| H672 | Count | The number of parameters identifiable from a single experiment is bounded by the number of independent instrument channels, not by the number of data points, because one channel's samples are repeated measurements of one projection — fits that claim more are trading parameters against each other. Fix the data, add one genuinely independent channel, and the identifiable count should rise by one rather than by the sample count. | 6.49 |
| H645 | Reversal | Adding training data can degrade a model when the data are heteroscedastic and the loss is unweighted: the noisiest regime dominates the gradient and the fit regresses toward noise. More data from the noisy regime is worse than less data from the quiet one. Weight the loss by inverse variance and the degradation should reverse — a test that separates data quantity from data quality. Volume is not a monotone good, and the inflection is measurable without new experiments. | 6.24 |
| H658 | Count | A record of N real samples carries at most N/2 independent spectral amplitudes, so a peak found by fitting more parameters than that is fitting the window's shape. The bound explains why "significant" narrow features proliferate as fitting freedom grows. Reduce the model to N/2 parameters and the feature count should collapse to the number of resolvable lines. | 5.71 |
| H601 | Bound | Accuracy in a long integration is bounded by the bits the state sheds per step, not by the integrator's order: the error grows linearly in step count at a rate proportional to machine epsilon times the state's magnitude. Doubling the word length should move that bound by exactly a factor of two, so a run that improves more than that was never round-off-limited. Comparing binary32, binary64 and binary128 trajectories at fixed truncation error separates the two cases. | 5.29 |
| H694 | Count | A digitised record's information content is bounded by the converter's resolution times the number of independent samples, and independence is set by the analogue bandwidth rather than the sample rate, so a twelve-bit converter at a megasample per second behind a one-kilohertz front end carries of order 2^12 × 2×10^3 distinguishable states per second (estimate), not 2^12 × 10^6. Extra samples are interpolation. Decimate by a hundred and the recoverable information should be unchanged until the decimation reaches the bandwidth. | 5.22 |
| H631 | Hidden term | The "convergence factor" printed by many solvers is a residual scaled by the preconditioner, so it measures the preconditioner and only partly the solution. Changing the preconditioner can make a stalled solve look healthy without improving the answer at all. Print the true residual from the original operator once every hundred iterations and compare the two histories. | 4.73 |
| H682 | Bound | The information rate through a measurement chain is bounded by its slowest stage's bandwidth, and a fast sensor behind a slow amplifier cannot resolve transients no matter what is done to the data afterwards. The lost information was never sampled, so no filter recovers it. Step a known fast edge through each stage and find where it broadens. | 4.71 |
| H661 | Identity | Baseline drift and low-frequency noise are the same process observed with and without a high-pass filter, so detrending changes the measured exponent and not the process — a claim that a signal "becomes white after detrending" is a statement about the filter's corner. Apply two detrending orders to one record and compare the spectra above the corner; they should agree there. | 4.64 |
| H688 | Conditioning | The condition number of a calibration matrix is a physical quantity equal to the ratio of the instrument's dynamic range to its resolution, so an ill-conditioned calibration is a hardware property and not a numerical accident — it predicts which parameter combinations the instrument cannot separate before any data are taken. Measure it beforehand and it should match the parameter covariance observed afterwards. | 4.6 |
| H678 | Threshold | Past some data volume the provenance record grows faster than the data itself, because per-sample metadata is fixed while per-sample information falls; beyond that point the archive's binding constraint is curation rather than storage. The crossover is computable from the schema's size and the sample's entropy. Plot provenance bytes per measurement byte against record length for one instrument and the knee should be visible. | 4.58 |
| H683 | Selection effect | Archives receive the results that were significant, so the archived distribution is truncated and any variance computed from it is biased low — a property of the archive rather than of the field's measurements. The bias is computable from the truncation point whenever near-misses were also deposited. Compare an archived spread with a pre-registered batch and the truncation appears as a deficit just below threshold. | 4.56 |
| H664 | Scale argument | The variance of a measured mean falls as 1/n only while the record is longer than the correlation time; for shorter records it falls as 1/n² with an offset set by the correlation, so extrapolating the 1/n law badly overstates precision. The crossover is the record length itself and is measurable. Split one long record into segments and plot the variance of the segment means against segment length. | 4.49 |
| H663 | Hidden term | Every correlation coefficient computed on an autocorrelated series has an effective sample size of roughly n/τ, with τ the integral timescale, so the significance threshold is wrong by a factor of about sqrt(τ). This is why geophysical correlations "significant at p<0.01" fail to replicate. Estimate τ from the autocorrelation and rescale the p-value; most of the surprise will go. | 3.69 |
| H660 | Misattribution | Low-frequency excess reported as 1/f noise is often the instrument's gain drift filtered by its own feedback loop, whose transfer function turns a slow exponential into a power law across two decades. The exponent then belongs to the loop, not to the noise process. Open the loop, or record the loop's error signal, and the power law should change slope or disappear. | 3.62 |
| H622 | Hidden term | Observed discretisation error carries a term from the boundary condition's own truncation, which for a one-sided implementation is one order below the interior scheme and does not shrink when the interior is refined. Reports of "sub-second-order" convergence in otherwise clean codes are usually this term. Swap in an exact boundary closure at fixed interior mesh and the slope should jump by an order. | 3.6 |
| H638 | Closure error | In a partitioned multiphysics solve the flux mismatch at the interface is not noise but a conserved quantity that accumulates as a bias, and its sign identifies which subsystem was solved last. Codes that report this mismatch as a generic "residual" hide the structure that would localise the defect. Reverse the solve order: the bias should change sign while its magnitude stays fixed. | 3.6 |
| H626 | Artefact mimicking effect | A staircase representation of a smooth boundary pins the contact line at cell corners, and the resulting apparent contact angle becomes a function of grid orientation rather than of surface chemistry. The artefact is small in angle but large in hysteresis, and hysteresis is the quantity usually reported. Rotate the interface relative to the grid: the hysteresis should change while the angle does not. | 3.58 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1583 | Count | Folded-beam flexure stages impose 6N constraints against six rigid modes, so they are over-constrained by construction and their surplus constraints appear as parasitic modes rather than as pure stiffness. Adding beams for stiffness adds modes in the same proportion. Test: count the modes in a measured frequency response and compare with the beam count. | 6.8 |
| H1593 | Unit trap | Jerk limits in metres per second cubed and in g per second differ by 9.80665, and trajectory generators often take the number without its unit, so a profile tuned on one machine is violent on another. The symptom is a spectral line at each segment boundary rather than a smooth limit. Test: compute commanded jerk as the third difference of the profile and compare it with the intended limit. | 6.47 |
| H1526 | Scale argument | Flexure stiffness falls as the cube of thickness while its range rises only inversely with thickness, so the force available at full deflection falls as the square and shrinking a stage tenfold cuts that force a hundredfold at constant strain. Miniature stages therefore cannot be compared with their large counterparts through any single similarity number. Test: build one topology at three thicknesses and plot available force against thickness on logarithmic axes; the slope should be two. | 6.31 |
| H1505 | Identity | Integrator windup and actuator rate limiting produce the same overshoot envelope after a large step, which is why they are confused during commissioning and both blamed on too much integral. They differ in what saturates — the integrator's state or the actuator's derivative — and that difference survives a controlled change. Test: clamp the integrator but keep the rate limit; overshoot that remains belongs to the rate limit, and its height should scale with the step's velocity content. | 6.27 |
| H1574 | Hidden term | Squareness error in the reference artefact folds into the machine's straightness measurement and cannot be separated without a reversal or a second artefact of different size. Most error budgets quote the machine's straightness and silently include the square. Test: measure with two squares of different provenance and compare the difference. | 4.89 |
| H1554 | Bound | Vision-servo gain is bounded by the product of exposure time and image velocity, since beyond a few pixels of blur the feature centroid's bias grows faster than gain can compensate. Frame rate is irrelevant to this bound; only exposure enters it. Test: hold frame rate fixed, vary exposure, and measure phase margin. | 4.84 |
| H1569 | Misattribution | The hysteresis of a micrometer is assigned to elastic recovery of the frame, but at hand speeds it is mostly flank friction in the thread, which is why it grows when the thimble is turned quickly. Elastic recovery would be rate-independent. Test: approach one reading at two speeds and compare the overshoot on reversal. | 4.8 |
| H1550 | Artefact mimics effect | Hysteresis measured on a torque sensor during a slow ramp is often the fixture's stiction rather than the sensor's compliance, and it appears only on the first ramp after a reversal. Genuine sensing hysteresis is rate-independent and survives a dwell at zero torque. Test: reverse at two rates, with and without a hold at zero. | 4.78 |
| H1599 | Misattribution | Lost motion at reversal is often the coupling's wind-up rather than the screw's backlash, and the two separate by locking the output: a coupling shows free input rotation with the output clamped, while backlash does not. Most machines are diagnosed without that clamp in place. Test: clamp the output and measure the input's free travel. | 4.78 |
| H1567 | Unit trap | An arc-second is 4.848 microradians, and stage specifications mix the two, so a straightness in microradians and an angular error in arc-seconds can differ by a factor of five inside one document. The mixture survives because both numbers are small. Test: recompute one specification chain entirely in SI and compare it with the vendor's error budget. | 4.76 |
| H1589 | Threshold | Above an acceleration threshold a recirculating guide's rolling elements transition from rolling to sliding, and the threshold is the ratio of inertial force to preload rather than any speed. The transition is sharp, produces a step change in friction, and looks like a controller fault. Test: ramp acceleration slowly at fixed speed and record friction force. | 4.76 |
| H1595 | Bound | Usable bandwidth of a ball-screw drive is bounded by the screw's first axial resonance, which falls inversely with length, so long-travel machines cannot be fast however strong the motor. The bound is mechanical and no controller moves it. Test: measure the axial resonance on two screws of different length and compare with the predicted inverse relation. | 4.76 |
| H1533 | Threshold | Below a preload threshold a linear guide's rolling elements never unload, so stick-slip disappears outright rather than merely diminishing, the threshold being the load at which the external force variation equals the preload. Above it the same guide shows ordinary Stribeck behaviour. Test: sweep preload at fixed external oscillation amplitude and find where the hysteresis loop collapses. | 4.73 |
| H1580 | Unit trap | Newton-metres and pound-feet differ by about 1.356, and a wrench calibrated in one unit used against a budget written in the other produces a preload error of that factor, which sits inside the scatter of most bolted joints and is therefore never diagnosed. Test: instrument one joint with a load washer and compute the preload both ways. | 4.73 |
| H1564 | Hidden term | Thermal uncertainty in a calibration is the artefact's expansion coefficient times the difference between calibration and use temperature, not its absolute value, so a high-coefficient artefact used where it was calibrated beats a stable one used five kelvin away. Datasheets quote the coefficient and conceal the difference. Test: calibrate at two artefact temperatures and compare the residual scale error. | 4.69 |
| H1570 | Artefact mimics effect | Dead-path error in a laser interferometer appears as a scale error when the dead path changes between measurements, and it is easily mistaken for the machine's calibration because it is proportional to distance. Measuring two lengths with the same dead path separates the two. Test: compare a short and a long measurement with the reference mirror moved between them. | 4.69 |
| H1585 | Scale argument | Angular stiffness in a flexure pivot falls steeply with blade thickness while range rises, so a pivot built for a large stage cannot be scaled down without changing topology, since the ratio of range to stiffness is not invariant. Miniature versions need more blades, not thinner ones. Test: build one topology at three thicknesses and plot range against stiffness. | 4.69 |
| H1538 | Identity | Detent torque and cogging torque are one physical quantity measured on opposite sides of the gearbox: one is the magnet's preference for a position, the other is the same preference referred through the reduction. Listing them as separate specifications counts the same energy twice. Test: measure at the motor shaft and at the output, refer both to one side, and compare. | 4.64 |
| H1548 | Unit trap | An accelerometer's sensitivity in millivolts per g differs from the SI value by 9.80665, and the error propagates into every gain that converts acceleration into current. Because g is close to ten the mistake survives a factor-of-ten sanity check and is blamed on the mechanism. Test: calibrate the same sensor on a shaker against laser vibrometry and compare the two sensitivities. | 4.64 |
| H1597 | Hidden term | The encoder scale's thermal expansion sits outside the loop, so a metre of steel scale at one kelvin moves about eleven micrometres — an estimate — which is larger than most stages' resolution and comparable to their accuracy. Feedback holds the readhead to the scale, not the tool to the part. Test: measure one artefact with the scale at two temperatures. | 4.64 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2277 | Count | Crack growth has two regimes, one dependent on stress intensity and one plateau, and a third threshold regime is usually a transient of the test rather than of the material. Constant-intensity tests held long enough for the transient to decay should erase the apparent threshold regime and leave only the two. | 6.87 |
| H2282 | Selection effect | Stress corrosion tests are run on alloys that survived fabrication, so heats that cracked during forming were removed before the laboratory saw them and the tested population is the survivors. Testing the failed heats alongside the survivors should leave the survivors' threshold above the service failures, which is the selection effect made visible. | 6.82 |
| H2261 | Bound | Protection potential is bounded below by hydrogen evolution at the local pH, and going further negative adds no protection while causing disbondment and hydrogen uptake. Stepping potential below the hydrogen line while measuring disbondment and permeation should raise disbondment and permeation together while the corrosion rate stays at zero, so the extra polarisation buys only damage. | 6.76 |
| H2285 | Hysteresis | The potential at which cracking stops is not the potential at which it starts, because the tip's own chemistry holds the crack open below the initiation threshold. Stepping potential down after growth is established should give an arrest potential below the initiation potential measured on a smooth specimen, and the gap should widen with crack depth. | 6.76 |
| H2291 | Bound | Ultrasonic thickness resolution on a hot wall is bounded by the sound velocity's temperature dependence rather than by the transducer's frequency, because the velocity change across the wall's gradient blurs the echo. A hot calibration block and a cold one measured with the same probe should give apparent thicknesses differing by more than the probe's specified resolution. | 6.73 |
| H2259 | Misattribution | The hundred-millivolt criterion is applied to the measured potential shift, which includes the IR drop, so the true polarisation is smaller than the number being compared with the criterion. Current-interruption measurement of the IR-free shift on the same structure would find the criterion met less often than the uncorrected survey suggested, and the shortfall should scale with soil resistivity. | 6.69 |
| H2275 | Threshold | A crack cannot keep growing by film rupture below a critical tip strain rate, because repassivation outruns the exposure of bare metal. A strain-rate series with repassivation kinetics measured separately should place the cessation of growth where the repassivation time equals the interval between ruptures, and the threshold should move with the alloy's repassivation rate rather than with its yield strength. | 6.64 |
| H2280 | Identity | Sulfide stress cracking and hydrogen embrittlement are one mechanism with a poison in the environment, and the sulfide's role is to raise the fraction of cathodic hydrogen that enters the metal. Permeation and cracking measured on one steel at matched charging current, with and without sulfide, should tie the cracking threshold to the permeation flux rather than to the sulfide concentration. | 6.64 |
| H2218 | Artefact mimicking effect | Pits counted after a standard coupon test include a population initiated at grinding scratches, which is a preparation artefact and not a material response. Coupons from one heat finished by grinding, by diamond polishing and by electropolishing should give three different pit densities at the same potential, and the ground specimen's excess should vanish when the scratches are removed without changing the alloy. | 6.6 |
| H2297 | Scale argument | Laboratory coupons and field structures differ in area and in time by factors that do not scale linearly. Initiation and propagation have different exponents in each, so the ratio of their rates drifts with exposure. Coupons and a full structure exposed together, measured the same way, should make the ratio of their rates change with exposure time, and the change should follow the initiation exponent. | 6.56 |
| H2283 | Scale argument | Solution chemistry at a crack tip is confined to nanometres, where a bulk pH and a bulk potential have no meaning. The relevant variable there is the potential of zero charge at that scale. Coupling an atomistic tip model to a measured crack-tip potential should predict a local pH differing from the bulk by more than the bulk's own variation, and the prediction should change sign when the tip's surface charge does. | 6.4 |
| H2232 | Derived quantity | Exchange current density is a derived quantity twice over, current divided by an area and area inferred from a roughness factor, so a porous electrode's value can be wrong by one to two orders of magnitude without any single measurement being wrong (estimate). Measuring roughness independently by double-layer capacitance and by gas adsorption should shift both the area and the derived density, and the two corrections should account for the discrepancy between laboratories. | 6.38 |
| H2202 | Hidden term | The passive current density measured in aerated chloride solution is the sum of film formation and the reduction of whatever oxidant the cell contains. Trace ferric iron from the salt and oxygen leaking past the seal both contribute, so the quoted value is partly a property of the cell rather than of the alloy. Holding potential and temperature fixed while the oxygen partial pressure is stepped from about 1 ppm to air should leave a true passive current almost unchanged and move an oxidant-dominated one in proportion (estimate). | 6.04 |
| H2233 | Bound | A reaction with a chemical step before electron transfer cannot show a Tafel slope steeper than about 120 mV per decade at room temperature. The preceding step caps the sensitivity of rate to potential. Anything steeper is uncompensated resistance or a change of mechanism, and an impedance-derived ohmic correction applied to the same data should bring the slope back inside the bound. | 5.62 |
| H2269 | Selection effect | Field protection surveys report potentials at accessible test posts, and the inaccessible sections, under roads, in casings and at river crossings, are precisely the ones that fail. Instrumented sections with remote monitoring should record potentials outside the survey's distribution, and the discrepancy should be largest where access is hardest. | 4.91 |
| H2236 | Artefact mimicking effect | An inductive loop at low frequency in a corroding system is often the potentiostat's own control loop interacting with a high-impedance reference, not an adsorbed intermediate relaxing. Repeating the measurement with a low-impedance reference and in galvanostatic mode should remove the loop, and an intermediate that is real would survive both changes. | 4.84 |
| H2287 | Symmetry claim | Static-load and cyclic-load crack growth at the same maximum stress intensity are not comparable, because the cycle adds a ratcheting mechanism a static hold cannot reproduce. A static hold and a cyclic test at matched maximum intensity should give different growth rates, and the difference should shrink as the load ratio approaches one. | 4.82 |
| H2262 | Threshold | Along a protected line, potential uniformity holds only above a critical current density, because the attenuation length set by coating conductance and soil resistivity is shorter than the line. Measuring the potential profile at several current levels should flatten the profile abruptly once current exceeds that value, and the value should track the coating's conductance. | 4.78 |
| H2276 | Reversal | Cathodic protection can accelerate cracking by generating hydrogen at the tip, so the safe condition is a window between underprotection and overprotection rather than a single potential. Slow strain rate tests across a range spanning both should recover ductility and then lose it again, with the second fall coinciding with the onset of hydrogen uptake. | 4.78 |
| H2286 | Conservation claim | Hydrogen entering the metal plus hydrogen evolved as gas must equal hydrogen produced cathodically. A permeation efficiency below one is therefore not an inefficiency but a measurement of the recombination kinetics. Simultaneous permeation and volumetric gas measurement on one specimen should close the balance, and the missing fraction should vary with the poison's surface coverage. | 4.78 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H995 | Scale argument | At very short measurement intervals, the dominant error in a differential measurement is timing skew between channels, and its effect scales with slew rate rather than amplitude. A constant-amplitude sine sweep should show error peaking where slew rate peaks, not where the signal does. | 6.98 |
| H936 | Reversal | Raising field strength improves MRI signal-to-noise only until body-dominated noise grows faster than signal, at which point the trend reverses. Coil-loading measurements at two field strengths with one coil geometry should show the reversal as a change in which noise source dominates. | 6.93 |
| H993 | Bound | The smallest resolvable change of a sensor is bounded by the stability of its excitation whenever that excitation is common to the measurement path. Two excitations of different stability driving one sensor should show resolution following the excitation, not the noise floor. | 6.93 |
| H983 | Bound | Long-term stability of a radiometric measurement is bounded by its reference source rather than its detector, so a better detector cannot rescue a worse reference. Swapping two references of different stability between two detectors should make stability follow the reference. | 6.89 |
| H966 | Bound | Accuracy of a differential pressure measurement across a constriction is bounded by the tap's own flow disturbance rather than by the transducer, once the tap's bore exceeds a fraction of the lumen. Taps of decreasing bore in one flow rig should show a plateau where the transducer's specification no longer limits anything. | 6.84 |
| H959 | Hidden variable | Measured D-values depend on how aggregated the spore population is on its carrier, since an aggregate shields its interior, and aggregation follows the drying method. Suspensions dried two ways to equal viable count should yield two D-values. | 6.82 |
| H977 | Bound | Cuff-based measurement has a temporal resolution bounded by inflation and deflation, so a device reporting beat-to-beat pressures from a cuff is interpolating. Two readings requested a second apart should be more correlated than any physiological signal allows. | 6.82 |
| H905 | Threshold | Cavitation in coupling gel begins at a negative pressure fixed by the size distribution of undissolved gas nuclei, which is why two gels of identical acoustic impedance can give different measured output. Degassing one aliquot under vacuum while leaving its twin at ambient should shift the inception threshold by a ratio set by the nuclei's Laplace pressure. | 6.73 |
| H910 | Inversion of standard practice | Textbook practice puts a conductive ground plane behind a body-worn antenna to cut absorbed power, yet at frequencies where the plane spans a fraction of a wavelength it can concentrate the near field at the feed. Measuring peak local field with and without the plane at fixed radiated power locates the crossover thickness, which should scale with wavelength in tissue. | 6.67 |
| H951 | Identity | Dry-heat kill and steam kill of one organism are the same protein denaturation at different water activity, so dry-heat resistance should follow from steam resistance plus the sorption isotherm. One strain tested at three controlled water activities should collapse both curves onto a single one. | 5.0 |
| H971 | Unit trap | Defibrillator energy in joules and charge in coulombs are linked by the waveform's voltage, so a fixed-energy setting delivers different charge into a 25-ohm load than into a 50-ohm one (nominal test loads). A load-resistor sweep with voltage and current logged makes the difference explicit. | 5.0 |
| H994 | Identity | Zero drift and span drift are one physical process seen at two points of the transfer curve, so a device cannot have a stable zero and an unstable span. Measuring at two known inputs over time should show both terms moving in a fixed ratio. | 5.0 |
| H928 | Threshold | Below some contact area, an electrode's own thermal noise exceeds the amplifier's at every usable bandwidth, so selectivity bought by shrinking the electrode costs signal-to-noise that no front end recovers. An area series measured through one amplifier should show where total noise stops falling. | 4.98 |
| H963 | Misattribution | An oscillometric cuff does not measure pressure directly; it records the amplitude of its own pressure oscillations and derives systolic and diastolic from an assumed envelope shape. A cuff of different compliance but identical bladder geometry should shift the estimate while the true pressure is unchanged. | 4.98 |
| H926 | Hidden variable | Scatter in impedance between nominally identical implanted electrodes is dominated by how much geometric area the encapsulation actually left exposed, not by the metal, and that area varies with the moulding step. Optical area measurement on the same units before implantation should absorb most of the scatter. | 4.96 |
| H941 | Selection effect | Uniformity and noise figures come from phantoms chosen to fit the field of view and to be homogeneous, so they exclude the edge conditions that dominate a body-sized object. The same measurements on an oversized phantom should shift the numbers in the direction the beam-hardening correction dictates. | 4.96 |
| H927 | Symmetry argument | Exchanging the two electrodes of a bipolar pair must invert the signal and leave the noise unchanged, so any residual difference belongs to the amplifier or the tissue, never to the source. Recording both polarities and averaging cancels everything source-independent and leaves the asymmetry as a measurable residue. | 4.93 |
| H975 | Selection effect | Published accuracy for wearable thermometer patches is computed from patches that stayed adhered for the whole wear period, which excludes exactly the cases where thermal contact failed. Retaining and measuring every patch, detached ones included, should widen and shift the reported distribution. | 4.93 |
| H911 | Selection effect | Reliability figures for implanted sensors come only from units that survived sterilisation and handling, so the reported failure rate is conditional on passing the step most likely to damage the encapsulation. Sacrificial test structures measured after sterilisation but never implanted should expose the early failures the conditional sample hides. | 4.91 |
| H931 | Misattribution | Beam-hardening streaks are blamed on the polychromatic source, but a small object shows the same streaks under a monochromatic beam when the detector's gain depends on energy. A monochromatic scan with the gain curve measured and flattened should leave the streaks in place if the detector is the cause. | 4.91 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5602 | Solvent effect collapsed into one scale | The shift of a dye’s absorption maximum between solvents is indexed to a single polarity scale, yet for a dye whose dipole moment barely changes on excitation the refractive-index term should dominate and the polarity term should nearly cancel. On that reading the scale measures the solvent’s polarisability in disguise. A series of solvents matched in polarity but differing in refractive index would test it, and the differences needed are modest, of order 0.05 as an estimate. | 8.29 |
| H5644 | Mobility constraint | Hindered amine stabilisers cycle between an amine and a nitroxide, and the cycle needs the molecule to reach the radical site, so in a film below its glass transition the stabiliser is present but immobile. Its measured effectiveness then reports the film’s mobility rather than its own chemistry. One stabiliser in two binders of different transition temperature should give protection that follows the transition. | 4.58 |
| H5659 | Counterintuitive direction | Aggregated dyes have short excited-state lifetimes because the excitation is delocalised over many molecules, so an aggregate is often more lightfast than the same dye molecularly dissolved, and diluting a dye in a print can lower its lightfastness. That reverses the usual assumption that less dye fades less. Prints of decreasing dye load, with the aggregation state measured spectroscopically, would show fade per molecule rising. | 4.56 |
| H5640 | Contaminant as colorant | Nobody counts grinding-media wear as a colorant, yet the zirconia entering an ink from the beads is a high-index contaminant that scatters, so a long-milled batch can be lighter and less chromatic than a short-milled one at the same pigment size. The colour change is then a wear measurement. Ashing two batches milled for different times, with colour measured on the same drawdown, would show the correlation. | 4.51 |
| H5656 | Reversibility splits the effect | A faded print that recovers colour in the dark is said to have faded less than it appeared, but the recovery is a separate phenomenon: a trapped photoproduct relaxes or a reduced dye re-oxidises. The total change is the sum of a reversible and an irreversible part, and only the second matters for permanence. Colour measured immediately, after a dark interval and after re-exposure would separate them. | 4.51 |
| H5660 | Ranking not preserved | Accelerated lightfastness testing ranks materials at an irradiance far above service, and a mechanism of higher order in intensity is over-weighted in that ranking, so two materials can invert their order between the test and the window. The test is then a filter rather than a predictor. A small set of samples exposed at three irradiances to the same total dose would show whether the order survives. | 4.51 |
| H5652 | Diffusion-limited quantum yield | Below its glass transition a polymer film has almost no oxygen diffusion, so a dye that fades by a singlet-oxygen route slows not because its photochemistry changed but because the quencher cannot reach it. In that region the temperature dependence of fading should follow a diffusion activation energy, which is much larger than a photochemical one. Fade across the transition, with oxygen diffusion measured separately, would show two activations. | 4.47 |
| H5699 | Stress rather than chemistry | An ultraviolet-cured ink that fails on a flexible substrate is reformulated for adhesion, but cure shrinks the film and puts the interface in tension, so a well-bonded film can still lift at a scratch. The failure is then a stress problem, and a more flexible ink treats it by lowering the modulus. Films of several moduli on one substrate, with stress measured by cantilever curvature, would show the lift threshold. | 4.44 |
| H5649 | Front propagation | Fading in a thick film proceeds as a front, because the top layer bleaches first and then transmits more light to the layer beneath, so the rate at depth is not the rate at the surface. The depth profile of colour is a moving boundary problem, and a film opaque at the start becomes translucent as it fades. Sectioning a faded film and measuring colour across the section would show the front. | 4.4 |
| H5653 | Non-unidimensional scale | Blue wool standards are used as a scale of lightfastness, but each standard is a different dye in a different wool, so the scale orders mechanisms by their sensitivity to the test spectrum rather than along one axis. A sample can match one standard under a xenon source and another under daylight, and both readings are honest. Exposing the standards themselves under two sources and comparing their order would show the dependence. | 4.4 |
| H5668 | Oxygen inhibition | An ultraviolet-cured ink is inhibited by atmospheric oxygen at its free surface, because oxygen quenches the radicals before they propagate, so the topmost layer stays tacky while the bulk cures, and that layer is a larger share of a thin film than of a thick one. The measured tack is a boundary effect rather than an incomplete dose. Films of several thicknesses cured at one dose, with the surface layer removed before testing, would show the inhibited depth. | 4.27 |
| H5605 | Role reversal | A lake pigment’s permanence is credited to the dye and blamed on the mordant when it fails, but in a lake the cation is the framework: the dye is held as an insoluble salt whose coordination decides whether water and light can reach the chromophore. Aluminium and calcium lakes of one dye should therefore differ in lightfastness for a reason that is structural rather than chromophoric. The test is a series of lakes of one dye with different cations, exposed side by side. | 4.16 |
| H5611 | Undefined quantity | Indicator dyes are calibrated in water and then used to report the acidity of a coating, where the dielectric constant, the ionic strength and the solvation all differ, so the number they report is a pH only in the operational sense of that film. The claim is not that the measurement is wrong but that the quantity is undefined away from its calibration medium. A film holding two indicators of different charge type should report two different acidities if the effect is real. | 4.16 |
| H5686 | Scaling argument | Gravure cell emptying is a capillary-number problem, so the transferred fraction should collapse onto one curve against that number, but only if the cell’s aspect ratio is held: a deep cell empties by a different route from a shallow one at the same number. Cells of two depths at matched capillary number should then disagree, and the size of the disagreement measures the aspect ratio’s contribution. A transfer series across speeds for cells of several depths would show whether the collapse is real. | 4.16 |
| H5629 | Percolation threshold | Gloss collapses in an emulsion paint at a pigment volume concentration treated as a formulation constant, but the collapse is the point at which air voids connect through the film, which is a percolation threshold rather than a chemical one. That makes it sensitive to the pigment’s packing geometry and nearly insensitive to the binder’s chemistry. One binder with pigments of different shape should show the threshold moving with the packing. | 3.42 |
| H5682 | Energy partition | The diameter of a dot on paper is modelled as a spherical cap, but the drop’s kinetic energy is spent in spreading, in deforming the coating and in rebound, and rebound should stop mattering above a threshold speed, leaving dot diameter independent of further speed. A plateau in dot diameter against impact speed would be evidence for the partition, and a continuing rise would be evidence against it. Drops of one ink at several speeds, imaged at short times, would show which. | 3.18 |
| H5666 | Surface with a shelf life | Printing on polyethylene requires an oxidised surface, and that oxidation is not permanent, because low-molecular-weight chains migrate into the bulk and bury the polar groups, so the surface energy falls over days. The treatment’s age is then part of the substrate specification. Adhesion tested on one treated film at intervals would show the decay. | 2.8 |
| H5689 | Charge-limited development | Electrophotographic density is set by the toner’s charge-to-mass ratio, which the field between developer and photoconductor develops, so a toner that has aged or taken up humidity develops less density although its pigment is unchanged. The drift is then a triboelectric measurement. Density plotted against measured toner charge over a long run would show the relation. | 2.76 |
| H5662 | Competing effects | Feathering looks like excessive wetting, but lowering the ink’s surface tension also increases its penetration into the paper, so a surfactant that sharpens the edge can lighten the dot. The two effects pull in opposite directions on one variable. Lines printed from a surfactant series, measured for edge width and for density, would show the curves crossing. | 2.73 |
| H5667 | Time-dependent surface | Corona treatment is specified by its watt density, but its effect on wetting decays with the film’s storage temperature and humidity, so two rolls treated identically can print differently. The treatment is a state rather than a finish. Wetting measured on treated film stored at two temperatures, at intervals, would show the decay rate and its dependence. | 2.73 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H346 | Bound | A well's capture zone is bounded by the ratio of pumping to ambient groundwater flux, not by the visible cone of depression, so a well can intercept water from a river it never appears to dry. Streamflow depletion should then be measurable well outside the cone. | 6.98 |
| H351 | Scale argument | Aquifer correlation length estimated from a single test is dominated by the test's support until that support is about ten times the true length, so estimates grow with the scale of the experiment. A nested pumping-test series at one site should show the growth and where it saturates. | 6.93 |
| H328 | Bound | A plume's height is bounded by the tropopause only while its excess temperature stays below a value set by the stratification; hotter plumes overshoot, and the overshoot then measures the vent rather than the sky. Two eruptions of equal mass flux but different exit temperature should separate cleanly on that basis. | 6.87 |
| H314 | Two-clock race | Whether a patch slips fast or slow at a given stressing rate is decided by which is quicker: shear-induced dilatancy pressurizing the gouge, or drainage depressurizing it. The crossover rate is a laboratory measurement on the same gouge, and it should map onto the fast/slow boundary observed in nature if the mechanism is the same. | 6.78 |
| H330 | Count | Heat leaves a hydrothermal system through three channels at most — conduction, liquid advection, vapour advection — so a four-term budget is double-counting, usually the vapour's latent heat, which is already inside the enthalpy flux. A system with measured gas and liquid discharge should close with three terms and miss by a factor with four. | 6.78 |
| H361 | Scale argument | Bed roughness measured at a radar footprint is not the roughness that governs sliding; the controlling scale is the wavelength the ice can conform to, set by ice thickness. The same bed measured by radar and by a plough should give two different roughnesses, each correct at its own scale and each misleading at the other. | 6.76 |
| H374 | Derived quantity | Diffusivity computed from a dye release is the initial patch size divided by a growth rate, so the same water yields different diffusivities for different patch sizes; at these scales the flow does not possess a diffusivity at all. Two releases of different initial size tracked simultaneously should disagree by construction, and the disagreement is the measurement. | 6.73 |
| H309 | Scale argument | Below a network's station spacing, the completeness magnitude is set by the coda of the largest nearby event rather than by the smallest detectable one, so the two scales cross at a spacing-dependent magnitude. Thinning a dense array to sparse geometry should move completeness with spacing while the noise floor stays put — a test that costs only bookkeeping. | 6.69 |
| H308 | Artefact mimics effect | The daytime dip in microearthquake counts at most networks is the detection threshold rising with cultural noise, not a tidal or thermal modulation of the crust. Re-picking the same continuous records with a matched-spectrum detector and injecting synthetic events at fixed amplitude through the day should flatten the cycle; if it survives the injection test, something real is modulating the source. | 6.64 |
| H337 | Threshold | Infiltration excess begins at an intensity set by sorptivity rather than by saturated conductivity, and the two thresholds differ by a factor that depends on storm duration. A rainfall simulator at constant intensity for two durations would locate the sorptivity threshold; measuring conductivity would miss it. | 4.98 |
| H365 | Threshold | The active layer deepens abruptly above a snow-depth threshold, because snow insulates until it is also thick enough to yield meltwater that refreezes and releases latent heat. The two effects cross at a depth measurable by a snow-fence experiment with a depth series either side of the fence. | 4.93 |
| H349 | Identity | Tritium and CFC ages are not two estimates of one age but two moments of one residence-time distribution, so their disagreement measures mixing rather than contamination. A suite along a flow line should show the gap widening with the distribution's variance. | 4.91 |
| H366 | Hysteresis asymmetry | Glacier length responds to a mass-balance step faster on retreat than on advance, and the asymmetry is set by the valley's widening down-glacier rather than by ice dynamics. Two glaciers with the same balance history and different valley geometry should show different lags, which historical terminus records can test. | 4.91 |
| H387 | Reversal | Added aerosol can suppress rain from a shallow cloud and enhance it from a deep one, because the coalescence delay matters only while the cloud is shallow enough to evaporate before glaciation. The sign should change with cloud depth at a crossover measurable by tracking individual cells on radar. | 4.91 |
| H397 | Null space | A single-channel radiometer's temperature and humidity retrieval has a direction along which the radiance is unchanged, so two retrievals can fit one measurement exactly and disagree in the boundary layer. Adding a surface humidity measurement, which projects onto that direction, is the test rather than a longer radiance record. | 4.91 |
| H319 | Bound | Airborne gravity resolution is bounded by traverse height rather than by instrument precision: at a hundred metres above a compact target, a tenfold better gravimeter buys nothing at all. Flying the same line at two altitudes is the test, and the higher line will show no improvement from the better sensor. | 4.87 |
| H320 | Threshold | Secular variation becomes visible at an observatory when baseline drift exceeds the signal, which for most instruments arrives after about a decade, so apparent geomagnetic jerks cluster at the epochs of absolute measurement rather than at the jerks. Observatories with continuous absolute control should show a smoother curve at exactly those epochs. | 4.84 |
| H316 | Boundary-condition ownership | For the first months after grouting, a borehole strainmeter measures its own curing rather than the rock, because the compliance that sets its calibration is the grout's. Two instruments at the same depth in different rocks but identical grout should share a decay curve; the shared curve is the signature, and its absence the refutation. | 4.78 |
| H370 | Threshold | Langmuir circulation appears above a wind speed set by the ratio of Stokes drift to wind drift rather than by wind speed alone, so two days with the same wind and different swell are not equivalent. A site with long-period swell should show cells at lower wind than one without, which surface imagery can test. | 4.78 |
| H322 | Reversal | Postglacial rebound raises the crust while the geoid falls above it, so the sign of the measured gravity change reverses between a station near the former ice centre and one on its periphery. Viscosity profiles fitted to one class of station alone are therefore unconstrained, and a profile fitted to both would be tighter than either. | 4.76 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1936 | Threshold | A split in a ground plane stops isolating and starts coupling below a certain width, where the two edges' capacitance dominates the split's inductance and the plane behaves as one. Narrow splits are capacitors; wide ones are chokes. Sweeping the split from 0.2 mm to 5 mm and measuring coupling at 100 MHz finds the width where the trend reverses. | 6.67 |
| H1946 | Count | Mode density in a cavity grows as the cube of frequency, so an enclosure that is well behaved at 300 MHz has roughly a thousand times as many modes to control at 3 GHz (estimate). Mode density, not the box's worst resonance, sets the difficulty of a broadband shield. Counting resonances in a swept measurement of one box at two bands checks the exponent. | 6.58 |
| H1976 | Scale argument | Shielding against electric fields scales with the shield's coverage, and against magnetic fields with its thickness relative to the skin depth, so a foil that is excellent at 1 MHz is poor at 50 Hz and the two specifications cannot be traded against each other. Coverage and thickness are different currencies. Measuring one foil against both kinds of field at both frequencies prices them separately. | 6.56 |
| H1923 | Scale argument | Excess noise in a film resistor scales inversely with the film's volume, so two resistors of identical value in different case sizes have different low-frequency excess noise, and the smaller one is the noisier at every current. Geometry rather than resistivity sets the volume that averages the fluctuations. Two case sizes of one value, measured at the same dissipated power, should differ by the volume ratio. | 6.53 |
| H1916 | Count | A converter with N bits has at most 2^N − 1 independent code transitions, so an integral-nonlinearity curve fitted with more parameters than that is fitting the test's own noise. Beyond that count the fit has no degrees of freedom left to distinguish the parts. Raising the polynomial order on one dataset and watching the residual stop falling at the predicted order shows where the information ends. | 6.51 |
| H1933 | Bound | The rejection a differential front end achieves against a ground loop is bounded by the match of its two source impedances, not by the amplifier's rejection ratio, because mismatch converts common-mode voltage into a differential one ahead of the amplifier. A 0.1% mismatch caps rejection near 55 dB (estimate) regardless of the part. Deliberately trimming one source resistor by 0.1% and measuring the rejection confirms the bound. | 6.51 |
| H1935 | Count | A system with N chassis bonded at M points has at most N + M − 1 independent ground paths, and the number that actually carries current is smaller than the number drawn, which is why schematics overstate the loop count. Paths in parallel share current according to impedance, and the low-impedance ones dominate. Injecting current at each bond in turn and measuring the voltage at every other bond recovers the real count. | 6.38 |
| H1929 | Identity | Popcorn noise and low-frequency excess noise in a bipolar input stage come from the same defect population seen at two time scales, so a part screened for popcorn at room temperature will still show elevated excess noise at another. The defect's emission time constant moves with temperature and swings through the measurement band. Screening a batch for popcorn at 25 °C and measuring excess noise on the survivors at 85 °C tests the link. | 6.36 |
| H1968 | Unit trap | A reference's temperature coefficient is quoted in parts per million per degree, but its effect on a converter is in codes, so the conversion needs the reference's value, the input range and the bit count together; one part per million per degree in a 16-bit system costs about two-thirds of a code over ten degrees (estimate). The same coefficient costs sixteen times as much in a 20-bit system. Measuring one reference in two converters over one temperature sweep shows the code error scaling. | 6.16 |
| H1949 | Identity | What is called electric-field shielding and what is called magnetic-field shielding differ at low frequency only in the shield's bonding and thickness, not in mechanism: a floating foil blocks the electric field and passes the magnetic one, and the same foil bonded at a single point blocks neither well. Eddy currents and charge redistribution are one response viewed through two boundary conditions. Testing a foil floating, bonded at one point, and bonded all round at 50 Hz and at 50 kHz gives the map. | 6.13 |
