¹ Odes iii, 5.

² Tacitus, Ann. i, II.

³ Suetonius, Gaius, 46.

⁴ Tacitus, Ann. xii, 31 : cunctaque cis Trisantonam et Sabrinam fluvios cohibere parat (accepting Bradley’s emendation for castris antonam, and assuming Trisantona= Trent). This interpretation of the Fosse was advanced by the writer in the journal of Roman Studies, 14 (1924)Google Scholar, where the whole question is discussed.

⁵ Limes, limitis : from limus=transversus (cf. limen threshold : so Festus) and the root-it ‘going.’ Cf. Fabricius, art. ‘Limes’ in Pauly-Wissowa’s Realencyclopädie.

⁶ Haverfield, , Agricola and the Antonine Wall, Proc. Soc. Ant. Scot., 1917-18, 174–81Google Scholar. Macdonald, Roman Wall in Scotland, 382-8.

⁷ Macdonald, op. cit., 385-6.

⁸ Excavations: Proc. Soc. Ant. Scot. 1901-2, 182–242 Google Scholar, and Jour. Rom. Stud. 9. 113–122.Google Scholar

⁹ Macdonald, The Agricolan Occupation of North Britain, J.R.S., 19, 111–138 Google Scholar. Home, Roman York, 36–37 Google Scholar, seems to express dissent, but makes no attempt to reinterpret the facts in any other way, and bases his case on false assumptions as to the implications of Macdonald’s theory. For Newstead, see Richmond in Proc. Soc. Ant. Scot. 1923-4, 309–321 Google Scholar. While this was in the press I heard of a further and apparently conclusive confirmation in the excavations going on at Mumrills.

¹⁰ C.I.L., vii, 241, where the date is given as 108-9 ; but see Cagnat, Cours d’Epigraphie latine, p. 194.Google Scholar

¹¹ For the date see Ritterling in Pauly-Wissowa, xii, 1606.

¹² Spartian, Hadr. v, 2

¹³ Fronto, 218 N.

¹⁴ Ritterling, op. cit. 1669 Google Scholar, argues that certain recorded careers (L. Aemilius Karus, L. Novius Saturninus) suggest a decidedly later date—after 120, and preferably after 125. I cannot think that this is consistent with the legion’s absence from the British mural inscriptions.

¹⁵ Pelham, Essays on Roman History, 162.

¹⁶ Mommsen, Provinces of the Roman Empire, E.T. i, 157.

¹⁷ C.I.L., iii, 3385 (=Dessau 395) rapam omnem burgis a solo extructis itempraesidisper loca opportuna ad clandestinos latrunculorum transitus oppositis munivit : ibid. 10312-3.

¹⁸ Fabricius, art. Limes, in Pauly-Wissowa, xiii.

¹⁹ The separate existence of this period is not proved. Nether Denton was certainly occupied then, but we do not know that it was then first occupied. Archaeologia, Ixiv, 303.Google Scholar

²⁰ Haltwhistle, Burn, and Throp, are the dated examples. For the establishment of their date, see Cumb. and West. Trans., N.S., Xiii, 379–381.Google Scholar

²¹ For a general description of the works, see Collingwood, Hadrian’s Wall, a history of the problem, J.R.S., xi, 37–66 Google Scholar, or Guide to the Roman Wall, Reid, Newcastle, 6d.

²² The whole view is set forth in detail in the article Hadrian’s Wall, cit. in the preceding note.

²³ I refer to Professor Fabricius, who took endless trouble to understand the English theories during the preparation of his invaluable Limes article for Pauly-Wissowa’s Realencyclopädie.

²⁴ Hull, The Excavations at Aesica, 1925, in Archaeoiogia Aeliana, 1926, 197-202.

²⁵ The broad foundation may be admirably seen in e.g. Cumb. & West. Trans. N.S. xi, fig. 6 facing p. 404.

²⁶ This fact is of importance. Great Chesters (Aesica) lies well away from the Vallum, which therefore it does not deflect. It is therefore open to anyone to hold that this fort, and conceivably one or two others, were built not simultaneously with the Vallum but at a later stage in the development of the frontier, when the broad-foundation Wall was added. In this case, the theory above described as holding the field to 1925 requires only two modifications : (a) the addition of the early broad-foundation Wall before the stone Wall ; (b) the hypothesis that certain forts were added at this stage. Personally, I regard this as the only working hypothesis likely to commend itself to those in possession of the facts as now known.

