Industrial distribution of decertification elections, winner, eligible voters, and valid votes cast, during fiscal year 1950

p. 251. CROP IS INCOMPLETE. The table continues below the bottom of the image: the "Total" line gives 112 elections and 9,474 eligible voters, but the industry lines shown (Manufacturing 80 + Mining 3 + Wholesale trade 11 + Retail trade 7 = 101; eligible voters 8,152 + 150 + 385 + 378 = 9,065) fall short, so roughly 11 further elections' worth of industry rows are cut off. The Manufacturing block itself is complete (its 17 detail rows sum to 80). The footnote keyed to the "Industrial group ^1" stub heading is also below the crop, so `footnotes` is empty. DECIMAL POINTS. This scan drops or nearly erases the period in a large share of the one-decimal percent figures, so cells print as e.g. "10 7", "71 4", "25 0", "66 2". These are decimal points in the original (every percent column in this table is carried to one decimal, and each reconstructed value checks out against the counts and the 100.0 totals), so they are transcribed as "10.7", "71.4", "25.0", "66.2", etc., in the same spirit as the space-after-comma rule. No value was otherwise altered. ZERO CELLS. The table mixes a bare "0" and a leading-point ".0" in the percent columns, and at this resolution the leading point is the single least reliable mark on the page. Each zero cell is transcribed as the dot appears to me, but any individual "0" vs ".0" call in the four Winner percent columns should be treated as low-confidence. Three such cells I read as having a point that the surrounding rows would not predict: Fabricated metal products, C. I. O. Percent (".0"); Professional, scientific, and controlling instruments, C. I. O. Percent (".0"); Retail trade, C. I. O. Percent (".0"). PRINTED AS SHOWN, NOT CORRECTED. Two cells depart from what the counts would give and are left as printed: Primary metal industries, A. F. of L Percent reads "33.4" where 1 of 3 is 33.3 (the C. I. O. and No union cells in the same row both read "33.3"); Retail trade, Elections Percent reads "6.3" where 7 of 112 is 6.25, while Automotive transportation equipment, also 7 of 112, reads "6.2". Lumber and wood products, Elections Percent reads "5.3" where 6 of 112 rounds to 5.4. Two values whose decimal point is wholly invisible and were set from column context: Apparel, Valid votes cast Percent (read ".4"; 36 of 8,485 = 0.42) and Leather / Fabricated metal / Professional, Elections Percent (read ".9"; 1 of 112 = 0.89). HEADINGS. "Elections", "Eligible voters" and "Valid votes cast" each carry Number/Percent pairs set across the lower two heading levels; those have been repeated down both levels, as has the stub heading, so that all three heading rows have 15 cells. The "A. F. of L" heading is transcribed without a closing period after the L, as printed.

p. 252. Continuation crop: the page carries no column headings (the table starts on an earlier page), so header_rows is empty. Column count (stub + 14) was taken from the vertical rules; the columns run as seven number/percent pairs. The scan is 1-bit and decimal points are frequently thin or dropped. Points restored where the printed value is unambiguous: col 10 of "Highway freight transportation", "Water transportation" and "Warehousing and storage" images as "100 0" and is transcribed 100.0; col 4 of "Heat, light, power, and sanitary services" images as "33 3" and is transcribed 33.3; col 2 of "Warehousing and storage" and of "Services" images as a bare "9" and is transcribed .9 (consistent with the other single-case rows and with 1 of 112). Col 6 (the percent beside col 5) prints as a bare "0" on the "Transportation, communication, and other public utilities", "Heat, light, power, and sanitary services" and "Services" lines, while every other row prints ".0" there; this may be the same dropped-point artifact. The value is zero either way, and the cells are transcribed as they image. "Services", col 5: a small speck sits to the left of the figure; it is read as a stray mark and the cell as "0". The row's own breakdown (0 + 0 + 0 + 1 = 1) supports counts, not a percent, in cols 5 and 7. "Warehousing and storage", col 12: the cell carries the parenthesized footnote reference (2) — "Less than 0.1 percent" — in place of a figure; transcribed (2), the raised 2 being the whole cell rather than a marker attached to a value. Doubtful cell: "Warehousing and storage", col 14 reads ".1". On that column's own base (4 against a total near 8,500, since 206 is 2.4 percent) it should fall under 0.1 percent and show (2) as col 12 does. Transcribed as printed; possible misprint. A double rule separates "Services" from the block above; "Services" is a major-division line at the same indent as the first row, not a subitem of it. Arithmetic checks on the indented block: cols 11 and 13 sum to the 395 and 358 printed on the "Transportation, communication, and other public utilities" line, and cols 5, 7, 9 and the col-3 entries sum to its 10. Note that the subitem percents in col 2 (.9 + 1.8 + .9 + .9 + 1.8 + 2.6 = 8.9) agree with the division line, though 3 of 112 would round to 2.7 rather than the printed 2.6; left as printed.