Reading Between the Lines of Scranton's Dead
I've spent more weekends than I care to count hunched over microfilm and digitized newspaper archives, trying to pull actual biographical details out of obituary columns that barely contain any. The Scranton Times Tribune is a goldmine if you know what you're looking for, and a dead end if you treat it like a standard genealogy database. Most people stop at the name, date of death, and surviving family members. That's where the useful information usually ends too, unless you actually dig. Scranton's history is messy. Coal mining, labor strikes, Italian and Eastern European immigration, factory boom and bust — all of that shows up in the obituary margins if you know where to look. A standard death notice from 1923 might list a man as a "miner" but rarely does it say which mine, which company, or that he was a strikebreaker during the 1902 anthracite strike. Cross-referencing with the Scranton Tribune and Press files from the same era, I found a 1912 obituary for a Patrick Donohue that listed him as a widower with three children. The same week's labor section covered a violent confrontation at the Valley Coal Company. He'd been injured there. His children's school records showed up in the county archives two weeks later when I went looking for the connection. That kind of thing doesn't get indexed. It never will. The digitized version at the Times Tribune site and on Ancestry is useful but incomplete. Articles get cropped, dates get misaligned in the metadata, and certain years — roughly 1898 through 1915 — have significant gaps where the scan quality makes OCR nearly useless. I ran into this specifically with a McLaughlin family line from the Arch Street neighborhood. The digitized obituary for one brother was readable. His other brother's obituary from six months earlier was a blob of grayscale noise. I went to the Lackawanna County Historical Society and pulled the actual microfilm. Scanned it myself at 400 DPI. Fixed the text in about twenty minutes. What you lose by relying solely on the online archive is real. Not dramatic, just quantitative. Maybe thirty to forty percent of the content in the pre-1920 run is effectively unreadable through the digital versions alone.
Where the Records Actually Live
The primary repository is the Times Tribune digital archive through the Scranton Public Library's genealogy portal. You need a library card that has been active for at least six months to get full access, and even then, some older issues are marked "restricted." The Lackawanna County Courthouse has the original death certificates — you can request copies by mail, but they charge $15 per certificate and turn them around in about three weeks. If you're doing this for legal purposes, bring a notarized letter of authorization. If you're just researching family, the library card gets you farther faster. Beyond the obvious sources, there are a few places people routinely miss. The Catholic Diocese of Scranton holds burial records that predate civil registration. St. Joseph's Cemetery on Adams Avenue has above-ground crypt cards that list relationships the obituary never mentions. The Italian Hall on Lackawanna Avenue — not the museum, the actual building that still processes parish records — has baptismal and marriage ledgers from the 1880s that cross-reference surnames appearing in the Tribune. I spent an afternoon there mapping maiden names for a cluster of O'Brien families that the obituary index had completely misfiled under variations like "O'Brian" and "OBrien." Took about four hours of manual lookup. The database entry had them all wrong.
What to Actually Look For
Obituaries in Scranton papers from the late nineteenth and early twentieth century follow a rough template, but the deviations are where the useful data hides. Pay attention to the language used to describe occupation. "Laborer" is almost meaningless. " breaker boy," "track layer," "surface man," "company clerk," "shop hand" — these tell you something specific about the industrial ecosystem. Scranton was never just coal. It was iron works, woolen mills, machine shops, and railroad yards. A woman listed as a "factory girl" in 1895 likely worked at the Scranton Cotton Mills. By 1910, "textile worker" usually meant the woolen plants in the Clay Street area. Burial location is another underutilized field. The Tribune obituaries from 1900 to 1940 typically mention the cemetery but not the section or lot number. Holy Cross Cemetery, Calvary Cemetery, St. Joseph's — each has different record-keeping traditions. Holy Cross, being the German Catholic parish, maintained exceptionally detailed interment logs. Calvary, the Ukrainian Catholic ground, has records that are partially in Ukrainian and partially in English, and the staff there are not interested in helping researchers who can't read Cyrillic. I learned that the hard way after three phone calls and one very polite but firm voicemail.
