^{page 1 note 1} Cf. Böcking, Not. dign. p. 223, note 35. This legion seems to have been raised by Severus Alexander for his intended eastern campaign. It was certainly in Egypt by Diocletian's reign. Under Constantine I it was disbanded as a frontier-legion and its name survived only as that of a legio comitatensis (Nol. dign., Or. viii, 51, Julia Alexandria).

^{page 2 note 1} ‘Das römische Militärwesen seit Diocletian,’ Hermes xxiv (1899), p. 228Google Scholar.

^{page 2 note 2} Gesch. des Untergangs der antiken Welt. ii, p. 33 and note.

^{page 3 note 1} Römische Militärgeschichte von Gallicnus bis zum Beginn der byzantinischen Themenverfassuung (1920), pp. 57 f.

^{page 4 note 1} Zosimus iii, 34; Vict. Caes, xli, 12; Lact., De mort. pers. 7; finally an extract in Suidas which apparently comes also from Zosimus.

^{page 4 note 2} De mens, 1, 27.

^{page 4 note 3} C.I.L. vi, 2759, 2787, 32965. See p. 55 infra.

^{page 4 note 4} C.I.L. iii, 6194.

^{page 5 note 1} The reading ‘Metis’ (Occ. v, 269) or, better, ‘Mettis’ seems to me preferable to ‘Martis’ (Occ. vii, 95).

^{page 6 note 1} Note, however, Upper Moesia (IV Flavia: 1 station), Dacia Ripensis (V Macedonica: 4 stations; XIII Gemina: 5 stations), Valeria (II Adjutrix: 6 stations), and Upper Pannonia (X Gemina: 1 station). These are exceptions.

^{page 7 note 1} Müllner, in Emona, 1879, p. 186; Argo viii, p. 201 f, 220 f; ix, II f, 29 f. Premerstein and Rutar, Römische Strassen und Befestigungen in Krain: 1899. Last surveyed during the European War by W. Schmid, some of whose results have been published.

^{page 8 note 1} Throughout this article the use of square brackets indicates that the unit concerned is not actually mentioned in any of the ancient authorities, although there is other evidence from which its existence may be inferred. In the present case, for example, we may safely argue back from a ‘II Flavia Constantiniana’ to a ‘I Flavia Constantiniana.’

^{page 8 note 2} The reasons for the disposition I here suggest for I Flavia Victrix, IV Italica, and I Flavia Constantia cannot be considered until we come to the reforms of Constantine with which they stand in close connexion : see pp. 21, note 7, 23 ff.

^{page 9 note 1} See note 2, page 8.

^{page 9 note 2} Dio, lv, 24.

^{page 10 note 1} Ann. iv, 5. The calculation of Tacitus includes among the auxiliaries the crews of the ‘sociae triremes.’

^{page 10 note 2} Cf. Cohors XIV Valeria Zabdenorum (Or. xxxvi, 36).

^{page 10 note 3} Hermes xxiv, p. 275.

^{page 11 note 1} e.g. Cohors I Jovia—Cohors III Herculia; Cohors I Flavia—Cohors XII Valeria.

^{page 11 note 2} e.g. Ala I Flavia Raetorum—Ala VIII Flavia Francorum.

^{page 11 note 3} e.g. Ala I Valeria dromedariorum—Ala II Herculia dromedariorum.

^{page 11 note 4} e.g. Cohors I Herculia Raetorum—Cohors III Herculia Pannonicorum—Cohors VI Valeria Raetorum.

^{page 11 note 5} e.g. Cohors I Flavia Sapandica—Cohors II Flavia Pacatianorum—Cohors V Valeria Frygum.

^{page 11 note 6} Gesch. des Untergangs, i, p. 255.

^{page 11 note 7} Lactantius, De mort. pers. 7.

^{page 11 note 8} There were in Noricum:

in 69 (Tac. Hist. iii, 5) at least 1 ala 8 cohorts

in 153 (Diploma no. lxiv) 4 alae 14 cohorts

c. 400 (Not. dig. Occ. xxxiv) 6 alae 3 cohorts

Simultaneously with the disappearance of the majority of the independent cohorts from Noricum a new legion, I Noricorum, makes its appearance there. We may conclude that this was no mere chance coincidence, but that Diocletian had thought it advisable to form a new legion by the fusion of a number of separate auxiliary cohorts. We may perhaps find a further indication of this in the fact that inscriptions and other finds relating to Legio I Noricorum begin at the exact date when those relating to the cohorts apparently come, for the most part, to an end.

