Late Roman Unit Sizes

[Nathan Ross]

Arithmos is 640 men... it comes out to something like 668

That might be about right. But I think we should be wary of making these definite statements; we're all working from the same small body of evidence, and really all we can do is construct hypotheses within certain parameters.

Hypothetically, then, I would estimate a full-strength numerus of palatine auxilia at 678 men all included: 6x110-man subunits, each led by a centenarius and a draconarius (following Vegetius), with three ducenarii each leading a double unit - one of the ducenarii perhaps given the grade of senator - with a leading standard-bearer, a tribunus in command and a primicerius/vicarius as his deputy. That, I think, accounts for all the ranks we find in our evidence for the 4th-5th century.

It would put Zosimus/Sozomen's six arithmoi at 4068 men rather than a round 4000, which I would say is feasible.

A numerus legionem might have a similar set-up, but with ten subunits, ordinarii rather than centenarii, no ducenarii or senatores, and additional subunit NCOs (optiones, senior augustales, imaginiferi), and minor and major tribuni, making up 1172 men in total.

However, this would make the expedition against Gildo (5 palatine auxilia, 2 palatine legions, following Claudian) 5734 men, which is too large to be rounded to 5000 (following Orosius). So either a palatine legion was smaller than a normal comitatensis/limitanei one, or the campaign strength of these units was reduced, or (most likely!) the estimates are wrong. All these units were drawn from the western army, though, and quite possible were at reduced strength in the years after Frigidus, which might offer a tenuous explanation.