The exact page text, read by your browser's voice.
Decennial census returns, instruments labelled; Census Bureau, CPH-2-16, Table 8
The decennial series, with each return's instrument stated and nothing merged: 142 in 1990; 144 in 2000; 133 in 2010 — all three read from the Census Bureau's printed 2010 counts volume — and 118 in 2020 from the PL 94-171 redistricting file. Housing units in the same printed table run the other way: 50 in 1990, 60 in 2000, 63 in 2010. Read whole, the series is a place essentially flat through 2000, then declining across two decades — down some seventeen percent from 2000 to 2020 — on a housing stock that grew: shrinking households, not a shrinking town plat. No decennial count before 1990 was reached in any printed carrier; whether the town was tabulated in the 1950–1980 volumes depends on the unrecovered incorporation date, and the earlier volumes stand as the settler named in the note.
Estimates are kept out of the series, as the house rule requires. The one estimate printed is labelled as one: a 2019 census estimate of 134, carried with its own footnote that it is an estimate and not an actual count. Aggregator figures that conflict with the printed record are rejected as carriers and recorded in the note's conflicts register: one widely copied aggregator tabulates 1990 as 150 and 2020 as 116, against the Census Bureau's printed 142 and the PL count of 118 — its out-year figures are estimates presented in the same column as counts, which is exactly the merge this build does not make.