^{page 1 note 1} Studies of the reliefs must be based upon Die Reliefs de Traianssäule, herausgegeben und historisch erklärt von Conrad Cichorius, Berlin, 1896–1900Google Scholar, in two folio wrappers of plates and three quarto volumes of text. There are three systems of numbering, by plates (Roman), by scenes (Roman) and by casts (Arabic), the second being mostly used, as here. The second and third volumes of the text, describing the scenes, though consulted, have been deliberately eschewed as a basis for descriptions here, in order to form a completely independent judgment of what the scenes portrayed. Reference must also be made to the plates and text of Lehmann-Hartleben's, Die Traianssäule, ein römischer Kunstwerk zu Beginn der Spätantike, vols. iGoogle Scholar (text) and ii (plates), Berlin and Leipzig, 1926, henceforward abbreviated as L-H. The plates are collotypes of large quarto size, but their small scale makes them inadequate for the finer detail. The text is indispensable. The scenes are quoted with Cichorius's second numbering, but the plates have also an independent Arabic enumeration.

^{page 1 note 2} Complete sets of casts exist in the Lateran Museum, Rome, the Casts Museum, Berlin, and the South Kensington Museum, London, being moulded from the matrices in this order.

^{page 1 note 3} No better illustration of a final victory, achieved, like that of the Mons Graupius, citra Romanum sanguinem (Agricola, 35), exists than scene xxiv.

^{page 1 note 4} L-H. i, p. vii: cf. Petersen, Trajans Dakische Kriege, i–ii; Von Domaszewski, Philologus, 1906, 321; Stuart-Jones, , PBSR, v, 442Google Scholar; Davies, , JRS, vii, 74–97Google Scholar.

^{page 1 note 5} Cf. Strong, Roman Sculpture, 202–3, 207–10.

^{page 1 note 6} L-H. i, pp. 5–6; the inscription shews this, and excavations have made the matter plain; cf. Boni, , British Academy Proc. iii, 93Google Scholar. One might compare the conversion of Victor Emmanuel's Monument into the tomb of the Unknown Soldier, in modern Rome.

^{page 2 note 1} L-H. i, 4: the coins shewing the Eagle are Cohen, , Médailles Imperiales, ii ², Trajan, 357–9Google Scholar.

^{page 2 note 2} Dio Cass. lxviii, 16; cf. Boni, loc. cit.

^{page 2 note 3} Hist. Rom. ii, 113–15.

^{page 3 note 1} L-H. i, 147.

^{page 3 note 2} Ibid., and Abb. 23, 24: the fact is obvious throughout the reliefs.

^{page 3 note 3} L-H. i, 1.

^{page 3 note 4} For Roman buildings, see amphitheatre, xxxiii; arches, xxxiii, lxxix, xcviii; block-houses, i; entrenchments, see note 15, p. 12; forts, see section iv, below; lighthouse, lxxxiii; porticoes, lxxx, lxxix and lxxxi with temples; temples, lxxix; theatres, iii and lxxxvi (stone), c (wood); town, iii; watch-towers, i. For costume and accoutrement, see cavalry, lxiv; slingers, lxvi, lxx, lxxii, cxiii; club-men, xxiv, xxxvi, xxxviii, lxx; archers, lxvi, lxx, cviii, cxv; physiognomy, archers, cxv, club-man, xxiv, cavalry, lxiv.

^{page 3 note 5} For Dacian buildings, see fortress, xxv, xciii; houses (stone), xxv (on legs), xxix, xlv; huts (wood), xxv, lix, lxxvi, cxix, cli, xxv and lvii on legs; pound, xxv; Sarmizegethusa, murus gallicus, cxiii–cxv; temples or tombs, lxii. For costume and accoutrement, ambassadors, c; cataphracts, xxxi, xxxvii; sword, lxxii, xciv, xcvi, cxv (on ground); Decebalus, cxlv, xxiv (cf. L-H., 68). In the case of the cataphracts (xxxvii) it is worth noting the intimate detail, extending to the eye-protector for the horse. The writer has noted many similar eye-protectors in the R-G. Central-Museum at Mainz and some in Bonn Museum. A fragment of one is in Cherters Museum.

^{page 3 note 6} It is, in fact, much more likely that the cartoon was derived from the scrolls.

^{page 4 note 1} L-H. i, 39 observes that only two out of all the scenes in the Column of Marcus illustrate building: in the Column of Trajan there are more than one hundred times as many.

^{page 5 note 1} See JRS xvi, 264, where the writer first ventured this conclusion. The identification of the log-walk had already been made by Jacobi, (Saalburg Jahrbuch, iv (1921), 70Google Scholar, n. 2) and Cichorius (ii, 61, Abb. 8), without perceiving that this implied an earthwork below, since duck-board walks have no place upon stone ramparts.

^{page 5 note 2} Cf. xlix, where the duck-board walk becomes a pattern. For other misconceptions, see pp. 21, 25, 27, 28–9, 32, 34.

^{page 5 note 3} See below, p. 14.

^{page 5 note 4} cxiv: see pp. 39–40 for a discussion of their purpose.

^{page 6 note 1} The fallen superstructure of this gate can still be seen in the grounds of Castle Park at Colchester.

^{page 6 note 2} This gate was discovered buried in the mound of the mediaeval castle, like the Roman walls of Cardiff and York. It collapsed soon after discovery, but not before a careful drawing, now in Lincoln City Museum, had been made by Samuel Tuke.

^{page 7 note 1} Suovetaurilia appear twice in the first war and once in the second (viii, liii, ciii), marking the opening of campaigns rather than the beginning of special tasks (cf. JRS xi, p. 13). The representation is to be compared with that on the famous terminal slab (cf. JRS xi, pl. i) of the Antonine Wall at Carriden. With the allocutiones, these scenes form the most important guides to the action of the reliefs. The procession goes all round the camp, headed by garlanded tubicines and cornicines, followed by the three decorated animals, with attendant popae, and by the troops in undress uniform. The general awaits it at the praetorium, sacrificing upon a portable altar or a turf altar, and attended by tibicines, camilli and the sprinkler of mola salsa (ciii).

^{page 7 note 2} See p. 28 below.

