Late Roman Unit Sizes

[Ross Cowan]

But might a figure of 340-400 men be about right, and if so how did Coello work it out?

It's probably been mentioned somewhere above in the epic thread, but isn't Coello drawing on Zuckerman's interesting note about the strength of Legio V Macedonica in AD 399?

http://tyche-journal.at/tyche/index.php/...ew/177/286

[Nathan Ross]

isn't Coello drawing on Zuckerman's interesting note[url=http://tyche-journal.at/tyche/index.php/tyche/article/view/177/286][/url]

Aha! Yes, now I check I see that Coello does reference Zuckerman's paper, but only as regards the point he makes in the previous sentence (about Duncan-Jones and his estimates), which leaves the connection to the V Mac document unclear...

Thanks for the link - fascinating stuff, and it neatly answers my question about where Coello gets his figure of 340-400 from. I do wonder (assuming the note at the end about the possible daily ration is incorrect) that the numbers might be less that Zuckerman estimates; we know from other documents that V Mac had Augustales (a whole schola of them, whatever that means!), and therefore perhaps also Flaviales, both grades on 3-6 annonae each, who between them could eat up a good chunk of the rations. The actual number of troops in the numerus might well be less than 300.

[Flavivs Aetivs]

The number of troops *at this stationing* was about 300 yes, probably. But much like on the Danube, V Macedonica had small one or two century large detachments all over Egypt at Antinoupolis, Panopolis, etc. etc. So we have an estimate of the strength of its main stationing, but not the detachments along the rest of the Nile.

[Nathan Ross]

But much like on the Danube, V Macedonica had small one or two century large detachments all over Egypt at Antinoupolis, Panopolis, etc. etc.

Possibly. Although the ND scrupulously lists all the various stations of the other legions in Egypt and the Thebaid - all have multiple garrisons, just as do the legions on the Danube. But there is only one post listed for V Macedonia - Memphis. Are there papyri mentioning other stations of this legion, do you know?

This could just have been because the unit was so small - Zuckerman's suggestion that it was the descendant of the (single-cohort?) vexillation of the Danubian legion brough to Egypt by Galerius in c293 seems quite plausible. In that case, of course, it wouldn't tell us much about the size of the other legions in the province!

[Flavivs Aetivs]

Are there papyri mentioning other stations of this legion, do you know?

Yeah, I think Ross Cowan's article on V Macedonica names the specific ones but we know they had detachments at various times in various places including Antinoupolis, Helioupolis, etc.

[Nathan Ross]

I was just having a look at some of the recorded sizes of late Roman fortifications, and trying to work out the possible numbers of men they could have housed (this is following Richard Duncan Jones in Structure and Scale in the Roman Economy, and Brewer's Birthday of the Eagle, mostly). This seems a reasonable way to try and calculate the possible size ranges of various limitanei units, and perhaps other units too.

Troesmis, main base of II Herculia, is 2.8 hectares. The legion also has subunits ('cohorts') at three other locations, one of which, Axiopolis, is 0.6 hectares.

Noviodunum, up the Danube, is 5.8 hectares, while El Lejjun is 4.6 hectares and the 'Camp of Diocletian' at Palmyra is 4 hectares. All of these, like Troesmis, appear to have been the main bases of Diocletianic frontier legions. Richborough, where II Augusta are based in the ND, is 2.4 hectares, while Divitia (Deutz) on the Rhine is 2.25. 

Then there are a lot of very small fortifications, listed in the ND as housing a cohort - Kellmunz at 0.86 hectares, Eining at 0.18 and Burgle at 0.16.

So how many men could actually be accommodated at these places?

The Principiate legion fortresses at Caerleon and Inchtuthil were presumably built to house c.6000 men - one is 20 and the other 20.5 hectares. So maybe we could assume a rough estimate of c.300 men per hectare, assuming the density of housing, stores and other buildings was about the same.

That would give c.840 men at Troesmis, c.1740 at Noviodunum, c.1380 at El Lejjun, c.1200 at Palmyra, c.720 at Richborough, and c.675 at Deutz.

The smaller 'cohort' forts have very small garrisons - c.258 men at Kellmunz, only c.54 at Eining and c.48 at Burgle.

Taking the Troesmis garrison, though: we know that II Herculia at one time had 10 cohorts (inscription from North Africa, c.AD298). The ND has three 'cohort' subunits away from base, presumably leaving 7 at Troesmis itself. Divide the 840 men there by 7 and we get 120.

The subunit base at Axiopolis would house 180 men - but the ND also has the Milites Superventores based there as well. If the latter unit was only 60 men strong, the legion 'cohort' could indeed be 120 men. At Troesmis we also have the Milites Secundi Constantini, however, who do mess up the calculations a bit!

