Honey Granulation Is a Sugar Ratio — Hypothesis Test 15
Supplied working report; original verdicts not validated.
Review status: Needs review. This is an archival source note, not a passed proof.
report.md
# Honey Granulation Is a Sugar Ratio — Hypothesis Test 15 **Working report — not peer reviewed. Educational information, not medical advice. AI-assisted test: DeepSeek.** ## Source Verbatim claim, prediction and refutation from `[local provenance path] Theories Vol XXIV - 2026-10-30/theories/honey-glucose-fructose.html` (Incoprea Theory 126 of 219, Vol. XXIV, "Honey crystallisation / Ratio lens", title: *Granulation Is a Sugar Ratio*): > **The claim** — "Granulation tendency is set by the ratio of glucose to fructose, so honeys of equal total sugar differ in how fast they set." > > **Prediction** — "Granulation should track the ratio." > > **Refutation** — "Granulation tracking total sugar would remove composition." Mechanism section: "Glucose crystallises readily while fructose stays in solution; the ratio determines the driving force." ## Why probable The mechanism is chemically sound: honey is a supersaturated glucose solution; glucose crystallises from honey as glucose monohydrate while fructose is 4.4× more soluble and stays in solution (Schiassi et al. 2025, citing Escuredo et al. 2014). The thermodynamic driving force is the supersaturation S = C_glucose / C_glucose,sat(T, C_fructose), and published ternary data show glucose solubility in water at 30 °C falls from 54.6 wt% (no fructose) to 32.6 wt% in a 39.4 wt% fructose solution (Jackson & Silsbee 1924). So the direction of the claim is plausible: more glucose relative to fructose (and to water) → higher S → faster granulation. ## Method ### Solubility model (glucose–fructose–water) Published relationship used: **Jackson, R.F. & Silsbee, C.G. (1924), "Saturation relations in mixtures of sucrose, dextrose, and levulose", Technologic Papers of the Bureau of Standards No. 259, Table 5** — the system dextrose–levulose–water at 30 °C with hydrated dextrose as solid phase (the same hydrate that crystallises from honey): | Levulose wt% | 0 | 8.94 | 14.50 | 23.23 | 33.09 | 35.69 | 37.10 | 39.39 | |---|---|---|---|---|---|---|---|---| | Dextrose (sat.) wt% | 54.64 | 49.34/49.32 | 45.97 | 41.01 | 35.76 | 34.48 | 33.67 | 32.55 | Linear fit (r = −0.9997): **C_g,sat(wt%) = 54.33 − 0.559·C_f(wt%)**. Re-expressed per 100 g water the isotherm is nearly flat (118.6 − 0.028·C_f,water g/100 g water) — i.e. fructose lowers glucose solubility mainly by replacing water. The task anchor (≈91 g glucose / 100 g water at 25 °C, pure water) is 0.755× the J&S 30 °C value (120.5 g/100 g water); a 25 °C-scaled S was also computed (rankings unchanged, since scaling is uniform). S = C_g,honey / C_g,sat(F_honey) was computed at the table temperature (30 °C); Kasolia (1991) documents that the J&S data were later extended to higher fructose by Lothrop (1943), but the extended numbers are not numerically available here. ### Data 34 composition rows in `data/honey_compositions.csv` (every row carries a source): - **White, Riethof, Subers & Kushnir (1962), "Composition of American Honeys", USDA Technical Bulletin 1261** (490 honey samples): Table 8 gives mean glucose/fructose/water for each of the 10 observed granulation classes (0 = completely liquid … 9 = complete hard, judged after 6 months undisturbed); Tables 5–7 give area means for alfalfa (Intermountain, Imperial Valley), cotton (Texas, Arizona, California) and orange (California, Florida) with their observed granulation state. Values extracted by OCR of the 400 dpi archive.org scan (rotated tables) and cross-checked. 