considerations led Mommsen to abandon his former view, to assume instead a single century of the kind; and recent writers are inclined to follow him.[1230] Lowest in rank of the supernumerary centuries was that of the proletarians.[1231] The government so designated those citizens who owned no land,[1232] and hence were poor. They were exempt from military duty, excepting in so far as they served with arms furnished by the state.[1233] Though few in the beginning, their number gradually increased till in the time of Dionysius[1234] it exceeded all the five classes together. At some time in the early history of the comitia centuriata they were formed into a century and given one vote,[1235] which was not counted with any class but was reported after all the others. Dionysius[1236] wrongly speaks of it as a sixth class. The existence of this century is due to the principle that no one should be excluded from the right to vote on account of poverty.[1237] Six supernumerary centuries have now been mentioned and the place of three—the two industrial and the one proletarian—in the voting system has been considered. With reference to the others Dionysius assigns the musicians to the fourth class, Livy to the fifth. The settlement of this question is aided by an examination into the total number of comitial centuries of the fifth class. It is given as thirty by the sources.[1238] Assuming this to be the correct number and adding to the sum of centuries in the five classes (170) the six supernumerary centuries and the eighteen centuries of knights to be considered below, we should have in all a hundred and ninety-four, which would be one too many. In an earlier chapter, however, the conclusion was reached that there were but fourteen military centuries in the fifth class.[1239] Two of the thirty centuries assigned to that class In the comitia centuriata must therefore have been in fact supernumerary. If one was the accensi, what was the other? Most probably it was the century of the tardy described by Festus,[1240] made up at each meeting of those who came too late to vote in their own classes. Obviously all writers who apply the discriptio centuriarum to the army view this century, as well as that of the proletarians, with suspicion.[1241] The two centuries of the accensi and the tardy should be included among, not added to, the thirty of the fifth class.[1242] Having reached this result, it might seem advisable for us to assume no further supernumerary centuries for the fifth class, but to follow the authority of Dionysius in assigning the musicians to the fourth. Or as the trumpeters preceded the hornblowers in rank, it might be plausibly argued that the former belonged to the fourth and the latter to the fifth. In this way a compromise could be effected between Livy and Dionysius, and Livy’s three supernumerary centuries of the fifth class could be explained. Absolute certainty is unattainable. The notion of Dionysius