^{page 7 note 3} The bridge-head, built in stone, still exists; the mosaic is figured in Notizie degli Scavi, 1914, 286, see Constans, Arles Antique, 343: cf. Ausonius, Urbes, 77, ut mediam facias navali ponte plateam.

^{page 7 note 4} De mun. castr., 24. I owe the comparison with navvies to Mr. C. E. Stevens.

^{page 7 note 5} As pointed out by Stuart-Jones, Companion to Roman History, pl. xxvii, 210–212.

^{page 8 note 1} viii, xxiv; cf. Tac., A. i, 18Google Scholar.

^{page 8 note 2} Cf. Stuart-Jones, Companion to Roman History, pls. xxxvii, xlviii.

^{page 8 note 3} Legionary standards: iv, viii, x, xxii, xxiv, xxvii, xxxiii, xlvi, xlviii, lxi, lxxvii, lxxxv, cvi, cxxviii; cf. standard on tombstone of L. Duccius Rufinus at York. Praetorian standards, v, viii, xl, xlii, l, liii, liv, lxiii, lxxv, lxxix, lxxxvii, lxxxvii, xcviii, cii, civ, cxiii, cxxiii, cxxxvii. There are also standards on xxiv, l, and lxxv which are of a peculiar type, perhaps connected with cavalry.

^{page 8 note 4} Germania Romana, iii², pl. vi, I: the Moscow flag is published by Rostovtzev, , Monuments of the Egyptian section in the Alexander III Museum, Moscow, iii, pl. xxivGoogle Scholar, from the Golenicheff collection. It is a scarlet flag with a Victory, on globe, holding a wreath and palm branch, all embroidered in gold. I owe this interesting reference to Miss Taylor., M. V. In England, Proc. Soc. Ant. Newcastle, ser. 3, iv, 18Google Scholar, figures the head of a staff or standard which is first and best described in op. cit. presentation of a centaur slaying a wolf.’ Total height, 8¼″. This object, as Mr. James Mclntyre points out to me, is undoubtedly akin to the standard from Sidmouth, figured in Illustrated London News, 14 Aug. 1926, which is a socket from the top of a pole or staff, provided with a strong hook at right-angles to the figure above, and meant to hold something, perhaps a flag, hanging parallel with it. On its long base stands a centaur ridden by a winged figure and attacked from the front by a dog or wolf. As Major Gordon Home, who re-discovered the Sidmouth object, notes, the centaur was the emblem of the Severan legiones Parthicae. The first of these objects is apparently an imago, the second a vexillum.

^{page 8 note 5} Germania Romana, iii², pl. iv, 2.

^{page 8 note 6} Cichorius, ii, 35. The tombstone of Favonius Facilis, now in Colchester Museum, shews a centurion wearing greaves; a double kilt with metal-trimmed lower lappets; a breast-plate; an elaborate belt, with which are connected a sword-strap and dagger-girdle; a dagger; a sword; shoulder-plates; sleeves to the elbow, and a cloak. He carries a vinestick. Caelius and Q. Sertorius Flaccus wear phalerae on top of their cuirasses, the former having a plain cuirass, the later one of scale armour; see Lindenschmit, Tracht und Bewaffnung des röm. Heeres während der Kaiserzeit, pl. i.

^{page 8 note 7} Scene x. Cf. Vegetius, ii, 16, ursinis pellibus: this was the disguise adopted by Germanicus, Tac., A. ii, 13Google Scholar, presumably because a signifer might move about the camp unquestioned.

^{page 8 note 8} The standard of clothing is invariable on the Column, and is to be compared with that on the tombstones cited below. Compare also the pile of clothes carried by the legionary fording a river, xxvi. For the helmets, see Schaeffer's valuable work, Le casque romain de Drusenheim.

^{page 8 note 9} Perhaps because it is summer. The evidence of the ombstones is for breeches or even trousers sometimes; cf. Germania Romana, iii², pls. xxviii, 3 and iii, 1 with breeches, iii, 2, 3, 4 without them; or pls. iv, 1 and 3 and v, 1 with rousers, v, 3 without them. Parker, The Roman Legions, 250, is perhaps too categorical.

^{page 8 note 10} See, however, scene xv, and notes below.

^{page 8 note 11} Jacobi, , Saalburg Jahrbuch, iv (1921), 90Google Scholar, n. 3, thinks the arrangement abnormal, as against Rice-Holmes. Neither commentator visualises the special occasion of a parade, nor the desire to shew all possible on the part of the artist.

^{page 9 note 1} Strat, iv, 50, 7.

^{page 9 note 2} xxv; see below, p. 36.

^{page 9 note 3} For these paterae in general, see Haverfield, , Arch. Journ. xlix, 228–231Google Scholar, and Bosanquet, , PSAS lxii, 246–254Google Scholar: they are especially plentiful in Scotland, where they were valued as booty, cf. Curle, , PSAS, lxvi, 298 sqqGoogle Scholar.

^{page 9 note 4} Newstead, 274, pl. liii, 3, 4, 5, 6, 8: for names, see Ibid. 274, fig. 37.

^{page 9 note 5} EE. 1324 a, b.

^{page 9 note 6} Spartian, Vit. Hadriani, 10.

^{page 9 note 7} See Curle, , PSAS, lxvi, pp. 310–314Google Scholar, figs. 18, 21.

^{page 9 note 8} B.J. iii, 5, 95.

^{page 9 note 9} Shewn to the writer by Miss A. Robertson, at the Hunterian Museum: she will shortly publish it.

^{page 9 note 10} Scene ex; cf. Newstead, 283–4, pl. lxi, 2, 5, where it is noted that they are an exact reproduction of this type. For the practice of foraging, cf. Caesar, , B.G. v, 17Google Scholar: the bedding occurs in Varro, de L.L. v, 166, s.v. lectica, and in Vegetius, ii, 10, as stramina.

^{page 9 note 11} See Newstead, pl. lvii, 1–5, and p. 279, fig. 3 for the sheaths.

^{page 9 note 12} An example is figured in Newstead, pl. lxviii, 6: as Mr. James McIntyre points out to me, its curved handle and double edge combine the advantages of the modern types of pruning-saw, which has either a curved handle and back, with one edge (the so-called ‘Grecian’ type), or a straight back and handle with two edges. The teeth in each edge are of different sizes, as now.

