To be a good historian you have to be a good statistician...

... or at least someone who's good at thinking about data, and in this case, survivorship bias.

From Bret Devereaux's highly recommended history blog A Collection of Unmitigated Pedantry.

I’ve mentioned quite a few times here that Roman Egypt is a perplexing part of understanding the Roman Empire because on the one hand it provides a lot of really valuable evidence for daily life concerns (mortality, nuptiality, military pay, customs and tax systems, etc.) but on the other hand it is always very difficult to know to what degree that information can be generalized because Roman Egypt is such an atypical Roman province. So this week we’re going to look in quite general terms at what makes Egypt such an unusual place in the Roman world. As we’ll see, some of the ways in which Egypt is unusual are Roman creations, but many of them stretch back before the Roman period in Egypt or indeed before the Roman period anywhere.
...
Instead what makes Roman Egypt’s uniqueness so important is one of the unique things about it: Roman Egypt preserves a much larger slice of our evidence than any other place in the ancient world. This comes down to climate (as do most things); Egypt is a climatically extreme place. On the one hand, most of the country is desert and here I mean hard desert, with absolutely minuscule amounts of precipitation. On the other hand, the Nile River creates a fertile, at points almost lush, band cutting through the country running to the coast. The change between these two environments is extremely stark; it is, I have been told (I haven’t yet been to Egypt), entirely possible in many places to stand with one foot in the ‘green’ and another foot in the hard desert.

That in turn matters because while Egypt was hardly the only arid region Rome controlled, it was the only place you were likely to find very many large settlements and lots of people living in such close proximity to such extremely arid environments (other large North African settlements tend to be coastal). And that in turn matters for preservation. When objects are deposited – lost, thrown away, carefully placed in a sanctuary, whatever – they begin to degrade. Organic objects (textile, leather, paper, wood) rot as microorganisms use them as food, while metal objects oxidize (that is, rust).¹ Aridity arrests (at least somewhat) both processes. Consequently things survive from the Roman period (or indeed, from even more ancient periods) in Egypt that simply wouldn’t survive almost anywhere else.

...

Now within the typical European and Mediterranean humidity, papyrus doesn’t last forever (unlike the parchment paper produced in the Middle Ages which was far more expensive but also lasts much longer); papyrus paper will degrade over anything from a few decades to a couple hundred years – the more humidity, the faster decay. Of course wood tablets and wax tablets fare no better. What that means is that in most parts of the Roman Empire, very little casual writing survives; what does survive were the sorts of important official documents which might be inscribed on stone (along with the literary works that were worth painstakingly copying over and over again by hand through the Middle Ages). But letters, receipts, tax returns, census records, shopping lists, school assignments – these sorts of documents were all written on less durable materials which don’t survive except in a few exceptional sites like Vindolanda.

Or Egypt. Not individual places in Egypt; pretty much the whole province.

In the extremely dry conditions of the Egyptian desert, papyrus can survive (albeit typically in damaged scraps rather than complete scrolls) from antiquity to the present. Now the coverage of these surviving papyri is not even. The Roman period is far better represented in the surviving papyri than the Ptolemaic period (much less the proceeding ‘late’ period or the New Kingdom before that). It’s also not evenly distributed geographically; the Arsinoite nome (what is today el-Fayyum, an oasis basin to the West of the Nile) and the Oxyrhynchus nome (roughly in the middle of Egypt, on the Nile) are both substantially overrepresented, while the Nile Delta itself has fewer (but by no means zero) finds. Consequently, we need to be worried not only about the degree to which Egypt might be representative of the larger Roman world, but also the degree to which these two nomes (a nome is an administrative district within Egypt, we’ll talk about them more in a bit) are representative of Egypt. That’s complicated in turn by the fact that the Arsinoite nome is not a normal nome; extensive cultivation there only really begins under Ptolemaic rule, which raises questions about how typical it was. It also means we lack a really good trove of papyri from a nome in Lower Egypt proper (the northern part of the country, covering the delta of the Nile) which, because of its different terrain, we might imagine was in some ways different.