Cavalry Cohorts

[Nathan Ross]

Zosimus creates the confusion by terming units of various sizes a tagmata.

Yes. But it's not just Zosimus - Cedrenus also uses inexact terminology, as does Ammianus and every other late antique writer. We need to distinguish the vague, colloquial or anachronistic terms used in literature from the official terminology of the day.

Luckily we have the evidence of inscriptions and the lists of the ND to tell us what Roman military units of the later period were actually called.

By the end of the 4th century, units of any size were called numeri (Greek arithmoi) - the term is near ubiquitous, and covers anything from a legion to a small unit of cavalry or frontier limitanei. Hence the apparent confusion in the literary sources: they are using the same words to describe units of variable types and sizes.

If a tagmata was 600 men, what would a 300 man unit be called? 

From Maurikios's Strategikon:

"A moira is made up up tagmas, arithmoi or bandons. A count or tribune commands the tagma, arithmos or bandon." (Strat, 1.3)

So the terms were interchangeable in the 6th century - I suspect bandon was the same sort of cavalry unit as a vexillatione (named for its banner or standard); arithmos (numerus) was the same thing. Either could be referred to by later writers as a tagma, or (presumably, by Ammianus) a 'cohort' or 'turma'.

"Tagmas should be formed in strength from three hundred to four hundred at most, and counts, also called tribunes... should be placed over them. ... The tagma should not exceed four hundred men, except in the bandons of the Optimates... All of the tagmas should definitely not be the same strength. If they are, the enemy can easily determine the size of the army by counting standards. Still, the statement we have made above should be observed, that the tagma should not contain less than two hundred men or more than four hundred." (Strat, 1.4)

So here we have the system of the 6th century - tagmata aka arithmoi (aka numeri!) of 200-400 men. We know from Zosimus that the units he calls tagmata of earlier centuries could number over a thousand: these are numeri of the 4th century type. But there was, I think, no unit in the Roman army of the 4th century officially called a tagma.

So Ammianus's 'cohorts' and 'turmae' are numeri (either equitum/vexillationes or auxilia).

Zosimus's 'tagmata' and Sozomen's arithmoi are also numeri (legions or auxilia)

Cedrenus's hippeon tagmata are also numeri - the numeri equitum or vexillationes that formed the mainstay of the later Roman cavalry force.