311 Josephus, Bell. Iud. iii. 5.
312 He was not, however, an ‘unmilitary historian’ in the sense that, for instance, Ephoros was. Ephoros made elaborate accounts of military operations an important feature of his work, although he was quite lacking in military knowledge (Polybius, xii. 25); Tacitus never pretends to concern himself with more than the moral and social aspects of war. The same attitude may be observed both in Dio and Herodian (ii. 15, 6). This attitude was perfectly justifiable, since there existed, as we learn from this passage in Herodian and from Lucian (De Hist. Conscrib.), a technical literature which would probably satisfy our needs. That we do not possess it is the fault, not of Tacitus and Dio, but of the Middle Ages.
313 This was still the case at Argentorate in 357; cf. Ammianus, xvi. 12.
314 Tac. Ann. ii. 16.
315 Tac. Ann. ii. 52 ‘Legio medio, leves cohortes duaeque alae in cornibus locantur’.
316 Tac. Ann. xiii. 38 ‘Socias cohortes et auxilia regum pro cornibus, medio sextam legionem constituit’. The defensive formation described in xiii. 40 is slightly different, since the legions in the centre formed a square.
317 Tac. Ann. xiv. 34 ‘Igitur legionarius frequens ordinibus, levis circum armatura, conglobatus pro cornibus eques adstitit’.
318 Tac. Hist. iii. 21 ‘Cohortes auxiliorum in cornibus, latera ac terga equite circumdata’.
319 The legionaries were to occupy the centre, the auxiliary infantry high ground on the wings, the cavalry to wait in the rear.
320 Dio, lxxiv. 7.
321 Tac. Agr. 35 ‘Legiones pro vallo stetere, ingens victoriae decus citra Romanum sanguinem bellandi, et auxilium, si pellerentur’. It was not necessary for Tacitus to invent this not very creditable excuse. The tactics are those adopted with equal success against a Highland army by the Duke of Cumberland at the battle of Culloden in 1746. The same idea of checking the impetus of a Celtic charge by successive obstacles has often been suggested as the reason for the seven ditches which protect the exposed side of the fort at Whitley Castle, and the lilia at Rough Castle on the Antonine Wall.
322 They are particularly noticeable in the battle shown on Cichorius, Die Traiansäule, Pl. 45, and to judge from the column, the auxilia did more than their usual share of fighting in this war.
323 In spite of the boasting of Antonius Primus, the achievements of the Pannonian cavalry in 69 were limited to a reckless advance followed by a disorderly retreat. Cf. Tac. Hist. iii. 2 with iii. 16. In the second century, however, we find heavy cavalry, contarii, who must have been intended for shock tactics.
324 For Haltern see Schuchhardt, Führer durch die Ausgrabungen von Haltern. It is, however, perhaps incorrect to limit the garrison to auxiliaries: for Hofheim cf. Ritterling, Das frührömische Lager bei Hofheim, 1912. It was occupied from about 40 to 60.
325 See Pelham, Essays in Roman History, p. 191.
326 e.g. Friedberg has an area of (roughly) 10 acres, Okarben of 14, Heddernheim of 13, and Kesselstadt of 35. A cohort of 500 infantry was usually allowed about 5 acres.
327 See Macdonald, The Roman forts at Barr Hill, pp. 11-15; Curle, A Roman Frontier-post, pp. 29, 349.
328 xi. 6344 ‘P. Cornelio P. f. Sab(atina) Cicatriculae prim(o) pil(o) bis, praefect(o) equit(um), praef(ecto) clas(sis), praef(ecto) cohortium quattuor civium Romanor(um) in Hispania, trib(uno) mil(itum)’.
329 iii. 14147².
330 W. D. Z. xxi. 186, where this theory of the first-century frontier system is further developed.
331 See Pelham, op. cit., p. 199. The forts on the North British frontier range in area from 2½ to 5½ acres, the largest (Amboglanna) being designed to hold a cohors miliaria peditata. The German forts seem to have been on a rather larger scale, and ran up to 15 acres, which is the area allowed at Aalen to an ala miliaria. It is not meant, of course, that forts of this type did not exist in the first century, but it was not until the reign of Hadrian that the dispersion of the auxilia in separate units was adopted as a general policy.
332 Cf. iii. 3385 ‘(Commodus) ripam omnem burgis a solo extructis item praesidis per loca opportuna ad clandestinos latrunculorum transitus oppositis munivit’. This is from the Danube frontier.
333 Historia Augusta, Vita Hadriani, 12 ‘In plurimis locis in quibus barbari non fluminibus sed limitibus dividuntur, stipitibus magnis in modum muralis saepis funditus iactis atque conexis barbaros separavit’.
334 The best recent account of this frontier in English is Pelham’s essay, ‘The Roman Frontier in Germany,’ in the work already cited.
335 Recent researches have, however, made it very doubtful whether a turf wall ever preceded the stone wall along the whole line.
336 For this wall see the admirable account by Dr. George Macdonald, The Roman Wall in Scotland, Glasgow, 1911.
337 The best description of the internal arrangements of a mile-castle is given by Mr. F. G. Simpson, in the Transactions of the Cumberland and Westmoreland Archaeological Society, vol. xi, New Series.
338 The German Limes-Commission has not yet published its report on these works and the course of the frontier line in general. On the Walldürn-Welzheim section, which was built by Pius, the towers are at intervals of from 250 to 400 metres. Pelham, op. cit., p. 204.
