¹ The author would like to record his thanks to all who have contributed to this project and, in particular, to the small team of excavators upon whose dedicated efforts, often in the most difficult conditions, all depended: G. Evans, W.H. Herring, C. Ravenhill, J. Richards, A. Thomas, and L.C. Zienkiewicz. The project was initiated by Mr R.J. Brewer of the Department of Archaeology and Numismatics at the National Museum of Wales, and I am grateful to him, as to Mr G.C. Boon and Dr S.H.R. Aldhouse-Green (as former and present keepers), and departmental colleagues for their support. Mr Boon and Mr Brewer have kindly read drafts of this report which benefits from their helpful comments. Many of the finds were drawn by Mrs A. Leaver, to whom also my thanks. Site plans etc. are drawn by the author.

² Summary reports appear in Britannia xv (1984), 269Google Scholar with fig. 2, and Britannia xvi (1985), 257–8.Google Scholar

³ Britannia xx (1989), 263.Google Scholar

⁴ Antiq. Journ. lxiv (1984), 403–7.Google Scholar

⁵ Britannia xvii (1986), 450–2 and pl. xxxiv.Google Scholar

⁶ Partial plans of timber barracks were recorded at the ‘Roman Gates’site in 1981; Evans, D.R. and Metcalf, V.M, Roman Gates, Caerleon, Oxbow Monograph 15 (1992).Google Scholar

⁷ Britannia xix (1988), 421–2.Google Scholar

⁸ A metrological study of these two sites has produced interesting results, the detail of which is wholly omitted here; the restored plans which accompany the discussion below (FIGS 21, 23–5) depend on that work, without further explanation, and there are implications for the layout of the fortress as a whole; but full discussion is beyond the scope of this report.

⁹ For ‘natural’ at the baths, and preparation of the site there, see J.D. Zienkiewicz, The Legionary Fortress Baths at Caerleon (1986), I, 51–2.

¹⁰ Britannia xviii (1987), 307.Google Scholar

¹¹ See note 7 above.

¹² Evans, E.M., ‘Excavations at ‘Sandygate’, Cold Bath Road, Caerleon, Gwent’, Britannia xxii (1991), 103–36CrossRefGoogle Scholar; the natural surface, seen by this author, was identical at all sites.

¹³ Lady Fox, however, was of the opinion that irregular holes in the natural ground surface, encountered on the Myrtle Cottage site, were left by the drawing of tree-stumps; Arch. Camb. cxv (1940), 105Google Scholar; but see the evidence of the plant remains from Wells 1 and 2, below, pp. 136–8.

¹⁴ The Priory Garden site; Arch. Camb. cxiii (1964), 28, and see comment below, p. 39.Google Scholar

¹⁵ Zienkiewicz, op. cit. (note 9), 1, 187f.

¹⁶ For similar construction in successive phases at the Valkenburg, see W. Glasbergen, De Romeinse Castella te Valkenburg Z.H. (1962), figs 23–5.

¹⁷ Compare preserved panelling (with slotted uprights etc., but on basal beam) from the Valkenburg (Castellum 1; Claudian); ibid., figs 12–14.

¹⁸ For coal in mortar of the primary fabric of the Fortress Baths, c. A.D. 75 (and further refs), see Zienkiewicz, op. cit. (note 9), 1,343. Coal adhered also to the iron bowl, No. 3 below (from Well 1), but fragments large enough to have been brought to the site as fuel occurred at the earliest in Phase II contexts. Coal also appeared in some quantity in connection with the smithy of Phase III–IV.

¹⁹ At Silchester, the square boarded frame of a well was ‘supported on an oak curb 7½ in deep'(Archaeologia lii (1890), 743)Google Scholar; G.C. Boon, Silchester: The Roman Town of Calleva (1974), 85–6 for discussion.

²⁰ Floors of the primary tabernae at the Telecom site (note 7 above) were similarly made up.

²¹ The very small quantity of slag which was otherwise recovered from contexts earlier than Phase IIIa, over the site as a whole, suggests that little iron-working was conducted either during the construction or occupation of the timber buildings.

²² Summarised (by J.D.Z.), with permission.

²³ Hillam, J., Current Archaeology 96 (1985), 21–6Google Scholar; Sheldon, H.L. and Tyers, I., London Archaeologist 4.13 (1983), 355–61.Google Scholar

²⁴ Baillie, pers.comm.

²⁵ G.C. Boon, lsca: The Roman Legionary Fortress at Caerleon, Mon. (1972), 25–6; and idem. The Legionary Fortress of Caerleon-Isca (1987), 27.

²⁶ Preliminary assessment by M. Maltby.

²⁷ This erroneous view is expressed in Britannia xv (1984), 269.Google Scholar

²⁸ About 10 kg was retained from this and other contexts in the smithy.

²⁹ This may be some reflection of the degree of attention that was generally paid to the official collection of scrap metals for re-use, as we are reminded by the discovery of much scrap bronze held in amphorae in the western aerarium of the legionary headquarters at Novae (Samowski, T. in Germania lxiii (1985), 521–40).Google Scholar

³⁰ e.g. in Neronian context (with bronze and iron dust) at Verulamium; S.S. Frere, Verulamium Excavations I (1972), 18–19Google Scholar, citing others in contexts down to c. A.D. 120, and parallels (Flavian) at Catterick. A large number o f such troughs, of similar size but with trace of plank linings, were arranged in orderly fashion at the fabrica of the Exeter fortress; P.T. Bidwell, Roman Exeter: Fortress and Town (1980), 31–4.

³¹ Frere (op. cit. (note 30), 18) notes a similar earth-fast timber (20 in by 6 in) between two of the trays at Verulamium.

³² At the Culver Street site, Colchester, (Britannia xiv (1983), 309–10Google Scholar and fig. 15) one tribune's house incorporated a brass-smelting workshop, which chamber shared an orderly arrangement of shallow pits (with burnt bases) around its perimeter.

³³ Zienkiewicz, op. cit. (note 9), I, 157–8 for this portico; excavations at the Telecom site now suggest that the columns were spaced at 4.6 m (not 5.5 m) centres and loose quadrant bricks indicate columns of 60 cm diameter.

³⁴ Much butchers' waste was dumped, apparently at similar date, into the natatio of the Fortress Baths; T.P. O'Connor in Zienkiewicz, op. cit. (note 9), 11, 230–4.