| H1983 | Misattribution | Drift in a four-wire resistance measurement is often the sense leads' thermal electromotive force rather than the resistor, because the sense pair carries no current and therefore contributes only its junction voltages. Reversing the current reverses the resistor's contribution and leaves the junction's. A current-reversal measurement repeated as the room warms separates them. | 4.82 |
| H1952 | Hidden term | A phase-noise plot includes the analyser's reference, which is not drawn on the same axes, and cross-correlation removes it only when the two internal references are independent. Otherwise the reference's skirt is attributed to the device. Measuring a known-quiet source at two correlation counts shows how much of the floor belongs to the instrument. | 4.8 |
| H1934 | Reversal | Star grounding suppresses low-frequency loop currents and worsens high-frequency behaviour, because the star's branches are inductances and the loop they form is large. Above a few hundred kilohertz a solid plane is the quieter topology. Building the same circuit both ways and measuring at 1 MHz decides which argument applies at the frequency of interest. | 4.78 |
| H1914 | Selection effect | Published effective-number-of-bits figures come from parts that passed a spectral-purity screen, so the survivors' distribution understates the risk of idle tones in a bought batch. Screening removes exactly the units a production line will meet. Testing every unit of a purchased lot with a DC input and a slow staircase counts the tone incidence without the screen. | 4.73 |
| H1985 | Bound | The accuracy of an alternating-current measurement is bounded by the thermal converter's own alternating-to-direct transfer difference, not by the meter's resolution, so extra digits do not buy accuracy at the top of the range. The converter is the standard. Comparing two converters against each other and against a direct-current standard gives the bound without a better reference. | 4.73 |
| H1944 | Threshold | An aperture in a shield stops leaking above a cutoff set by its longest dimension, and below that cutoff the leak falls steeply as the aperture becomes a waveguide beyond cutoff. A hole that ruins a 1 GHz measurement is irrelevant at 100 MHz. Sweeping one aperture's size at a fixed frequency locates the cutoff and shows the slope. | 4.69 |
| H1947 | Scale argument | Shielding effectiveness measured in a small transverse-electromagnetic cell is not the free-space value, because the cell presents a wave impedance that differs from 377 Ω and the coupling to the box depends on that ratio. The ratio is computable and is the correction. Measuring one box in the cell and on an open-range site at the same frequency gives the factor. | 4.64 |
| H1975 | Reversal | Twisting a pair more tightly lowers magnetic pick-up and raises the pair's capacitance, so the tighter pair is worse above the frequency where the capacitance's loss dominates. The two effects scale in opposite directions with the lay length. Two pairs of the same gauge and different lay lengths, measured for both pick-up and attenuation, cross at a frequency set by the lay. | 4.62 |
| H1918 | Identity | Charge injection and sampling-network nonlinearity in a switched-capacitor input are one mechanism seen at two source impedances, since both are the signal-dependent division of the switch's charge between source and hold capacitor. Nothing else in the input path depends on the source that way. Sweeping source resistance at fixed input frequency should collapse the two measured curves onto one when plotted against the impedance ratio. | 4.56 |
| H1922 | Unit trap | Noise figure is defined at a 50 Ω source, so a device whose optimum is 1 kΩ and whose datasheet says 0.5 dB is not 0.5 dB at the impedance actually presented to it. The number carries its reference impedance invisibly. Measuring the same device through a transformer at 50 Ω and at 1 kΩ, and comparing with the datasheet's noise contour, settles the question. | 4.56 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H254 | Bound | Bubble coverage, not catalysis, is what stops an alkaline electrolyser: when the gas fraction at the electrode approaches the percolation threshold for the electrolyte, the ionic path is cut regardless of catalyst quality. Raising pressure shrinks the bubbles and should raise the current at which the cut occurs, which is a test the catalyst cannot pass. | 6.76 |
| H290 | Bound | Ask a battery store for fast frequency response and the answer is set by the inverter, not the cells, because current limit and measurement latency bound the ramp before any electrochemical process does. A step test at two current limits, same cells and same state of charge, should show the ramp changing with the limit and not with the cell. | 6.67 |
| H228 | Scale argument | Below about 10 nm (estimate) the interphase's electronic transport is tunnelling-dominated, so the electronic conductivity fitted from a parabolic growth law is a barrier width in disguise. Films of controlled thickness spanning that scale should show a change from exponential to linear thickness dependence of leakage current. | 5.96 |
| H205 | Threshold | Above an areal capacity of roughly 3 mAh cm⁻² (estimate), a graphite anode plates at 2C however its particles are engineered, because the ionic potential drop through the electrode thickness exceeds the plating margin. Electrolyte conductivity and tortuosity should move the threshold, while a surface coating should not. | 5.93 |
| H291 | Threshold | Below a state of charge estimated near 10% (estimate) for many lithium cells, internal resistance rises steeply because electrolyte is depleted in the electrode's pores. An impedance series across state of charge will show the rise as a change in slope, and the threshold should shift with electrolyte volume. | 5.91 |
| H271 | Misattribution | Porosity is the confounder in nanostructured thermoelectrics: a pore scatters phonons more strongly than a grain boundary of the same spacing, so part of the lattice-conductivity reduction credited to nanostructure belongs to voids. An independent porosity measurement, plus one fully dense nanostructured sample, splits the credit. | 4.98 |
| H294 | Artefact mimics effect | Harmonic distortion measured at a point of common coupling can originate in the voltage sensor's own nonlinearity when the sensor is a resistive divider exposed to the same field. Two sensors of different types, measuring simultaneously, should disagree only if the artefact sits in the instrument. | 4.98 |
| H210 | Inversion of standard practice | Maximizing electrode density is the standard route to volumetric energy, but for cells that must operate below about −20 °C it is the wrong objective, because tortuosity's penalty on ionic transport outweighs the density gain. A density series measured cold and warm will rank the electrodes oppositely. | 4.96 |
| H220 | Artefact mimics effect | Self-discharge reported for a charged cell at open circuit can be the electrometer's leakage acting on the cell's capacitance, which matters most below roughly 1 mAh (estimate). Two electrometers of different input impedance, or one guarded measurement, will give different self-discharge rates if the artefact is present. | 4.96 |
| H275 | Threshold | There is a carrier concentration above which the temperature coefficient of the Seebeck coefficient changes sign, because transport crosses from non-degenerate to degenerate statistics. A doping series measured across a temperature range should locate the sign change at a concentration that shifts with the density-of-states effective mass. | 4.96 |
| H279 | Scale argument | Thermal conductivity measured on nanocrystalline samples is a property of the sample plus its boundaries, so two laboratories using different dimensions will report different values for the same powder. Reporting dimensions with the value, or measuring two thicknesses, turns the ambiguity into data. | 4.96 |
| H236 | Misattribution | A soft short in a lithium-metal cell is frequently a pressure-induced contact through the separator's largest pore, not a filament crossing the bulk. Narrowing the pore-size distribution while holding mean pore size constant should reduce soft shorts, which is the discriminating test. | 4.93 |
| H247 | Selection effect | The highest gravimetric capacitances cluster at the lowest mass loadings, because thin electrodes wet and measure more easily, so the reported record is partly a record of loading. Reporting capacitance at a technical loading, of order 10 mg cm⁻² (estimate), beside the thin-electrode value makes the selection visible. | 4.93 |
| H261 | Selection effect | The best electrolyser catalysts are usually measured on small electrodes where bubbles detach easily, and the ranking can change at technical scale where bubbles are confined. A parallel test at two electrode sizes, same catalyst and same loading, is the minimal check. | 4.93 |
| H202 | Hidden term | The separator is not inert; trace metal, binder, and conductive carbon in it add a residual current that is then charged to the material's coulombic efficiency. A separator-only symmetric cell in the same electrolyte should reproduce that residual, in which case the material's efficiency is a difference of two measurements. | 4.89 |
| H223 | Identity claim | Dead lithium and interphase-encapsulated lithium are one substance counted twice, and a titration for either necessarily finds both. The test is an inventory sum: if the two methods together account for more lithium than the cell held, the categories overlap. | 4.89 |
| H277 | Count | Three instruments measure a thermoelectric: a four-point probe for conductivity, a thermocouple pair for the Seebeck coefficient, a heat-flow meter for thermal conductivity. Every other number in the paper is algebraic in those three, so a reported fourth transport property should be predicted from the first three measured on the same sample. | 4.84 |
| H242 | Reversal | Thickness trades foil for active material, so energy per device rises while power falls and the product peaks somewhere between. Two thicknesses measured for both quantities bracket the peak, whose position should move with electrolyte conductivity. | 4.8 |
| H293 | Identity claim | Virtual impedance, virtual inertia, and the voltage-loop gain of a grid-forming inverter are not three independent knobs but one two-dimensional object reparameterized, which is why tuning one disturbs another. A transfer-function measurement at three settings that lie on a line in parameter space confirms the degeneracy; settings that scatter off it would refute it. | 4.8 |
| H264 | Conserved-quantity accounting | Water balance in a proton-exchange electrolyser closes only when back-diffusion from cathode to anode is counted, and claims of dry operation usually omit it. A cell with a thinner membrane, where back-diffusion is larger, should tolerate lower anode water stoichiometry if the balance is right. | 4.0 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2594 | Null claim | Two identical co-located sensors should differ only by their calibrations; if the difference is larger and structured, the cause is their positions within the shelter, and swapping the positions should reverse its sign. The null is a siting test that costs one swap. Applying it to a pair before trusting either as a reference is the cheapest check in a network. | 6.64 |
| H2583 | Null claim | At a site with steady flow and chemistry, moving an in-situ sensor a few centimetres within the water column should not change its reading, and any change is a siting artefact that a swap of two identical sensors resolves. If the reading follows the position rather than the sensor, the column is not well mixed at that scale. The test costs one dive. | 6.6 |
| H2517 | Identity | Breakthrough volume and retention volume of a sorbent tube are the same quantity approached from opposite ends, so a method that measures breakthrough directly and quotes a safe sampling volume at half of it is using one number, not two. The half is a safety factor, not a second measurement. Running a tube in reverse after a frontal run should release the analyte over the same volume at which it broke through. | 6.4 |
| H2566 | Count | Size bins earn their keep only when they separate modes: subdividing one mode adds no independent information and multiplies counting noise per bin, so more bins with fewer counts can make a unimodal distribution look multimodal. Rebinning one dataset at half the resolution and testing whether the modes survive shows which features were real. | 4.8 |
| H2515 | Bound | A denuder's capacity is bounded by its coated area times the monolayer coverage of the reagent, and a coating thicker than a few tens of micrometres adds capacity but also adds a particle loss that rises with thickness. The optimum maximizes capacity minus loss, not capacity. Denuders coated at several thicknesses, tested for both breakthrough and fine-particle transmission, locate it. | 4.67 |
| H2527 | Bound | A peristaltic sampler's contamination floor is bounded by the tubing's leaching rate per unit wetted area, not by the analytical method's sensitivity, and the bound scales with the tube's residence time. A longer tube at the same flow leaches more. Pumping ultrapure water through tubes of two lengths and two materials, and measuring the plateau, yields the bound without a field sample. | 4.62 |
| H2588 | Scale argument | Footprints scale with wind speed times averaging time, so a one-minute average and a one-hour average at one site describe different areas, and comparing either with a model grid of fixed size compares two different things. The mismatch is largest on windy days. Computing the footprint for both averaging times from the site's wind record and testing which predicts a co-located reference sizes the mismatch. | 4.56 |
| H2541 | Scale argument | Gas in a cylinder stratifies by molecular weight, and the relevant scale is the cylinder's height, so a tall cylinder's first draws differ from a short one's at the same nominal mixture. Stratification relaxes by diffusion, slowly compared with the interval between draws. Drawing from the top and the bottom of a tall cylinder, or inverting it for a day, tests homogeneity on the height scale. | 4.53 |
| H2508 | Identity | Inlet efficiency and line loss are the same deposition measurement made at two positions, so a study that measures both and adds them into a transmission budget charges the wall twice. The wall does not know where the inlet ends. A single deposition tracer injected at the inlet mouth and counted at the analyser, compared with the sum of the two reported terms, exposes the double count. | 4.51 |
| H2567 | Identity | PM2.5 from an optical sensor and PM2.5 from a filter are not the same quantity: one is a calibrated proxy for mass derived from scattering, the other is a mass, and their agreement at one site is a property of the calibration and the aerosol there. Applying both at a second site with a different size distribution and refractive index, without refitting, tests whether the agreement travels. | 4.51 |
| H2590 | Count | A count of hours above a level depends on the sampling interval, because a five-minute sampler can miss a two-minute plume that a one-minute sampler catches, so the count is partly an instrument's property. The missing hours are not random; they are the shortest events. Running two samplers at one site at different intervals through a season and comparing counts, and the durations each missed, quantifies the dependence. | 4.49 |
| H2536 | Conservation argument | Alkalinity measured on a stored sample closes only if the bottle's headspace carbon dioxide is accounted for, because the headspace is a reservoir that exchanges with the water and shifts the titration's endpoint. A bottle with a large headspace can lose or gain measurable alkalinity with no biological activity at all. Titrating samples bottled at two headspace fractions, with total inorganic carbon measured on the same bottles, tests the closure. | 4.44 |
| H2512 | No-go claim | No passive inlet can simultaneously preserve coarse particles and semi-volatile vapours, because the measures that keep particles airborne — short residence time, warm walls, no impaction — are the ones that let vapours pass or evaporate from the deposit. The two requirements pull the same variables in opposite directions. A single inlet tested for both coarse mass and semi-volatile recovery across temperatures and flows should trace a frontier rather than show an optimum. | 4.38 |
| H2553 | Reversal | Raising sample flow through a photoionization detector is supposed to speed the response by shortening residence time, but above a flow the dilution inside the lamp housing lowers the concentration at the electrode, so the steady amplitude falls while the time constant has already stopped improving. The two effects have different flow exponents. Stepping concentration at several flows and measuring both the time constant and the steady amplitude puts the reversal in view. | 4.38 |
| H2556 | Count | Manual interference lists overstate the correction burden, because the species that matter at a site are the ones that co-vary with the target there, and that is usually two: one tracking the target's source and one tracking its sink. Listings enumerate everything the sensor responds to, not everything that matters. Regressing the sensor against a suite of measured co-pollutants at one site and counting coefficients that survive out-of-sample prediction gives the site's number. | 4.38 |
| H2563 | Bound | The smallest particle an optical counter can size at a given wavelength is bounded by stray light in the collecting optics rather than by diffraction, so quoting the diffraction limit overstates the instrument's reach. Stray light sets a background that a small particle's scattering cannot exceed. Measuring the pulse-height distribution for a filtered airstream and for a monodisperse aerosol at the quoted smallest size shows where the background sits relative to the signal. | 4.38 |
| H2560 | Null claim | For a diffusion-limited electrochemical cell the temperature coefficient of sensitivity should be small, because the diffusion coefficient rises with temperature while solubility falls and the two partly cancel; a large measured coefficient therefore indicates a blocked diffusion path or a dry electrolyte, not a temperature effect. The null is a diagnostic. Measuring the coefficient before and after deliberately partial blockage of the membrane shows the sign and size of what a blocked path does. | 4.36 |
| H2528 | Reversal | Low-flow purging is meant to reduce turbidity and preserve in-situ redox, but below a certain rate the pump's suction lowers pressure enough to degas carbon dioxide, raising pH and shifting the redox state the method was trying to preserve. The effect grows as flow falls, opposite to the assumption that slower is gentler. Sampling at several pumping rates while measuring pH and dissolved oxygen in a flow-through cell, before any air contact, shows the reversal. | 4.33 |
| H2531 | Count | Purging to a fixed count of well volumes assumes the count stands for a residence time, and in a low-yield well the proxy fails because the water reaching the pump has had time to equilibrate with the casing and the atmosphere. Time, not volume, is the variable that matters there. Purging one low-yield well to a fixed volume and to a fixed time, and measuring a conservative parameter in both, shows which reaches formation water first. | 4.31 |
| H2529 | Scale argument | Purge volume is usually expressed in well volumes, which hides that the volume to be removed scales with the square of the well radius while the time to remove it scales inversely with pumping rate. A wide well and a narrow one therefore need the same number of well volumes and different times. Purging two wells of different diameter at equal flow, with a tracer at the screen, tests which quantity tracks the arrival of formation water. | 4.29 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5314 | Electrode-loss artefact | Minimum spark ignition energy is read as a chemical quantity, but the kernel loses heat to the electrodes during its first instants of growth. The measured minimum should then scale with the electrode's thermal conductivity at fixed gap, and swapping electrodes without changing the mixture should move the minimum in a predictable direction. | 6.82 |
| H5394 | Area-scaled re-ignition | The chance that a suppressed fire re-ignites is tied to how much fuel remains. It should scale with the exposed hot surface area instead, so a thin spread-out residue should re-ignite more readily than a compact one of the same mass, and tests on consolidated samples should not predict the behaviour of distributed ones. | 6.8 |
| H5305 | Local-versus-mean control | A fuel carrying a distributed inert is characterized by its mean composition, yet ignition begins at the hottest particle-scale site. The critical flux for such a mixture should depend on the width of its particle-size distribution at fixed mean size, so two samples with identical averages but different widths should differ by more than run-to-run scatter. | 6.76 |
| H5388 | Mixing-time bound | The time to extinguish a compartment fire is bounded below by the time needed to mix the agent with the upper layer. A less effective agent that mixes faster can therefore extinguish sooner than a more effective one, and agent rankings from enclosure tests should depend on the discharge's momentum as much as on the agent's chemistry. | 6.73 |
| H5317 | Preheat-length dilution | In concurrent spread the flame preheats the fuel over a length set by its own boundary layer. Part of that heat is conducted ahead of the front and lost into the bulk, so a fuel of higher thermal conductivity should spread more slowly at equal volatility, with the penalty scaling as the ratio of conductivities. A conductivity series at fixed chemistry would test this without moving the pyrolysis temperature at all. | 6.42 |
| H5333 | Momentum-versus-buoyancy regime | Under a ceiling the flame is fed by a horizontal jet whose momentum, not the plume's buoyancy, sets the inflow. Spread should then scale with a Froude number built on the jet velocity rather than with the fuel's orientation, and the same fuel should spread faster on a ceiling than on a vertical wall at equal incident flux. Repeating the ceiling test at two ceiling heights would separate the jet's momentum from the plume's buoyancy. | 6.33 |
| H5347 | Optical-thickness bound | The fraction of a flame's heat released as radiation is taken as a fixed value for a given fuel. That fraction is bounded by the flame's optical thickness, so it should rise with flame size while the flame is optically thin and saturate once it is thick, and any tabulated constant is then a coincidence of the sizes that were measured. Two pan diameters would test the claim without any knowledge of the soot yield. | 6.33 |
| H5350 | Sub-thermal chemical marker | Alkali metals suppress flames at doses that also depress temperature. At lower doses they quench soot luminosity without measurably cooling the flame, so a change in visible emission should precede any thermocouple response and can serve as a chemical marker of inhibition at sub-thermal dose. | 5.0 |
| H5344 | Residue property change | The black residue behind a smolder front is treated as inert char. It is partly oxidized material of higher surface area and lower ignition temperature than the virgin fuel, so a re-ignition test on the residue measures a different substance and cannot return the original fuel's threshold. | 4.96 |
| H5312 | First-run bias | A freshly prepared surface carries adsorbed water and a labile fraction that are driven off on first heating. The ignition temperature measured on the first run should therefore be higher than on later runs, which makes that first number a statement about the sample's storage history rather than about the material. | 4.87 |
| H5359 | Aperture-dependent coefficient | Transmissometers report a mass-specific extinction coefficient as a property of the smoke. Forward-scattered light enters a wide-aperture receiver, so the coefficient measured with a wide aperture should be lower than with a narrow one by an amount computable from the soot's size parameter, which makes the coefficient partly a property of the instrument. | 4.82 |
| H5374 | Missing geometric variable | Switching between fuel-controlled and ventilation-controlled burning is indexed on the ventilation factor alone. The plume's residence time also depends on the compartment's height, so two compartments with equal ventilation factors but different heights should fall on different sides of the transition. Fixing the ventilation factor while changing only the height is a test no existing correlation would pass. | 4.71 |
| H5372 | Steady-state selection | A fire travelling along a long corridor should settle at a speed set by the ratio of heat release per unit length to heat loss per unit length, without reference to spread kinetics. The transient before that speed is reached is where spread physics still matters, so corridor tests short enough to remain in the transient measure a different quantity. | 4.67 |
| H5323 | Lateral-loss width limit | A spreading flame loses heat sideways to the sample's edges as well as downward into the fuel. Below a critical width the lateral loss should exceed the flame's heat production and opposed spread should stop entirely, which would make width a hidden variable in every narrow-sample measurement. A width series at fixed thickness should show spread failing abruptly rather than fading gradually. | 4.62 |
| H5321 | Confinement enhancement | A flame spreading along a narrow channel entrains less than one on an open surface, so its preheat zone is longer and its spread faster. The enhancement should peak at a gap of a few flame thicknesses, an estimate rather than a measured optimum, and decay on both sides, which would make channel width a control variable as strong as the fuel's orientation. | 4.6 |
| H5395 | Path-integral misreading | Obscuration measured across a compartment is a line integral through a stratified layer. A beam at the layer's midpoint reads a value between the layer's concentration and the lower zone's, so reported obscuration should change with beam height at constant smoke production, and a single beam cannot define a concentration. The vertical profile of obscuration, rather than one reading, is therefore the quantity comparable between compartments. | 4.53 |
| H5325 | Radiation-limit crossover | Where buoyancy is absent, opposed spread is limited by radiative loss from the flame rather than by conductive gain to the fuel. The crossover between the two limits should sit at a fuel thickness where loss and gain balance, which makes thickness, not gravity, the parameter deciding whether a microgravity spread test is representative. Varying thickness at fixed pressure in a drop tower would locate the crossover directly. | 4.49 |
| H5331 | Superposed fronts | A thick fuel can carry a fast surface flame and a slow in-depth front at the same time. A reported spread rate is then a weighted mean of the two, and its value should shift with the sensitivity of the instrument used, which makes cross-study comparison of the spread rate meaningless for thick fuels. Separating the two fronts needs a depth-resolved measurement, which a surface camera cannot supply. | 4.49 |
| H5329 | Chamber-flow aliasing | Spread experiments run in still chambers whose residual buoyant circulation is never zero. The scatter between repeats should then be a function of the chamber's aspect ratio rather than of the material, and two laboratories with different chamber heights should report systematically different rates for the same sample. Rotating the chamber's aspect ratio about a fixed sample should change the measured rate, which is a cheap check on any material difference. | 4.42 |
| H5381 | Stokes-number optimum | The optimal droplet size for a water mist is usually chosen from heat-transfer arguments. Droplets below a critical Stokes number follow the gas into the reaction zone while larger ones cross it ballistically, so the optimum should shift with the fire's strain rate, and a mist tuned to a small fire should underperform on a large one. One nozzle tested at two fire sizes should show the predicted loss of performance. | 4.42 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3648 | Unit trap | Flow coefficient is quoted as Cv in gallons per minute at one pound per square inch in one tradition and as Kv in cubic metres per hour at one bar in another, and the two differ by a factor near 0.86, close enough to unity to be mistaken for rounding. A valve specified in one and checked in the other will pass a capacity test it should fail, or fail one it should pass. | 6.36 |
| H3641 | Count | The flow-force characteristic of a spool at fixed geometry is shaped by at most two independent parameters, an entrance angle and a discharge angle. A computational study reporting five sensitive geometric variables is reporting combinations of the two. Two spools built to vary the angles independently would test the count. | 4.89 |
| H3691 | Hidden variable | Cleanliness differences between two machines of one design are usually charged to their filters, but the breather's location relative to the reservoir's air space and to the plant's dust sources accounts for more of the difference than the filter's rating. Swapping breathers between two machines, without touching the filters, would test it. | 4.84 |
| H3675 | Conservation audit | A circuit's energy balance closes only if the relief valve's dissipation is counted. That dissipation is invisible in pressure and flow logs taken at the actuators, so it is usually omitted and the balance is closed by inflating the actuator's efficiency. A calorimetric measurement of the reservoir's heat rejection recovers the missing term. | 4.82 |
| H3681 | Misattribution | Filter efficiency measured by counting particles upstream and downstream is charged to the medium, but at high concentration the downstream counter undercounts because two particles cross the sensing volume together, and the undercount reads as efficiency. Diluting the downstream sample, or counting at two concentrations and extrapolating to zero, separates the medium from the counter. | 4.78 |
| H3613 | Conservation audit | A calorimetric efficiency can exceed a torque-derived one because the temperature rise is measured across a casing that loses heat to the room, and the missing heat is counted as work delivered. The discrepancy grows as the power falls, so a small pump tested calorimetrically looks best exactly where the loss is largest relative to the signal. | 4.73 |
| H3669 | Count | The behaviour of a bladder accumulator in a circuit needs at most two independent gas parameters, the precharge and the polytropic exponent, so a model with separate bladder stiffness, gas solubility and heat-transfer coefficients is fitting a curve that two numbers already describe. A discharge test at two rates identifies both, and a third rate should then be predicted rather than fitted. | 4.62 |
| H3680 | Selection effect | Reported water-hammer events are the ones that broke something, so the literature's distribution of surge pressures is truncated at the damage threshold and its mean is biased high, while events measured and found harmless are rarely written up. A campaign that records every surge, damaging or not, would move the distribution's body and leave the damaging cases where the literature put them. | 4.62 |
| H3662 | Unobservable | A lip seal's leakage is intermittent at the scale of a shaft revolution, so an average leak rate cannot be inverted to a film thickness: the same average can come from a thin film everywhere or a thick film for part of the cycle. Only a time-resolved measurement, or a tracer with a fast response, can distinguish the two, and flowmeters integrate both. | 4.6 |
| H3602 | Hidden term | A pump's suction margin measured on a pressurised test stand includes a term from the gas the stand's blanket dissolves into the oil, because that gas leaves solution in the eye and occupies volume the impeller cannot pump. Degassing the same loop should raise the measured margin with no change to the pump, so a bench figure and a field figure are not the same quantity. | 4.53 |
| H3659 | Identity | Extrusion damage and nibbling of an elastomer seal are one failure seen at two gap sizes. At a large gap the material leaves in a slug and at a small gap in fragments, but the stress state at the edge is the same. A test at three gap widths on one compound should show a continuous damage rate across the two appearances rather than a step. | 4.53 |
| H3603 | Bound | The head a volute can accept at its best point is bounded by the diffusion the volute tolerates, not by the impeller's exit angle, so opening the exit angle past that bound buys head at the design flow and gives it back on both sides. One impeller tested in two volutes of different throat area would show the ceiling moving with the volute and not with the blade. | 4.49 |
| H3685 | Threshold | A hard particle causes wear only when its size exceeds the film thickness at the contact, and below that size the count matters little however high it is, because the particle passes without touching both surfaces. That is why cleanliness codes correlate with life in one machine and not in another, and the correlation should be plotted against the code divided by the machine's film thickness rather than against the code. | 3.42 |
| H3605 | Threshold | Suction recirculation sets in at a flow coefficient that is almost independent of rotative speed but falls steeply as blade number rises. The threshold is a balance between the blade's incidence and the diffusion the vaneless space will tolerate, so it belongs to the geometry rather than to the duty. One impeller run at three speeds and with two blade counts should collapse the inception points onto the blade-number curve and not onto a speed curve. | 3.31 |
| H3631 | Degeneracy | Two pumps of identical displacement and pressure rating can differ in flow ripple by a factor of three (estimate) because their porting plates differ, and no line on the datasheet distinguishes them. A ripple measurement at two speeds and one pressure separates the designs, and it is almost never specified. | 2.89 |
| H3635 | Misattribution | Relief-valve chatter is diagnosed as spring surge and cured with a stiffer spring, yet the frequency usually matches a quarter-wave mode of the downstream line rather than the spring's own. Changing the line's length while holding the spring fixed moves the chatter, and that is the test that distinguishes them. | 2.89 |
| H3646 | Time-scale mismatch | A servo valve's torque motor responds faster than the hydraulic stage it drives, and the difference appears as a phase lag that is then attributed to the actuator the valve is driving. A pressure-step test with the actuator disconnected isolates the valve's own lag, which is usually smaller than the system's. | 2.87 |
| H3649 | Degeneracy | Two spools with different land widths and the same nominal flow gain can differ in null leakage by an order of magnitude, so a datasheet that lists gain and leakage as separate guarantees is describing a family and not a part. Only a null-leakage measurement at the specified gain distinguishes them. | 2.87 |
| H3647 | Conservation audit | Valve leakage measured by tank-level rise over an hour includes the reservoir's thermal expansion, which over a few degrees in that time is comparable to the leakage being measured. A level sensor on a cooling tank reports a leak that is not there, and the error carries the sign of the temperature change. | 2.82 |
| H3615 | Degeneracy | Two impellers of different blade angle can be trimmed to give the same head-flow curve while differing in axial thrust by a factor of two (estimate), so an acceptance test on performance alone cannot certify the bearing's duty. A thrust measurement on both impellers at one flow would show the degeneracy that the curve hides. | 2.78 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1023 | Degeneracy | Strong acceleration and strong convex curvature both relaminarize a boundary layer by thinning the stress-bearing layer, so a flow with both should be no more stable than the more stabilizing of the two. A curved accelerating channel with independent control of acceleration and curvature would test that sub-additivity. | 6.93 |
| H1002 | Hidden term | Temperature-dependent density puts a term into the shape factor of a heated boundary layer that is proportional to the wall heat flux and is normally absorbed into the pressure-gradient parameter. Two isothermal walls held at different temperatures, matched in momentum-thickness Reynolds number and in Clauser parameter, should then give shape factors differing by an amount that tracks the temperature ratio rather than the Reynolds number. The term is computable from the mean density profile, so the correction needs no new measurement. | 6.2 |
| H1087 | Misattribution | Shear thickening in a dense suspension is attributed to particle contacts, though dilatant migration of particles away from the shearing surface may account for the remainder by raising concentration in the bulk. Concentration measured across a Couette gap changes at the onset of thickening. | 4.98 |
| H1022 | Conservation audit | Turbulent kinetic energy budgets measured with hot wires fail to close because the wire rectifies the fluctuating velocity, and the rectification is largest where the intensity is largest. Integrating the budget twice, once with each instrument, localizes the missing term and shows whether it accounts for the residual. | 4.93 |
| H1032 | Inversion of standard practice | Screens are installed upstream to homogenize a test section, yet a screen also imposes its mesh as a preferred length that can lock the downstream turbulence to it. Swapping the screen for one of different mesh, everything else fixed, moves the integral scale while leaving the mean profile alone. | 4.91 |
| H1034 | Unit trap | Mixing efficiency is a ratio of potential to kinetic energy and is dimensionless only when both are integrated over the same control volume. Recomputing reported efficiencies with a common volume, or measuring both fluxes on one grid, should remove the occasional values that appear to exceed the theoretical ceiling. | 4.91 |
| H1048 | Scale argument | Coalescence in a turbulently stirred emulsion is governed by the ratio of film drainage time to the eddy turnaround time at the drop scale, so rates collapse on that ratio rather than on the Weber number. Stirring a cell at variable viscosity ratio tests the collapse over a range of drop sizes. | 4.84 |
| H1008 | Symmetry | The log law is fitted with one pair of constants for both walls of a channel, but unequal wall temperatures break the wall-normal symmetry of the mean profile and make the constants wall-specific. Asymmetric heating at matched friction Reynolds number is the clean test: the centerline should kink, and the kink must vanish when the two wall temperatures are equal. | 4.67 |
| H1099 | Bound | No Reynolds-stress closure can be both local and exact, because the stress at a point depends on the pressure field, which is nonlocal, so every local model carries an error bounded below by the pressure-strain integral scale. A local model against direct simulation at two Reynolds numbers shows the error failing to fall with resolution. That floor would appear as a resolution-independent bias in every a-posteriori test. | 4.51 |
| H1004 | Threshold | A roughness element alters drag only once it reaches far enough into the log region to displace the virtual origin, so the controlling ratio is element height to boundary-layer thickness and not height to viscous length. Building the same height-to-thickness ratio at two Reynolds numbers in different facilities tests it directly, since the roughness function must then agree even where the viscous-scaled height differs by a factor of three (an estimate). | 4.42 |
| H1010 | Degeneracy | Riblets and streamwise travelling waves are thought to act by different routes, but at low amplitude both merely displace the virtual origin, so a surface carrying both should perform no better than the better one alone. Running each element on and off in one channel exposes sub-additivity if the routes are degenerate. The same test at high amplitude, where the wave forcing turns nonlinear, should show them ceasing to be degenerate. | 4.42 |
| H1024 | Unobservable | Dissipation is not measurable by any single probe, because the smallest scales are smaller than the wire and what is reported is a filtered dissipation whose filter depends on wire diameter. Three matched wires of different diameter, in one flow, return three estimates differing by a factor set by the filter rather than by the flow. That would make quoted dissipation values instrument-specific at the level of tens of percent (an estimate). | 4.4 |
| H1069 | Unit trap | Permeability is reported in darcy, a unit of area, so a tight core is not a different kind of rock but the same rock with smaller pores, and comparisons that treat permeability as a material class conflate geometry with scale. Cores sintered from different grain sizes to the same permeability differ in Forchheimer coefficient if geometry is being read as scale. The pair is easy to make, and the test would show whether the darcy is being read as a material label. | 4.18 |
| H1077 | Identity | The coffee-ring deposit and the Marangoni deposit are the same edge-directed capillary flow driven by two different gradients, so reversing the surfactant gradient should convert one into the other without changing the flow topology. Particle tracking in a drop with a controlled surfactant source catches the flow reversing while the deposit pattern stays recognizably similar. | 3.78 |
| H1012 | Unit trap | Drag reduction is quoted per part per million of polymer, but the controlling variable is concentration times molecular weight raised to a power near four-fifths (an estimate), so two polymers ranked per ppm can reverse order when ranked at equal overlap concentration. Two molecular weights compared at matched ppm and again at matched c M^0.8 demonstrate the reversal in a single sitting of rheometry. | 3.58 |
| H1064 | Hidden term | The effective stress law for a soft poroelastic solid carries a term from the solid's own compressibility that is absorbed into the Biot coefficient and then varies with strain. Drained and undrained tests across a stress range, on one sample, reveal the coefficient drifting. | 2.98 |
| H1095 | Conservation audit | Solids flux in a horizontal slurry pipe is not the product of mean concentration and mean velocity, because the two profiles are correlated and the covariance can be a substantial fraction of the total. Both profiles measured at once expose the missing covariance term. | 2.98 |
| H1051 | Count | A monodisperse bubbly flow at low void fraction needs three independent closure quantities, void fraction, slip velocity and interfacial area, so a model adding a fourth without a new measurement is unidentifiable. A parameter-identifiability analysis on one dataset shows the fourth trading off against the other three. | 2.96 |
| H1063 | Identity | Early tracer breakthrough and long tailing in a column are two ends of one residence-time distribution shaped by dead-end pores, so a model fitting only one describes half the flow. A column with its dead-end fraction measured by imaging would let the same distribution predict both ends. | 2.96 |
| H1094 | Count | A dilute suspension has at most two independent viscometric functions, viscosity and first normal stress difference, so a third reported at the same shear rate is a combination of the two. One suspension measured in cone-and-plate and in Couette exposes the third quantity as a derived combination. | 2.96 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1800 | Scale argument | A single-kernel drop test at high velocity cannot predict breakage in a bucket elevator, because the elevator applies many low-velocity impacts whose damage accumulates subcritically, and the scaling variable is the number of cycles rather than the velocity. A rig reproducing the elevator's impact count at its velocity should predict the plant's breakage. | 8.82 |
| H1766 | Hidden term | Measured oxygen transmission through a film includes the boundary layers on both sides, so a high-permeability film reads below its true value and the bias depends on the carrier gas flow. Raising the flow at fixed film should move the reading until the boundary layer is swept away, and the plateau is the film's own value. | 8.76 |
| H1784 | Unit trap | Relative humidity describes air and water activity describes the product; the two are numerically equal only at equilibrium, and using one where the other belongs reverses the predicted direction of moisture transfer. A storage experiment measuring both should show the reversal exactly where they diverge. | 7.0 |
| H1758 | Reversal | Raising impeller speed in wet granulation can reduce homogeneity above a Froude number, because the regime shifts from distributive mixing to intensive densification and granules grow faster than they are dispersed. A speed series with granule size measured at fixed liquid content should show the reversal. | 6.98 |
| H1774 | Derived quantity | Shelf life derived from a package is not a physical constant: it contains an acceptability limit that is a specification, so two shelf lives measured against different limits are not comparable. Recomputing one dataset against the other's limit should either reconcile the numbers or expose the difference as definitional. | 6.91 |
| H1785 | Misattribution | Apparent yield stress of a food gel from a vane measurement is credited to the network, but the vane's end effects add a term scaling with its aspect ratio. Vanes of two aspect ratios in the same gel should give two yield stresses that converge once the end correction is applied. | 6.87 |
| H1753 | Misattribution | Mixing time measured by a conductivity probe is credited to the impeller's pumping capacity, but a substantial part of the recorded time is the probe's own response and its position in the flow. Two probes at mirrored positions, and one with a faster response, should return different mixing times for the same batch. | 6.82 |
| H1742 | Reversal | Lengthening a holding tube raises the mean residence time but narrows the distribution, so the coldest element's thermal load can rise while the hottest element's falls; more holding is not uniformly more cook. Residence-time distributions measured at two tube lengths with the same tracer should show the two ends of the distribution moving in opposite directions. | 6.76 |
| H1716 | Turning point | The surface temperature of a drying solid rises sharply at the moment the surface stops being wet, and that turning point locates the transition with no mass measurement at all. The turn should coincide with the end of the constant-rate period in a companion gravimetric run; if it does not, the thermocouple is reading the boundary layer rather than the surface. | 6.64 |
| H1721 | Threshold | Supercooling before ice nucleation in a food matrix is set by the largest pore of the most active foreign particle present, not by the matrix's water activity, so the nucleation temperature should be insensitive to solute concentration above a certain level. Filtering the mix through graded membranes and measuring nucleation temperature should track the filter cut-off rather than the solids content. | 6.64 |
| H1768 | Bound | Achievable carbon dioxide in a package is bounded by the film's selectivity and the product's respiratory quotient, so no film can push the level past a computable ceiling at a given oxygen level. Films of graded selectivity should approach that ceiling without crossing it. | 5.98 |
| H1769 | Threshold | Panel collapse of a sealed package happens below a pressure deficit set by the laminate's bending stiffness and the headspace volume, which makes the threshold a structural quantity rather than a gas quantity. Packages of two film gauges and two headspace volumes should collapse at deficits predicted by plate theory. | 5.89 |
| H1756 | Bound | Powder mixing cannot homogenise below the scale of the largest cohesive cluster in the feed, so the achievable coefficient of variation is bounded by the feed's cluster size distribution rather than by the machine. Sieving the feed to remove clusters should lower the achievable variation with no change to the mixer. | 5.87 |
| H1707 | Artefact mimics effect | The long tail of a drying curve, read as a slowly relaxing bound-water fraction, can be the chamber's own headspace coming to equilibrium with the sample. Shrinking the headspace tenfold should shorten the tail tenfold while leaving the early curve unchanged, and if it does the tail was never in the sample. | 5.84 |
| H1795 | Bound | The filling rate of a silo without segregation is bounded by the percolation velocity of the fine fraction through the flowing coarse matrix rather than by the conveyor's capacity, and above that rate the fines are trapped where they land. Filling at two rates with tracers in the fine fraction should show the transition. | 5.8 |
| H1765 | Misattribution | Rising headspace oxygen in a package is usually charged to film permeability, but a seal leak produces the same rise and scales with seal length rather than film area. A pressure-decay test plus packages of two seal lengths at equal area separates the two contributions. | 5.0 |
| H1759 | Artefact mimics effect | Emulsion stability scored by turbidity depends on the dilution used for the reading, since dilution leaves time for coalescence before the measurement is complete. Measuring one emulsion at two dilution ratios and extrapolating to zero dilution separates the sample's stability from the measurement's. | 4.96 |
| H1778 | Hidden term | Crumb firmness used as a staling measure contains a moisture-redistribution term in addition to starch retrogradation, so a firm loaf is not necessarily a retrograded one. Hermetic and permeable storage at one temperature should diverge in firmness while retrogradation measured calorimetrically stays equal. | 4.96 |
| H1743 | Count | A tracer experiment resolves only as many residence-time moments as its sampling interval permits, and higher moments fitted beyond that are interpolation rather than measurement. Resampling the same record at half the rate should preserve the claimed moments if they are real and destroy them if they are not. | 4.91 |
| H1783 | Scale argument | The thermal time constant of a grain bulk scales with the square of its dimension, so a two-hundred-gram laboratory sample equilibrates in days while a two-hundred-tonne silo takes months (estimate). A trial that does not run for the larger time constant cannot observe the phenomenon it claims to test. | 4.91 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5455 | Plateau after a regime change | Penetration of a non-deforming bullet in gel should rise with impact energy at fixed presented area, but a bullet that upsets on impact should show depth independent of velocity above the upsetting threshold. The transition should coincide with the onset of visible deformation. Firing a graded velocity series and measuring both depth and recovered diameter locates the transition. | 6.71 |
