¹ RCAHM Glamorgan vol. 1 part 2, 120Google Scholar; JRS li (1961), 158.Google Scholar

² Information from Messrs S. A. Luen and Jack Stephens.

³ Glamorgan County History vol. 1, 129; 245.Google Scholar

⁴ Bedford Record Office GY 10 ‘Survey of the manor of Barry in the county of Glamorgan by Evans Mouse 1622’, 81. The accompanying estate map is deposited in Glamorgan Record Office.

⁵ Glamorgan Record Office D/DF vol. 40 ‘Penmark, Fonmon and Barry manors Court Books 1569–1615’.

⁶ Ordnance Survey, Field Archaeology (4th edn., 1966), 5, 9.Google Scholar

⁷ Rodwell, W., The Archaeology of the English Church (London, 1981), 116.Google Scholar

⁸ cf. Davy, N. and Ling, R., Wall Painting in Roman Britain Britannia Monographs iii (1981), 54.Google Scholar

⁹ A sample was taken of this material but it was unfortunately destroyed by fire before any analysis could determine how it was related to the mortars.

¹⁰ It is difficult to be sure which tile deposits show signs of this activity as much of the tile was discarded before it had been examined for reworking, but likely contexts are 167, 168, 194, 203, 268.

¹¹ We are indebted to P. M. Barford for some of the suggestions in this section.

¹² See Fowler, D., ‘Temporary works, their construction and archaeology’ in Drury, P. J., Structural Reconstruction BAR 110(1982), 125.Google Scholar

¹³ Iolo Morgangwg MSS in National Library of Wales MS 13, 117E p. 52 under ‘Glamorgan Topography’ names Bullcliff together with St. Athan, Ysgwrgar and Newton as noted oyster beds. A newspaper article in Barry Dock News November 1st 1912 by D. T. Alexander mentions a fleet of ships from Oystermouth fishing for oysters in the 1880's in Porthkerry Bay.

¹⁴ See Williams, J. H., ‘Roman Building Materials in the South West’ Trans. Bristol Gloucs. Arch. Soc. xc (1971), 95–119.Google Scholar

¹⁵ Kindly identified by Dr S. Shackly of the Department of Oceanography, University College of Swansea.

¹⁶ Mr S. Luen of Barry has described how, as a boy c. 1900, he watched labourers descend the rock flats when the tide was out, break off the stone with iron bars and wedges and pile it in a great cairn with a pole stuck in the top as a sea marker. A sloop would come in with the tide and drop anchor; the assembled stones would be thrown in the hold at low tide and the sloop be floated off with the next high tide. See also Glam. Record Office, Fonmon MSS₁/183 ‘Lease of limestones and building stones 1774’, and NLW MS 13 115B pp. 312–14.

¹⁷ Blagg, T. F. C., ‘Tools and Techniques of the Roman Stonemason in Britain’ Britannia vii (1976), 156.Google Scholar

¹⁸ Dr R. M. Owens of the National Museum of Wales suggests the Bath area as a possible source.

¹⁹ Information from Dr R. M. Owens.

²⁰ A fragment of ancient woodland still survives in Birth Grove Wood, less than 0·5 km from the site.

²¹ Information from the masons of S. Glamorgan Building Maintenance Section employed on the reconsolidation of the building.

²² Thresholds are rare in Britain, but such arrangements are well-known in places with better preserved Roman remains e.g. Pompeii.

²³ Young, C. J., ‘The Processing of Roman Tile’ in McWhirr, A. D. (ed.), Roman Brick and Tile BAR S. 68 (1979), 402.Google Scholar

²⁴ Webster, G., ‘Tiles as a Structural Component in Buildings’ in McWhirr, op. cit. (note 23), 287–9.Google Scholar

²⁵ The only rooms for which a hypocaust cannot actually be disproved are M, T and U.

²⁶ The only type which was not recorded was half box-tile, but this is probably because these tiles were originally thought to be tegulae.

²⁷ cf. Boon, G. C.Isca (Cardiff, 1970), 68.Google Scholar

²⁸ During experiments in working tiles to shape carried out by E. M. Evans, J. M. Daly and M. Sullivan this was found to be the most effective way of breaking Roman tile along a predetermined line.

