Working With Historical Newspaper Archives
Most people treating genealogy as a hobby run into the same wall pretty quickly. They find a reference to a death notice in some digitized database, click through, and immediately hit a paywall or a blurry scan that makes everything illegible. I've been doing this long enough that I don't bother with the obvious paths anymore. The Niagara Gazette collection, for instance, has layers most researchers never get past because they stop at whatever genealogy website auto-fills for them. Here's what actually happens when you try to pull substantive information from these records. The Niagara Gazette (which covers the Niagara Falls and Buffalo corridor) has its obituary section going back over a century, but the digital versions are not organized the way anyone would design them if they had a choice. They're scanned images with OCR that occasionally catches the date but regularly misses half the body text. The names of surviving relatives, the funeral home, the service details — those often get dropped entirely by automated transcription. My workflow doesn't start with Ancestry or Find a Grave. I go straight to the Niagara County Library system or the microfilm holdings at the Buffalo & Erie County Public Library if I'm dealing with older material. There's a specific reason for this. The microform copies frequently contain content that was cut from the digital version — different layouts, additional columns, local advertising that ran adjacent to obituaries and sometimes referenced the same families. You miss those details by relying exclusively on the scanned-and-indexed versions online.
When I'm working with a specific obituary from the Gazette, I pull the original issue first. If I'm doing this remotely, I use Newspapers.com's image viewer rather than any OCR-dependent interface, zoom into the actual article, and transcribe it myself. It takes longer upfront but saves hours later when you realize the index entry had the wrong maiden name or the wrong death date. I recently spent about forty-five minutes on a single obituary from 1923 because the index listed the deceased as "Margaret H." but the actual text said "Margaret O'Sullivan-Hart." The marriage record I'd already traced was under O'Sullivan, not Hart, and that discrepancy would have derailed the whole line if I'd accepted the index at face value.
What to Actually Look For Beyond the Basic Facts
Obituaries in regional papers like the Gazette rarely follow a consistent format before the 1950s. A death notice from 1898 might be two paragraphs of boilerplate. One from 1931 could include a detailed list of survivors, the employer, the church, and the location of the gathering. The format shift itself is useful information — it tells you something about how the family positioned themselves publicly, which reflects their social standing in the community at that time. The real value in these records isn't usually the death date. It's the network around it. The neighbors mentioned as "cousins" who aren't actually cousins. The funeral home that appears repeatedly across unrelated families in the same neighborhood, which can help you place where a family lived. The church or lodge affiliation that connects to organizations whose records survived separately. I once traced a family line forward through a series of obituaries where the only consistent clue was the same Italian bakery on Rainey Street appearing as the meal provider for funeral gatherings across three different branches of the family. That bakery detail didn't appear in any indexed metadata, and no automated search would have connected it.
Get the Full Details
The Problems That Actually Come Up
There are genuine bottlenecks with this particular collection. The Gazette didn't publish every death notice that occurred in the coverage area. Local papers of this size had selective practices, especially before the 1940s. If someone died in a nearby town like Lewiston or Youngstown, the notice might have run in the local paper instead, or not at all. You need to cross-reference with the Buffalo News and the Batavia Times Record depending on which part of the county the person lived in. The geographic overlap is messy and poorly documented in any single index. Naming conventions are another issue. Women's names are inconsistently recorded. Before the 1920s, a woman might appear as "Mrs. John Doe" with no first name attached. In some cases, the Gazette didn't publish obituaries for women at all unless they came from a prominent family or the death was unusual. This means your research on female ancestors in this area will have gaps that no amount of searching will fill. You'll need birth records, cemetery records, or church registers to compensate for the silence in the newspaper. Another practical problem: the Gazette's own website has a search function, but it's notoriously unreliable for anything before 1970. The digitization quality drops off significantly, and the page load times make manual browsing impractical. I use the library's digital newspaper portal instead, which gives better image resolution and actually functional keyword searching across the full text, not just the OCR layer.
Where to Access This Material
The Niagara County Library system offers free remote access to their digitized newspaper archive for cardholders. If you live outside the county, you can still get a card by paying the non-resident fee, which is substantially cheaper than subscribing to multiple commercial genealogy databases. The Buffalo & Erie County Public Library has the most complete microfilm run, including supplement sections and funeral advertisements that aren't in the main text scans. Their reading room is open to the public, and they allow photography of microfilm pages, which is faster than copying. For anyone who can't travel, the library system also provides interlibrary loan of microfilm reels. It's a slow process — usually two to three weeks for a single reel — but it's free and covers gaps that no commercial database has. I've gotten copies of issues that were missing from every online version because the microfilm had physical damage that the digitizers flagged and skipped. If you want a straightforward starting point, pick a known ancestor, pull the Gazette issue from their death date, and read the entire page, not just the obituary column. The surrounding ads and community announcements will contain references that help you place the family in the broader social context of the area. That context is what separates a name-and-date search from actual historical research, and it's the part that automated tools consistently fail to reproduce.
