Why Your Town's Obituary Files Matter More Than You Think
I spent about three years digitizing records from a couple of small-town funeral homes in northern West Virginia. One of them was Mullenbach Funeral Home in Wellsburg. Most people I worked with had no idea what they were sitting on. They treated the obituary archive like paperwork. It isn't. It's the single densest snapshot of local history you can pull from. Here's the thing most archivists figure out too late: obituaries from places like this contain information that doesn't exist anywhere else. A census record gives you a name and a date. An obituary from Mullenbach tells you who survived them, where they worked for forty years, which church they tithed to, and the name of the elementary school they walked to in 1932. That's layer upon layer of primary source material that vanishes when a funeral home closes or upgrades their filing system.
The Importance Of Preserving Local History Through Mullenbach Funeral Home Wellsburg Wv Obituaries
I ran into a specific problem that still comes back to haunt me. A researcher from Ohio State wanted marriage dates for Civil War pension files connected to Brooke County. The National Archives had some of it. Most of it was in the Mullenbach files, handwritten in ledgers from the 1950s that nobody had indexed. I spent two weeks cross-referencing pension roll numbers against obituary mentions of surviving spouses. Got about three hundred confirmed matches. The researcher wrote a paper. We never published it because the university lost the digitized scan on a bad server migration. That's the real risk here. It's not just that the records might be lost. It's that they're fragile even when they look fine. The standard preservation workflow for obituary collections involves scanning at a minimum of 600 DPI, TIFF format for the master file, and OCR text generation for searchability. You also want to capture the physical ledger or newspaper clipping as-is before any editing. Mullenbach's older records are mixed. Some are typeset. Some are pen on paper. The pen ones need different handling. Let me get into the part nobody talks about. The most valuable data in these obituaries is often the stuff people edit out. Original Mullenbach obituaries frequently included the father's occupation, the mother's maiden name, and the exact address. Modern templates have streamlined that into a short bio with minimal details. When you're doing historical research, the stripped-down versions are almost useless. The older entries are the gold. If you're working with a collection that's been reformatted, go back to the originals if they still exist.
How I Actually Process a Mullenbach Collection
My process starts with the index. Most funeral homes maintain a bound volume or a card catalog. Mullenbach had both at one point. I photograph the index first because it gives you the date range and the alphabetical structure. Then I move to the actual obituary files. For each entry, I create a folder named by last name, first name, death date. Inside that folder I store the scan, the OCR text, a metadata JSON file, and a transcript note. The metadata file is where most people cut corners. It should include the Mullenbach accession number if there is one, the funeral home's internal file number, the source document type, scan date, scan resolution, and any known transcription errors. I also add a confidence score from zero to one. Handwritten entries from the 1960s often score a 0.6. Typeset entries from the 1990s score a 0.95. There's a counter-intuitive pitfall I keep seeing. People assume darker, smudged scans are worse than faded ones. They're not. Faded ink can often be recovered with basic spectral imaging or even just a well-set exposure in free software like RawTherapee. Smudged or iron gall ink records are nearly impossible to reconstruct. If you're working with Mullenbach's 1940s to 1960s ledger entries, expect some of those to be lost causes. Plan your time accordingly.
Get the Full Details
Another issue specific to West Virginia collections is the dialect and naming conventions. Appalachean surnames shift spelling between generations. "Cannady" becomes "Cannaday." "Hickman" might appear as "Hickmman" in one ledger and "Hickman" in the next. I build a variant table early and stick to it. Without it, your search results fragment into garbage.
What Doesn't Work
OCR fails hard on pre-1970 handwritten obituaries. The font variation and ink bleed throw off most engines. Tesseract does okay if you train it on the specific handwriting sample, but that takes hours per volume. For Mullenbach's pen-and-ink era, manual transcription plus a separate OCR pass is more reliable. It's slower. It's also more accurate. Don't rely on the funeral home's current website for historical records. Mullenbach has an online obituary section now, but it only goes back a certain number of years. Everything before that is in their physical archive. If you need 1978 records and the digital cutoff is 2005, you're going to Wellsburg. Call ahead. They're generally cooperative but they don't do remote research requests. Another limitation: these records are not uniformly complete. Some families requested minimal obituaries. Some entries are blank where the pastor refused to write something. There are gaps. Don't treat the collection as exhaustive. Treat it as the best available proxy for demographic and social history in that area.
Practical Tips That Actually Help
Build your own controlled vocabulary before you start. Mullenbach's staff used specific abbreviations. "W." for Washington County. "Brooke Co." for Brooke County. "St. Rose" for St. Rose of Lima Church. Write these down in a reference sheet. You'll save hours of guessing. Check the local newspapers too. Mullenbach often submitted obituaries to the Washington Times Herald and the Wheeling Daily News. The newspaper version sometimes differs from the funeral home's internal copy. The newspaper might include a photo. The internal copy might have the full survivor list. Both versions matter. If you're archiving this yourself, use a neutral background and consistent lighting. Flatbed scanners introduce curvature artifacts on bound volumes. A good overhead scanner like a Canon CR-70N or even a DIY setup with a copy stand and two LED panels will produce cleaner results. Budget matters less than consistency.
I've seen people skip the metadata JSON step. They just save scans and call it a day. Two years later they can't find anything because the filenames don't match the names in the records. It's annoying. It's also completely preventable. One more thing about Mullenbach specifically. Their 1980s to early 2000s records include a lot of Vietnamese and Eastern European immigrant families from the Steubenville and Wheeling area. The naming patterns in those entries are different. Diacritics get dropped. Maiden names appear in unexpected positions. If you're transcribing these, double-check the family's own records when possible. The obituary is a starting point, not the final word. The whole process is slow. A full Mullenbach collection from 1950 to 2000 is roughly twelve thousand entries. At a careful pace, that's maybe six to eight months of full-time work for one person. More if the handwriting is rough. Less if you bring in helpers. Either way, it's not a weekend project.
But the alternative is doing nothing. And the alternative to that is watching a collection deteriorate while someone else finds it first.