²⁹ Blake, M. E., Ancient Roman Construction in Italy from the Prehistoric period to Augustus (Washington, 1947), 302–3 gives chipping, with sawing as one of the usual methods of shaping tiles.Google Scholar

³⁰ The association of pebbles of a convenient size and shape to hold in the hand with broken/reworked tile in 196, Room J, suggests that such pebbles might have been used for breaking or abrading tile.

³¹ Where lengths are given in feet, pedes Monetales are used.

³² For the use of proportion as the basis of plan design see Vitruvius esp. III, 1; V introduction and VI, 3, 3–6.

³³ cf. Fishbourne, Cunliffe, B.Excavations at Fishbourne, vol. i (London, 1971), 80.Google Scholar

³⁴ The lack of repointing to walls where the mortar had a life-expectancy of 20 years at most (pers. com. D. Evans, Ancient Monuments Branch Welsh Office) would lead to the conclusion that at best the building had only a short life-span.

³⁵ Or 2·17 m above the cellar floor. It is perhaps worth noting in this context that the level of the Deep Room at Lullingstone was about 2·4 m lower than the neighbouring room with the Bellerophon mosaic (Arch. Cant. (1952)Google Scholar, fig. 18). At Verulamium the cellar in XXII 2 is 2·9 m below ground level Ant. Journ. xxxvi (1956), 2.Google Scholar

³⁶ Lowther, A. W. G., Antiq. Journ. xvii (1937), 34 and pl. xii, fig. 1, pl. xxiii, fig. 2.Google Scholar

³⁷ Blagg, T. F. C., ‘Architectural patronage in the western provinces of the Roman Empire in the third century’ in King, A. and Henig, M. (eds.), The Roman West in the Third Century BAR 110 (1982), 175, 177.Google Scholar

³⁸ This is the only arrangement which falls within the parameters of the British material. See Blagg, T. F. C. ‘Reconstruction of Roman decorated architecture’ in Drury, P. J. (ed.), op. cit. (note 12), 136 f.Google Scholar

³⁹ For the purpose of the reconstruction drawing (FIG. 6), the column diameter has been taken as 0·375 m (with 3·75 mm/I½ʺ layer of rendering) and the height as 7 diameters of the full-length columns. The miniature columns (in FIG. 8) have been shown with a height of 4 · 5 diameters loc. cit., 140–2.

⁴⁰ Traditional post-medieval rubble construction in this area allows 2ʹ for load-bearing walls and 1½ʹ for non-load-bearing walls (pers. com. A. Tomlinson).

⁴¹ The Dark Age occupation is discussed in greater detail in the Glamorgan County History vol. II (Cardiff, 1984), 344.Google Scholar

⁴² Barry Castle ST 10086719 – tiles and large blocks of tufa found during demolition in 1956 of the late thirteenth-century E range. Barry Village House B ST 10306720, House C 10276722, House D 10276722, all late thirteenth/early fourteenth century – fragments of flat tile. St Nicholas' church and priest's house ST 10456697 thirteenth century – complete flat tiles.

⁴³ Rickman, G., Roman Granaries and Store Buildings (Cambridge, 1971). The only warehouse with rooms of similar proportions to the Cold Knap building is the Horrea Epagathiana et Epaphroditiana at Ostia (op. cit., 30–8; esp. 37). The doorways are rather narrow for a warehouse but comparable examples are known op. cit., 79.Google Scholar

⁴⁴ Drury, P. J., ‘Form, function and the interpretation of the excavated plans of some large secular Roman-British buildings’ in Drury, (ed.), op. cit. (note 12), 289.Google Scholar

⁴⁵ cf. the N range of the Godmanchester mansio loc. cit.

⁴⁶ The size is similar to the watchtower on Holyhead Mountain (Arch. Wales xxi (1981), 36–7), although the walls are not as massive.Google Scholar

⁴⁷ This mortar is not known from any other site in the area. The tile fabric was one of those known from Cold Knap (Fabric 1).