| H5478 | Inversion of a sequence | Aromatic fractions in porous substrates should persist longer than alkanes, inverting the usual weathering order, because aromatics adsorb more strongly to polar sites and are therefore less available to evaporate. A residue that looks heavily weathered by its alkane loss can then be comparatively fresh. Weathered residues in several substrates, quantified against a deuterated standard, test the inversion. | 6.71 |
| H5492 | Orientation dependence | Cooling in water should be faster with a body's long axis vertical than horizontal, because the convective boundary layer is thinner at the leading face and the plume rises along the body. The difference should be large enough to matter over the first hours. Immersing body analogues in a tank at both orientations, with thermocouple arrays, tests the orientation dependence. | 6.67 |
| H5422 | Processing history recorded in geometry | Twist-reversal spacing in cotton should record the tension at which the fibre dried, because a reversal forms where the collapsing lumen buckles under drying stress. Two fabrics woven from the same yarn count can then differ in spacing without differing in variety. Measuring spacing along many fibres from each fabric, with drying tension varied in a control batch, tests the record. | 6.64 |
| H5431 | State at the moment of contact | Scale impressions pressed into a soft surface should depend on the hair's hydration at the moment of contact, because cuticle scales are softest when swollen, so two contacts of the same head of hair at different humidities should give different impression depths. Pressing one hair into a polymer at controlled humidities and profiling the impressions tests whether depth records the environment. | 6.64 |
| H5493 | Attenuation by the preparation | Carbon coating applied for charge control should attenuate low-energy X-ray lines more than high-energy ones, because attenuation scales steeply with wavelength and the coating's thickness varies with the coater's geometry. Two laboratories coating the same particles to different thicknesses can then report different compositions. Measuring one particle at several coating thicknesses, with thickness read from a quartz monitor, tests the attenuation. | 6.64 |
| H5459 | Direction reversal from hardness | Striations on a soft workpiece should sometimes be the harder tool's own surface features transferred in positive relief, because the harder body indents the softer one along the sliding direction. The mark's asymmetry then reverses compared with the usual negative reading. Sliding hardened and softened tools of identical shape across lead and brass at matched loads, and imaging striation cross-sections, tests the reversal. | 6.6 |
| H5479 | Inversion problem with a unique solution | Thickness profiles of soot on a cold surface should follow boundary-layer growth along that surface, so the profile can be inverted to give the smoke's duration and the surface's temperature. The inversion should be unique because deposition velocity is monotone in surface temperature. Depositing soot on cooled plates at graded temperatures in a controlled smoke stream and comparing profile shapes tests the inversion. | 6.6 |
| H5438 | Time-dependent count | Fragment count from tempered glass should keep rising after the break, because stored strain energy continues to drive branching for a short interval. A count made immediately and a count made after handling will therefore differ, and the difference belongs to the measurement's delay rather than to the glass. High-speed imaging of the fragmentation, with counts extracted frame by frame, gives the count's time dependence. | 6.56 |
| H5481 | Boundary condition assigned to the wrong surface | Early post-mortem cooling should be dominated by conduction into the supporting surface rather than by convection to the air, because the contact's thermal resistance is much lower than the air's. The early slope should then change with the surface material and be steeper on tile than on carpet. Cooling body analogues on surfaces of graded conductivity, with core and surface thermocouples, tests the boundary assignment. | 6.56 |
| H5473 | Alternate cause for a pattern | Burn patterns on a floor should sometimes be the joint layout controlling airflow rather than the liquid's spread, because the liquid burns where vapour mixes with air and the joints feed air into the boundary layer. A poured line crossing a joint-free panel should then burn differently from one along a joint. Pouring identical quantities across jointed and joint-free panels and mapping burn depth tests the alternative cause. | 6.49 |
| H5454 | Saturation of a deformation | Final diameter of an expanded bullet should be bounded by the jacket's mouth thickness rather than by impact velocity above the velocity at which the petals fold back, because the petals can open only as far as the mouth allows, so a velocity series should show diameter rising and then plateauing. Recovering bullets fired into a recoverable medium at graded velocities, and measuring each mouth thickness, tests the bound. | 6.47 |
| H5413 | Geometric similarity argument | Deposit profiles along a vehicle's underbody should collapse onto one curve when plotted against the length of the recirculation zone behind the front axle, rather than against distance from the bumper. Two vehicles of different size driven at different speeds should then show the same profile in that length scale. Measuring the recirculation length by tuft or smoke visualisation and rescaling the deposit profile tests whether one geometric variable organises the data. | 6.4 |
| H5476 | Material-dependent geometric signature | Apex angles of V patterns on walls should depend on the wall's thermal effusivity as well as on the flame's standoff, because soot deposition requires the surface to stay below a threshold temperature and effusivity sets how fast that threshold is reached. Two walls of different material should then show different apex angles for the same fire. Burning a controlled fire against walls of graded effusivity and measuring the pattern's angle tests the dependence. | 6.36 |
| H5410 | Shear-direction record | Shedding from skin at a contact should depend on the shear direction as well as the pressure, because flakes detach when the interface's shear exceeds their intercellular adhesion and the detachment front propagates down-shear. A contact made with a twisting motion should deposit an asymmetric patch whose long axis follows the twist. Depositing on glass with controlled normal load and twist angle, then counting flakes per field, tests whether the asymmetry is a direction record. | 6.33 |
| H5414 | Inversion of an expected monotonicity | Pick-up of fibres on a blade should rise with penetration depth only until the fabric's elastic recovery begins to wipe the blade on withdrawal, beyond which the blade can leave with fewer fibres than it carried in. The signature is a maximum in pick-up at a depth comparable to the fabric's thickness. Stabbing a layered fabric stack at graded depths and counting fibres on the blade, with the stack's thickness varied independently, tests the inversion. | 6.33 |
| H5408 | Sign reversal from a compliance ratio | Smear taper on a transferred paint flake is read as a direction indicator, but the taper should reverse when the painted substrate is more compliant than the tool, because the transfer becomes a peel rather than a scrape. In a peel the softest layer is left behind last, so the feather points opposite to the motion. Painting rigid panels and elastomer panels with the same coating and dragging an identical tool across both separates the two cases. | 6.29 |
| H5429 | Conditional coincidence | Metameric pairs should be more common among dyed blacks than among saturated colours, because a black is built from several absorbing dyes whose combined spectra can be matched in more than one way. Two blacks that match under daylight can then diverge under tungsten, and the divergence is a property of the dye mixture as much as of the illumination. Measuring reflectance under a daylight simulator and under tungsten for a set of blacks, and counting pairs whose difference changes sign, tests the claim. | 6.13 |
| H5496 | Non-monotone roughness response | Low-energy X-ray intensity from a pressed powder should vary non-monotonically with surface roughness, because roughness both shadows the detector and lengthens the effective path, and the two effects oppose. Light-element quantification should then depend on pressing pressure. Pressing one powder at graded pressures, and measuring roughness by profilometry alongside intensity, tests the non-monotonicity. | 4.82 |
| H5443 | Narrow population as a discriminator | Road-marking beads should be distinguishable from container glass by their narrower index distribution, because the beads are drawn from a batch selected for retroreflectivity, and their surfaces should also carry an alkali-depletion layer formed during flame spheroidisation. Measuring index distributions and surface alkali by XPS for both populations tests whether the two criteria agree. | 4.8 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5786 | Viscosity of a second liquid phase controlling release | Complex coacervation between two oppositely charged species forms a dense liquid phase that holds an active far above its solubility, and release from it is governed by the coacervate's own viscosity rather than by diffusion through water. Rate should therefore scale with that viscosity, which is measurable directly. Coacervates of two viscosities at one loading should give rates differing in proportion. | 6.64 |
| H5714 | Inversion is a kinetic event, not a phase event | Phase diagrams predict where an emulsion should invert, and they are wrong about it, because inversion is triggered by the dispersed phase coalescing into a connected structure rather than by packing alone. The critical fraction therefore depends on the shear history that produced the sample, not only on composition. Two emulsions of identical composition made by different orders of addition should invert at measurably different fractions. | 6.53 |
| H5721 | Solubilised fraction subtracted from the dispersed fraction | Nominal loading overstates the dispersed phase, because part of the oil is solubilised in micelles rather than dispersed as droplets, so the effective fraction and the droplet count are both lower than the recipe says. The solubilised part grows with surfactant concentration and with the oil's own solubility. Measuring droplet volume fraction by a method that excludes micelles, against the nominal loading, gives the gap directly, and it should widen as surfactant rises. | 6.38 |
| H5743 | Definition, not precision, behind a spread | Yield stresses quoted for one formulation differ by an order of magnitude, and the difference is a definition rather than an error: the stress at which the material flows on the test's timescale against the stress at which it has flowed indefinitely. The two are joined by the material's relaxation time, so a long relaxation opens a wide gap between them. A three-rate sweep extrapolated to zero shear, set against a constant-stress creep test, should close the gap as the relaxation shortens. | 6.18 |
| H5735 | Reversibility used as the discriminator | Flocculation by depletion is reversible on dilution while flocculation by bridging is not, since the bridges must break before the droplets can separate. That difference is a test requiring no knowledge of the formulation's composition. Diluting a flocculated sample in its own continuous phase, and watching the size return, separates the two without a model. | 4.78 |
| H5785 | Loss to the container counted as decay | Container walls take a fraction of a dilute active, so the concentration delivered is lower than the concentration made and the loss is usually booked as degradation. The test is a mass balance on the emptied container: desorbing the wall with a solvent should recover the missing material if adsorption rather than breakdown is the cause. | 4.76 |
| H5799 | Proxy that cannot discriminate | Free surfactant, not pH, is what a formulation's stability depends on, and pH serves as a specification because the two are linked through a buffering equilibrium that shifts with raw material purity. Two batches can meet the pH specification and differ in free surfactant. Measuring free surfactant alongside pH should identify which of them tracks stability. | 4.76 |
| H5757 | Chemical and physical loss distinguished by heat | Irreversible thinning after pumping is called shear degradation, but the loss can be a permanent change in the thickener's association state rather than chain scission. Heating a sheared sample above the association temperature and cooling it restores a physically disrupted network and cannot restore a cleaved chain. That difference is the test, and it needs no chromatography. | 4.73 |
| H5790 | Calibration valid only in one rheology | Calibration with a Newtonian oil does not transfer to a yield-stress fluid, because the flow field changes once the sample has a plug. The error is largest near the yield point, which is exactly where the measurement is wanted. A certified non-Newtonian reference, or a change of geometry, should catch the calibration failing where the plug sits. | 4.73 |
| H5783 | Cancellation between mesh size and path length | Swelling shells gain mesh size and path length together, so the diffusion coefficient can stay nearly constant across a range of crosslink densities and release appears insensitive to crosslinking. That cancellation explains why crosslinking sometimes fails to change release at all. Shells of three crosslink densities, with swelling measured, should expose both the cancellation and where it breaks. | 4.71 |
| H5792 | Mechanism change under acceleration | An elevated-temperature stability test predicts shelf life at ambient, but the dominant destabilisation at elevated temperature can be a different mechanism, such as surfactant desorption or oil crystallisation, so the extrapolation predicts a mechanism that never occurs at ambient. Running one formulation at three temperatures and identifying the mechanism at each, rather than only the rate, should catch the change. | 4.67 |
| H5734 | Maximum in flocculation before restabilisation | Bridging flocculation is said to need polymer beyond monolayer coverage, yet for a high-molecular-weight adsorbing polymer the opposite holds: at partial coverage one chain can attach to two droplets, and flocculation is strongest where coverage is incomplete. Raising the dose can therefore restabilise a flocculated emulsion. A dose series measured for turbidity and low-shear viscosity should pass through a flocculation maximum before restabilising. | 4.6 |
| H5741 | Gradient along the sample overlooked | Polymeric surfactants can themselves cream or sediment, since their density differs from the continuous phase's, and the coverage gradient that results is what produces the top-to-bottom difference in stability. A well-mixed sample cannot reveal it, because mixing removes the gradient the measurement needs. Samples drawn from the top and bottom of a stored emulsion should differ in interfacial coverage if the mechanism is real. | 4.58 |
| H5752 | Process ranked by the wrong deformation | Performance in spraying or roller application is predicted from shear viscosity, but the deformation that matters is extensional, and the two rank differently because a thickener built from long chains resists extension far more than one built from compact associations. Two formulations matched in shear viscosity and differing in extensional resistance should atomise differently. Filament breakup time measured alongside the shear curve separates them. | 4.58 |
| H5756 | One property saturates while another grows | The plateau modulus a thickener can deliver is limited by the number density of elastically active strands, which for a given chemistry is set by the molecular weight between crosslinks rather than by concentration above overlap. Adding polymer past the entanglement concentration therefore raises viscosity while the modulus barely moves. A concentration series measured for both should flatten the modulus while viscosity keeps climbing. | 4.58 |
| H5770 | Stabiliser becoming a destabiliser above a dose | A long-chain alcohol added to a surfactant foam usually increases stability, but above a threshold it destabilises, because at high surface coverage the alcohol forms a condensed layer that cracks rather than flows. Stability therefore has a maximum in alcohol content. A series at fixed surfactant with rising alcohol, measured for foam life and surface shear modulus, would put the maximum where the modulus changes character. | 4.53 |
| H5727 | Two states with one size signature | Growth in measured droplet size on storage is read as coalescence, but flocculation gives the same rise in an instrument that counts an aggregate as one particle. The states differ in whether the size returns on dilution or gentle shear, and in whether the interfacial area has changed at all. Measuring interfacial area by surfactant titration alongside the size separates them without any assumption about the mechanism. | 4.51 |
| H5711 | A force attributed to the wrong layer | Repulsion measured between two surfactant-coated surfaces in water is attributed wholly to hydration, but part of it is the osmotic pressure of counterions confined between the charged layers. The two respond differently to salt: hydration is nearly indifferent to ionic strength while the osmotic term collapses at high salt. Force profiles at three salt concentrations would put the osmotic part's disappearance at a concentration the hydration term ignores. | 4.49 |
| H5724 | Budget closed over the wrong inventory | Surfactant budgets are closed by assuming the whole charge sits at the interface, but a significant part sits in micelles or in a separate lamellar phase, and only the interfacial part sets stability. Two formulations with the same total surfactant but different partitioning should therefore differ in stability at identical nominal coverage. Measuring the interfacial excess directly and comparing it with the total measures how much is unavailable. | 4.49 |
| H5781 | Amphiphile that modifies its own barrier | An amphiphilic active accumulates at the droplet interface, where its local concentration can exceed the bulk by orders of magnitude, and at sufficient coverage it changes the film that the same molecule must later cross. Release is then governed by a barrier the active has itself built. Measuring release at two interfacial areas at fixed bulk loading tests whether the rate scales with area, as an interfacial reservoir requires. | 4.47 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2339 | Feedback loop | Shear heating during slip raises pore pressure, which lowers effective stress and promotes further slip, giving a positive feedback whose onset is a critical slip rate. The feedback should produce a strain-rate threshold for localization that is independent of total displacement. Experiments at controlled slip rates bracketing the threshold, with pore pressure measured during slip, show whether localization begins at a rate rather than at a strain. | 6.51 |
| H2379 | Count | Blast fragment counts are set by the in-situ block size, not by the explosive energy, because the energy propagates existing discontinuities and comminutes only the intact bridges between blocks. A massively jointed rock should give a fragment count close to its block count regardless of powder factor. Counts from two blasts of different powder factor in the same blocky mass, with the block size surveyed, test the claim. | 6.49 |
| H2394 | Count | Bolt counts for a roof are fixed by the number of kinematically admissible blocks, not by the roof area, so a larger span can need fewer bolts if its blocks are larger. The count should then be predictable from the joint survey alone, without a stress analysis. Comparing bolt counts that performed well across openings of different span in one jointed mass, against surveyed block counts, tests that. | 6.49 |
| H2360 | Reversal | Frother concentration usually increases recovery by stabilising bubbles, but beyond a point it decreases recovery because smaller bubbles raise the detachment probability in the froth and reduce each bubble's carrying capacity. The reversal should occur at a lower concentration in a deep froth than in a shallow one, since residence time in the froth is longer. A concentration series at two froth depths tests whether the reversal moves as predicted. | 6.44 |
| H2378 | Reversal | Heavier weight on bit normally drills faster, but in a deep, small-diameter hole it can drill slower, because the added load buckles the string, misaligns the bit, and converts axial load into lateral friction. The reversal should occur at a weight set by hole diameter and string stiffness, and should be absent in a large vertical hole. A weight series at two diameters, with survey data showing string curvature, tests it. | 6.42 |
| H2322 | Hysteresis | In a roll-type uranium deposit the redox front advances and retreats along different pH-Eh paths, so its position depends on the direction of the last fluid event rather than on the present chemistry. A front that retreated should sit inside its own alteration halo, leaving an outer reduced zone with no ore. Mapping the redox boundary and the ore boundary separately, in a deposit with a documented flow reversal, tests the prediction. | 6.4 |
| H2369 | Boundary layer | Thickener performance is set by the permeability of the hindered-settling bed, which is fixed by floc size rather than by feed rate, so a thickener can be stable over a range of fluxes bounded by two permeabilities. Within that range, changing the feed rate should change bed height but not underflow concentration. A flux series with bed height and underflow density measured together shows whether the operating window has the predicted bounds. | 6.4 |
| H2377 | Threshold | No chip forms below a minimum weight on bit, because the bit's contact stress stays below the rock's ductile-to-brittle transition, and the threshold is a pressure rather than a force, so it scales with bit area. A larger bit should need proportionally more weight, not the same weight, to start chipping. Testing two bit sizes on one rock, with the threshold identified from cuttings morphology, shows whether the threshold is a stress. | 6.4 |
| H2389 | Hidden term | Support pressure of a grouted rockbolt includes the plate's bearing capacity, so a bolt in soft rock delivers less than its steel capacity because the plate punches into the surface. Delivered load should therefore scale with plate area and the rock's bearing strength, not with the bolt's diameter. Pull tests with two plate sizes on one rock and bolt, measuring the load at which the plate begins to embed, quantify the term. | 6.4 |
| H2393 | Reversal | Denser bolting is assumed to improve stability monotonically, but in a highly fractured rock it can reduce it by clamping the surface and preventing a natural arch from forming, so load transfers to a thinner zone. The reversal should occur at a density set by block size rather than by span. A density series at a blocky test site, with convergence and bolt load measured together, shows whether there is an optimum. | 6.4 |
| H2331 | Scale argument | Strength scaling from laboratory to field is often fitted as a single power law, but the exponent changes at the crossover between flaw-controlled and fabric-controlled failure, so an exponent fitted over two decades mispredicts at four. The crossover should appear as a break in slope when strength is plotted against size for specimens spanning the transition. Testing specimens from centimetre to metre scale in one rock mass, rather than extrapolating, locates the break. | 6.38 |
| H2363 | Scale argument | Laboratory flotation cells cannot reproduce plant hydrodynamics because bubble size and turbulence scale differently from cell size, so a scale-up factor derived from residence time alone is not geometric and will mispredict recovery. The mismatch should grow with cell size and be largest for coarse particles, whose detachment depends on turbulence. Running one ore in cells of three sizes at matched bubble size, rather than matched impeller speed, tests which variable governs the scale-up. | 6.36 |
| H2305 | Reversal | Cooling is the default explanation for sulfide precipitation, yet several metal complexes have retrograde solubility, so precipitation should occur where the fluid is hottest along its path. Ore shoots should then sit at the thermal maximum, near the intrusion or at the deepest part of the conduit, rather than at the distal cool tip. Homogenization temperatures measured along one continuous vein settle it: a positive correlation between temperature and grade reverses the standard distal-cooling picture. | 6.33 |
| H2330 | Artefact-mimics-effect | Time-dependent strain in a triaxial cell is often read as creep, but a saturated specimen keeps compacting as pore fluid drains from the dead volume of the platens, and the resulting displacement looks exactly like primary creep. The artefact should scale with the system's storage volume and vanish for a dry specimen at the same stress. Testing a dry and a saturated specimen of the same rock, with the drainage path instrumented, separates the two. | 6.33 |
| H2324 | Hidden term | Uniaxial compressive strength measured in the laboratory contains a term from the loading system's stiffness, because the machine's compliance adds to the specimen's measured displacement and alters the post-peak behaviour from which the peak is read. The same rock tested in a stiff frame and a soft one should differ, and the difference should scale with the machine-to-specimen stiffness ratio. Testing one homogeneous block in two machines, with the platens instrumented separately, quantifies the term directly. | 6.31 |
| H2337 | Memory effect | Rock retains a memory of the maximum stress it has experienced, expressed as a Kaiser effect in acoustic emission, and this is used to bound palaeostress. The memory is erased by pore fluid at elevated pressure, so a wet specimen should show no Kaiser effect above a threshold saturation. Testing one rock dry and at several saturations, with the stress memory imposed under controlled conditions, establishes the saturation limit beyond which the palaeostress bound is invalid. | 6.31 |
| H2344 | Threshold | Below a critical particle size, fracture cannot propagate because the flaw population is exhausted, so grinding ceases to be a fracture process and becomes one of plastic wear. The threshold should scale with the largest intrinsic flaw in the material, which is a microstructural length rather than a grindability number. Grinding a series of materials with known flaw sizes to successively finer sizes, looking for the size at which the energy-fineness curve flattens, tests the scaling. | 6.31 |
| H2376 | Bound | Maximum rate of penetration in a given rock is bounded by the rate at which cuttings can be removed from the face, not by the weight on bit, so beyond the cleaning limit additional weight only re-grinds debris. The bound should scale with annular flow area and the fluid's carrying capacity, not with the bit's cutting structure. Drilling at a series of flow rates and weights, with cuttings concentration measured at the return, locates the bound. | 6.31 |
| H2313 | Sign error | Skarn zonation is conventionally drawn with garnet proximal and pyroxene distal to the intrusive contact, but the sense of the zonation is fixed by the sign of the coupled calcium-iron flux, which reverses in reduced intrusions. A reduced skarn should therefore carry the iron-poor mineral proximally, making the textbook pattern a redox indicator rather than a universal sequence. Mapping one skarn from marble to pluton in both an oxidized and a reduced system tests the prediction directly. | 6.29 |
| H2318 | Feedback loop | Mineral precipitation lowers permeability, which raises fluid pressure until the fracture re-opens, and the cycle's period is set by the poroelastic diffusivity of the wall rock rather than by tectonic forcing. If so, band spacing multiplied by the number of bands should give a time that scales with the square of the diffusivity length. Measuring band spacing in veins hosted by wall rocks of contrasting diffusivity, at the same depth, tests the scaling without needing an absolute date. | 6.27 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3469 | Identity | Reaction sintering and reactive melt infiltration are one process at two melt fractions, both dissolving a solid into a liquid that then precipitates a new phase in place. If so, one reaction couple should give the same product morphology at either fraction, scaled by the liquid's volume. One couple run at two fractions, with morphology compared, tests it. | 6.71 |
| H3406 | Identity | The seeds counted in a melt and the blisters counted after forming are one bubble population seen at two times, not two defects. If so, a glass clean at the forehearth but blistered in the mould must have gained its bubbles in between, and the gain should follow the forming temperature. Sampling one ladle before and after forming, at a single temperature, tests it directly. | 6.56 |
| H3480 | Bound | The strength of a polycrystalline ceramic is bounded above by the largest facet crack a grain of its size can host, so no processing change raises strength past a value set by grain size alone. The bound is geometric and does not depend on boundary chemistry. Two ceramics of one grain size but different boundary chemistry, tested in bending, should share the bound while differing below it. | 6.51 |
| H3425 | Causal inversion | Cracking in gel-cast bodies is blamed on shrinkage, but the crack spacing is set by the gel's permeability, which decides how fast a shrinking body can drain and therefore how much stress builds before a crack relieves it. A more permeable gel should crack less often and with wider spacing, which inverts the usual reading. A permeability series at fixed solids, with crack spacing counted, tests the inversion. | 6.49 |
| H3478 | Environmental threshold | A fatigue limit in glass exists only below the stress at which the crack's opening becomes narrower than the water molecule's access, so a threshold measured in humid air is a threshold of the environment rather than of the glass. In vacuum the same glass should show none. One glass, one crack, tested in air and in vacuum across a stress range, places the threshold in one environment and not the other. | 6.4 |
| H3479 | Identity | Indentation strength and scratch strength are one residual-stress-controlled flaw measured in two geometries, so a ranking by one should predict the other for the same glass. The residual field is the same and only the contact shape differs. Both tests on one set of glasses, with the residual field read by birefringence, tests the identity. | 4.78 |
| H3489 | Identity | Damage from a thermal quench and damage from a sharp contact are the same crack population produced by two loading paths, so a thermal-shock ranking should predict a contact ranking. Both nucleate at the surface and grow on the same planes. One set of ceramics ranked both ways, with the crack patterns compared, tests the identity. | 4.76 |
| H3461 | Trade-off | Glaze hardness and glaze crack resistance cannot both be raised in one composition, because hardness needs a stiff, well-connected network while crack resistance needs a compliant one that can deform at the tip. A scratch-resistant glaze is therefore a chip-prone one. A composition series across the trade-off, with both measured on one firing, should trace a single curve. | 4.71 |
| H3496 | Conservation argument | The energy released when a tempered plate fragments is its stored elastic energy, so fragment count per unit area is a direct reading of tempering stress that does not need the glass's toughness. Counting fragments is then a cheaper stress measurement than photoelasticity. One plate fragmented, with fragment count and birefringence measured, calibrates the count against the stress. | 4.71 |
| H3422 | Identity | Die swell in a ceramic paste and springback in a pressed compact are one elastic recovery seen in two geometries, and the same shear modulus should predict both. If the identity holds, a paste that swells by a given fraction in extrusion must spring back by a corresponding fraction in pressing. One paste, both measurements, no change of formulation, tests it. | 4.64 |
| H3481 | Count | A fracture surface carries at most three independent features marking crack-velocity regimes, mirror, mist and hackle, so a fractographic report claiming five regimes is counting roughness rather than velocity. Each feature maps to a velocity range, and the ranges do not overlap. One surface measured for roughness and for velocity along a travelling microscope shows whether the extra features have distinct velocities. | 4.62 |
| H3482 | Scale argument | The rising toughness seen in alumina is credited to a material R-curve, but its plateau is set by the specimen's ligament length, so a larger specimen keeps rising where a small one has flattened. The plateau is then a specimen property, not a material one. One alumina on three specimen sizes, with crack length measured, should move the plateau with the ligament. | 4.62 |
| H3495 | Symmetry argument | An isotropic glass should fragment isotropically, so a preferred crack direction records a frozen-in stress from forming or annealing, which makes the fracture pattern a stress map no polarimeter is needed to read. The direction follows the stress's principal axes. One plate broken on a support, with fragment orientations measured, should recover that direction and can be checked against a birefringence map. | 4.62 |
| H3499 | Threshold | Crack healing in glass needs a temperature at which surface silanol groups condense, so a healed crack's strength depends on the humidity during annealing and not only on the temperature. A dry anneal heals less than a humid one at the same temperature. One glass, indented and annealed at one temperature in two humidities, with strength and interfacial water measured, places the threshold. | 4.6 |
| H3423 | Bound | Tape thickness in doctor-blade casting is bounded above by the slurry's yield stress rather than by the blade gap, because a thick wet tape flows under its own weight before it dries. Past the bound, opening the gap only makes a wider puddle. A gap series at two yield stresses, with dried thickness measured, shows thickness saturating where rheology takes over from the gap. | 4.58 |
| H3454 | Two-sided bound | Opacity from an opacifier is bounded at both ends of particle size: below roughly a tenth of the wavelength the particles scatter too little, and above a few wavelengths there are too few of them at a given loading. The maximum lies between, and its position depends on the refractive-index contrast. A milling series at fixed loading, with reflectance measured, shows the two-sided bound. | 4.58 |
| H3450 | Energy threshold | A glaze shivers only when its compressive stress multiplied by its thickness exceeds the energy needed to separate it from the body, so a badly matched glaze applied thin stays put indefinitely. The threshold is a thickness, and it moves with the interface's toughness. A thickness series of one mismatched glaze on one body places it and shows it scaling with the measured interface energy. | 4.56 |
| H3455 | Identity | A matte glaze and a devitrified glaze are one surface at two crystal sizes, the matte being the one whose crystals are too small to scatter light individually. If so, every matte glaze should show a crystalline phase on diffraction, and its surface should roughen on overfiring as the crystals grow. Diffraction and microscopy on one matte and one crystalline glaze from a common base test the identity. | 4.49 |
| H3460 | Conservation argument | Metal release from a glaze into acid is bounded by how fast the silica network dissolves, not by how much metal the glaze contains, so a lead-free glaze with a more soluble network can release more of a less toxic metal. Total content is the wrong variable. Leaching one glaze at two acid strengths, with the network's dissolution followed by silica in solution, tests whether release tracks dissolution. | 4.49 |
| H3408 | Count | Property maxima in alkali borate melts, including the viscosity maximum, can all be carried by one structural variable, the fraction of boron in four-fold coordination, so a model needing two species for two maxima is over-parameterised. The count is one, and it is measured by resonance rather than inferred. Coordination measured across a composition series, compared with the position of each maximum, shows whether one count carries them all. | 4.47 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1423 | Reversal | Channel ribs, as they grow taller, eventually reduce the Nusselt number, because the recirculation cell behind a tall rib blankets the wall it was meant to scrub. The optimum height should track boundary-layer thickness rather than rib spacing. A height series at two channel heights, with local flux measurement, puts the optimum where the boundary-layer argument predicts. | 6.73 |
| H1490 | Bound | Critical heat flux cannot be raised indefinitely by wettability, because beyond a contact-angle threshold the vapour film's stability is governed by the surface's effusivity, so a superhydrophilic coating on a low-effusivity substrate can underperform bare high-effusivity metal. One coating on two substrates tests the bound. | 5.0 |
| H1477 | Bound | A latent module's volumetric energy density is bounded by PCM packing fraction times latent heat, and no encapsulation scheme exceeds it; a reported value above the bound means the shell's sensible heat was counted. Recomputing from composition and measured shell mass tests the bound. | 4.98 |
| H1475 | Misattribution | Subcooling measured in a DSC pan is partly nucleation kinetics and partly the pan's own resistance delaying the sensed onset; the pan part scales with scan rate while nucleation does not. Two scan rates and two pan masses on one material separate them. | 4.96 |
| H1476 | Hidden term | Every latent heat measured by the T-history method includes the sensible heat of the thermocouple sheath, which sits inside the insulated boundary and is not sample; the term grows as samples shrink. Two sheath diameters at the same sample mass would show it. | 4.96 |
| H1497 | Identity | The 'contact' and 'constriction' resistances reported for a bolted joint are the same quantity at two length scales: the joint resistance equals the microcontact constriction resistance in vacuum, and only interstitial gas separates the two numbers. One joint measured in vacuum and in helium tests the identity. | 4.96 |
| H1441 | Count | An N-surface diffuse enclosure has N(N-1)/2 view factors to determine, but radiosity measurements on each surface can fix at most N-1 per surface, so experimentally 'measured' matrices are usually rank-deficient. Taking the rank of a measured three-surface matrix and comparing it with its geometric prediction is the test. | 4.93 |
| H1450 | Hidden term | Effectiveness-NTU for a condenser hides the desuperheating zone, whose area is counted in the two-phase total while its coefficient is gas-like; the error peaks at part load and no single NTU absorbs it. Effectiveness measured at three inlet superheats for the same duty would size the hidden term. | 4.93 |
| H1481 | Artefact mimics effect | Conductivity enhancement claimed for graphite-loaded PCM can be the measurement cell's own conduction when the sample is thin, and the tell is enhancement that scales with cell gap rather than with loading. Two cell gaps at fixed loading would expose it. | 4.91 |
| H1483 | Conservation audit | Thermochemical cycle efficiencies must close an energy balance including reactor sensible heat; cycles that omit it can appear to exceed the second-law bound, which is the audit's red flag. Recomputing each reported efficiency from measured masses and temperatures is the audit. | 4.91 |
| H1464 | Hidden term | Guarded hot plate measurement of a low-conductivity blanket carries an edge term whose area grows with thickness while the metered area does not, and the term presents itself as a thickness-dependent conductivity. Two plate diameters at one thickness would expose an edge term that does not scale with metered area. | 4.89 |
| H1493 | Count | A boiling surface with N cavity-size families shows at most N changes in the boiling curve's slope, so counting slope changes gives the family count. A surface with deliberately graded cavity sizes would test the count. | 4.8 |
| H1410 | Degeneracy | Step-heating a slab leaves effusivity and contact resistance inseparable whenever both act on the same time scale, and the two are effectively degenerate wherever the resistance time constant lands within a decade of the slab's diffusion time. A second thickness breaks it. Test: two thicknesses, one contact, joint fit with and without the second specimen. | 4.78 |
| H1469 | Count | An N-shield stack has at most N+1 steady temperature nodes, so a measured profile with more inflections contains a layer whose spacer has detached. Profile measurement with sensors threaded between shields is the test. | 4.76 |
| H1428 | Bound | Passive enhancement cannot escape the pumping-power penalty's cube law: at fixed pumping power the gain in Nusselt number is bounded by roughly the one-third power of the area increase (estimate from the Reynolds analogy). The bound is geometry-independent and would explain why reported gains cluster below about two. Replotting literature data at fixed pumping power rather than fixed velocity tests it. | 4.64 |
| H1414 | Bound | Multilayer insulation has a lower bound on resistance at a given areal density: below roughly ten layers per centimetre the layer-to-layer contact spots conduct more than the interlayer gas, so adding layers raises conductance instead of lowering it (estimate for typical aluminised films). The bound is set by contact-spot fraction, not by foil emissivity. Measure conductance against layer count at fixed compression. | 4.62 |
| H1406 | Reversal | Thin-film conductivity, with the heater deposited on the film, can first fall and then rise with thickness, because boundary scattering rules the film while the substrate's spreading resistance rules the measurement at small thickness. The minimum sits where those two resistances are equal, so it moves with substrate conductivity. Separate the regimes with one thickness series deposited on two substrates of very different conductivity. | 4.58 |
| H1404 | Threshold | Coordination number near four puts a knee in a granular bed's effective conductivity: below it the load rides on a few force chains and conductivity becomes strongly pressure-dependent, above it the network is redundant and pressure hardly matters. The knee is mechanical in origin, so it should appear in the bed's stiffness at the same load. The decisive measurement is conductivity and Young's modulus on one bed while uniaxial load is swept. | 4.4 |
| H1431 | Derived quantity | The 'convective velocity' obtained by cross-correlating two heat-flux gauges is derived, not measured, and equals the flow speed only if the thermal disturbance travels at that speed; in a boundary layer it travels at a fraction of it. That fraction is the same function of wall distance across boundary layers, so it can be calibrated once and reused. Two gauges at different wall distances, checked against hot-wire truth, would fix the calibration. | 4.4 |
| H1402 | Hidden term | Any transient hot-wire value of conductivity in a moist porous solid carries a hidden latent-heat term, because vapour distilled off the hot boundary re-condenses a few millimetres away and returns by capillary flow, so the wire cools by mass transfer as well as by conduction. That term scales with heating rate while conduction does not. Extrapolate apparent conductivity to zero wire power; in a water-bearing sand the term could plausibly reach a third of the reading (estimate, untested). | 4.27 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5571 | Coincidental correlation | Thread count correlates with painting date only because the textile trade's economics changed, so dating a canvas from its weave assumes the trade's history and then claims to recover it. The correlation is real but circular, and it should not be used where the trade's local history is unknown. A documented local trade turns the assumption into a measurement. | 6.69 |
| H5538 | Counterfactual claim | A treatment is reversible only if the solvent can reach the interface it must undo, and in a porous substrate the diffusion path can be longer than the interval between treatments. Reversibility is thus a claim about plumbing rather than about chemistry, and it should be stated together with the path length that makes it true. The path length is measurable by tracer penetration, and it belongs in the treatment record. | 6.42 |
| H5544 | Underdetermined inverse | Many varnish histories produce the same gloss and colour, so recovering a history from appearance is underdetermined. Cleaning decisions framed as restoring the original therefore amount to choosing among equally consistent histories, and the choice should be argued on grounds other than the measurement. | 4.98 |
| H5564 | Coupled ageing | Hydrolysis shortens the cellulose chains and the shorter chains plasticise the fibre, so a canvas can become weaker and more extensible at the same time. The two changes are coupled, and measuring one while assuming the other constant inverts the interpretation of the other. | 4.98 |
| H5569 | Gatekeeper variable | Whether a canvas's degradation products escape depends on the ground's pore structure rather than on its chemistry, so a tight ground traps the acids it was meant to keep out. The ground's permeability is the gatekeeper, and measuring it should precede any choice of lining adhesive. | 4.98 |
| H5537 | Error dominance | In tensile testing of a painted canvas the reported scatter in Young's modulus is dominated by the extensometer's own compliance, not by the material. The check is to test the same strip at two attachment lengths, which should shift the apparent modulus systematically if compliance is the cause. | 4.93 |
| H5548 | Inverse-size scaling | The shrinkage stress a drying consolidant exerts on a particle scales inversely with the particle's radius, so a formulation safe on coarse pigments will crack fine ones. This explains the common pattern in which whites survive a treatment that destroys lakes, and it predicts which passages to test first. | 4.91 |
| H5578 | Competing annealing | Light generates radicals in a varnish while moisture anneals them, so yellowing should be greatest at intermediate humidities and fall at both extremes. A maximum in the damage-versus-humidity curve would explain why some well-ventilated stores show less yellowing than expected. | 4.89 |
| H5558 | Optical artefact | Apparent brightening after aqueous treatment is often an artefact of measuring a wetted sheet against a dry reference, since water changes the refractive index match between fibres and air. The paper has not changed and the geometry of the measurement has, so a re-measurement after full equilibration should remove the effect. | 4.87 |
| H5539 | Cross-contamination | Solvent gels leave residues below the detection limit but not below the threshold of chemical activity, and those residues alter the next treatment's penetration. The effect is a hidden term in every later measurement, and it can be exposed by treating a blank with the gel and measuring what it does to a fresh film. | 4.78 |
| H5580 | Limit crossover | A gypsum crust on stone may protect the stone or destroy it, according to which dominates, the crust's diffusion resistance or its content of soluble salt, and the outcome is opposite in the two regimes for a crust of identical thickness. The crossover thickness is measurable and should decide whether a crust is removed or left. | 4.76 |
| H5533 | Mass-budget closure | The consolidant that matters is the fraction that stayed where it was placed, while surface uptake measurements count the fraction that wicked away as success. Closing the mass budget requires weighing the interior, which is destructive, so most reported efficiencies are upper bounds. Quoting the wicking ratio beside the uptake would make two studies comparable that now are not. | 4.69 |
| H5591 | Interface artefact | An optical coherence tomogram's apparent varnish layer can be a refractive-index interface within the paint, so layer counts overstate the varnish and understate the paint. The ambiguity is resolved only by an independent thickness measurement at the same point, which is rarely made. A second angle of incidence would separate the two, since only one of them moves with it. | 4.67 |
| H5506 | Sub-threshold null | Prussian blue fading by reduction is balanced by re-oxidation, and below some light intensity the two rates could cancel so that no net fading occurs at all. The threshold would be set by the binder's oxygen permeability, which predicts that the same pigment under a low-permeability varnish has a higher threshold, and that a sample called light-fast may only have been tested dimly. | 4.6 |
| H5502 | Hidden term | Every measured rate of vermilion blackening contains a term proportional to the chloride available from the substrate, the ground, or the binder, and that term never appears in pigment-only light tests. One then expects the same pigment from the same maker to blacken at rates differing by an order of magnitude, estimated, when it is bound in a chloride-bearing glue ground rather than in oil. | 4.56 |
| H5517 | Surface reservoir saturation | Indigo fading saturates because once the indigo at the air interface is consumed, the remaining pigment lies behind a binder layer that absorbs the actinic wavelengths. A fading rate measured on a long-exposed sample is therefore an underestimate of the initial rate, and accelerated tests should report the early slope. The early slope is the only part of the curve that reports the material rather than the lighting. | 4.49 |
| H5503 | Reaction front as clock | Blackening of lead white by hydrogen sulfide cannot penetrate deeper than the sulfide front, whose position is set by diffusion rather than by exposure time, so the depth of the dark layer is a clock and not a measure of damage. Cleaning a darkened lead white surface therefore exposes paint that has never reacted, which is why cleaned passages look younger than the passages beside them. Two paintings of equal exposure and different ground porosity should then show different front depths. | 4.2 |
| H5530 | Transport latency | Nanolime appears inert for weeks and then hardens, because carbonation waits on carbon dioxide diffusing into the pore. The delay is a transport time and should scale with the square of the penetration depth, which makes it longer in dense stone even when the chemistry is identical. | 3.96 |
| H5590 | Spot-size bound | The resolution of a scanning image is bounded by the source's spot size rather than by the pixel count, so oversampling adds file size and no information. A knife-edge target scanned at two pixel pitches should show the same edge width if the spot is the limit. | 3.96 |