^{page 10 note 1} It would, however, be unsafe to base an historical deduction upon this point.

^{page 10 note 2} Vegetius, ii, 22: iii, 5 (description of instruments). The trumpeters in a body were aeneatores, CIL, xiii, 6503.

^{page 10 note 3} loc. cit.; tubicen ad bellum vocat milites et rursum receptui canit … sive ergo ad vigilias vel agrarias faciendas, sive ad opus aliquod… tubicine vocante operantur.

^{page 10 note 4} Cf. Bonn Museum, no. 685 (15319); Museum Vindo-bonense, Vienna, tombstone of M. Praeconius Iucundus, tubicen leg. xv. Ap(ollinaris).

^{page 11 note 1} Veget. iii, 5, quod ex uris agrestibus argento nexum.

^{page 11 note 2} id. ii, 22, classicum item appellatur quod buccinatores per cornu dicunt: hoc insigne videtur imperii.

^{page 11 note 3} loc. cit. quotiens autem pugnatur et tubicines et cornicines pariter canunt.

^{page 11 note 4} They appear, however, in the allocutio of civ, immediately after the suovetaurilia opening the second war.

^{page 11 note 5} See, however, Newstead, pl. xxvi; B.M. Guide, Antiquities of Roman Britain, pl. iv, the Ely helmet.

^{page 11 note 6} Cf. cvi–vii, cxxix, lxi, lxvi. Scene lxi shews oxen and mules as draught-animals. In lxvi such a cart bears a ballista towards new gun-emplacements. These carts are not to be confused with the Dacian four-wheelers, with yoke-poles and harness, carrying miscellaneous objects, including a dragon-standard, in xxxviii.

^{page 11 note 7} This may be deduced from the space allowed to iumenta, in front of each tent, which is nine feet by twelve. Cf. Columella, , de R.R. i, 6, 6Google Scholar, giving a minimum of nine feet wide to oxen.

^{page 12 note 1} Vegetius, ii, 10.

^{page 12 note 2} Vegetius, ii, 11.

^{page 12 note 3} Vegetius, ii, 25; nam per singulas centurias singulas carroballistas habere consuevit, e quibus muli ad trahendum et singula contubernia ad armandum vel dirigendum…deputantur.

^{page 12 note 4} De mun. castr. 21.

^{page 12 note 5} Vegetius, iii, 8, tamen pulchriora creduntur quibus ultra latitudines spatium tertia pars longitudinis additur.

^{page 12 note 6} Royal Comm. Anc. Mon., Dumfriesshire, pp. 27–8, no. 45, fig. 28.

^{page 12 note 7} Arch. Journ. lxxxix, 21, fig. 3, 76, pl. xx.

^{page 12 note 8} MacLauchlan, Map of the Watling Street, sheet vi.

^{page 12 note 9} Roy, Military Antiquities of the Romans in N. Britain, pl. ix.

^{page 12 note 10} Ibid. pl. xiii.

^{page 12 note 11} PSAS, l, p. 352, fig. 10; cf. Bericht xix R-G. Komm. abb. 17.

^{page 12 note 12} Ibid. 344, fig. 6; cf. Bericht cit. abb. 18 = Brit. Acad. Suppl. Pap. vi, fig. 8.

^{page 12 note 13} Vegetius, iii, 8, pro necessitate loci … castra constitues; nec utilitati praeiudicat forma.

^{page 12 note 14} Vegetius, iii, 8.

^{page 12 note 15} De mun. castr. 48; Jacobi, , Saalburg Jahrbuch, iv (1921), 68–72Google Scholar, distinguishes, (1) the work in unius noctis transitum; and (2) the castra stativa, ‘supra quem ad similitudinem muri et pinnae et propugnacula componuntur;’ cf. Caesar, B.G. v, 40, vii, 72, viii, 9, and the same type, but composed of corpses for sods and arms for a breastwork, in the emergency of Bell. Hisp. 32. To these may be added (3) the semi-permanent type so often figured on the Column, as Jacobi notes (op. cit. 70, n. 2), without observing that the first and second types also appear. The complicated types are considered below, p. 21, n. 3. Examples of this kind appearing on the Column are xciv, ciii, possibly cvii, cxiv, cxxxi, cxxxiv (with emergency towers).

^{page 12 note 16} The traditional stake is described by Livy, xxxiii, 5, 9; Romanus leves et bifurcos plerosque et trium aut cum plurimum quattuor ramorum vallos caedit. The Column exhibits only regularly-cut stakes for stockades, which are not associated upon it with ramparts.

^{page 12 note 17} Cf. xxxiii; this type of bundle is presumably the normal papilio, called ‘butterflies’ by the soldiers either because they folded out like one, or because they came out of a long roll like a caterpillar; cf. Isid., Orig. xv, 10Google Scholar. The bundle is described by Cichorius (iii, 184) as lange mit Stricken umschnürte Ballen and mit Riemen verschnürte Ballen, mistaking the edges of the leather panels for binding-ropes. For closer views, see lxi–ii, and lxxv. Tent-pitching was one of the three important drills; cf. Appian, Iberica, 88, , .

^{page 12 note 18} See a preliminary note in Cumb. and Westmorland. A. & A. Soc. Trans. xxxiv, 62–90.

^{page 12 note 19} Quintil. viii, 2, 8; cf. Tac., A. ii, 13Google Scholar: the tent appears in scenes viii, xii, xiii, xvii, xxi, liii, lxi, lxii, xcviii, cxxv, cxli, cxlvii.

^{page 12 note 20} There is no evidence that the term was used distinctively for these tents, but they were at least neither praetoria nor papiliones. See viii, xii, xiii, xvii, xxi, xxviii, xliii, cv, cxiii, cxxv, cxxviii, cxlvii.

^{page 13 note 1} For the name, see p. 12, n. 17. They occur in xvi, lvi, lxi, lxii, lxvi, cii, ciii, cvii, cix, cx, cxiii.

^{page 13 note 2} Script. Hist. Aug. Pescennius Niger, 10: cf. Schulten, Masada, pls. iv, vi, viii; for prohibition, see Caesar, B.G. vi, 29, monet ut ignis fieri in castris prohibeat; cf. S.H.A. above.

^{page 13 note 3} See Newstead, pl. liii, and PSAS, lxvi, 310, fig. 18.

