Heritage at Risk runs the opposite way to ancient woodland: it clusters in deprived areas.

What we measured

We used Historic England’s Heritage at Risk register for 2022 (the layer was last edited in June 2023), not the current register. The assets we matched were listed buildings, scheduled monuments and registered parks and gardens. We counted sites within 500 m of an at-risk asset, excluding Conservation Areas so they aren’t counted twice. The 500 m radius is an analytical choice, not a statutory distance. 45 of 652 sites (6.9%) fall inside it.

What it shows

16.1% (14 of 87) of sites in the most deprived band sit beside an at-risk asset, against 2.5% (6 of 242) in the most affluent. Odds ratio 7.54 (95% CI 2.80–20.34), Fisher p = 3.2 × 10⁻⁵, robust under whole-thesis Bonferroni. It is the near mirror image of Ancient Woodland, which skews affluent (OR 4.88).

Why deprived areas have more at-risk assets is not tested here. The strongest protections are the preventative ones that cluster in wealth; the register of assets at risk of decay clusters in deprivation. On the 652-site cohort, adding heritage to the composite designation count removes the wealth gap: five protections plus heritage gives U = 9,379.5, p = 0.099; six protections plus heritage gives U = 9,609.5, p = 0.20. If flood zone is also counted as a hazard layer the gap partly returns (p = 0.028). That decomposes the gap rather than overturning it.

Caveats

The affluent cell is only 6 sites, although the confidence interval excludes 1 by a wide margin. The 500 m radius is an analytical choice. Re-queried at 250 m and 1,000 m the skew holds: 6 of 87 against 1 of 242 sites at 250 m (odds ratio 17.9) and 26 of 87 against 28 of 241 at 1,000 m (odds ratio 3.2). The composite with heritage loses its gap at 500 m and 1,000 m, but not at 250 m for five designations (p = 0.013). Each site is placed at the postcode centroid given on the campaign form, so these distances carry a position error that we have not measured.

Revised 12 November 2026: The composite with heritage is recomputed on the 652 sites (the earlier v5 figures are replaced). The earlier claim about 250 m and 1,000 m radii is withdrawn until it can be re-run.

Corrected again 12 November 2026 after a check of the data pipeline: the register is the 2022 one, not the live one; the 250 m and 1,000 m check has now been run; an unsupported explanation for the link with deprivation is removed.