⁴⁸ cf. Richborough

⁴⁹ Moore, D., ‘Maritime aspects of Roman Wales’, Studien zu den Militargrenzen Roms ii (1977), 34–7.Google Scholar

⁵⁰ Nash-Williams, V. E. and Jarrett, M. G., The Roman Frontier in Wales (2nd. edn., Cardiff, 1969), 70.Google Scholar

⁵¹ Glamorgan-Gwent Archaeological Trust, Annual Report 1983–1984, 113Google Scholar; Bull. Board Celtic Stud. xxii (1966–1968), 431.Google Scholar

⁵² RCAHM Glamorgan vol. i, part 2, 82.Google Scholar

⁵³ Bull. Board Celtic Stud. xxii (1966–1968), 291–300Google Scholar; Shiel, N., The Episode of Carausius and Allectus BAR 40 (1977), 204–5Google Scholar. Other military sites occupied during this period are Caerhun, Caernarvon, Gelligaer, Leintwardine (Nash-Williams and Jarrett, 59, 62, 91, 95) and now also Caerleon and Neath. (Glamorgan-Gwent Archaeological Trust Annual Report 1980–1981, 10 and 1983–84, 120Google Scholar). The fortifications at Holyhead, however, are probably not now to be dated to this period (Arch. Wales xxi (1981), 36–7).Google Scholar

⁵⁴ Even if the site did possess some sort of enclosure wall, the nature of the SW facade with its terrace on the sea suggests an essentially open site, unlike the massive fortifications at Cardiff. Cardiff Castle may however have been built of stone from the same source as Cold Knap: the stones making up the exposed section of wall are of roughly the same size and one or two have traces of pholodidae borings, while the consultant engineer during the rebuilding work for Lord Bute noted that the stones had ‘rounded and sea-worn edges’. Allen, S. W., Reminiscences (Cardiff, 1918), 182.Google Scholar

⁵⁵ Pierce, G. O., The Place-names of Dinas Powys Hundred (Cardiff, 1968), 198–200.Google Scholar

⁵⁶ Arch. Camb. (1888), 138–162; (1892), 167. Professor Rhys citing the Penmachno inscription in Caernarvonshire, and early literary sources, suggests that the name evolved thus: Cara-usius, Cereuus, Ceris, but admits that ‘the phonology of the change offers considerable difficulties.’Google Scholar

⁵⁷ Glamorgan Record Office, Fonmon MSS.

⁵⁸ Arch. Journ. cxxii (1965), 12.Google Scholar

⁵⁹ Williams, D. F. ‘The Romano-British Black-burnished ware industry: an essay on characterization by heavy mineral analysis’ in Peacock, D. (ed.), Pottery and early commerce (London, 1977), 163–220.Google Scholar

⁶⁰ Birley, E., ‘Excavations on Hadrian's Wall West of Newcastle upon Tyne in 1929’ Arch. Ael.⁴ vii (1930), 152–64.Google Scholar

⁶¹ Young, , op. cit., 128.Google Scholar

⁶² 1975 excavations by J. L. Davies, Feature 1.1. For a discussion of the fabric in a N Wales context cf. Gillam in Dinorben and Webster in G. Guilbert (forthcoming).

⁶³ Young op. cit.

⁶⁴ ibid.

⁶⁵ Fulford, M. G., New Forest Roman Pottery BAR 17 (1975).Google Scholar

⁶⁶ Boon, G. C., ‘“Legionary” ware at Caerleon’ Arch. Camb. cxv (1966), 45–66.Google Scholar

⁶⁷ Harden, D. B., Roman Glass from Karnis (1936), 65–6, 83 nos. 166–8, pl. xii.Google Scholar

⁶⁸ Boon, G. C., ‘Roman Glassware from Caerwent 1855–1925’ Monmouthshire Antiq. iii (1972–1973) pt. 2, 111–23Google Scholar; Harden, D. B., ‘The glass in the Roman villa at Ditchley, Oxon’ Oxoniensa i (1936), 62–4Google Scholar; Harden, D. B. and Price, J., ‘The glass’ in Cunliffe op. cit. (note 33), vol. ii 331–2 nos. 26 and 28 fig 138Google Scholar; Maxwell, G., ‘Objects of glass in Roman fort at Cramond Edinburgh, Excavations 1954–66’ Britannia v (1974), 198–9, nos. 6 and 7, fig. 16Google Scholar; Price, J., ‘Glass in Maryport, Cumbria’ Cumb. West. Ant. Arch. Soc. Extra ser. xxii (1976), 49–54Google Scholar; Townsend, P. and Hinton, P., Glass in 1–7 St. Thomas Street Southwark Excavations 1972–4 vol. ii (1978), 388–9 mo. 101 fig. 176Google Scholar; Charlesworth, D., ‘Glass vessels in Excavations at the New Market Hall Site, Gloucester 1966–7’ Trans Bristol Glos. Arch. Soc. xciii (1974), 74–6, nos. 4–5, fig. 29Google Scholar; Price, J., ‘Roman Glass from 1 Westgate Street’ Britannia xi (1980), 111 no. 1, fig. 17.Google Scholar

