Elections conducted by the National Labor Relations Board during the fiscal year ending June 30, 1939, by regions

p. 60. 19 columns (stub + 18). The stub has no printed heading; the first body line "Region:" is a label-only section line. Empty cells in the body are printed as short dash rules. AMBIGUOUS GLYPHS RESOLVED BY INTERNAL ARITHMETIC. This is a 1-bit scan in which "9" and "0" (and in one case "6" and "0") are hard to separate. For seven cells my first visual read gave a "0" that every available internal check contradicts; I transcribed the arithmetically confirmed digit and list them here so they can be re-checked at higher magnification: - 1. Boston, Valid votes cast Total: read as 14,107, transcribed 14,197 (row components sum to 14,197; column sums exactly to the printed 177,215; the printed 54.6 and 31.5 percentages both require 14,197). - 18. Minneapolis, Valid votes cast Total: read as 5,303, transcribed 5,363 (row sum, column sum, and all three printed percentages require 5,363). - 19. Seattle, For trade unions: read as 7,500, transcribed 7,590 (column sums exactly to the printed 121,643; row sum and the printed 95.2 require 7,590). - 20. San Francisco, Against all unions: read as 1,809, transcribed 1,899 (column sums exactly to the printed 34,085; row sum and the printed 30.9 require 1,899). - 20. San Francisco, Number of elections Total: read as 20, transcribed 29 (= Consent 12 + Ordered 17; column sums to the printed 746; elections won 18+1+0 plus lost 10 = 29). - 13. Chicago, Percentage against all unions: read as 10.3, transcribed 19.3 ((1,314+378)/8,749 = 19.34). - 15. New Orleans, Percentage for trade unions: read as 80.5, transcribed 89.5 (8,131/9,083 = 89.52). Supporting evidence: with these values, ten columns (Total/Consent/Ordered elections, Voting, Valid votes Total, For trade unions, National, Local, Against all unions, For neither, and all three elections-won columns plus elections lost) each sum exactly to their printed totals, and every row's vote components sum exactly to its row total. UNRESOLVED DISCREPANCY. The "Number of employees eligible to vote" column as transcribed sums to 206,697, which is 900 less than the printed total 207,597. A shortfall of exactly 900 points to one cell whose hundreds digit is a 9 I read as 0; the candidates are 2. New York 46,063, 5. Baltimore 18,011, and 13. Chicago 10,055. I could not tell which, so all four cells are transcribed as read. Every other column in the table reconciles, so this is the one column to verify. Minor rounding mismatches between printed percentages and the components (all transcribed as printed, none affecting the digit choices above): 1. Boston unaffiliated 13.9 vs. computed 13.84; 3. Buffalo trade 56.4 vs. 56.46; 7. Detroit unaffiliated 5.7 vs. 5.64; 21. Los Angeles unaffiliated 1.5 vs. 1.44. The source appears to mix rounding and truncation. Scan specks, not footnote markers: a small mark sits immediately left of "1,500" (3. Buffalo, Voting) and left of "276" (3. Buffalo, Local), a stray mark appears under 5. Baltimore in the percentage-for-unaffiliated column, and a dot precedes the "1" in 20. San Francisco's elections-won National cell. The column totals confirm the underlying values in each case, so I treated all of them as dirt.