^{page 13 note 1} Probable additions are enclosed within square brackets; see supra p. 8, note 1. The placing of the title ‘seniores’ within parentheses indicates that, when the field army was first constituted, each half of the Empire contained only a single unit bearing the name to which the title is attached, When similarly named units were subsequently raised in the same half of the Empire, the latter were distinguished as ‘iuniores,’ while the original units naturally became ‘seniores.’

^{page 17 note 1} As it is uncertain to which half of the Empire the four missing regiments of Equites Dalmatae (numbered I, II, IV, VII) belonged, they have been equally divided between East and West in the lists and in the above total.

^{page 17 note 2} e.g. Equites promoti seniores (pal.)—Equites promoti iuniores (com.).

^{page 18 note 1} Honoriani felices Gallicani (Occ. v, 247 = vii, 89); Lanciarii Gallicani Honoriani (Occ. v, 239 = vii, 81); Honoriani Mauri [iuniores] Gallicani (Occ. v, 220 = vii, 52).

^{page 19 note 1} Honoriani Atecotti seniores and iuniores; Honoriani Mauri seniores and iuniores.

^{page 19 note 2} Having regard to the fact that the old legions were certainly not up to strength, I reckon those in the West and those in the Eastern Danube provinces at 4,000 men apiece, those elsewhere in the East at 3,000.

^{page 19 note 3} Only I Isaura was disbanded. II and III continued to exist as frontier-legions with their old organization and establishment.

^{page 20 note 1} While the vexillationes palatinae and comitatenses of the Notitia are cavalry regiments 500 strong, the ‘vexillations’ here referred to are detachments similar to those which it was customary in earlier centuries to draw from one or more regular units for some definite purpose (a particular campaign or more than usually important field-works) and to group under a single standard, the vexillum, whence their name. This was done when for one reason or another it was thought inadvisable to withdraw complete units from a province. Cf. C.I.L. iii, 600, ‘praepositus in Mesopotamia vexillationis equitum selectorum alarum …, item cohortium …’; C.I.L. iii, 1980Google Scholar, ‘vexillationis leg(ionis) II Piae et III Concordiae.’

^{page 20 note 2} A probable explanation of the omission of ‘seniores’ is that the seniores and the iuniores were stationed in different halves of the empire, and there could thus be no confusion. I think we should be justified in connecting the Primani and Secundani—seniores and iuniores alike—with the I and II Adjutrix, respectively. Apart from the fact that this is the only way in which they can conveniently be fitted into the general system of vexillations, the circumstance that in the West they appear as a pair points to the same conclusion.

^{page 20 note 3} The title of the vexillation of the II Augusta combines the legionary number with the name of the province from which it came.

^{page 20 note 4} The legiones riparienses (frontier-legions) II Traiana and III Diocletiana, which were stationed in Egypt, were originally detachments of the similarly named frontier-legions stationed in the Thebaid. When the Egyptian portion of the frontier was made into two provinces, the detachments that were in Egypt remained there and became independent frontier-legions, though with a strength of only 1,000 men each.

^{page 20 note 5} The legiones riparienses V Macedonica and XIII Gemina, which we find in Egypt, are vexillations of the similarly named Dacian frontier-legions. They were not incorporated in the field-army, but were assigned to the Egyptian provincial army as independent frontier-legions, with a strength, however, of only 1,000 men each.

^{page 20 note 6} In Occ. v, only three regiments called Septimani are enumerated, while of the four mentioned in Occ. vii three are ‘iuniores’ and only one ‘seniores.’ The order of names in the Notitia suggests that the ‘iuniores’ of vii, 31 is a mistake for ‘seniores.’ Thus we have in v and vii:

v, 228 Septimani seniores vii, 31 Septimani

iuniores

229 Regii 32 Regii

The mistake of writing ‘iuniores’ for ‘seniores’ in vii may have been caused by the fact that in that passage the Septimani are immediately preceded by the Mattiarii iuniores (vii, 30). As for the other inconsistency, a unit has dropped out in chap, v—unless, indeed, this legion disappeared during the few years which separated the compilation of the two chapters. As the missing unit bore the same name as the other set of ‘seniores,’ it was omitted by the copyist either purposely—because he thought the two units were one and the same, and wished to correct what he took to be a textual error—or by accident. A similar instance of the conflation of two units occurs in chap. vii. In v (Occ.) we have:

198 Honoriani Marcomanni seniores.

199 Honoriani Marcomanni iuniores.

in vii, on the other hand, we find only.

