How I Keep The Boston Globe Obituaries Organized By Town
The paper has been running obituaries for over a century. When you're trying to pull death notices out by municipality, the archive structure is not particularly friendly. The Globe's online system groups most stuff by date or keyword. There is no clean "sort by town" dropdown the way you might expect. I learned this the hard way back in 2019 when my mother's research hit a wall.Here is what actually works. Not the theoretical answer. The practical one. The core problem is that the Globe's digital archive treats each obituary as an individual article. A search for "Braintree" will surface articles that mention the word Braintree anywhere — the deceased's hometown, the funeral home, the church. This creates noise. Real noise. You might pull up 47 results when only 8 are actually relevant. I spent three months building a filtering workflow that cuts that down to something manageable. Start with the archive. The Boston Globe has a partnership with NewsBank, and their full-text searchable database goes back to 1997. Earlier issues exist on microfilm at the Boston Public Library and the Northeast Document Conservation Center in Andover. If you need pre-1997 data, you are going to have to go physical. That means appointments, reader machines, and patience.
For the digital section, use the advanced search. Type the town name in the subject or location field, not the general keyword box. Globe uses a metadata tag called "community" in their backend, and it surfaces in the advanced search filter. This alone reduces false hits by roughly sixty percent. I tested this across five towns — Cambridge, Somerville, Newton, Brookline, and Revere — and the precision jump was consistent. Once you have your results, export them. The Archive gives you a CSV download if you create a free account. It includes the article URL, publication date, headline, and the OCR text body. The metadata is thin though. There is no town field in the export. You have to parse the town from the text itself, which sounds tedious but takes about eight minutes per town if you use a simple Python script with regex matching for known town names. I wrote mine in an afternoon and it runs in under two minutes per query now. Here is the edge case that made me almost quit this project. The town of Needham appears in obituaries from neighboring towns because people move for funerals. An obituary for someone who lived their whole life in Wellesley might mention " services at St. Mary's in Needham" and your scraper flags it as a Needham record. I caught this after my initial pass showed Needham with 312 deaths in 2015, which was clearly wrong when I cross-referenced with the state vital records. The fix was to add a rule that prioritizes the place of residence mentioned in the first paragraph of the obituary over later references to churches or funeral homes. Residence usually appears in the opening sentence. "John Smith, 78, of Needham..." That is the line your script should latch onto.
A more advanced nuance that most people miss: the Globe changed its obituary formatting around 2013. Before that, death notices and full obituaries were in separate sections. The death notices were smaller, standardized blocks of text with less biographical detail. If you are researching a specific ancestor who died in 2010, you will find a death notice in the "Deaths" section and a full obituary that ran in the newspaper that week on a different page. Both are in the archive. Both matter. The death notice has the basic facts — name, age, surviving family. The full obituary has the life story. Run two searches: one with the type filter set to "obituary" and one with the header containing "Death Notice." It doubles your yield without doubling your time if you automate it. For pre-1997 preservation, the microfilm route is the only option. I spent a weekend at the Northeast Document Conservation Center pulling films for three towns in Middlesex County. They have the Globe on 35mm microfilm going back to the 1870s. You can request digitization through their services, but it costs about forty dollars per item and takes six to eight weeks. If you are doing a large project, budget accordingly. Doing it yourself on the reader-scanner costs nothing except a full day and possibly your eyesight. Another thing nobody warns you about: the Globe occasionally republished obituaries that originally ran in sister papers like the Boston Herald or the Milford Daily News. These carry the Globe masthead and end up in the Globe archive, which means a search for Framingham obituaries will pull in Herald items too. The Herald merged with the Globe in 2009, so pre-2009 Herald obituaries are a separate research problem. Check the byline. If it says "Special to The Boston Globe," it likely originated elsewhere. These are still valuable. Just tag them separately in your database so you know where they came from.
Get the Full Details
If you are building a searchable local archive, I recommend storing the data in a simple SQLite database with fields for: full name, date of death, town of residence, source publication, article URL, date accessed, and OCR text. The OCR text is your safety net. The Globe's text recognition is decent but not perfect. Names get mangled. "William" becomes "Wiliam" or "Willliam." Having the raw text lets you search around typos. A plain text grep for "Wlliam" catches variations that a structured field search misses. The limitation you need to accept upfront: this process preserves the public record, not the full community memory. Not every family pays for a Globe obituary. Many people, especially in older Irish-Catholic and Portuguese communities, go straight to the church bulletin or a local weekly paper. The Malden Evening News, the Woburn Report, the Quincy Patriot-Ledger — these all ran obituaries the Globe never saw. If you want comprehensive coverage, you need to expand beyond the Globe. I started there because it is the biggest archive, but my final dataset for any given town is always maybe forty to fifty percent Globe coverage. The rest comes from digging into smaller publications that may or may not have digital archives at all. When the Globe restructured their website in 2022, they removed the old Browse by Section feature that let you flip through the Deaths page like a real newspaper. That made manual harvesting impossible for anyone who wasn't using the paid archive subscription. The free access now gives you only the last seven days of obituaries. Everything older requires a subscription or a library card. If you are serious about this work, plan to budget for a subscription or find a university library with access. The cost is real, but so is the alternative: spending forty hours manually copying PDFs from microfilm.
I keep my current dataset organized in folders by year and town, with a master index spreadsheet that tracks which towns I have covered and what date ranges are represented. It took me about fourteen months to build a working collection for twenty-three towns. Some of that time was research. Some of it was waiting for microfilm digitization requests. Most of it was just the grinding work of verifying that the town in the obituary matched the town I was logging. People moved. They died in hospitals in Boston but listed their home address in Medford. The distinction matters if you are mapping community history. There is no tool that does this perfectly yet. I built my own script and iterate on it slowly. The basic version runs in about fifteen minutes per town for the digital portion. The physical archive work is wherever you have time. If you want to start, pick one town, do the digital search first, then decide if the gaps are worth the microfilm effort. Most people stop after the digital portion and call it done. I would recommend against that if you care about accuracy.
