The Roman census, conducted every five years in the Republic and irregularly under the Empire, is both a valuable historical source and a source of persistent confusion. The figures preserved in ancient sources are hard to interpret because we're uncertain what they count.

The Republican census counted adult male citizens — a relatively clear category. The consular census of 70-69 BC showed 900,000 or so; the census of 28 BC, the first under Augustus, showed 4,063,000. The jump is enormous. The 'high count' interpretation — that Augustus counted all citizens, including women and children, possibly including those across the empire — would mean the Republican citizen population was relatively modest. The 'low count' interpretation — that Augustus counted adult males only — implies the citizenship rolls had grown enormously, possibly through mass enfranchisement.

The argument has practical consequences for how we understand Roman demographic history. The low-count interpretation implies a serious population decline in the 1st century BC (consistent with the civil wars); the high-count interpretation implies a more stable population trend. The two interpretations imply very different population sizes for Roman Italy and the empire as a whole.

P.A. Brunt's Italian Manpower and Eitan Morley's Rome and the Barbarians represent the best scholarly treatments of this problem. The honest answer is that we cannot determine the correct interpretation with current evidence.