17 rows with observed granulation rank. - **Schiassi et al. (2025), Food Science and Technology (Campinas), doi:10.5327/fst.552** — 8 Brazilian honeys (T0 glucose, fructose, moisture) with 180-day crystallization observations (absorbance/microscopy) and the paper's F/G-based fast/slow labels. - **Omran et al. (2025), J. Sohag Agriscience 10(2):129–140** — 7 Egyptian honeys with composition and F/G; granulation speed only *inferred* from F/G (not observed) — excluded from speed correlations, kept for composition. - **Vilgis (2015), Physik in unserer Zeit 46(6):310** — two synthetic ternary model "honeys": acacia-like 36 % G / 40 % F / 24 % water stays liquid; rape-like 46 % G / 30 % F / 24 % water crystallises quickly. ### Analysis `code/honey_test.py`: compute G:F, G:W (glucose/water), S for every row; Spearman rank correlation of each index against observed granulation rank (n = 17 real-honey rows with observed speed; plus the n = 10 class means alone); fast (rank ≥ 7) vs slow (rank ≤ 1) threshold comparison; check of White's published rule (D/W ≤ 1.7 nongranulating, ≥ 2.1 rapid) and of the claim's example threshold (G:F > 1.3). ## Results **Granulation tracks composition, not total sugar (refutation criterion met).** In TB 1261 Table 8, mean glucose rises monotonically from 27.70 % (code 0, liquid) to 35.22 % (code 9, complete hard) while mean fructose is essentially constant (37.5–38.9 %) and moisture falls from 17.5 % to 15.7 %. White et al.'s own ANOVA found granulating tendency regresses on dextrose and moisture, not on total sugar — granulation would *not* track total sugar. **Correlations (Spearman, observed-speed rows):** | Index | rho (n = 17 real honeys) | rho (n = 10 class means) | |---|---|---| | G:F ratio | +0.891 (p = 1.6e−6) | +0.988 | | G:W ratio | +0.924 (p = 1.2e−7) | +0.976 | | S (30 °C, J&S solubility) | +0.843 (p = 2.1e−5) | +0.976 | **Threshold analysis (fast = rank ≥ 7: ¾ of depth / complete; slow = rank ≤ 1: liquid / few scattered crystals):** | Index | fast range (n = 7) | slow range (n = 2) | Separation | |---|---|---|---| | G:F | 0.892 – 0.956 | 0.739 – 0.798 | clean (gap 0.798→0.892) | | G:W | 2.059 – 2.385 | 1.583 – 1.760 | clean (gap 1.76→2.06) | | S (30 °C) | 1.028 – 1.152 | 0.830 – 0.922 | clean (S = 1 sits in the gap) | - White's own D/W rule: all slow rows ≤ 1.7; fast rows ≥ 2.06 (the class-7 mean, 2.06, sits marginally under the 2.1 line; cotton 2.25–2.39, classes 8–9 = 2.16–2.24). Rule holds with one borderline row. - **The claim's example threshold G:F > 1.3 is never reached by any real honey** (max observed G:F = 0.956, cotton). Only the Vilges *rape-like synthetic syrup* (G:F = 1.53) exceeds 1.3 — and it does crystallise fast, so the example works for model syrups, not honeys. - The Vilges acacia-like syrup has G:F = 0.90 — the same G:F as the complete-granulation class means (0.90–0.91) — yet stays liquid because its water content is higher (G:W = 1.50 vs 2.16–2.24). **Ratio alone does not fix granulation; water does part of the work.** - Individual-level overlap: in the Schiassi 2025 dataset the published fast criterion (F/G ≤ 1.14, i.e. G:F ≥ 0.877) labels coffee (F/G 1.10) and eucalyptus (1.12) as rapid, but the honeys *observed* to crystallise most over 180 days were assa-peixe (F/G 1.30, G:W 1.43) and cipó-uva (F/G 1.20, G:W 1.31) — 2/8 rows where the ratio criterion and the observation disagree (caveat: intensity there was scored by absorbance/microscopy, and initial nucleation state differs between honeys). ## Verdict **PARTIAL** — the direction is strongly supported (higher G:F, G:W and S all rank higher granulation speed; the extremes separate cleanly in the 490-honey TB 1261 data, and granulation demonstrably tracks composition rather than total sugar), but the claim as stated has a weak threshold: its own example (>1.3) is unreachable by real honeys, the published data show glucose/water — not G:F alone — is the better separator, and individual-honey observations (Schiassi 2025; Vilges model syrups) show