Results of Rerun Elections Held in Representation Cases Closed, Fiscal Year 1995

p. 161. Eight data columns (four spanner heads, each over Number / Percent by type); the stub has no printed heading. The scan is a 1-bit image in which many decimal points have dropped out, so several percent cells print as e.g. "100 0", "66 3", "40 4", "38 0"; these are decimal points and are transcribed as "100.0", "66.3", "40.4", "38.0". Doubtful cell: row "RC cases", column "Outcome of original election reversed / Percent by type" — I read the glyphs as "38 0" (transcribed 38.0), but 33 of 83 is 39.8, and every other reversal percent in the table is internally consistent (38/94 = 40.4; 5/11 = 45.5; 26/69 = 37.7; 11/24 = 45.8), so this is either a printed misprint or an 8/9 confusion in the degraded scan; transcribed as read, not corrected. All em-dash rules in percent columns of the 0-valued rows (RM cases, Expedited elections, Board-directed elections) and in the "No union chosen / Percent by type" cell of "Consent elections" are rendered as empty cells. The section line reads "By type of case" followed by a faint colon-like mark, transcribed as a colon to match "By type of election:". A small ink speck sits above the "67.0" cell in the "All representation elections" row; it is dirt, not a footnote marker.