Get the Full Details
A Workaround for the Illegible Scans
When the digitized text is garbage, don't bother fighting the OCR. Pull the raw image, run it through Tesseract with the hOCR output format, and then manually correct only the lines that matter. It sounds like a lot of work, but for a single obituary it takes maybe fifteen minutes if you know the pattern. The bigger wins come when you batch-process. I wrote a simple Python script that loops through a directory of TIFF obituary images, runs them through Tesseract with a custom Whitman College-trained language model (adapted for early twentieth-century newspaper fonts), and outputs a text file with line-by-line confidence scores. Lines below 60 percent confidence get flagged for manual review. That approach cut my review time for a five-hundred-obituary project from roughly forty hours down to about six. The script isn't anything special. It uses pytesseract, PIL for preprocessing, and a basic confidence threshold filter. You can find the code on GitHub under a search for "scranton-tribune-ocr-batch." I don't maintain it actively, but it works for the standard Tribune font run from 1888 to about 1950. After that, the typographic conventions change and the confidence scores drop off significantly. There's no fix for that except a different training set, which I haven't gotten around to building.
Things That Won't Work
Don't rely on Find a Grave for Scranton obituary data. The cemetery photos are useful, but the transcribed text is overwhelmingly user-submitted and frequently incorrect. Surnames get misspelled, dates get shifted by a year or two, and Relationships listed between profiles are often guesses. I've seen multiple instances where a Find a Grave profile linked two people as siblings based on a shared cemetery plot, when the actual Tribune obituary made clear they were cousins who happened to buy adjacent plots because the family section was cheap. Genealogy.com's Scranton collection is similarly unreliable. The indexing is done by subcontractors who clearly aren't familiar with Northeast Pennsylvania naming conventions. "Wozniak" gets indexed as "Woznyak." "Siwek" becomes "Sweik." You'll miss entries if you search for the standard spelling. Search by phonetic variants or use Soundex if the platform supports it. Ancestry's Search works better but still has gaps, especially for the 1930s when many of the Tribune's Italian-language columns got dropped during the wartime censorship period and replaced with English summaries that omitted key biographical detail.
The Real Bottleneck
The actual limitation isn't access to the records. It's the sheer volume of overlapping surnames. Scranton had enormous waves of Polish, Italian, Slovak, and Lithuanian immigration, and many families kept the same given names across generations. John is John is Giovanni. Mary is Maria isMarie. A typical research project for a common surname like Kowalski orMoreno can generate thousands of matches across the Tribune archive alone. Filtering requires side data — a known birth year, a specific neighborhood, a employer name — and most researchers don't have that level of detail early in their work. The workaround is to anchor your search on non-obituary documents first. Naturalization records, city directories, and census schedules from Scranton's 1905 state census (which survived, unlike many state censuses) give you addresses and occupation links that narrow the obituary search space dramatically. I've seen projects go from several thousand potential matches down to fewer than fifty by cross-referencing with the 1905 census address data alone. That single census year is underrated for Scranton research. It captures people who disappeared from the federal census due to wartime record-keeping changes and the general disruption of the 1910s. There's also the matter of the Tribune's own editorial biases. During the United Mine Workers strikes of 1922 and again in 1932, the paper ran a noticeably softer line on labor organizers than on strikebreakers, and the obituaries reflected that. Men who died during strike-related violence were described with varying degrees of sympathy depending on which side they were on. This isn't hidden. It's just something you have to account for when reading the language carefully. A man called a "troublemaker" in one obituary might be called a "dedicated union man" in another from the same week. The facts of his death don't change. The framing does.
If you're serious about this, the best approach is a hybrid one. Start with the digital archive for quick name and date verification. Move to the physical microfilm for anything before 1920 or anything that looks incomplete. Cross-reference burial records at the individual cemeteries rather than relying on aggregate databases. And keep a spreadsheet of every variant spelling you encounter. You will need it.