| H5584 | Selective penetration | Ozone attacks the surface of rubber but the bulk of paper, because the materials differ in how fast the oxidant is consumed relative to how fast it diffuses. A single oxidant dose therefore cannot rank materials, and the ranking reverses with thickness. The reversal follows from the ratio of reaction rate to diffusion coefficient, and it should be predictable. | 3.69 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4172 | Hot-spot migration | In a packed bed of sulfide concentrate the reaction zone can localise because product formation changes the bed's permeability, so gas bypasses the hot zone and reaction migrates elsewhere. The hot spot should move when gas flow is reversed, tracking the gas rather than the solids, and temperature mapping in a bed run in two directions tests which it follows. | 6.67 |
| H4166 | Boundary-layer control | Volatilisation of a metal from a melt is modelled as first-order loss, but the rate is controlled by mass transfer through the gas boundary layer, so loss should scale with gas flow and surface area rather than with metal concentration. Once the interfacial reaction is fast the rate becomes concentration-independent. Loss measured at two gas flows and two surface areas, at one concentration, identifies the control regime. | 6.51 |
| H4146 | Autocatalytic growth | Nodules grow because they sit at locally higher current density, and that higher density is itself caused by the nodule's protrusion into the boundary layer, so growth is autocatalytic and the final size distribution is set by initial roughness. Periodic current reversal, or a particle blocking the nodule tip, should interrupt it, and starting cathodes at controlled roughness while following the size distribution over time tests the feedback. | 6.49 |
| H4102 | Hidden reservoir | Acid demand measured in a short bottle-roll test leaves out the protons that clay and iron oxyhydroxide surfaces adsorb and release over weeks, so a heap's true acid consumption is set by a reservoir the test never fills. Washing the residue with several bed volumes between stages releases that stored acid and exposes the short test as a lower bound. If the reservoir is real, its size should scale with the gangue's cation-exchange capacity and not with its carbonate content. | 6.22 |
| H4180 | Autothermal threshold | A sulfide roaster or smelter becomes autothermal only above a feed sulfur content set by heat losses and off-gas volume rather than by reaction enthalpy alone. Recycling off-gas lowers the threshold by cutting the heat that leaves in the gas. Two campaigns with and without recycle test the threshold's dependence. | 4.89 |
| H4193 | Effective dead time | A control loop's dead time is estimated from a step test, but in a circuit with recycle the effective dead time is longer than the physical transport time because the recycle returns a delayed disturbance. Effective dead time scales with recycle ratio. Step responses measured at two recycle ratios test the scaling. | 4.84 |
| H4184 | Masking | An impurity can sit above its specification without causing a defect because another impurity masks it, so tightening one limit without regard to the other can make the product worse. In the two-impurity plane the boundary is diagonal, not two independent limits. Depositing cathodes across a grid of the two concentrations tests whether the boundary is diagonal. | 4.73 |
| H4138 | Poisoning threshold | An organic can tolerate a strongly bound impurity up to a point without losing capacity, because the impurity occupies sites off the main extraction path, and capacity falls steeply only beyond that point. A plateau, then a drop, is the shape the capacity-versus-impurity curve takes. Loading a series to different impurity levels and then measuring target capacity maps it. | 4.69 |
| H4168 | Coupled operating window | Matte grade and slag magnetite are treated as independent operating variables, but they are coupled through oxygen potential and silica activity, so raising grade raises magnetite unless silica is raised too. In the grade-magnetite plane the coupling draws a boundary that cannot be crossed without a phase change. Operating at two silica levels and mapping the accessible region tests it. | 4.67 |
| H4177 | Le Chatelier reversal | A flux is expected to lower a slag's liquidus, but if it raises the activity of a component already at its solubility limit, the addition can raise the liquidus instead. The reversal sits at the flux composition where a new phase becomes saturated. Liquidus measured across a flux series, with the phase assemblage identified at each point, locates the reversal. | 4.67 |
| H4175 | Stratification | A molten slag can stratify into two layers of different composition when silica activity crosses a boundary, and the stratification is invisible in a single sample taken from the top. The layers differ in viscosity, so tapping behaviour changes before the analysis does. Sampling at two heights during tapping and measuring viscosity at each would show two layers if they exist. | 4.64 |
| H4198 | Surface-to-volume scaling | An impurity's transfer rate is estimated from a laboratory rate constant and a plant residence time, but the product can be wrong by an order of magnitude (estimate) because the laboratory rate was measured at a much higher surface-to-volume ratio. The discrepancy scales with the ratio of the two surface-to-volume ratios. Running one reaction at two such ratios tests the scaling. | 4.64 |
| H4164 | Concentration-scale illusion | Replacing weight per cent with mole fraction changes the apparent effect of an additive on slag viscosity, and a model fitted in weight per cent can predict the wrong direction when the additive's molecular weight differs from the slag's average. Heavy-oxide additions carry the largest error. Refitting one viscosity dataset in both scales and comparing predictions at an untested composition tests the sensitivity. | 4.6 |
| H4135 | Phase-ratio constraint | A circuit's throughput is often limited not by mixer residence time but by the settler's dispersion band at the required phase ratio, and the band height scales with the flux of the dispersed phase rather than with total flow. Invert the dispersion at the same total flow, and the constraint lifts. Band height against flux, measured for both dispersion types, maps the real limit. | 4.58 |
| H4148 | Split sink budget | Glue and other levelling additives are consumed both by incorporation into the deposit and by degradation at the anode, and the two sinks scale differently with current density, so additive concentration cannot be inferred from deposit appearance. Incorporation scales with cathode current while anodic degradation scales with anode current, so their ratio moves when anode area moves. Decay measured in cells with two anode-to-cathode area ratios separates the sinks. | 4.47 |
| H4188 | Steric fouling | An ion-exchange resin loses capacity because a foulant blocks pore mouths rather than because it occupies functional sites, so a solvent wash that removes no metal can still restore capacity. The capacity loss should be larger for the target than for a smaller ion of the same charge, and comparing the uptake of two ions of different hydrated radius on a fouled resin tests whether the block is steric. | 4.47 |
| H4106 | Moving-boundary bound | Acid cannot reach a sulfide surface faster than the neutralisation front in the surrounding gangue advances, so the leach rate is bounded by the gangue's acid capacity per unit volume rather than by the mineral's own kinetics. Double the valuable mineral's surface area at constant gangue, and the rate does not move. Leaching a dilute and a rich sample of one gangue at equal acid supply is the direct test. | 4.44 |
| H4140 | Recycle enrichment | In a circuit that recycles strip liquor to extraction, an impurity whose distribution ratio sits just below the extraction threshold can still accumulate to a high steady state, because the recycle multiplies a small extraction. The steady state is slow to arrive and follows the bleed ratio, not the extractant concentration. A staged run with an impurity spike, continued until the concentration plateaus, tests whether it follows the recycle ratio. | 4.44 |
| H4159 | Boundary crossing | A leach poised near a redox boundary can cross it when pH moves, so the same solution can dissolve a mineral at one pH and precipitate it at another with no change in oxidant supply. At constant ORP with drifting pH, the metal's concentration reverses direction at the crossing. Holding ORP with an oxidant controller while letting pH drift at a set rate tests whether precipitation is a boundary effect. | 4.44 |
| H4152 | Ionic-strength effect | An equilibrium constant measured at low ionic strength does not transfer to the ionic strengths of plant liquors, and the error is a function of composition rather than a constant offset, so the predicted direction of an impurity's response can reverse. A specific-ion interaction model fitted to the major ions makes the correction predictable. A precipitation boundary measured at two ionic strengths at the same free ion separates activity from equilibrium. | 4.42 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2020 | Count | Fully flooded isothermal line-contact film thickness collapses onto two dimensionless groups rather than the four usually tabulated, because the load and speed exponents are not independent once supply is unlimited; the extra groups describe the starved branch. Flooded and starved data should therefore need different counts on the same axes. One contact taken from flooded through starved, with the groups counted on each branch, tests whether the second pair earns its place. | 6.4 |
| H2024 | Hidden term | Every boundary friction measurement in a reciprocating rig contains a term from the reversals, where the film is squeezed out and re-formed twice per cycle, and that term grows as the stroke shortens. Coefficients measured at short strokes therefore include a squeeze contribution that is not boundary friction at all. Two stroke lengths at one sliding speed and one nominal pressure separate them, and the difference should scale with the reversal frequency. | 6.4 |
| H2025 | Threshold | An anti-wear additive stops protecting a steel contact above a temperature near 150 °C (estimate) not because it decomposes, but because its film's growth rate falls behind the wear rate there, and the two rates cross. The film is still present above the threshold; it is simply thinner than the wear it must replace. Measuring film thickness in situ against temperature at fixed load puts the crossing where wear rate overtakes growth. | 6.4 |
| H2099 | Null claim | For a contact already limited by its boundary film, raising the base oil's pressure-viscosity coefficient changes nothing, because the film's rheology is not the bulk oil's; the coefficient matters only where a hydrodynamic film survives. Any measured effect in a boundary contact is a measurement of how much hydrodynamic film remained. Two oils a decade apart in coefficient, tested at a load high enough to remove the film, should give equal friction. | 6.4 |
| H2028 | Bound | A boundary film cannot lower friction below its weakest plane's shear strength divided by the contact pressure, so there is a floor that no additive chemistry passes at a given load, and a reported coefficient below it implies that part of the load was carried by a fluid film. The floor rises with pressure, which is why low-load tests look better than service. One additive at several loads, with film fraction measured independently, should approach the floor asymptotically. | 6.27 |
| H2043 | Null claim | In a contact already fully boundary, adding a viscosity modifier to the base oil changes nothing about friction, because the film's rheology is set by the additive layer and not by the bulk polymer; any measured change is the bulk oil's contribution to the few hydrodynamic pockets that remain. Testing at a load high enough to remove those pockets should make the modifier irrelevant. One oil with and without the modifier, at three loads, shows the difference converging to zero as load rises. | 6.16 |
| H2076 | Reversal | Lubricating an abrasive contact can increase wear, because the lubricant prevents debris from compacting into a protective third-body layer and keeps it free to cut. Dry abrasion of some pairs is therefore slower than oiled abrasion at equal load. The reversal needs debris harder than both surfaces; one pair run dry and oiled, with debris hardness measured, shows it. | 4.69 |
| H2033 | Misattribution | Friction anisotropy on a rubbed boundary film is usually assigned to molecular alignment, but much of it belongs to the topography that the rubbing itself created, which is aligned in the same direction and cannot be separated by chemistry alone. Alignment relaxes on mild heating; topography does not. Annealing one specimen after rubbing and re-measuring along both directions separates them. | 4.67 |
| H2090 | Count | At high contact pressure the number of mechanisms that can act is smaller than at low pressure: adhesion, abrasion and oxidation collapse into one because real area saturates and debris cannot escape. A high-pressure test therefore cannot distinguish mechanisms that a low-pressure test separates. One pair across a pressure range, with debris populations counted at each pressure, shows them merging. | 4.67 |
| H2031 | Count | Coverage, shear strength and thickness are the only three independent chemistry variables that a boundary film's friction can depend on at fixed temperature and load. A study reporting four or more independent chemical drivers is reporting collinear proxies for those three, which is why such studies rarely reproduce. Measuring two of the three independently and predicting the third tests the count. | 4.64 |
| H2023 | Misattribution | The friction reduction credited to a friction modifier is often the modifier changing the surface's running-in rather than the boundary film's shear strength, because the same additive added after running-in gives a much smaller effect. If the film were the mechanism, the order of addition would not matter. Adding the additive before and after running-in in two otherwise identical tests separates the two. | 4.6 |
| H2034 | Hidden term | Friction measured in humid air includes the capillary term of water bridges at the contact's periphery, and in a boundary contact that term can exceed the film's shear term at low load. Humidity control alone does not remove it, because the bridges re-form on unloading. Two loads at one humidity, with friction plotted against load, give an intercept that is the capillary term. | 4.6 |
| H2040 | Artefact mimicking effect | Reports of superlubricity in boundary contacts sometimes rest on a friction force smaller than the rig's own drift, which a null run with the load removed would have measured. A drift of a few micronewtons per minute (estimate) is enough to manufacture a coefficient. The null test, same motion with no contact, is the discriminator, and it is rarely reported alongside the claim. | 4.6 |
| H2051 | Artefact mimicking effect | Apparent friction reduction from a soft metallic coating is often the coating conforming to the counterface's roughness and reducing real contact area, which ends as soon as the coating wears through. A short test records the conforming phase and reports it as the coating's property. Running one specimen past coating removal, with wear depth measured, shows the step back to the substrate's friction. | 4.6 |
| H2098 | Scale argument | Under extreme pressure the film thins exponentially with pressure while the asperities flatten only in proportion to pressure over hardness, so the ratio of film to roughness falls faster than either, and extrapolating protection from low-pressure data overestimates it. The two quantities have different pressure dependences and cannot be scaled together. Measuring both at three pressures gives the two slopes and the crossing. | 4.6 |
| H2092 | Threshold | A solid lubricant's friction rises irreversibly above a contact temperature lower than its oxidation onset, because the rise is caused by its own decomposition products accumulating at the interface rather than by bulk oxidation, and those products are stickier than the parent. A temperature ramp with the interface composition measured at each step, and a second ramp in an inert carrier, separates the two temperatures. | 4.56 |
| H2063 | Count | A coated contact's friction needs at most two independent parameters, the interface's shear strength and the pressure at which it yields, so a study reporting friction varying with five surface treatments is reporting proxies for two quantities. What is measurable is not what matters. Coating one series with graded chemistry, with shear strength and yield pressure measured independently, should collapse the friction data onto one curve. | 4.53 |
| H2081 | Selection effect | Published wear maps are drawn from tests that reached steady state, and tests that seized or transferred material are excluded as not being wear, so a map's boundaries mark where the experiments stopped rather than where mechanisms changed. The excluded region is the interesting one. Testing deliberately into that region at the map's boundary conditions shows whether the boundary is a mechanism change or an apparatus limit. | 4.51 |
| H2055 | Misattribution | The scuffing resistance of a phosphated surface is credited to the phosphate's own low shear strength, but the phosphate's main function is to hold oil in the pores it creates, and its benefit disappears in a dry contact where no oil is available to hold. Phosphate without oil should then be no better than bare steel. Two phosphated specimens, one run dry and one oiled, test the attribution. | 4.49 |
| H2077 | Misattribution | The wear-in of a new machine's first hours is attributed to asperity removal, but much of the metal released in that period is manufacturing residue, loose grit and burrs and lapping compound, and the machine would settle faster if it were cleaned first. The residue's share is measurable by flushing before start-up. Two identical machines, one flushed and one not, with debris counted hourly, separate the two contributions. | 4.49 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H585 | Aliasing | A linear encoder with a 20 µm grating and 1 µm interpolation aliases grating-scale errors into a slow drift. Moving the read head by half a pitch should reverse the drift's sign. The test distinguishes an optical scale error from a thermal one. | 8.96 |
| H544 | Scale argument | Heat-affected-zone width scales with the square root of thermal diffusivity over travel speed, so doubling speed shrinks it by 1/√2 rather than by half. A weld made at twice the speed should show a zone about 71% as wide, not 50%. Measuring across a speed series tests whether the scaling is diffusional at all. | 8.8 |
| H567 | Symmetry argument | A scan strategy with rotational symmetry cannot produce net in-plane residual stress unless the part geometry breaks the symmetry. Asymmetry measured in a nominally symmetric part therefore localizes the cause to the baseplate or the gas flow. Reversing the gas-flow direction should flip the asymmetry. | 7.0 |
| H528 | Artefact mimics effect | Grain refinement measured on an etched section is often an etching artefact that reveals cell boundaries as grain boundaries. EBSD on the same fields will reveal a larger true grain size, and the discrepancy should grow with solute content. Metallography that never checks this is measuring its own etch. | 6.91 |
| H524 | Reversal | Raising pouring temperature improves fluidity up to a point and then degrades it, as oxide films and dissolved gas raise the apparent viscosity. A spiral-fluidity test at constant superheat but varying melt cleanliness should locate the reversal. If the reversal is real, 'hotter is more fluid' is a rule with an expiry temperature. | 6.82 |
| H580 | Inversion of standard practice | A long gauge is the reflex choice for CMM calibration, but short-range work is better served by a short gauge, because the long one averages the scale's periodic error into its factor. The calibration length should match the measurement length. Comparing scale factors from two gauge lengths on the same machine tests it. | 6.82 |
| H575 | Threshold | A stylus cannot detect a scratch narrower than its tip radius regardless of depth, so defect detection has a threshold set by tip geometry rather than by vertical resolution. The same scratch measured with two tip radii should appear in one and vanish in the other. It is a geometric filter, not a sensitivity limit. | 6.78 |
| H539 | Threshold | A nugget forms only when the interface reaches melting before the water-cooled electrodes remove the heat, a threshold set by the stack's thermal effusivity. The same current that welds a thin stack will not weld a thick one, and the threshold should scale with effusivity rather than with thickness alone. A stack-thickness series at fixed current tests that scaling. | 6.69 |
| H525 | Scale argument | Feeding distance is limited by the pressure drop through the mushy zone, which scales with the square of distance and inversely with permeability. Because permeability rises with the square of dendrite arm spacing, coarsening the structure should extend feeding more than enlarging the riser does. A plate cast with and without a grain refiner, at equal riser size, tests it. | 6.67 |
| H555 | Threshold | Lack of fusion appears below an energy density that depends on the ratio of hatch spacing to melt-pool width, not on energy density alone. The threshold should be crossed when hatch exceeds roughly 70% of pool width (estimate). Two builds at equal energy density but different hatch separate the two explanations. | 5.87 |
| H590 | Coincidence | When the same dimensional error appears on two unrelated parts, a common datum or a shared program is more likely than a shared material. Measuring both parts on a different machine with a different datum should separate them. Coincidence in manufacturing usually has an address. | 5.0 |
| H566 | Saturation | Recoating quality saturates with speed: above a threshold the layer is starved, below it the powder is over-compacted. The usable window narrows as layer thickness decreases. A speed ladder at two layer thicknesses, with layer density measured optically, maps the window and its narrowing. | 4.98 |
| H577 | Count | A rigid body has six degrees of freedom, so a fixture constraining more than six independent directions over-constrains the part and induces elastic distortion. Measuring the part free and fixtured should differ by exactly that distortion. The difference is a fixture property, not a part property. | 4.98 |
| H514 | Selection effect | Tool-life tests that discard inserts which fractured early report the life of the survivors' runout distribution, not of the tool grade. Recording each insert's runout before the test and re-weighting the survival data should shrink the apparent difference between grades. A difference that survives re-weighting is real. | 4.96 |
| H564 | Selection effect | Mechanical properties reported for additive materials are usually measured on coupons built alongside the part but at a different location. Sampling the part's worst location instead should lower the reported elongation. Building coupons at the top and bottom of a tall job would quantify the location effect. | 4.96 |
| H571 | Misattribution | Most CMM length error at short probe extensions is the probe bending under stylus force, though the scale takes the blame. Two stylus lengths measured at the same scale position separate the two. A longer stylus showing proportionally more error is evidence that the scale is innocent. | 4.96 |
| H574 | Bound | The uncertainty of a form measurement is bounded below by the number of independent points on the surface, not by the instrument's point-to-point noise. Adding points along one line does not reduce it. A grid of increasing areal density, analysed for convergence, shows where the bound sits. | 4.96 |
| H598 | Inversion of standard practice | Inspecting every part feels like protection, yet it can lower outgoing quality when inspectors' attention degrades and escapes rise. A lot-by-lot comparison of outgoing defect rates under sampling and under 100% inspection tests the inversion. The mechanism is attentional, and it is measurable. | 4.96 |
| H531 | Bound | A sand mould's maximum unsupported cavity is bounded by the green strength's ability to carry the metal's hydrostatic pressure, not by the sand's refractoriness. The bound scales with the cavity's smallest dimension and inversely with pour height. A stepped cavity test poured at two heights would locate it. | 4.93 |
| H552 | Symmetry argument | A butt weld between equal-thickness plates has a symmetric residual-stress field only if the weld is centred and the heat sinks are symmetric. Any asymmetry in a measured stress map therefore points to fixturing or sequencing, not to the metal. Swapping the clamping order should move the asymmetry. | 4.93 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3070 | Hidden term | Wave-in-deck loads measured on a model include the deck's trapped air, which is not Froude-scaled because air compressibility depends on absolute pressure. Venting the deck should change the load more at model scale than at full scale, testable with two vent areas. | 6.93 |
| H3053 | Scale argument | A model mooring line cannot satisfy Froude and Cauchy scaling at once, so every model test chooses one and biases the dynamic tension. Testing the same system at two model scales with the same choice should show the bias shrinking predictably. | 6.91 |
| H3056 | Interference | Thruster-hull interaction reduces effective thrust more at low advance speed, so a dynamic-positioning capacity curve is not a scaled bollard-pull curve. Thrust measured at three speeds and two headings should show the interaction changing with speed rather than with thrust alone. | 6.91 |
| H3072 | Reversal | A heave plate lowers the natural frequency by adding mass and can move the structure into a worse wave band, so a plate that reduces the resonant response in one sea state can increase it in another. A plate-size series tested in two spectra should show the ranking reversing. | 6.91 |
| H3081 | Limiting case | At zero current a strake's drag penalty vanishes but its added mass still shifts the natural frequency, so a drag-only comparison at model scale misses the frequency shift. A still-water decay test with and without strakes should show the period change. | 6.91 |
| H3075 | Scale argument | The Keulegan-Carpenter number governs the drag-inertia transition but does not fix Reynolds number, so tests at the same KC and different Re give different coefficients. Running one KC at three Re should show the coefficients converging above a threshold. | 6.87 |
| H3063 | Scale argument | VIV model tests at low Reynolds number cannot be scaled by a single factor because wake mode transitions depend on Reynolds number, so amplitude ratios between model and prototype are not constant. Two model scales at the same reduced velocity should disagree by more than the uncertainty if the dependence is real. | 6.84 |
| H3004 | Reversal | In head seas the bulb that saves resistance in calm water can add to it, because pitching changes the bulb's submergence and therefore the phase at which its wave cancels the bow wave. Two bulbs with equal calm-water gain, tested at the same wave period, should reverse their ranking if the phase argument holds. | 6.8 |
| H3038 | Symmetry argument | A symmetric hull in head seas must have zero mean yaw moment, so a measured nonzero mean indicates model misalignment or an oblique wave field rather than a physical asymmetry. A repeat with the model rotated by one hundred and eighty degrees should reverse the sign of the mean if alignment is the cause. | 6.8 |
| H3100 | Bound | No tank comparison can resolve a difference smaller than the facility's run-to-run repeatability, and that repeatability is set less by the dynamometer than by the water's temperature history over the day. Two identical runs separated by a full diurnal cycle should disagree more than two back-to-back runs, which makes the bound measurable rather than assumed. | 6.78 |
| H3005 | Scale argument | Hull roughness penalties are not Froude-scaled, so a model tested two orders of magnitude below the ship's Reynolds number cannot reproduce the roughness regime the ship lives in. The penalty has to be built from a roughness function evaluated at full-scale Reynolds number, which makes model-scale roughness comparisons a comparison of trip behaviour rather than hull condition. A geosim series spanning a factor of four in length, with one roughness applied to all, should show the penalty changing with scale in the direction the roughness function predicts. | 6.07 |
| H3031 | Hidden term | The linear roll damping coefficient from a decay contains the frequency dependence of added inertia, so a still-water decay and a forced oscillation at the same amplitude do not measure the same coefficient. Forcing at two frequencies while holding amplitude should separate the two terms. | 5.0 |
| H3050 | Derived quantity | Quasi-static and dynamic mooring stiffness differ by the line's added mass and drag, so using the static value to compute the vessel's natural period gives a period that is too long. A free-decay test in still water and a forced-oscillation test should bracket the true period. | 4.98 |
| H3064 | Count | In a sheared current the VIV response has at most a few excited modes because the shedding correlation length limits participation, so a fit with many modes is aliasing the response. Reconstructing the response from the first three modes should reproduce the measured curvature. | 4.98 |
| H3036 | Count | A linear seakeeping response at a given heading is fixed by six degrees of freedom and two wave parameters, so an empirical model with more than about eight fitted coefficients is absorbing nonlinearity or noise. A hold-out test on a second heading should expose the extra coefficients. | 4.96 |
| H3042 | Misattribution | Damping in taut polyester moorings is usually assigned to line drag, but most of the dissipation is the rope's internal hysteresis. A dry bench test at the same tension amplitude and frequency should reproduce most of the measured damping with no water present. | 4.96 |
| H3099 | Inversion of standard practice | Model tests are used to predict full-scale behaviour, yet the largest uncertainty usually sits in the full-scale environment, so a short instrumented trial can constrain the model more than another model iteration. Running prediction and trial in parallel and comparing uncertainty bands, not means, is the test. | 4.96 |
| H3030 | Misattribution | Roll damping from a small-amplitude decay test is assigned to skin friction, but the dominant term there is eddy-making at the bilge keels' roots. Fitting root fairings, which change the vortex but not the wetted area, should cut the small-amplitude damping while leaving the large-amplitude value almost unchanged. | 4.93 |
| H3058 | Hidden term | A rotating rig's apparent current has a vertical component wherever the model sits off the rig's centreline, so measured VIV amplitude includes an axial flow the analysis does not model. Offsetting the model by a known amount should change the response in a way predictable from the geometry. | 4.93 |
| H3062 | Reversal | Raising a riser's top tension lowers its static stress but can raise its VIV fatigue by moving the natural frequency into a more energetic current band. A tension sweep in a measured current profile should show fatigue reaching a minimum before rising. | 4.93 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H863 | Threshold | A ply thinner than a critical value cannot crack transversely, because its neighbours constrain the crack's opening. The threshold should rise with the matrix's fracture energy and fall with the constraint stiffness, so a tougher matrix should raise it — a thickness series in two matrices would test that. | 8.91 |
| H815 | Conservation argument | The driving force supplies three sinks — new surface, plastic work and elastic waves — while toughness computed from load and displacement counts only the first two. The overstatement should equal the acoustic fraction, a few per cent by estimate, and acoustic emission recorded during a toughness test would close the balance. | 8.84 |
| H824 | Count | Four independent parameters suffice for a fatigue life model — initiation resistance, growth coefficient, growth exponent and a threshold — and a published model with more is trading parameters for fit. Refitting several data sets with four, and counting how many extra parameters fail a held-out test, states the case numerically. | 6.87 |
| H842 | Misattribution | The Bauschinger effect is attributed to back stress from dislocation pile-ups, but part of the reverse-yield offset belongs to the load train, since grips, extensometer and load cell store elastic energy that unloads asymmetrically. Reversing the load cell's position, or moving to a much stiffer frame, should change the measured offset if that share is real. | 6.76 |
| H846 | Threshold | Below some strain nothing permanent survives unloading, and the proportional limit is the conventional place to put that threshold; if the mechanism is bowing of dislocations between pinning points, the threshold should track pinning spacing rather than yield strength. Cold work to several strain levels and measure both, spacing and threshold, to see which one follows. | 6.76 |
| H866 | Scale argument | The in-situ strength of a ply inside a laminate differs from that of the same ply tested alone, because neighbouring plies constrain its cracking. The constraint acts over a length comparable to the ply thickness, so the in-situ strength should scale with ply thickness relative to the laminate's thickness rather than with the ply's own properties. | 6.76 |
| H803 | Hidden term | Machining damage leaves a fixed-depth layer of residual tension in every specimen, and its share of measured toughness grows as the ligament shrinks. The predicted shift is layer depth times yield stress — of order ten micrometres times a few hundred megapascals for a polished sample, an estimate rather than a measurement. Electropolishing away successive surface layers at constant in-plane dimensions should then drive apparent toughness toward a width-independent value. | 6.44 |
| H817 | Hidden variable | Specimen misalignment puts a bending moment on what is supposed to be a tension test, and the resulting stress gradient raises apparent toughness by an amount that varies between laboratories using the same material. The gradient is largest where the specimen is thinnest, so the inflation should scale with the ratio of bending to axial strain rather than with load. Strain-gauging four sides of the gauge section and shimming until they agree removes a between-lab scatter normally blamed on the alloy. | 6.2 |
| H807 | Count | Three numbers exhaust the mode I crack-tip field in an isotropic body — intensity, T-stress and one length — so a toughness model carrying a fourth fitting parameter is re-describing one of them. If that is right, the fourth parameter must come out indistinguishable from zero whenever the data are clean, and it should grow only as the fit absorbs scatter. Refitting a published data set with three parameters and with four, then comparing residuals on held-out specimens, settles the matter without new testing. | 6.13 |
| H877 | Bound | Compressive strength is bounded by fibre misalignment: a fibre off-axis by an angle kinks at a stress proportional to matrix shear yield strength divided by that angle. Measured strengths exceeding this bound for the measured misalignment are measuring something else, most likely the fixture's lateral support. | 4.98 |
| H816 | Degeneracy | One fracture test cannot separate cohesive strength from toughness: raising one and lowering the other can leave the load-displacement curve almost unchanged. Only a size series breaks the degeneracy, because the two parameters scale differently with dimension. Hence a single-size fit can look excellent and still be non-unique. | 4.93 |
| H852 | Derived quantity | Uniform elongation is derived from the Considère construction on the true stress-strain curve, so it is a consequence of hardening exponent and rate sensitivity rather than an independent measurement. Two alloys quoted with different uniform elongations may differ only in rate sensitivity, which a rate-jump test would reveal. | 4.93 |
| H813 | Selection effect | Grain-boundary chemistry correlates with toughness largely because the boundaries that cracked are the ones sent for analysis, while the boundaries that resisted are rarely characterized at all. Mapping cracked and intact boundaries within one specimen converts a correlation into a test. | 4.91 |
| H856 | Degeneracy | A uniaxial stress-strain curve cannot determine the shape of the yield surface: isotropic and kinematic hardening fit the same curve while predicting different behaviour on reversal. Only a test containing a reversal, or a superposition of hydrostatic pressure, breaks the tie. | 4.91 |
| H886 | Artefact mimicking effect | Near-zero friction reported for some dry sliding pairs can come from the tribometer's own stiffness. A stiff load cell in stick-slip records the static friction and misses the slip, so the average falls far below the kinetic value. A compliant rig, or a direct high-speed force measurement, should recover the slip-phase friction. | 4.84 |
| H814 | Path dependence | Where a mixed-mode crack goes depends on how the mode mix was reached, not only on its final value, because a non-proportionally loaded zone retains damage that biases the next increment of direction. Two load paths arriving at the same intensity ratio, one rotating and one proportional, should produce measurably different deviation angles. | 4.82 |
| H896 | Hidden term | A pull-off force measured in air carries the meniscus of adsorbed water at the contact periphery. That contribution scales with humidity rather than with surface chemistry, so it can masquerade as a change in the interface. Pull-off at several humidities, same apparatus and same surfaces, separates the meniscus term from the intrinsic one. | 4.82 |
| H878 | Identity | Fibre kinking and shear banding in compression are described as separate instabilities, yet a kink band is a shear band viewed in the fibre frame, so the two should share an orientation relationship. Digital image correlation at the moment of band formation can check whether the measured band angle matches the predicted kink angle. | 4.8 |
| H838 | Degeneracy | Two surface finishes that give different fatigue curves can give identical curves once residual stress and roughness are accounted for, which would mean the finish itself is not the variable. Turn, grind and polish a specimen set, measure both quantities on each, and compare lives at matched values; anything left over is the finish's own effect. | 4.76 |
| H822 | Misattribution | Step-loading fatigue produces lives far longer than constant-amplitude testing at the final stress, and the gain is credited to coaxing, a supposed strengthening. A simpler account: most damage accumulated at the lower stresses, which the final stress then inherited. Repeating the staircase with the count of low-stress cycles fixed, varying only their amplitude, separates history from strengthening. | 4.73 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2184 | Identity | Quenched-in vacancies in a crystalline alloy and excess free volume in a metallic glass are one defect population at different packing fractions, and their annihilation should follow the same function of temperature. Measuring dilatation against annealing time in both, at the same homologous temperature, gives curves that superimpose. | 4.93 |
| H2171 | Kinetic bound | Lamellar growth stops below a critical spacing, because the diffusion fields of adjacent lamellae overlap and the interface can no longer advance, so the bound is kinetic and independent of undercooling. Directional solidification of a eutectic at a series of velocities, with spacing measured where growth ceases, locates it. | 4.91 |
| H2149 | Count | A recrystallisation texture produced by a single strain path contains at most three independent components, because the orientation space reachable by the active slip systems is bounded, so a decomposition into more components is fitting the measurement's noise. Fitting published orientation distributions with three and then four components leaves the residual unchanged. | 4.84 |
| H2161 | Threshold | A dislocation cannot climb past a particle below a critical stress, because the osmotic force from the vacancy supersaturation at the particle's interface opposes the mechanical force, and the threshold scales with the particle's misfit. Creep testing alloys of different misfit at one stress and temperature exposes a sharp transition in creep rate. | 4.82 |
| H2129 | Crossover claim | Nucleation on dislocations becomes faster than nucleation on grain boundaries only above a dislocation density, because the elastic interaction that makes a dislocation a good site is screened when the lines are sparse. Deforming a series to different strains at one ageing temperature, and locating where the dominant nucleation site changes, tests the crossing. | 4.8 |
| H2175 | Identity | Sluggish diffusion in concentrated alloys and the late onset of precipitation are one effect seen at two length scales, since the same chemical complexity that slows the jump also delays nucleation. An alloy designed for sluggish diffusion shows a longer incubation at fixed supersaturation, and the two measurements scale together across a composition series. | 4.8 |
| H2194 | Derived quantity | The Zener–Hollomon parameter absorbs temperature and strain rate into one variable, but the compensation holds only while the mechanism is unchanged, so a plot that is linear in it over a wide range indicates two mechanisms rather than one. Looking for a slope change where the mechanism should change, instead of fitting a single line, tests the assumption. | 4.71 |
| H2147 | Threshold | A critical strain exists below which recrystallisation does not occur at any temperature, and it is not a constant, since it scales with the initial grain size because the stored energy per unit boundary area needed for nucleation falls as the grain grows. Measuring the critical strain across a grain-size series at one temperature gives a line rather than a point. | 4.64 |
| H2140 | No-go claim | No thermoelastic martensite can show perfect superelasticity above the stress at which the austenite yields by slip, because slip destroys the lattice correspondence the transformation needs and the reverse path is then lost. Testing a series of orientations in one alloy, so that the resolved shear for slip varies while the transformation stress does not, shows superelasticity failing at a predictable stress. | 4.62 |
| H2102 | Hidden term | A cooling curve recorded in a casting carries the thermocouple's own response time, which for a half-millimetre bead in a metal sheath is of order a second (estimate), so any recalescence undercooling read from the curve is a lower bound. Deconvolving the bead's step response, measured in the same bath, recovers the true arrest, and the correction grows with cooling rate rather than staying constant. | 4.56 |
| H2125 | Inversion of standard practice | Two-step ageing normally starts low to nucleate and finishes high to grow, but for a system with a wide metastable gap the reversed order, high first and low second, gives a finer dispersion, because the high step fixes a large number density and the low step grows it without further nucleation. Ageing one alloy both ways to the same total time and comparing size distributions tests the inversion. | 4.49 |
| H2182 | Scale argument | Critical casting thickness and the critical cooling rate measured by calorimetry are not interchangeable, because thickness scales with the square root of the ratio of thermal diffusivity to the product of nucleation rate and growth velocity, so two alloys with equal calorimetric rates can differ in castable thickness by the ratio of their growth velocities. Casting one alloy at two thicknesses and comparing the result with its measured critical rate tests the scaling. | 4.38 |
| H2144 | Identity | Tempering of martensite and ageing of a supersaturated ferrite are one process at two carbon levels, and the sequence of carbide precipitation is continuous across the boundary that separates them in nomenclature. Comparing a low-carbon martensite tempered at 200 °C with a ferrite of the same carbon content aged at the same temperature gives the same carbide type and kinetics. | 3.67 |
| H2179 | Threshold | The face-centred to body-centred transition in concentrated substitutional alloys is set by the valence-electron concentration at which the Fermi surface touches the zone boundary rather than by the atomic size ratio, so alloys with equal size ratio and different electron count should sit on opposite sides. A composition series holding the size ratio constant while varying the electron count decides between them. | 3.62 |
| H2197 | Hidden variable | The annealing atmosphere fixes the surface's nitrogen or carbon content, so the case depth attributed to a steel's hardenability is partly the atmosphere's doing. Annealing one steel in two atmospheres of different potential, then measuring case depth by hardness traverse, separates the two contributions. | 2.98 |
| H2189 | Hidden term | Grain sizes measured by electron backscatter diffraction include the step size's rounding, so maps with a step comparable to the grain size under-report the fine fraction and over-report the mean. Re-mapping the same region at half the step changes the reported distribution if that term is significant. | 2.96 |
| H2132 | Hidden term | Martensite-start temperatures measured by dilatometry include the sample's thermal lag, which appears as an extra undercooling of a few kelvin (estimate) and differs between furnace designs. Measuring one steel in two dilatometers with different thermocouple placements, and correcting by the recorded lag, removes most of the inter-laboratory scatter. | 2.93 |
| H2196 | Identity | Shear band formation in a metallic glass and adiabatic shear band formation in a crystalline alloy share one instability criterion, in which thermal softening overtakes strain hardening, and differ only in the state variable that carries the softening. Measuring band width against thermal diffusivity in both classes gives one scaling. | 2.89 |
| H2122 | Count | Three parameters predict the yield strength of a given precipitate microstructure: mean size, volume fraction and the fraction of particles that remain coherent, so a strengthening model with more terms is folding measurement scatter into physics. Fitting published datasets with three parameters and then adding a fourth leaves the residual unchanged. | 2.87 |
| H2126 | Derived quantity | Precipitate volume fractions inferred from electrical resistivity use a rule of mixtures that fails for coherent particles, whose strain field scatters electrons independently of the particle's own resistance, so the derived fraction overestimates and the excess grows with misfit. Comparing resistivity-derived fractions with atom-probe measurements across a misfit series supplies the correction. | 2.84 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2650 | Selection effect | Emission databases are built from failures that emitted, and a crack that grew by a mechanism producing little emission leaves no record, so the apparent sensitivity of the method is a selection effect on the archive rather than a measured detection capability. The denominator is missing. Following a cohort of specimens to failure with continuous monitoring, and counting failures with and without recorded emission, supplies that denominator. | 4.51 |
| H2695 | Unit trap | Strain-based indices normalized by microstrain and by raw bridge voltage differ by the gauge factor and the bridge excitation, so a reported five percent change means different physical strains in the two conventions and thresholds do not transfer between laboratories. The normalization is rarely stated. Measuring one specimen with both a calibrated strain gauge and the raw bridge, and computing the index both ways, fixes the conversion. | 4.51 |
| H2644 | Bound | The measurable emission rate has a ceiling set by the system's dead time, and above it events overlap and are counted as one, so a rise in activity can produce a fall in the recorded rate. The peak of the recorded curve marks saturation, not the material's maximum. A controlled source pulsing at increasing rates, logged through the same acquisition chain, measures the dead time directly and locates that peak. | 4.44 |
| H2645 | Threshold | Crack advance emits detectable emission only above an increment set by the grain size and the size of the plastic zone at the tip; below it, growth is aseismic and the silence is not evidence of arrest. The threshold is a property of the microstructure as much as of the loading. Fatigue tests at two stress-intensity ranges on one alloy, with emission per cycle compared against measured striation spacing, locate it. | 4.42 |
| H2643 | Identity claim | An event and a hit are the same physical occurrence only on a single channel; once several channels are used, the count of events becomes a function of the array's detection threshold and geometry rather than of the source. The number is then a property of the instrumentation. Recording one emission source with two array geometries at matched thresholds, and comparing event rates, measures how far the count has drifted from the source. | 4.38 |
| H2611 | Unit trap | Decibels per millimetre is a one-way figure in most tables and a round-trip figure in most instruments, so a sizing correction built with the wrong convention is out by a factor of two while looking entirely plausible; the neper-to-decibel factor of 8.686 lurks in the same arithmetic. Measuring a step wedge in one setup, applying the correction both ways, and comparing each against the known step heights identifies which convention the instrument is using. | 4.36 |
| H2690 | Identity claim | If a damage index is a linear projection of a measured vector, damage and temperature change are the same transformation unless the projection is orthogonal to the temperature direction, so no threshold on the index can distinguish them. The index has no information the projection discarded. Projecting the same data onto the temperature direction and onto its complement, and comparing damage response in each, tests whether the index was ever capable of separating them. | 4.36 |
| H2618 | Energy budget | Reverberation in a thin-walled component decays mainly by leakage at the boundaries and only secondarily by absorption, so a decay rate inverted as material damping returns a number that changes when the component is clamped differently. In thick sections the two terms trade places and the inversion becomes reasonable again. The same alloy as a thin plate and a thick block, with decay measured before and after a boundary is loaded, separates leakage from absorption. | 4.33 |
| H2671 | Dimensional argument | At fixed dose, signal-to-noise in a voxel scales with the square root of the voxel's volume, so halving the voxel size in all three directions demands roughly eight times the dose to hold the same noise. (estimate) A claim of finer resolution at unchanged dose is spending contrast, and contrast is what detects the flaw. Reconstructing one acquisition at two voxel sizes, and measuring noise and contrast in each, identifies where the cost was paid. | 4.33 |