³⁵ Legionary stamps with the Antoniniana suffix occur only on roofing-tiles, and never on bricks; G.C. Boon, Laterarium Iscanum: The Antefixes, Brick and Tile Stamps of the Second Augustan Legion (1984), 34.

³⁶ Fortress Baths: Zienkiewicz, op. cit. (note 9), 11, fig. 27,17.43; Telecom site, unpublished.

³⁷ Arch. Comb, xcii (1937), 324–5.Google Scholar

³⁸ Zienkiewicz, op. cit. (note 9), 1, 184–5 and fig. 57.

³⁹ ibid., 1,160, Section 49.

⁴⁰ ibid., 1, 184.

⁴¹ ibid., I, 28, fig. 3 and passim.

⁴² ibid., I, 150–1; finds (1965–1967) indicating that the area of the NW end of the basilica had previously served as an armourer's workshop were identical in character to those recovered from the adjacent timber tabernae in 1987.

⁴³ The existence of such a street was predicted by Boon, G.C.; Arch. Camb. cxiii (1964), 36.Google Scholar

⁴⁴ Plan in Boon, op. cit. (note 25; 1972), perpetuated in Zienkiewicz, op. cit. (note 9), 1, fig. 2 etc.; the effect of ‘moving’ this street is to enlarge the area of the hospltal significantly, as FIG. I above.

⁴⁵ Arch. Camb. cxiii (1964), 28–40.Google Scholar

⁴⁶ ibid., 32.

⁴⁷ Plan in Boon, op. cit. (note 25; 1972).

⁴⁸ ibid., 31; minor excavations referred to there as having failed to establish the presence of a tribune's house coincide with the Telecom site of 1987.

⁴⁹ ibid., 33, repeated in L.F. Pitts and J.K. St Joseph, lnchtuthil: the Roman Legionary Fortress (1985), 139.

⁵⁰ ibid., 129–36 and 140, Table iv for comparative (and comparable) dimensions at other fortresses.

⁵¹ The plan reconstruction of Timber Building A which is offered in FIGS 23 and 26 allows it a width of 52 m; but this results in an available area (1925 m²) well beyond the normal range (compare Inchtuthil III at 1460 m², and Vetera I at 1560 m²); perhaps a street separated Houses 3 and 4 (and 5 and 6, by symmetry).

⁵² Pitts and St Joseph, op. cit. (note 49), 129–36.

⁵³ Polybius tells us (vi. 27.1) that the general's tent was to be set in a 100 ft square so as to face the most convenient general view, and best for giving orders, and continues (vi. 27.5), ‘They set the tents of these [officers] all in one straight line, which is parallel to the side of the chosen square, but 50 feet distant from it to allow room for their horses, baggage animals, and equipment. Moreover these tents of the aforesaid range are pitched turned the opposite way towards the external aspect’ (my italics; I am grateful to Mr G.C. Boon for his translation, and for helpful discussion). Since it appears from Polybius' description that the tribunes’ tents were to be placed on the same side of the via principalis as the general's tent, his remark presumably means that the tribunes’ tents faced either towards the rear of the camp (and so with their backs to the via principalis) or outwards, towards the lateral gates; it does not imply that the tents opened directly on to that thoroughfare.

⁵⁴ Pitts and St Joseph, op. cit. (note 49), 136–41.

⁵⁵ H. von Petrikovits, Die Innenbauten römischer Legionslager während der Prinzipatszeit (1975). 64–7 and Bild 12.

⁵⁶ ibid., 64; i.e. probably at Lambaesis and Inchtuthil, and at Neuss, if we include a residence in the adjacent scamnum.

⁵⁷ E. Birley, ‘The equestrian officers of the Roman army’, Durham University Journal, December 1949, 8–19; and Holder, P.A., The Auxilia from Augustus to Trajan, BAR int. ser. 70 (1980), 72–85.Google Scholar The basic pattern of the tres militiae under Nero (as praefectus cohortis, tribunus militum, praefectus alae) was to become firmly established under the Flavians (ibid., 76), and is exemplified e.g. by Stlaccius Coranus and Ti. Claudius Alplnus who both held tribunates in II Augusta, the latter probably in the 80s, before going on to the command of alae (A.R. Birley, Officers of the Second Augustan Legion in Britain (1990), 10–12).

⁵⁸ The primipilate and equestrian careers were divorced by Claudius (Holder, op. cit. (note 57), 82), but the relative status of praefectus castrorum and praefectus alae, and other equestrian officers, may be perceived in their respective annual rates of pay: before Domitian these were (P.A. Holder, The Roman Army in Britain (1982), 144; in thousands of sesterces): praef. Coh.-15; trib.mil.-30; praef. alae-45; praef. cast.-60. Apportionment of accommodation to the senior officers may have reflected a similar progression.

⁵⁹ Von Petrikovits, op. cit. (note 55), 67; and Pitts and St Joseph, op. cit. (note 49), 137–8; see also useful discussion of officers' accommodation at Usk, in W.H. Manning, Report on the Excavations at Usk 1965–1976: The Fortress Excavations 1972–1974 and Minor Excavations on the Fortress and Flavian Fort (1989), 121–4 and 165.

⁶⁰ Each half of the scamnum could neatly accommodate three houses each notionally 140 ft wide and 100 ft deep, and a fourth 170 ft wide and 120 ft deep, adjustments being made throughout for the minor alleys which separated the buildings; the extrapolation of the tabernae is supportive of this resolution.

⁶¹ Vegetius, Epitoma rei militaris II. 9.

⁶² At Inchtuthil, the suspicion remains that the senatorial officers (legate and tribunus laticlavius, as general and ‘general to be’) were initially housed together in the dual residence of the temporary compound (Pitts and St Joseph, op. cit. (note 49), 220–2) where they might be isolated from the disturbance of construction work.

⁶³ Britannia xvii (1986), 450–2.Google Scholar

⁶⁴ Tacitus, Annals XII. 38.

⁶⁵ At Neuss, a residence of appropriate size wa s incorporated in the same scamnum of the praetentura as the auxiliary barracks; but see the hypothetical restoration of FIG. 26, discussed below.

⁶⁶ A.R. Birley, op. cit. (note 57), 36.

⁶⁷ The ill-fated Poenius Posfumus, who had charge of the legion at the time of the Boudiccan revolt, and P. Anicius Maximus (ibid., 9–11).

⁶⁸ RIB 317.

⁶⁹ RIB 326.

⁷⁰ RIB 324.

⁷¹ RIB 334.

⁷² Pitts and St Joseph, op. cit. (note 49), 207–22.

