^{page 24 note 1} JRS xxi, p. 60 and pl. v, 2.

^{page 25 note 1} The epigraphic copy here given is not a facsimile: it reproduces the form and the arrangement of the letters but not the style of the script.

^{page 25 note 2} Ύπατικὸς ἡγεμών is a description already used by Strabo (iii, 4, 20, p. 166) for the legatus consularis of Hispan. Tarracon.

^{page 25 note 3} Eusebius, Hist. Eccl. viii, 6, 8.

^{page 25 note 4} Fastes des prov. asiatiques (1872), p. 25.

^{page 26 note 1} Dessau, ILS, 1374.

^{page 26 note 2} The Rise of the Equites in the Third Century, Princeton diss., 1915.

^{page 27 note 1} Caess., c. 33, 33 f. and c. 37, 5–7.

^{page 27 note 2} ILS, 537, an inscr. of Caerleon. It may be observed that, if the reform was effected by edict and not by administrative action simply, we can hardly regard the title praef. leg. a.v.l. as a pretence or subterfuge. It should rather be interpreted with Keyes (p. 37), as a method of distinguishing the office from that of praefectus (castrorum).

^{page 28 note 1} Der röm Ritterstand, p. 453.

^{page 28 note 2} He translates: ‘Had the edict of G. become a dead letter, the military service could have been reformed.’ One may confidently say that ‘if the edict had become obsolete’ could not be expressed by amisso edicto. Moreover, amittere and refici are plainly correlates, both referring to militia; the variant amisso arose by assimilation to edicto.

A. Stein has evidently failed to grasp Baynes' argument as a whole, when he calls it a Rückschritt, though it is in essential agreement with his own (except as regards Tacitus). On the other hand, E. Stein judges the ‘Vorschlag zu Vict., Caes. 37, 6,’ to be ‘sehr beachtenswert’ (Gesch. spätröm. Reiches, i, p. 68, note).

^{page 28 note 3} CIL ii, 4102 (ILS 599) and 4103.

^{page 28 note 4} CIL iii, 3418 (ILS 3654).

^{page 28 note 5} Some examples, collected from CIL iii, may be given here: 202, leg. Aug. pr. pr. praeses prov. Syriae Phoenices (milestone, A.D. 213); similarly 205 (milestone of A.D. 198); 12213, praes. prov. Clodio Saturnino leg. Aug. pr. pr. (milestone, Gal.—Capp.); 13724, leg. ac [p]r(aeses) provinciae (dedication by the governor of Lower Moesia in the reign of Severus Alex.); 90, v.c. leg. Augg. pr. pr. praes. provinc. Arabiae (set up by a praef. alae in the reign of Valerianus and Gallienus).

^{page 29 note 1} On praeses, see Mommsen, Staatsr., ii³, p. 240, n. 2.

^{page 29 note 2} CIL iii, 3525 (10492), ILS 2457; IGRR iii. 1286–7.

^{page 29 note 3} Röm. Militärgeschichte (1920), p. II.

^{page 29 note 4} CIL viii, 2529, ILS 2291.

^{page 29 note 5} CIL viii, 2572.

^{page 29 note 6} CIL viii, 9041, 9324 (ILS 627–8).

^{page 30 note 1} Seeck, Untergang, ii³, p. 508, and Pauly-Wissowa, v, 1870; Grosse, op. cit., pp. 153, 171, 174.

^{page 30 note 2} The existence of duces in A.D. 289 is attested by the panegyrist of Maximian (Panegyr., ii, 3). The earliest epigraphic mention of them is perhaps in A.D. 293 (dux limitis prov. Scythiae, CIL iii, 764, compared with 14450; Costa in Ruggiero, Dizion. Epigr., s.v. Diocletianus, p. 1833Google Scholar); the next in 303 (dux Pannon. secundae Saviae, CIL iii, 10981).

^{page 30 note 3} Ges. Schr. v, 561 ff.

^{page 30 note 4} CIL viii, 8924, 20215 (ILS 6866). For the corrupt (Mauretania) Tabiainsidiana of the Verona list Costa suggests Tubusuctitana (CIL viii, 8836),— a temporary designation of M. Sitifensis.

^{page 30 note 5} CIL viii, 608, Ruinart, Acta Sincera, p. 309 ff.; Cagnat, Mélanges Louis Havel, 1909, p. 67 ff. The argument depends on the genuineness of the statement in the Acta concerning the martyrdom of St. Maximilian that the saint was tried at Theveste by the proconsul of Africa. Costa, op. cit., p. 1835, distrusting the Acta, would place the division towards 305–6, on the ground that the name Valeria (Byzacena)—attested in the period 293–305 Ann. épigr. 1908, no. 197—does not appear in the Verona list and soon fell into disuse.

^{page 30 note 6} Cp. Sir G. Macdonald in XIX Bericht d. Röm.-germ. Komm., p. 45 ff., and Brit. Acad. Supp. Papers, vi, p. 64 ff.

^{page 30 note 7} Costa, p. 1833 f. The second Numidian province appears in inscriptions as N.M., which guarantees the first letter of the strange name Miliciana given by the Verona list and puts out of court several conjectures, discussed by Bury, JRS xiii, p. 144 f.

^{page 31 note 1} ClL viii, 7067.

^{page 31 note 2} Mommsen's conclusion does not seem to me to be shaken by the arguments of Bury, op. cit., who thought that ‘the Verona List reflected a division of the Empire which as a whole existed only for a few years from A.D. 308 at earliest to A.D. 315’ (p. 146). This post quem limit is based on an interpretation of Aur. Victor, Caess. 40, 9–10, which is linguistically possible but not necessary and is historically improbable. Stein, Cp. E., Rhein. Mus. 74 (1925), p. 367Google Scholar, note, and Gesch., i, p. 102, n. 3.

^{page 31 note 3} CIL iii, 14191.²

^{page 31 note 4} CIG 3607=IGRR iv, 214.

^{page 31 note 5} ClL iii, 449, set up in the period of the tetrarchy (293–305).

^{page 31 note 6} Mommsen's view that it was Diocletian who divided Phrygia has been disputed on the basis of the evidence furnished by the subscriptions to the Council of Nicaea (A.D. 325). The objections have been answered by Ohnesorge, Provinzliste von 297, p. 21 f., and Ramsay, Cities and Bish., p. 80 ff. Their discussions show that the evidence of the Council lists must be used with caution.

^{page 31 note 7} ClL iii, 13661, compared with 6807.

^{page 32 note 1} ibid. 450; cp. Costa, op. cit., p. 1843.

^{page 32 note 2} Ramsay, Cities and Bish., p. 529, no. 373, =IGRR iv, 731, ILS 8881.

^{page 32 note 3} JRS xvii, p. 32.

^{page 32 note 4} ibid., p. 156.