This computation is not very different from that which M. Brotier, the last editor of Tacitus (tom. ii. p. 380), has assumed from similar principles; though he seems to aim at a degree of precision which it is neither possible nor important to obtain.