⁷³ Discussion, ibid., 220–2, concluding that the building had most probably been, successively, sole residence of the praefectus castrorum and of the legate; contra Richmond's view (JRS lv (1965), 200)Google Scholar that it had housed both praefectus castrorum and ‘praefectus fabrum’ together; and see my view expressed at note 62 above.

⁷⁴ Pitts and St Joseph, op. cit. (note 49), 221.

⁷⁵ See below (copper-alloy object No. 1; and PL. v B), and discussion by Boon, G.C. in Antiq. Journ. lxiv (1984), 403–7Google Scholar and pl. LV, of which some detail is necessarily repeated here, with acknowledgement.

⁷⁶ Boon (ibid., 406) considers various explanations for the presence of the trulleus, which, thus marked and still serviceable, would have been retained in official ownership and issued to personnel against an appropriate stoppage of pay. The suggestion that the vessel may have originated in an earlier fort on the same site can be entirely dismissed, for it had apparently been lost whilst the well was still in use; nor is there yet any hint of occupation or structures earlier than the establishment of the Flavian fortress.

⁷⁷ That the vessel had frequently been used in a scooping action is shown by vertical abrasions at the distal wall of the bowl and by very considerable wear to the rim there.

⁷⁸ Arch. Comb, cii (1952), 17–19Google Scholar, concluding that the units listed there (including three other alae) had formed part or all of the Caerleon command; also V.E. Nash-Williams, The Roman Frontier in Wales, ed. M.G. Jarrett (1969), 15. The diploma is C1L xvi, 48 (now RIB ii, I, 2401.1).

⁷⁹ Richmond, L.A., Hod Hill 11 (1968).Google Scholar

⁸⁰ W.H. Manning, Report on the Excavations at Usk 1965–1976: The Fortress Excavations 1968–1971 (1981), and idem, op. cit. (note 59).

⁸¹ Manning, op. cit. (note 80), 34–9 for discussion of the movements of legio XX and its identification as the likely garrison at Usk.

⁸² Britannia xix (1988), 424Google Scholar, and xx (1989), 265–8; Marvell, A., ‘Recent work on the Neronian fortress at Usk’, in Burnham, B. and Davies, J.L. (eds), Conquest, Co-existence and Change: Recent Work in Roman Wales, Trivium 25 (1991; for 1990), 19–27.Google Scholar

⁸³ RIB 201.

⁸⁴ RIB 109; for discussion see M. Hassall, in J. Wacher and A. McWhirr, Early Roman Occupation at Cirencester (1982), 69 and note 10.

⁸⁵ D.B. Saddington, The Development of the Roman Auxiliary Forces from Caesar to Vespasian (1982), 183 for the relation of the auxilia to legions; idem, 74 and 139 for the ala I Thracum.

⁸⁶ Von Petrikovits, op. cit. (note 55), 55–7; Pitts and St Joseph, op. cit. (note 49), 144–6.

⁸⁷ Von Petrikovits, op. cit. (note 55), 55–7.

⁸⁸ Hygini qui dicitur De Metatione Castrorum liber (ed. Grillo, 1977), fig. 11.Google Scholar

⁸⁹ Von Petrikovits, op. cit. (note 55), 57 and 167, n. 38 for refs.

⁹⁰ Pitts and St Joseph, op. cit. (note 49), 137 and 144–6; a total area of 5,226 m² was available, being rather less than the 6,150 m² taken up by auxiliary barracks at Neuss (ibid., 145).

⁹¹ The possibility that a detachment of the fleet may have been accommodated at Chester is unproven (Nash-Williams, op. cit. (note 78), 35–6), but the large size of that fortress would appear to demand some additional garrison.

⁹² As, e.g., at Carnuntum and Aquincum (M. Nemeth, ‘Forschungen im Alenkastell von Aquincum’ in H. Vetters and M. Kandler (eds), Akten des 14. lnternationalen Limeskongresses 1986 in Carnuntum (1990), 675–79); both are timber constructions of the late first century. In this context, attention may be drawn to the rectangular earthwork (158 by 162 m; 2.6 ha overall) at Penrhos (ST 342917), barely 1 km north-east of the fortress; although latterly dismissed as being a Civil War fort, on account of projecting bastions, perhaps added, the site has produced Roman remains (W. Coxe, Historical Tour in Monmouthshire (1801), 90–1).

⁹³ contra Boon, op. cit. (note 75), 406.

⁹⁴ It has been suggested that buildings in the sinistral part of the latera praetorii may have housed mounts for the equites of the legion (Boon, op. cit (note 25; 1972), 15, 59 and 131, n. 214; Plot XVI of his definition), but these are not barracks. Latterly, Boon prefers to identify the plot as one of several fabricae (idem, op. cit. (note 25; 1987), 54).

⁹⁵ Boon, op.cit. (note 25; 1972), 15–16 and fig. 79.

⁹⁶ Each cohort in the retentura occupied a plot of some 6,990 m ²(and rather less in the praetentura; maximum c. 6,803 m ²); Plot XVIII probably corresponds to the full cohort area, but Plot XXII covers only 5,760 m².

⁹⁷ There is, as yet, no warrant for the street, apparently separating these plots, which is elsewhere restored to the fortress plan; together they occupy a maximum area of 12,960 m².

⁹⁸ At Inchtuthil these same plots are taken up in part by pairs of granaries set per scamnum (i.e. to give four of an intended eight; Pitts and St Joseph, op. cit. (note 49), 117–22 and fig. 79); at Isca, eight horrea of the same size might be restored in two groups of four in similar position.

⁹⁹ Holder, op. cit. (note 58), 33–4.

¹⁰⁰ RIB 403; Nash-Williams, op. cit. (note 78), 15, 48–51, 150 and 180.

¹⁰¹ Calculated by taking the area of the ‘building zone’ within the via sagularis and subtracting from it the areas apportioned to the principia, latera praetorii and the principal streets.

¹⁰² That the allocation to auxiliaries at Neuss was very closely one half of this figure may perhaps be significant.

¹⁰³ CIL XVI, 70 (now RIB ii.i, 2401.6) gives the last British mention of the unit and, by reference to units known to have been stationed in the north, shows it under the command of the York legion.

¹⁰⁴ Holder, op. cit. (note 58), 111.

¹⁰⁵ J.D. Zienkiewicz, ‘The early development of the legionary fortress at Caerleon-Isca and consequences for Flavian policy in Britain’, in Burnham and Davies, op. cit. (note 82), 27–35.

