The Local Shell

01The evidence

• The local distance ladder has survived every attempt to break it. JWST observations by the SH0ES team reproduce a local value near 73 kilometres per second per megaparsec, far from the 67 inferred from the cosmic microwave background.SH0ES, JWST update
• The ladder itself is suspected of environmental sensitivity. Analyses of fifth forces using local distances find that the low-redshift calibration is exactly where a screening mechanism would do the most damage.Physical Review D, local distance ladder

02The tension

Every proposed fix changes the early universe: extra radiation, early dark energy, decaying particles. Each one fits the microwave background slightly worse while fixing the local value. Nobody has demonstrated that the local measurement is wrong, and nobody has shown that early-universe physics can be added without breaking something else.

03The theory

The measurement is not wrong and the early universe is not modified. We are inside a region roughly 300 megaparsecs across with a matter density perhaps twenty per cent below the cosmic mean. Inside such a region, expansion runs fast because the local curvature and the outflow both contribute. Beyond the shell, the metric relaxes to the background value.

This makes the discrepancy a function of distance rather than a constant offset. The local ladder samples the inner part of the shell, where the bias is largest. BAO and supernova samples that reach beyond 150 megaparsecs average over the wall and see something much closer to the cosmic value. The prediction is a smooth decline of inferred expansion rate with redshift, not a step.

The distinctive signature is directional. A shell has a centre, and we are off-centre in it, so the bias must be larger in the direction of the nearest wall and smaller opposite. That produces a dipole in the Hubble diagram at the few per cent level, with an axis uncorrelated with the kinematic dipole of the cosmic microwave background.

The idea is old and has been dismissed on the grounds that voids this large are rare. That objection is weaker than it looks: we have one sample of one, and the observed local galaxy counts already show a deficit. The point is not that the void is certain. It is that the hypothesis makes a sharper prediction than any early-universe fix, and the data to test it arrive with the next supernova survey.

04What this is not

05Predictions

• P1The expansion rate inferred from standardisable candles should fall from about 73 to about 67 between redshift 0.02 and 0.15, a trend detectable with a few thousand supernovae.
• P2There should be a three to six per cent dipole in the Hubble diagram aligned with the direction of the nearest large-scale structure, and it should not match the kinematic dipole.
• P3Measurements that do not use a local ladder, such as redshift drift or cosmic chronometers, should show no tension at all.

06What would refute it

A flat relation between inferred rate and redshift, at one per cent precision, kills it. So does an isotropic Hubble diagram at the predicted amplitude. Either result would send the problem back to the early universe.

07What it would change

It removes the pressure to invent new physics before the universe, changes how local measurements are quoted, and makes the Hubble diagram a map of our own neighbourhood rather than a probe of inflation.