The Look-Elsewhere Ledger: A Trials-Factor Accounting for the Large-Angle Anomaly Catalogue as a Whole

The large-angle anomalies are usually argued one at a time, each with its own percentile, and the catalogue as a whole is then described as either damning or negligible according to taste. This paper does the catalogue-level arithmetic. Taking five headline anomalies at their calibrated marginal percentiles, from 0.005 for the correlation deficit to 0.06 for parity, a Fisher combination under an independence assumption gives 0.00002, a figure this paper prints only to set aside, because the statistics share the same few low-multipole degrees of freedom and are not independent. The honest calculation is a trials accounting: in a pool of one hundred anomaly definitions examined on one sky, chance alone supplies on average 5.0 results below the five per cent level and produces at least one result at the 0.005 level with probability 0.3942. The catalogue's true surprise is therefore concentrated in its single strongest member, and the correct collective statement is that the sky contains one strong anomaly and a supporting cast of the size chance predicts. The study is computed, not measured.