| H2651 | Unit trap | Absolute energy is reported in attojoules by one instrument family and in arbitrary units by another, and a threshold set in one is meaningless in the other because the arbitrary unit has no fixed reference to the input. A comparison across instruments then measures the instruments. Injecting a calibrated electrical pulse into both acquisition chains at a known energy fixes the conversion and settles whether the two scales differ by a constant or by a gain. | 4.31 |
| H2691 | Reversal | Adding sensors to a monitoring network improves localization up to a point and then degrades it, because more channels bring more cabling capacitance, more connectors, and more opportunity for one bad bond to inject coherent noise into every path that uses it. The optimum is finite and depends on the structure's size. Building up a network one channel at a time on one structure, and measuring localization error at each count, finds where the curve turns. | 4.31 |
| H2639 | Limiting case | As an adhesive bond becomes perfectly rigid, the interface mode's frequency should approach the free-surface value, and watching that approach is a way to measure bond stiffness without a reference. A bond that is stiff but lossy will approach the frequency limit while keeping its attenuation, so the two properties separate along the approach. Bonds cured for different times, measured repeatedly through the cure, trace the approach and show where stiffness stops changing while loss does not. | 4.29 |
| H2696 | Count | One baseline gives a differential index exactly one degree of freedom with which to absorb environmental variation, so a structure whose response depends on three independent environmental drivers cannot be corrected by it, and the residual will be dominated by whichever driver varies fastest. Correction requires as many independent references as there are drivers. Regressing the baseline's variation against measured temperature, load, and humidity, and counting how many coefficients are significant, establishes how many the index actually needs. | 4.27 |
| H2625 | Threshold | Beneath a stiffener, a disbond escapes guided-wave detection below a size set by the ratio of the stiffener's mechanical impedance to the plate's, and that ratio, not the wavelength, fixes the threshold. The stiffener carries the wave around the disbond, so the shadow only forms once the bypass path is longer than the direct one by enough to matter. Bonded plates with stiffeners and inserts of graded size, tested in one setup, put the threshold where the impedance ratio predicts. | 4.22 |
| H2628 | Scale argument | Group velocity measured over a short path is dominated by the excitation bandwidth, so the value obtained depends on the path length until the path exceeds the dispersion length of the mode. A velocity measured on a 200 mm path and applied to a 2 m path is then a different number, and the error looks like a material change. The same mode measured over three path lengths on one plate, plotted against length, marks where the value stops moving. | 4.22 |
| H2676 | Bound | Pulsed eddy current reaches only as deep as the skin depth at the lowest frequency the coil can usefully drive, and no amount of time-domain averaging extends that, because the late-time signal is the decay of a current that never went deeper. The bound is set by diffusion, not by noise. Measuring a step wedge with the coil driven at three low frequencies, and plotting the depth at which the signal vanishes against frequency, confirms the bound moves as predicted. | 4.22 |
| H2641 | Misattribution | Emission recorded during a proof test on a large frame is often dominated by the load train rather than the specimen, since every pin, shim, and hydraulic grip settles and frets under load and each of those events radiates. The events cluster at the load holds and at load reversals, which is also where real damage would be expected. Loading a dummy specimen of matched stiffness through the same schedule gives the load-train signature, and anything surviving subtraction is a candidate. | 4.2 |
| H2681 | Inversion of standard practice | Coating a surface matte black to fix its emissivity is standard, but a thin coating of unknown thickness adds a thermal resistance in series with the defect and can hide the very feature being sought, so the better practice is often to measure the surface's emissivity in place and correct, rather than to paint. The coating's resistance scales with thickness over conductivity. Comparing a painted and an unpainted region of the same disbonded panel, with emissivity measured on each, settles which approach preserves contrast. | 4.13 |
| H2694 | Scale argument | Coupon-scale validation does not transfer to a structure because attenuation, mode density, and boundary reflections all grow with size, so a damage index calibrated at 300 mm behaves differently at 3 m with no change in the physics of the damage. The index is scale-dependent even when the damage is not. Instrumenting two geometrically similar panels at different scales, with matched relative damage, demonstrates the index's scale dependence. | 3.49 |
| H2699 | Self-reference | An index normalized by the current excitation amplitude reads healthy when the amplifier's output falls and the received signal falls with it, because the ratio is preserved while the measurement degrades. The failure is silent and looks like stability. Tapping the drive signal independently and comparing it with the commanded value over months exposes the amplifier's drift, which the normalization has been hiding. | 2.6 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5936 | Hidden state in the oxide's stress | Two coupons of the same oxide thickness can oxidise at different subsequent rates, and the difference is the oxide's residual stress, which relaxes by creep on a time scale comparable with the transition time. Curvature measurements at intervals during oxidation should show the stress falling before the rate changes, which puts the stress before the rate in causal order. | 6.69 |
| H5983 | Threshold in cooling rate | Sensitisation of a weld's heat-affected zone requires the cooling rate through the carbide precipitation window to be below a threshold, above which the carbon stays in solution and the same composition resists. Three coupons cooled at three rates, then tested for intergranular attack, should show attack appearing below one rate, and the threshold should move with the alloy's carbon content. | 6.67 |
| H5988 | Reversal of a surface stress | The surface residual stress of a weld can reverse sign after a stress-relief treatment, because the treatment relaxes the through-thickness gradient faster than it relaxes the surface layer's own stress, so a surface that was compressive becomes tensile. Diffraction measurements before and after the treatment, at two depths, should show the reversal and locate the depth at which it happens. | 6.67 |
| H5918 | Hidden reservoir that discharges | Grain boundaries absorb fission gas early in life and release it later, so measured release is the difference between production and the boundaries' net filling, which makes early release lower than production and late release higher. A power ramp after a long low-power hold should produce a release transient above the instantaneous production rate, and the excess measures the reservoir that was filled. | 6.6 |
| H5967 | Ranking inverted between initiation and propagation | The ranking of alloys by resistance to stress-corrosion cracking in a slow-strain test can invert in service, because the test measures initiation in a smooth specimen while the tube fails by propagation through a cold-worked layer. A fracture-mechanics test on the same alloys in the same environment should give the opposite ranking, and the two together identify which stage the tubes have reached. | 6.6 |
| H5923 | Ranking inverted by the oxide's electronic character | Alloys ranked by autoclave corrosion resistance can invert under irradiation, because the ranking is set by the oxide's electronic conductivity rather than by its thickness. A thin oxide that conducts electrons passes oxidant to the interface, and a thick oxide that does not, blocks it. A photoelectrochemical measurement of each oxide's carrier type should predict the in-reactor ranking better than the autoclave thickness does. | 6.58 |
| H5965 | Misattributed excitation frequency | U-bend tube vibration at low flow is attributed to vortex shedding, but the dominant excitation there is the pump's pressure pulsation transmitted through the inlet. The two separate because the vibration frequency follows the pump's speed and does not scale with cross-flow velocity. A test at two pump speeds and one flow should move the frequency with the pump and not with the flow. | 6.58 |
| H5969 | Threshold of flux for consolidation | A deposit on a boiling surface consolidates and loses porosity only above a local heat flux. Below it the deposit stays porous and can be removed by the flow, so one chemistry produces either a sludge or a scale depending on local flux. A test section with a flux gradient should show the deposit's porosity changing at one position, and the position should move with total power. | 6.51 |
| H5910 | Reversal of a trend where two carriers compete | Gadolinium substitution lowers the fuel's thermal conductivity at low temperature by scattering phonons. At high temperature the phonon mean free path is already limited by intrinsic scattering while the electronic contribution grows, so the dopant's relative penalty falls and may reverse. Two dopant levels measured from room temperature to the operating range should show the curves crossing if two carriers compete, and staying parallel if the dopant only scatters. | 6.47 |
| H5966 | Bound from the tube's own sag | The span between tube supports is bounded by the tube's sag under its own weight and the two-phase drag, and beyond that bound the tube contacts the support and the wear rate rises, so widening the span to reduce pressure drop trades directly against wear. A rig with three spans and a wear measurement at the support should show wear rising sharply beyond one span, and the bound should scale with the tube's bending stiffness. | 6.33 |
| H5996 | Same amplitude loss, two causes | Probe wear and probe fouling both reduce the echo amplitude, but wear changes the probe's frequency spectrum while fouling changes only the amplitude, so the spectrum names the cause. A probe run until its amplitude falls, then cleaned and re-measured, should recover its amplitude if fouled and not if worn. | 4.89 |
| H5957 | Two sources behind one measured species | The oxygen measured in a sodium system's cover gas is the sum of the dissolved oxygen and that released from the cold trap's deposits when the trap's temperature changes, so one reading does not identify one source. Stepping the cold-trap temperature at fixed sodium flow should give two plateaus whose difference is the deposit's contribution. | 4.78 |
| H5982 | Hidden term in a magnetic phase measurement | The ferrite number of a duplex weld measured magnetically includes a contribution from deformation-induced martensite in the austenite, so the reading measures the specimen's strain history as well as its phase balance. A measurement before and after stress relief, on one specimen, should change if martensite is present and should not change if the reading is truly ferrite. | 4.71 |
| H5937 | Load-history interaction that voids a sum | The fatigue life of cladding under strain-controlled cycling depends on the order of the blocks, because hydrides reorient during the tensile half-cycle and not the compressive one. A life-fraction sum over blocks of equal amplitude is therefore invalid. Two block orders with identical amplitudes should give two lives, and the difference should track the hydride reorientation measured between blocks. | 4.69 |
| H5939 | Trade-off whose optimum moves with cycling | A chromium coating resists oxidation in proportion to its thickness, but the same thickness stores more elastic energy in the coating's thermal expansion mismatch, so the coating that survives a steady exposure is not the one that survives cycling. Two thicknesses tested isothermally and cyclically should rank oppositely, which makes a single recommended thickness meaningless without the cycle shape. | 4.69 |
| H5942 | Quantity fixed by the section plane | The thickness of the fuel-clad reaction layer is defined by a metallographic section, and the same bond measures thick or thin depending on the plane's obliquity and the etch, so comparisons between laboratories compare preparations. One bonded specimen sectioned at three angles should give three thicknesses that converge when corrected for obliquity, which tests the correction rather than the bond. | 4.67 |
| H5922 | Bound from solid solubility | The hydrogen available to build a hydride lens is bounded by the solid solubility at the local temperature, so a lens larger than that bound can only have grown by transporting hydrogen from elsewhere. Measuring the lens volume together with the surrounding concentration profile distinguishes growth by local precipitation from growth by diffusion, and the first cannot exceed the solubility bound. | 4.64 |
| H5959 | Partition that is a flow property | The distribution of iodine between liquid and steam in a boiling channel is treated as an equilibrium partition coefficient. The volatile species forms at the interface at a rate competing with the liquid's mixing, so the measured carry-over is a property of the flow. Two channel lengths at one void fraction should give two carry-over fractions if interfacial formation is rate-limiting. | 4.64 |
| H5960 | Circular definition in a calibration | The high-temperature pH reported by a glass electrode is defined by the buffers used to calibrate it, whose liquid junction potentials are assumed rather than measured, so an apparent pH excursion can be manufactured by changing the reference's concentration. Two references of different concentration, calibrated against the same buffers, should give two pH values that differ by the junction potential alone. | 4.64 |
| H5956 | Lag from a buffered inventory | The boron concentration measured downstream lags a makeup because the ion-exchange resin buffers borate, and the lag time is set by the resin's distribution coefficient and the flow rather than by the loop's residence time. A step change in makeup across two bed volumes should give two lag times in proportion to the resin's mass, which a well-mixed loop model would not. | 4.62 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3977 | Rate dependence | Tear resistance is rate-dependent because the tear tip's process zone is viscoelastic, so a pendulum test at one speed ranks materials differently from a slow tear. The crossover speed is where the process zone's relaxation time meets the test's time scale, and three speeds locate it. | 4.98 |
| H3991 | Reversal | In an accelerated shelf-life test, moisture gain and oxygen ingress accelerate by different factors, so two packages can rank one way in the test and the other way in storage. The reversal follows from the two activation energies, and a test at two conditions is the minimum that detects it. | 4.89 |
| H3915 | Conservation audit | If a laminate's measured permeability is lower than every constituent layer's permeability divided by its thickness fraction, the accounting is impossible under series resistance, since interfaces can only add resistance where they are barriers. The audit is arithmetic and costs nothing, and it catches adhesion claims that would otherwise survive to a product trial. | 4.8 |
| H3906 | Threshold | A barrier coating cannot compensate for its own defect density: below a certain thickness the transport is set by the pinholes, whose areal density belongs to the substrate's roughness and not to the coating process. The threshold is where defect density times coating permeability equals the substrate's permeability, and thinning below it changes nothing measurable. | 4.78 |
| H3910 | Artefact | A large barrier gain credited to a platelet filler is often measured against a film of a different thickness, because the filled film is opaque and its thickness is taken from a different gauge than the clear control's. A matched-thickness comparison should halve the apparent effect, and the residual is the part that tortuosity can actually explain. | 4.73 |
| H3934 | Creep claim | A seal that passes a burst test can fail in creep at a lower stress held for a long time, because the seal layer is viscoelastic and its failure envelope is time-dependent. Burst strength therefore does not rank two sealants the way a warm storage trial does, and the ranking can invert within the range of warehouse temperatures. | 4.71 |
| H3930 | Count | A seal's leak rate is usually quoted as one number, but there are at least three parallel paths, the interface, the film's own permeability, and the cut edge's wicking, and they scale differently with pressure. A single-pressure test cannot apportion them, while a pressure series can, because only the interface path is viscous and only the edge path is humidity-sensitive. | 4.67 |
| H3932 | Geometric limit | A contaminated seal fails through a channel whose conductance scales with the cube of the gap height in the viscous regime, so a particle that doubles the gap raises the leak by nearly an order of magnitude. Leak-test sensitivity is therefore not a linear measure of seal quality, and a seal that passes a coarse tracer can still hold a channel a finer one would find. | 4.53 |
| H3904 | Identity claim | The curvature of a glassy polymer's sorption isotherm and the slow tail of its permeation transient are one phenomenon seen from two sides, both being the relaxation of excess free volume toward its equilibrium occupancy. If that identity holds, a single long sorption run on a freshly quenched film should recover the permeation transient's time constant, which is a test no permeation cell can perform on its own. | 4.49 |
| H3998 | Regime change | Oxygen stored in a headspace scales with volume while ingress scales with area, so a large headspace protects by dilution early and is dominated by ingress late. The crossover time follows the volume-to-area ratio, and it is the number that should be compared when package sizes are changed. | 3.93 |
| H3971 | Reversal | A thicker package can migrate more total mass while releasing less per unit area, because it holds a larger reservoir that replenishes the depleted surface layer. The two measures therefore rank packages differently, and only one of them is the quantity a specification usually limits. | 3.0 |
| H3927 | Hysteresis claim | The first measurement after a specimen change is biased by the previous specimen's sorbed gas desorbing into the carrier, and the bias is largest when the highest-barrier specimen follows the lowest. The order of measurement is therefore a variable, and randomising it costs less than modelling it. | 2.96 |
| H3972 | Count | A migration result is a sum over at least three resistances, the polymer's diffusion, the interface's partition, and the food's own diffusion, and one test cannot separate them. Two film thicknesses and two stirring rates give four equations, after which the resistances are identifiable without any new instrument. | 2.93 |
| H3983 | Thermal mismatch | A rigid package's lid deflects in retort because lid and body expand differently, and the deflection concentrates stress at the seal's edge. The mismatch belongs to the material pair, so a lid that passes a pressure test at one temperature can fail at another with the same pressure. | 2.93 |
| H3978 | Load path | A corner drop loads the package through its stiffest path, so the corner sees a higher acceleration than a face at the same height and cushion. A test that drops only on faces therefore underestimates the demand, and the corner is the design case even when every face passes. | 2.91 |
| H3926 | Degeneracy | An oxygen rate and a water-vapour rate measured on one film cannot separate pinholes from intrinsic permeability, because both mechanisms raise both rates in the same direction. A helium leak test breaks the degeneracy, since helium's diffusivity is high enough that a defect path dominates where a polymer path does not. | 2.87 |
| H3989 | Buffer capacity | A desiccant extends shelf life by its capacity divided by the transmission rate, so once the capacity exceeds what the shelf life needs, the extension is linear in desiccant mass and independent of chemistry. That linearity is the test, and a desiccant that fails to show it is rate-limited rather than capacity-limited. | 2.84 |
| H3925 | Extrapolation error | A differential-pressure cell measures at an activity far above one atmosphere, where plasticising gases no longer follow Henry's law, so extrapolating to ambient fugacity overstates the permeability and therefore understates the barrier. The deviation is a function of fugacity rather than of pressure as such, and two fugacities are enough to detect it. | 2.82 |
| H3961 | Solubility matching | Aroma scalping is largest for compounds whose solubility parameter matches the sealant's, so a polymer chosen as a good moisture barrier can be a poor aroma barrier for the compound that matters. Polymer rankings by scalping therefore do not transfer between aroma compounds, and a single-compound test is not predictive of a mixture. | 2.82 |
| H3964 | Activation mismatch | Migration's activation energy exceeds permeation's because migration is diffusion in a matrix that the migrant itself plasticises, so an accelerated test at elevated temperature overestimates the room-temperature result by an amount that grows with migrant size. The overestimate is not a constant factor, and two temperatures are the minimum needed to detect it. | 2.82 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5045 | Saturation ceiling in strength | Tablet tensile strength is bounded by interparticle bond density, which rises with relative density and saturates once the contacts are fully formed, so no formulation can be strengthened indefinitely by pressure. The bound should appear as a plateau in strength against porosity, with the plateau level set by bond area per unit volume. Compressing a series of tablets to porosities between 5% and 40% (estimate) and locating the plateau tests it. | 8.42 |
| H5073 | Optimum with attrition | Increasing pan speed improves uniformity by increasing passes through the spray zone, but beyond a point it damages the film by attrition and impact. The optimum should move to lower speed for friable cores, whose attrition threshold is lower. Coating a friable and a robust core over a speed ladder, with uniformity and defect count measured separately, tests the reversal. | 6.67 |
| H5099 | Error cancellation | Two scale-dependent errors can cancel at one scale and not another, so a process validated at pilot scale can pass for the wrong reason. The classic pair is a longer drying time that over-dries and a lower heat transfer coefficient that under-dries. Measuring both terms separately at two scales shows whether agreement was cancellation, and perturbing one term at pilot scale should break it. | 6.58 |
| H5051 | Sign reversal with speed | For brittle materials, faster compaction can give stronger tablets, because particles fracture rather than deform and fresh surfaces bond before the load is removed. The reversal should appear as a positive slope of strength against compression speed up to some speed, turning negative once plastic deformation dominates. Compressing a brittle and a plastic grade over the same speed ladder and comparing slopes tests the sign. | 6.56 |
| H5067 | Percolation bound on barrier | The minimum film thickness that resists acid penetration is bounded by the percolation threshold of the coating's pores: below a critical thickness, pores connect through the film and the barrier fails whatever the polymer chemistry. The threshold should scale with pigment volume concentration, because pigment particles create the pore network. Coating tablets at a series of thicknesses and measuring both acid uptake and film porosity locates it. | 6.51 |
| H5010 | Count | A wet-granulation regime map has at most three independent coordinates — pore saturation, a collision-intensity measure, and a binder-to-solid ratio — so a map drawn with more axes is projecting the same space twice. Dimensional analysis supports the count, because the remaining candidate variables enter only through these groups. Fitting a growth model to data from a two-dimensional grid should show the third axis adding no independent variance. | 6.49 |
| H5098 | Regime crossing | A process can cross a regime boundary during scale-up even when every dimensionless group is held constant, because the boundary depends on a ratio that is not among the groups being matched. The crossing should appear as a discontinuity in a response that was smooth at the smaller scale. Running a scale series through the suspected boundary, with the response measured continuously, shows whether the discontinuity is real. | 6.49 |
| H5036 | Percolation threshold | Lubricant blending has a percolation threshold: below a critical surface coverage the lubricant sits in isolated patches and compaction is unaffected, and above it a continuous film forms and strength falls sharply. The threshold should appear as a break in the strength-versus-blending-time curve rather than as a gradual decline. Compacting blends at a series of lubricant loadings and times, with strength measured on one press, tests for the break. | 6.47 |
| H5035 | Counting limit | The best uniformity attainable in a blend is bounded by the number of minor-component particles in a tablet-weight of blend and by the size ratio of the components, so a blend cannot be more uniform than counting statistics allow. The bound is a counting limit, not a mixing limit. Measuring RSD at two tablet weights from the same blend should show the smaller failing at a uniformity the larger achieves. | 6.44 |
| H5007 | Drying artefact in the ranking | Size distributions compared after drying rank formulations partly by how much each binder shrinks as it loses water, so a film-forming binder can look coarser than a sugar binder when wet and finer after drying. Measuring the same batch before and after drying, by wet sieving or by in-process imaging, should show the ranking invert. The shrinkage term is computable from the binder's solids volume fraction, which makes the correction testable. | 6.42 |
| H5024 | Critical-point threshold | A drying curve has a critical moisture at which the surface stops being fully wetted, and that point is a property of the pore structure rather than of the air. Below it the rate falls, above it the rate is set by the external mass transfer coefficient. Measuring the curve at two air velocities should shift the constant-rate branch but leave the critical moisture unchanged, which separates a pore threshold from an equipment effect. | 6.36 |
| H5060 | Irreversibility | Work hardening of punch tips is irreversible, so a press that has run for years at production speed has different tooling from a new one even at the same dimensions. The change should appear as a gradual rise in the force needed to reach a given tablet thickness rather than as a dimension change. Measuring punch tip hardness and the press's force-thickness curve at intervals over a long run shows whether the two drift together. | 6.33 |
| H5016 | Lower bound on porosity | The lowest porosity a ribbon reaches at a given force is bounded by the nip's release geometry, because material that has not relaxed when the rolls separate retains elastic strain and springs back. The floor should move with the material's elastic recovery, measurable on a compacted slug in a press, not with the mill setting. Compacting one blend at a force series and measuring ribbon porosity immediately and after a day separates instantaneous from recovered porosity. | 6.31 |
| H5046 | Transition threshold | The brittle-to-ductile transition in a compact occurs where the particles' yield stress is reached before fracture, and it should show as a change in slope of a Heckel plot rather than as a change in tablet strength. The transition pressure moves with temperature and moisture, both of which lower the yield stress. Compressing the same powder at two temperatures and locating the slope change tests whether the transition is a material threshold rather than a press artefact. | 6.29 |
| H5047 | Single-cause identity | A lubricant level that lowers tablet strength almost always lowers ejection force too, which suggests both are readouts of one boundary film of magnesium stearate rather than two independent effects, since that film lowers interparticle bonding and die-wall friction together. If so, no lubricant level should lower strength without also lowering ejection force, and the two effects should scale together across lubricant grades. Compacting a lubricant ladder and measuring both responses on the same tablets tests the coupling. | 6.27 |
| H5003 | Ceiling set by breakage | The largest granule a high-shear mixer sustains is set by the balance between coalescence and breakage at the impeller tip, and at a fixed tip speed that ceiling belongs to the wet mass rather than to the blade pattern. Changing blade angle should alter how quickly the ceiling is reached, not where it lies. Running two impellers of different blade angle at the same tip speed, to the same revolution count, and sieving at the same elapsed time tests the distinction. | 6.2 |
| H5006 | Scale argument | Geometric scale-up at constant impeller tip speed preserves collision velocity but not spray flux, because liquid delivery scales with batch mass while powder surface scales with mass to the two-thirds power. The quantity to hold constant is liquid volume per unit powder surface per unit time, which forces a tenfold batch to run about 2.2 times longer at the same relative fill rather than ten times longer (estimate). Two batch sizes run at matched spray flux and at matched tip speed should differ in granule porosity only in the first case. | 4.98 |
| H5092 | Limiting-mechanism crossover | Primary drying rate against shelf temperature crosses over from heat-limited to vapour-limited behaviour, and beyond the crossover extra heat adds no flux because transport through the cake limits it. The crossover should move with cake thickness. Drying cakes of two thicknesses over a temperature ladder, finding where flux stops responding, tests the shift. | 4.82 |
| H5089 | Observer effect | Thermocouples inserted into product vials act as nucleation sites and change the freezing event they are meant to measure, so a monitored vial is not representative of the batch. The artefact should be largest for vials that would otherwise nucleate at high supercooling. Comparing product traces from instrumented and non-instrumented vials, the latter measured remotely, quantifies the bias. | 4.71 |
| H5009 | Non-monotone response | Raising impeller speed first increases granule size by promoting coalescence and then decreases it once breakage dominates. The turning point should move to higher speed as binder concentration rises, because a thicker binder film raises the granule's yield stress. A speed ladder at two binder concentrations, with size measured at fixed revolutions, tests whether the curves shift or cross. | 4.69 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H718 | Hidden term | A transimpedance measurement accounts for the feedback resistor's thermal noise and the amplifier's current noise, and habitually drops the photodiode's own shunt-conductance noise, which has the same spectral form as the first and dominates for large-area devices. At high shunt resistance the dropped term is negligible, so calibrations performed on small diodes do not transfer. Two diodes of one material and different area, read through the same amplifier, should differ in that term by exactly the area ratio. | 6.27 |
| H771 | Count | N pump frequencies generate at most N(N−1) distinct products of the form two-of-one-minus-another, so a measured spectrum containing more lines than that count has recorded cascaded products and is not a single mixing event. The count is a diagnostic that costs nothing and is rarely applied, which is why spectra with unexplained extra lines are sometimes read as new processes. Reducing the input to two pumps and counting the output lines tests whether the extra structure is cascade. | 6.27 |
| H757 | Scale argument | Edge enhancement in propagation-based phase contrast has a width that grows as the square root of the propagation distance times the wavelength, so an image taken at one distance carries a contrast that depends on the apparatus and not only on the object. Two distances invert the relation and remove the artefact, which is why a single-distance phase image should not be quoted as a material contrast. The test is to predict the second distance's image from the first and compare. | 6.2 |
| H705 | Rank argument | A pulse's spectral phase across N resolved bands has N−1 independent coefficients, so a shaper with N pixels cannot set more than N−1 of them whatever its mask resolution. Claims of arbitrary waveform synthesis are therefore claims about N, and N usually goes unreported because the shaper's pixel count is mistaken for the pulse's information content. The test is to request two target waveforms whose difference lies in the null space of the shaper's response matrix and observe that the outputs are indistinguishable. | 6.16 |
| H748 | Conservation argument | A phase-insensitive amplifier fed with a coherent input and measuring only one quadrature cannot have a noise figure below 3 dB, because the unmeasured quadrature's vacuum enters through the same inversion that provides the gain. Any report of a lower figure for such an amplifier has assigned part of the input coupling loss to the gain, or has measured a phase-sensitive configuration under a phase-insensitive name. Blocking the input and comparing the output noise with the shot-noise reference exposes which accounting was used. | 6.16 |
| H739 | Invariance argument | The nonlinear coefficient of a waveguide scales as the inverse mode area, but so do the two-photon absorption penalty and the free-carrier density generated per unit power, so a figure of merit for wavelength conversion is set by the bandgap and not by the confinement. This is why shrinking a converter improves efficiency in the low-power limit and changes nothing at the power where two-photon absorption saturates, and why reports of record efficiency cluster at low power. Measuring conversion efficiency against input power on two mode areas should produce curves that cross rather than separate. | 5.89 |
| H753 | Threshold | Coherence gating reaches a depth at which the ballistic fraction has fallen below the shot-noise floor, and because that fraction decays exponentially while the floor falls only as the square root of integration time, doubling the dynamic range buys a fixed increment of depth and not a proportional one. The increment is the transport mean free path times the logarithm of the range ratio — a scale, not a factor. Measuring the same sample at two detector gains, one decade apart, should move the maximum useful depth by a fixed amount, which is the prediction. | 5.89 |
| H712 | Bound | No cavity delivers a linewidth below the Schawlow–Townes value set by its outcoupling rate and its intracavity photon number, and at low power that bound, not the servo, is what limits a laser locked to a high-finesse reference. The bound falls with power and with length, so longer and lower-loss is the only route, and any sub-hertz claim should be checked against the photon number it implies (estimate: a few milliwatts in a metre-long cavity sits near the hertz level). Measuring the same laser at two output powers tests it, because the bound scales inversely with power while technical noise does not. | 5.73 |
| H728 | Identity | Thermally generated counts and optically generated counts leave a single-photon detector identically: same amplitude, same shape, same time tag. Only the arrival-time distribution against a known optical gate separates them, so a dark-count figure taken in continuous light measures both populations under a name that promises one. Gating the source and histogramming arrival times relative to the gate recovers the split without touching the detector. | 4.56 |
| H729 | Selection effect | Quantum efficiencies are reported for devices chosen because they are high, and the choice is invisible because the rejected devices are never counted as rejections. Process yield and best-device efficiency thereby become one number, and a "reproducible" ninety per cent can be a forty per cent yield. Measuring every device on one wafer and reporting median with maximum is the only test that distinguishes them. | 4.56 |
| H754 | Reversal | In dark-field illumination the collected signal falls when the illumination numerical aperture is raised past the collection aperture, because the geometry that excludes the specular beam also excludes part of the scattered light. Below that point the signal rises with aperture; above it, it falls. Sweeping one aperture while holding the other, on a fixed scatterer, should show the maximum and thereby measure the geometry rather than the scatterer. | 4.47 |
| H716 | Hysteresis | The threshold of a laser that has been running sits below its cold-start threshold, because the gain medium's temperature profile and the mode's thermal alignment carry the previous run's imprint. The gap exceeds shot-to-shot jitter, so a threshold quoted without a thermal history is not a material property. Measuring a cold start and then an hour at full power, with the mount held at fixed temperature, sizes the history term. | 4.44 |
| H747 | Identity | Polarization-dependent loss and birefringence are reported as separate defects, but they are two projections of one random perturbation tensor along the fibre, so a component with large birefringence and small PDL is one whose perturbation happens to be aligned. Twisting the fibre rotates the projection and should trade one for the other at fixed total perturbation. A twist series with both quantities measured at each angle tests the identity directly. | 4.44 |
| H709 | Identity | The relaxation-oscillation peak in a laser's intensity noise and the corresponding peak in its frequency noise are one resonance, coupled by the linewidth enhancement factor, not two mechanisms that happen to share a frequency. Their amplitude ratio at the peak is therefore a direct reading of that factor, needing no modulation and no separate experiment. A simultaneous amplitude-and-phase noise measurement at two bias currents should return the same factor from both. | 4.42 |
| H769 | Threshold | Self-focusing overtakes diffraction at a critical power that depends on the material and the wavelength but not on the beam's width, so a measured threshold that scales with beam size is not self-focusing at all but damage or impurity breakdown wearing its name. The invariance is the test, and it is cheap: the same material at two beam sizes. A threshold that moves with the size has already answered the question. | 4.42 |
| H725 | Derived quantity | Photon-counting efficiency is derived — quantum efficiency times collection times threshold acceptance — and it is normally fitted from a Poisson assumption about the source. If the source is not Poisson the fit buries the difference in the efficiency, which is how values slightly above the separately measured quantum efficiency appear. Calibrating against a source whose statistics are known independently, an attenuated mode-locked train with a measured photon-number distribution, tests the decomposition. | 4.4 |
| H789 | Count | A comb with N teeth across a span gives N optical frequencies but only N−1 independent differences, so a dual-comb spectrometer reporting more independent spectral points than it has teeth is interpolating, and its resolution is the tooth spacing rather than the reported point spacing. The interpolation is legitimate only when a line's width exceeds the spacing. Measuring a known two-line source at two comb spans shows whether extra points resolve anything. | 4.4 |
| H797 | Sign argument | The sign of a measured displacement in a heterodyne interferometer depends on whether the reference is taken from the upper or the lower sideband, so a laboratory that changes its driver's sideband reports the opposite sign with identical hardware. The convention is invisible until two datasets are compared. Reversing the sideband deliberately, and confirming that the magnitude is unchanged while the sign flips, documents the convention in one line and prevents the confusion. | 4.38 |
| H721 | Count | A photon-number-resolving detector built from N on-off elements returns at most N+1 distinguishable outcomes, so it can report no more than N+1 photon numbers whatever its quantum efficiency. Reports of resolving far more photons than there are elements are reports about a fit, and the fit's prior does the resolving. Feeding the detector a source whose statistics are deliberately flat over the first N+2 numbers, and watching where the reported distribution stops changing, tests it. | 4.33 |
| H786 | Hidden term | Every heterodyne phase measurement includes the acousto-optic modulator's drive phase, and that drive's temperature coefficient enters the result as a scale error rather than an offset, so it survives a zeroing procedure. The term is proportional to the drive's electrical length and therefore to the cable as much as to the crystal. Swapping two drive cables of different lengths, at fixed frequency and power, changes the measured scale by an amount that identifies the term. | 4.33 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1273 | Selection effect | Laminates peeled from the edge under-report bond quality in the middle, because edge-initiated failure selects the weakest path available at the boundary. A narrow strip cut from the centre and peeled should give a different, usually higher, value. The difference measures how much of the specification is edge preparation. | 6.91 |
| H1239 | Scale argument | The quench boundary layer grows as the square root of distance over speed, so cooling rate scales with the square root of speed and inversely with diameter; doubling speed raises cooling by only 1.41, not 2. A diameter series at two speeds, with an inline infrared temperature, tests both exponents at once. | 6.84 |
| H1237 | Threshold | Fibrillation begins above a critical draw ratio set by the ratio of amorphous orientation to crystalline bridging, and the threshold moves with quench air. A colder quench should raise it by making the skin more oriented before drawing. A draw series at two quench rates, with fibrillation counted on abraded filaments, tests the shift. | 6.8 |
| H1240 | Reversal | Higher take-up speed usually raises orientation and modulus, but past a point the surface cools so fast that a brittle skin forms and the effective draw falls, so modulus should plateau and then decline. The turnover should scale with the polymer's glass transition and the quench rate. A speed ladder with modulus and skin thickness measured together locates it. | 6.69 |
| H1282 | Count | A symmetric balanced laminate's ABD matrix has at most the independent constants its distinct ply angles allow, so a characterisation reporting more elastic constants than angle groups is measuring the test fixture. Imposing the angle-group count on the same coupon data should not worsen the fit. If it does, the extra constants are real and the count is wrong. | 6.69 |
| H1243 | Inversion of standard practice | Drawing warm is the standard route to drawability, but a fibre quenched into a skin-core structure may gain more from a cold first stage and a warm second, because the skin is oriented before the core is mobilised. The order should matter at the same total draw. Two-stage draws at reversed temperatures, with strength and skin orientation measured, test the inversion. | 6.64 |
| H1207 | Artefact mimics effect | Gels counted in extruded film are assumed to be crosslinked polymer, though a common class are unmelted resin skins that entered the feed throat late and never reached the melting zone. On that account their count should track feed-throat temperature and pellet size rather than the melt's history. Dropping cold slugs into a running machine and sectioning the resulting gels tests the attribution. | 6.6 |
| H1290 | Null result | If the ply interface is well bonded, transverse strength should not depend on stacking sequence at fixed ply angles, and a null result across sequences would rule out a constraint mechanism. Testing five sequences at one angle set, with the same fibre and resin, is a cheap way to find nothing. A difference appearing only where 90-degree plies are adjacent is the mechanism showing itself. | 6.56 |
| H1255 | Bound | Post-cure cannot raise Tg beyond the value set by the network's crosslink density, so a reported Tg above that bound is measuring the sample's thermal lag or the pan's. The same crosslink density is what the rubbery plateau modulus measures, so the two can be checked against each other. A series of post-cure times on one resin should flatten at the Tg that the modulus predicts. | 6.53 |
| H1258 | Scale argument | The thermal-runaway threshold for a curing part scales with thickness squared over diffusivity times reaction rate, so a 10 mm section that cures safely says nothing simple about 20 mm: the margin falls roughly fourfold. A thickness series in an instrumented mould, with the exotherm's peak recorded, tests the exponent. If the peak rises linearly with thickness, the mechanism is adiabatic and the squared scaling does not apply. | 6.49 |
| H1210 | Unit trap | Specific mechanical energy is quoted both in kWh per tonne and in J per gram, which differ by 3.6; a scale-up that mixes them is wrong by that factor and the error usually gets absorbed into a scale-up constant. Recomputing one published dataset in consistent units should make the constant vanish. The check costs a spreadsheet and catches an error class that survives review. | 5.58 |
| H1205 | Scale argument | Residence-time width scales with the square root of the number of flights traversed, not with L/D, so geometric scale-up at constant L/D narrows the distribution while every drawing stays identical. Axial dispersion along a longer helical path is the mechanism. Two machines of 27 and 52 mm at one L/D and one specific output, with a single tracer injection each, should show the narrower curve on the larger machine. | 5.47 |
| H1266 | Hidden term | Peel strength includes the substrate's plastic bending work, which scales with thickness cubed, so a stronger bond on a thicker film can be the same interface with more bending. Peeling one laminate at two substrate thicknesses, and correcting for the measured bending work, tests it. Without the correction, adhesion specifications compare thicknesses. | 4.84 |
| H1254 | Identity claim | Cure shrinkage from dilatometry and residual stress from curvature are one quantity seen at two constraints: a low-shrinkage resin with high modulus can generate more stress than a high-shrinkage resin with low modulus, so the two rankings need not agree. Measuring both on the same four resins should show whether the rankings cross. | 4.82 |
| H1232 | Derived quantity | Cooling time from the classic plate formula is a derived number: it assumes a fixed ejection temperature and a constant diffusivity, while the real constraint is the part's stiffness at ejection. The formula therefore gives a lower bound. Ejecting at increasing times and measuring deflection, rather than temperature, shows where the bound actually sits. | 4.8 |
| H1235 | Misattribution | Fibre strength is credited to draw ratio, yet at fixed draw ratio the quench rate sets the skin-core split and with it the strength. Two fibres drawn identically but quenched in air at different temperatures should differ. A quench-rate series at one draw ratio, with skin thickness measured by microscopy, separates the two causes. | 4.8 |
| H1241 | Artefact mimics effect | The apparent skin-core difference in fibre birefringence can be an artefact of radial averaging through a cylindrical lens, which weights the centre more than the edge. Confocal measurement or a microtomed section should reduce it. One fibre measured both ways, at the same draw ratio, shows how much of the profile belongs to the optics. | 4.78 |
| H1234 | Threshold | Below a minimum injection speed the fountain flow does not develop and the surface stays dull whatever the melt temperature, and the threshold scales with the gate's hydraulic diameter. Gloss then measures a flow regime rather than a thermal one. Gates of three diameters, each run over a speed ladder, place the threshold on a capillary number. | 4.73 |
| H1252 | Hidden term | Dielectric cure monitoring's signal contains the sensor's own electrode polarisation, which grows as the resin vitrifies, so the ion viscosity reported late in cure is partly an electrode property. Two sensors of different electrode area in one resin should diverge at the same conversion. Until that divergence is measured, dielectric cure indices near vitrification are instrument readings. | 4.73 |
| H1281 | Reversal | Adding plies usually adds strength, but at fixed thickness substituting a 90-degree ply for a zero-degree one can cut tensile strength by more than the ply's load share, because the transverse ply cracks first and seeds delamination. A laminate series with one ply rotated, tested in tension with acoustic emission, shows the loss and its onset strain. | 4.73 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2430 | Scale argument | Subsurface damage depth scales with the abrasive's penetration depth, which goes as the square root of load per particle, so doubling the load deepens damage by only about 1.4 times while widening its lateral extent by about two. Depth-only measurements therefore under-report the damaged volume. | 7.0 |
| H2417 | Bound | Deterministic figuring cannot converge below the reproducibility of its own removal function, so the achievable figure is bounded by run-to-run variation of the tool influence function rather than by metrology. Iterating a known-good part while requesting zero correction, and recording the residual that appears, measures that floor. | 4.96 |
| H2469 | Threshold | Coating adhesion fails above a deposited stress-thickness product that depends on the substrate's surface energy, so two identical substrates with different cleaning histories have thresholds differing by up to a factor of two (estimate). Paired coupons and a scratch series would test it. | 4.96 |
| H2411 | Selection effect | Reported polishing yields are biased upward because parts that fail inspection are reworked before the formal measurement, truncating the error distribution at the tolerance limit. Recording the first measurement of every part, unreworked, should move the distribution's shoulder outward and widen it. | 4.93 |
| H2416 | Identity | Edge roll-off after sub-aperture polishing and the turned edge of a machined part are the same convolution of a tool footprint with a step in the workpiece boundary. The claim is testable by checking whether the roll-off width equals the tool footprint at half maximum in both processes. | 4.93 |