¹⁰⁶ Perhaps the ala was then displaced to the Penrhos fort (see note 92) which would have been sufficiently large to have accommodated it.

¹⁰⁷ Zienkiewicz, op. cit. (note 9), 1, 37–8.

¹⁰⁸ Boon's sequence of plots (VII–XIII) numbers seven only, the projection of the baths basilica into the scamnum being included with the baths as Plot V.

¹⁰⁹ This possibility is discussed in Zienkiewicz, op. cit. (note 105).

¹¹⁰ ibid., passim.

¹¹¹ For the notion of an aborted Severan abandonment, prompted by evidence at the Fortress Baths, see Zienkiewicz, op. cit. (note 9), 1, 47–9. Dating at the Museum site is not sufficiently precise as to encourage further speculation, but the notion of a premature dismantlement is supported now by observations at the British Telecom site.

¹¹² i.e. at least down to A.D. 253–5 (RlB 334), and probably to the closing years of the third century; Boon, op. cit. (note 25; 1972), 62–4, and Zienkiewicz, op. cit. (note 9), 1, 49–50.

¹¹³ The finds are held in the collection of the National Museum of Wales under accession no. 84.43H. Phase attributions are given I, II, III etc., and context numbers are italicised. Unless otherwise stated, finds reports are by the author (J.D.Z.).

¹¹⁴ For the symbolism of this type, commonly found in Britain, and a discussion of the minting and supply of this and related issues of A.D. 154-5, see D.R. Walker, ‘The Coins’, in B. Cunliffe (ed.), The Temple of Sulis Minerva at Bath, Vol. 2. The Finds from the Sacred Spring (1988), 281–358, at pp. 293-9.

¹¹⁵ Mr P.V. Webster's efforts in reducing a large collection to this essential minimum are gratefully acknowledged; his detailed record of the plain forms is held with the site archive.

¹¹⁶ Atkinson, D., JRS iv (1914), 27–64, no. 19.Google Scholar

¹¹⁷ Fiches, J-L., Figlina iii (1978), 67, no. 87.Google Scholar

¹¹⁸ Atkinson, op. cit. (note 116), no. 49.

¹¹⁹ B.R. Hartley, in Frere, op. cit. (note 30), 216–62, D23.

¹²⁰ D. Atkinson, Report on Excavations at Wroxeter (1942), pl. 68, 52A.

¹²¹ Atkinson, op. cit. (note 116), nos 4 and 19.

¹²² J.P. Bushe-Fox, Fourth Report on the Excavations of the Roman Fort at Richborough (1949), pl. LXXIX, 39.

¹²³ Museum of London, 5880G.

¹²⁴ Hartley, op. cit. (note 119), D73.

¹²⁵ Atkinson, op. cit. (note 116), no. 76.

¹²⁶ J. Jacobs, Jahrbuch für Allertumskunde (1913), nos 13 and 25.

¹²⁷ J.P. Bushe-Fox, First Report on the Excavations of the Roman Fort at Richborough (1926), pl. xix, 2.

¹²⁸ Atkinson, op. cit. (note 116), nos 26–30.

¹²⁹ Terrisse, J.-R., Lés Ceramiques sigillées gallo-romaines des Martres-de-Veyre (Puy-de-Dôme), Gallia suppl. xix (1968), pl. XLIV, 10154.Google Scholar

¹³⁰ ibid., 804.

¹³¹ ibid., PL. XXVII 706 and PL. XXX, 1018, respectively.

¹³² Saalburg-Jahrbuch xxvii (1970), 25, no. 8.Google Scholar

¹³³ Dickinson, B.M. in Buckland, P.C. and Magilton, J.R., The Archaeology of Doncaster, BAR 148 (1986), 141, 33 and 133, 29, respectively.Google Scholar

¹³⁴ Saalburg-Jahrbuch xxvii (1970), 25, no. 10.Google Scholar

¹³⁵ Patemus II of J.A. Stanfield and Grace Simpson, Les Potiers de la Gaule Centrale (1990), 235.

¹³⁶ Atkinson, op. cit. (note 116).

¹³⁷ This ‘hiatus’ corresponds to the period of the earliest occupation of the stone buildings (i.e. before Phase IIIb), when there is little or no deposition on site (J.D.Z.).

¹³⁸ Again local circumstances may apply, there being no deposits on site referable to the interval between Phase IVa and the demolition, Phase V (J.D.Z.).

¹³⁹ B.R. Hartley in S. Frere and J.J. Wilkes, Strageath: Excavations within the Roman Fort 1973–86 (1989), 218.

¹⁴⁰ Boon, G.C., Antiq. Journ. xlvii, part 1 (1967), 39–42.Google Scholar

¹⁴¹ J.D. Zienkiewicz, Journ. Roman Pottery Studies (forthcoming).

¹⁴² Anderson, A.C. in Anderson, A.C. and Anderson, A.S. (eds), Roman Pottery Research in Britain and North-West Europe, BAR int. ser. 123 (1981), 11Google Scholar, 332 and fig. 19.3, nos 19–20 for form; also Gallia xlvi (1989), 253–9Google Scholar for recent discussion of the industry.

¹⁴³ For discussion of the local industry, see Boon, G.C., Arch. Comb, xcv (1966), 45–456Google Scholar, and S. Greep in Zienkiewicz, op. cit. (note 9), 11, 55.

¹⁴⁴ J.P. Bushe-Fox, Excavations on the Site of the Roman Town at Wroxeter, Shropshire, in 1912 (1913), fig. 19. I am grateful to Mis K. Hartley for the identification of non-local mortaria and the comments as to date which are reported here.

¹⁴⁵ Zienkiewicz, op. cit. (note 9), 11, fig. 26.

¹⁴⁶ Young, C., Oxfordshire Roman Pottery, BAR 43 (1977).Google Scholar

¹⁴⁷ Zienkiewicz, J.D., Monmouthshire Antiquary viii (1992), 1–9.Google Scholar I am grateful to Mr G.C. Boon and Dr Jennifer Price for helpful discussion and comment.

¹⁴⁸ D.B. Harden, Glass of the Caesars (1987), 105–8 for discussion, and pp. 130–1, nos 59–60 for similar trail terminals in colourless glass.

¹⁴⁹ D. Allen in Zienkiewicz, op. cit. (note 9), 11, 108–9 and fig. 43, no. 59.

