The authors of a recent article in this journal suggest that ‘water supply arrangements can be interpreted to provide urban population figures’. They offer conclusions constructed on this basis about the population of cities in Roman Cyrenaica. (J. A. Lloyd, P. R. Lewis ‘Water supply and urban population in Roman Cyrenaica’ Eighth Annual Report 1976–7, 35–40, at 36.)

Useable information about ancient city population remains perennially elusive. If a valid criterion for deducing population from physical remains could be established, it would be a highly important addition to the tools available to the archaeologist and the ancient historian. But before any such criterion can be achieved, we must escape from faulty methods. Demographic inference from aqueduct capacity should be recognised as a blind alley.

Difficulties can be indicated briefly.

1. The authors note that the ratios of aqueduct capacity to urban area at Ptolemais and Berenice differ widely. They explain this by noting that the ancients could not measure water flow at all accurately. If that is conceded, the attempt to derive useable population figures from aqueduct capacity fails on its own terms, since we have no means of discovering which ancient hydraulic calculations (if any) were accurate. The authors choose to regard the Berenice figure as accurate and the Ptolemais figure as inaccurate (by a factor of 2). But there is no real reason why one should be preferred to the other, once the likelihood of erroneous calculation has been admitted.