Character and results of representation elections conducted during the fiscal year, by regions

p. 36. 21 columns. Column 6 ("Total") is placed under the "Number of employees" spanner, not under "Valid votes cast": the one-line heading "Number of employees" is about 228 px wide at this scan's scale, wider than the two-column span (~185 px) and matching the three-column span (~279 px), and its text sits entirely to the right of the group rule after "Ordered". Note for the reader that this "Total" column is nevertheless arithmetically the total of the five valid-vote components to its right (e.g. Boston 18,698+1,981+1,703+7,579+838 = 30,799; column total 532,355), and the "Percent of total votes" columns are percentages of it. Two cells where the scanned glyph and the table's own arithmetic disagree, and I transcribed the arithmetic-supported digit: - Indianapolis, "Number of employees / Voting": the scan reads as 22,020, but I entered 22,620. 22,020 is impossible because the same row's valid votes (22,048, confirmed by its own components and by the 532,355 column total) would exceed it, and 22,620 makes the Voting column sum to the printed 540,544 exactly (22,020 leaves it 600 short). - St. Louis, "Against single union on ballot / Number": the scan reads as 1,009, but I entered 1,099. 1,099 makes the row sum to the printed valid-votes total 4,222 (1,009 leaves it 90 short), makes the column sum to the printed 52,063, and matches the printed percent 26.03 (1,099/4,222 = 26.03; 1,009 would be 23.90). Both are 1-bit-scan digit degradations (6 read as 0, 9 read as 0) rather than misprints, but they should be spot-checked against the page. Every other value is cross-checked and consistent: all three election columns sum to 1,192 / 676 / 516 and Consent+Ordered equals Total in each of the 22 regions; Eligible to vote sums to 595,075; valid-votes Total and all five component columns sum to their printed totals; each region's five components sum to its valid-votes total; and the four "Result of elections" columns sum to 793/45/83/271 and, row by row, to that region's number of elections. The small raised digits before the "Number of elections / Total" values are faint. I read them as 3 for Boston, New York, Buffalo, Philadelphia, Baltimore, Detroit, Cincinnati, Indianapolis and Milwaukee; 2 for Cleveland and Atlanta; and two markers, 2 and 4, for Pittsburgh (written 27^2^4). Atlanta's marker is the least legible and could be a 3. Three cells in "Against contesting unions / Percent of total votes" read one hundredth above the value implied by the printed numerator and denominator: Atlanta 1.88 (261/13,953 = 1.87), Los Angeles 3.45 (480/13,954 = 3.44) and Denver 3.25 (80/2,467 = 3.24). These may be last-digit misreadings on my part or the compositor's rounding; other cells in the same column round both up and down inconsistently (New York 6.16 for 6.166, Baltimore 1.21 for 1.217, Minneapolis 1.08 for 1.086, San Francisco 10.25 for 10.256), so inconsistent rounding is present in the original either way. The "Board^1" row is printed with dash rules across every column and carries no values. Blank cells elsewhere (e.g. the National unaffiliated columns for Milwaukee, Fort Worth, Kansas City and Minneapolis, the Local columns for Seattle, and several "Result of elections" cells) are printed as dash rules and are given as empty strings.