^{page 13 note 4} For similar stagings on a river-bank, cxxxi. A watering-party with skins appears on cxxiv.

^{page 13 note 5} Rangordnung des röm. Heeres, 45.

^{page 13 note 6} Auxilia of the R.I.A., 43–44. Tac., A. i, 64Google Scholar implies that fomenta sauciis were a normal provision.

^{page 13 note 7} Vegetius, ii, 10.

^{page 13 note 8} Novaesium, cf. Legacy of Rome, 295, fig. 15: Haltern, Germania, 1928, 74: Vetera, Bonn. Jahrb. 1932, 277. An interesting summary of evidence, inspired by Haverfield, occurs in C. Allbutt, Greek Medicine in Rome, pp. 467–9.

^{page 14 note 1} Block, ix, Arch. Ael. 2, xxv, p. 239Google Scholar, pl. xix; see Companion to Roman History, p. 255.

^{page 14 note 2} Masada, 128.

^{page 14 note 3} This interesting work is rarely studied. It contains an important section of directions to army-surgeons for extracting missiles, and is remarkably up-to-date and practical. See remarks by J. S. Milne, Surgical Instruments in Greek and Roma Times, 6–7; also C. Allbutt, Greek Medicine in Rome, p. 414.

^{page 14 note 4} Vegetius, iii, 24.

^{page 14 note 5} Idem, ii, 25.

^{page 14 note 6} Idem, loc. cit., see note 5, p. 39.

^{page 14 note 7} Schramm, Antiken Geschütze der Saalburg, pls. 9 and 11.

^{page 14 note 8} Schramm, op. cit., 36, fig. 10: also Companion to Roman History, pl. xxxviii, a better reproduction, from the original stone.

^{page 15 note 1} Scene lxiv.

^{page 15 note 2} Cagnat, L'armée romaine d'Afrique, 332. Cheesman, Auxilia, 128–9; Stuart-Jones, Companion to Roman History, pp. 208–9.

^{page 16 note 1} Cf. Cheesman, op. cit. p. 65; for the parade masks, see Newstead, pls. xxix, xxx, and British Museum Guide to Roman Britain, pl. v, for the Ribchester helmet. The very great beauty of these articles has been quite insufficiently appreciated.

^{page 16 note 2} Agricola, 35.

^{page 16 note 3} Vegetius, ii, 2.

^{page 16 note 4} Cichorius, , Paully-Wissowa, iv, 329Google Scholarsqq. records, s.v. cohors; I sagittariorum, I sagittariorum miliaria, I s. m. equitata, I Ulpia s. eq., III sagittariorum; I Apamenorum, I Flavia Chalcidenorum, III Cyrenaica, I Flavia Damascenorum miliaria, I Hamiorum, I Hemesenorum miliaria, I Aug. Ituraeorum, I Ituraeorum, I miliaria nova Surorum, I Thracum and I Tyriorum. Apart from the equitate contingents, alae are also known, three from inscriptions and one by inference. For the bow, cf. Lindenschmit, Tracht und Bewaffnung des röm. Heeres während der Kaiserzeit, pl. v, 3, shewing Monimus Ierombali f., of coh. I. Ituraeorum.

^{page 16 note 5} Scene lxvi.

^{page 16 note 6} Scene cxv, shewing best of all the Eastern facial types.

^{page 16 note 7} Scene cviii.

^{page 16 note 8} CIL vii, 748, 758, 774, from Carvoran (Magna) on Hadrian's Wall: Ibid. 1110 and EE. ix, 1242, from Bar Hill, on the Antonine Vallum.

^{page 16 note 9} Roman forts on the Bar Hill, 122, fig. 44.

^{page 16 note 10} Germania, xvii, 110–114, Beinplatten zur Bogenversteifung aus römischen Waffenplätzen. The matter is further discussed by Sir Geo. Macdonald, Roman Wall in Scotland (ed. 2), 283 sqq.; cf. also Arch. Camb. lxxxvii, 1932, 95–6Google Scholar for a number of these objects from Caerleon.

^{page 17 note 1} It must be said, however, that all the records of special bodies are of earlier date. E.g. Caes., B.G. ii, 7, 10, 19, 24Google Scholar; viii, 40; Bell. Afr. 78; cohorts appear in B.C. iii, 4, 44, 88, 93, 94, and Appian, , B.C. ii, 49, 71Google Scholar. Livy describes the Balearic leather sling, xxxviii, 29. Tacitus mentions them twice, A, iv, 20, xiii, 39, the latter in A.D. 58.

^{page 17 note 2} For the Continental examples, see Zangemeister, , EE. viGoogle Scholar, glandes plumbeae. Those from Birrenswark appear in PSAS xxxiii, 213–16, and one from Birdoswald appears in C. & W. Trans. O.S. xv, 200; cf. Agricola, ed. Furneaux-Anderson, liii. The idea that in this country the glandes were of early date was started by Haverfield, in The Antiquary, March, 1899.

^{page 17 note 3} Cagnat, Les deux camps de la légion troisième auguste à Lambèse, 43.

^{page 17 note 4} Brachycephalic, wavy hair, nose broad and snub.

^{page 17 note 5} Cf. CIL vii, 838, of cohors I Aelia Dacorum at Birdoswald, which is decorated with this kind of sword and a palm-branch. Bruce, Handbook to the Roman Wall, ed. 9, 174.

^{page 17 note 6} De mun. castr., 19, 29, 43.

^{page 18 note 1} L'Année épigraphique, 1926, p. 21, no. 88, mentioning C. Sulpicius Ursulus, praefectus symmachiariorum Asturum belli Dacici. The inference is that this class of soldier was also drawn from other communities than the Astures. I owe this reference to Mr. E. B. Birley. For the use of clubs, cf. the Aestii of Tac. Germania, 45.

^{page 18 note 2} Scenes xxiv, xxxviii, xl, lxvi, lxxii, cviii.

^{page 18 note 3} Vegetius, iii, 8; nam singulae centuriae … accipiunt pedaturas, et scutis ac sarcinis suis in orbem circa propria signa dispositis, cincti gladio fossam aperiunt.

^{page 18 note 4} For stacked arms, see scenes, xii, xx, lvi, lxviii, lxxiii, cxxvi, cxxviii.

