hegnauer.family
hegnauer.family

The workshop

Newspapers captured into the record index

· 7 min read · 25 people named

Offline data task, no browser. The compiler's rule of 27 Sep 2026 is that every Hegnauer mention found must be in the study's index of records. A mention is linked to a person where the evidence supports it; otherwise it is kept unlinked with a note. This covers every mention, not only those about the person being researched. No master, platform tree or export was edited, and nothing was committed.

Result

  • 616 rows added: O06098–O06713. They came from the study's harvest files through python tools/add_occurrences.py. That script calls occ_ids.assign(), which updated the study's data. (the study's tools has no command-line entry point, so it did not need to be run separately.)
  • Links: 564 rows are linked (449 confirmed, 99 probable, 16 possible) and 52 are unlinked. The rows cover 321 Newspapers.com page images and 136 study persons.
  • Check: python tools/check.py gives 0 errors. Its 5 warnings were there before this work.
  • Newspaper rows in the index: 5 before this capture (O06093–O06097), plus 7 Ancestry-tree rows that cite Newspapers.com but have no image URL. There are now 621 newspapers.com rows.
Source readRowsLinkedUnlinkedImages
Masters, people/H*.md (74 image URLs, plus the plain-text "page image 183512513" of alice1935)236235172 new + 3 already indexed
(harvest and review §8)1237548118
(appeared during this task; 72,142 bytes, md5 b2d50a79…, 20:25)1951950110
(Rochester and Akron notices 1d–1i)444407
the study's list of leads (Edward Hagnauer, 1898)1011
Pages recorded only in sources/ancestry/* and1715213

Proposals for Luzius Hegnauer and Anna Margreth Bernet, from Newspapers.com adds no image that is not already in the harvest file.

By decade (linked / unlinked):

DecadeLinkedUnlinked
1860s11
1870s40
1880s1210
1890s8134
1900s886
1910s1830
1920s491
1930s530
1940s140
1950s270
1960s140
1970s70
1990s150
2010s130
2020s30

By paper (rows):

PaperRows
Appleton City Journal181
Rich Hill Mining Review30
The Republican Press30
Bates County Democrat29
The Butler Times-Press / Bates County Record24
Butler Daily Democrat23
The Times-Union (Rochester)23
Carthage Evening Press20
Democrat and Chronicle19
Unidentified Butler paper (Luzius §3b-22)17
Longview Daily News16
The Rich Hill Tribune16
The News Tribune (Tacoma)13
The Herald-News (Joliet)13
Rockville Booster11
55 other titles1–9 each

Most-linked people:

PersonRows
H0003 Leonhard145
H0010 Christian34
H0012 Leonard29
H0007 Christian24
H0038 Susanna18
H0017 Robert L.18
Working notes How the rows were made
  • Columns:
    • source_site is newspapers.com.
    • url is https://www.newspapers.com/image/<id>/.
    • collection is Newspapers.com: <paper>.
    • citation gives the title, the paper with its place, the date and the page. It is the same for every row on one article.
    • date is the publication date (D Mon YYYY).
    • place is the place of publication.
    • event is the kind of item, for example "obituary of father Martin".
    • role is the person's relation to the item's subject.
    • surname_as_written and given_names are as printed. Brackets mark supplied text, for example [Hegnauer] when only the given name is printed, and styles are kept: Fred (Mrs.), (wife of Criss).
    • notes give the quote, the master key or the file's item number, and the grade source.
  • One row per person named as an individual in the item. People cited only as context get no row: for example, H0001 and H0036 on their children's obituaries, or survivors who are not named.
    • Named only in the spouse's style: a husband named only this way ("Mrs. John Grill") gets no row of his own. His ID is put in the wife's note. This affects H0070, H0253, H0095, H0318, H0017 (1911), H0044 (1916), H0256 and H0375.
    • Coverage: 65 master citations have no own row for these reasons. Every one of the 75 master-cited images has at least one row.
  • Grades:
    • Masters: taken from the master's own reasoning. A plain statement gives confirmed; "probably" or "taken to be" gives probable; "possible, not proven" gives possible.
    • Harvest files: certain → confirmed, probable → probable, possible → possible. Where the file names a person but gives no grade, the row is probable, or confirmed when the item's own details are unique (full name, lot number, Co. E 4th Mo.).
    • Unlinked: items marked "lead", "or", "unidentifiable" or "another family" in the file.
  • Items left out:
    • Items about other surnames: the Kraettli "Luzi", Luzi Allemann, Mrs. Mary Luzius and Domeyer items.
    • Non-study people in the notices.
    • Living people who are withheld in the masters.
  • Reprints:
    • Barbara Kamm's 1923 obituary:
      • Butler Daily Democrat printing: indexed for her only.
      • Bates County Democrat printing: its survivors come from the Ancestry index of that page.
      • Republican Press printing ("shortened"): indexed for her only.
    • Sylvester Krattly's reprint in the Carthage Press (18 May 1916): indexed for him and Christine only.
Working notes For the lead to decide

  1. Ancestry-tree duplicates. These pairs were not added again, although the url + person dedupe would have allowed it:

    These rows have no URL. Adding the image URL by an edits/*.json entry is preferable to a second row.

