How this history treated its sources — and the trap with four doors

The method, stated once, at the end of the shelf it governed

First-hand, or flagged: this history was built by reading its sources, not summaries of them — county histories in their scans, census volumes from their page images, opinions in their full texts, finding aids in the archives' own words. Where a source could not be read first-hand, the file says so at the citation: one Michigan Road citation carried from search-surfaced text and flagged as a re-read item; one newspaper article printed attributed from behind its paywall; one study's back half unread after a failed fetch. The flag is part of the citation, and a checker should treat an unflagged claim here as one the build read where it says it read it.

Conflicts printed, never averaged: where two sources in this section disagree, both stand. The centennial volume is dated 1922 in the catalogs and 1923 in this build's own Events citation, and Part One prints both dates with their homes. The Library of Congress dates its county map 1866 against an 1863 imprint, and the Maps file prints both. The library's historian is Mary Waterson in the catalogs and Edith C. M. Waterson in the library's own timeline, and the archives entry above prints both. The executions record puts a "Henry Jones" in Marion County where the Department of Correction puts a "Harry Jones" in St. Joseph County, and the Crime file prints the pair unresolved. The Dillinger take stands printed as a conflict — $28,439 by the Associated Press against $29,890 in Newton's account — because the record contains both numbers and no honest arithmetic makes one. Even the city's own peak-era counts carry a live pair: the Story of South Bend section file's 1970 population of 125,850 against the census volume's 125,580, which this history's assembly conforms to the volume. Averaging a conflict manufactures a third fact that no source holds; this history declines to manufacture.

Gaps at true size: what the record would not yield is printed at the size of the failure — a section that found nothing prints its nothing, a sole carrier is named as a sole carrier, a rejected source is named as rejected. Three stand for the class: an invented founding story for the zoo, circulated by a travel aggregator, was tested and excluded; an uncorroborated fraud article was excluded for want of any second record; and the advocacy literature on a contested modern case was refused in both directions, the file resting on the court and the prosecutors instead. Negative results are results: the Serial Killers file's three tests all came back negative, and the file is the record of the tests.

The four South Bends: the last hazard is a name. There are four South Bends in the American record, and this build met all four, usually mid-sentence in somebody else's database. South Bend, Washington, on Willapa Bay in Pacific County, is the dangerous one, because its record is real and parallel: its Carnegie library stands in federal and museum records that a careless search returns for Indiana's — the Libraries file caught the National Park Service and CLIO records doing exactly that — its First National Bank (charter 4467) is the South Bend bank of the famous 1895 failure in McClaine v. Rankin, 197 U.S. 154, which a Federal Reserve-derived compilation misfiles under Indiana, so that the Banks file's Indiana failure record had to be rebuilt from the OCC list to keep a Washington bank out of it; and its dead stand in the casualty lists beside Indiana's, close enough to add a thirty-second name to a memorial search. South Bend Township, Indiana County, Pennsylvania, is the fiction trap: the "vanishing man" Oliver Lerch, who walks out of South Bend in a famous tale, traces to Ambrose Bierce's fiction and to that Pennsylvania township — the Unsolved file prints the case as literature that wandered into the record wearing this city's name. St. Joseph, Michigan, forty minutes north, is the twin-city trap: a United Way site for Michigan's St. Joseph was caught in the Charities file's searches (none of its material was used), a sheriff candidate of Michigan's St. Joseph County surfaced in the Public Servants file's (he is not printed), and Michigan's St. Joseph County has its own Lewis Publishing county history (Cutler, editor, 1911) in the same dress as Howard's — same publisher, same title form, wrong state.

Two smaller traps of the same kind stand beside them: Atlanta's St. Joseph's Hospital School of Nursing (1902–1969) is not South Bend's school (1907–1952), and the Saint Mary's College Museum of Art that search returns first is in Moraga, California, not at Saint Mary's College here. The rule the build worked by, offered to the next checker: a South Bend fact is not a South Bend fact until its state, county and institution have each been read, in the source, with the name.

Where a checker starts: with the doors in the order this history found them worth the walking: the coroner's index and the Tribune photographs at Local & Family History; the Hesburgh Libraries' complete Tribune run for everything the mid-century files left thin; the League collection at the History Museum and the notebooks at Hesburgh, kept separate; the pull list above, which is the map of the record's remaining dark. The shelf is open. Check us.

Sources:

1. south-bend-full/ — the conflicts, refusals, exclusions and traps as documented in the Chapman, World Famed, Libraries, Banks, Crime, Unsolved Mysteries, War Heroes, Charities, Public Servants, Hospitals and Museums entries and files.
2. Comptroller of the Currency, Alphabetical List (1917) — charter 4467 against the Indiana charters; McClaine v. Rankin, 197 U.S. 154.