What about pictures? We call this representational evidence. Representational evidence can be quite good at telling you what something looked like (but beware of artistic conventions!), but is of course little help for the names-and-dates kind of historical work. The larger problem though is that representational evidence especially becomes difficult to interpret without literary or archaeological evidence backing it up. The problem of correlating an image to a specific person or object can be very hard (by way of example, the endless debates about what is meant by kotthybos in the Amphipolis military regulations). Representational evidence gets a lot more useful if you can say, “Ah, X depicts Z events from B-literary-source” but obviously to do that you need to have B-Literary-Source and B is going to do most of the heavy lifting. To see just how hard it can be to use representational evidence without a robust surviving literary tradition, one merely needs to look at work on pre-historic Gaul (it’s hard!).

Which brings us at last to the big dog, archaeological evidence (although all of the aforementioned also show up in the archaeological record). Archaeology is wonderful, easily the biggest contributor to the improvement in our knowledge of the ancient world over the last century; my own research relies heavily on archaeological evidence. And the best part of it is we are getting more and better archaeological evidence all the time. Some archaeological finds are truly spectacular, like the discovery of the remains of the wrecks from the Battle of the Aegates Islands (241), the decisive engagement that ended Rome’s first war with Carthage (underwater archaeology in general in a young part of archaeology, which is itself a young field so we may well expect more marvels to come).

But (you knew there would be a but), archaeological evidence is really only able to answer certain specific questions and most research topics are simply not archaeologically visible. If your research question is related to what objects were at a specific place at a given time (objects here being broad; “pots” or “houses” or “farms” or even “people” if you are OK with those people being dead), good news, archaeology can help you (probably). But if your research question does not touch on that, you are mostly out of luck. If your object of study doesn’t leave any archaeological evidence … then it doesn’t leave any evidence. Most plagues, wars, famines, rulers, laws simply do not have archaeologically visible impacts, while social values, opinions, beliefs don’t leave archaeological evidence in any case.

Take, for instance, our evidence for the Cult of Mithras in the Roman Empire. This religion leaves us archaeological evidence in the form of identifiable ritual sanctuaries (“mithraeums“). Archaeology can tell us a lot about the normal size and structure of these places, but it can’t tell us much about what people there believed, or what rituals they did, or who they were, with only a handful of exceptions, which is why so much of what we think we might know about Mithraism is still very speculative.

Moreover, archaeology only works for objects that leave archaeological remains! Different materials preserve at different rates. Ceramic and stone? Great! Metals? Less great; these tend to get melted down when they don’t rust. Wood or textiles? Worse, almost never survives. This is why we have so much data on loom weights (stone, ceramic) but less on looms (wood, textile), and so much data on spindle whorls (stone, ceramic) but less on spindle-sticks or distaffs (wood). Compounding this are preservation accidents, in that things that survive tend to be things thrown away or buried with bodies and those practices will impact your archaeological record.

But the best part about archaeology is that it has network effects, which is to say that the more archaeology we do, the more useful each find becomes. New discoveries help to date and understand old discoveries and with lots of archaeological evidence, you can do really neat things like charting trade networks or changing land-use patterns. The problem is that you really do need a lot to generate a representative sample so you know you aren’t wrongly extrapolating from exceptions, and for right now, only the best excavated regions (Italy, to a lesser extent Greece and Egypt) are at the point where we can talk about, for instance, changing patterns of land use and population with any detail. And even then, uncertainties are huge.

Finally, archaeology, like everything else, works best with literary evidence. Take, for example, pre-Roman Gaul. The Gauls, due to their deposition practices are very archaeologically visible. Rich burial assemblages, large ritual deposits and archaeologically visible hill-fort settlements mean that the archaeological record for pre-Roman Gaul is very robust (in some cases more robust that the equivalent Roman context; we can be far more confident about the shape and construction of Gallic weapons than contemporary Roman ones, for instance). But effectively no literary sources for Gaul until contact with the Romans and Greeks. Consequently, almost everything about their values, culture, social organization in the pre-Roman period is speculative, with enormous numbers of questions and few answers.

If you want to ask me, “When did the Gauls shift to using longer swords” I can tell you with remarkable precision, in some cases, region by region (but generally c. 250 BC, with the trend intensifying in the late second century). But if you want to ask, “what was it like to rule a Gallic polity in c. 250 BC?” The best we can do is reason from what we see Caesar describing in c. 50 BC and hope that was typical two hundred years earlier.

Bret Devereaux, “Fireside Friday: March 26, 2021 (On the Nature of Ancient Evidence”, A Collection of Unmitigated Pedantry, 2021-03-26.