^{page 18 note 5} Vegetius, ii, 25; Josephus, , B.J. ii, 5, 95Google Scholar.

^{page 19 note 1} Vegetius, iii, 8 cespes autem circumciditur ferramentis, qui herbarum radicibus continet terram: fit altus semissem, latus pedem, longus pedem semis. Quod si terra solutior fuerit, ut ad similitudinem lateris cespes non possit abscindi, tunc … fossa percutitur.… For illustrations of turf-work, see JRS, xxi, pl. iii.

^{page 19 note 2} loc. cit.: cf. Tac., A. i, 64Google Scholar, amissa magna ex parte per quae egeritur humus aut exciditur caespes: see Newstead, pl. lxi, 3, for a turf-cutter.

^{page 19 note 3} Caesar, , B.G. v, 42Google Scholar; sed nulla ferramentorum copia quae esset ad hunc usum idonea, gladiis caespites circumcidere, manibus sagulisque terram exhaurire videbantur.

^{page 19 note 4} See Cichorius, ii, 61, Abb. 8; Jacobi, , Saalburg Jahrbuch, iv, 70 n. 2Google Scholar.

^{page 19 note 5} Stone forts with doors and towers appear in scenes xxxii, xlvii, and xxxiii: stone forts in xcviii and c.

^{page 19 note 6} Scene xxxv.

^{page 20 note 1} O.R.L. xlvi, Kastell Munningen, 68a, pl. ii, 1: pp. 3–4.

^{page 20 note 2} PBSR x, pls. vi and viii: the criticism by von Gerkan, of this view in Germania, xv, 303Google Scholar, is based upon insufficient acquaintance with the problem: there is no doubt about the way in which the facts present themselves on the spot.

^{page 20 note 3} City Wall of Imperial Rome, p. 58, figs. 7 and 9, pl. iii, a, b, c.

^{page 20 note 4} The sector to which reference is made is unpublished; it is to be found just north of the mediaeval castello, where the wall presents a polygonal base capped by Sullan rubble facing, crowned in turn by late-Roman merlons and two sets of post-Roman embrasures.

^{page 20 note 5} Visible to the east of the south gate of the modern town. The merlons are 1·12 m. long, with an interval of 2·39.

^{page 20 note 6} Each merlon preserved in the curtain south of the Golden Gate is T-shaped in plan, with the down-stroke coming back on to the rampart-walk. In the opposite angles of adjacent T's, some machinery has been dismantled and torn out. The holes now left indicate that the vanished object was massive: cf. Procop., B.G. i, 21Google Scholar, and City Wall of Imperial Rome, p. 29.

^{page 20 note 7} See this volume, p. 75.

^{page 20 note 8} c. 48, munitio aestivalium observatur generibus quinque, fossa, vallo. …It is the failure to realise that these are described as alternatives which has led editors to misunderstand the corruption in c. 50, about the lorica and the vallum; though it should be noted that the vallum is expressly described as made of material not derived from a ditch. The text reads et lorica perbesis (pbes MSS.) similiter ante portas ut titulus ad fossas ad vallum causae instructionis sanctum est cognominatum. If we admit in the first place the accepted emendation et lorica parva fit, all that is required is to replace the second ad with a full-stop. For the tumultuaria fossa, see Vegetius, iii, 8.

^{page 20 note 9} Nicolson, and Burn, , History of Westmorland, i, 610Google Scholar.

^{page 20 note 10} MacLauchlan, Memoir, 32.

^{page 20 note 11} Rauthmell, History of Overborough (ed. 2, 1824), p. 134: Watkin, Roman Lancashire,, p. 82.

^{page 20 note 12} As yet unpublished. One may add the Kirk Beck, at Bewcastle, C. & W. Trans. (N.S.), xxiv, 115: cf. Tac., A. i, 56Google Scholar for bridging on campaign, and A. i, 20 for this work as a permanent routine.

^{page 21 note 1} Vegetius, ii, 25 Item ad fossarum opera facienda bidentes, ligones, palas, rutra; alveos, cophinos quibus portetur terra. Habet quoque dolabras, secures, ascias, serras, quibus materia ac pali dedolantur atque serrantur. Cf. Jacobi, op. cit. 84–92; also Newstead, lxi, 9, an entrenching tool.

^{page 21 note 2} Bruce, The Roman Wall, ed. 3, 80; it might reasonably be assumed that this rope was that mentioned as ἱμἀντα in Josephus, , B.J. iii, 5, 95Google Scholar; cf. Jacobi, op. cit. 88. Cichorius, ii, 272, fails to note that two blocks, not one, are thus bound over the man's shoulder.

^{page 21 note 3} Jacobi, loc. cit. The type occurs very frequently on the Column, where it is to be carefully dissociated from the examples of camps with a rampart-walk behind the merlons; see scenes, viii, xxvii, xlix (with towers), liii, lvi, lx, lxi, lxii, lxv, cv (annexe only), cx, cxiii, cxxv, cxxviii–ix, cxli.

^{page 21 note 4} Arch. Journ. lxxxix, 25, fig. 5, pl. v, a.

^{page 22 note 1} Arch. Cambrensis, 1914, series vi, vol. 14, 207Google Scholar.

^{page 22 note 2} Cf. scenes xii, lx, lxv, cvi, cxxviii, cxli, cxlvii.

^{page 22 note 3} Newstead, 14, plan: Inchtuthil, , PSAS, xxxvi, 207, fig. 6Google Scholar.

^{page 22 note 4} Chew Green, MacLauchlan, Map of the Watling Street, sheet vi; Bosanquet, , Hist. of Berwickshire Nat. Club, xxv, pp. 63–4Google Scholar, pl. ii. Reycross, , C. & W. Trans. (N.S.). xxxiv, p. 50Google Scholar.

^{page 23 note 1} Arch. Journ. lxxxix, 30, pl. vii; Arch. Camb. 1914, ser. vi, xiv, 213; Jacobi, op. cit., 51, fig. 23.

^{page 23 note 2} B.C. i, 42.

^{page 23 note 3} Vegetius, iii, 8.

^{page 23 note 4} Appian, Iberica, 86.

^{page 23 note 5} CIL viii, 2532, ‘Exercitationes militares quodam modo suas leges habent, quibus si quid adiciatur aut detrahatur, aut minor exercitatio fit aut difficilior.’