| H2452 | Overconstraint | A thin mirror becomes overconstrained at a support count that depends on thickness-to-diameter ratio rather than on weight. Below the threshold, adding supports improves figure; above it, adding supports worsens it, and the crossing point is measurable by a support-count series. | 4.91 |
| H2481 | Hidden term | Every linear interferometer measurement carries a cosine error from beam misalignment that is second order in the angle, so a milliradian of misalignment gives about half a part per million of scale error. It is small but systematic, and on long-range calibration it is often the largest single term. | 4.91 |
| H2444 | Threshold | A coating changes a mirror's figure only above a stress-thickness product, below which the substrate's stiffness absorbs the stress without measurable bending. The threshold follows from plate stiffness and explains why small mirrors survive coatings that visibly warp large ones. The product, not either factor alone, is the design variable. | 4.89 |
| H2451 | Hidden term | Every touch-probe measurement carries a term from the tip radius and the calibration sphere's form error, and on curved surfaces the term enters as a scale error rather than an offset. It is therefore invisible on flats and conspicuous on spheres of small radius. Calibrating the probe on a flat therefore cannot reveal it. | 4.8 |
| H2482 | Threshold | Etch-rate measurements reveal subsurface damage only above a depth set by the etch's selectivity, and below that depth damage is present but invisible. That threshold is why a polish certified damage-free can still fail in service. | 4.8 |
| H2403 | Identity | The tool-mark band and the residual-waviness band that shops report separately after sub-aperture polishing are one continuous spectrum seen through two different instrument transfer functions. A single measurement with the transfer function stated should reproduce both bands from one power spectral density, and the claim fails where the two bands do not join at their common spatial frequency. | 4.71 |
| H2433 | Threshold | Regenerative chatter in diamond turning has a threshold set by the ratio of tool stiffness to workpiece stiffness, and below a critical ratio the process is stable at any spindle speed. A stiffness series at fixed depth of cut locates that ratio. The ratio can be measured on a static rig before any cutting, which makes the threshold cheap to check. | 4.64 |
| H2461 | Threshold | Every sub-aperture process has an edge exclusion zone wider than the tool radius, set by the tool's inability to hold pressure at the boundary. Dwelling on the edge and observing the sign of the change measures that width, and polishing inside the zone produces roll-off rather than correction. The width belongs to the tool and the edge geometry together, not to the tool alone. | 4.58 |
| H2402 | Hidden term | Preston-type removal models leave out the lap's thermal expansion, yet a pitch lap warmed by friction along the work track develops a bump that removes extra material at the track centre. Running an identical dwell schedule at two spindle speeds, matching the Preston product while changing the frictional heating, isolates the term; its magnitude should follow the measured lap temperature rise and not the nominal pressure. | 4.53 |
| H2401 | Misattribution | Astigmatic figure error that appears only after a mirror is clamped is routinely charged to the polishing run, but on ring-mounted blanks it is usually the mount's own out-of-plane flatness printed through the part. Measuring the same optic on three supports of different stiffness should settle it: the astigmatism's axis follows the mount, not the polish tool path, and the amplitude tracks support flatness rather than the dwell schedule. | 4.47 |
| H2431 | Bound | The decenter tolerance of a surface is bounded by the coma it generates per unit decenter, and that sensitivity rises with the cube of numerical aperture. A system aligned at f/4 and re-housed at f/2 therefore has one eighth of the tolerance, not one half. | 4.0 |
| H2419 | Aliasing | Running a smaller polishing tool faster to remove mid-spatial frequency error can increase that error, because the tool's footprint period aliases against the dwell grid. The reversal should occur where the dwell grid period approaches twice the tool footprint, a condition a dwell-pitch series would locate. | 3.98 |
| H2420 | Count | A rotationally symmetric system with N surfaces has a sensitivity matrix of rank at most 2N−1, so a tolerance analysis that adjusts more compensators than that is fitting metrology noise and calling it alignment. A singular-value count of the matrix settles how many adjustments are real. | 3.98 |
| H2449 | Count | RMS figure error is one number and manufacturability is not: two surfaces carrying the same RMS can need three times the polishing time, set by how many Zernike terms exceed a threshold amplitude. A term-count statistic, not RMS, should be the acceptance number. | 3.96 |
| H2405 | Threshold | Brittle materials grind ductilely below a depth-of-cut threshold governed more by the wheel's effective edge radius than by the material's fracture toughness. Two wheels of identical grit size but different bond hardness should therefore show thresholds differing by a factor of two or more, an estimate a plunge-grinding series would settle. | 3.87 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4959 | Imperfect reference ceiling | When the reference in an attribute study is a second inspector or a destructive test, the gauge's agreement with the reference is bounded above by the reference's own agreement with truth. A gauge can be perfect and still fail a study whose reference is not. | 5.0 |
| H5000 | Bottleneck bound | A line's measurement capability is bounded by the slowest traceable reference in its chain, not by the gauge at the point of measurement, so improving the gauge cannot improve the line beyond that bound. The bound can be located by tracing which reference each acceptance decision actually depends on. | 4.91 |
| H4916 | Degeneracy of two causes | A gauge's magnification change with temperature and its change with pivot wear produce the same slope error when only one master is used, so a single master cannot separate them. Two masters measured at two temperatures do separate them, because wear changes the slope near one end of the range while temperature scales the whole range. | 4.76 |
| H4921 | Constraint-state transfer | A formed part springs back when the gauge's rollers release it, so a measurement taken under constraint describes a geometry that exists only inside the fixture. Acceptance limits should be written on the state the part will occupy in service, and a gauge that measures a different state needs an explicit transfer rather than a tighter tolerance. | 4.73 |
| H4928 | Field-position bias | A vision gauge whose scale varies across its field, from a slight perspective or a stage calibration, imposes an apparent taper that reverses when the part is rotated end-for-end. A real taper does not reverse in part coordinates, so measuring the same part in two orientations separates a field-scale error from a part error without an additional standard. | 4.71 |
| H4922 | Illumination-defined dimension | The measured width of a feature depends on the illumination's angular distribution, so a diffuse dome and a coaxial light give different widths for the same part. The difference scales with the feature's sidewall angle, which makes the illumination part of the dimensional definition rather than a viewing condition. Two plants using different lights are then measuring different dimensions from identical drawings. | 4.62 |
| H4915 | Size-independent dead band | A lever gauge has a dead band at reversal set by pivot friction, and the width of that band does not scale with the displacement being measured. A linearity check across the range therefore cannot see it, while a measurement taken near the tolerance limit is dominated by it. The dead band is a property of the mechanism, not of the range it is used across. | 4.53 |
| H4906 | Tip-radius convolution | A scanning probe with a finite tip radius cannot enter a curvature tighter than its own, so a convex edge is reconstructed as an arc whose radius is the true radius plus the tip radius. The error is a constant offset rather than a scale factor, which makes it removable on a convex edge by subtracting the tip radius, while a concave feature tighter than the tip is not measured at all. | 4.4 |
| H4975 | Hierarchy inversion | A working gauge can be more stable than the reference used to calibrate it, in which case every calibration adds scatter instead of removing bias. For such a gauge the optimal chain is inverted, with the working gauge as the reference and the laboratory standard checked against it. The inversion should be checked whenever a gauge's history shows better stability than its calibration reference does. | 3.56 |
| H4943 | Absolute threshold on a relative signal | Fixed in absolute contrast, a defect threshold misses shallow defects on bright surfaces and false-calls on dark ones, because the informative signal is contrast relative to the local background. The threshold should scale with the background, and the scaling exponent is set by the sensor's noise model rather than by the defect's physics. A single global threshold is therefore a statement about the lighting as much as about the defects. | 3.44 |
| H4955 | Overlapping-average correlation | If the gauge averages a buffer of readings and the buffer overlaps between replicates, the replicates are correlated and repeatability is underestimated by a factor set by the overlap. The study's precision then depends on a firmware setting that is absent from the measurement record. | 3.0 |
| H4934 | Space mismatch in guard banding | A guard band applied symmetrically to the measured value becomes asymmetric on the part when the measurement function is curved near the limit. Applying the band in measurement space therefore over-rejects on one side and under-protects on the other, and the error grows with the curvature. | 2.98 |
| H4935 | Value-dependent test uncertainty | The conventional test-uncertainty margin treats the gauge's uncertainty as constant, but a gauge with a span error has an uncertainty that grows with distance from its calibration point. The effective margin is worst at the specification limit, which is exactly where the accept-or-reject decision is made. | 2.98 |
| H4973 | Traceability of the artefact, not the quantity | An in-line gauge calibrated against a master part is traceable to the master's geometry, but if the master's surface differs from the production surface the chain is traceable in name and not in measurand. The chain then certifies a length on a surface the line never measures. | 2.96 |
| H4982 | Orphaned value | A measurement stored without its time, temperature, part identity, and gauge identity is traceable to a gauge but not to a condition, which for acceptance is the same as being untraceable. The metadata is usually the first thing lost when systems are joined. | 2.96 |
| H4984 | Consensus by weight | In a comparison among laboratories the consensus is dominated by the laboratories claiming the smallest uncertainties, so a laboratory that is right and precise can be flagged as an outlier because the consensus is wrong. The exercise measures agreement with the majority, not agreement with the definition. | 2.96 |
| H4986 | Wrong thermal node | A painted part measured in-line has a skin temperature different from its core, and an optical gauge with shallow penetration sees the paint while a contact gauge sees the metal. The two gauges then disagree by the difference between two thermal states rather than by their accuracy. | 2.96 |
| H4992 | Environmental noise floor | An ultrasonic gauge's noise floor rises near compressed-air exhausts and other broadband sources at its operating frequency, so the same gauge has different detection limits at different points on the same line. A specification measured in a quiet room does not transfer to the line's acoustic environment. | 2.96 |
| H4957 | Assumption replacing replication | For a destructive measurement, repeatability must be replaced by a within-batch uniformity assumption, because two nominally identical specimens differ by more than the gauge's error. That assumption is rarely tested, and when it fails the study attributes specimen variation to the gauge. | 2.93 |
| H4990 | Film as part of the measurand | A wet part's optical measurement includes the coolant film, whose thickness depends on geometry and drainage time, so the gauge reports film plus part. The film contributes most where the part is flattest, which is usually where the dimensional tolerance is tightest. | 2.93 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1630 | Bound | Energy resolution cannot be better than the photoelectron Poisson limit, and the excess above it measures intrinsic non-proportionality, so resolution squared against inverse energy should be linear with an intercept set by that excess. Non-proportionality adds variance that is not counting variance. Fitting that line across several energies separates the two contributions. | 6.84 |
| H1604 | Bound | Maximum stable gas gain is bounded by the fraction of avalanche photons that return to the gas rather than by the Raether spark limit, so an opaque cathode stops the counter earlier than geometry alone predicts. Feedback photons seed secondary avalanches whose spread destabilises the discharge before the charge criterion is met. Cathodes differing only in ultraviolet reflectivity should reach measurably different gains. | 6.6 |
| H1661 | Misattribution | Build-up measured behind a shield can include scatter from the source capsule itself, which is present before the shield and does not scale with shield thickness. Capsule scatter enters the detector alongside the primary. An unshielded measurement with two capsule thicknesses separates the capsule's contribution from the shield's. | 4.93 |
| H1608 | Count | No single proportional-counter pulse constrains more than four shape parameters, so a five-parameter fit is underdetermined however densely that pulse is sampled. Drift, avalanche and ion tail are not independent generators of the shape. Fitting synthetic pulses with known parameters and inspecting the covariance matrix's rank tests the claim. | 4.91 |
| H1687 | Misattribution | Modulation transfer function degradation blamed on the detector can be the focal spot's, whose size and shape enter the same edge image and cannot be separated from it by one measurement. Both convolve the image, but only one scales with magnification. Two edge images at different magnification separate them. | 4.91 |
| H1664 | Bound | The attenuation a shield can deliver is bounded by the dose from its own activation, so beyond a certain thickness added material makes the measured dose rate rise rather than fall. Activation grows with the same fluence that attenuation removes. A thickness series with a spectrum-sensitive detector shows the minimum. | 4.89 |
| H1678 | Threshold | Ratio measurements have a rate above which the denominator's fluctuation dominates the variance, and beyond it adding counts to the numerator buys nothing. Variance contributions add in quadrature with weights set by the ratio. A rate series of one ratio, with both channels recorded, shows where the denominator takes over. | 4.89 |
| H1699 | Threshold | Self-absorption in a pressed source pellet becomes the dominant term below a certain photon energy, where the attenuation correction stops being a correction and becomes the measurement. The pellet's own mass sets the path length. Pellets of two thicknesses carrying the same activity, measured at one position, bracket the threshold. | 4.89 |
| H1665 | Threshold | In neutron shielding an energy exists below which moderation dominates and above which inelastic scattering does, and the crossover decides whether hydrogenous or high-Z material should come first. The two processes have different energy dependences. A two-layer shield measured in both orders across a range of source energies locates the crossover. | 4.87 |
| H1652 | Count | Independent calibration factors in a reference chain, chamber factor, electrometer, pressure, temperature, humidity and beam quality, bound how small a combined uncertainty can honestly be, and any quoted figure below their quadrature sum is arithmetic rather than measurement. Each factor is separately determined. Propagating each one's own uncertainty to the final dose shows the floor. | 4.78 |
| H1690 | Bound | Sharpness and noise are bounded jointly by the absorber's thickness, because making a scintillator thinner improves the point spread while raising quantum noise per unit exposure, so no design escapes the product. The two move in opposite directions with one parameter. A thickness series measured for both modulation transfer and noise shows the product is nearly constant. | 4.73 |
| H1644 | Irreversibility | Radiation damage in a semiconductor detector is not a state function of fluence, because annealing between exposures changes the damage constant that later measurements infer, so two devices given equal fluence can differ by history alone. Defect complexes form and dissolve on different timescales. Irradiating in interrupted steps, with and without annealing between them, separates fluence from history. | 4.71 |
| H1682 | Scale argument | Histogram bin width interacts with count rate: at fixed acquisition time a coarse bin averages over gain drift while a fine bin resolves it, so one dataset supports different conclusions depending on a choice that looks cosmetic. Binning is a filter. Rebinning one dataset at three widths and repeating the same test shows whether the result is bin-independent. | 4.71 |
| H1626 | Conservation | A wavelength shifter cannot add photons, only relocate them, so any gain in detected signal over the unshifted case is bounded by the ratio of detector quantum efficiency at the two wavelengths; a larger gain means the shifter is scintillating on its own. The bound follows from photon conservation. A shifter-only sample measured under identical excitation exposes the excess. | 4.69 |
| H1654 | Inversion of standard practice | For a parallel-plate chamber used in a low-energy beam, calibration in air can be more transferable than calibration at depth in water, because the phantom's own scatter changes the spectrum more than the air path does. Standard practice inverts this at high energies. Calibrating one chamber both ways, then comparing its response in a third field, tests the claim. | 4.69 |
| H1614 | Misattribution | The temperature coefficient of a sealed hygroscopic scintillator's light output is usually assigned to the crystal, but much of it can belong to the reflector, whose contact with the crystal changes as the housing expands. A pressed reflector loses optical coupling as it cools. The same crystal measured with a rigid and with a compliant reflector should give different coefficients. | 4.67 |
| H1618 | Threshold | There is an overvoltage below which afterpulsing in a silicon photomultiplier is negligible and above which it grows faster than the gain, so the best signal-to-noise sits below maximum bias. Trapped charge release scales with total avalanche charge, which rises faster with overvoltage than the gain does. Afterpulse probability and excess noise factor measured together across bias locate the knee. | 4.67 |
| H1656 | Unit trap | A neutron survey meter reading in microsieverts per hour reports a derived quantity whose conversion coefficient depends on the spectrum, so the same reading corresponds to fluences differing by an order of magnitude (estimate) between a bare source and a moderated one. The instrument applies one coefficient. Two spectra giving one reading, with fluence measured by an independent counter, expose it. | 4.64 |
| H1610 | Inversion of standard practice | Raising a tritium proportional counter's pressure to improve efficiency is self-defeating past a point, because wall-adsorbed tritium then supplies most of the count rate and the instrument measures its own contamination. The wall monolayer exchanges with the gas, so the gas-phase signal saturates while the wall term keeps growing. Filling at several pressures from one reservoir shows the rate departing from gas density. | 4.6 |
| H1670 | Inversion of standard practice | Putting the high-Z layer first is the conventional order for a laminated shield, but for a broad spectrum the order that minimises dose can be the reverse, because the low-Z layer scatters photons out of the beam before they generate secondaries in the high-Z layer. The optimum is spectrum-dependent rather than material-dependent. Both orders measured with one detector and one source settle it. | 4.6 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4539 | Precursor claim | Acoustic emission from a rolling contact should precede a detectable surface crack by an interval that scales inversely with the crack growth rate. That interval is the useful quantity, because it converts an emission threshold into a maintenance lead time and can be calibrated on a test rig with artificial defects. | 6.87 |
| H4536 | Scaling claim | Adhesion recovers after water is applied at a rate set by squeeze-out of the film, so the recovery distance scales with speed rather than with time. A train at twice the speed should need twice the distance to regain the same adhesion, which an instrumented axle with a water applicator can measure directly. | 6.82 |
| H4547 | Symmetry assumption | Carbody ride quality is often computed with a secondary suspension that damps equally in bump and rebound, but real dampers do not, and the asymmetry feeds energy into the carbody at the wheelbase frequency. Halving a damper's rebound rate should change measured ride quality more than changing the spring rate by the same nominal amount. | 6.78 |
| H4579 | Regime boundary | Traction's contribution to rail wear is piecewise, since below a threshold tractive effort wear follows the normal load and above it wear follows the tangential work, with the threshold at the point where creep enters saturation. Wear measured at a series of tractive efforts on one rail should show a kink at the threshold rather than a smooth curve. | 6.69 |
| H4591 | Error-budget claim | Position uncertainty in train odometry is dominated by wheel diameter wear rather than by tachometer resolution, because a diameter error is a scale error that accumulates over distance while tachometer error is random and averages out. A train with worn wheels should show a position error growing linearly with distance travelled rather than with its square root. Wheel diameter should then be measured on the same schedule as the tachometer is calibrated, not on a separate one. | 6.29 |
| H4560 | Turning-point claim | Hunting stability does not degrade monotonically with wheel wear, because effective conicity first falls as the tread hollows and then rises as the flange thickens. A wear series tested on one vehicle should show stability passing through a maximum rather than declining from the first millimetre. | 4.98 |
| H4572 | Recovery law | Adhesion after a water film is applied recovers with elapsed distance rather than elapsed time, because the film thins by squeeze flow under a load applied for a distance proportional to speed. A test at two speeds should therefore show two recovery times but one recovery distance. | 4.96 |
| H4524 | Amplitude-dependence claim | Normal contact stiffness measured at small amplitudes is higher than a Hertzian calculation gives, because asperities carry load before the nominal area is fully established. The stiffness should fall with amplitude toward the Hertzian value over a range of a few micrometres, an order-of-magnitude estimate rather than a specification. | 4.91 |
| H4546 | Apparatus bias | Hunting critical speed measured on a roller rig runs higher than on track because the rollers restrain the wheelset longitudinally in a way that rail does not. The offset should grow with roller diameter, so a small-roller rig and a large-roller rig carrying the same vehicle should bracket the track value. | 4.87 |
| H4576 | Saturation mechanism | Increasing axle load improves adhesion only while the third-body film's thickness stays proportional to the load, and once the film is thin enough that its hydrostatic pressure carries part of the load, extra load buys nothing. The turnover is where measured adhesion stops rising with load, and it should move with the film's viscosity. | 4.8 |
| H4573 | Overlooked cost | Friction modifiers applied for noise control reduce the traction available at the same site, and the reduction is invisible to hand-pushed friction testers because they operate at a slip the vehicle never uses. Measuring with a vehicle at realistic slip before and after application would quantify a cost that the noise case does not count. | 4.78 |
| H4506 | Threshold estimate | Below a train speed of roughly four tenths of the formation's shear wave speed, sleeper deflection should be indistinguishable from the static case, so dynamic amplification measured below that speed points to a defect rather than to soft ground. The four tenths is an order-of-magnitude estimate from the shape of the dispersion curve, not a measured constant. | 4.73 |
| H4535 | Misplaced cause | Gauge-face wear is commonly blamed on flange contact, but on the high rail of a sharp curve the tread's outer edge reaches the gauge corner before the flange does, at a pressure high enough to remove metal. Profiles measured across the high rail should show the wear centred on the tread's edge rather than on the flange root. | 4.71 |
| H4600 | Ageing-order claim | Battery internal resistance rises with age faster than capacity falls, so a battery that passes a capacity test can still fail a ride-through, because ride-through depends on the voltage sag under a short high-current load. A test applying a brief high-current pulse would rank batteries differently from a capacity test, and the ranking should match field failures better. | 4.71 |
| H4568 | Mechanism substitution | The speed dependence of adhesion is usually treated as friction falling with sliding velocity, but it is more likely the water film being thicker at higher speed. A locomotive whose wheels are polished by tread brakes should therefore show a different speed curve from one with disc brakes, and the same rail at the same speed is the test. | 4.69 |
| H4529 | Site-of-initiation claim | Squats should begin at the trailing edge of the contact rather than at its centre, because that is where the pressure gradient is steepest during braking and where the surface's residual stress is most tensile. Sections cut through squats and measured for crack origin along the contact's length would place the origin, and a centre origin would refute the claim. | 4.67 |
| H4522 | Test-rig artefact | High adhesion values from a twin-disc rig are partly an artefact of test duration, because a freshly turned disc carries no oxide or water film and the film that matters in service takes time to establish. Running the same pair of discs for hours at constant load, with the film allowed to form, should bring the coefficient down toward field values without any change of material. | 4.53 |
| H4565 | Contact-partition claim | Flange-climb derailment margin is set by friction at the flange contact rather than at the tread, and the two differ because their contact pressures differ by an order of magnitude. Lubricating the tread alone should leave the margin unchanged, while lubricating the flange should raise it sharply. Flange lubrication is then a derailment measure and not only a wear measure, which is not how it is usually justified. | 4.49 |
| H4507 | Non-monotonic claim | Tamping first loosens the ballast and then stiffens it as the crib compacts, so track stiffness passes through a minimum within days of the work. A stiffness measured immediately after tamping is therefore not the design value, and it can invert the ranking of two sites maintained to different standards. Scheduling the survey to miss the first week after tamping removes the effect without changing the maintenance at all. | 4.47 |
| H4504 | Hidden term | Rail decay rates measured by the hammer method contain a term for energy radiated into the ground through the sleepers, and that term follows sleeper contact area rather than pad damping. The same rail mounted on two sleeper widths, with identical pads and ballast, should show decay rates differing by roughly the width ratio before any damping change is invoked, and the ratio itself is an estimate rather than a measurement. | 4.42 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2744 | Scale argument | The width of an outlet distribution from a tubular unit is set by the number of effective mixing cells, which for a fixed length-to-diameter ratio does not change with scale; the absolute spread therefore grows with the square root of the cell count while the relative spread falls. Two geometrically similar units, tracer-tested at the same residence time, should share a normalised variance if cell count is the invariant. | 6.47 |
| H2790 | Conservation argument | Ignoring dissolved gas, a gas balance around a gas-liquid reactor undercounts consumption by the amount held in the liquid leaving, and at high pressure that inventory can be a substantial fraction of one pass. Depressurising the outlet liquid and measuring the released gas closes the balance. | 4.98 |
| H2776 | Derived quantity | Subtracting a clean coefficient that was itself measured once, often on another day and at another flow, is how a fouling factor is derived, so the fouling number inherits the clean measurement's error entirely. Re-measuring the clean coefficient immediately before and after a fouling run bounds the inherited error. | 4.91 |
| H2783 | Bound | Conversion per pass in a gas-liquid reactor is bounded by gas solubility at the reactor's coldest point whenever a cold zone can hold dissolved gas unreacted, not by the catalyst's turnover rate. Imposing a deliberate cold spot and measuring outlet conversion tests whether the bound moves with the cold spot's temperature. | 4.87 |
| H2779 | Hysteresis claim | The gas rate at which a packed column floods differs going up and coming down, and the loop's width measures how long the liquid film takes to reorganise rather than the packing's capacity. Cycling gas rate slowly through the transition at fixed liquid rate gives the width, and cycling faster should shrink it. | 4.82 |
| H2765 | Threshold | Soft deposits stop consolidating below a wall shear set by the deposit's yield stress, and above it consolidation proceeds at a rate nearly independent of shear. Bracketing the threshold by running one foulant at three velocities and measuring the deposit's shear strength, rather than its mass, avoids confusing the threshold with the induction period. | 4.8 |
| H2791 | Null claim | Above the velocity at which the flow becomes fully developed, sparger type has no measurable effect on the mass-transfer coefficient, because the initial bubble size is forgotten within a few column diameters. Two spargers of different hole size, run at the same high velocity with holdup measured at two heights, should return a null. | 4.8 |
| H2735 | Threshold | Particulate fouling on a coolant side stops accumulating above a wall shear at which deposition and removal rates cross, and that threshold is largely independent of particle concentration above a few parts per million. A velocity ladder at fixed concentration, weighing the deposit rather than measuring duty, gives the crossing; duty alone confuses the threshold with the induction period. | 4.69 |
| H2748 | Null claim | Above four baffles of standard width, adding more has no measurable effect on the outlet distribution of a single-phase tank, because the mean flow is already axisymmetric. Two tanks with four and eight baffles, tracer-tested at the same impeller speed, should return the same normalised variance, and a difference would point to a wall effect rather than to baffling. | 4.69 |
| H2713 | Bound | Bounded by the local stoichiometric ratio at the plume, the selectivity gain available from redistributing feed does not grow with the number of feed points: once the plume's local ratio sits at the optimum for the wanted product, adding points buys nothing. A feed-split ladder taken past the point of no improvement, with local composition measured at the plume, shows the ceiling. | 4.62 |
| H2721 | Reversal | Lowering the jacket temperature does not always lower the hottest metal temperature: when the process side can crystallize or freeze at the wall, the layer that grows adds resistance faster than the extra driving force it buys, and the peak moves up. A jacket-temperature sweep with simultaneous wall-layer thickness measurement locates the turning point, which should move with the process-side water content. | 4.62 |
| H2731 | Hysteresis claim | On a fouled surface the boiling curve follows different paths when flux is raised and when it is lowered, and the loop's width measures how much of the surface stays wetted at the lower flux. Stepping the flux up and down in one sequence, logging wall temperature at each step, gives the loop, and a clean surface should give a narrower one. | 4.62 |
| H2714 | Scale argument | The excess temperature and concentration in a feed plume scale with the plume's surface-to-volume ratio, so a geometrically similar scale-up at constant power density lengthens the plume faster than it widens it, and the large unit's plume is thin relative to its own circulation. Two sizes with matched power density and matched feed velocity ratio, probed along the plume axis, test whether the excess is conserved. | 4.53 |
| H2753 | Inversion of standard practice | Constant power per unit volume is the standard scale-up rule, but for reactions whose selectivity is decided before the feed plume disperses the invariant that matters is the ratio of circulation time to plume dispersion time, which constant power density does not hold. A two-scale pair with equal power density but different circulation times, at fixed chemistry, tests whether the rule or the ratio predicts the selectivity. | 4.51 |
| H2800 | No-go claim | No inerting test in a small vessel predicts the oxygen threshold of a large one when the small vessel's mixing time is comparable to the flame's residence time, because the threshold depends on local oxygen at the ignition point rather than on the mean. Two vessels of different size with matched mean oxygen but different mixing times, ignited at one location, would show thresholds that fail to agree. | 4.49 |
| H2757 | Bound | Impeller tip speed has an upper bound set by the onset of gas cavities that change the power draw, and past it added power no longer improves dispersion; the bound moves with gas rate, so a single number quoted for an impeller belongs to one gas rate. A tip-speed ladder at three gas rates, with power measured by torque rather than by motor current, shows the family of bounds. | 4.47 |
| H2752 | Hysteresis claim | In a trayed column with a downcomer, the froth-to-emulsion transition sits at different liquid rates depending on whether the liquid rate is being raised or lowered, and the gap between the two branches is a measure of the downcomer's inventory rather than of the tray's capacity. Stepping the liquid rate through the transition slowly at fixed gas rate gives the gap, and stepping faster should narrow it if inventory is the cause. | 4.4 |
| H2704 | Null claim | At matched local energy dissipation and matched feed position, impeller type has no measurable effect on the selectivity of a fast second-order reaction; the differences reported between impellers are differences in where the feed lands relative to the blades. Two geometrically different impellers run at equal dissipation with the feed at the same dimensionless radius should return a null, and if they do not, the matched quantity was not the one that matters. | 4.38 |
| H2701 | Misattribution | The blend time of a stirred tank is read from the disappearance of colour, but in liquids above roughly 100 mPa·s the last colourless patch vanishes by molecular diffusion through striations rather than by bulk circulation, so the number returned is a diffusion time wearing a mixing time's name (the 100 mPa·s figure is an estimate, not a measurement). Two indicators of different molecular diffusivity, at identical impeller speed and viscosity, separate the two: if the measured times differ by the ratio of diffusivities, bulk blending was already finished. | 4.02 |
| H2766 | Selection effect | Compiled fouling factors come from exchangers that were cleaned, and cleaning is scheduled when duty becomes intolerable, so the record's tail is truncated by maintenance policy rather than by physical law. Two sites with different cleaning thresholds, same duty and same water, should show different apparent distributions over the same underlying deposit. | 3.84 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5129 | Two measurements of one quantity | Microactivity conversion and riser conversion disagree systematically because the fixed bed lets cracked products re-crack during a long contact while the riser’s short contact does not. The discrepancy should grow with test contact time. That growth makes the microactivity test a ranking tool rather than a scale model. | 6.93 |
| H5173 | Unconverted-fraction term | Delayed coker coke yield contains a term from recycle heavy material that is not converted but deposited. Yield against recycle ratio therefore measures that deposit rather than the feed’s coke-forming tendency. A zero-recycle run should give the true feed yield, and the difference should scale with recycle Conradson carbon. | 6.91 |
| H5151 | Ratio-defined threshold | Below a minimum hydrogen partial pressure HDS stops following normal kinetics because the surface becomes sulfur-rich, and the threshold is set by the H2S-to-H2 ratio rather than pressure alone. A high-pressure unit with high H2S should behave like a low-pressure unit. | 5.91 |
| H5117 | Heat integration constraint | Crude preheat train inlet temperature is bounded by the fouling threshold of the hottest crude-side surface, not by furnace duty, and the bound moves with crude velocity. Raising velocity raises the threshold, which is why the same train can run hotter after a throughput increase. | 5.0 |
| H5176 | Reversal | Raising furnace outlet temperature can either raise or lower coke yield depending on whether added heat drives cracking or vaporization. The sign is set by feed Conradson carbon. A high-carbon feed should lose yield with temperature while a low-carbon feed gains it, a testable crossover. | 5.0 |
| H5121 | Yield on the wrong basis | Vacuum gas oil yield is often reported on fresh feed, but actual recovery is reduced by gas oil entrained in the residue, so the two bases diverge as the wash section degrades. The gap between them measures wash-section entrainment and should widen before overhead colour changes. | 4.98 |
| H5137 | Conservation invariant | Hydrogen in an FCC’s products plus coke must equal hydrogen in feed plus steam, and a closure gap measures unaccounted heavy material rather than analytical noise. A hydrogen balance that closes only after the heavy oil’s assay is adjusted is hiding the gap in that assay. | 4.98 |
| H5140 | False Arrhenius | Apparent activation energy of FCC cracking falls with conversion because the reaction becomes diffusion-limited inside zeolite micropores, and the curvature is a transport signature rather than a change in chemistry. A rate measured on smaller crystals should show less curvature, which is the test. | 4.98 |
| H5141 | Apparent order shift | Apparent order with respect to hydrocarbon partial pressure shifts from first to half order as acid sites saturate, so a rate law fitted at one pressure fails at another. The shift should be visible as a pressure-dependent apparent rate constant in a fixed-bed test. | 4.98 |
| H5106 | Two mechanisms conflated | Loss of tray efficiency is usually charged to poor contacting, but vapour backmixing in the downcomer and liquid crossflow maldistribution produce the same apparent efficiency. A tracer test at two liquid loads separates them, because backmixing scales with downcomer residence time while maldistribution scales with flow path. | 4.96 |
| H5130 | Measurement contaminating its own quantity | Coke measured on FCC catalyst by combustion includes carbon from the binder and hydrocarbon entrained in pores, and a single burn-off cannot separate them. A two-step burn at different oxygen concentrations should resolve the entrained fraction first, leaving catalytic coke as the difference. | 4.96 |
| H5109 | False linearity | Overhead temperature responds linearly to reflux only while the condenser is fully flooded. Once it is partly drained the response acquires a hold-up lag that a linear fit misreads as gain. A step test at two condenser inventories should show the break. | 4.93 |
| H5111 | Theoretical-stage ambiguity | Stage count cannot be inferred from terminal compositions alone, because different combinations of stages and reflux reproduce the same separation. A stage efficiency derived from terminal data is not unique, and an independent tracer residence-time measurement is needed to break the degeneracy. | 4.93 |
| H5128 | Optimum that does not coincide | Peak gasoline yield and peak octane occur at different catalyst-to-oil ratios, because octane is set by olefin retention while yield is set by over-cracking. A unit tuned for yield is not tuned for octane, and the two optima should be separated by a measurable interval on one circulation curve. | 4.91 |
| H5131 | Selectivity misread as activity | A catalyst that looks more active in a lab test may simply be more selective, preserving gasoline that a less selective catalyst would have cracked to gas. Apparent activity is then a selectivity artefact. Comparing conversion at fixed gasoline yield rather than fixed contact time should separate the two. | 4.91 |
| H5143 | Solubility limit | Coke becomes a separate phase on catalyst only above a solubility limit in the riser vapour; below it, carbon is carried as a solute and catalyst colour is not a measure of coke. The limit should move with vapour aromatic content, which is why a more aromatic feed darkens catalyst at constant coke. | 4.82 |
| H5184 | Diffusion-limited activation energy | Coke formation’s apparent activation energy falls with residence time because the reaction becomes diffusion-limited in the coke layer. A tube with a thinner initial layer should show a higher apparent activation energy. That difference is the test. | 4.82 |
| H5186 | Method disagreement | Volatile matter by proximate analysis and volatile content inferred from drum off-gas disagree because proximate analysis includes surface moisture. Vacuum-drying the sample before testing should close the gap, and the residual difference is the true analytical bias. | 4.82 |
| H5146 | Pressure drop diagnostic | Riser pressure drop is not only hydraulic; its deviation from the expected two-phase gradient measures catalyst inventory and therefore circulation rate. Calibrating pressure drop against a known circulation rate should make inventory observable without a tracer. | 4.8 |
| H5134 | Threshold | A minimum riser outlet temperature should exist below which gasoline olefins lack the thermal energy for hydrogen transfer, so olefinicity rises sharply as the unit cools through that point. The threshold is chemical rather than hydraulic and should not move with catalyst circulation rate. | 3.98 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3784 | Identity claim | Kapitza resistance at a solid-helium boundary and acoustic mismatch at a bolted joint are one phonon transmission problem at two scales, so a joint's conductance and a coating's interface conductance should collapse onto one curve against the ratio of acoustic impedances. One cryostat and two specimen types would test the collapse. | 6.87 |
| H3714 | Energy audit | Any claimed COP above the Carnot value for its own measured temperatures is an audit failure at the boundary, and the usual culprit is an air-side temperature used where a refrigerant temperature belongs. Recomputing the claim against its own stated boundary settles it without new experiments. | 4.98 |
| H3780 | Hidden term | Thermal expansion measured by dilatometry at cryogenic temperature includes the push-rod's own contraction, and below about 20 K the rod's term can exceed the sample's (estimate). The correction is not a constant but a function of the rod's history, so one room-temperature calibration does not remove it. | 4.96 |
| H3787 | Finite-size effect | At specimen sizes below the phonon mean free path the measured conductivity becomes size-dependent, so a foil's conductivity at liquid-helium temperature is not the bulk value. The discrepancy is a boundary effect rather than a materials property, and reporting it as a property is why foil data scatter. | 4.93 |
| H3724 | Impossibility | No simple single-refrigerant circuit can match a zeotropic blend's glide to a counterflow load without a composition shift somewhere in the loop, so glide matching and composition stability cannot both be had in one circuit. Designs claiming both have moved the shift into the accumulator, where it is not measured. | 4.89 |
| H3742 | Consistency requirement | A reversible unit's two coils must each be sized for the worse of their two duties, so a claim of symmetric performance implies one coil is oversized and its extra area is being paid for in fan power. Measuring both duties on the same unit shows which coil carries the penalty. | 4.87 |
| H3762 | Saturation | A cryopanel's capture coefficient for a gas falls once the adsorbed layer becomes mobile, at a coverage of roughly a monolayer (estimate). That threshold is why pumping speed measured on a fresh panel does not hold on a loaded one, and why regeneration is scheduled by coverage rather than by time. | 4.87 |
| H3796 | Duration artefact | Coatings that look frost-free in a laboratory test usually show a longer induction time rather than a different equilibrium frost layer, so a short test measures the wrong quantity. Running coated and uncoated coils until blockage, rather than for a fixed duration, reverses the ranking about as often as it confirms it. | 4.84 |
| H3701 | Misattribution | Measured cycle COP is usually debited to the compressor's isentropic efficiency, but in a laboratory rig a comparable share of the shortfall can be heat picked up by the suction line between the evaporator outlet and the compressor inlet, which raises mass flow without raising refrigerating effect. The two separate because the line gain scales with the difference between line and ambient temperature while the efficiency term does not. Instrumenting both ends of the suction line at two ambient temperatures decides it. | 4.18 |
| H3726 | Hysteresis claim | Hysteresis in a compressor map between increasing and decreasing speed is thermal rather than aerodynamic, arising from casing expansion changing tip clearance. The loop width should therefore scale with the casing's time constant and vanish on a water-cooled shell, a comparison no map currently reports. | 4.0 |
| H3741 | Lag | Coil response to a valve step lags by the refrigerant inventory's residence time, and the lag grows with charge, so a fast test underestimates the steady gain from any valve change. The underestimate is systematic and grows with the very charge that improves part-load behaviour. | 3.0 |
| H3750 | Operator artefact | Cooldown curves taken with a heater clamped to the cold tip contain the heater's heat capacity and its lead conduction, and both masquerade as a slow machine. A curve with the load applied radiatively and no heater block is a different measurement, not a repeat. | 3.0 |
| H3799 | Reversal | Raising the defrost termination temperature shortens the next frost cycle while lengthening the current defrost, and the sum of the two has a minimum. Controllers set for the shortest single defrost are usually running the longest total, which a week of logged cycles will show. | 3.0 |
| H3709 | Lag | Capacity measured at start-up lags the steady value by the evaporator's own thermal mass, and for a coil whose metal outweighs its refrigerant charge the lag is minutes rather than seconds. Short-cycle tests therefore compare machines on their thermal mass while claiming to compare their capacity. | 2.98 |
| H3768 | Degeneracy | Leak rate and outgassing rate are degenerate in a rate-of-rise test, since both produce a linear rise, and only a helium spray or a repeated pump-down separates them. A test reporting a leak rate without either is reporting a sum with an assumed split. | 2.98 |
| H3779 | Artefact | Cryogenic embrittlement is attributed to a ductile-to-brittle transition, but at liquid-helium temperature part of a specimen's lost toughness can be pre-stress from grips contracting more than the gauge length. Testing the same material with compliant and rigid grips separates the material from the fixture. | 2.98 |
| H3786 | Hysteresis claim | A composite's thermal expansion on cool-down differs from its expansion on warm-up because the matrix yields in one direction, so the loop area measures interfacial shear strength at temperature. That makes a dilatometer a mechanical tester for interfaces no other instrument can reach. | 2.96 |
| H3734 | Identifiability limit | Evaporator UA and refrigerant maldistribution cannot both be identified from one air-side traverse, since a uniformly underfed coil and a badly distributed one produce similar profiles. A second traverse at a different airflow breaks the degeneracy, so installations reporting both from one test are reporting a fitted ratio. | 2.93 |
| H3736 | Symmetry argument | Optimising a coil for cooling anti-optimises it for defrost, because both duties are set by fin spacing, so a design reporting improvement in both has changed its defrost termination criterion rather than its geometry. Running both tests on one coil, with the criterion held fixed, exposes the substitution. | 2.93 |
| H3785 | Exponent misassignment | A metal's heat capacity is electron-dominated below a few kelvin and phonon-dominated above, and the crossover moves with impurity content, so a single power-law fit across it misassigns the Debye temperature. Fitting two terms and reporting the crossover as a fitted parameter turns a nuisance into a measurement. | 2.93 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4353 | Wrong state | Warpage measured on a free part is not the warpage that matters, because the socket constrains the part and that constraint sets the stress in the joints. A specification written on the free part can pass a package that fails when it is clamped. Measuring the same part free and in its socket would give the correction, and the difference should scale with the socket's contact stiffness. | 6.51 |
| H4386 | Rate competition | A solder joint's life is the minimum of three rates, creep, intermetallic growth, and pad consumption, and the minimum changes with temperature, so a qualification at one temperature ranks alloys for that temperature alone. A ranking is therefore a statement about the test and not about the joint. Two temperatures that swap the limiting rate invert the ranking. | 4.71 |
| H4380 | Test-dependent ranking | Adding bismuth to a tin-silver-copper alloy improves thermal fatigue and worsens drop resistance, so the ranking of alloys reverses between the two tests and no alloy is best at both. The same solid-solution strengthening that helps under slow strain embrittles the joint under fast strain. One alloy series through both tests crosses at the alloy that is second-best at each. | 4.67 |
