[244] Between 70 and 107 the garrison of the Danubian provinces was increased to ten legions, chiefly at the expense of the Rhine army, in which the legions were reduced from eight to four. [245] In 69, when Antonius Primus boasted of the superiority of the Danubian cavalry, there were, according to Tacitus, sixteen alae in Pannonia and Moesia (Tac. _Hist._ iii. 2). In the second century seventeen regiments can be traced in the two Pannonias and Moesia Inferior, while in Dacia, which covered Moesia Superior, were ten more. These figures, moreover, are probably below the real total. See Appendix I. [246] It appears in a Pannonian diploma for 98 (D. xxvii) and in the first Dacian diploma for 110 (D. xxxvii). [247] See _Archaelogiai Ertesitö_, 1905 and following, and for the inscriptions _A. E._ 1906 and following. [248] D. lviii (138-46), iii. 3331. [249] iii. 10316, 10318. _A. E._ 1906. 110. Ib. 1909. 150. Ib. 1910. 137. [250] D. lviii. The name of the town is missing. [251] _A. E._ 1910. 141. Cf. 133. [252] D. xxvii, xxxvii; iii. 4371. Another inscription (iii. 4368) mentions a Batavian, but he is a decurion who may have been transferred on promotion from another corps. [253] For orientals on the Rhine, cf. xiii. 7512, 7514. [254] Throughout the Empire the archer regiments seem to have been exclusively Thracians or orientals, but the latter alone preserved their national character in the second century. [255] iii. 10315, 10316. [256] The difficulty is to establish clear cases of men who must have entered a regiment after its original formation. The ‘Britto’ of the Dacian diploma for 145-61 (lxx) seems to be one. [257] ‘Nectovelius natione Brigans’ in the Cohors II Thracum. The inscription comes from Mumrills and probably dates, therefore, from between 142 and 180. _Eph. Ep._ ix. 623. [258] The _numeri_ have been discussed by Mommsen in the latter part of the _Conscriptionsordnung_, a discussion which naturally forms the basis of the following pages. [259] xiii. 6526, 6542, 6592, 6622, 6629, 6642, 7749. Elantienses—6490. Gurvedenses—7343. Murrenses—6471. Triputienses—6502, 6511, 6514, 6517, 6518, 6599, 6606. [260] _A. E._ 1910. 152. [261] viii. 2486, 2505, 18007, 18008, 18026, &c. [262] iii. 837, 907, 7999, 14216. [263] The Palmyrene _vexillarius_, whose tombstone was found at Corbridge in 1911, is most likely to have belonged to a _numerus_ formed from his countrymen. _Eph. Ep._ ix. 1153a. [264] D. lxvii (158). [265] _Eph. Ep._ ix. 1191, where all the references are collected. [266] iii. 8032. [267] viii. 21015, 21017. [268] iii. 7493. [269] If Mommsen’s interpretation of _Eph. Ep._ vii. 957 as n(umerus) m(ilitum) S(urorum) S(agittariorum) be correct. [270] Hyginus, 29. Accepting this emendation for the meaningless Getati of the manuscripts. [271] iii. 12601 a and b, 12605. The inscriptions date from the reign of Hadrian, showing that this usage was an early one. [272] Hyginus, 30.