¹⁵⁰ Isings, C., Roman Glass from Dated Finds, Archaeologia Traiectina 11 (1957), 103.Google Scholar Compare e.g. Fremersdorf, F., Römische Gläser mit Fadenauflage in Köln, Die Denkmäler des römischen Köln v (1959), Taf. 15Google Scholar; and see J. Price in M.G. Jarrett and S. Wrathmell, Whitton: An Iron Age and Roman Farmstead in South Glamorgan (1981), fig. 66 and pp. 154–5, no. 8, for a vessel of similar form and references to other British occurrences.

¹⁵¹ Isings, op. cit. (note 150), 102–3.

¹⁵² e.g. Charlesworth, D. in Frere, S.S., Verulamium Excavations III (1984), 156–7Google Scholar; and Harden, D.B. and Price, J. in Cunliffe, B., Excavations at Fishbourne 1961-1969 II (1971), 352.Google Scholar

¹⁵³ Allen, op. cit. (note 149), In and fig. 43, nos 68–74.

¹⁵⁴ Harden, D.B., Proc. Camb. Antiq. Soc. li (1958), 12Google Scholar, fig. 6 and pl. IIIB; pp. 14-5 for discussion of the type.

¹⁵⁵ Boon, G.C., Journ. Glass Studies viii (1966), 41–5Google Scholar; Harden, D.B., Antiq. Journ. liv (1974), 280–1.Google Scholar

¹⁵⁶ Window-glass was employed at the Fortress Baths from their foundation, c. A.D. 75 (Zienkiewicz, op. cit. (note 9), 1, 337), but it occurs only very sparsely in ‘timber’ levels elsewhere at the fortress.

¹⁵⁷ Boon, op. cit. (note 155), 45, for other panes in decolourised glass from Caerleon. Re-examination of the large quantity of glass from the Fortress Baths now confirms the distinction between natural blue-green and these colourless panes which are always much thinner (1.8–2.0 mm); a chronological factor may apply, with colourless panes occurring there only from c. A.D. 150–60, doubtless as a reflection of the quality of cullet available, for panes can rarely have been made from primary ingredients.

¹⁵⁸ Arch. Comb, xcix (1970), 18Google Scholar and 57, fig. 17, no. 13; probably second century.

¹⁵⁹ Unpublished, in context dated c. A.D. 200–300.

¹⁶⁰ e.g. Fishbourne Harden and Price, op. cit. (note 152), 221; Portishead, , JRS xlvii (1957), 221.Google Scholar

¹⁶¹ J. Price, pers. comm.

¹⁶² Antiq. Joum. lxiv (1984), 403–7Google Scholar, for detailed discussion by G.C. Boon.

¹⁶³ ibid., 405.

¹⁶⁴ A. Radnoti, Die römischen Bronzegefässe von Pannonien (1938), Taf. IV.18.

¹⁶⁵ H. Norling-Christensen, Aarbogen (1952), 182, fig. 13 (Valkenburg) and figs 18–19 (Avenches, Villeneuve), for this Gaulish stamp on similar handles. I am grateful to Mr G.C. Boon for this reference.

¹⁶⁶ cf. R. Robinson, The Armour of Imperial Rome (1975), 153ff.

¹⁶⁷ Archaeologia lxxviii (1928), 166Google Scholar, fig. 14, no. 29.

¹⁶⁸ As Robinson, op. cit. (note 166), fig. 75 etc.

¹⁶⁹ ibid., 181 for function; Arch. Comb, cxviii (1969), 109Google Scholar, fig. 9, no. 9, for an identical Caerleon specimen.

¹⁷⁰ Arch. Camb. cxiii (1964), 20Google Scholar, fig. 6, no. I, for a similar, though more highly decorated, example from Caerleon; and for closer parallel, seeBishop, M.C., ‘Cavalry equipment of the Roman army in the first century A.D.’ in Coulston, J.C. (ed.), Military Equipment and the Identity of Roman Soldiers, BAR int. ser. 394 (1988), 67–196Google Scholar, at fig. 50, particularly the examples from Gloucester and Sea Mills.

¹⁷¹ As e.g. the complete harness-set from Xanten; Britannia xvi (1985), 141–64Google Scholar, compare figs 7–8.

¹⁷² Compare Newstead specimens; J. Curle, A Roman Frontier Post and its People: The Fort at Newstead (1911), pl. 73.

¹⁷³ Bishop, M.C., ‘The evolution of certain features’, in Dawson, M. (ed.), Roman Military Equipment: The Accoutrements of War, BAR int. ser. 336 (1987), 109–40, at 118.Google Scholar

¹⁷⁴ Cirencester: J. Wacher and A. McWhirr, Early Roman Occupation at Cirencester (1982), fig. 35, no. 98; Doorweth: Bishop, op. cit. (note 170), table 6, Type ic; Strageath: Frere and Wilkes, op. cit. (note 139), 145, fig. 73, no. 40.

¹⁷⁵ e.g. at the Fortress Baths; R.J. Brewer in Zienkiewicz, op. cit. (note 9), 11, fig. 56, nos 19–20.

¹⁷⁶ The terminology follows E.T. Leeds, Celtic Ornament in the British Isles (1933).

¹⁷⁷ For a discussion of enamel colours, see Bateson, J.D., Enamelworking in Iron Age, Roman and Sub-Roman Britain: The Products and Techniques, BAR 93 (1981), 76.Google Scholar

¹⁷⁸ As Künzl, E., ‘Medizinische Instrumente aus Sepulkralfunden der römischen Kaiserzeit’, Banner Jahrbücher clxxxii (1982), 1–132, at Abb.30, no. 30.Google Scholar

¹⁷⁹ idem, ‘Operationsräume in römischen Thermen’, Bonner Jahrbücher clxxxvi (1986), 491-510, at 493.

¹⁸⁰ For discussion of knives with decorated handles, see W.H. Manning, Catalogue of the Iron Tools, Fittings and Weapons in the British Museum (1985), 108f.

¹⁸¹ I am grateful to Mr G.C. Boon for this suggestion; see Boon, G.C., Britannia xxii (1991), 21–32.CrossRefGoogle Scholar

¹⁸² Bateson, op. cit. (note 177), 57, for discussion of enamelled designs. Chequered patterns are found on other handles e.g. at Caerleon (Jenkins' Field; Arch. Camb. lxxxiv (1929), 255Google Scholar, fig. 16, no. 4), Carlisle (Tullie House Museum; L. Allason-Jones, ‘Small-finds from turrets on Hadrian's Wall’, in Coulston, op. cit. (note 170), 197–233, at fig. 4, no. 49a), and Chichester (Down, A. and Rule, M., Chichester I (1971)Google Scholar, fig. 3.16, no. 8); but the curvilinear elements do not occur.

