A site-frequency spectrum is a compact histogram of genetic variation. Even if its expected values were known without sampling error, it would still reveal only finitely many exponential averages of population history. This corrected release asks a sharper question: can those averages determine the mean population size inside a fixed calendar-time window? Under a neutral one-population coalescent model and a broad positive class of histories, the answer is no. Calendar time must be recomputed separately for every candidate history. The package gives two strictly positive histories with exactly the same finite expected spectrum, but their ratios for the same illustrative calendar windows are about zero point four four one and one point seven nine four. One implies a depression; the other implies an expansion. Folding the spectrum loses further coordinates, and the paper characterises the bounded linear functionals that remain identifiable. The priority audit also corrects the earlier framing. Myers, Fefferman and Patterson already established broad same-spectrum ambiguity, calendar versus genetic time and exponential projections. Bhaskar and Song supplied the finite Laplace coordinates and folded parity. This release therefore claims a target-specific sharp-window synthesis, explicit witness and replayable assurance package, not first discovery. The result concerns an exact expected summary. It does not establish equality of linked genomic laws, decide the ancient human bottleneck, or substitute producer replay for independent reproduction, specialist review or peer review.