Every few months history Twitter runs a thread about what software to use for note-taking. (See, e.g., threads recently started by Austin McCoy and J.Meléndez-Badillo.)

I wrote a whole chapter of The Princeton Guide to Historical Research on this topic, the gist of which is that historians have mostly abandoned our simple, easy-to-teach system based on 5 x 8 inch notecards, but we have yet to settle collectively on a computerized system to replace it.

Here are some more thoughts along those lines, and an explanation of the centrality of FileMaker Pro to my workflow for major projects.

Zotero is fantastic for collecting metadata and PDFs from a range of online databases, which is great for books, journal articles, and some newspaper collections. It also offers a function for taking notes on those sources, but that’s pretty rudimentary. For example, there’s no good way to sort notes within Zotero. And it’s not particularly well suited for archival research, since entering the metadata manually is such a pain.

I write in Scrivener, which can also store PDFs and give you a space to take notes on them. For shorter projects, these two packages may suffice. Alexandra Samuel explains a simple workflow at “How to Use Zotero and Scrivener for Research-Driven Writing,” JSTOR Daily, December 17, 2019.

But what if you have hundreds or thousands of notes to store for a book-length project? (The Fires of Philadelphia database ended up with 12,750 notes.)  Here, it’s nice to have a database in between Zotero and Scrivener, or whatever reference manager and writing software you use. While the reference manager can store your metadata and PDFs, and the writing software stores your arguments and stories, the database will store the information itself, and help you retrieve what you need.

In particular, database software should be good at finding and sorting. For example, I might want to see all my notes that mention “Levin” before 1843, sorted in chronological order. Or, all notes that mention the Constitution, drawn from sources published since 1995.

I’ve experimented with three programs for this intermediate step: FileMaker Pro, DEVONthink, and Obsidian. I used FileMaker Pro to write my first, second, and fourth books, and DEVONthink for The Princeton Guide to Historical Research. For my current project, I tried Obsidian foe a few months, but I am now back to relying on FileMaker Pro.

Any of these database programs can find and sort notes far better than Zotero or Scrivener. The real power of FileMaker is its ease of customizability for users with minimal knowledge of coding. To give a hint of its power, I post screenshots from the databases I created for Fires of Philadelphia and for my current research on the Dulles Corridor Metrorail Project.

The two databases have roughly similar layouts, with notes on the far left, a bunch of metadata in the middle, and, elsewhere, various buttons to activate scripts. (Top right for the Philadelphia book, top center for the Dulles Corridor.)  Thanks to FileMaker’s easy tools for creating scripts and layouts, I was able to customize each for the particular project.

FileMaker Pro database designed for research on the Philadelphia riots of 1844, with buttons that generate new citations to several of the newspapers I most commonly used.

The 1844 database research screen has a section of buttons devoted to the several newspapers I used most frequently. As I worked through a run of the American Advocate, the Catholic Herald, or some other paper, one click on the correct button would create a new record with the same periodical and the same date. A different button would start me on the next issue, whether it was one, seven, or fourteen days later.

In addition, a set of fields toward the bottom of the screen let me enter the names of anyone killed, wounded, or arrested. Then additional FileMaker layouts and scripts can convert that into a table showing me all the names, sorted by last name. This table helped me understand that John Loesher, John Lescher, and John Lusher were all the same person.

FileMaker Pro table showing the names of people arrested, wounded, or killed in the riots of 1844.

The Dulles Corridor database, by contrast, has fewer newspaper buttons. But because most of the newspapers I am using for this project can be found in Zotero-compatible online databases, I usually just hit the “Zotero article” button and paste the full citation into that. Note that FileMaker can parse the date of articles or other documents, allowing me to sort notes chronologically. For this project, I don’t need to track people killed and wounded, but I do want to track the projected opening dates at various times. Again, with FileMaker Pro it’s easy to add new fields and tables that are specific to a project.

FileMaker Pro database designed for research on the Dulles Corridor Metrorail Project, with specialized fields for projections of opening dates and costs.

Obviously the layout is inelegant. At some point, tidying it up will help me avoid writing.

My sense from Elena Razlogova’s work on Obsidian for Historical Notetaking is that Obsidian is equally customizable, but only if one has a pretty solid grounding in TypeScript. And even then, I think you need to remember a lot of commands, rather than creating buttons.

I am already familiar with FileMaker Pro, so I face high transaction costs in trying to learn a new system, and neither DEVONthink nor Obsidian appear to justify that effort. For researchers just starting off, I might suggest Obsidian, which has several advantages. It’s open source, it’s free, and the Markdown files it generates will be readable by future generations. But I fear that getting the most out of Obsidian requires learning enough TypeScript to write your own plugins so that it serves your needs.

All of this is subject to change, and five or ten years from now, I can imagine that Razlogova and others will have integrated Zotero and Obsidian into an open-source, cross-platform system that will meet the needs of most historians out of the box, and will provide power users the change to customize their tools without learning a programming language.

But we aren’t there yet, so for now each historian must decide how much time and money they are able to invest, and what features matter the most to them.