38 Marcomanni.

^{page 21 note 1} A similar case to that of the Primani and Secundani, cf. p. 20, note 2.

^{page 21 note 2} Cf. p. 20, note 5.

^{page 21 note 3} Ammianus's description seems to suggest that this legion was comitatensis.

^{page 21 note 4} Not ‘Truncensimani.’ The order in which they appear m the Notitia and their assignment to Gaul suggest that they were pseudocomitatenses. As the above list shows, this is not the only instance in which a vexillation is assigned the position of a legio pseudocomitatensis.

^{page 21 note 5} These vexillations were drawn from the legions in Lower Pannonia or from those in Scythia (cf. p. 9), or partly from the one and partly from the other.

^{page 21 note 6} cf. p. 20, not e 4.

^{page 21 note 7} I regard this legio comitatensis as a vexillation of the I Flavia Victrix staioned in Uppear Britain: cf. p. 8. note 2, and pp. 4, ff.

^{page 21 note 8} Cf. p. 40.

^{page 21 note 9} Cf. p. 41.

^{page 24 note 1} Ann. iv, 5.

^{page 24 note 2} Bell. Jud. ii, 16, 4.

^{page 24 note 3} C.I.L. vi, 3492.

^{page 24 note 4} lv, 23.

^{page 24 note 5} The inscription found at Bordeaux in 1921 (Rev. des Études anc., 1922), showing that Lincoln was in Lower Britain, proves that this view is no longer tenable, and that the boundary must really have run further south (J.R.S. xi, p. 104). That, however, does not affect the main argument here.

^{page 25 note 1} Cf. Or. viii, 42 Constantini seniores; viii, 45 Constantini Dafnenses.

^{page 26 note 1} These deductions are of a purely hypothetical character and represent an attempt to establish the name of this fourth British legion. Its existence I regard at proven.

^{page 26 note 2} xxix, 6, 13.

^{page 27 note 1} Mommsen, Hermes xxiv, p. 230, dates the separation to the Constantinian period. Grosse, Militärgeschichte p. 15, assumes that it is as early as Gallienus.

^{page 28 note 1} According to Vegetius (ii, 6) each legion had 726 mounted men on its establishment.

^{page 28 note 1} Cf. Mommsen, ‘Die Conscriptionsordnung der römischen Kaiserzeit’ in Hermes xix, pp. 22 f.

^{page 29 note 1} Occ. xxv, 20; xxvi, 12.

^{page 29 note 2} Or. xxxix, 28; xl. 29.

^{page 29 note 3} With the exception of the two ‘cunei equitum’ (Or. xxxi, 23 and 24) and the single regiment of ‘milites’ (Or. xxxi, 35) in the Thebaid.

^{page 29 note 4} 72 ‘equites’ and 73 ‘alae.’ Here, and in the comparisons that immediately follow, I include the auxiliary regiments that date from Valentinian and Theodosius, because in the main they were raised only in substitution for units that had been annihilated or disbanded, a process that entailed no substantial alteration in the general perspective as contrasted with the period of Constantine.

^{page 30 note 1} Or. xxviii, 14 and 15.

^{page 30 note 2} Or. xxviii, 18 and 19.

^{page 30 note 3} Or. xxxi, 36 and 39.

^{page 30 note 4} So far as they were not ‘milliaria,’ unless indeed this name too had degenerated into an empty form.

^{page 30 note 5} Hermes xxiv, p. 209.

^{page 31 note 1} Mommsen here means the field-army or, to his own phrase, the imperial troops.

^{page 31 note 2} e.g. Mangold, , ‘Legionen des Orients auf Grund der Notitia Dignitatum’ in Rheinisches Museum N.F. 57 (1902) p. 262Google Scholar; Delbrück, , Geschichte der Kriegskunst im Rahmen der politischen Geschichte, ii ² (1909), p. 225Google Scholar; Wilcken in Mitteis, and Wilcken's, Grundzüge and Chrestomathie der Papyruskunde vol. i, pt. 1 (1912)Google Scholar; Grosse, Militärgeschichte, pp. 58, 90 f.

^{page 31 note 3} Partly too by legions, as in Spain (Occ. xlii, 26) and Isauria (Or. xxix, 7 and 8); also by legionary detachments.