⁶⁹ Charlesworth, D., ‘A Roman cut glass plate from Wroxeter’ Ant. Journ. lv (1975), 414–6.Google Scholar

⁷⁰ Isings, C., Roman Glass from dated finds (1957), 63, form 50.Google Scholar

⁷¹ Guido, M., The glass beads of the prehistoric and Roman periods in Britain and Ireland (1978), 100, 230, fig. 35, nos. 21 and 22.Google Scholar

⁷² cf. Lawson, A., ‘Shale and jet objects from Silchester’ Archaeologia cv (1976), fig. 4, 21.Google Scholar

⁷³ Painter, K., ‘The British Museum collection of Roman head-stud brooches’ Brit. Mus. Quart. xxxiv (1970), 164–5 nos. 13–4 and fig. 3.Google Scholar

⁷⁴ Fowler, E., ‘The origins and development of the penannular brooch in Europe’ PPS xxvi (1960), 151, fig. 1, 152 and 176.Google Scholar

⁷⁵ Clarke, G., Pre-Roman and Roman Winchester. Part II. The Roman Cemetery at Lankhills Winchester Stud iii (1979), 303.Google Scholar

⁷⁶ idem., 302.

⁷⁷ For a classification of nails from Roman sites see Manning, W. H. in Frere, S. S., Verulamium Excavations i (London, 1972), 186.Google Scholar

⁷⁸ Manning, W. H., Catalogue of Romano-British ironwork in the Museum of Antiquities Newcastle upon Tyne (Newcastle, 1976), 71–5.Google Scholar

⁷⁹ Allason-Jones, L. and Miket, R., The catalogue of small finds from South Shields Roman Fort (1984), 5 · 99, also 10. 13.Google Scholar

⁸⁰ I am indebted to Mr G. Brodribb who read an earlier draft of this report and made many helpful suggestions.

⁸¹ cf. Darvill, T. C. ‘A petrological study of LHS and TPF tiles from the Cotswold Region’ in Roman Brick and Tile BAR 68 (1979), 309–49.Google Scholar

⁸² cf. Terry, R. D. and Chilingar, G. V., ‘Summary of “Concerning some additional aids in studying sedimentary formations” by M. S. Shvetsou’ Journ. Sedimentary Petrology xxv (1979), 229–34.Google Scholar

⁸³ Whereas the sizes of individual tiles are given in metric measurements, standard sizes of tile are referred to, as types, by measurements in Roman feet and inches where the Roman name is not known. Roman tiles were made to a series of standards, mostly relating to the foot or to multiples of it, though lack of control over shrinkage due to water loss means that the targets were rarely exactly achieved.

⁸⁴ McWhirr, A. and Viner, D., ‘The production and distribution of tiles in Roman Britain with particular reference to the Cirencester region’ Britannia ix (1978), 361 f.Google Scholar

⁸⁵ See Evans, E. M. and Stapenhorst, C., ‘Roman boxtile manufacture in SE Wales’ Glamorgan-Gwent Archaeological Trust Annual Report 1983–1984, 140–4.Google Scholar

⁸⁶ Kindly identified by Mr G. C. Boon. See Meiggs, R., Roman Ostia (2nd edn., Oxford, 1973), pl. xia.Google Scholar

⁸⁷ Driesch, A. von den, ‘A guide to the measurement of animal bones from archaeological sites’ Peabody Mus. Bull. i (1976).Google Scholar

⁸⁸ The writer is grateful to Mark Robinson (Oxford University Museum) for his help in the identification.