²⁷ I should like to discuss a piece of circumstantial evidence on this head ; though to avoid breaking the thread I relegate it to a footnote. The ‘centurial stones’ which marked the beginning and end of each section of the Wall built by a century acting as an independent working-party certainly belong to the stone Wall, not to that of the broad foundation. I say this because in character and style they resemble the materials of the stone Wall. Now in Wales, centurial stones are quite common ; they seem to belong to a period going down to, perhaps, about the end of Trajan’s reign, when forts were being built in stone. The fashion of dividing up a fort-rampart into lengths and assigning each length to a century may thus be called a Trajanic fashion ; and the Hadrianic forts of the Wall, which belong to a type evolved in Trajan’s reign, are thus not the only Trajanic feature on the Wall—the centurial stones constitute another. Now the Antonine Wall was, as I point out below, divided not into centurial lengths but into much longer sections, plainly more economical to build. I understand the transference of a centurial division of labour from Trajanic fort-walls to a great stone wall under Hadrian, and I understand how, being there found cumbrous, it should be replaced by a better division of labour under Pius ; but I cannot believe that, after the Antonine Wall, anyone would exhume a Trajanic method of building forts and apply it in sheer wantonness and perversity to the stone Wall. I infer that the stone Wall is Hadrianic.

²⁸ Archaeologia Aeliana, 1925, p. 103 Google Scholar and plate xiv.

²⁹ My paper on The British Frontier in the Age of Severus (J.R.S., xiii), was written before the new evidence at Aesica came to light. But I do not see need to withdraw its conclusions in consequence of the new discoveries.

³⁰ The Purpose of the Roman Wall, in the Vasculum, Oct. 1921.Google Scholar

³¹ The chief work is Macdonald, The Roman Wall in Scotland, which summarizes all knowledge to 1911. Later additions by the same author are published in Proc. Soc. Ant. Scot. 1914-15 and 1924-5, and by Miller, The Roman Fort at Balmuildy, 1922.

³² Macdonald, The Building of the Antonine Wall, in J.R.S. 11, 1–24.CrossRefGoogle Scholar

³³ lxxii, 8.

³⁴ Antonine Wall Report (Glasgow Archaeological Society).Google Scholar

³⁵ Foord, Last Age of Roman Britain.

³⁶ Foord, op. cit. pp. 32, 50.

³⁷ J.R.S. 10, 172.Google Scholar

³⁸ History of England, ch. xxi.

³⁹ See the reviews by Oman in J.R.S., xvi, Macdonald in History, April 1926, and Collingwood in Eng. Hist. Review, April 1926.

⁴⁰ Haverfield in Proc. Soc. Ant. Scot., 1904, 454 seqq.

⁴¹ viii, 43, 4.

⁴² Marcus Aurel. 8, 7.

⁴³ Or at any rate, that not long after 182 it was rebuilt. Miller, Roman York, J.R.S. xv, 185.

⁴⁴ The British Frontier in the Age of Severus, J.R.S., xiii.

⁴⁵ Haverfield, Roman Coast Defences of Britain, J.R.S., ii. Collingwood, The Roman Signal-station on Castle Hill, Scarborough (pamphlet, 2d.)

⁴⁶ Mothersole, The Saxon Shore : Haverfield, Litus Saxonicum in Pauly-Wissowa.

⁴⁷ Wheeler, Segontium (Y Cymmrodor, 33, 97–101).Google Scholar

⁴⁸ Wheeler, Roman and Native in Wales (Trans. Cymmrodorion Soc. 1920-1).

⁴⁹ See e.g., Bury, , The Notitia Dignitatum, J.R.S. Google Scholar, x. ; Collingwood, , The Roman Evacuation of Britain, J.R.S., xiiGoogle Scholar; Foord, op. cit. and reviews by Macdonald (History Jan. 1926), Oman (J.R.S. xiv), and Collingwood (Eng. Hist. Rev. Jan. 1926)