^{page 174 note 1} Those here reproduced were taken partly by myself and partly (at my direction) by M. Barral of Arles. A rather poor photograph of the Roman gate is reproduced in the useful collection of illustrations published at Avignon under the title of Les villes romaines de la vallée du Rhône, 1926.

^{page 175 note 1} For a more ample history, see Constans, op. cit.

^{page 175 note 2} Cassiodorus, , Variae, III, 44Google Scholar.

^{page 176 note 1} Congrès arch. 1876, pp. 270 ff.

^{page 179 note 1} For the conduit, see M. Clerc, Aquae Sextiae, 1916, p. 432. For the Aix gate see ibid. pp. 263 and 423. The sketch plan in Fig. 59 of the present paper is an approximation to the fifteenth and eighteenth century ‘plans’ collected by M. Clerc.

^{page 179 note 2} This gate is well known from an inaccurate perspective-drawing which has often been reproduced (e.g. V. Petit, Congrès arch. 1866, pp. 280 ff.; Blanchet, op. cit., Pl. xix; Schultze, Bonner Jahrb. 118, Fig. 5). More recently (1920) M. Jules Formigé, chief architect to the Monuments Historiques, has partially cleared the base of the structure, but it is not easy to square the visible remains with his plan, which, with other illustrations, may be found in Donnadieu, A., Le Pompéi de la Provence: Fréjus (Paris, 1927). p. 95Google Scholar. Whilst the main principles of the design are clear enough, there are details which require further investigation: in particular, the excavation revealed on the north side (and failed to reveal on the south) the fragmentary foundation of a semicircular tower or guardroom which is definitely of the same build as the rest of the structure but seems never to have been completed. The base of the adjacent gate pier is carried across it, and a vertical seam in the core of the pier stops at a height of rather less than 3 m. suggesting that the provision for an intended guardroom-wall was discontinued at this point. Moreover, there is no sign of the abutment of the intended guardroom against the curved town-wall on the north side. It is only possible, I think, to postulate a change of purpose during the progress of the work.

The plan here reproduced (Fig. 59) is completed from the surviving fragment of the similar gate known as the Porte de Rome at the opposite end of the town. The following additional notes on the Porte des Gaules may here be placed on record: the flanks of the footways are faced with small cubes similar to those of the town-wall (average ·12 m. deep by ·24 m. long). whereas the flanks of the main roadway are lined with large blocks (up to ·50 m. deep by ·70 m. long). The north tower has been wholly or largely rebuilt as a cottage; the south tower is represented by a fragment hewn out of the live rock.

^{page 181 note 1} Bursian, C., Mittheilungen der antiquar. Gesellschaft in Zürich, xvi, 1867Google Scholar, I^re série, fasc. 1 à 5.

^{page 181 note 2} Henberger, S., Arch. Anzeiger xxiii, 1921, pp. 76 ff.Google Scholar; Schulthess, O., Röm.-germ. Kommission, (Deutsches archäologisches Institut), Bericht xv, 1923–1924, p. 19Google Scholar.

^{page 181 note 3} Der römische Limes in Oesterreich, 1901, p. 54.

^{page 181 note 4} O.R.L. Lief. 24, Theilenhofen, p. 4.

^{page 181 note 5} Cagnat, M. R., ‘Les deux camps à Lambèse,’ Mémoires de l'Acad. des inscr. et belles-lettres, xxvii (1908)Google Scholar, figs. 3 and 5.

^{page 181 note 6} O.R.L. Lief. 35, Faimingen, p. 12.

^{page 182 note 1} Fougères, G., Mantinée el l'Arcadie orientale, 1898, p. 153Google Scholar; Clerc, M., Aquae Sextiae, p. 433Google Scholar.

^{page 183 note 1} A. Véran, Congrès arch. 1876, p. 271; Constans, op. cit., p. 224, has failed to find this fragment.

^{page 183 note 2} Blanchet, op. cit., p. 263. In late Roman times rectangular towers were not uncommon, as in the Aurelian walls of Rome and at Richborough in Kent. The Roman or sub-Roman towers of Carcassonne have round fronts built on square bases which have been regarded (wrongly, I think) as of earlier Roman date; and the semicircular bastion at All Hallows on the London wall stands on a square platform.

^{page 184 note 1} Small-stone and large-stone construction are, it is true, frequently associated in Roman work: e.g. the Porte St. André at Autun is itself built of large stones, but its flanking towers (contemporary) are of small cubes. At Nîmes the Porte d'Auguste is of large stones, the town-wall is of small stones, and the Porte de France is partly of the one and partly of the other, yet all these works may be regarded as contemporary—the inscription on the Porte d'Auguste, dated 16 B.C., commemorates the construction of ‘gates and walls.’ Nevertheless, as one stands in front of the walls and towers of Arles, it is difficult to suppose that the towers of the gateway could have been the work of the builders of the small-stone wall, quite apart from the other considerations discussed below,

^{page 184 note 2} The side shown in the photograph (Plate XXII) is weathered and otherwise damaged, and does not therefore indicate the precision of the original jointing. This can still be seen on the other, less accessible, side of the tower.

^{page 184 note 3} See a drawing by Gibelin, c. 1785, reproduced by M. Clerc, Aquae Sextiae, pl. xv. This excellent drawing probably gives a fair impression of the elaborate appearance of the towers of the Arles gateway in their prime.

^{page 186 note 1} As has been indicated above, the association of the two systems may have been envisaged at Fréjus.

^{page 186 note 2} Op. cit., pp. 227 ff.

^{page 187 note 1} Véran's full description in Le Musée, 1873–4, pp. 49–50, is a valuable record.

^{page 187 note 2} Clair, H., Les monuments d' Arles, 1817, p. 69, noteGoogle Scholar; cited by Fassin, , Le Musée, 1873–4, p. 51Google Scholar.

^{page 187 note 3} The walls of Nîmes are dated by the inscription; the early walls of Autun, though not dated, can hardly be other than those of the Augustan city, the wealth of which is referred to by Pomponius Mela, and Fréjus, as the great Augustan naval base of Gaul, must certainly have been walled at an early date.

^{page 188 note 1} Rivoira, G. T., Roman Architecture, 1925, p. 70Google Scholar.

^{page 188 note 2} Constans, op. cit. pp. 293–4.

^{page 189 note 1} Cited by Fassin, in Le Musée, 1873–4, p. 51Google Scholar.