^{page 131 note 1} Cod. Th. 6, 10, 2.

^{page 131 note 2} Not. Or. xviii. omnium is Schonhov's correction of omnis; the corresponding text in Not. Occ. xvi has Notitia omnium, but omits the words scholas—tractat.

^{page 131 note 3} Carm. Min. 25, 83 sqq. The father of Celerina is the primicerius referred to, and he held office in the west. Cp. Birt in the index to his ed., sub Palladius.

^{page 132 note 1} Cp. Justinian, Nov. 24 ad fin., 25 ad fin. etc. Karlowa, , Gesch. d. röm. Rechts, i, 991Google Scholar. It is not probable that mandata principis were issued to provincial governors in the fourth century, nor (if they were) is there any evidence that the primicerius had anything to do with them. We only know that this practice had fallen out of use before the age of Justinian and that he revived it (Nov. 17).

^{page 132 note 2} Not. Or. xviii; Not. Occ. xvi. They were called laterculisii (cp. e.g. Justinian, locc. citt.).

^{page 132 note 3} Cod. Th. 1, 8.

^{page 133 note 1} Canon. Cat. Misc. 378.

^{page 134 note 1} Cp. Gelzer, M., Studien zur byz. Verwaltung Aegyptens, pp. 8, 9.Google Scholar

^{page 135 note 1} There was only a different lettering on the codicilli; see Not. Occ. xliv, xlv.

^{page 135 note 2} See Mommsen, , Chron. Min. i, p. 533Google Scholar.

^{page 135 note 3} This writer also records only one Cappadocia, but we know from Cod. Th. 13, 1, 11, that there were already two Cappadocias before A.D. 386. In enumerating the provinces of western Illyricum (Diocese) along with those of eastern Illyricum. (Prefecture) Polemius reproduces the conditions existing before A.D. 379. Then there was only one province of Dacia; in Not. Or. there are two Dacias, but we cannot say whether this change was due to Theodosius or to Eutropius.

^{page 136 note 1} Cp. Mommsen, , Hist. Schr. iv, 558Google Scholar.

^{page 136 note 2} Die Zeit des Vegetius, Hermes, xi (1876), p. 72Google Scholar.

^{page 137 note 1} It is possible that some of the units named Valentinianenses were also formed in the early years of Valentinian III (cp. Not. Occ. vii, 47, 61, 71, 165).

^{page 138 note 1} I do not include the clarissimate of the praeses in Not. Or. xliv, 4, and the perfectissimate of the praeses in Not. Occ. xlv, 4. It is pretty certain that perfectissimi in the latter passage is an error. The perfectissimate had probably been entirely done away with in A.D. 412, as Hirschfeld has shown, Die Rangtitel der römischen Kaiserzeit, in S. B. of the Berlin Academy, 1891, p. 592. Compare Mommsen, , Hist. Schr. i, 558sqq.Google Scholar

^{page 139 note 1} The inconsistencies noted above, § 4, show indeed that it was not meticulously corrected. There is only one case of the inadvertent inclusion of a name which had been corrected in the margin xl. 48; and in the same place the troops of the laterculum minus, 44–49, ought to have been transferred immediately after 36, so as to precede the officium (as in other sections). Some small inconsistencies, pointed out by Mr. Seeck, op. cit. p. 73, have also been allowed to remain in xxxix and xli.

^{page 140 note 1} Cod. Th. 7, 8, 3.

^{page 140 note 2} Ib. 11, 28, 9, in the note as to sending copies of the law to certain officials.

^{page 140 note 3} Ib. 6, 8, 1.

^{page 140 note 4} A.D. 372, ib. 6, 9, 1; A.D. 380, ib. 2.

^{page 140 note 5} Because in Cod. Th. 1 the title 8 on the quaestor precedes title 9 on the mag. off.

^{page 140 note 6} Cod. J. 1, 30 and 31.

^{page 140 note 7} See Mommsen's, ed. of Cod. Th. vol. i, p. clxxxviiGoogle Scholar. Before Feb. 22, 430, Helio was replaced, Cod. Th. 7, 8, 15.

