The Roman army was not a “demographically closed institution.” While veterans’ sons formed an important recruitment block (evidenced by the number of recruits who listed their birthplace as “in castris”, or “in the camps”), the army did not reproduce itself demographically. During the Principate, most recruits would not have been born to a veteran.

While the exact rational behind the Augustan marriage ban is unclear, it was not a policy designed to create a body of illegitimate children whose only path to citizenship lay in military service, as there is a lack of evidence suggesting the granting of citizenship upon enlistment. The marriage ban thus did not serve to create a permanent class of recruits, and the idea of “internal replacement”, or recruits largely coming from veterans’ sons, does not stand up demographically.

Determining the number of annual recruits needed by the army of the Principate is difficult, as it revolves around the issue of military mortality. While the exact odds of a recruit surviving his twenty to twenty-five year term is impossible to calculate, infectious disease, combat, and early discharge for invalidity would have been significant factors affecting survivability. It is not unrealistic to assume that about half of all recruits would not live to be discharged. Those who did survive would have spent roughly three-quarters of their adult lives in service, often in difficult, hostile environments. Many would not live long after discharge. Such circumstances are not conducive to the raising of families, even though liaisons with local women, often stable and long-term, were common.

Assuming a population of 60-70 million, and an annual discharge rate of 6,000-7,000, Walter Scheidel estimates the army of the Principate required roughly 15,000 recruits per annum. This figure represents roughly 2.5% of the “military demographic” (young men about twenty years of age); though there was likely much regional variation. This should not be taken as an exact figure, rather an estimate arrived at through various assumptions based upon unclear source data. While the annual discharge figure would have resulted in a sizeable population of former soldiers (most of whom appear to have remained in their area of service), it is unlikely the army was able to reproduce itself. Most veterans would have been discharged at a relatively late age (Scheidel’s calculations assume an average age of 45, though epigraphic evidence attests to men serving until they were much older), and epigraphic evidence also suggests many soldiers formed de-facto marriages at such a late age. The high mortality levels due to infectious disease and combat necessitated uneven recruitment flows, with certain units having potentially higher needs. The army also recruited from non-Romans, often targeting specific groups (like the Batavians) because their martial cultures were deemed to produce quality-recruits.

There was never a systematized policy of, nor historical precedence for, internal recruitment. Soldiers did not form an hereditary caste. While veterans’ sons formed an important recruitment block, most recruits would have been locals, drawn to the army by a combination of push and pull factors. Local recruitment remained the dominant form throughout the Principate.

For further reading, see Walter Scheidel’s chapter “Marriage, Families, and Survival: Demographic Aspects” in The Blackwell Companion to the Roman Army (2007).