But could the Legion II Herculia originally have been raised 1200-strong, with ten 120-man 'cohorts'? That would make a new 'cohort' roughly the size of an old maniple. The limitanei legions further up the Danube, split into five 'cohort'-sized subunits, would be 600 men strong when united, and the field army detachments of these legions would therefore be 600 men strong as well (five 120-man 'cohorts')

Both El Lejjun and Palmyra, meanwhile, look like they could house 1200 men quite comfortably - probably the undivided Legions IV Martia and I Illyricorum respectively.

The small limitanei forts could easily house multiples and divisions of a 'cohort' - 240 at Kellmunz, perhaps, with 60 each at Eining and Burgle.

There is the problem of Legio I Iovia at Noviodunum, which has two 'cohorts' out-stationed, and despite sharing its base with the Milites Primi Constantini would appear to have about twice as much space as it would need, if it really was the same size as the legion at Troesmis! And why are Richborough and Deutz so small by comparison?...

So perhaps there's nothing in all these calculations after all!

[Flavivs Aetivs]

Interesting hypothesis, I actually like the general idea.

But the fort at Al-lejjun (Caparcotna) dates to like 132-244ish, it's not late Roman. And was inhabited by VI Ferrata.

[Nathan Ross]

the fort at Al-lejjun (Caparcotna) dates to like 132-244ish, it's not late Roman. And was inhabited by VI Ferrata.

You're thinking of a different fort - al-Lajjûn in Israel.

El-Lejjun (Roman Betthorus) is in modern Jordan, and Diocletianic. It was probably the base for IV Martia.

Udrah, also in Jordan and with similar appearance and size (4.7 hectares) was probably the later base for VI Ferrata.

This is Lejjun. Barrack size and number of rooms definitely suggest smaller subunit sizes (c.64 per century?). I don't see more than about 1200 men fitting in there either:

[Flavivs Aetivs]

Yes they've found roof tiles in Udruh with VI Ferrata's roof tile mark.

[Nathan Ross]

I've just noticed that the Valentinianic fort at Boppard on the Rhine is almost exactly the same size (4.7 hectares) as Udrah and Lejjun!

With the camp at Palmyra also a close match, might this have been the standard size for a late legion fortress? It would appear about right to accommodate c.1200 men.

[CaesarAugustus]

Really interesting, I would add a note. Principiate legion fortresses where much larger then following period fortresses, with a quite different organization and space arrangement, especially comparing to the fortresses of late period, more similar to middleage castle. So, they old fortresses are not a good base to try to estimate the number of men contained in late empire fortresses.

[Nathan Ross]

comparing to the fortresses of late period, more similar to middleage castle.

That was what I'd thought - but if you look at that ground plan of Lejjun, the internal space doesn't look to be used all that economically. In fact, Principiate fortresses like Novaesium or Inchtuthil look far more densely occupied. Troesmis is very dense, but only a few of the internal buildings are barracks.

Granted, the plan above probably shows the situation in the 6th century, after an earthquake c.500 I think and a partial rebuilding of some areas of the fort. But the barrack sizes and spacings are, I think, original. Even if there were originally large barracks in Areas O and M we'd still only have 20 blocks, enough for 1200-1600 men. That gives a very similar 'men per hectare' ratio to the 1st-2nd century fortress at Caerleon (approx 300 men per hectare).

There's also a possibility that parts of the legion were out-stationed. Khirbet el-Fityan fort, at 0.6 hectares, is the same size at Axiopolis near Troesmis and might have played a similar role in housing a reduced-size cohort.

But I don't think we can assume any more (as several historians once did) that a limitanei legion was substantially larger than a legion of the comitatensis - 3000 men or so. Instead, they might well have been exactly the same size - 1200 or so at full strength.

How would this alter the various calculations for troop numbers in the ND? If frontier legions were 1200 men, and frontier cohorts and milites units were between about 60 and 240 men, we'd have to adjust our estimates downwards.

[Flavivs Aetivs]

But I don't think we can assume any more (as several historians once did) that a limitanei legion was substantially larger than a legion of the comitatensis - 3000 men or so. Instead, they might well have been exactly the same size - 1200 or so at full strength.

This was Jones' assumption and I have no idea what the historical basis for it was. I've always assumed 12 centuries totalling 960 men.

[Nathan Ross]

I've always assumed 12 centuries totalling 960 men.

That seems a fairly radical notion! I don't necessarily disagree, but where does the idea comes from? I don't think we've discussed anything quite like that on this thread so far.

[Flavivs Aetivs]

Two cohorts, each 6 centuries. Seemed standardized enough and fit within the source numbers (c. 750-1350 men) for the purposes of my estimates.