The 2000 census's 1,809 — the count that was corrected away

The claim: Census 2000 tabulated products, as carried in the reference record's historical table. The test: the Census Bureau's Count Question Resolution record, read whole

The tabulated claim is 1,809 people in 2000 — a 156 percent jump over 706 in 1990, printed in the reference record's own historical table — against the same record's census text: "there were 645 people (corrected count), 311 households, and 171 families residing in the town." The Census Bureau's Count Question Resolution program "handles external challenges to particular official Census 2000 counts of housing units and group quarters population received from state, local, or tribal officials" — and its design explains how both numbers legitimately coexist: corrected CQR counts flow into the estimates base, but, in the Bureau's own words, "the base files for the census will remain unrevised, so that none of the standard Census 2000 data products will reflect the corrections." The tabulated 1,809 is therefore the unrevised product's figure and the 645 the corrected count, and neither is a misprint of the other. What no reached carrier supplies is Roseland's own determination: a full-text search of the Bureau's CQR errata document finds no Roseland, and no Census, press or reference carrier reached names the dormitory, block, tract or boundary error involved. The tabulated count's own profile — two-thirds of the counted population aged 18 to 24, median age 21, in a town whose every other count is a small family town — is printed as internal evidence only; the explanation it suggests, a university population counted into the town, is not reached and not printed as fact. Verdict: the mechanism is settled, the case is not; the pair stands whole and unmerged, exactly as Pass R-1 printed it, with the CQR file for place 18-65988 as its settler (G-FR-5).