Roman architectural Ornament in Kent

ROMAN ARCHITECTURAL ORNAMENT IN KENT T.F.C. BLAGG, M.A., Ph.D., F.S.A. When Britain became part of the Roman Empire, it experienced the greatest architectural revolution it has known, except those of the Norman Romanesque and the present century. Lacking 'any indigenous architectural tradition of consequence',1 the first two generations of Britons after the Conquest saw their surroundings transformed, with the construction in towns, forts and the countryside of rectilinear buildings made of dressed stone and brick, with tile roofs, columns, arched openings and vaults. In addition to the introduction of new principles of planning and techniques of construction, there also came a highly evolved manner of decoration, internally with painted plaster walls and tessellated floors, and both internally and externally with ornamented stonework. Architectural decoration in stone differs from wall-plaster and mosaic in that it was usually part of the structure of the building. It, therefore, provides evidence for archaeological reconstructions of the buildings themselves. Also, unlike wall-plaster and mosaic, stonework can be re-used in later buildings: occasionally, for ornamental effect, but more usually, broken up to provide building stone, or in the case of limestone and marble, burned in lime-kilns to yield quick-lime for mortar. Much has, therefore, been lost, and what survives may well be found elsewhere than on the site of the building to which it originally belonged, particularly in towns. Local geological factors affect the extent of this survival. Most Roman architectural ornament in Kent was executed on oolitic limestone, which had to be imported. Less carving survives than in Roman towns like Cirencester and Bath, or such villas as Chedworth, where limestone is locally abundant and the motive for robbing stone from Roman buildings is consequently weaker; less, too, than in 1 G.C. Boon, Isca (Cardiff 1972), 75. 65 T.F.C. BLAGG Chester, Wroxeter and York, where most Roman stonework was executed on sandstone and gritstones and has thus escaped the lime-kiln. Nevertheless, what does survive in Kent is of particular interest, because of the range of sites and buildings from which it comes, some of which were decorated with unusual richness. The material to be discussed here includes Corinthian column capitals from public monuments in Canterbury, Richborough and Springhead; marble veneers from Canterbury and Richborough; and Tuscan capitals and column bases from a number of sites, including Dover, Eccles, Farningham and LuUingstone. A comprehensive catalogue must await another occasion.2 This paper will review the material in general, but will concentrate on those items which are most significant in relation to their archaeological context and to the architectural and economic history of Roman Britain. CANTERBURY Until the Canterbury Archaeological Trust began large-scale excavations in 1975, little Roman architectural stonework was known from the city. Although there are records of some earlier discoveries, for example the column bases and ornamented cornices found in 1861 in front of what is now no. 35 High Street,3 nothing found before 1976 has been preserved, nor adequately recorded, apart from some marble mouldings and veneers from Professor Frere's excavations in the yard of the former Fleur de Lis Hotel.4 Excavations of the Cakebread Robey sites between Castle Street and Stour Street and on the Marlowe car park sites since 1976 have now produced material of considerable importance, which decorated public buildings in the centre of the Roman town.5 Perhaps the most significant among these finds are five small pieces of Corinthian capitals from the Cakebread Robey sites. Although fragmentary, they preserve sufficient detail in their carving to demonstrate that they belong to a well-known Romano-British type which 2 In publication of the writer's Corpus of Roman Architecture in Britain. Records of unpublished material referred to below are contained in the archive of the Corpus at the University of Kent at Canterbury. See also, T.F.C. Blagg, Roman Architectural Ornament in Britain (University of London unpublished Ph.D thesis, 1981). 3 Arch. Cant., iv (1861), 35; VCH (Kent) hi, 68, no. 2. 4 Publication forthcoming in Canterbury Archaeological Trust, The Archaeology of Canterbury, viii. 'Arch. Cant., xcii (1976), 238-40; xciv (1978), 275-77; xcv (1979), 267-70; xcvi (1980), 402-10; xcvii (1981), 279-81. Full publication in Canterbury Archaeological Trust, The Archaeology of Canterbury, v and vi, in preparation. 