339 The view with which I find myself in most agreement is that of Delbrück, Geschichte der Kriegskunst, ii. 155-60.
340 See Pelham, op. cit., p. 201 and appendix.
341 The torches which project from the upper stories of the block-houses represented on the Trajan column have often been noted as indicating some method of fire-signalling.
342 The palisade was not, of course, a board fence which could be torn down in a few minutes. It was made of oak-trees split in halves and bedded in a ditch four and a half feet deep. See Pelham, op. cit., p. 200. Some idea of its appearance can be gathered from the representation at the beginning of the reliefs on the Marcus column.
343 Oman, Art of War in the Middle Ages, p. 209. He quotes Nicephorus Περὶ παραδρομῆς πολέμου.
344 The stone wall on the Raetian frontier is of very inferior construction.
345 The garrisons were: at Netherby, the Cohors I Aelia Hispanorum M. E., vii. 954, 963, 964, 965; at Bremenium, the Cohors I Fida Vardullorum M. E. and exploratores, vii. 1030, 1043; at Habitancium, the Cohors I Vangionum M. E., exploratores and Raeti Gaesati 1002, 1003. The latest inscription at Netherby dates from the reign of Severus Alexander; at Bremenium, the most advanced of the Dere Street forts, from the reign of Gordian III.
347 Cf. Historia Augusta, Vita Marci, 12 ‘Dum Parthicum bellum geritur, natum est Marcomannicum, quod diu eorum, qui aderant, arte suspensum est’.
348 The only changes which we know of are that Legio V Macedonica was transferred from Moesia Inferior to Dacia in the reign of Marcus and that Legio III Augusta was sent by Gordian III as a punishment from Africa to the Rhine, whence, however, it returned in 253. Cagnat, L’Armée romaine d’Afrique, pp. 156-61.
349 The date of this section of the Notitia is disputed, but it can hardly be earlier than the end of the third century. The diplomata are xxix (98), xxxii (103), xxxiv (105), xliii (124), lv (ante 138), lvii (146).
350 Cf. D. ii and iii. 3664, D. lviii and iii. 3331, and D. lx and iii. 11333. These regiments may of course have been temporarily absent, but the evidence is fairly continuous in each case. The Cohors V Lucensium et Callaecorum, for example, appears on diplomata for 60, 84, 85, 133, 138/48, 148, 149 and 154.
3511 I. G. R. R. i. 1348, ib. 1363, iii. 14147², D. xv, iii. 14147¹. In the last, which dates from 39, the number of the cohort is not given, and possibly another in the same series is meant.
352 For the date of the Egyptian section of the Notitia see my article, The Garrison of Egypt, in the account of the excavations at Karanóg by the Eckley B. Cox Junior Expedition to Nubia published by the University Museum, Philadelphia.
353 iii. 4379 (3 Ulpii, 2 Aelii), 4360 (Aelius), 4369, 4370 (Aurelii), 11081.
354 iii. 1371, 1372, 1374, 1379; A. E. 1903. 66.
355 vii. 818 (Severus Antoninus), 819 (Gordian III), 820 (Postumus), 823 (Tetricus). Another inscription (808) dates from the reign of Maximin.
356 Tac. Ann. iv. 47; id. Hist. ii. 22.
357 Cf. Dio’s remarks on the impression made by the provincial legionaries in Rome in the reign of Septimius Severus, lxxiv. 2. The following sketch applies only to the troops on the Western frontiers, concerning whose life we have considerable evidence. The locally raised troops in the East started as a rule at a higher level of culture, but possibly a similar advance was made by Trajan’s regiments of Paphlagonians, Galatians, and Arabians, although here Hellenization, not Romanization, was of course the goal.
358 For the importance of Mithras in the army, cf. Cumont, Les Mystères de Mithra. Toutain, Les Cultes païens dans l’Empire romain, cc. ii and iv, gives a classified list of the inscriptions of Mithra and Dolichenus.
359 H. Jacobi, Führer durch das Römerkastell Saalburg, 1908, gives a summary of the latest results.
360 Excavations so far have been confined to the fort itself, in which the buildings were in an exceedingly good state of preservation, and the Mithraeum.
361 xiii. 7800 ‘Petronius Athenodorus prae(fectus) coh(ortis) I Fl(aviae) horologium ab horis intermissum et vetustate colabsum suis inpendis restituit’. The date is 218.
362 Herodian iii. 8, 5.
363 _Rheinisches Museum_, xlviii. 616 ff.
364 viii. 18067.
365 _A. E._ 1910. 75.
366 Cagnat, _L’Armée romaine d’Afrique_, pp. 380-3 and 505-7.
367 e.g. D. iii (64), a wife, son, and daughter; D. xcviii (105), a wife, son, and two daughters; D. xxxvii (110), three sons. See above, p. 32.
368 iii. 3271. The approximate date of the inscription is sufficiently indicated by the names employed.
369 This very comprehensive dedication comes from the shrine of Mandoulis, the source of many military inscriptions.
370 The possibility that cives had already been admitted into the auxiliary regiments before this date has already been discussed. See above, p. 33.
371 iii. 10316 and _A. E._ 1910. 144.
372 The campaigns between Severus and his rivals (193-7) were fought out by vexillations; hence at the end of the war we find all the regiments on both sides, so far as they can be traced, in their old quarters.
373 Historia Augusta, Vita Alex. Sev. 58 ‘Sola, quae de hostibus capta sunt, limitaneis ducibus et militibus donavit, ita ut eorum essent, si heredes eorum militarent, nec umquam ad privatos pertinerent, dicens attentius eos militaturos, si etiam sua rura defenderent’. The theory of the self-sufficiency of each provincial garrison could not be more clearly expressed.