^{page 61 note 2} Vol. xv, pp. 176–194. This is referred to as York, 1925.

^{page 61 note 3} This date was inferred from the character of the pottery associated with the clay rampart. I may note here that in 1926 Mr. T. Davies Pryce made a careful examination of the early Samian ware in the York Museum and concluded that it gave evidence of a pre-Agricolan occupation of the site.

^{page 61 note 4} Throughout I use the term ‘mound’ only in referring to this post-Roman structure. Such evidence as our excavations produced for its date and character and for other post-Roman structures is to be discussed elsewhere by Mr. Raine.

^{page 62 note 1} The inflow of water made it impossible to dig beneath the foundations and complete our Section D¹ by determining the line and mode of junction of the inner concrete with the outer cobbles and clay. For the same reason we could not ascertain if the foundations rested upon piles as at C, fig. 20 (Section G–H of York, 1925, J.R.S. xv, pl. xxvi).

^{page 64 note 1} In York, 1925, J.R.S. xv, p. 188, I described this foundation of cobbles as run with mortar. This description was based upon observation of the upper part of it, and should not have been applied to the whole. The foundation of cobbles and clay at the corner also showed mortar in its upper part. This had no doubt been run into it when the bed of grouted rubble was laid upon it.

^{page 67 note 1} So far it has been our experience that foundations of cobbles and clay belong to the earliest stone structures, concrete being used later.

^{page 68 note 1} The sharp stepping-down of the dressed stonework suggested a staircase. Cf. Cagnat, L'Armée romaine d' Afrique, 2nd ed., p. 458.

^{page 68 note 2} In plate VII, no. 3, the backmost blocks are concealed by the burnt matter packed around them. When this was cleared away, the lower course of the row was found to be complete from wall to wall, as shown in fig. 15, Section D.

^{page 68 note 3} It will be seen that the upper course is at about the level which I have suggested above (p. 67) for the original gun-platform.

^{page 71 note 1} Here and there in the discussion that follows, I develop suggestions which have been made to me, notably by Mr. C. R. Peers and Dr. R. E. M. Wheeler.

^{page 75 note 1} Eph. Epig., vii, 935.

^{page 75 note 2} C.I.L., vii, 241.

^{page 75 note 3} The surviving letters are The fragment measures only 9 in. by 6 in., and the restoration TRAI[ANVS A]VG or TRAI[ANO A]VG would imply for the stone in its original condition an improbably long and narrow shape. The inscription is therefore to be restored TRAI[ANVS HADRIANVS A]VG(ustus) P(ater) [P(atriae)]. The title pater patriae was not received by Hadrian officially till 128, but its occurs on inscriptions dating from the early years of his reign (De Ruggiero, s.v. ‘Hadrianus,’ pp. 616–8. Cf. the Moresby and Great Chesters stones, C.I.L. vii, 362, 730).

^{page 76 note 1} The same would be true of the interval-tower. There indeed it is uncertain i f the readjustment of alignment had any relation at all to the line of the rampart-wall. It may have been simply a change from a rectangular plan to a tower with converging side-walls (see fig. 16). Our excavation there was not extensive enough to decide this. In either case the change of alignment gives no adequate reason for a complete rebuilding of the tower, and there, as at the corner, the destruction of the first stone tower would seem to be implied by the view which would connect the building of the second stone tower with the building of the first rampart-wall.

^{page 81 note 1} Probably after the Roman period. Some small fragments of tiles appeared not to be Roman, and it was doubtful if the piles, which differed from those found elsewhere, were Roman. In any case the piles and concrete, with the pottery associated with them, were clearly intrusive.

^{page 83 note 1} It is possible that the concrete, if not the piles, found in the ditch in our trench at F may have come from an earlier Roman structure.

^{page 83 note 2} I have too much material from our own excavations to dispose of to have space here for a discussion of the character and date of this building.

^{page 87 note 1} Mr. C. R. Peers suggested to me that, at the time when the fourth-century restoration was carried out, repairs or other work may have been going on at the harbour.

^{page 89 note 1} For the details that follow see plate x, Section H and plate ix, no. 3.

^{page 90 note 1} For the details that follow see, besides Section G, the Section across the Interior of Room 1 of Site VI and the plan of Site VI—all on plate x.

^{page 91 note 1} See plate x, plan of Site VI, for the details here.

^{page 92 note 1} For the details that follow see plate x, Section across Interior of Room 1 of Building VI; also plate IX, no. 5.

^{page 94 note 1} Final proof that building VI was an interior building and that the defences of that period lay outside, and not inside, the fourth-century line can come only with further excavation. Meanwhile, in support of the view expressed in the text, I cite an interesting piece of documentary evidence unearthed by Mr. Raine. This is a charter of the twelfth century granted to St. Leonard's Hospital by Clement, Abbot of St. Mary's, and relating to a piece of land que terra incipit a cruce lapidea tendens versus aquilonem usque ad fossam de la barre et a regia via usque ad vetus fossatum regis quod dicitur Wirchedic (Farrer, , Early Yorkshire Charters, vol. I, p. 215Google Scholar, no. 276). Regia via is Bootham (see plate xi), and Wirchedtc is the ditch of the Aldwark, a term already used in pre-Norman times to indicate the Roman defences on the north-east side of the fortress. The description of the boundaries implies—and there is later documentary evidence to confirm it—that the Roman ditch on the north-east side (vetus fossatum regis quod dicitur Wirchedic) had continued past the north corner of the fourth-century fortress until it met another ditch (fossa de la barre) running south-west towards Bootham (regia via).

^{page 95 note 1} At least the earth for the back of the mound had been so obtained. That for the front of it had apparently been got by stripping the ground outside, as indicated by the fact that the earth excavated from the front part of the mound in our Section L (plate XI) contrasted sharply with that from the back in containing little or no pottery or other débris.

^{page 96 note 1} I am indebted to Mr. Philip Corder for the use of his drawings of this collection and for notes comparing the pieces with the types which he has found at Crambeck.

^{page 96 note 2} In addition to the evidence referred to above the barrack appeared to show traces of more than one floor level and to present other marks of reconstruction.

^{page 98 note 1} Or that other troops were associated with it. But for that there is no evidence.

^{page 98 note 2} Herodian, iii, 8; Wilmanns, Étude sur le camp et la ville de Lambèse, p. 12, and C.I.L. viii, p. 284; Cagnat, L' Armée romaine d' Afrique, 2nd ed., pp. 380–1. For the interior arrangement of a Severan fortress see G. Lugli's study of the camp of the legion II Parthica at Albano, in Ausonia, vol. ix 1919), pp. 211–265Google Scholar. I am indebted to Mr. Ian Richmond for some notes from this article and for a plan.

^{page 98 note 3} The south-west gateway, as taken from the O.S. map of York, shows one or two features difficult to understand. This is still truer of the north-west gateway,which I have omitted altogether. The projecting interval-towers along the river-front have been taken from Mr. George Benson's Archæological Plan of York.