overlap where the ratio criterion misclassifies. ## Limitations - TB 1261 rows are class *means*; White et al. note large individual variation within types, so clean separation of means overstates individual-level separation. - S uses the 30 °C J&S isotherm while honey granulates at ~15–25 °C; the uniform temperature scaling preserves rankings but absolute S values are not storage-temperature-exact. - The J&S fit is extrapolated beyond 39.4 wt% fructose; Lothrop's extension to higher fructose was not numerically accessible (cited via Kasolia 1991 and White 1962). - Nucleation kinetics (seed crystals, the dominant practical factor — cf. Vilges' metastable syrup) are not part of the S/G:F model. - Schiassi 2025 speed is scored from absorbance/microscopy over 180 days, not days-to-granulate; Omran 2025 has no observed speed at all (excluded from correlations). - Table extraction relied on OCR of rotated tables in a 1962 scan; values were verified at 400 dpi with cross-readings (one ambiguous cell, row-7 levulose, resolved as 38.54 by column-pattern). - Composition data are dominated by 1956–57 U.S. honeys; European unifloral types (rape, acacia, heather) are covered only qualitatively (Bodor 2022 thesis classes; Vilges model syrups). ## References 1. Incoprea Theory 126 (claim source), `Original Theories Vol XXIV - 2026-10-30/theories/honey-glucose-fructose.html`. 2. Jackson, R.F., Silsbee, C.G. (1924). Saturation relations in mixtures of sucrose, dextrose, and levulose. Technologic Papers of the Bureau of Standards No. 259. https://archive.org/details/saturationrela1924259jack 3. White, J.W., Riethof, M.L., Subers, M.H., Kushnir, I. (1962). Composition of American Honeys. USDA Technical Bulletin No. 1261. https://archive.org/details/usda-composition-american-honeys-1962 4. Schiassi, M.C.E.V. et al. (2025). Evaluation of honeys of different botanical origins stored under average day and night temperatures. Food Science and Technology (Campinas) 45, e552. doi:10.5327/fst.552 — https://fstjournal.com.br/revista/article/download/552/408 5. Omran, N.S., Abdel-Rahman, M.F., Al-Amir, R.O.A., Abouzid, A.K. (2025). Evaluation of the Physicochemical Properties of Selected Honey Types Produced in Southern Egypt and Their Compliance with International Quality Standards. Journal of Sohag Agriscience 10(2):129–140. https://jsasj.journals.ekb.eg/article_461888_3c503b29cae3cb754e07d8b8dafb5d2a.pdf 6. Vilgis, T. (2015). Honig: Zucker an der Phasengrenze. Physik in unserer Zeit 46(6):310. http://wwwhomes.uni-bielefeld.de/~tkohers/Bienenprojekt/Vilges,%20Honig%20an%20der%20Phasengrenze.pdf 7. Kasolia, D.J. (1991). MSc thesis, University of Nairobi (reproduces the glucose–fructose–water solubility discussion: J&S 1924, Lothrop 1943, Kelly 1954). https://erepository.uonbi.ac.ke/bitstream/handle/11295/21780/KASOLIA%2C%20D.J.%20M.%20SC.%201991.pdf 8. Bodor, Z. (2022). PhD thesis, Hungarian University of Agriculture and Life Sciences (crystallization-speed classes for Hungarian honeys; G/W > 2.1 accelerates, after Tosi 2008). https://uni-mate.hu/documents/20123/336900/Bodor_Zsanett-ertekezes.pdf 9. Escuredo, O., Dobre, I., Fernández-González, M., Seijo, M.C. (2014). Contribution of botanical origin and sugar composition of honeys on the crystallization phenomenon. Food Chemistry 149:84–90. doi:10.1016/j.foodchem.2013.10.097 (paywalled; F/G ≤ 1.14 criterion as cited by Schiassi et al. 2025). ## Files - `report.md` — this report - `code/honey_test.py` — solubility fit, index computation, Spearman correlations, threshold analysis, figure - `data/honey_compositions.csv` — 34 composition rows with source column - `figures/granulation_ratio.png` — granulation rank vs G:F, G:W and S with thresholds Working report — not peer reviewed. Educational information, not medical advice. AI-assisted test: DeepSeek.