Analysis of Elections Conducted in Representation Cases Closed, Fiscal Year 2005

p. 136. Grid is 17 columns: the stub plus four 4-column panels (All R, RC, RM, RD elections), each panel spanned by "Elections conducted". These are repeated column groups of one table, not side-by-side panels, so no stacking was needed. Dash cells: the table prints a short double hyphen ("--") in many cells, used as a "not applicable" marker that is deliberately contrasted with a printed "0" elsewhere in the same row (e.g. the "Rerun required" line under "All representation elections" has "--" in the All R / RC / RD rerun columns of some panels but "0" in the RM rerun column). Because that distinction carries meaning, I transcribed it literally as "--" rather than treating it as a blank dash rule. If the convention for this corpus is to blank such cells, every "--" above should become "". Footnote: the caption carries a superscript 1 ("Fiscal Year 2005^1"), but no footnote text appears below the table in this crop — the image ends at the bottom rule of the last data row. The footnote text is presumably cut off or printed elsewhere, so `footnotes` is empty. Doubtful cells (all small, low-contrast marks in the "Resulting in a rerun or runoff" columns of the bottom two blocks): under "Board–directed", the "Rerun required" and "Runoff required" rows read "0" in the All R rerun column but "--" in the RC, RM and RD rerun columns; under "Expedited–Sec. 8(b)(7)(C)", the same two rows read "0" in both the All R and RM rerun columns and "--" in the RC and RD rerun columns. The 0-vs-"--" calls in these four rows are the least certain part of the transcription. Row labels use en dashes as printed ("Regional Director–directed", "Board–directed", "Expedited–Sec. 8(b)(7)(C)"). Nothing was computed or corrected; no printed total appeared inconsistent.