^{page 31 note 4} Or. xl, 46, 48 and 49.

^{page 31 note 5} Occ. xlii, 27-30, and 32.

^{page 31 note 6} Occ. xlii, 17 and 19.

^{page 31 note 7} Occ. xlii, 6.

^{page 32 note 1} Still the numbers given here will not be far removed from the actual establishment in Constantine's time. Just as the frontier-legions, in consequence of their immobility, remained very much the same from Constantine to Honorius, so too the garrisons of the fortified zones in rear of the frontiers must have continued very much the same. Indeed, this would apply to them in an even higher degree, since their position was on the whole distinctly less exposed than that of the frontier-garrisons. Nevertheless, that we have got to reckon with the loss of units of the kind is proved by the instances of the V Parthica, and of the Praeventores and Superventores.

^{page 32 note 2} Amm. Marc. xviii, 9.

^{page 32 note 3} Amm. Marc. xviii, 9. The Superventores in the West (v, 220=vii, 96) are pseudocomitatenes, and we must assume that those in the East were so also. The correspondence of name, as well as the name itself (cf. Defensores, Exploratores, Insidiatores), suggests that the Praeventores belonged to the same category.

^{page 32 note 4} Analogy favours the existence of a legio pseudocomitatensis of this name: Occ. xxxvii, 16, Praefectus militum Maurorum Benetorum (i.e. Venetorum); Occ. xxxvii, 17, Praefectus militum Maurorum Osismiacorum.

^{page 33 note 1} Julia Alpina, II Julia Alpina, I Armeniaca, II Armeniaca, V Parthica, VI Parthica, I Isaura sagittaria.

^{page 33 note 2} The legio IV Jovia provided 2 auxilia palatina (Jovii seniores and iuniores, Occ. v, 168 = vii, 16, and Occ. v, 184 = vii, 42) instead of a single field-legion.

^{page 34 note 1} Raeti, Sequani, Jovii seniores and Jovii iuniores.

^{page 34 note 2} With the exception of Britain (Occ. xxviii and xl) and Occ. xxxv, 32.

^{page 35 note 1} Thus, for instance, only the northern part of Noricum (Noricum Ripense) remained under the command of the Dux, while the southern part (Noricum Mediterraneum), was placed under the direct control of the Comes of the field-army.

^{page 35 note 2} e.g. I Flavia Mettis; Romanenses (=cohors civium Romanorum).

^{page 35 note 3} e.g. Exploratores: Superventores.

^{page 35 note 4} e.g. Cornacenses; Scampenses.

^{page 39 note 1} C.I.L. iii, 5756, 11847.

^{page 39 note 2} C.I.L. iii, 5765 (Mautern), 11870 (Enns, Mauer-on-the-Url), p. 2328⁴² (Vienna); Mitteilungen der Zentralkommission, N.F., xvi (1893), p. 232Google Scholar (St. Pölten).

^{page 39 note 3} C.I.L. iii, 11349 (Ragendorf in the county of Wieselburg close to Gerulata, Sirmium), 11848 and p. 2328²⁰⁰ (Mauer-on-the-Url),p. 2348⁴² (Pürbach at the north end of the Neusiedler See).

^{page 39 note 4} ‘Vom norischen Donauufer, 2. Ad Iuvense’ Mitt. der Zentralkom., 3 F, v (1906), pp. 49 f.Google Scholar

^{page 39 note 5} Nischer, ‘Untersuchungen über die Römerstrasse von Wien nach Wels’ in Mitt. d. geographischen Gesellschaft in Wien, 1919, p. 110.

^{page 40 note 1} Miller (Itineraria Romana, col. 285) equates thus: Adiuvense = Castrum Iuvense = Salzburg. But this, apart from anything else, would split the legion up completely, and would involve a quite lop-sided division of the frontier-defence. Betides, even then we should still have to assume n corruption of the name.

^{page 40 note 2} C.I.L. iii, 11870, FIG IVES.

^{page 40 note 3} C.I.L. iii, 11848 and p. 2328²⁰⁰, FIGVLINAS IVENSIANAS LEG I NOR; iii, 11870, FIG IVES.

^{page 40 note 4} Jahrb. der Zentralkom. v (1911), p. 253Google Scholar.

^{page 40 note 5} All ten cohorts of the Legio II Italica are enumerated in the Notitia (Occ. xxxiv, 38, 39); only the text is partly corrupt.

