^{page 38 note 1} In a temporary marching-camp Hyginus allows 21,600 square feet to a cohors quingenaria, which is about 1,000 men per acre (De Munit. Castr. §2): in permanent forts the accommodation seems to have been 200–250 men per acre. In the text I assume that Newcastle really does lie astride of the wall and Stanwix south of it, as authorities agree to be probable. That Burgh lies astride of it was proved by excavation in April, 1922.

^{page 39 note 1} Itinerarium Antonini Augusti, ed. Parthey and Pinder, 1848, pp. 222, 223, 227.

^{page 39 note 2} The abandonment of Birrens about 180 gives a terminus ante quem for the compilation of this portion of the Itineraries; and since the Forth-Clyde wall was built about 143, and lasted till 180, the compilation would seem to be placed in the years c. 125—c. 143. This inference was pointed out to me by Dr. Macdonald.

^{page 40 note 1} Notitia Dignitatum, ed. Seeck, 1876; Occid. xxviii, pp. 180–181. A list of civil officials follows.

^{page 41 note 1} Ibid. Occid. xl, pp. 209–212. Some emendations derived from inscriptions have been added in square brackets. A few of these are already adopted in Seeck's text.

^{page 41 note 2} This paper was in type before Prof. Bury's article in J.R.S. x appeared, and I have thought it best to let the above passage stand.

^{page 41 note 3} I suppose στρατηγὸς to mean Praetor, in which case the Governor of Britain is meant.

^{page 42 note 1} The MSS. have post Maurum; it is generally assumed that murum is a safe emendation, the corruption being perhaps due to the presence of an Aethiops in the context. Professor Stuart Jones suggests to me that post Martem apud vallum commissum would be an easy and satisfactory emendation; the writer of vit. Aurelian., 21, 2, borrows aperto Marte from Ov. Met. 13, 27; so Spartian may have borrowed committere Martem from Sil. Ital. 13, 155.

^{page 43 note 1} Professor A. C. Clark kindly tells me that whereas the phrase alio muro cespiticio would most naturally imply a previous turf wall, in a late and inartistic writer it might possibly be intended to convey the other meaning given in the text. He further points out that an easy palaeographical correction would be alto. This suggestion is very attractive: for Capitolinus has mentioned no previous Wall, and it is unlikely that he would here allude to Spartian's Hadrian when Spartian himself, speaking of Severus's Wall, does not refer back to his own mention of Hadrian's; and alto might easily be altered to alio either accidentally, or purposely by any copyist who thought it clever to introduce into Capitolinus's text the allusion to Spartian.

^{page 48 note 1} See for example the unworn brass of Carausius, dating before 289, dropped at Castlesteads (C. & W. Trans. N.S. xxii, pp. 204, 229).

^{page 48 note 2} ‘Sir’ of course indicites the status not of a knight but of a Bachelor of Arts. Ridley was a local clergyman.

^{page 48 note 3} Camden, Britannia: The early editions, e.g. ed. 2 (1587), pp. 532–543, give his views before visiting the remains; ed. 5 (1600), pp. 710–724, gives the fruits of his visit. There is some conflict of evidence as to whether Camden visited the Wall in 1599 or 1600. But (a) there is no doubt whatever that Camden was in Cumberland in 1599, (b) if the journey was in 1600, it was late in 1600 (so Wood, followed by Gibson and Gough), which makes it difficult to see how the results could have appeared in the 1600 Britannia.

^{page 48 note 4} Camden heard of forts at ‘Iverton, Forsten, and Chester in the Wall’: the addition ‘near Busy Gap’ is not in his original text of 1600, and even if it were, it would be too slender a foundation for identifying that site with Housesteads in face of the certain use of the name for Aesica. No commentator, so far as I know, has identified Iverton and Forsten: but Everton is the ruined farm half a mile west of Chesterholm, and Forsten is obviously Fourstones. ‘Iverton and Forsten’ are therefore Chesterholm and the prehistoric camp on Warden Hill: unless there was a Stanegate fort, which has now disappeared, near Newbrough. Mr. F. G. Simpson tells me that a fort is marked there on an early map. Camden's informant knew the line of the Stanegate, not that of the Wall; this is, of course, what we should expect.

^{page 49 note 1} Gibson's Camden, ed. 3 (1753), vol. ii, 1051–1060. Bruce (Handbook to the Roman Wall, p. 2) seems to imply that the writer of this article is not Gibson himself: I do not know on what grounds. Gibson's ed. i is 1695.

^{page 50 note 1} Itincrarium Curiosum, Centuria ii, pp. 55–68.

^{page 51 note 1} ‘There is a vallum and ditch … studiously chusing the southern declivity of rising ground. I observe, too, that the vallum is always to the north.’ This evidently implies that the vallum in question is the north mound, and that the other two escaped Stukeley's attention (op. cit. p. 59).

^{page 51 note 2} Ibid. p. 67.

^{page 51 note 3} Op. cit. ch. viii–x; the sites which Gordon can claim as proved by inscriptions are Carrawburgh = Procolitia, Housesteads = Borcovicium, Chesterholm = Vindolanda, and Birdoswald = Amboglanna (p. 83). Benwell is the easternmost fort which he actually saw.

^{page 52 note 1} Britannia Romana, 1732: pp. 98–158.

^{page 54 note 1} Second edition, 1806, in four volumes: see of this ed. vol. iii, pp. 467 seqq.

^{page 54 note 2} The History of the Roman Wall: ed. i, 1802: ed. 2, 1813.