66 ROMAN ARCHITECTURAL ORNAMENT 4 / •: % r^ i ^ c / <. r Fig. 1. Sketch Reconstruction of Corinthian Capital from Canterbury, showing the Position occupied by two of the Fragments (shaded). was introduced from north-east Gaul in the last third of the first century A.D.6 The closest parallels are with the capitals from the forum basilicas at Silchester and Caerwent.7 The main diagnostic feature is that the volutes (on an orthodox Corinthian capital, the spirals beneath the corners of the abacus, the topmost moulding) are wholly encased in foliage, with a leafy spray arching over the tips of the acanthus calyx from which the volutes normally rise (Fig. 1). The deep hollowing of the leaves and their spade-shaped tips are very close to the manner of carving used on the Silchester and Caerwent 6 T.F.C. Blagg, 'Schools of Stonemasons in Roman Britain', in (Eds.) J. Munby and M. Henig, Roman Life and Art in Britain (BAR 41, 1977), 51-70, pp. 64-9; idem, 'Roman civil and military Architecture in the Province of Britain: Aspects of Patronage, Influence and Craft Organization', World Archaeology, 12 (1980), 27-42, pp. 31, 37 ff. 7 Silchester: G.E. Fox and W.H. St. John Hope, 'Excavations on the Site of the Roman City at Silchester, Hants,' Archaeologia, lii (1890), 757 and PI. xxxiv, and ibid., liii (1893), 552, 555 and Pis. xxxvii and xxxviii; B.W. Cunliffe and M.G. Fulford, Corpus Signorum Imperii Romani, Great Britain 1.2: Bath and the Rest of Wessex (Oxford 1982), 53-5. Caerwent: V.E. Nash-Williams, 'The Forum-and-Basilica and Public Baths of the Roman Town of Venta Silurum at Caerwent in Monmouthshire,' Bull. Board of Celtic Studies, xv.2 (1953), 161 and PI. v. 67 T.F.C. BLAGG capitals, though the detail of the rosette in the centre of the abacus is distinctive. A minor variation in treatment between two of the Canterbury volute fragments suggests that they are the work of two different craftsmen. The decorative features are somewhat smaller in proportion to those of the Silchester capitals, which were 0.85 m. high overall, and 0.77 m. in diameter at the junction with the column shaft. It may be estimated that the Canterbury capitals were about 0.75 m. high and 0.70 m. in diameter at the shaft. Fragments of fluted column shafts of that diameter were also found in the excavations, as well as part of the torus moulding of a column base of appropriate size. The overall height of the column would have been about 7 to 8 m. Clearly, a monumental building is implied. The excavations revealed a gravelled courtyard of considerable extent, surrounded by a portico. The columns are too large to have stood on the stylobate of the portico, and it seems most likely that they came from a large building which stood within it, in the central area which has not been explored archaeologically. It is probable that this building was a temple. The arguments for this must await presentation in the final excavation report, which is now in preparation, but if they are acceptable, then from these fragments it may be inferred that Canterbury had a classical temple, similar to that of Sulis Minerva at Bath. The date, however, is a little later: the portico is said to have been constructed in the late first or early second century,8 and on stylistic grounds an early second-century date for the capital is likely.9 The typology of the Canterbury capitals shows that the building to which they belonged was one of the series.decorated by the same school of stone masons working in towns throughout southern Britain. The establishment of this, the first recognisable resident school of masons in the province, was a result of the urban building programme which was in its early stages during the governorship of Agricola,10 and was largely achieved by the end of Hadrian's reign. Fragments of smaller fluted columns and pilasters from the Cakebread Robey sites presumably belonged to subsidiary features either in the main building or in others within the precinct, possibly the small shrine discovered in 1981." More noteworthy is the evidence for lavish enrichment with veneers and panel mouldings of white and coloured marble (Fig. 2). Well over 600 fragments were recovered, 8 Arch. Cant., xcvii (1981), 280. 9 By inference from the capitals of the Silchester forum basilica, built in stone early in the second century: Britannia, xiv (1983), 331. '"Tacitus, Agricola, 21. " Arch. Cant., xcvii (1981), 280. 68 ROMAN ARCHITECTURAL ORNAMENT 23 Ii ii ii i % ^ Fig. 2. Marble Mouldings and Strips from the Cakebread Robey Sites, Canterbury: (Scale 1:2.) Nos. 1, 2, 4 and 5, Carrara Marble; nos. 3 and 6, Purbeck Marble. 