^{page 41 note 1} Occ. xxxiv, 43, Praefectus classis Lauriacensis; 42, Praefectus classis Arlapensis et [Co]maginensis.

^{page 41 note 2} See supra p. 17.

^{page 43 note 1} Zosimus ii 32: δύο γὰρ τῆς αὐλῆς ὄντων ὐπάρχων (praefecti praetorio), καὶ τὴν ἀρχὴν κοινῇ μϵταχϵιριζομένων, οὐ μόνον τἀ πϵρί πὴν αὐλὴν τάγματα τῇ τούτων ᾠκονομϵῖτο φρόντιδι καὶ ἐξουσίᾳ, ἀλλὰ γὰρ καὶ τὰ ἐπιτϵτραμμένα τὴν τῆς πόλϵως φυλακὴν καὶ τὰ ταῖς ἐσχατιαῖς ἐγκαθήμϵνα πάσαις.

^{page 43 note 2} Zosismus ii 33: ἐφεστώτων γὰρ τοῖς ἁπανταχοῦ στρατιώταις ού μόνον ἑκατοντάρχων (centuriones) καὶ χιλιάρχων (tribuni), ἀλλὰ καὶ τῶν λϵγομένων δουκῶν (duces) οῖ στρατηγῶν ἑν ἑκάστψ τόπψ τάξιν ὲπϵῖχον, στρατηλάτας καταστήσας, τὸν μὲν τῆς ἵππου, τὸν δἐ τῶν πϵζῶν (‘magister equitum’ and ‘magister peditum prasentalis’), ϵἰς τούτους δὲ τὴν ἐξουσίαν τοῦ τάττϵιν στρατιώτας καὶ τιμωρϵῖσθαι τοὺς ἁμαρτάνοντας μϵταθϵίς.

^{page 43 note 3} Dagalaiphus, 365 A.D. (Amm. Marc. xxvi, 5); Sebastianus, 377 A.D. (Zosimus iv, 17). Cf. Amm. Marc. xxviii, 6; xxx, 5 and 10; xxxi, 8. Also the order of succession in the Notitia, Occ. i. 5 and 6, as well as chap, v and vi.

^{page 44 note 1} Amm. Marc. xxii, 3.

^{page 44 note 2} Amm. Marc. xxvi, 5.

^{page 44 note 3} Amm. Marc. xiv, 9; xxi, 8; xxiv, 1 and 4; xxvi, 1, 4, 5 and 9; xxvii, 2. Also Zosimus iii, 21.

^{page 44 note 4} Amm. Marc. xviii, 5.

^{page 44 note 5} Amm. Marc. xxvi, 5.

^{page 45 note 1} Cf. Amm. Marc. xxvi, 7, and xxxi, 16.

^{page 46 note 1} Authorities for the subordination of the ‘duces’ to the three local ‘magistri militum’ are : Theodosiani libri i, 7, 2; vii, 1, 9 and 18; vii, 17, 1 and 20, 13; xii, 1, 113; and xv, 11, 1; Nov. Theodosian. xxiv, 1 and 2; Cod. Justin. xii, 59, 8.

^{page 47 note 1} Cf. the action of Charietto, the ‘comes per utramque Germanism,’ Amm. Marc. xxvii, 1; also Zosimus, iii, 7.

^{page 51 note 1} Or. xxxi, 36, legio I Valentiniana, Copto; 39, legio II Valentiniana, Hermunthi.

^{page 51 note 2} Cf. Mommsen, , Hermes, xxiv, p. 214Google Scholar, Anm. 1.

^{page 52 note 1} Field-army: 4,000 horse, 7,000 foot; pseudo-comitatenses: 6,000 foot. To these have to be added 2,500 horse and 1,000 foot for the frontier-troops (5 alae, 2 cohorts). A total, therefore, of 20,500 men.

^{page 52 note 2} As regards the ‘bucellarii’, cf. Böcking, Notitia Dignitatum pp. 207 f. Mommsen, Hermes, xxiv, pp. 234. f. (with list of contemporary references to the ‘bucellari’); Delbruk, , Geschichte der Kriegskunst ii ², p. 471Google Scholar; Guilhiermoz, . Essai sur l'origine de la noblesse en France au moyen age (Paris, 1902), p. 21.Google Scholar

^{page 53 note 1} Cf. supra p. 8.