^{page 55 note 1} Op. cit. Part ii, Parishes: vol. iii, pp. 149–322. This essay was also, however, printed separately.

^{page 56 note 1} Ed. i, 1851: ed. 2, 1853 : ed. 3, 1867 (greatly enlarged).

^{page 56 note 2} Ed. 1, 1863. Ten editions have appeared.

^{page 56 note 3} As a specimen we may mention the Rev. J. Maughan of Bewcastle: Mural Controversy, by A Cumbrian: 1857.

^{page 56 note 4} The Roman Wall (atlas) 1857: Memoir, 1858 (both privately printed).

^{page 57 note 1} German ed. pp. 169–171: Eng. tr. vol. i, p. 186–188.

^{page 58 note 1} A later and almost equally unsuccessful German attempt to review the subject is that of Krüger in Bonner Jahrbücher, cx. It is to be regretted that Kornemann, in a comprehensive paper on Die neueste Limesforschung (Klio, vii, 1907, pp. 73–121), seems to rely entirely on Krüger for his knowledge of the British Limes.

^{page 59 note 1} Archaeol. Aeliana, ser. ii, xxiv, pp. 13–18.

^{page 59 note 2} Ibid. pp. 19–62.

^{page 59 note 3} Reports in C. & W. Trans., O. S. xiii–xvi, N. S. i–iv, (1895–1904). The brilliant results of these explorations owe much of their value to the skilled surveying and draughtsmanship of Mr. and Mrs. T. Hesketh Hodgson.

^{page 59 note 4} Op. cit., pp. 28–29.

^{page 60 note 1} C. & W. Trans. O. S. xiv, pp. 185, 188.

^{page 60 note 2} Letter to T. McK. Hughes, Sept. 15, 1895, in Bates' Letters, p. 34.

^{page 60 note 3} C. & W. Trans. O. S. xv, p. 340 (summary): N. S. iii, p. 339.

^{page 60 note 4} Letter in op. cit. p. 58. to Prof. Hughes.

^{page 60 note 5} Chesters: C. & W. Trans. N. S. i, pp. 84–89; Arch. Ael. ser. ii, xxiii, pp. 9–22. Birdoswald: C. & W. Trans. O.S. xv, 180–182.

^{page 61 note 1} First formulations, C. & W. Trans. O.S. xiv, 190–191, xv, 342–3 (1897–9): accepted in appendix to Eng. tr. of Mommsen's, Provinces (vol. ii, p. 351, 1909Google Scholar) and Encycl. Brit. art. Roman Britain (1910).

^{page 61 note 2} C. & W. Trans. N.S. xi, pp. 390–461 : esp. pp. 459–60.

^{page 62 note 1} C. & W. Trans. N.S. xiii, pp. 297–397.

^{page note 62 note 2} At present no explanation of this structure has been proved by excavation, and whatever may be offered is a mere hypothesis. The turf wall is in fact, as Mr. Simpson says to me, ‘the skeleton at the feast,’ and his projected excavations at Birdoswald, to begin in 1923, will either break up the feast or turn the skeleton into a flesh and blood convive.

^{page 63 note 1} Proc. Soc. Ant. Newcastle, ser. iii, vol. ix, 1920, p. 295Google Scholar.

^{page 63 note 2} Proc. Soc. Ant. Newcastle, Ser. iii, vol. x, 1922, pp. 216–218Google Scholar.

^{page 63 note 3} Simpson and Shaw, The Purpose and date of the Vallum and its Crossings, C. & W. Trans. N.S. xxii. This paper, which is in the press as I write, summarises the whole of the last ten years' advance, and is the next landmark in the study of the Wall after the report in C. & W. Trans. N.S. xiii (diggings of 1911). The ‘crossings,’ as stated by Messrs. Simpson and Shaw, were noticed in certain cases by Horsley, but his observations were incomplete and were not followed up.

^{page 64 note 1} The only other theory not now disproved by excavation has not, I think, been stated in print. It is, that the crossings were intended as a ‘formal obliteration’ of the Vallum frontier-line carried out not when the Antonine Wall was built (that has been publicly suggested and definitely disproved) but when Hadrian's Wall was built. But this theory leaves unanswered the question asked in the text.

^{page 64 note 2} Macdonald, . The Agricolan Occupation of North Britain, J.R.S. ix, 111–138Google Scholar.

^{page 64 note 3} Haverfield, in Hist, of Northumberland, x, p. 478Google Scholar: for evidence, cf. Arch. Ael. ser. iii, vol. viii, pp. 22, 24, 32, etc.

^{page 64 note 4} Bushe-Fox, , Archaeologia, lxiv, p. 303Google Scholar, considers Nether Denton to have been founded a few years later than Agricola's time, but in any case it was held during Trajan's reign.

^{page 64 note 5} C.& W. Trans. N.S. xiii, pp. 379–381.

^{page 65 note 1} Maclauchlan, Memoir etc., p. 40: cf. pp. 28, 53. It runs ESE; Horsley thought he could trace a south-westerly branch from the same fort.

^{page 66 note 1} In The Purpose of the Roman Wall (Vasculum, vol. viii, Oct. 1921) I have given reasons for thinking that the Wall was not, as one is apt at first sight to suppose, a military work intended to give tactical advantage to troops on the defensive, but a police work, intended to facilitate the patrolling of the frontier-line against unauthorised crossing.

^{page 66 note 2} There is some evidence for the employment of large working-parties composed not of legionaries but of pressed men under the command of legionary officers; but this is a very obscure subject.