48.6 kg. in weight. White Carrara marble from Italy predominates (26.2 kg.), with Purbeck marble in second place (10.8 kg.). There are small quantities of imported coloured marbles, notably red and green porphyry, pavonazza, verde antico, and also Tournai slate and Hastings marble.12 The excavations of the Marlowe sites, which included areas adjacent to the Roman theatre, and part of the public baths, also produced marble mouldings and veneers, though in smaller quantity (148 fragments). It is quite possible that both buildings were decorated with this material, but since most of it came from post-Roman contexts, certainty is impossible. Even within the Roman period, migrant fragments are recorded from the sites of modest private houses, to which they are unlikely to have belonged. 12 Identifications by Martyn Owen, The Geological Museum, South Kensington. 69 T.F.C. BLAGG A few small and very thin (4-7 mm.) pieces may have decorated furniture. They include a pink, possibly Numidian, marble, green porphyry, a brown and yellow breccia from Skyros, and dark grey slate. Otherwise, inlay was used in wall veneer and, as opus sectile, on floors. Although, as a general rule, the thinner and more preciselycut slabs and strips may be considered more likely to have been used on a wall, and thicker pieces, particularly when the exposed surface is worn and/or the unexposed surface is rough, can be attributed to a floor with some confidence, there are many pieces which could equally well belong to either category. Most of the imported coloured marbles are veneers of between 9 and 15 mm. in thickness. They include strips with rectangular or oblique ends, in red porphyry (Egypt), green porphyry and verde antico (Greece), pavonazzetto (Turkey), a pink marble (possibly north African) and white Carrara, and parts of large slabs in verde antico and Carrara marble. There are also strips and thin slabs in Purbeck and Hastings marble, slate and siltstone. Most, if not all, of these probably came from a wall decoration consisting of veneer panels set within rectilinear frames of contrasting colours. Elements which are likely to belong to geometric floor decoration are the triangles and rectangular strips of Carrara and Purbeck marble which range in thickness from 16 to 25 mm. Slabs of greater thickness, most of them exceeding 30 mm., must have come from floors, and the marble is mainly Carrara and Sussex, with some Purbeck and a few pieces of Egyptian onyx. The range and quantity of marble from Roman Canterbury is comparable only with that from Fishbourne and the Temple of Claudius at Colchester, and it shows that the town had some buildings of outstanding magnificence. The stones imported from outside Britain are all from the central or east Mediterranean; there are no examples of the Pyrenean and other French marbles found at Fishbourne.13 All these exotic marbles were in regular use in Italy; all could have been brought to Canterbury in one or two shipments from Rome; alternatively, the Canterbury builders may have selected what they required from the stockpile of a London entrepreneur. The Marlowe excavations also produced some larger architectural fragments in limestone. They include a column drum 0.50 m. in diameter, and part of the upper moulding of a column base of equivalent size. The dimensions suggest that these pieces belonged to 13 B.W. Cunliffe, Excavations at Fishbourne 1961-1969, Research Report xxvi of The Society of Antiquaries of London (Leeds, 1971), ii, 16 ff. 70 ROMAN ARCHITECTURAL ORNAMENT the portico of a public building,14 presumably a feature of the baths or of a neighbouring street colonnade. Canterbury lacks a local source of stone suitable for this type of architectural decoration. Apart from the marble, most of the items discussed so far were made from a pelletal limestone, the origin of which has yet to be established, but may have been northern Gaul. One expedient used in places where good building stone was not conveniently available was to make columns of segmental tiles, rendered externally in mortar. It was conjectured that a 'circular column of Roman bricks' found at Butchery Lane might have supported a hypocaust,15 but that was certainly not the case with the quarter-circular tiles found in the Marlowe excavations, which are of a size to have made a column between 0.55 and 0.65 m. in diameter, large enough for a substantial portico or for use in a major public building. Smaller segmental tiles, one of them semicircular, the others fragmentary, were also found, and would have made a column 0.35-0.45 m. in diameter, suitable for a smaller portico or the porch or verandah of a town house. Tile-built