^{page 141 note 1} Op. cit. 71 sqq.

^{page 141 note 2} Marcellinus, Chron. sub. a.

^{page 141 note 3} This activity is referred to by Vegetius in De re Militari. Mr. Seeck has made it probable that those were right who, like Gibbon, saw Valentinian III in the emperor of Vegetius.

^{page 141 note 4} No arrangement had yet been made for the civil government of Valeria; the province does not appear in section ii, nor a praeses in the Index. Probably it was left temporarily under the military rule of the dux.

^{page 142 note 1} The other spectabiles (except the proconsuls and consulares) have, instead, volumes with various inscriptions on the covers, which present great difficulty. Eight different inscriptions can be distinguished. In the most important of these (which appears in the insignia of the castrensis, primicerius notariorum, magistri scrin., vicarii, comites and duces), FL|INTALL|COMORD|PR Böcking (Not. Dig. ii, 528, n. 9) rejects the most obvious explanation of the last words, comes ordinis primi, on insufficient grounds. The true reason for rejecting it—on the assumption that the inscription designates the rank of the particular official—is that it appears in the insignia of the simple duces and they were not comites ordinis primi.

^{page 143 note 1} Valentinianenses, to one of which Valentinianenses iuniores, 11, v. 190, will correspond.

^{page 143 note 2} 203 equites stablesiani. It might be conjectured that in vi <equites stablesiani Britanniciani> fell out after 82 equites st. Italiciani. It is probable enough that 204 equites Taifali may be the same as vi, 59, equit. Honoriani Taifali iuniores, but these may possibly be identical with vii, 172, equit. Hon. iun. who were in Gaul.

^{page 144 note 1} Mountain peaks are outlined in his insignia.

^{page 144 note 2} There is no evidence for a comes Britanniarum in the fourth century. Theodosius, the emperor's father, had the rank of comes when he was sent to Britain in 369, but he did not take the place of any one else, nor is there anything to show that he was succeeded by a comes permanently stationed there. In consequence of his success he received the post of a magister equitum.

^{page 144 note 3} στρατηγός Zosimus, v, 46, 2.

^{page 145 note 1} Ammian, 26, 5, 3; 28, 6, 3.

^{page 145 note 2} Carm. Min. 50.

^{page 145 note 3} Mommsen held that from the time of the supremacy of Arbogastes and throughout the fifth century, the two masterships in praesenti were combined (under the title Mag. utriusque mil.), Hist. Schr. i, 556. This view is certainly wrong. It is refuted by the Notitia, and it is quite clearly untrue of the later part of the reign of Valentinian III when we have the clearest evidence for two magistri, both of whom have the title mag. u. m.

^{page 145 note 4} Renatus Prof. Frig, . in Gregory of Tours, Hist. Fr. ii, 8Google Scholar.

^{page 145 note 5} Chron. Gall. 100, p. 658 (Chron. Min. vol. i).

^{page 145 note 6} Cod. Theod. ii. 23, 1. It is quite arbitrary to alter Crispino to Castino.

^{page 145 note 7} Prosper, Chron. sub a., Aetius magister militum factus est, interpreted, rightly as I think, by Mommsen (ib. 535). In the previous year Aetius had only the title comes. In the Life of St. Hilary of Arles we read of an inlustris Cassius qui tunc (A.D. 429) praecerat militibus (c. 6, in Migne, P.G. 50, col. 1227) and evidently stationed at Arles. Inlustris might suggest that he was a mag. mil. but that seems highly improbable. He is otherwise unknown.

^{page 146 note 1} Taking the strength of the legion as 1,000, of the auxiliary cohort and the vexillatio as 500 each. Of course, some of the units may not have been up to full strength.

^{page 147 note 1} See Historische Schriften, iii, 214, n. 2Google Scholar: ‘die britannischen Abschnitte der Notitia gehören der vordiokletianischen Epoche’; ib. 117, they belong ‘der vorconstantinischen Militärordnung.’ Mommsen's theory has been accepted, e.g. by Mr.Sagot, in La Bretagne romaine (1911), p. 230Google Scholar, and Mr.Grosse, in Römische Militärgeschichte (1920), p. 28Google Scholar. It is rejected by SirOman, C., England before the Norman Conquest (1910), p. 151Google Scholar.

