The census volumes

Read from rendered page images, not from machine text — the rule the Farms leg set

How they were read: the census volumes behind the Farms and Industrial Revolution files were read from rendered page images of the printed volumes, because the machine text under those images cannot be trusted with a table. The test case stands in the Farms file: the 1954 Census of Agriculture, Volume I, Part 4 (Indiana) — Internet Archive item unitedstatescens14unit — has a text layer so column-scrambled that every figure in the file was read from the page images and cross-checked by the tables' own arithmetic before it was printed. The same discipline carried the Census of Manufactures' city rows for 1890, 1900 and 1909 in the Industrial Revolution file — 1890: 208 establishments, $9,812,518 of products; 1909: South Bend second in Indiana manufacturing — read from rendered images of the volumes themselves.

The population series: the decennial counts and their composition rest on the Bureau's published working apparatus: Gibson and Jung, Census Bureau Working Paper No. 76 (February 2005), Table 15, from which the Population file's series was settled, and against which the Black decade series was cross-checked, decade by decade, with Carol O. Rogers, "Black and White in Indiana," Indiana Business Review (Summer 2005) — every percentage matched, and the match is printed in the Black History file because a checker deserves to know the two series were compared, not merely copied.

The holes: two volumes this build needed are not on the open shelf, and the Population and Farms files leave the gaps open at exactly their size. The country-of-birth tables — the 1920 Census, Volume III, Indiana chapter, and the 1910 Volume I equivalent — would settle the Polish-, Hungarian- and German-born counts for the city's two peak immigrant censuses; the Bureau's own chapter PDFs for the Indiana span return 404, and the copies of the volume found on the Internet Archive proved on inspection to be manufactures bulletins wearing the volume's name. The pre-1950 county farm tables (the 1900 Twelfth Census, Volume V; the 1920, 1930, 1940 and 1945 Volume I Indiana parts; the 1950 preliminary county report in the Bureau's ac50 series; and the 1954 volume's County Table 4, printed at pages 70–71) are in the same state. One HathiTrust or Allen County Public Library pull settles each of them; they stand consolidated on the pull list below.

Sources:

1. 1954 Census of Agriculture, Vol. I, Part 4 (Indiana), Internet Archive item unitedstatescens14unit — page images read; text layer documented as scrambled.
2. Census of Manufactures volumes, 1890, 1900, 1909 — city tables read from rendered images.
3. Gibson & Jung, Census Bureau Working Paper No. 76 (2005), Table 15; Rogers, Indiana Business Review (Summer 2005).