Late Roman Legion size based on the Perge Inscription

[FlaviusB]

the Notitia Dignitatum lists 99 Equites, 14 Cuneus and 10 Ala regiments in the Western Roman Army, and 117 Equites, 30 Cuneus and 73 Ala in the East; assuming an Ala of 512 and a Equites of 360

Using the ND to calculate army sizes is very tricky, I would say - it's a mishmash of fragmentary info from various periods. Parts of the western list might (I believe) date back to Valentinian and Valens, while other parts have been updated to the 420s. The eastern list may date to c.392, or be an entirely new construction of c.450 (if we follow Anthony Kaldellis's recent work), or be another mix of various periods. Either way, the units listed almost certainly did not exist at the same time.

Units sizes are another unknown, of course. I've written in another thread about the difficulties of attempting to judge the size of late units by the size of later forts. Quite possibly these traditional unit titles no longer connoted a set number of troops by this date.

While we're on the subject of the Notitia, the idea of the Legions being divided into thirds seems to fit with the numbers we're given, at least by my calculations. In DB Campbell's article "Did Diocletion overhaul the Roman Army," he mentions that Emil Ritterling identified 56 Legions that were in existence by the end of Diocletian's reign. If we exclude 9 which are only evidenced much later according to Campbell, 47x3 gives us 141 reduced Legions; by my count, the ND lists 103 Legions of all kinds (Palatinae, Comitatenses, Limitanei, Pseudocomitatenses) in the East and 90 in the West, which means there should have been 193 in either half by the 440s, going by Kaldellis' arguments. 52 1600-man Legions raised between, say, AD 324 when Constantine became sole ruler of the Empire and c. 440, is a rate of c.2 per year. That seems quite feasible alongside the raising of other units and incorporating new barbarian numeri.