The last question to be settled in this connexion is that of the _cohortes equitatae_, in which a proportion of the men were mounted, which form a peculiar and interesting feature of the imperial army. Corps in which infantry and cavalry fought together had of course always been common,[54] but the idea was probably revived by the Romans from observing the practice of the German tribes, from whom Julius obtained a contingent accustomed to fight in this manner.[55] It is certainly significant that one of the earliest of these regiments known to us from inscriptions is a _cohors Ubiorum_.[56] There is, however, no later evidence for the employment of these tactics, and the continued use of _cohortes equitatae_ is due rather to the necessity of having detachments of mounted men at as many frontier stations as possible. The _equites cohortales_ should be reckoned rather as mounted infantry than cavalry, since we learn from a fragment of Hadrian’s address to the army in Africa that they were worse mounted than the _equites alares_, and less skilled in cavalry manœuvres.[57] As regards the strength of these regiments and the proportion of mounted to unmounted men, Hyginus states that the _cohors miliaria equitata_ contained 760 infantry and 240 cavalry, while the _cohors quingenaria_ contained six centuries, and in other respects, ‘_in dimidio eandem rationem continet_’—that is to say, it apparently had 380 infantry and 120 cavalry.[58] The figures for the mounted men are probably correct, and, since we learn from an inscription that there were four decurions to a _cohors quingenaria_, we may presume that the _turmae_ were 30 strong.[59] This agrees very well with the Egyptian vexillation cited above, which included 61 _equites cohortales_—that is to say, 2 _turmae_. On the other hand, there is considerable reason for supposing that the figures for the infantry are schematic and incorrect. It is sufficient here to remark that centuries of 76 could not be divided into _contubernia_ of either 8 or 10, and that the 380 men of Hyginus’s _cohors quingenaria_ could not even be divided evenly among six centuries. The question cannot be settled with certainty until forts occupied by regiments of this class have been planned, but it seems probable that while the number of the centuries remained unaltered the complement of each was reduced from 80 to 60, or possibly to 64, if it was thought desirable to retain the division into _contubernia_ of 8.[60]