^{page 23 note 6} By Professor Fabricius, whose opinion is quoted in xix Bericht der R-G. Kommission, 1929, p. 25, by Sir Geo. Macdonald.

^{page 23 note 7} Arch. Journ. lxxxix, 45–7, 74–6, 77–8: see also the work last cited, loc. cit., and fig. 21.

^{page 24 note 1} Caes. B.G. viii, g, ‘portis fores altioresque turns imposuit.’

^{page 24 note 2} Arch. Cambrensis, 1914, ser. vi, vol. 14, 207Google Scholarsqq.

^{page 24 note 3} Ibid. p. 207.

^{page 24 note 4} Ibid. plan, 207, 210–11.

^{page 24 note 5} Ibid. 213.

^{page 24 note 6} Ibid. 211.

^{page 24 note 7} For the identification of these objects see below, p. 36, n. 1.

^{page 24 note 8} Bar Hill, Roman Forts on the Bar Hill, 12, fig. 2: Hill, Croy, PSAS, lxvi, 263Google Scholar, fig. 13, also Bar Hill, Ibid. 264, fig. 14.

^{page 24 note 9} J. Ward, Roman Fort at Gellygaer, 1903, plan; cf. Arch. R.B. fig. 6.

^{page 24 note 10} Arch. Journ. lxxxix, 21, fig. 3.

^{page 24 note 11} Vegetius, ii, 11: habet praeterea legio fabros tignarios, structores, carpentarios, ferrarios, pictores, reliquosque artifices ad hibernorum aedificia fabricanda.… Horum iudex erat proprius praefectus fabrorum.

^{page 24 note 12} Caligae are consistently worn by the legionaries on the Column: but the Eastern archers wear a kind of low shoe, while the other auxiliaries do not wear the caliga, though it is not clear what type of boot or shoe they use.

^{page 25 note 1} I am indebted for these details to Mr. James McIntyre, who dissected examples from Birdoswald and Chesterholm.

^{page 25 note 2} These were thickly studded, and could give a hard kick: cf. Juvenal, , Sat. xvi, 24–25Google Scholar, Cum duo crura habeas, offendere tot caligas, tot Milia clavorum!

^{page 25 note 3} Cf. cxxxviii, where pack-horses bring in the treasure of Decebalus.

^{page 26 note 1} The term is derived from clavis, with the Latin diminutive suffix. The most primitive form of key was a hook on the end of a bar, inserted through a hole to lift up the latch. The collar-bone receives its name from its resemblance to this form of key. Cf. Diehls, , Antike Technik, 41, Abb. 7–11Google Scholar, where every sense except this military one is mentioned.

^{page 26 note 2} Napoleon, Atlas, pl. 9: Grenier, , Manuel d'Archéologie gallo-romaine, v, I, 193Google Scholar, fig. 12: Rice-Holmes, Caesar's Conquest of Gaul (ed. 2), pp. 71, 661–4, 666, summarises various objections which would be removed by refusing to accept this camp as Caesarian at all. The type is so isolated that there is much to be said for Kahrstedt's transferring it to the Flavian age, on typological grounds, Bonner Jahrbücher, cxxxviii, 146, 151–152.

^{page 26 note 3} Josephus, , B.J. vii, 275–9Google Scholar; cf. Masada, pls. iv, v, vii.

^{page 26 note 4} Arch. Journ. lxxxix, pl. vii, xxi.

^{page 26 note 5} Newstead, plan, p. 38, also fig. 3, p. 80: Dealginross, Roy, Military Antiquities, pl. xi: Cadder, Clarke, The Roman Fort at Cadder, 26, and general Ardoch, plan., PSAS, xxxii, 438Google Scholar, pl. v, east gate.

Newstead I and Bar Hill exhibit devices akin to the clavicula, but not quite the same thing: see Newstead, plan, p. 38: Roman Forts on the Bar Hill, 12, fig. 2.

^{page 26 note 6} Y Pigwn, Ward, R-B. Buildings and Earthworks, 9, fig. 3 a: Glenwhelt, MacLauchlan, Survey of R. Wall, pl. iv; idem, Watling Street, Four Laws and Dargues, pl. v, Chew Green III, Featherwood and Bellshields, pl. vi; Sills Burn South, P.S.A.N. ser. 4, vi, 241, 243Google Scholar; Birrenswark, , Brit. Acad. Suppl. Pap. viGoogle Scholar, fig. 9 = Bericht xix. R-G. Komm. fig. 19.

^{page 26 note 7} c. 55, puncto mediae portae, adaperto circino ad cardinem portae. This will only give a quarter-circle covering one-half of the gate.

^{page 26 note 8} Roman Forts at Castleshaw, Interim Report ii, pl. 17. Ribchester, Smith and Short, History of Ribchester, 1890, part i, p. 11. Examples appear in scenes xiii, xlix, cxxviii, of intermediate courses of logs: in cix and cxli, the logs are at the base of the rampart.

^{page 27 note 1} Newstead, 24: Coelbren, , Arch. Camb. ser. vi. vol. 7, 139–40Google Scholar.

^{page 27 note 2} Macdonald, Roman Wall in Scotland (ed. i), pl. vi, 1, 2: Slack, , YAJ xxvi, 12Google Scholar, pl. iii, fig. 6; Ilkley, , YAJ xxviii, 156Google Scholar, pl. i, figs. 1, 2.

^{page 27 note 3} The cervoli are evidently upright and used alone, as an alternative to the rampart; this is decisively shewn by Jacobi, commenting upon Livy, xxxiii, 5, 9 (see note 16, p. 12) and Polybius, xviii, 18, op. cit. 57–59: cf. also Frontinus, , Strat. i, v, 2Google Scholar, vallum cervolis et alio materiae genere constructum incendit.

^{page 27 note 4} De mun. castr. 3.

^{page 27 note 5} Newstead, plan, p. 14.

^{page 27 note 6} Roman London, Anc. Mon. Comm. Report, 176, and, for illustration, Antiquaries' Journal, vi, pl. xxxv, 2.