¹⁸³ As e.g. Curle, op. cit. (note 172), 337, pl. 92, no. 22.

¹⁸⁴ For a similar tie-hook from the Corbridgehoard, see L. Allason-Jones and M.C. Bishop, Excavations at Corbridge: The Hoard (1988), fig. 94, no. 232; cf. Robinson, op. cit. (note 166), fig. 183.

¹⁸⁵ Lee, J.E., Isca Silurum (1862), pl. xxxi, no. 20.Google Scholar

¹⁸⁶ See Bishop, op. cit. (note 170), fig. 48, Type 9.

¹⁸⁷ For a similar Caerleon example, rather more rectangular in form, see J.E. Lee, Delineations of Roman Antiquities found at Caerleon (1845), pl. xvi, nos 4-5.

¹⁸⁸ Brewer (op. cit. (note 175), 183, no. 134), in discussing a similar group from the Fortress Baths, also Hadrianic-early Antonine, concludes that such might represent a hanger from the protective apron of armour, formed not of leather straps but of chains decorated with studs and pendants.

¹⁸⁹ Inspection shows that chain links from the Fortress Baths (Brewer, op. cit. (note 175), fig. 61, no. 146) and the Caerleon amphitheatre (Archaeologia lxxviii (1928), pl. XXXIIGoogle Scholar. 2, no. 13) were also formed in this way.

¹⁹⁰ Compare elements of the Xanten harness; Britannia xvi (1985), 141–64, at figs 12–14.Google Scholar

¹⁹¹ For an identical specimen from London, see Guildhall Museum Cat. (1908), pl. xxiv, no. 4.

¹⁹² H. Chapman in L. Miller et al. The Roman Quay at St Magnus House, London (1986), 235, fig. 14, no. 2; citing other London examples.

¹⁹³ A. Gansser-Burckhardt, Das Leder und seine Verarbeitung im römischen Legionslager Vindonissa (1942), 18, Abb.8.

¹⁹⁴ As suggested by Chapman, op. cit. (note 192).

¹⁹⁵ As one from Verulamium, M. Henig, Religion in Roman Britain (1984), 60f.; for two ram figurines from York, see Green, M., Small Cult-Objects from Military Areas of Roman Britain, BAR 52 (1978), 26.Google Scholar

¹⁹⁶ Examples from Pompeii (Ward-Perkins, J. and Claridge, A., Pompeii A.D. 79 (1976)Google Scholar, cat. no. 249) and London (R.E.M. Wheeler, London in Roman Times (1930), fig. 22.1).

¹⁹⁷ As one from the Fortress Baths; Brewer, op. cit. (note 175), 186, fig. 61, no. 150.

¹⁹⁸ As e.g. Wheeler, op. cit. (note 196), figs 22–3.

¹⁹⁹ Arch. Comb, lxxxvii (1932), fig. 36, nos 4f.Google Scholar

²⁰⁰ As Oldenstein, J. ‘Zur Ausrüstung römischer Auxiliareinheiten’, Bericht der Röm.-Germanischen Kommission lvii (1976), 49–284, at Taf. 41.Google Scholar

²⁰¹ For a number firom the Fortress Baths, in context dated c. A.D. 160–230, see Brewer, op. cit. (note 175), fig. 57, nos 49–52.

²⁰² But see Allason-Jones, L., ‘Bell-shaped Studs’, in Bishop, M.C. (ed.), The Production and Distribution of Roman Military Equipment, BAR int. ser. 275 (1985), 95–108.Google Scholar

²⁰³ Fortress Baths, with iron shank (Brewer. -op. cit. (note 175), fig. 62, no. 172); and a matching pair, with integral shanks, from the Roman Gates site, J. Webster in Evans and Metcalf, op. cit. (note 6), 135–6, nos 141–2.

²⁰⁴ As e.g. Brewer, op. cit. (note 175), fig. 61, no. 142.

²⁰⁵ Arch. Camb. lxxxvii (1932), fig. 36, nos 2–9.Google Scholar

²⁰⁶ For a comparable specimen from Faimingen, see Oldenstein, op. cit. (note 200), Taf. 45,444.

²⁰⁷ Manning, op. cit. (note 180), 120–2 and fig. 30.

²⁰⁸ W.H. Manning in A. Gailey and A. Fenton (eds), The Spade in Northern and Atlantic Europe (1970), 25–6.

²⁰⁹ Gansser-Burckhardt, op. cit. (note 193), 17–19; and Manning, op. cit. (note 180), 39, citing examples.

²¹⁰ Newstead: Curle, op. cit. (note 172), pl. LXI, no. 3 (blade length 115 mm); Colchester: N. Crummy, Roman Small Finds from Excavations in Colchester 1971–9 (1983), 136 and fig. 155, no. 4232 (98 mm).

²¹¹ L. Allason-Jones and R.F. Miket, The Catalogue of Small Finds from South Shields Roman Fort (1984), 288, ironwork no. 5.11 (blade 166 mm).

²¹² Manning, op. cit. (note 180), 104 and pl. 50, no. 3 4 for a ladle of typlcal form.

²¹³ For a flanged iron bowl, possibly of industrial use, from Verulamium see W.H. Manning in Frere, op. cit. (note 30), 176 and fig. 65, no. 47.

²¹⁴ Zienkiewicz, op. cit. (note 9), 1,318 and fig. 104 (95–107 mm).

²¹⁵ Manning, op. cit. (note 180), 139.

²¹⁶ ibid., 139 and pl. 64,59.

²¹⁷ Robinson, op. cit. (note 166), 171ff. for mail armour generally.

²¹⁸ Arch. Comb, lxxxvii (1932), 68 and fig. 16.Google Scholar

²¹⁹ e.g. from the Prysg Field; ibid., fig. 39, nos 5 and 6 (the latter in Flavian context).

²²⁰ Allason-Jones and Bishop, op. cit. (note 184), 83, no. 287 and cover illustration. Similar objects are noted in A. MacGregor, Bone, Antler, Ivory and Horn (1985), 179. For others, from the ‘Roman Gates’site, Caerleon, see S. Greep in Evans and Metcalf, op. cit. (note 6), 189–90.