| H4349 | Process window | Vacuum-assisted moulding reduces voids up to a point and then pulls resin away from the surface, leaving a filler-rich skin and a resin-depleted layer that bonds poorly. There is therefore a vacuum window whose upper edge is a surface chemistry effect and not a flow effect. Measure void count and surface resin content together across a vacuum series and both edges appear. | 4.62 |
| H4392 | Energy criterion | The right criterion for discharge robustness is the energy delivered at the device's own failure threshold and not the passing voltage, because two devices with different current-voltage shapes can pass the same voltage with different energies. Voltage is a convenient number and a poor ranking variable, so a test that records the current waveform as well as the voltage would reorder a supplier's catalogue. | 4.58 |
| H4311 | Scaling collapse | Warpage from dies of different thickness and attaches of different thickness should collapse onto one curve when plotted against the ratio of the two, because that ratio alone sets how much of the mismatch the attach absorbs. If the points do not collapse, the attach is not the compliance in the stack and the search should move to the leadframe; the test costs three die thicknesses, three attach thicknesses, and one plot. | 4.4 |
| H4317 | Boundary-condition error | Models of die stress in an attached die usually leave the top and edges traction-free, but the moulding compound bonds there and adds a constraint that can cut the peak tensile stress by an estimated third. The error is conservative for void growth and unconservative for die cracking, because the constraint also changes the stress triaxiality. A model run with the compound removed should disagree with a measured die curvature by about the size of that term. | 4.29 |
| H4323 | Test-created artefact | Bond lift in a pull test can be manufactured by the test itself, because if the wire is stiffer than the bond the hook lifts the ball before the wire breaks and the failure is recorded as an interfacial weakness. A shear test on the same bond either finds a normal force or finds nothing, which distinguishes a weak bond from a strong one tested the wrong way. Every bond-lift rate should be checked against shear before it is believed. | 4.22 |
| H4365 | Spreading bound | There is a minimum thermal resistance per unit footprint set by spreading in the heat spreader, and below that bound added copper or added vias do not help because the heat has nowhere to go. The bound comes from the spreader's thickness and conductivity rather than from the die, so a spreader-thickness series at fixed footprint should approach it and stop. | 3.64 |
| H4364 | Inflection claim | At low power the junction temperature rises sublinearly with power because spreading improves, and at high power it rises superlinearly because the silicon's conductivity falls, so the curve has an inflection that sets the usable range. The inflection is not a failure threshold but the point where two trends exchange dominance. A power sweep with the case held at two temperatures should locate it. | 3.58 |
| H4347 | Press bound | There is a package thickness above which the mould's clamp force cannot keep the two halves closed, so the limit on thin packages becomes the press rather than the compound. The bound scales with projected area and cavity pressure, so it arrives suddenly as body size grows and should appear as a step in flash rather than as a trend on a thickness series at one press. | 3.51 |
| H4341 | Wetting threshold | There is a filler content above which a compound stops wetting the leadframe, because the resin that must reach the surface is excluded by the packing of the filler network. The threshold is governed by the filler's surface treatment rather than by loading alone, so two compounds at the same loading can sit on opposite sides. A wetting-balance series at constant loading and varying treatment should find it. | 3.49 |
| H4307 | Threshold | There is a die size above which a given attach thickness cannot absorb the silicon-to-leadframe expansion without the die seeing more tensile stress than it can take. A useful estimate is that the threshold grows with the square root of the thickness, so doubling the attach buys roughly one and a half times the die edge. A size series at fixed thickness should find the threshold as a sharp onset rather than a gradual trend. | 3.33 |
| H4361 | Load-line threshold | There is a junction temperature above which a power device's operating point becomes unstable, and the threshold is fixed by the product of the thermal resistance and the temperature coefficient of the on-state loss. The instability belongs to the load line and not to the material, so the same die can be stable in one package and not in another. A power sweep at two case temperatures should find the threshold as a fold in the voltage-current locus. | 3.27 |
| H4396 | Bake penalty | Baking a package before assembly normally improves moisture resistance, but over-baking can worsen it by oxidising the leadframe and lowering the adhesion that the moisture test later exploits. The penalty begins where the oxide's growth outruns the water's removal. A bake-time series read with adhesion and moisture sensitivity together has a minimum. | 2.84 |
| H4374 | Inherited expansion | The strain range in a solder joint includes the board's own expansion, and for small packages that term can exceed the package's, so the joint measures the board's properties as much as the package's and a board material change can dominate a package change. Two board materials with one package give the term directly. | 2.8 |
| H4328 | Measurement drift | The loop height measured optically before moulding is not the loop height in service, because the wire sags between bonding and encapsulation and the sag depends on the span and on the annealing the wire received during bonding. Two wires bonded identically at different spans therefore do not sit at the heights the machine recorded. | 2.78 |
| H4382 | Parameter count | A solder joint under thermal cycling has two independent strain components, shear and peel, so a life model with more than two independent parameters is fitting the test rather than the joint. The number of independent parameters is a property of the geometry and not of the alloy. A deliberately asymmetric pad separates the two components. | 2.76 |
| H4358 | Self-heating term | A junction temperature inferred from a forward-voltage calibration includes the self-heating of the calibration current itself, and for a small die that term is not negligible. The calibration is measured at currents low enough to seem harmless and not low enough to be isothermal. Repeating it at two sense currents and extrapolating to zero removes the term. | 2.73 |
| H4346 | Instrument lag | The mould's temperature is controlled and measured at the platen, but the cavity lags by a time comparable to the cure, so the degree of cure at ejection is set by a temperature nobody records. Cycle-time changes attributed to cooling are therefore partly cure changes. Instrument the cavity and the lag appears, with its effect on eject strength. | 2.71 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1301 | Misattribution | Recovery of an analyte from a frozen sample often appears to rise over months in storage, and the rise is put down to better extraction of an aged matrix. A likelier carrier is enzymatic release: hydrolases still active well below zero convert a bound conjugate into the free analyte, and the free fraction is what the method measures. Two aliquots stored side by side, one of them held at 95 degrees for ten minutes before freezing, should diverge by the size of the enzymatic term while the total after strong hydrolysis stays constant. | 7.93 |
| H1341 | Bound | Any improvement in a limit of detection claimed after a detector upgrade is bounded by the improvement in the blank, and the blank is usually set by the container rather than by the detector. If the blank did not move, the gain is in background suppression and should not be expected to survive a dirtier matrix. Measuring slope and blank variance on both instruments in the same matrix assigns the gain to its source and tests the bound. | 6.27 |
| H1319 | Conservation budget | Published recoveries rarely close the mass balance, and the missing fraction is usually on the vessel wall rather than lost to the atmosphere; for basic analytes in glass, a rinse of the emptied vessel and a separate injection of that rinse should recover a double-digit percentage of the spiked mass, an estimate from surface silanol capacity rather than a measurement. Reporting the rinse as part of the method converts an unaccounted loss into a measured one and costs one injection. | 6.22 |
| H1366 | Scale argument | A hollow fibre's wall thickness is a substantial fraction of its lumen, so bench-scale fibre data cannot be scaled to modules by area alone; the wall-to-lumen ratio sets the polarisation and has to be matched. Modules with the same area and different fibre dimensions should therefore give different rejection for the same feed, and the difference will be read as a chemistry difference. Two modules differing only in fibre diameter test the scaling rule before any pilot work is committed. | 6.22 |
| H1306 | Bound | Uptake by a solid-phase microextraction fibre is bounded by delivery to the fibre rather than by the coating's affinity, once the boundary layer around it is depleted. The signature would be an uptake rate identical for coatings of different thickness at the same stirring rate: if a thick and a thin fibre take up the same mass per unit time, the coating is not the limiting element. A stirring series turns the test into a slope, since the rate should follow the square root of stirring speed while the boundary layer governs. | 5.96 |
| H1353 | Hidden term | A matrix-effect factor measured by post-column infusion differs from one measured by spiking the extract after extraction, because infusion never lets the analyte equilibrate with the matrix's binding sites. The ratio of the two factors is therefore a measure of bound analyte, which otherwise stays invisible and gets folded into recovery. Running both on the same extract converts an unmeasured binding term into a number. | 4.56 |
| H1379 | Artefact mimicking effect | Apparent aggregation of nanoparticles in field-flow fractionation is often the focusing step's own concentration polarisation at the accumulation wall, which produces a second and larger apparent population. A lower focusing flow, or a shorter focusing time, should dissolve the aggregate if it is polarisation and leave it if it is real. The test is cheap and is rarely performed before a size distribution is reported. | 4.56 |
| H1391 | Misattribution | Recovery losses in automated sample preparation are charged to the sorbent, but a film retained in the pipette tip can account for a double-digit percentage of the loss at small volumes. The film is recoverable by rinsing, and that is the test: rinse the tip into a separate vial and inject it. If the rinse carries the missing mass, the sorbent was never the problem. | 4.56 |
| H1397 | Reversal | Raising the internal standard's concentration can worsen precision, because the standard's own suppression begins to affect the analyte's ionisation, reversing the usual advice to match the standard to the analyte. The optimum is below the matched concentration for analytes that co-elute with their standard, which is where the advice is most confidently given. A standard-concentration series with precision measured at each level shows the minimum. | 4.56 |
| H1326 | Unit trap | Gradient retention is quoted as the percent organic at elution, a composition measured at the detector, while the column inlet sees that composition displaced by the dwell volume. Two instruments quoting the same number therefore run different gradients, and the difference is then attributed to the column or to the chemistry. Computing inlet composition from a measured dwell volume, and re-quoting retention against it, should collapse the inter-laboratory scatter. | 4.47 |
| H1375 | Bound | Once electro-osmotic flow dominates, the resolving power of a capillary separation is bounded by the ratio of the analytes' mobility difference to the electro-osmotic mobility, and raising the voltage cannot move that bound. The bound explains why certain pairs never resolve however long the capillary is made. Measuring mobility difference and electro-osmotic mobility separately, then predicting resolution for a longer capillary, tests the bound before the capillary is bought. | 4.47 |
| H1385 | Reversal | Raising the modulation frequency increases peak capacity up to a point and then decreases it, because each modulation dilutes the first-dimension peak and the second dimension's detection limit becomes the binding constraint. The maximum is therefore a property of the detector as much as of the separation. Measuring the same separation at two detector sensitivities should move the maximum, and the direction of the movement identifies the limiting term. | 4.47 |
| H1394 | Bound | No method's uncertainty can be lower than the uncertainty of the reference material used to calibrate it, so a limit of detection reported below the calibrant's purity is an arithmetic claim rather than a measurement. The bound is rarely stated because the certificate is filed rather than read, and the reported figure is then quoted onward without it. Comparing any published uncertainty with the calibrant's certified expanded uncertainty is a one-line check. | 4.4 |
| H1365 | Bound | The rejection of a neutral solute by a tight membrane is bounded by the ratio of its diffusivity in the membrane to its diffusivity in water, and that bound does not depend on applied pressure. Pressure tests therefore cannot distinguish a leak from a low bound, which is why conductivity-based integrity tests sometimes pass a membrane that leaks. Rejection measured at two pressures with two solutes of different diffusivity separates the two cases. | 4.38 |
| H1305 | Identity | Third-phase formation and a stable emulsion are treated as different failures of an extraction, but both appear when a surface-active co-extractant accumulates at the interface past its own critical concentration. If that is one phenomenon, an emulsion-prone solvent pair should be predictable from the matrix's interfacial tension alone, without a trial extraction. Measuring interfacial tension for the candidate pairs before extraction, and checking that the ranking matches the emulsification ranking, is the discriminating test. | 4.36 |
| H1357 | Scale argument | Over the range where the spray is saturated, matrix effect scales with the total carbon load injected rather than with the concentration of any single interferent, so identity matters less than mass. That makes the effect predictable across matrices of similar carbon content and unpredictable within a matrix whose carbon varies through the run. Adding a non-interfering carbon source at matched load, and watching the signal fall by a comparable amount, tests the scaling. | 4.36 |
| H1360 | Identity | Retention time is used as an identity criterion orthogonal to mass, but matrix shifts retention by amounts comparable to the spacing between isomers, so the criterion is not independent of the sample being identified. Matrix-heavy samples, which are exactly the ones where identification matters most, are the ones where the criterion is weakest. Injecting the same extract at two dilutions and measuring how the shift scales with load separates matrix-induced shift from column-induced drift. | 4.36 |
| H1377 | Unit trap | Plate counts in field-flow fractionation are quoted per channel length or per retention time, and the two conventions rank channels of different thickness in opposite orders, while the convention itself is rarely stated. A thicker channel has more plates per unit length and fewer per unit time, so better depends on the unit chosen. Recomputing published plate counts into the other convention is enough to test whether a claimed improvement is a unit artefact. | 4.36 |
| H1325 | Scale argument | The wall region of a packed bed is roughly a fixed distance thick, so its share of the cross-section grows as particle diameter shrinks, which predicts that sub-two-micron columns are radially less homogeneous than five-micron columns of the same chemistry. The consequence is more retention scatter for early peaks on the narrower particle, not less as the plate-count comparison implies. Two particle sizes at matched bed density, with the wall region deliberately perturbed, test the prediction. | 4.31 |
| H1384 | Threshold | There is a second-dimension injection volume above which the first dimension's resolution is destroyed, and the threshold is a fraction of the second-dimension column's void volume, of the order of a tenth as an estimate. Modulator loops are chosen for sensitivity rather than for this threshold, which is why some comprehensive methods look better in the contour plot than in the reconstructed trace. A loop-volume series at fixed gradient measures the threshold rather than assuming it. | 4.31 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H137 | Threshold | Above a critical crosslinker valence a gel's modulus stops being proportional to crosslink density, because multifunctional crosslinkers cluster and only clusters beyond the percolation threshold carry stress. The threshold should appear as a kink for tri- and tetrafunctional crosslinkers and be absent for difunctional ones. A valence series at fixed concentration, with modulus and cluster size measured together, tests that. | 6.67 |
| H155 | Threshold | Above a critical membrane tension a protein's hydrophobic mismatch is accommodated by tilting the protein rather than by deforming lipids, because beyond some tension tilt becomes the cheaper mode. The threshold should depend on the protein's aspect ratio rather than on its hydrophobic length alone. Proteins of matched mismatch and different aspect ratio, reconstituted in tension-controlled vesicles, would test that. | 6.67 |
| H159 | Scale argument | Below a pore diameter of roughly two hydration shells (estimate), continuum electrostatics is undefined because the dielectric constant is not a local property, and conductance should collapse onto one curve when scaled by the hydration-shell diameter. The collapse would say the relevant variable is the number of water molecules across the pore, not the pore's charge. Sub- nanometre pores of different chemistry, measured in one apparatus, would test the collapse. | 6.44 |
| H113 | Sign trap | The electroviscous contribution to a suspension's viscosity changes sign when the double layer grows thicker than the particle radius, because the primary and secondary electroviscous terms scale differently with that ratio. A measurement that averages across the crossover can therefore report either a rise or a fall from the same physics. A salt series at fixed volume fraction, with conductivity taken alongside viscosity, should show the sign change at a predictable ratio. | 6.4 |
| H183 | Topological obstruction | An active nematic confined to a sphere cannot reach a defect-free steady state, because the Euler characteristic forces a net topological charge whatever the activity. The defect number is fixed by geometry and activity only sets their motion. Spherical active nematics at two radii, with defect counts measured, would show whether the count is quantized as predicted. | 4.73 |
| H192 | Count | A molecular machine's cycle has at most as many independent rates as it has distinct conformational states minus one, so a model with more rates is underdetermined by dwell-time data alone. The bound concerns identifiability, not mechanism. Fitting dwell-time distributions with increasing numbers of states, and asking whether the data constrain the extra rates, would test it. | 4.73 |
| H187 | Threshold | Below a critical cell size the nucleus sets the mechanical response instead of the cytoskeleton, at roughly the nuclear diameter (estimate). The prediction is that cells of one type but different ploidy, and therefore different nuclear size, cross the threshold at different cell sizes. Micropipette aspiration on cells of graded size, with nuclear size measured, would test that. | 4.71 |
| H149 | Hidden term | A synthesized gel's measured swelling pressure includes the Donnan contribution of residual counterions from synthesis, and that term depends on washing history rather than on the network. Two gels of identical composition but different washing are therefore different materials in the osmotic sense. Swelling pressure against concentration for washed and unwashed aliquots, with residual ion content assayed, would test it. | 4.67 |
| H195 | Selection effect | Single-molecule studies keep the motors that stay bound long enough to be seen, which biases dwell-time distributions toward long times, and the bias grows as the observation window shortens relative to the mean dwell. Comparing single-molecule dwell times with a bulk assay at identical conditions, with the window varied deliberately, would measure the selection rather than correct it by guesswork. | 4.67 |
| H182 | Bound | A low-Reynolds-number swimmer's efficiency is bounded by its shape's slenderness rather than by its waveform, so a sperm cell and a helical bacterium of equal slenderness share the same maximum efficiency. The bound comes from the resistance matrix's structure, which a waveform can approach but not exceed. Measured efficiencies across swimmers matched in slenderness and varied in waveform would test it. | 4.64 |
| H171 | Hidden term | A measured active stress always contains the stresslet of the swimmer's own body, which scales with shape anisotropy rather than with propulsion, so two swimmers of equal speed and different shape give different measured stresses. Only varying shape at fixed speed isolates the propulsion term. A swimmer family of graded aspect ratio at matched speed, in a stress-measuring geometry, would separate them. | 4.62 |
| H177 | Conservation audit | The stress in an active gel should balance the traction its motors exert on their surroundings, and measured stress fields generally do not balance once the motor-density gradient is included. That gradient is rarely imaged, so the missing term gets absorbed into a phenomenological activity parameter. Imaging motor density and stress simultaneously in one sample would show whether the balance closes without it. | 4.6 |
| H146 | Selection effect | Distributions of self-healing efficiency are conditioned on samples that survived the cut and the rejoin, so gels that fail catastrophically never enter the statistics and the reported mean is biased upward. The bias should grow with the severity of the test, which makes it measurable rather than merely acknowledged. An experiment that records every attempt on one recipe, failures included, would size the bias. | 4.58 |
| H167 | Unit trap | Permeability coefficients circulate in both centimetres per second and metres per second, and the factor of a hundred occasionally gets absorbed into a claimed anomalous resistance. Two studies disagreeing by roughly two orders of magnitude should be checked for that conversion before any physical explanation is attempted. The test is bookkeeping against the method sections, and it is the cheapest test on this page. | 4.58 |
| H174 | Count | An active nematic has at most three independent elastic constants once activity is allowed to renormalize them, so a measurement reporting four has folded a flow-alignment term into one of the three. The count is bookkeeping about which deformations cost energy independently. Fitting the same defect structures with three- and four-constant models, and comparing residuals, shows whether the fourth is required or merely absorbed. | 4.58 |
| H178 | Threshold | Traction forces stop rising with substrate stiffness above a critical value that is set by adhesion-cluster size rather than by cytoskeletal stiffness. A cluster transmits force only up to the load at which it ruptures, so the plateau should move with cluster size and not with myosin activity. Cells with clusters of different size on stiffness gradients, read by traction microscopy, would test that. | 4.58 |
| H142 | Time-scale mismatch | A reversible covalent gel is reversible only on the timescale of its exchange reaction, so an activation energy from rheology equals the exchange energy only when the network relaxes faster than exchange. Push the stress up and the same measurement starts reporting bond rupture, with a different apparent energy. Activation energies measured at several stresses, set against a spectroscopic exchange rate, would locate the crossover. | 4.56 |
| H170 | Reversal | More activity can mean a colder tracer: at high activity the swimmers' collisions become correlated and a passive tracer turns subdiffusive, so the effective temperature read from its long-time diffusion falls even as the bath's energy input rises. The non-monotonicity should survive at fixed tracer size. Tracers of two sizes in baths of graded activity, with short- and long-time diffusion both measured, would show it. | 4.56 |
| H168 | Misattribution | Giant number fluctuations are assigned to activity itself, but in a dense suspension the same exponent can arise from a broad distribution of swimming speeds, which produces the same variance without any collective effect. The two are distinguishable because a population engineered to one speed should show a different exponent at the same density. Speed-sorted swimmers, with the exponent measured across sorted fractions, would test it. | 4.53 |
| H172 | Scale argument | Active turbulence has a characteristic length set by the swimmer's length only when confinement exceeds a threshold ratio of system size to swimmer length; below it, the pattern wavelength is set by the box. This predicts that the same suspension in two cells shows two vortex sizes, which is a statement about boundary conditions rather than biology. A confinement series with in-situ imaging would test it. | 4.53 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H5837 | Competition of two timescales | Shielding of a penetration electric field takes the time needed for the region-2 current to build, so its duration should track the Alfven transit time from the equatorial plane to the ionosphere rather than the duration of the solar-wind driver. Two events with the same driver duration but different plasma density should shield on different timescales. Comparing penetration events with the ambient density inferred from a magnetometer chain's ULF response would test the competition. | 8.36 |
| H5885 | Coupling into an engineered network | Track circuits in railway signalling respond to telluric currents at frequencies below the signalling band, so geomagnetic disturbances should appear as false occupancies whose rate scales with the electric field's rate of change rather than with its magnitude. The coupling is inductive through the rails. Comparing false-occupancy logs with the electric field's derivative would test the scaling. | 6.73 |
| H5838 | Coupling to a secondary population | Cold ion outflow should accompany a subauroral polarization stream, with a flux that scales with the stream's electric field, because the same field that drives the stream also lifts the ions. The outflow is a consequence, not an independent phenomenon to be explained separately. Measuring the outflow above a stream with a radar and the stream's field with a magnetometer chain would test the scaling. | 6.56 |
| H5820 | Resonance condition met intermittently | Pi2 pulsations are usually treated as a cavity resonance excited at onset, but the cavity's frequency depends on the plasmasphere's size, so the same broadband onset impulse should ring at different periods as the plasmasphere erodes through a storm. A fixed period across a storm would argue against a cavity mode. Tracking Pi2 period against an independently measured plasmapause location through one storm separates the two. | 6.53 |
| H5843 | Threshold in a distribution's shape | Chorus growth requires an electron anisotropy above a threshold that depends on the ratio of plasma frequency to gyrofrequency, so the same anisotropy should produce chorus at one altitude and not another. The altitude dependence comes from the density gradient along the field line. Comparing chorus observations at two spacecraft separated in altitude along the same field line, during the same event, would test the threshold. | 6.53 |
| H5842 | Consistency between two independent measurements | The ring current's pressure inferred from the ground magnetic field and that inferred from energetic neutral atoms must agree if both are correctly calibrated, and any disagreement should scale with the assumed current geometry rather than with the pressure itself. Ground-based estimates assume a symmetric current, while the neutral-atom measurement does not. Comparing the two through the recovery phase, when asymmetry is small, isolates the geometry error. | 6.51 |
| H5827 | Energy conversion bounded by geometry | Flow braking in the near-Earth tail converts kinetic energy into field-aligned currents, and the conversion efficiency should be bounded by the ratio of the braking region's transverse size to the flow's inertial length; that bound is a property of the plasma, not of the driver. Measuring current density and braking length across several events would place the efficiency below the bound in every case and near it in the strongest flows. | 6.42 |
| H5828 | Budget that should close | The mass carried away in plasmoids should account for the mass lost from the closed tail during the same interval, and the budget is rarely checked because plasmoid mass is poorly measured. A systematic excess of plasmoid mass over the closed-tail loss would point to a source inside the tail rather than at its boundary. Comparing the two over a series of substorms with reliable plasmoid crossings would make the budget diagnostic. | 6.4 |
| H5803 | Threshold in a geometric angle | Magnetosheath jets are usually counted as a high-Mach-number phenomenon, yet their occurrence should be governed mainly by the angle between the interplanetary field and the bow-shock normal, since that angle decides whether the shock is quasi-parallel and can transmit coherent structures downstream. Jet counts at fixed solar-wind speed across a range of cone angles would test this. If the angle controls occurrence, the jet rate will collapse at small cone angles without any change in Mach number. | 6.29 |
| H5802 | Bound from total pressure balance | The standoff distance of the dayside magnetopause is set by the solar-wind ram pressure plus the magnetosphere's internal pressure, so a boundary that reaches geosynchronous orbit during a storm should require an inflated ring current as well as strong ram pressure; the bound is one of total pressure balance, not of ram pressure alone. Sorting deep boundary crossings by the ring current's simultaneously measured pressure would place them all at high internal pressure, while ram-pressure-matched crossings without it would stop farther out. | 6.18 |
| H5866 | Structure set by a gradient | A field-aligned current sheet's width is set by the transverse gradient of the ionospheric conductance rather than by the current's magnitude, so a narrow sheet can carry a small current. The width scales inversely with the gradient. Comparing sheet widths from a radar's conductance map with those from a magnetometer inversion would test the scaling. | 4.78 |
| H5886 | Voltage that integrates a field | Submarine cable voltages integrate the electric field over thousands of kilometres, so they measure the field's large-scale average rather than its local value, and a cable can show a signal when coastal magnetometers show none; the integration is a spatial filter. Comparing cable voltages with an average of coastal magnetometers along the route would test the filter. | 4.73 |
| H5899 | Slow signal a baseline cannot see | A secular change of a few nanoteslas per year, an estimate, is smaller than the baseline's uncertainty between absolute measurements, so the change can be distributed arbitrarily between the baseline and the record; the ambiguity is a property of the differencing, not of the field. Comparing records from two observatories with independent absolute schedules would bound the ambiguity. | 4.71 |
| H5870 | Threshold for an instability | Equatorial spread F requires a seed perturbation whose amplitude exceeds a threshold set by the background density gradient, so the same gradient can be stable on one night and unstable on the next according to the seed. The seed comes from gravity waves and from the evening terminator's gradient. Comparing nights with matched gradients but different gravity-wave activity would test the threshold. | 4.62 |
| H5833 | Composition control of a timescale | The recovery rate of the ring current depends less on the solar-wind driver than on the oxygen fraction of the ring current, because oxygen charge-exchanges more slowly than hydrogen at the same energy. A storm with high oxygen content should therefore recover more slowly at matched pressure. Comparing recovery times for storms with composition measured by a passing spacecraft would test the ordering. | 4.6 |
| H5850 | Budget of precipitation against a reservoir | Microburst precipitation removes electrons from the outer belt at a rate that should be detectable as a decrease of the trapped population when the bursts are summed over a drift orbit. The budget is rarely tested because the bursts are measured at one point. A spacecraft in a drifting orbit that crosses a fixed microburst region repeatedly would show the trapped flux decreasing after each crossing. | 4.53 |
| H5814 | Selection bias in a crossing census | Reports that the magnetopause is usually found near a particular standoff distance inherit the orbits of the spacecraft that found it, because crossings are recorded where spacecraft linger. A census weighted by dwell time rather than by count would shift the reported mean outward. Recomputing the distribution with each crossing weighted by the time spent in that radial bin is a test that needs no new data. | 4.51 |
| H5823 | Hysteresis in a loading cycle | The relation between tail stretching and interplanetary driving should be hysteretic, because the tail remains stretched after the driving stops and the same driving level then corresponds to two different configurations on the way in and on the way out. The mechanism is the finite lifetime of the closed flux against reconnection. Plotting stretching against driving for a full loading-unloading cycle would show a loop whose width measures that lifetime. | 4.44 |
| H5844 | Two names for one loss process | Hiss-driven loss and the formation of the slot region are usually described as separate phenomena, but the slot is where the hiss loss timescale is shorter than the radial diffusion timescale. The identity means that any change in the slot's depth must be accompanied by a change in local hiss power, not only in the source population. Comparing slot depth with measured hiss power over a season would test the identity. | 4.42 |
| H5801 | Misattribution of a measured decay | Ring-current decay rates measured after a storm's minimum are usually attributed to charge exchange with geocoronal hydrogen, yet a substantial part of the apparent decay is adiabatic: as the magnetosphere deflates, drifting particles move to weaker field and their energy falls. The two processes differ in their dependence on solar-wind dynamic pressure, which sets the inflation, while charge exchange depends on density and species. Recomputing decay during two moderate storms with the same pressure history but different oxygen content of the ring current would separate them, because charge exchange scales with composition and the adiabatic term does not. | 3.82 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4882 | Reversal after densification | Softer foams should transmit less force at low impact energy and more at high energy. They bottom out and transmit the remainder through a densified layer, so the ranking of two foams reverses with energy. Drop tests across an energy series on both foams, with transmitted force measured, show the crossover. | 6.87 |
| H4862 | Series cap | Shoe and track behave as springs in series, so the track’s energy return is capped by the shoe’s. The softer element dominates the series compliance, which means a track can only help if it is softer than the shoe it is paired with. Measure the pair’s rebound and each element’s separately; a stiff shoe should make the track’s stiffness irrelevant. | 6.67 |
| H4868 | Angle dependence | Court restitution should depend on the ball’s inbound angle, not only on its speed. The court’s deformation and the ball’s slip at contact change with the angle, while the usual vertical drop test fixes the angle at ninety degrees and reports a single number. An angled drop rig on the same court, at several angles and speeds, shows the coefficient varying. | 6.64 |
| H4867 | Threshold owned by the instrument | A false-start threshold should be set partly by the block’s own dynamics. A compliant block can register a force transient from the athlete’s setup and trigger a decision the athlete did not cause, because the instrument sits inside the measurement it polices. Apply known impulses to rigid and compliant blocks, with no athlete present; a threshold that moves belongs to the hardware. | 6.62 |
| H4813 | Compensation between terms | Swing should fall more slowly with altitude than carry does. The seam’s effectiveness depends on boundary-layer thickness, and the transition point shifts as density falls, partly compensating for the smaller dynamic pressure; carry has no such compensation. Test one ball at two air densities in a pressure tunnel, matched on speed: the side-force coefficient should rise as density falls while lift does not. | 6.6 |
| H4829 | Null for overall roughness | Overall surface roughness should not change the direction of a cricket ball’s swing. The seam fixes the asymmetric separation, while roughness only sets how sharply the layer responds, so polishing one side and roughening the other should change magnitude without reversing sign. Prepare balls with four roughness combinations, test at one speed and seam angle, and the sign should stay with the seam. | 6.6 |
| H4865 | Anisotropy from manufacture | Calendering should leave a track stiffer along the direction of manufacture than across it. The polymer chains align with the calender; a landing angled slightly off the lane returns a different fraction of its energy than one straight down the lane. Rotate a drop test in increments on the same surface; the difference, and its reversal on a differently laid track, is the anisotropy. | 6.58 |
| H4881 | Threshold in the interface | Sliding interfaces in helmets should reduce rotational acceleration only while friction stays below a threshold. Above it the shell sticks to the head and the interface does nothing; the threshold is set by the ratio of interface friction to the foam’s shear resistance. Strike a helmet obliquely at a series of interface frictions, with rotational acceleration measured; the reduction should vanish at the threshold. | 6.58 |
| H4847 | Ratio, not magnitude | Blade rebound should depend on the ratio of contact time to the blade’s natural period, not on stiffness alone. That period scales as the square root of the stiffness-to-mass ratio, so two blades of equal stiffness and different mass rank differently as the ball’s inbound speed changes. Measure rebound at several speeds on blades chosen to straddle the ratio; the ordering should follow the time ratio. | 6.53 |
| H4846 | Solvent as the active ingredient | Speed glue works by solvent swelling that lowers the rubber’s modulus and moves its loss peak. The effect should decay as the solvent leaves and should be absent if the rubber is pre-swollen to the same extent; the solvent is part of the measurement’s timescale, not a permanent treatment. Track mass and restitution against time after gluing; the two curves should fall together as the solvent leaves. | 6.51 |
| H4850 | Storage that outlasts contact | Thin barrels should be slower at high impact speed even though they are faster at low speed. The barrel’s ovalisation takes longer to return than the ball stays in contact, and the barrel keeps energy the ball never receives; the reversal is a timing failure rather than a stiffness failure. A barrel-thickness series tested at two speeds, with the barrel filmed during contact, should show the reversal. | 6.51 |
| H4854 | Perception tracking uniformity | The point players call the sweet spot should be where the stringbed’s contact pressure is most uniform rather than where rebound is greatest. The hand senses the frame’s vibration pattern, and a uniform pressure excites it least; a denser pattern can move the pressure centre without moving the rebound maximum. Pressure film at the contact patch, compared with a rebound map on the same frame, separates the two points. | 6.47 |
| H4806 | Saturation at a geometric limit | Raising a cricket ball’s seam helps only until the seam stands as tall as the boundary layer at the separation point. Beyond that height it trips the layer too far forward and the asymmetric region collapses, which should make side force against seam height peak and fall rather than flatten. A printed seam-height series, tested at one speed and angle, places the peak where the seam first exceeds the layer thickness. | 6.42 |
| H4872 | Torsion, not studs | Release from turf should depend on a shoe’s torsional stiffness rather than on its stud count. The foot rotates about the ankle as it leaves, and a stiff plate cannot follow, which lets a shoe with few studs on a flexible plate release more cleanly than one with many on a stiff plate. Shoes matched on stud count but differing in plate torsion, tested on the same turf, should rank by torsion. | 6.4 |
| H4801 | Regime split | Tilt a seamed ball’s seam plane away from the flight direction and its drag crisis should split into two minima rather than one. The seam trips the boundary layer on the two sides at different speeds; a single drag sweep reports the average of two transitions while a side-force balance can resolve them. Sweep speed at a fixed seam angle near 45 degrees (an estimate) and the side-force trace should show two events where a drag-only sweep shows one. | 5.24 |
| H4883 | Quasi-static misranking | Quasi-static compression tests should misrank foams for impact. The foam’s stiffness rises with rate and two foams can cross over between the two rates, which a slow test cannot see. Compress the same foams quasi-statically, then drop them at matched strain; the ranking should change. | 5.0 |
| H4864 | Shear strength tied to pulling rate | Spike grip is a soil-mechanics problem. The track’s shear strength is rate-dependent: a spike that holds in a slow pull can slip in a fast one, and length beyond a threshold stops helping because the confining stress is reached. Pull-out tests at several rates and lengths expose the rate dependence and the length threshold. | 4.8 |
| H4875 | Camber as the design variable | Camber should set a ski’s turn radius more than sidecut does at low edge angles. The camber’s pressure distribution decides where the edge engages, while sidecut matters more as the ski is tilted. Measure the pressure distribution along the edge at several tilt angles; the geometry that dominates at each angle is the answer. | 4.8 |
| H4851 | Creep under humidity | Tension loss in a strung frame should be mostly viscoelastic creep rather than a sudden change at impact. The creep rate depends on moisture, so a racket stored humid loses tension faster than one stored dry; the polymer’s relaxation is plasticised by water. Store identical strung frames at two humidities and measure tension at intervals; the rates should separate. | 4.69 |
| H4878 | Test condition above use condition | Foam returns less energy at running impact speeds than a drop test suggests. Its loss rises with strain rate and the drop test strikes at a lower rate, so the ranking of two foams can reverse. Instrument a shoe in running and in a drop rig at matched peak force; the returns at the two rates are the comparison. | 4.69 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H2907 | Count | Pairwise measurement of n artefacts yields at most n-1 independent differences, and what remains is a single scale factor the differences cannot constrain, so any reported matrix with more independent entries has been assembled with an assumption, usually that one artefact is exact. A rank test on the measured difference matrix identifies which assumption was made and whether the report admits making it. | 6.6 |
| H2958 | Hysteresis | A laboratory's result can depend on what it measured immediately before, through carry-over in the instrument or the operator's expectation, and being small per sample the dependence appears as scatter rather than as a trend. Re-measuring a subset in reverse order, and comparing each sample's two results against its position in the run, exposes the carry-over. | 4.76 |
| H2970 | Bound | No combined uncertainty can be smaller than the calibration uncertainty of the reference behind its largest contributor, because that reference's error is common to every later use; a budget reporting a total below that figure has either omitted a term or double-counted a cancellation. Comparing each budget's total against the calibration uncertainty of its dominant input flags the impossible ones without any new measurement. | 4.58 |
| H2991 | Count | Among tolerance width, measurement uncertainty, and the two risks, at most two are free: fixing tolerance and uncertainty determines both risks, and fixing a risk and a tolerance determines the greatest permissible uncertainty. A specification stating all four independently is over-determined and at least one statement is redundant. Checking a specification's four numbers against the relation that should connect them shows which one is decorative. | 4.56 |
| H2946 | Hidden term | The denominator of a z-score omits the uncertainty of the assigned value, so participants are scored as though the reference were exact; when few laboratories contribute, that omission can halve the effective denominator. Consequently scores near a decision threshold are overstated in small comparisons. Recomputing every score with the assigned value's uncertainty added in quadrature, and counting the verdicts that change, measures the omission without any new measurement. | 4.49 |
| H2909 | Scale argument | The dominant error in short-length calibration is angular: the comparator's tilt multiplied by the offset from the measuring axis grows relative to the length as the artefact shortens, so an instrument excellent at one metre can be poor at one millimetre. Measuring a graduated series from a millimetre to a metre on one comparator, and plotting error against the offset-to-length ratio, isolates the angular term from the scale term. | 4.47 |
| H2920 | Reversal | The sign of the temperature correction in a comparison depends on whether the artefact expands more or less than the comparator's scale, so a correction validated on steel blocks reverses on ceramic blocks of the same nominal length. Applying the steel-derived correction to ceramic doubles the error instead of removing it. Measuring a mixed set at two temperatures and checking the sign of each material's slope detects the reversal. | 4.47 |
| H2990 | Reversal | Tightening an acceptance limit to protect the consumer can loosen effective control once measurement uncertainty dominates, because more truly conforming items fail and are re-measured until they pass, and re-measurement selects the favourable error. The delivered population then holds a higher fraction of items whose true value sits near the limit. Comparing the true-value distribution of accepted items under two limits, using a calibrated reference set, shows the reversal. | 4.47 |
| H2996 | Unit trap | A tolerance quoted as a plus-or-minus number without a stated coverage is ambiguous by about a factor of two, because one reader takes it as a standard uncertainty and another as an expanded uncertainty at ninety-five per cent, and both readings are defensible while yielding different verdicts on the same measurement. Recomputing a batch of verdicts under both readings, and counting disagreements, gives the cost of the missing word. | 4.47 |
| H2919 | Threshold | Wringing of gauge blocks holds only above a contact pressure at which the adsorbed film is displaced; below it the blocks appear to wring yet separate under the comparator's probe, and the measured length is short by roughly the film thickness, of order tens of nanometres (estimate). Measuring a stack at several wringing loads and looking for a step in length against load locates the threshold without needing the film's properties. | 4.42 |
| H2967 | Misattribution | Repeatability measured on a fixture is usually entered as the instrument's own repeatability, but on a compliant fixture most of it is the fixture's resonance and the mounting's creep. The figure is then a property of one bench and will not transfer to a user with a different mount. Measuring the same instrument on two mounts of different stiffness, and checking whether repeatability scales with the mount's compliance, assigns the term correctly. | 4.4 |
| H2927 | Threshold | Homogeneity of a powder fails below a minimum sample intake set by the largest particle's mass relative to the test portion rather than by the certified concentration, so two materials at the same concentration but different particle sizes have different intakes and a method validated at one fails quietly at a smaller one. Measuring one material across a series of intakes and finding where between-unit variance rises above within-unit variance locates the threshold. | 4.38 |
| H2971 | Threshold | Instrument non-linearity is negligible below a curvature threshold set by the ratio of second derivative to resolution, and above it the term grows with the square of the range used. A budget validated over a narrow range therefore understates the term when the instrument is used across its full span. Comparing residuals from a linear fit at two range spans, and checking whether residual grows as the square of span, tests the threshold. | 4.38 |
| H2982 | Observability | A term whose influence was never varied cannot be estimated from the data at hand, however many measurements were made, so a small value claimed for a term held constant claims knowledge the data do not contain, and its contribution is bounded only by the range over which it was held. Deliberately varying the neglected influence, at some cost, and measuring the change, supplies an empirical bound where the budget previously had an assumption. | 4.36 |
| H2999 | Hysteresis | A conformity decision made near a boundary depends on the previous decision for the same item, because failure triggers re-measurement and the re-measurement's error is not independent of the first, so the process has a memory no single-measurement model represents, and the memory shows up as reduced variance among accepted items. Comparing the variance of accepted items against the variance a memoryless model predicts, using items measured twice regardless of outcome, exposes the dependence. | 4.36 |