^{page 27 note 7} Germania Romana, iii², pl. xxxix, 2, xlii, 1; a further selection of these representations is handily collected in Loeschcke's, Denkmäler vom Weinbau aus der Zeit der Römerherrschaft an Mosel, Saar und Ruwer, Trier, 1933, 21–26Google Scholar. Cf. Camille Jullian, v, 233, footnote 2, a very interesting discussion on barrels.

^{page 28 note 1} Cf. Collingwood, R. G., Vasculum viii, 4–9Google Scholar.

^{page 28 note 2} Oberaden, , Jahrbuch K.D. Inst. xxiii, 79 ff.Google Scholar: Castleshaw, , Interim Report, ii, pl. 17Google Scholar; pl. 25 of the latter work illustrates also the Oberaden pila muralia.

^{page 28 note 3} The mere preparation of boards for the purpose would have been a formidable task.

^{page 28 note 4} Bearing these towers in mind, it is easy to reconstruct Caesar's extraordinary fortifications of B.G. viii, 9: the towers had three floors, one at the rampart-walk level, and two above: they were placed so close that the intermediate floors could be joined (pontibus traiectis) and a second rampart walk thus laid (constratisque coniungi). With towers which were only frameworks this was easy to construct.

^{page 28 note 5} Vegetius, iii, 7, between horse-lines; by dividing the stream; rafts; swimming.

^{page 28 note 6} Vegetius, i, 10: cf. scene xxvi, where the army is crossing a river.

^{page 29 note 1} Cf. Caes., B.G. v, 11Google Scholar, B.C. i, 36; Tac., A. ii, 6Google Scholar; Frontinus, , Strategemata, iv, 1, 20Google Scholar.

^{page 29 note 2} The coins are Cohen, , Méd. Imp. iv, Sept. Sev. 522Google Scholar, an elaborate bridge, with portico, perhaps not the same thing; Caracalla, 603, a bridge of boats. Cf. SirOman, Charles, Num. Chron. 1931, iii, 137Google Scholar.

^{page 29 note 3} Dacian tortures are shewn in scene lxv: cf., however, amputation of hands, Bell. Gall. viii, 44.

^{page 29 note 4} Cf. Caes., B.G. ii, 33Google Scholar; iii, 16; vii, 89; Tac., A. xiii, 39Google Scholar.

^{page 29 note 5} Germania Romana, ed. 2, i, pl. ix, 3.

^{page 29 note 6} Scene xxx: a lady of distinction is followed by a crowd of women with babes in arms; they move towards a moored boat on the river-bank, while Trajan and his soldiers extend their hands in farewell.

^{page 29 note 7} This is the type of hat worn by Decebalus, xxiv, cxlv, and other chiefs, lxxv, cxxxv: ordinary Dacians go bareheaded.

^{page 30 note 1} Arch. Camb. ser. vi, vol. vii, 149.

^{page 30 note 2} C. & W. Trans. ser. ii, vol. xiii, 378, fig. 35.

^{page 30 note 3} Willowford, Ibid., 396; Gellygaer, Arch. R.B. fig. 6.

^{page 30 note 4} Arch. R.B. fig. 5: JRS xiii, pl. viii.

^{page 30 note 5} Inscription of A.D. 99–100, JRS xviii, 1928, 211Google Scholar, fig. 71; for date of wall, see Arch. Camb. lxxxv, 1930, 196Google Scholar.

^{page 30 note 6} CIL vii, 241, from the gateway in King's Square.

^{page 31 note 1} Cf. the advantages urged for the canal connecting Belgica and Lugdunensis, Tac., A. xiii, 53Google Scholar, sublatis itineris difficultatibus, navigabilia inter se Occidentis Septentrionisque litora fierent. For river-travel in general, C. E. Stevens, Sidonius Apollinaris, 75, note 1.

^{page 31 note 2} JRS xv, 177–8, pl. xxxi, 2.

^{page 31 note 3} Caerleon, , Arch. Camb. 1929, lxxxiii, 4Google Scholar, Archaeologia, lxxviii, pls. 17 (general plan in relation to fortress), 18 (perspective), 20 (detailed plan): Chester, , Journ. Chester and N. Wales Architect. Arch. and Hist. Soc., N.S. xxix, pl. xviGoogle Scholar.

^{page 31 note 4} Gladiators' training-schools (ludi) were small Amphitheatres, with a limited seating and large arena, like these; and the connection between gladiatorial technique and the soldier's sword-drill, introduced by Rutilius Rufus (Val. Max. ii, 3, 2) will be recalled. Another feature which the Roman War Office may be thought to have borrowed from the world of athletics were the naves lusoriae, no doubt at first connected with the naumachia: their small size is to be calculated from Amm. Marc, xviii, 2, 11–12. For a discussion of ludi, see Lundström, Undersökningar i Roms topografi, 22–25.

^{page 31 note 5} Cf. Arch. R.B., fig. 4.

^{page 31 note 6} As at Xanten, Bonner Jahrbücher, 126, p. 1: even in the smaller castella, the principia contained a high cross-hall.

^{page 31 note 7} Cf. Ostia, Calza, Guida storico-monumentale, plan.

^{page 32 note 1} Cf. the Arch of Dativus at Mainz or the Heidenthor at Carnuntum.

^{page 32 note 2} B.G. v, 8, where the skill of the soldiers working cargoboats is compared favourably with that of those working warships. The baggage is represented by tents in bundles and soldiers' kit hanging from the awning of a cargo-boat. Caesar expected a navis oneraria to take about 100 men, as emerges from B.G. iv, 22.

^{page 32 note 3} Scene lxxix shews a rich variety of these cabin-awnings.

^{page 32 note 4} In the classis Britannica triremes seem to have been common, cf. CIL xiii, 3540–3546: Desjardins, , Géogr. de la Gaule, i, 367Google Scholar; Atkinson, D., Classis britannica, in Hist. Essays in Honour of James Tait, Manchester, 1933Google Scholar.

^{page 32 note 5} Cf. Caes., B.G. iii, 14Google Scholar, mentioning low fortified prows. The improvement was introduced by Gonatas, see Tarn, Hellenistic Military and Naval Developments, 149. The early colonial coins of Vienne give a fine representation of the storeyed towers on the prows of the vessels that fought at Actium and were sent to Fréjus.

^{page 32 note 6} For sea-going craft with sails, see scenes lxxx, lxxxvii. The Roman sails, with which it was impossible to tack, would be of little service except upon long straight reaches.

