Size of Units in Representation Election Cases Closed, Fiscal Year 1972

p. 277. 15 columns: stub, 4 single columns, four two-column union groups under the spanner "Elections in which representation rights were won by", and a final two-column group. The crop shows only panel A; the table continues (panel B etc.) on material outside this crop. Doubtful cell: the "Other local unions / Number" figure in the "Total RC and RM elections" row is a degraded glyph that could read 237 or 287. I transcribed 237 because the column's 30 size-class entries sum to exactly 237, the row's five union/no-representative numbers then sum exactly to the printed 8,472, and 49/237 = 20.7 matches the printed first-row percent (49/287 would be 17.1). The scan drops many decimal points in the percent columns (e.g. "3 8" and "77 1" in the 50 to 59 row, "94 0" in 190 to 199, "99 8" in 1,000 to 1,999, "0 8" in 10,000 and over). Every such cell was transcribed with the decimal point that the column's typography requires; no digits were altered. Cells printed as a dash rule are given as "": the Cumulative percent cell of the total row, and the "Percent by size class" cells wherever the adjacent count is 0 (80 to 89 and 180 to 189 other local; 190 to 199, 800 to 999, 2,000 to 2,999, 3,000 to 9,999 and 10,000 and over Teamsters; 2,000 to 2,999 AFL-CIO; 3,000 to 9,999 and 10,000 and over other national; 500 to 599 other local; 10,000 and over no representative). Arithmetic checks that passed exactly against the printed totals: Number eligible to vote (570,846), Total elections (8,472), AFL-CIO Number (2,583), Teamsters Number (1,461), Other national Number (372), Other local Number (237), No representative Number (3,819), and the Percent of total column against the printed Cumulative percent column at every row. A few printed percents do not match the ratio they should express, transcribed as read: "10 to 19" other local 17.4 (41/237 = 17.3); "20 to 29" no representative 12.6 (484/3,819 = 12.7); "30 to 39" no representative 8.0 (310/3,819 = 8.1); "200 to 299" no representative 3.3 (130/3,819 = 3.4). The other local percent column sums to 100.4 as printed.

p. 278. Structure: 15 columns (stub + 14). The crop prints no column headings — it continues a table whose headings are on the preceding page — so header_rows is empty. From the data, the 14 columns are: a count column (total 20,790, apparently employees eligible), then a number/percent/cumulative-percent trio for all RD elections (total 451), then five number+percent pairs with totals 79, 36, 13, 6 and 317. Those five sub-totals sum exactly to 451, and each column's body sums to its printed total (20,790; 451; 79 — see exception below; 36; 13; 6; 317), which was used to confirm the column assignment of every value. "B. Decertification elections (RD)" is printed centred above the table's top rule, not in the stub; it is transcribed as a section row with empty value cells. Doubtful / inconsistent cells: - Row "800 and over", 5th data column (the count column whose total is 79): the glyph reads as 5, but the percent printed beside it (1.3) and the column total both imply 1. The column as transcribed sums to 83, not the printed 79. Transcribed as read (5); this is the single arithmetic break in the crop and is either a source misprint or a misread of a degraded digit. - Row "30 to 39", 10th data column (percent of 13): printed 15.3, while the other two cells in that column with a count of 2 (rows "90 to 99" and "300 to 399") print 15.4. Transcribed as printed. - The 1-bit scan drops or faints some decimal points: "60 to 69" percent-of-451 appears as "5 1", "70 to 79" as "3 8", and "500 to 799" as "0 7". Written with the decimal point (5.1, 3.8, 0.7), matching every other cell in that column and the cumulative column. - The last two columns (count 317 and its percents) are the most degraded part of the scan; 3/8 and 6/8 are hard to separate there. Readings were resolved against the counts (e.g. 102/317 = 32.2, 84/317 = 26.5) and all 21 percentages then agree with their counts. - Dash rules ("------") in percent columns beside a zero count, and in the cumulative-percent cell of the total row, are transcribed as "". Size classes "140 to 149" and "400 to 499" are not printed on the page (no elections in those ranges); the rows run 130 to 139 → 150 to 159 and 170 to 199 → 200 to 299 → 300 to 399 → 500 to 799 as shown.