²²¹ e.g. Crummy, op. cit. (note 210), 134, no. 4227, for a Colchester find. Greep (in M.C. Bishop (ed.), Roman Military Equipment: Proceedings of a seminar… at the University of Sheffield, 21st March 1983 (1983), 16–21) has drawn together a number of examples, and demonstrates that cattle metapodia were almost invariably employed. For an example from the ‘Roman Gates’ site (in Hadrianic-early Antonine context) see S. Greep in Evans and Metcalf, op. cit. (note 6), 188–9.

²²² Such ornaments have been discussed by Boon, G.C. (Arch. Camb. cxxiv (1975), 62–4Google Scholar) with respect to a specimen from Segontium; the present example is from a Hadrianic context.

²²³ As, for example, is leather from a contemporary well-filling at Caernarvon; Arch. Camb. cxxxiv (1985), 88–101.Google Scholar

²²⁴ e.g. J.W. Waterer in D. Strong and D. Brown (eds), Roman Crafts (1976), 179–94; W. Groenman-van Waateringe, Romeins Lederwerk uit Valkenburg Z.H., Nederlandse Oudheden II (1967), 207; and Boon, G.C., Arch. Camb. cxxiv (1975), 60.Google Scholar

²²⁵ e.g. from Caernarvon (Arch. Camb. cxxxiv (1985), 91–3)Google Scholar; London (Britannia xviii (1987), 173–82); and Vindonissa (Gansser-Burkhardt, op. cit. (note 193), 99f.).Google Scholar

²²⁶ e.g. Boon, G.C., Arch. Camb. cxxiv (1975), 52–67, at fig. 5.Google Scholar

²²⁷ Circular applied patches, slit to take thongs, were employed for the attachment of tie-or toggle-loops, and the observed pulling of stitch-holes here confirms the general identification; compare examples from Newstead (Curle, op. cit. (note 172), pl. xix, nos 1,6–8, 10 and 13), and the Saalburg (Saalburg Jahrbuch xxii (1965), 158–210Google Scholar, at Taf. 27, no. 608; patch at hem, as here).

²²⁸ Groenman-van Waateringe, op. cit. (note 224), fig. 5, nos Ia, 2 and 3.

²²⁹ G.C. Boon (ed.), Cambrian Archaeological Association, Monographs and Collections: I – Roman Sites (1978), 4, fig. 10.6 and p. 22, note 20.

²³⁰ Arch. Camb. lxxxvii (1932), 104 and fig. 45.Google Scholar

²³¹ Now at the Newport Museum; C. van Driel-Murray inCoulston, op. cit. (note 170), 51–66.

²³² The evidence of shield-covers from the Valkenburg suggests that the largest useful pieces of goat leather generally available were 650-750 mm across (Groenman-van Waateringe, op. cit. (note 224), 209), nor do the complete panels of tents often exceed that range (ibid., 79–105 and 210).

²³³ For similar pleces from Zugmantel, see Saalburg Jahrbuch xxii (1965), 158–210, at Taf. 40, no. 888.Google Scholar

²³⁴ Britannia xvii (1986), 450–2 and pl. xxxiv.Google Scholar

²³⁵ For the format of such tablets, see A.K. Bowman and J.D. Thomas, Vindolanda: The Latin Writing Tablets (1983), 44–5, fig. 9 and pl. XIII.

²³⁶ See discussion by Lawson, A., Archaeologia cv (1976), 263–5Google Scholar; and by Biddle, M., Antiq. Journ. xlvii (1967), 205.Google Scholar All may originate at a single workshop, see N. Sunter and P.J. Woodward, Romano-British Industries in Purbeck (1987), 32–4.

²³⁷ For shale bracelets from the Fortress Baths, also Flavian-Trajanic, see Zienkiewicz, op. cit (note 9), II, fig. 77, nos 1–18.

²³⁸ For production, e.g. at Norden, see Sunter and Woodward, op. cit. (note 236), 36–9.

²³⁹ cf. G.C. Dunning in B.W. Cunliffe (ed.), Fifth Report on the Excavations of the Roman Fort at Richborough, Kent (1968), pl. LXVI, no. 1, and 110–14 for discussion of the type; Crummy, op. cit. (note 210), 76 and fig. 79, no. 2084 for a Colchester specimen.

²⁴⁰ Arch. Camb. lxxxvii (1932), 93 and fig. 14, no. 5.Google Scholar

²⁴¹ For contemporary examples of similar size, see B.W. Cunliffe, Excavations at Fishbourne, 1961-9, II: The Finds (1971), fig. 71, nos 4–7.

²⁴² e.g. from Colchester, Crummy, op. cit. (note 210), no. 2054.

²⁴³ For a slate palette from the Fortress Baths, and refs to other Caerleon specimens, see Zienkiewicz, op. cit. (note 9), 11,214, no. 34.

²⁴⁴ These results are incorporated here in summary form, with permission.

²⁴⁵ Bayley, J. in Niblett, R., Sheepen: An Early Roman Industrial Site at Camulodunum, CBA Res. Rep. 57 (1985), 115.Google Scholar

²⁴⁶ idem in E.A. Slater and J.O. Tate (eds), Science and Archaeology, Glasgow 1987, BAR 196 (1988), 193–208.

²⁴⁷ The form, demonstrably used for silver, occurs at St Sepulchre Gate, Doncaster (idem in P.C. Buckland and J.R. Magilton, The Archaeology of Doncaster: I, BAR 148 (1986), 196 and 199), and at Walton-le-Dale (idem, A.M. Lab. Report 93/88).

²⁴⁸ Zienkiewicz, op. cit. (note 9), 11, 252–3 for crucibles from the Fortress Baths, of which one was for gold; and Frere, op. cit. (note 30), 365–6 and fig. 141 for gold-working crucibles from Verulamium (dated c. A.D. 60-70).

²⁴⁹ Bayley, op. cit. (note 245).

²⁵⁰ Zienkiewicz, op. cit. (note 9), 1,325.

²⁵¹ ibid., 1,327 and fig. 108, nos 4–7; occurring there earliest in Hadrianic or Antonine context.

²⁵² Boon, op. cit. (note 35), 3-4, 10.

²⁵³ Boon (ibid., 5–7, 10) cites unpublished examples from ‘timber’ levels at the ‘Roman Gates’ site; and a fine, complete specimen (also unpublished) now comes from levels associated with the dismantlement of primary timber buildings at the Telecom site.

