Voting in decertification elections, fiscal year 1952

p. 304. Grid is 11 columns: stub plus two five-column spanner groups ("Elections in which a representative was redesignated" and "Elections resulting in decertification"). This is a single table, not side-by-side panels: the two groups are different measures for the same stub rows, so each row carries all 10 values. Decimal points: in the two "Percent casting valid votes" columns the point survives the scan for 92.1, 87.6, 90.3, 86.9 and 87.0, but is dropped (printed as a gap) in three cells — Total/redesignated percent ("88 1"), Total/decertification percent ("88 9") and Unaffiliated/redesignated percent ("86 6"). I have transcribed these as 88.1, 88.9 and 86.6 on the basis of the column's one-decimal format; the digits themselves are unambiguous. Row labels "A. F. L." and "C. I. O." are followed immediately by leader dots, which swallow some of the abbreviation periods (the image reads roughly "C I. O"); I have written the standard pointed forms. Suspected misprint, transcribed as printed: the last cell, Unaffiliated / "Elections resulting in decertification" / "Votes cast for no union" = 333. It fails both arithmetic checks that every other cell in the table passes — 151 + 333 = 484, not the printed 550 total valid votes cast for that row, and 1,453 + 538 + 333 = 2,324, not the printed column total 2,390. Both checks would be satisfied by 399. The glyph reads as 333 in the image, so 333 is what appears above; a user reconciling totals should treat it as doubtful. All other cross-checks hold: both panels' union + no-union splits equal their total valid votes cast, and all three affiliation rows sum to the printed Total row in all other columns.