^{page 32 note 7} Cf. Tac., Hist. v, 22Google Scholar, praetoriam navem vexillo insignem.

^{page 32 note 8} This unsound structure must be due to over-simplification by the copyist.

^{page 33 note 1} Tac., A. i, 20Google Scholar.

^{page 34 note 1} Caes., B.G. iv, 17Google Scholar.

^{page 34 note 2} Strabo, iv, 1, 12, ; Cf. the Rhine bridge at Mainz (Koepp, Römer in Deutschland, 89, Abb. 99, 102) and the Mosel bridge at Trier (op. cit. Abb. 81).

^{page 34 note 3} Cf. Mélida, Catalogo monumental de España, Provincia de Cáceres, pls. xxx, fig. 51, xxxi, fig. 52.

^{page 34 note 4} Arch. Journal, lxxxvii, 113, pl. viii.

^{page 34 note 5} Alcántara, Mélida, op. cit., pls. xxvii (plan), xxviii (elevational view), xxvi (general view); cf. St-Chamas, Bonner Jahrbücher, 118, p. 316, fig. 8.

^{page 34 note 6} Arch. Ael. Ser, 4, xi, figs. 1–3, pls. xxii–xxv.

^{page 34 note 7} Cf. Clarke, Roman Fort at Cadder, 18, for an intermediate stage rather of this kind.

^{page 34 note 8} Dio Cassius, lxviii, 13.

^{page 35 note 1} Collingwood, R. G., C. & W. Trans. (N.S.), xxix, 151Google Scholar.

^{page 35 note 2} E.g. Maryport, , CIL vii, 379–82Google Scholar; and Bowness on Hadrian's Wall.

^{page 35 note 3} PSAS, 1900–1, 25 sqq.

^{page 35 note 4} Cf. the German examples dating to Domitian.

^{page 35 note 5} Die Marcussäule, text. 109, note.

^{page 36 note 1} Roman Wall in Scotland, ed. II, 355–8, pl. lx.

^{page 36 note 2} Vegetius, iii, 5.

^{page 36 note 3} Cf. the alphabetical code invented for use with torches by Polybius and two colleagues, Polybius, x, 45–7.

^{page 36 note 4} Vegetius, iii, 5: Aliquanti in castellorum aut urbium turribus appendunt trabes, quibus aliquando erectis aliquando depositis indicant quae geruntur.

^{page 37 note 1} Schuchhardt, , Sitzungsbericht der Preussischen Akademie der Wissenschaften, Phil-Hist. Klasse, 1931, xxiii, 608–634Google Scholar.

^{page 37 note 2} See Dio Cassius, lxviii, 9, 5; also Roman artisans in Dacia, Dio, lxvii, 7, 4, . Cf. the Gauls in the estimate of Caesar, , B.G. vii, 22Google Scholar, genus … ad omnia imitanda et efficienda quae a quoque traduntur aptissimum.

^{page 37 note 3} B.G. vii, 73.

^{page 37 note 4} B.G. vii, 73; cf. Bell. Afr. 31, stili caeci.

^{page 37 note 5} For water shewn in an artificial duct or dike, cf. scenes l, lxxiv, cxxiv.

^{page 37 note 6} Cf. scenes xciii, cxii, cxix, cxx, cxxiv: this is exactly the state of the Welsh hill-forts described by Willoughby-Gardner, , Arch. Camb. lxxxi, 221–282Google Scholar, which all have great post-holes for towers covering the entrance.

^{page 37 note 7} Scenes cxiii–cxv (rubble), cxix (ashlar and wood), lxii (ashlar).

^{page 38 note 1} Catalogue du musée de Langres, 1931, nos. 54, 94, 112, 113, 114; cf. Espérandieu, , Bas-reliefs, iv, 3301, 3293, 3295Google Scholar.

^{page 38 note 2} For inferior Dacian bows, see scenes xxiv, xxxii.

^{page 38 note 3} Cf. scenes lvi, xxiv, lxxii; in xxiv note the Roman trooper fighting with a head, carried by the hair, in his mouth.

^{page 38 note 4} Scenes xliii–xlv.

^{page 39 note 1} Jacobi, , Saalburg Jahrbuch, iv, 24Google Scholar.

^{page 39 note 2} Caes., B.G. vii, 23Google Scholar; cf. Childe, , PSAS, lxvii, 387Google Scholar, Ibid. 369, fig. 3, Castlelaw, , and Antiquaries' Journal, xiiiGoogle Scholar; Hawkes, , Antiquity, v, 71–2Google Scholar, figs. 6 (Uffington), 8 (Uffington and St. Catherine's Hill), pl. ii, 85 (Corley, ); Antiq. Journal, xi, 22Google Scholar (Cissbury); ibid, xiv, 17 (Titterstone Clee).

^{page 39 note 3} Scene cxvi.

^{page 39 note 4} Cf. Cichorius, iii, 229–232.

^{page 39 note 5} Scene lxvi, a ballista mounted on a tribunal.

^{page 39 note 6} Scene lxxv.

^{page 40 note 1} Schramm, Die antiken Geschütze der Saalburg, Abb. 33, 34, Taf. 10, based upon Amm. Marc. xxiii, 4, 5.

^{page 40 note 2} Vitruvius, i, 5, 5; Veget. iv, 14; cf. scene xxxii.

^{page 40 note 3} Procop., B.G. i, 21Google Scholar; see Cichorius, iii, 229–232.

^{page 40 note 4} Cf. Caes., B.G. iv, 17Google Scholar; Vitruv. iii, 3 intervalla … solidanda festucationibus.

^{page 40 note 5} See p. 37, note 2, for the recent importation of Roman builders.

^{page 40 note 6} Scene cxix: the Romans finally planted their camp in one angle of the native fortress, see scene cxxv; cf. Hod Hill, Wessex from the Air, pl. i.

^{page 40 note 7} Scene cxvii, cf. lxvi.

^{page 40 note 8} Frontinus, , Strategemata, iii, 5, 2Google Scholar: ‘Ti. Gracchus, Lusitanis dicentibus in X annos cibaria se habere et ideo obsidionem non expavescere, ‘undecimo’ inquit, ‘anno vos capiam'.’