Anatomy of an Excess: The q = 0.50 Residues, Taken Apart

Paper #144 computed the scatter of the Closure Identity's residues and found the author's sentence standing: the residue factor is a draw, not a constant. But #144 left one form with two soft spots, and it named them. At q = 0.50 — the shallowest form in the family — the residues centre at −0.070 dex, the closest any form comes to demanding a constant, and 0.048 dex of their scatter sits unexplained above the measurement budget. This paper takes that form apart. I rebuilt the residues with the deposited machinery, checked the rebuild against the printed record, and formed the difference series: each galaxy's q = 0.50 residue minus the median of the other six forms' residues on the same galaxy. The difference centres at −0.060 dex, and it is not a scatter problem and not a data problem. It is the form's own construction, priced to the third decimal: the difference is, galaxy by galaxy, the form's correction-factor advantage over its siblings minus the larger frozen intercept its own fit set — two terms that nearly cancel and do not quite. Worse, the centring is not even a shift. It is a tilt. It runs from −0.127 dex in the shallowest quarter of the sample to +0.034 dex in the deepest, because this form's correction rises faster with depth than any sibling's while its intercept was frozen highest. No artifact covariate — inclination, distance, mass error — comes within reach of the bar. And the excess scatter does not localize: in the difference series there is no excess, in either half of the sample, because the noise the forms share cancels when you subtract them. The residue #144 named has an address. It lives in the seam between the form's two ends.