The exact page text, read by your browser's voice.
What it is: what this build could not settle, it did not smooth over: it named the gap, numbered it in its owning leg's series (GAP-SPINE-2, GAP-FARMS-1, GAP-BMV-6 …), printed the state of the record, and named the source that would settle it. The harvest gathers them in one place: The rows are classed by their settling source's door: NEWSPAPER 26 · ANCESTRY 4 · OPEN-WEB 36 · COUNTY-RECORD 32 · OTHER 76. Duplicates are merged by the catalog's own rule (one row per gap; restatements cross-referenced, not re-rowed), settled items are not rows, and cells the Population file rules "not gaps in the story" are excluded on its say-so.
Its standing, from this build's experience: the harvest is a pointer, and this apparatus follows its own rule: the rows are NOT reprinted here. The harvest is built to merge into the History program's Pending Research Catalog, in that catalog's row shape, when the parent merges it — and until then it is the single index of everything this history still owes.
Where it fails: a harvest can only gather gaps the passes thought to name. The method's limit is stated in the open: a question no pass asked does not appear, and the residue entry below is the build's honest attempt at the largest of those.
Sources:
- granger-full/.