Years ago, Sterling Dow determined that Athenians who appeared next to one another with the same patronymic in ephebic lists were not true twins but brothers born close in time, commenting 'of course the presence of non-twins in even one list proves that at a very early period the age-limit was altered. Brothers served together as (full) epheboi' (TAPA 91  391). Although Dow's insight has been commonly accepted, its implications have not yet been explored. After confirming through Hellin's Law that these ephebes with homonymous patronymics at Athens and elsewhere in the Greek world were not actually twins, I will use comparative evidence from demographic studies of fertility, infant mortality, and birth spacing to argue that the presence of two or more siblings in the same ephebic list substantially affects the interpretation of the nature and frequency of Greek ephebic training. For example, even without the post-partum amenorrhoea characteristic of breastfeeding, average birth spacing (for offspring of both sexes) among young, healthy Western mothers was 18 months in the early 1960s (Potter, Population Studies 17.2  157). In the highly unlikely event that all the ephebes in extant lists with homonymous patronymics were male siblings born consecutively who survived the perils of disease and nutrition besetting children in antiquity to reach ephebic age in their middle to late teens, the age spread between them would thus have ranged from over one year in the case of two brothers to more than four in the several cases of 'triplets' (e.g. IG II2 2193; SEG 20 741; I. Kios 16) and 'quadruplets' (e.g. Michel, Recueil 643 II). More plausible, however, is that a good number of infants born consecutively either died young or were girls, so ineligible for the ephebate. Consequently, many if not most ephebic brothers may have been further apart in age.
While the age bunching seen in ephebic lists would have had no particular significance at cities like Athens that had large numbers of potential ephebes, whether native or foreign, each year during the later Hellenistic and early Roman periods, the situation was different in towns of more modest population, which would likely have mounted ephebates only when a sufficiently large cohort of young men was available. This phenomenon would account for the appearance of multiple siblings in the same list and could have been one of the factors underlying the dramatic reduction in the quantity of evidence for ephebates in the centuries of Roman domination. Ephebic systems might be interrupted or even abandoned (e.g. Sparta in the 3rd and 2nd centuries BCE [Kennell, Gymnasium of Virtue, 9-13], Priene after the First Mithridatic War [I.Priene 114], possibly Beroia in the second century CE [Nigdelis & Souris, jAnquvpato~ Levgei, 2005], and Antioch by the third century CE [Libanius, Or. 11.157), making irregularly held ephebates less unusual than modern scholars may suppose. Some evidence in fact suggests that Macedonian Beroia held its ephebate every four years in the third century CE (E. Bevroia" 137, 138). Finally, the high proportion of 'twins,' 'triplets,' and 'quadruplets' in several late ephebic lists (IG II2 2193, 2239, 2245) indicates that even Athens was forced to reduce the frequency of its ephebate during that same tumultuous century.
Paideia and Polis: The Ephebate and Citizen Training