²⁵⁴ The type is of the commonest occurrence at Caerleon, and Boon, op. cit. (note 35), tentatively suggests a second-century date; but it appears that antefixa of this die had decorated the Fortress Baths from their foundation in the later 70s (Zienkiewicz, op. cit. (note 9), 1,333).

²⁵⁵ Boon, op. cit. (note 35), 13ff.; for Type A.ii.8, also additional to that series, see P. Lennox in D.M. Evans and V.M. Metcalf, op. cit. (note 6), 86.

²⁵⁶ With evidence from elsewhere at the fortress, this supports the view that Types A.i and A.ii were the earliest of the series (Boon, op. cit. (note 35)); there is nothing here or elsewhere to allow their relative dates of use to be established and it may be (contra Boon) that they denote the products of two workshops operating contemporaneously to supply (complementary varieties of) brick and tile for the general rebuilding.

²⁵⁷ ibid., 26.

²⁵⁸ ibid., 26.

²⁵⁹ ibid., 34.

²⁶⁰ And may be compared, e.g., with Nash-Williams's list for the Prysg Field barracks; Arch. Camb. lxxxvii (1932), 53–8.Google Scholar

²⁶¹ Identified by T. Tipper, Department of Botany, National Museum of Wales.

²⁶² Bone from subsequent phases is not reported here, but is retained for future study. Measurements follow Driesch, A. von den, A Guide to the Measurement of Animal Bone from Archaeological Sites, Peabody Museum Bulletin 1 (1976)Google Scholar; withers heights are based on idem and J. Boessneck, Kritische Anmerkungen zur Widerristhböhenberechnung aus Längenmassen vor-und frühgeschichtlicher Tierknochen, Säugetierkundliche Mitteilungen 22 (1974).

²⁶³ Mostly cow, but may include horse and red deer.

²⁶⁴ Probably mostly sheep/goat, roe, and pig.

²⁶⁵ Includes: eagle (probably white-tailed, Haliaeetus albicilla), woodcock (Scolopax rusticola), pigeon (probably Columba livia), small duck (probably teal, Anas crecca), wigeon (Anas penelope), and an unidentified small passerine.

²⁶⁶ Not identified to species.

²⁶⁷ Maltby, J.M., The Animal Bones from Exeter, 1971-1975, Exeter Archaeological Reports 2 (1979).Google Scholar

²⁶⁸ T.P. O'Connor in Zienkiewicz, op. cit. (note 9), 11, 225–48.

²⁶⁹ J.M. Maltby, Ancient Monuments Laboratory Report 9/90 (unpublished).

²⁷⁰ idem in D. Searjeantson and T. Waldron (eds), Diet and Crafts in Towns, BAR 199 (1989), 75–106.

²⁷¹ T.P. O'Connor in P.V. Addyman, The Archaeology of York; Vol. 15: The Animal Bones, fasc. 2 (1988).

²⁷² Noddle, B.A. in Bell, M. and Limbrey, S. (eds), Archaeological Aspects of Woodland Ecology, BAR int. ser. 176 (1982), 315–53.Google Scholar

²⁷³ And compare abundant fowl bones from the Fortress Baths, O'Connor, op. cit. (note 268).

²⁷⁴ G.S. Cowles kindly allowed access to the collections at Tring, and advised on cranes.

²⁷⁵ Harrison, C.J.O. and Cowles, G.S., Journ. Arch. Sci. iv (1977), 25–7.CrossRefGoogle Scholar

²⁷⁶ Apicius, , De Re Coquinaria vi. 2, passim; cf. Flower, B. and Rosenbaum, E., The Roman Cookery Book (1958), 140–5, passim.Google Scholar

²⁷⁷ J.M. Maltby in P.J. Fasham (ed.), The Prehistoric Settlement at Winnal Down, Winchester (1985), 97–112.

²⁷⁸ idem, Ancient Monuments Laboratory Report 6/87 (unpublished).

²⁷⁹ St David's University College, Lampeter.

²⁸⁰ Tattersall, W.M., Glamorgan County History 1 (1936), 128–9, 243–6.Google Scholar

²⁸¹ Jones, M. in Jones, M. and Dimbleby, G. (eds), The Environment of Man: The Iron Age to the Anglo-Saxon Period, BAR 87 (1981), 95–127Google Scholar, for the occurrence of such damp ground species (particularly the spike rush, Eleocharis palustris) as impurities in carbonised grain assemblages from the Upper Thames Valley.

²⁸² D.G. Wilson in M.G. Jarrett and S. Wrathmell (eds), Whitton: An Iron Age and Roman Farmstead in South Glamorgan (1981), 240–3.

²⁸³ Dates are otherwise unreported from Roman contexts in Britain, see Davies, R. in Britannia ii (1971), 122–42.CrossRefGoogle Scholar

²⁸⁴ Compare e.g. evidence for horse dung from Roman Leicester; Wilson, D.G., Archaeol. Physica. viii (1979), 331–50.Google Scholar

²⁸⁵ Hall, A.R. et al. Environmental Evidence from Roman Deposits in Skeldergate, The Archaeology of York 14 (1980), 3.Google Scholar

²⁸⁶ For rushes etc. in a demolition deposit at Bearsden, and furtherrefs, see D. Breeze in idem (ed.), Studies in Scottish Antiquity (1984), 37.

²⁸⁷ Allen, J.R.L. and Fulford, M.G., Britannia xvii (1986), 91–117.CrossRefGoogle Scholar

²⁸⁸ Compare, e.g., walling 1.04 m thick at the Fortress Baths, where a vaulted superstructure some 15 m high is indicated.

²⁸⁹ Lambaesis (built or restored in A.D. 268): M.R. Cagnat, Les Deux camps de la Légion de Lambése (1908), 232ff. and idem, Armée romaine d'Afrique (1912), 466–70; Aquincum (Severan): Studien zu den Militärgrenzen Roms III (1986), 398–9Google Scholar. Similar structures were provided also at Lauriacum, Dura Europos (Severan), and (Diocletianic) at El Lejjun (Parker, S.T. in Vetters, H. and Kandler, M. (eds), Akten des 14. lnternationalen Limeskongresses 1986 (1990), I, 216–7).Google Scholar

²⁹⁰ Arch. Comb, cxix (1970), 10–63Google Scholar, for the basilica principiorum. Mr G.C. Boon tells me that he is no longer satisfied with his late dating (ibid.) for the stone principia. Trial excavations (1992) in the garden of the new Legionary Museum have uncovered piers of an arch at the via praetoria frontage of the groma-structure which is now shown to be cruciform in plan.