columns are rare in Britain, occurring, as might be expected, mainly in the south-east.16 The architectural fragments from Canterbury have all been found in the area of public buildings, to which most if not all of them must have belonged. With the possible exception of the smaller tile-built column just mentioned, we know nothing yet about the architectural embellishment of private houses in the town in the manner of those at Silchester, Cirencester and Caerwent.17 RICHBOROUGH In contrast with Canterbury, the decoration of one public building at Richborough is well enough known: the Monument itself.18 There 14 Cf. T.F.C. Blagg, 'Reconstructions of Roman decorated Architecture: Proportions, Prescriptions and Practices', in (Ed.) P.J. Drury, Structural Reconstructions (BAR British series 110, 1982), 131-51, p. 137. 15 VCH (Kent) Hi, 69, no. 5; Archaeologia, xliii, 163, no. 34. 16 J.H. Williams, 'Roman Building Materials in south-east England,' Britannia, ii (1971), 166-95, p. 192, for a list and references. 17 The column base in the arcaded screen between the nave and the presbytery of St. Pancras' church is of appropriate size to have come from a private building, but there are strong reasons for doubting the common assumption that it was a re-used Romano-British piece: T.F.C. Blagg, 'Some Roman architectural Traditions in the early Saxon Churches of Kent,' in (Ed.) A. Detsicas, Collectanea Historica: Essays in Memory of Stuart Rigold (Kent Archaeological Society, Maidstone, 1981), 50-3. 18 D.E. Strong, 'The Monument', in (Ed.) B.W. Cunliffe, Fifth Report on the Excavation of the Roman Fort at Richborough, Kent, Research Report xxiii of The Society of Antiquaries of London (Oxford 1968), 40-73; hereafter cited as 'Strong (1968)'. 71 T.F.C. BLAGG are, however, some new observations to be made. The great fourway Arch was erected between 80 and 90 A.D., the decade in which Roman armies under Agricola reached the north of Caledonia. Although we lack an inscription to prove it, it is a fair assumption that the Richborough Arch was built, at the place where the Roman army of conquest landed in A.D. 43, to commemorate the culmination of that conquest, Britannia perdomita. It may also be seen as an appropriate dynastic memorial to the emperor Vespasian, who had commanded a legion in the original invasion force, who had appointed Agricola as governor to renew the conquest, and who had died in A.D. 79. If any building in Britain (military fortifications apart) was erected as an act of direct imperial benefaction, the Richborough Arch is the most obvious candidate. That is suggested, not only by its context in time and place, but also by its decoration. It was encased in Carrara marble,19 embellished with fluted columns and pilasters, and in its superstructure with bands of bead-and-reel ornament. It is the only building in Roman Britain known to have been decorated externally in marble. Such lavishness is itself an indication that the Arch was an imperial building project, with the marble supplied direct from the quarries which were under imperial control. It would also have been necessary to import the skilled marble masons from Italy or southern Gaul, for the carving of marble requires techniques which the masons, who at this time were starting to carve limestone decoration for the towns of Britain, are not likely to have mastered. Although some masons with marbleworking skills must have been available to carve the veneers for the internal decoration of the Fishbourne palace and urban public buildings, that work was mainly limited to plain mouldings and geometric shapes. The fluting of marble casings and the carving of bead-and-reel ornament required greater expertise;20 there is no certain evidence that such more elaborate ornament was produced at any other site in Britain,21 and in any case the Arch was such a 19 Examples of Pentelic marble have apparently also been identified in the past, but if the identification was correct, the quantities cannot have been significant, for Professor Strong found only Carrara marble among the several thousand fragments available to him for examination: Strong (1968), 64. 20 It has been remarked, however, that 'the quality of the workmanship is fair only': Strong (1968), 65. 21 There is a piece of fluted marble casing for a large column in the British Museum (CRS 56 7-1 7a), unprovenanced, but from Roach Smith's collection; he may be suspected to have found it at Richborough. The rest of the marble from Roman Britain known to the writer is from the interior decoration of buildings like the Canterbury material discussed above, or figure sculpture which was probably imported into Britain in a finished state. 