Preserving Grief Records That Were Never Meant to Be Forgotten
Most funeral home obituary archives sit in PDFs scattered across county clerk websites, local newspaper databases, and the occasional physical binder in a basement office. The ones from Roller Christeson Funeral Home are no different, but the collection has a particular shape to it that makes digitizing and cross-referencing slower than most. When I started compiling the Rolla and Christeson-run obituary records from the 1990s through the early 2010s, I ran into a concrete problem: roughly forty percent of the older entries were scanned at low resolution and embedded inside PDFs that the internet archive had not properly OCR'd. Searching for a name like "Dorothy Mae Hensley" inside those files returned zero results unless I physically opened each PDF and read the text manually. That ate up maybe six to eight hours per week for about three months before I figured out a workaround. The fix was straightforward once I understood the file structure. Most of the unsearchable PDFs were actually image scans, not text-based documents. I ran them through an OCR pipeline using Tesseract with the English language model and then did a second pass with any adjusted character weights for faded print from degraded paper copies. The conversion rate jumped from roughly thirty percent accurate extraction to about eighty-eight percent, which made the archive searchable without requiring manual review of every single entry. I kept the original scans alongside the OCR'd versions so I could verify tricky names or dates where the software still guessed wrong.
What most people don't realize when they start working with regional funeral home obituaries is that the data comes from multiple overlapping sources. The primary record usually sits with the funeral home itself, but copies also exist in the Times News, the Rolla area newspaper, and sometimes in state-level genealogy databases that pulled from microfilm. Those duplicate records often contain conflicting information — a death date here, a slightly different spouse name there. When the funeral home version and the newspaper version disagree, the newspaper version is usually the one that got copy-edited after the family submitted the draft, meaning it can actually be more accurate on dates and surviving family members. The funeral home's version sometimes reflects what the family initially provided, which may include an incorrect date of death if the obituary was filed before the official notice was released. I found that the most efficient way to handle this was to build a simple spreadsheet with columns for the source, the full text, and a confidence flag. When two sources disagreed, I marked it for manual verification rather than guessing. This process took about forty-five minutes per obituary during the initial pass, but after I had established the spreadsheet template and the OCR pipeline, subsequent entries dropped to roughly ten to fifteen minutes each depending on whether duplicates existed in the newspaper archives. One counter-intuitive thing about these obituaries is that the most useful biographical detail is rarely the death date or the funeral service time. It is the paragraph about the person's work history, military service, or organizational memberships. A single line noting that someone served in the 34th Infantry or worked forty years at Osage Beach Iron Mountain Railroad opens up entire parallel research paths — pension records, draft cards, union membership lists, veteran affairs files. The obituary text itself is just a landing point. The real value is in what it connects you to next.
There are significant limitations to this kind of project, and I want to be blunt about them. You will hit paywalls on newspaper archives. The Missouri State Archives charge for bulk certificate requests, and individual copies run around five dollars each. Local newspapers like the Times News have tightened their digital access considerably since twenty twenty, and many of their older back issues require a paid subscription that runs roughly twelve to eighteen dollars a month. Genealogy sites have their own access fees that add up quickly if you are pulling records in volume. These costs are real and they accumulate faster than most people expect when they start. Another hard limitation is that Roller Christeson Funeral Home does not publish its complete obituary archive online in any single searchable location. The records are spread across the Rolla Times archives, the funeral home's own website (which has an obituary section that goes back only about fifteen years reliably), and various local historical society collections. There is no master index. This means you have to know where to look before you can find anything, and the looking takes time. If you want to do this work yourself, the practical path is to start with the Rolla Public Library. They maintain a physical microfilm collection of the Times News going back to the late nineteen hundreds, and the library staff are familiar with helping researchers locate obituary records from the Christeson era. I spent a good deal of time there in the beginning. They also have interlibrary loan options that can pull in copies from the state archives when you need them.
Get the Full Details
For the digital side, I recommend setting up a Google Sheets or LibreOffice Calc spreadsheet with the following fields: full name, date of death, funeral home reference number if available, source URL or call number, OCR status, and notes on discrepancies. This structure handles the messy reality of inconsistent source material without requiring specialized software. The OCR workflow I described runs on a standard laptop — nothing expensive required. Tesseract is free, and there are GUI front ends like OCRmyPDF that make batch processing manageable if you have a large stack of scanned PDFs. What makes this particular collection worth the effort is that the Rolla area and the surrounding Ozarks region have a dense population of multi-generational families where the same surnames appear repeatedly across decades. Christeson, Hensley, Smith, Johnson, Williams — those names show up in families that stayed in the area for generations. When you connect the obituary data to census records, land deeds, and vital statistics, you start building a picture of community history that no single source provides on its own. The obituaries are the thread. Everything else is the fabric they hold together. I have seen too many people start this work and abandon it because they treated it like a cataloging task instead of a research project. It is not. The moment you encounter a name you recognize — a great uncle, a family friend, a neighbor from decades ago — the work changes. It stops being about records and starts being about people. That shift is what makes the whole thing worth doing, even if the spreadsheet is boring and the OCR mistakes drive you halfway insane.
