Most History Writing Fails Because People Confuse Categories With Tools
I spent about six years working on a digital archive project for regional economic records from the late Ottoman period, and the single biggest source of headaches wasn't the digitization or the metadata schema. It was deciding what category to file things under when those records clearly belonged to multiple categories at once. A tax register could be economic history, administrative history, or demographic history depending on who was looking at it. We lost about three weeks just arguing about where to file a batch of customs receipts from 1892. The category system we ended up using, and what I still recommend, is what most methodology textbooks call the periodization-plus-domain framework. You define two axes: a chronological periodization and a thematic domain. The intersection becomes your filing unit. It's straightforward once you've hit the wall of ambiguity the first time, which everyone does.
A Useful Category Of Historical Analysis
Let me explain how this actually works in practice rather than how the literature describes it, because the literature tends to make it sound more rigid than it needs to be. Start with your domain categories. For most work, the standard set covers political/institutional history, economic history, social/demographic history, cultural/intellectual history, and environmental/material history. You don't need all five. Pick the ones your material actually speaks to. I once worked with someone who tried to force everything into political and economic categories because they didn't know what to do with letters, diaries, or parish records. That's why the five-domain default exists. Then layer periodization on top. Here's where beginners mess up. They either use overly granular periods — breaking something into "early modern period early phase" — or they use periods that don't match their source material. If you're analyzing 17th-century French tax records, the concept of "early modern" means almost nothing unless you're doing comparative work. Use the periodization that your sources are embedded in. That might mean sticking to reign dates, colonial administrative periods, or pre/post war boundaries depending on what you're studying. The real value of this framework isn't the filing system itself. It's that it forces you to acknowledge when a single source spans multiple domains or periods. That tension is where the actual historical argument lives. Most historians who claim to be doing "multicausal analysis" are really just stacking domain labels without showing how they interact. The framework makes you see that gap.
I should mention a specific problem I ran into that the standard texts don't cover. When working with colonial records from the Belgian Congo, the archival categories imposed by the colonial administration didn't map onto any of the five domains cleanly. A single file labeled "native affairs" contained labor contracts, land disputes, marriage records, and criminal cases. The workaround I used was to create a sixth ad-hoc domain called "colonial governance" and then cross-reference those files across the existing domains using a secondary index. It added about two hours of initial setup per archive but saved maybe forty hours of re-file work later. The cross-referencing matrix became the most useful part of the whole project. There are real limitations to this approach that people rarely talk about. The framework assumes your sources are discrete enough to be categorized, which falls apart with oral histories, fragmentary evidence, or heavily redacted documents. It also creates a false sense of closure — once something is filed under "economic" and "1920s," it's easy to treat that categorization as the end of the analysis rather than the beginning. I've seen graduate students spend months producing beautifully organized datasets with no actual historical argument attached because the categorization task felt productive enough to substitute for one. Another counter-intuitive point: more categories usually make your analysis worse, not better. When I reviewed a colleague's work on Weimar-era newspaper archives, they had created seventeen sub-domains under economic history alone. The granularity was so fine that each category contained three to five sources. At that point the system wasn't organizing anything. It was just a labeling exercise. Stick to the smallest number of domains that still captures the variation in your material. Four is often enough.
Get the Full Details
If you're starting a new project and want a quick reference sheet for the domain-periodization framework, I've put a basic version together. It's not a formal publication, just a working document I use before I start tagging anything. Download the reference sheet here. It includes the standard five-domain breakdown, examples of periodization mismatches, and a simple cross-reference table format that works in Excel or any spreadsheet program. One more thing that nobody warns you about: the framework works differently depending on whether you're the primary researcher or whether you're working with someone else's categorized material. When you build the categories yourself, you notice the edge cases in real time and adjust. When you inherit categories, you have to reverse-engineer the logic, and you'll find gaps that weren't intentional — just the result of whoever built the system running out of time. Always check the metadata or ask the original researcher how they handled the ambiguous cases. If they didn't handle them and just picked a domain arbitrarily, your analysis will inherit that arbitrariness.