72 ROMAN ARCHITECTURAL ORNAMENT large-scale project as to justify the bringing of masons to Britain for the sole purpose of working on it. The bulk of the monument was constructed from native materials: mortared flint, and blocks of greensand and oolitic limestone. The structure measured 87 by 47 ft. in plan, and Strong estimated its height as 86 ft.22 Allowing, of course, for the arched passageways through it (the heights of which are also estimated) the volume of material can be calculated to be 290,184 cubic feet, or 8201 cubic metres. That would have weighed about 20,500 tonnes. No doubt most of this would have been brought to Richborough by water, but it must have been carted at some point on its journey, which would have required 41,000 waggon loads.23 The surface area of the monument was approximately 20,600 ft. sq. The flat wall veneers of marble vary in thickness from \\ to 2\ in., but the mouldings are thicker. Allowing also for wastage in the cutting, an average thickness of 3 in. will be assumed. On these figures, the volume of marble was 5150 cubic feet, or 145.6 cubic metres, a weight of 378.5 tonnes.24 How many shiploads this represents depends, of course, on the size of ship, but this is unlikely to have been less than 100 tonnes burthen or more than about 400 tonnes;25 that is, the marble for the Arch could have been carried in from one to four shiploads. A calculation of this sort is inevitably based on a number of variables, but it does indicate a rough order of magnitude for the logistics of the building operation. In making his study of the marble fragments for the publication of the fifth Richborough report, Donald Strong observed that 'although a large number of different kinds of mouldings survive and their general purpose is clear enough, it is not possible to assign them precisely to positions on the building'.26 His reconstruction, therefore, showed the decorative scheme of the monument in very simple 22 Strong (1968), 60. 23 G.C. Boon, in making similar calculations about the town walls at Silchester, takes the weight of oolite as 2480 kg./m.3 and that of flint as 2560 kg./m.3: Silchester, the Roman Town of Calleva (Newton Abbot 1974), 101, 319, note 11. It is impossible to calculate the relative quantities of different materials in the Richborough Arch; the figure used here for an approximation of the weight is 2500 kg./m.3. The estimate for a waggon load of 500 kg. is that used by Boon (loc. cit.) based on the 1500 librae maximum load specified in Cod. Theod., viii.3.8. 24 Assuming a bulk density of 2.6 for marble, i.e. 2600 kg./m.3: see E.M. Winkler, Stone: Properties and Durability in Man's Environment (Vienna, New York, 1975). 25 Study of seven Mediterranean wrecks indicated that they ranged from 200-450 tons burthen: P. Pomey and A. Tchernia, 'Le Tonnage maximum des Navires de Commerce romains', Archaeonautica, ii (1978), 233-51. 26 Strong (1968), 65. 73 T.F.C. BLAGG terms, indicating fluted columns at the corners and adjoining the archways, with fluted pilasters supporting the arches. The capitals are represented by a trapezoidal outline; one would expect those of the columns to have been Corinthian or perhaps Composite, as on nearly every other Roman arch, at least at the corners; no fragment of such capitals survives, however. The bead-and-reel mouldings may have served as panel framings or to divide the three fasciae of the architrave, but are too small to be shown on the reconstruction drawing. Recent transfer of the material to Dover Castle for storage has permitted a re-examination of some of the marble, and a few new observations about the decoration can be made.27 First, in addition to there being fluted columns, there are also some unfluted column casings of the same diameter (1.20 m.). Secondly, although Strong did not find any pieces which he could identify as column or pilaster capitals, there are some column casings which have an ovolo and a cavetto moulding and which seem likely to have come from Tuscan column capitals (Fig. 3, no. 1). These would have surmounted the unfluted columns. It may therefore be suggested that the monument had fluted columns at the corners (with Corinthian or Composite capitals), and unfluted columns with Tuscan capitals alongside the archways, though not necessarily on both the long and the short sides. They may have been employed only on the short sides, where their relative plainness would have lightened the visual appearance of their being set rather close to the corner columns; or, for similar reasons of spacing, columns