    Done 27 Sep 2026 (lead decision): the study's edit files sets the URLs. O00131, O00132, O00235 and O00236 get image 201681616 and O00133 gets image 688565858, with the paper and page in their notes. O00118 (1968) and O00134 (1955, clipping 184336216) still have no page-image URL.

  2. Unlinked pages that need re-reading. Each page carries one of several items, and the harvest does not say which:
    • Luzius §3b-20: five pages from Apr–Jun 1893 ("Chris … sick" or "Leonard … Chicago").
    • Luzius §3b-22: 17 pages from 1893–97, Leonhard or the grandson Leonard.
    • Luzius §3b-35: six 1900 "township officers" pages, H0003 or H0012.
    • The 1896 Bryan-club item, 999415845.
  3. Identity questions carried in the notes:
    • Thelma E. Drees is linked to H0549 Ottilie as probable. H2184 Tillie (b. c.1898) looks like the same girl: a possible duplicate person.
    • "Miss Kratly", 1927: unlinked.
    • Marie Hegnauer "of Pierce City", 1913: H0023, possible.
    • "John Hegnauer" at Highland, 1881: H0003, possible; H0002 is the alternative.
    • "R. L. [sic] Hegnauer", corn exhibitor 1912: H0003, possible, or H0017.
    • "Mrs. L. Hegnauer's quilting", 1916: H0443, probable, per the file's "probably the grandson's wife".
    • "Mrs. Richard W. Yeo": H0252, possible (index only).
    • "Will Hammer", 1923: H0061, possible.
    • A6, 1885: the review said probably Leonhard; the Leonhard file says possible. Kept at probable.

  4. Dates inferred from the order of a harvest list, not read:

    • Luzius §3b-15 (first and last ACJ pages) and §3b-32 (Henry County).
    • Leonhard #52, #60, #63 and #92 (later insertions of notices).

    Each is marked in its notes.

  5. Mentions found but not capturable (no image id), for a browser pass:
    • Elided runs: the Westliche Post / Mississippi Blätter "Dr. N. Hegnauer" advertisement (35 more issues), ACJ 1891–96 (27 more), the Rich Hill letter list of 1895 (10 more), the 1896 Butler Times-Press candidate card (3 more), and the 1908–09 Rockville Booster bank advertisement (about 10 more).
    • Luzius §3b-21: the grandson's university items.
    • Leonhard §5: "W. F. Grob and Leonard Hegnaur", ACJ 7 Jul 1892, p. 5.
    • Leonhard §6: nine hits the viewer could not show (999286386, 999649521, 863280329, 863281573, 863282723, 863058376, 863287596, 1000263915, 999418556).
    • Leonhard §8: about 150 tier-2 pages not opened.
    • Master citations with no image link:
      • H0012 oregonpress and acj1891 (the latter is probably image 863245385, which is linked as possible).
      • mobit1978, Hazel L. Curtis 1978, which names "Adelaid H[egnauer]".
      • H0017 sbsun1955 and rg1968, already O00134 and O00118.
      • The Ancestry obituary-index-only citations on H0297–H0299.
  6. Living person. H0028, the 1977 engagement notice, is indexed with minimal data. Its citation does not name the fiancée. The living sons named in the 1995 and 1999 obituaries are not indexed.
Working notes Procedure for later harvest files (repeat exactly)
  1. List families/research/newspapers/*.md and any other research file that holds Newspapers.com items. Compare each with the table above, or with its md5, so that only new or changed files are captured. If a file with the same name has changed, look only at the new or changed items.
  2. For each item with an image id, write one row per person named as an individual, following the conventions and grading rules above.
    • Link a study person (check the study's list of people) at the file's grade.
    • Otherwise leave the row unlinked, with the file's verdict in notes.
    • Keep the variant spellings the file treats as this name (Hegnaner, Hegenauer, Hegnaur; Hagnauer only when the file or the study's list of leads treats it as a possible variant).
    • Skip other surnames.
    • A page whose persons cannot be told apart gets one unlinked row, with the candidate IDs in notes.

  3. Deduplicate before writing: skip every (url, person_id) pair already in the study's index of records. The natural key includes the citation text, so alone would not catch a second row for the same pair. The check:

    import csv; have = {(r["url"], r["person_id"]) for r in csv.DictReader(open("data/occurrences.csv", encoding="utf-8"))}
rows = [r for r in rows if (r["url"], r["person_id"]) not in have]

    For a changed verdict on a row that already exists (a new grade or a new link), do not add a row. Record the change in edits/*.json, as requires.

  4. Write all new rows to one file, data/raw/newspapers-capture-<date>.json: {"about": …, "defaults": {"source_site": "newspapers.com", "added": "<date>"}, "rows": […]}. Then run:
    • python tools/add_occurrences.py data/raw/newspapers-capture-<date>.json --dry-run
    • the same command without --dry-run
    • python tools/check.py, which must give 0 errors
  5. Add a section to this log. For each file, record its size or md5, the counts (linked and unlinked, by paper and by decade), and the open questions.
  6. After a rebuild from the importers, re-run every data/raw/newspapers-capture-.json. This is idempotent: rows already present are skipped by their natural key. Then re-apply edits/.json in order.

People in this article