| H2929 | Count | The number of independent certified values a material can carry is bounded by the number of independent realization methods available, not by the number of analytes listed, since values from one primary method share its bias. A certificate listing fifteen values from two methods carries two independent claims and thirteen correlated ones. Measuring a subset of analytes by a wholly independent method and checking whether the listed values move together shows how many claims are actually supported. | 4.29 |
| H2931 | Scale argument | Demonstrating homogeneity at a one-gram intake says nothing about a one-milligram intake, because between-unit variance for a powder scales inversely with the number of particles in the portion and that count scales with the cube of the portion's linear size. A material can be acceptable for macro analysis and hopeless for speciation work on the same batch. Measuring variance at four intakes spanning three decades and checking whether it follows the predicted inverse-count law tests the scaling directly. | 4.27 |
| H2985 | Misattribution | A failed conformity decision is usually attributed to the instrument's accuracy when it is the decision rule's position relative to the tolerance that failed, and the same measurement would pass under a rule with a different guard band. The attribution is invisible because a test report states the measurement and the verdict, not the rule. Recomputing the verdict from the same measurement under two rules, and counting the cases that change, separates the instrument's contribution from the rule's. | 4.27 |
| H2975 | Scale argument | Whether a term is negligible depends on the value being measured rather than on the instrument, because absolute terms hold fixed while relative terms scale, so resolution dominates at small values and calibration at large ones. A budget stated as one set of numbers therefore misdescribes the instrument across its range. Measuring a standard at three points spanning a decade and computing the budget at each from the same terms shows the crossover and where the single-number budget fails. | 4.24 |
| H2989 | Threshold | Conformity testing approaches a coin flip as measurement uncertainty approaches the tolerance half-width, and the transition is sharp enough to locate: for an item truly at the limit the pass probability falls from near certainty to near chance over a factor of about two in that ratio (estimate). Below the threshold more testing helps; above it more testing only sharpens the coin. Simulating pass probability against the ratio, or measuring it with a calibrated artefact set at the limit, locates the transition. | 4.18 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4087 | Wear-mapping claim | Scale phase maps onto roll wear. Harder scale abrades the roll more, so a strip with mixed scale should wear the roll in bands that follow the phase. Rolling strips of different phase mixes, with the roll's profile measured after each campaign, should map the bands. | 6.98 |
| H4066 | Inversion | Argon shrouding reverses above a flow. Bubbles coalesce and the gas layer breaks, so more argon can mean more clogging. Increasing argon in steps at one casting speed while measuring deposit growth should expose a minimum rather than a monotonic improvement. | 6.91 |
| H4075 | Impingement claim | Jet angle, not casting speed, drives mould-level fluctuation. Penetration scales with the angle's sine and the fluctuation follows impingement, so two speeds with one nozzle and one speed with two angles should separate the variables. The second pair is the test. | 6.91 |
| H4085 | Model-failure claim | Roll-force formulas assume a friction that high strain rates remove. The roll's surface heats and the scale softens as rate rises, so measured force should fall below prediction. A strain-rate series at constant reduction is the test. | 6.82 |
| H4094 | Orientation claim | Quench orientation fixes distortion direction. The vapour blanket's asymmetry sets the thermal gradient, and the gradient sets the distortion, so identical parts quenched in two orientations should distort in opposite directions. That reversal is the test. | 6.8 |
| H4026 | Stoichiometric bound | Chromium sets the carbon floor in argon-oxygen decarburisation, not the argon's dilution. Below a critical carbon, chromium's activity buffers the oxygen potential and decarburisation stalls, so a heat with more chromium should stall at a higher carbon at the same argon flow. Decarburising two chromium levels to the stall, with carbon and chromium sampled together, tests the bound. | 6.71 |
| H4002 | Hidden term | Coke's reactivity test heats its sample faster than a blast furnace heats its burden, and the ramp itself cracks pore walls before the gas can attack them. That cracking adds a strength loss the test attributes to the gas. Heating one coke at three rates to the same temperature, with the same hold, should give three losses, and two cokes that reverse rank under different ramps would prove the term real. | 6.42 |
| H4071 | Path dependence | Width transitions leave a crack trail that remembers speed. Strain accumulates while the mould's taper lies between the two widths, so two transitions to the same final width at different speeds should leave cracks in different places. Mapping crack position against transition speed tests the path. | 4.98 |
| H4078 | Coupling claim | Crown and edge drop answer to different roll terms. Bending sets crown; wear sets edge drop, so a mill can hold one and lose the other. Profile measured at two bending settings with wear fixed, then at two wear states with bending fixed, should separate them. | 4.98 |
| H4090 | Spall-depth claim | Spall depth follows case depth. The spall runs along the case-core interface, so a deeper case gives a deeper spall at the same thermal fatigue. Two rolls of different case depth under one mill load, with spall depth measured, should put spall depth with case depth. | 4.98 |
| H4093 | Delamination claim | Cooling rate, not layer thickness, spalls the white layer. Layer and substrate expand differently, and a fast cool puts the layer in tension. Two nitrided samples of equal layer thickness cooled at different rates, with residual stress and spall count measured, separate the two. | 4.98 |
| H4054 | Sectioning bias | Polishing removes soft inclusions before the microscope sees them. Relief hides them in the focal plane, so a polished section under-reports the soft population. One sample measured both by polished section and by electrolytic extraction exposes the deficit, which should grow with softness. | 4.96 |
| H4100 | Geometric-blanket claim | Vapour blanket lifetime follows part geometry. Vapour escapes most easily from upward-facing surfaces, so a concave face holds its blanket longer. Two parts of one alloy with different cavity orientations quenched in one oil, with hardness measured at the cavity's base, should make blanket lifetime follow geometry. | 4.96 |
| H4036 | Post-combustion mixing claim | Jet entrainment of carbon monoxide, not oxygen flow, sets the post-combustion ratio. A higher lance mixes with more of the vessel's gas and burns more carbon monoxide, so two lance heights at one flow should give different off-gas ratios. Off-gas analysis at both heights tests the mixing explanation. | 4.93 |
| H4082 | Scale argument | Orange peel tracks grain size relative to sheet thickness, not roll roughness. A fine-grained sheet stays smooth even under a rough roll. Two sheets rolled to equal roughness but different grain size, with surface waviness measured, should make waviness follow grain size. | 4.93 |
| H4098 | Creep claim | Load weight creeps the bottom of a carburising stack. At austenitising temperature the steel is weak enough for the stack's own weight to flatten the lowest parts, so distortion should follow stack height rather than thermal gradient. One tray loaded at two heights, with bottom-part ovality measured, tests it. | 4.91 |
| H4055 | Thermodynamic-path claim | Slag basicity, not aluminium total, fixes silica reduction in the ladle. The slag sets silica activity and therefore the oxygen potential the aluminium sees, so a high-basicity slag should suppress reduction even at high aluminium. Two heats with equal aluminium but different basicity, with silicon pickup measured, should follow basicity. | 4.89 |
| H4058 | Area claim | Slag-eye area, not gas flow, controls reoxidation. The eye's area is set by gas flow and slag thickness, and two stirring rates that produce the same area should produce the same reoxidation. Matching eye area by adjusting slag thickness at two gas flows, then measuring nitrogen pickup, tests the claim. | 4.89 |
| H4074 | Entrainment threshold | A vortex air core, not slag depth, sets entrainment at the tundish nozzle. The core forms when the outlet's Froude number passes a value set by nozzle diameter, so two diameters at one slag depth should entrain at different heights. Water-model tests would show the critical height scaling with diameter. | 4.89 |
| H4083 | Buckling-wavelength claim | A flatness wave is a buckling wavelength, set by strip thickness and the residual stress's gradient. The same stress can give edge waves in one thickness and centre waves in another. Two thicknesses rolled to the same measured stress profile, with flatness mapped, should move the wave's location with thickness. | 4.89 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3889 | Threshold | Contamination below roughly a monolayer changes bond strength very little, so a cleanliness specification tighter than that threshold costs money without buying strength, and the threshold is what a specification should be written against. Above it strength falls steeply and the cost of contamination appears suddenly. Bonding a graded series that spans sub-monolayer to multilayer, with strength measured, locates the knee and gives the specification a physical basis. | 6.49 |
| H3892 | Identity claim | A weak boundary layer and a contaminated surface are the same failure mode described from two ends, one from the fracture path and one from the chemistry, so either diagnosis should predict the other. Where they do not, one of the two measurements is wrong, and the disagreement is more informative than either result. Taking joints that fail interfacially and analysing their fracture surfaces for the contaminant the weak-boundary-layer model requires tests the identity. | 6.36 |
| H3827 | Identity claim | Desmutting and pickling are treated as separate steps, but both are dissolution of a surface layer at a potential set by the bath's oxidising power, and the boundary between them is a convention about how much metal is removed. A bath that desmuts is a weak pickle, and the two names describe doses rather than mechanisms. Running one bath at graded oxidising power and measuring both smut removal and metal loss should give one continuous curve, not two regimes. | 6.24 |
| H3875 | Count | A contamination budget resolved from one surface measurement can distinguish at most three independent sources, because the measurement returns a total and each source contributes through two ratios. Budgets that name five sources from one sample are accounting rather than measurement. Tagging one suspected source with a tracer and refitting shows how many the data can actually support. | 4.71 |
| H3832 | Bookkeeping deficit | Descaling acid is consumed by two reactions, dissolving scale and dissolving base metal, and most bath-life accounting counts only the first, so the acid looks longer-lived than it is while the base-metal term quietly dominates at long dwell. The error is small per part and large per campaign. Measuring iron in solution and separately measuring scale removal over a full bath life splits the two consumptions. | 4.53 |
| H3872 | Hidden term | A cleanroom's particle count includes the contribution of the sampling probe's own tubing and of the disturbance the probe creates, so the class number is partly a property of the monitor and its placement. Two rooms can be certified differently because they were measured differently, and neither measurement is wrong. Measuring one room with two probe lengths and two sampling positions quantifies the instrument's share of the number. | 4.49 |
| H3817 | Conservation claim | Solute leaving on the part equals what the flow carries away in a counterflow rinse, so the stage that limits the line is the one with the largest product of local flow ratio and local drag-out, not the dirtiest tank. Operators clean the dirtiest tank and leave the true bottleneck alone. Injecting a tracer at each stage and measuring the outlet concentration identifies the limiting stage without dismantling anything. | 4.47 |
| H3826 | Inversion | A pickling inhibitor can raise a bath's total acid consumption while lowering its metal loss, because it slows general attack but concentrates what remains at inclusions and grain boundaries, and those pits consume acid locally and seed roughness. Inhibitor efficiency measured by weight loss then reads as improvement while surface quality falls. Comparing inhibited and uninhibited baths at equal metal loss, with pit density counted, separates the two effects. | 4.47 |
| H3899 | Falsifiability claim | A claim that a cleaned surface stays clean indefinitely cannot be tested, because any observed failure is attributed to handling, so the claim absorbs its own counterexamples. Restating it as a dose to failure under a specified exposure makes it testable and makes the number comparable between suppliers. Exposing coupons to a controlled organic atmosphere and measuring the dose at which adhesion falls converts the claim into a measurement. | 4.47 |
| H3847 | Bound | The best roughness a mechanical polish can reach is bounded by the coarse tail of the abrasive's size distribution rather than by its mean, because the largest particle in the slurry sets the deepest scratch that survives the pass. Buying a tighter mean without tightening the tail therefore buys nothing measurable. Polishing with two abrasives of equal mean but different tail widths, and counting deep scratches, tests the bound directly. | 4.44 |
| H3854 | Criterion claim | Scratch visibility has a threshold in depth set by the observer's contrast sensitivity and the illumination, not by a physical transition in the material, so a specification of no visible scratches names a viewing condition rather than a surface. Change the light and the same part passes or fails. Showing one scratched coupon to a panel under two illuminations, and measuring the scratch depths, calibrates the wording to something physical. | 4.44 |
| H3848 | Inversion | Electropolishing can raise a surface's roughness average while lowering its contamination, because it removes the disturbed layer and exposes grain facets whose height variation exceeds that of the smeared layer it replaced. A rougher electropolished part can then outperform a smoother buffed one downstream, which inverts the usual reading of the number. Measuring roughness and surface carbon on buffed and electropolished coupons of one alloy separates the two directions of change. | 4.42 |
| H3862 | Null claim | On a gold surface, whose only contamination is adsorbed hydrocarbon, ultraviolet-ozone treatment should do nothing that a clean argon sputter does not, because gold grows no oxide and the ozone's action is confined to carbon removal. The claim is that the treatment is generic carbon removal and nothing else. Running both treatments on one gold film and comparing carbon by spectroscopy tests the null and, if it holds, simplifies the recipe. | 4.42 |
| H3868 | Regime identity | Corona treatment and dielectric-barrier discharge are names for one discharge regime at two current levels, and the differences reported between them are differences of gas chemistry and dwell time rather than of discharge physics. If so, the two should be interchangeable once gas and dose are matched, and the literature's disagreement dissolves into bookkeeping. Treating one polymer by both methods at matched dose and gas, and comparing the surfaces, tests the identity. | 4.4 |
| H3834 | Lower limit | There is a particle velocity below which an oxide is not removed however long the blast runs, because the crack a particle starts does not reach the scale-metal interface, and the threshold depends on scale thickness and toughness. Coverage, the variable operators control, cannot substitute for velocity at all. A velocity series at fixed mass flow and fixed time, with removal measured rather than coverage, finds a floor that coverage-based specifications cannot express. | 4.38 |
| H3856 | Conservation claim | Mechanical removal in polishing obeys a work relation in which rate is proportional to the product of pressure and relative speed, so neither variable alone can be optimised and two processes with the same product should remove the same amount. This is why speed can be traded for pressure only within the damage limit. A pressure-speed grid at fixed slurry, comparing removal at equal products, tests the relation and finds where it breaks. | 4.38 |
| H3880 | Bound | Removal of a particle from a surface by fluid flow is bounded by the boundary layer: a particle much smaller than the layer sits in nearly stagnant liquid and never sees the free-stream velocity, so no increase in flow rate dislodges it. The bound is set by particle size against layer thickness, not by pressure. Measuring removal against flow rate for three particle sizes shows the size at which the curve goes flat. | 4.38 |
| H3808 | Causal ordering | In solvent cleaning the residue left behind is decided by the last solvent that evaporates, not by the total volume used, so the order of the steps carries more information than their number. A pure final solvent that evaporates cleanly erases the history of the earlier ones; a contaminated final solvent writes that history back. Two sequences with identical solvent consumption but reversed final stages give different residue, and that is the whole test. | 4.36 |
| H3825 | Threshold | Hydrogen embrittlement from pickling has a threshold in absorbed hydrogen rather than in immersion time, and that threshold is set by the steel's hardness and trap density, so a mild steel can be pickled for hours while a high-strength one fails after minutes. Time-based specifications therefore transfer badly between alloys. Charging at a fixed current density to graded hydrogen contents, then testing notched tensile strength, gives the threshold independently of how the hydrogen arrived. | 4.36 |
| H3871 | Misattribution | Particles found on a part after cleaning are usually debited to the cleaning process, but most are deposited after the last wet step, during drying, transfer and handling, so the count measures the room more than the bath. Cleaning harder then changes nothing, and the effort is spent in the wrong place. Sampling at three points along the line with the same method at each attributes the count to the stage that added it. | 4.36 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H1124 | Thermodynamic bookkeeping | Mechanically alloyed intermetallics described as metastable are usually metastable only with respect to themselves at a larger grain size, because the enthalpy stored in boundaries of a ten-nanometre grain is estimated at a large fraction of the formation enthalpy. The phase is stable against the elements and unstable against coarsening. Annealing to coarsen without transforming is the test, and transformation should follow coarsening rather than accompany it. | 6.51 |
| H1197 | Count | A quenched phase-diagram section has one independent tie line per two-phase field, so a study reporting several distinct two-phase regions between the same end members is more likely resolving kinetics than thermodynamics. Equilibrating one composition for two durations and checking that phase fractions stop changing tests it. Counting regions without counting tie lines overstates the diagram. | 4.76 |
| H1132 | Artefact mimics effect | Negative thermal expansion measured by diffraction in a hydrated framework is frequently the beam driving water off the sample, which changes the lattice without any phonon mechanism and mimics a real effect convincingly. A sealed capillary at controlled humidity and an open one, measured through the same temperature series, separate them. Only the open sample should contract. | 4.73 |
| H1196 | Path dependence | An oxidising anneal heals defects until surface oxidation outruns bulk diffusion, after which the same anneal creates a new defect population by oxidising the skin while the core stays reduced. The condition depends on the path taken to the anneal, not only on its endpoint. Two thicknesses tested at one temperature show it, since only the thin sample reverses. | 4.69 |
| H1161 | Misattribution | A conductivity maximum in a doped oxide is read as defect association when it is often the dopant's solubility limit, beyond which a second phase coats the boundaries and blocks them. Measuring the boundary composition directly, rather than inferring it, decides between them. A maximum that coincides with the first appearance of a second phase is solubility and not association. | 4.67 |
| H1164 | Bound | Ionic conductivity cannot exceed the host's attempt frequency weighted by the best pathway's migration barrier, and doping raises the carrier concentration rather than lowering that barrier, so no dopant beats the bound. The bound is computable from the host's phonon spectrum. Best-doped and undoped host compared at one temperature, with carrier concentration measured independently, would show whether the barrier moved. | 4.67 |
| H1115 | Identity | Environmental and intrinsic embrittlement of a nickel aluminide are one mode seen at two strain rates, the environmental version being the intrinsic one with hydrogen supplied faster than it can escape. If so, the ductility minimum should fall at the strain rate where diffusional supply matches mechanical demand. Testing in vacuum at that rate should reproduce the curve obtained in air. | 4.64 |
| H1140 | Bound | Epitaxial perfection cannot exceed the substrate's subsurface damage, because the damage's strain field propagates through the film and seeds dislocations there, so the deposition method sets the ceiling only when the substrate is better than the film. Growing identical films on substrates prepared to different depths of damage tests it. The best film should be no better than the best substrate. | 4.64 |
| H1175 | Identity | Mechanochemical activation and a nanoscale precursor are two names for raising the contact area between reactants, so a milled mixture and a nanoparticulate one follow identical reaction curves once normalised to measured interfacial area. Gas adsorption supplies that normalisation. Identical curves would show that the two routes differ only in how the area was made, not in what the area does. | 4.64 |
| H1186 | Bound | Pressure in a multi-anvil cell is capped by the anvil material's yield strength rather than by the press's force, so a larger press does not guarantee a higher pressure. An internal standard in two cells of different truncation measures where the cap sits. Pressure should saturate at the load where the anvil deforms, and the saturation should follow the anvil's strength. | 4.64 |
| H1189 | Selection effect | High-pressure phase diagrams are assembled from products that survive decompression, so they are biased toward phases with high kinetic barriers and against those that revert on unloading. In-situ diffraction at pressure, with no unloading step, is the only way to see the difference. Running both on one composition and counting the phases missing from the quench diagram measures the bias directly. | 4.64 |
| H1118 | Reversal | Boron refines grains in an intermetallic and improves ductility only until the borides themselves become crack initiators, after which toughness falls with further addition. Where it turns over is set by particle spacing relative to the plastic zone size, not by boron content alone. Measuring that spacing across a boron series, rather than assuming it scales, is what makes the reversal visible. | 4.62 |
| H1169 | Scale argument | When the space-charge layer at a boundary grows thicker than the grain, the boundary no longer depletes a finite region but shifts the whole grain's Fermi level, and the conductivity's dependence on grain size inverts. The crossover is estimated in the tens of nanometres for a typical oxide. A grain-size series with the boundary potential measured independently puts the inversion at that size. | 4.6 |
| H1152 | Bound | A solvothermal yield is capped by the solvent's own decomposition at the reaction temperature, because past that point the solvent's reducing power shifts the product's oxidation state and the target phase stops forming. The cap is a property of the solvent, not of the precursor. A temperature series with the solvent's decomposition measured independently by pressure rise should show yield falling where decomposition begins. | 4.58 |
| H1199 | Scale argument | Once the boundary's own width becomes a non-negligible fraction of the grain, at roughly ten to twenty nanometres, the boundary composition stops behaving as a monolayer and the material's effective stoichiometry shifts (estimate). Nanocrystalline and coarse samples of identical nominal composition then differ in lattice parameter by an amount that scales with the boundary width, measurable at the tenth-of-a-percent level in a careful scan (estimate). | 4.56 |
| H1153 | Threshold | A hydrothermal reaction needs a minimum water activity even at high temperature, because dissolving the oxide requires a hydrated surface intermediate, and the threshold is set by water content rather than by total pressure. Water-in-alcohol mixtures held at fixed pressure test it: conversion collapses at a water fraction, not at a pressure. A threshold that follows mole fraction rather than fugacity rules out a purely mechanical explanation. | 4.51 |
| H1112 | Unit trap | Per cent theoretical density is ambiguous whenever the reference cell differs from the cell the sample adopts, and a phase transition during sintering moves that reference by several per cent (estimate: two to four for a perovskite). Two laboratories can then dispute whether a body reached 95 per cent while measuring identical porosity. Reporting image-analysis porosity beside the density ratio, for one sample straddling the transition, resolves the disagreement. | 4.47 |
| H1163 | Identity | The two defect regimes reported for a perovskite, one at low oxygen partial pressure and one at high, are one regime viewed from two sides of the same neutrality condition, since the majority defect changes sign of charge without changing mechanism. Conductivity and Seebeck coefficient measured in one cell across the range should cross where the majority changes. The transition belongs at the crossing, not at either slope change. | 4.47 |
| H1102 | Hidden term | Every measured sintering shrinkage contains a term from the dilatometer pushrod's own expansion, and for a sample shorter than about five millimetres that term is comparable to the shrinkage being reported (estimate, order of magnitude). The correction is not a constant, because it scales with the pushrod's thermal gradient, which moves when the furnace hot zone drifts. Two sample lengths in one run separate the contributions exactly, since only the sample term scales with length. | 4.33 |
| H1110 | Inversion of standard practice | Reducing atmospheres are the standard prescription for oxides with easily reduced cations, but near a redox boundary a reducing gas stabilises a vacancy population that suppresses densification, and the correct practice is a mildly oxidising atmosphere that fixes the cation valence. A pO2 series at fixed temperature locates the density maximum instead of assuming it. If that maximum sits above the equilibrium pO2, the prescription is inverted. | 3.51 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3322 | Equifinality | Two spinning routes can yield yarns of identical count, twist and tenacity that behave differently in weaving, because the difference lives in the distribution of fibre ends rather than in any measured mean. If so, an end-distribution metric should predict weavability while the standard means do not. Measuring end distributions on both yarns and following them through the loom is the test. | 6.62 |
| H3340 | Disguised mechanism | Bagging recovery is reported as an elastic property, but the loops that deform during bagging are loaded at their apices, where the fibre is already near its yield strain, so recovery is a fatigue measurement in disguise. Repeated bagging at one load should therefore degrade recovery cycle by cycle even though no single cycle is excessive. Following recovery against cycle count on one specimen decides it. | 6.53 |
| H3301 | Misattribution | Bundle tenacity is read as a fibre property diluted by irregularity, yet a bundle also suppresses the lateral contraction that a free filament performs just before it breaks. If the deficit belongs to that constraint, it should scale with the number of neighbours rather than with the test length, which is exactly what an irregularity explanation does not predict. Bundles of deliberately varied count, all at one gauge length, disentangle the two explanations. | 6.38 |
| H3313 | Optimum | Friction-spun yarn strength rises and then falls with drum speed, because faster drums wrap more fibres while each wrapper grips less. The optimum therefore belongs to the fibre's finish rather than to the machine. Two finishes of different friction should place the optimum at different speeds, a prediction one machine can test in a day. | 4.78 |
| H3389 | Aspect ratio | Wetlaid formation depends on the fibres staying dispersed until the water is removed, and dispersion stability is set by the fibre's aspect ratio rather than by the surfactant dose above a threshold. Long fibres entangle and flocculate whatever the chemistry. Forming handsheets from a length series at fixed chemistry, and measuring floc size, locates the limit. | 4.76 |
| H3331 | Moiré | Reed marks are a moiré between the dent pitch and the weave repeat, so their spacing is set by the pair and not by either alone. The same reed should therefore mark at one pick density and not at another. Weaving a pick-density series with a single reed, and locating the densities that mark, tests the arithmetic. | 4.73 |
| H3309 | Hysteresis | A core-spun yarn remembers the tension at which its core was fed, because the sheath's residual torque is set by that tension and is not erased by winding. Two yarns identical in count and twist can therefore snarl differently, and spinning one core at three tensions while measuring torque and snarl length turns the memory into a calibration. | 4.71 |
| H3382 | Coupled transport | Vapour transmission is measured on a dry fabric, but in use the cloth contains liquid water, which changes both the vapour path and the heat path, so the dry measurement is the wrong boundary condition. A fabric can be excellent dry and poor wet. Measuring transmission at several internal water contents, on one cloth, gives the in-use curve. | 4.71 |
| H3383 | Blurred input | Spunbond uniformity is attributed to the spinneret's hole pattern, but the filaments are still fluid when they enter the quench, so the pattern is blurred by the air's asymmetry before the web lands. A perfectly drilled spinneret can therefore produce a non-uniform web. Measuring basis-weight uniformity from one spinneret under two quench symmetries would show which account holds. | 4.71 |
| H3359 | Scale argument | Beam dyeing passes liquor axially, so the pressure drop grows with the beam's length and levelness along the beam becomes a pressure-drop problem rather than a chemistry problem. A long beam should be less level than a short one at equal flow per unit mass. Dyeing beams of two lengths with all chemistry fixed measures the axial gradient directly. | 4.69 |
| H3304 | Bound | No filament can be torsionally stiffer than a homogeneous rod of its own axial modulus and section, and every measured torsional rigidity exceeds that floor. The excess is the signature of a radially graded modulus, a stiffer skin over a softer core, rather than of experimental error. Filaments spun with different skin thicknesses should show the excess tracking that thickness. | 4.67 |
| H3350 | Multiple minima | A draped fabric settles into one of several energy minima, so a specimen placed gently and one dropped can take different shapes and both be correct. Single-shape drape measurement is therefore a sampling of one basin. Releasing the same specimen from several initial configurations, and counting the distinct shapes it reaches, characterises the landscape instead of a point in it. | 4.67 |
| H3318 | Inversion of standard practice | Worsted combing is designed to remove short fibres on the theory that they weaken yarns, but short fibres also fill interstices and raise packing, so removing them can cost abrasion resistance while gaining tenacity. Combing intensity should therefore be set by the intended wear mode rather than by the fibre-length distribution. A combing series tested for both tenacity and abrasion would locate the crossover. | 4.58 |
| H3305 | Reversal | Drawing raises a fibre's modulus, but past a certain draw ratio the modulus can fall again, because the tie molecules that carry load are pulled so straight that further drawing breaks them instead of orienting them. The turnover should move with drawing temperature rather than with polymer grade. A draw-ratio series at two temperatures, with modulus and orientation measured on the same specimens, places it. | 4.56 |
| H3311 | Unit trap | The hairiness index counts protruding fibres per unit length and is blind to how far they protrude, so two yarns with equal indices can differ in the mass of fibre outside the core. A claim that a treatment reduces hairiness is therefore a claim about a count until protruding length is measured too. Both measurements on the same bobbins, before and after treatment, settle what the treatment did. | 4.49 |
| H3338 | Bound | The bending rigidity of a woven fabric has a floor set by the free bending of its interlaced yarns, and every measured value sits above it. The gap is the friction moment at the crossings, which the weave can change without changing the yarns. A series of weaves from one yarn with different float lengths should show the gap growing with the number of crossings per unit area. | 4.49 |
| H3400 | Unification | Three quantities measured by different industries (yarn irregularity, fabric drape and nonwoven pore size) may all be set by one length, the mean distance a fibre travels between contacts. If so, one measurement of contact spacing on an assembly should predict all three, and the test is a set of assemblies spanning the three industries, measured for spacing and for each property, checking whether one scaling collapses them. | 4.49 |
| H3324 | Reframing | The spinning triangle is drawn as a triangle, but a fibre under tension between the nip and the yarn hangs in a catenary, so tension across the ribbon is not linear in position. Predictions that assume a triangular profile are therefore biased at the edges, where the curvature is greatest. Recomputing the profile as a catenary changes the predicted edge tension without changing any measured input, which is what a reframing should do. | 4.38 |
| H3397 | Metric boundary | A fabric's water footprint is reported as water withdrawn per kilogram, but the environmental effect is set by the water returned and its quality, so a process with high withdrawal and full recycling can score worse than one with low withdrawal and poor effluent. The metric measures the wrong end of the pipe, and computing both withdrawal and effluent load for two dyehouses, ranking them, shows whether the metric inverts the true order. | 4.38 |
| H3308 | Saturation | Adding air pressure in a jet-spinning nozzle stops adding wrapper fibres past a certain point, because the supply of free fibre ends at the nozzle is already exhausted. The saturation pressure is therefore set by the drafting zone's control of floating fibres, not by the nozzle's geometry. Two machines with different break drafts should saturate at different pressures, which is both a test of the claim and a way to compare their drafting zones. | 4.36 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H4656 | Guild-count ceiling | The number of functional guilds a community can hold is bounded by the number of limiting resources, so adding a carbon source adds a guild only if it adds a resource dimension. A readily degradable carbon that feeds the same guilds should raise rates without raising diversity, and that is the test. | 6.84 |
| H4619 | Artefact mimicking effect | The compression phase seen in a settling test is produced partly by friction against the cylinder wall, and its duration should then scale with the column's diameter. A material property would not care about the diameter. Three columns of different diameter are enough to decide whether the compression coefficient is a sludge property or an apparatus one. | 6.73 |
| H4602 | Hidden term | Alkalinity consumed during coagulation is charged to hydrolysis, but a floc that consumes protons also draws carbon dioxide across its boundary layer, and the measured drop includes that transfer. The extra term should vanish in a closed vessel with no headspace and should grow with the floc's surface area at fixed dose. A pair of tests, open and closed, separates the two contributions without any assumption about speciation. | 6.49 |
| H4659 | Sampling artefact | The anoxic microzone reported inside a floc can be created by the probe that measures it, since inserting a microelectrode displaces the very water whose gradient it reads. A bulk sample relaxes toward the bulk oxygen within a diffusion time, so bulk and microelectrode measurements differ by construction. The test is a floc whose size is varied at constant biology, where the apparent zone should scale with the disturbance. | 6.47 |
| H4610 | Null claim | Bridging contributes nothing once the adsorbed coverage passes the point where loops overlap, and every improvement seen beyond that point is charge neutralisation. The null is testable with two polymers of equal charge and very different molecular weight, which should converge at high dose if the null holds. | 4.93 |
| H4698 | Aging null | The decline in a membrane's performance over a campaign can be explained entirely by irreversibly adsorbed humic material, so the null is that the polymer itself is unchanged. A surface analysis of the cleaned membrane tests the null, and a null that survives removes aging as an explanation. | 4.93 |
| H4691 | Induction-time control | The induction time before scaling begins is set by the density of nucleation sites on the surface, not by the bulk supersaturation, so the same water scales sooner on one membrane than another at equal supersaturation. A coupon test inherits the coupon's surface and does not transfer to a module. | 4.89 |
| H4605 | Interior optimum | Floc size passes through a maximum as shear rises, because collisions become more frequent while breakup becomes more severe. The shear at which the maximum sits is set by floc strength, so two coagulants at the same dose should peak at different shears and the separation of the peaks is a strength measurement. | 4.82 |
| H4633 | Consistency test | The limiting flux obtained from a batch settling test and the limiting flux inferred from a full-scale thickener's dynamic response must agree if both measure the same sludge. A disagreement identifies an artefact in one of the measurements rather than a scale effect, and the batch test is usually where the artefact sits. | 4.82 |
| H4637 | Dimensional collapse | Media size and bed depth act on capture through their ratio, so two beds with the same depth divided by grain size should give the same normalised breakthrough. A collapse across beds chosen to match the ratio is the test, and a failure of the collapse would mean the collector efficiency depends on size separately. | 4.78 |
| H4629 | Polydispersity limit | The interface between sludge and water is as thick as the spread of settling velocities allows, so a sharp interface in a polydisperse suspension means the fines have been built into the coarse flocs. Interface sharpness is then a measure of flocculation and not of settling, and reading it as a settling indicator inverts the cause. | 4.76 |
| H4665 | Oscillation claim | Gas production from a digester oscillates with a period set by the hydraulic flush time, because volatile acids accumulate and are washed out on that time scale. A stable reading over a window shorter than the period is therefore not evidence of stability, and the window has to exceed the flush time for the test to mean anything. | 4.71 |
| H4686 | Fine-fraction ceiling | Irreversible fouling begins at a flux set by the back-transport of the smallest foulant and not by the mean, so a membrane can be run above the critical flux of the bulk and below that of the fines. The safe flux is then a property of the fine fraction, and a test on a filtered feed will overstate it. | 4.69 |
| H4650 | Misread symptom | Bulking is read as a filamentous outbreak, but the filaments respond to the residual substrate profile, and the selector's mixing is what sets that profile. Changing the community while holding the mixing constant should therefore leave the bulking unchanged, while changing the mixing at constant community should remove it. The test is a pair of continuous reactors with the variables crossed. | 4.64 |
| H4613 | Hysteresis claim | A floc consolidated at high shear does not re-expand when the shear is lowered, so the size at a given shear depends on whether the shear is rising or falling. A jar test run downwards therefore reports larger flocs than the same test run upwards at the same speed. The loop area measures consolidation and should vanish for a floc that never exceeded its yield stress. | 4.53 |
| H4670 | Mixing limitation | When the disinfectant decays faster than it mixes, the achievable inactivation is bounded by the mixing time, and additional dose is consumed in the inlet rather than delivered to the far wall. The bound is the ratio of the decay rate to the mixing rate, so it moves with the tank's geometry and not with the chemistry. A tracer and a residual profile measured together locate it. | 4.51 |
| H4651 | Surfactant term | The oxygen transfer coefficient measured in mixed liquor carries a term from the coalescence of bubbles coated with soluble microbial products, so the alpha factor depends on the bubble size distribution and not only on viscosity. Two liquors of equal viscosity and different product concentration should differ, and the difference should track bubble size rather than viscosity. That is a measurement of the coating, not of the blower. | 4.49 |
| H4606 | Identity claim | Sweep floc and patch-charge attachment are described as separate mechanisms, yet both are the attachment efficiency of a surface at some polymer coverage, and coverage is the variable they share. If that is so, attachment efficiency plotted against coverage should be single-valued across the two regimes, turning down where the adsorbed layer saturates. The test needs two polymers of different charge density and a coverage measurement that does not assume the mechanism. | 4.4 |
| H4653 | Maintenance threshold | A biofilm cannot sustain itself below a substrate concentration set by its maintenance energy and its diffusion length, and that threshold is not the half-saturation constant. It should be independent of the history of the bulk concentration, which distinguishes a maintenance limit from an acclimation effect. | 3.98 |
| H4667 | Redundancy as rate | Two communities can carry the same function and still differ in how quickly they recover from a shock, so redundancy is a rate property and not a composition property. A diversity index cannot predict resilience, while a recovery rate measured on the same inoculum can. | 3.0 |
| ID | Form | Claim | Score |
|---|---|---|---|
| H3248 | Two percolation thresholds | Conductive fibres percolate electrically at a volume fraction set by aspect ratio, but the same fibres percolate mechanically at a lower fraction because stress transfers over a longer range than charge. One sheet series should show both thresholds, and their ratio measures the transfer range. | 7.0 |
| H3265 | Ranking reversal between scales | The ranking of two pulps by tensile strength can reverse between a handsheet and a machine-made sheet, because machine orientation amplifies the length difference in one direction while the handsheet averages it. The reversal should vanish when both sheets are tested at equal orientation ratio. | 7.0 |
| H3267 | Boundary that is not a material property | The moisture at which stacked sheets block is a property of the stack's pressure and dwell time rather than of the paper, so blocking results do not transfer between reels wound at different tensions. A blocking test at two winding tensions should give two thresholds for one paper. | 6.93 |
| H3280 | Energy accounting | Energy absorbed in wood fracture is dominated by friction during fibre pull-out rather than by new surface, so raising surface energy does little for toughness while raising fibre length does much. The pull-out term should scale with the square of length in a fracture test across span-to-depth ratios. | 6.93 |
| H3282 | Sign that looks like error | A Poisson's ratio above 0.5 in the radial-tangential plane need not be an error, because in compression the lumens collapse and the volume can decrease. Simultaneous lateral strain and volume measurement should show collapse exactly where the ratio passes 0.5. | 6.89 |
| H3227 | Misattribution of joint strength | In a thin glue line the measured shear strength is the wood's rolling shear strength, not the adhesive's, so adhesives ranked by lap-shear tests are partly ranked by the wood beneath them. The crossover thickness, where failure moves from the bond line into the wood, should scale with the adhesive's modulus. | 6.87 |
| H3261 | Instability threshold | Wrinkling in a wound roll is a buckling instability whose onset is set by the ratio of winding tension to bending stiffness, and above a critical diameter the critical tension stops depending on diameter. The threshold should be measurable by winding one paper at several tensions and reading the onset radius. | 6.87 |
| H3284 | Irreconcilable tests | Two standard shear tests disagree by a factor that depends on grain angle, and the disagreement is a difference in imposed stress state rather than experimental error, so no single shear strength exists. Strain-instrumented specimens should show different strain fields at equal nominal stress, and the difference should account for the factor. | 6.84 |
| H3255 | Limiting case that fails | In the limit of zero bonding a network's modulus should vanish, yet measured values extrapolate to a positive intercept, which is the fibres' own bending stiffness. The intercept should scale with the cube of fibre length and with sheet density, and it bounds how low a modulus can go at a given density. | 6.82 |
| H3209 | Two-regime claim with a diagnostic | Drying time should scale with the square of thickness below fibre saturation and linearly with thickness above it, because the transition from bound-water diffusion to capillary flow changes the exponent. A log-log plot should therefore have a break, and the thickness at the break is a more direct measure of fibre saturation than the sorption isotherm. | 6.76 |
| H3221 | Rate reversal | The transient strain of loaded wood under a moisture change should shrink and then reverse sign as the ramp is slowed below the cell wall's relaxation time, because slow sorption lets the wall relax the stress that fast sorption converts into strain. One specimen tested at two ramp rates through the same moisture interval is the test. | 6.73 |
| H3250 | Irreversibility of curl | The compressive behaviour of a fibre mat differs between the first loading and later ones because fibre curl is set irreversibly on first compression, so caliper depends on how often the sheet has been calendered. A sheet measured before and after one extra nip should differ at equal load, and the difference should not recover on standing. | 6.73 |
| H3274 | Threshold then power law | There is a stress below which wood does not creep measurably over decades, set by shear yielding in the cell wall, and above it the creep rate follows a power law whose exponent is a material constant. The threshold is where the log-rate against log-time curve stops being flat, and it should be independent of specimen size. | 6.73 |
| H3291 | A standard that is not a bound | The small clear specimen's strength is not a lower bound for a board's strength, because the specimen removes the defects that govern the board, so the 'characteristic' value is neither characteristic nor a bound. Paired tests of small specimens and full boards from the same logs should show the board falling below the specimen as defect severity rises. | 6.71 |
| H3215 | Sign reversal | The temperature dependence of bound-water diffusion should reverse above roughly 60 °C (estimate), because thermal softening lowers tortuosity faster than it raises the activation barrier. A temperature ladder should therefore show a maximum in diffusivity, and the maximum should move to lower temperature as moisture content rises. | 5.96 |
| H3254 | Property-specific reversal | Increasing drying restraint raises tensile strength but lowers extensibility, so toughness has an interior maximum in restraint, and 'restraint is good' is a statement about one property only. A restraint series measured for all three should place the toughness maximum away from the strength maximum. | 5.0 |
| H3281 | Specimen-thickness limit | The rising R-curve of wood plateaus at a crack extension set by the fibre bridging zone, and below a specimen thickness comparable to that zone the rise cannot be measured. A thickness series should show the plateau appear and then stabilize once thickness exceeds the zone. | 4.98 |
| H3294 | Hidden mass effect | Part of the property gain from more resin is the resin's own mass raising panel density, so at constant density the gain is smaller by a measurable factor. Panels pressed to equal density at two resin levels should differ less than panels pressed to equal thickness. | 4.98 |
| H3231 | Count of identifiable parameters | A wood bond has three possible failure loci — adhesive, interphase and wood — so strength data alone identify at most three interfacial parameters. A four-parameter fit to three species should leave one parameter unidentifiable, and the test is whether its interval spans the physically possible range. | 4.96 |
| H3277 | Non-monotonic limit case | Strength should rise as moisture falls, yet measured strength peaks at low moisture and falls again in the oven-dry state because micro-cracks open on desorption. The peak should move with drying severity, so a slowly dried specimen should peak at a lower moisture than a fast-dried one. | 4.96 |
What this page establishes: of the five most-repeated claims across the 6,000, two held, one was a definition masquerading as a result, and two failed. That ratio is the honest summary of the whole collection so far.
Incoprea Research · the practical 1,200 · 3 November 2026 ·
~/Desktop/Nov3/ · selection in incoprea_nov3/results/selected1200.json