^{page 147 note 2} For the cohort I or II Asturum (xl, 42) and the ala Sabiniana (xl, 37) cf. Mr.Cheesman, , The Auxilia of the Roman Imperial Army, p. 147.Google Scholar

^{page 148 note 1} See Mr.Craster's, study of ‘The Last Days of the Roman Wall’ in Archaeological Journal, lxxi, 25sqq. 1914CrossRefGoogle Scholar.

^{page 148 note 2} ‘Roman Britain in 1914’ (Supp. Papers, iii, of the British Academy, 1915), p. 40. In the same way the archaeological evidence is consistent with the destruction of Calleva about A.D. 420, but does not disprove its survival till a later year.

^{page 148 note 3} Presumably a detachment of legio II Augusta.

^{page 149 note 1} See also his Romanization of Roman Britain (ed. 3, 1915), p. 80Google Scholar: ‘After 407 the Romanized area was cut off from Rome.’

^{page 149 note 2} Bell. Got. 416.

^{page 149 note 3} Cambridge Medieval History, i, p. 379.Google Scholar

^{page 149 note 4} Mr. Haverfield (ib.) explains the assumed ‘departure of Romans’ after 406–7 as meaning, not a great departure of persons, but that the central government ceased to send ‘the usual governors and other high officials and to organize the supply of troops.’ If the Caesar Carausius of the Richborough coin published by Sir Evans, Arthur (Num. Chron. vii, 191, 1887Google Scholar) was, as he suggested, a colleague of Constantine, this would support the probability that Constantine did not propose to let Britain slip from his hand.

^{page 150 note 1} vi, 5, 2–3.

^{page 150 note 2} Ib.. 10, Ὁνωρίου δὲ γράμμασι πρὸς τὰς ἐν Βοεττανίἲ χρησαμένου πόλεις Φυλάττεσθαι παραγγέλλουσι. Gibbon dated the end of Roman government in Britain not to the departure of Constantine but to this rebellion (chap. xxxi). Of a Saxon invasion at this period we have a record in the Chronica Gallica (Chron. Min. ed. Mommsen, i, p. 654): Britanniae Saxonum incursions devastatae in the sixteenth year of Honorius, i.e. A.D. 410. But we cannot depend on the accuracy of the dates in this chronicle to a year. The death of Arcadius, e.g. is entered under xii instead of xiv. That of Honorius comes in the thirty-second instead of the twenty-ninth year of his reign, and some of the other dates are three years wrong. (Possibly there was a confusion between 395, and the year of Honorius's creation as Augustus in 393.) The true date of the invasion may thus be two or three years earlier, 408 or 407, or it may be later.

^{page 151 note 1} C.I.L. vii, 268. See Evans, A. J., Numismatic chronicle, 1887, p. 208Google Scholar and cp. Haverfield, , Ephemeris epigraphica, ix, p. 561Google Scholar. For Justinian, see olympiodorus, fr. 12 Ἰουστῖνον for Ἰουστινανόν the source of Zosimus, vi, 2.

^{page 151 note 2} That he was responsible for fortifying some of the defences of Britain we know from Claudian, de cons. Stil. ii, 250–5.

^{page 151 note 3} Op. cit. p. 42.

^{page 152 note 1} Not. Occ. v. 241, secunda Britannica = vii, 156, secundani Britanniciani.

^{page 152 note 2} See above, p. 148, n. 3.

^{page 153 note 1} Constantius, Vita Germani, c. 17. See Levison, , Bischof G. von Auxerre, in Neues Archiv, xxix, 97sqq. 1903Google Scholar.

^{page 153 note 2} Historia Brittonum (in Chron. Minora, iii), c. 66, p. 209, cp. c. 31, p. 171.

^{page 153 note 3} Chron. Gall. 128, p. 660.

^{page 153 note 4} So I read for latae; the correction seems simpler than Mommsen's la <te uexa> tae.