Bret Devereaux

This. The Roman Peace was neither Roman, nor peaceful.https://twitter.com/jpnudell/status/1266148060725420038 …

Yes, but it is possible for the Pax Romana to not be terribly peaceful (by modern standards) and at the same time represent a reduction in overall levels of violence. Just going by demographic evidence, it very likely was the latter - a relative, if not absolute, peace.

Is this anecdotal or supported by statistical analysis of data? Was the Julian-Trajanic period numerically less violent than the Republican or Marian eras?

Nature of the evidence makes statistical comparisons of deaths impossible - too many battles without casualty reports, or where casualty reports are unreliable. But use militarization (% pop under arms per year) as a proxy, and yes, supported by data.

Could you use length of battle as a proxy for casualty reports?

That information is actually less likely to be reported in our sources than the casualties. But also - lots of confounding variables. Length of engagement not clearly related to casualty figures.

Interesting. I figured that would be more likely reported because you’d get contemporaneous accounts. I was actually questioning if casualties is the right figure. It feels like it introduces complications, like tactics. I was hoping that was mitigated somewhat with length.

Contemporaneous accounts for ancient battles are quite rare. Even long-after-the-fact eyewitness testimony is rare and valuable. Far more common that we are dealing with a condensed report by a non-eyewitness of official records/eyewitness testimony, often at centuries distance