alongside the arches on the short sides may have been omitted, and the Tuscan columns would then have been employed in a corresponding position on the long sides. Thirdly, there are fluted pilasters and pilaster bases of two sizes. The larger sizes of each are of similar dimensions to the fluting and base mouldings of the columns, and presumably supported the arches on the long sides at least; the smaller pilasters (Fig. 3, nos. 2 and 3) may have carried the lower and narrower arches on the short sides, or have decorated the corners of the attic, placed vertically above the columns. Finally, although in the absence of decorative mouldings other than the bead-and-reel the decoration appears to have been much less elaborate than is characteristic of such contemporary monuments in Rome as the Temple of Vespasian or the Flavian palatium, the two surviving pieces of marble sculpture, which have 271 am grateful to Mr. Andrew Petrie and to Dr. C.J. Young of the H.B.M.C. for arranging for me to see the material and for various points made in discussion. There are several thousand fragments and re-examination has of necessity been selective. 74 ROMAN ARCHITECTURAL ORNAMENT BS T 3 w I Fig. 3. Architectural Fragments from Richborough: (Scale 1:3.) No. 1, Marble Column Capital Moulding from the Arch; nos. 2 and 3, Marble Pilaster Base Mouldings from the Arch; no. 4, decorated Moulding in oolitic Limestone. been somewhat disregarded, give a hint of greater elaboration, possibly a figured frieze or relief carving in the panels between the columns. One piece represents the fore-quarters of a horse, seen frontally, the head and legs of which are missing; the other is the back of a human head, badly damaged, but with some locks of hair visible.28 It should be noted that such relief carving could not have 28 Presumably these are the 'two fragments of marble sculpture, neither of them clearly identifiable', mentioned by Strong (1968), 62. He thought that the former was perhaps a frontal representation of a centaur, and that it might have decorated one of the keystones of the arches; that would be an odd place to find a centaur, though we may agree with Strong's supposition that the keystones were decorated. He made no further comment on the second fragment. 75 T.F.C. BLAGG been executed on thin veneers, but will have required whole blocks of marble, a particularly attractive quarry for the lime-kilns, which may explain why more does not survive; likewise, with the posited Corinthian capitals. Richborough has also produced several pieces of architectural decoration in oolitic limestone and greensand which have received relatively little attention. Some of them indicate that there was at least one other major monumental building there. There are four fragments of a Corinthian capital of limestone. They are insufficient to permit a restoration of the whole capital or to suggest a date; but there are two fragments of volutes on which, unlike the Canterbury capitals mentioned above, the spirals are exposed, showing that the capital is not one of the type normal in southern Britain, and is therefore not likely to have been produced by the established school of masons who were responsible for that type. It is not likely that the capital belonged to the Arch; limestone capitals, even if stuccoed, would seem inappropriate for a building decorated otherwise entirely in marble, and in any case the proportions suggested by the fragments seem considerably too small. They are better suited to the four pieces of fluted column of limestone, approximately 0.55 m. in diameter, which also survive. The association may be extended further to include a cornice fragment decorated rather crudely with an egg-andtongue and a foliate moulding (Fig. 3, no. 4). That all these pieces came from one building is a hypothesis, but an economical one: for each of them by itself indicates a building of importance. Since fluted columns were normally used, in Britain, as exterior decoration, the building may have been a small classical temple. An architrave block, 0.315 m. high and 0.865 m. wide, with a cyma moulding and three fasciae, is of greensand, however, and probably belonged to another building, though one of similar monumental character. There are nine pieces of small columns with Tuscan capitals, the shafts ranging in diameter from 0.145 m. to 0.225 m. These are as likely to have come from the porches and verandahs of private houses and Romano-Celtic temples as from military buildings. Among these pieces, a pair of bases and a pair of capitals are of interest because they have quite close parallels elsewhere in Britain. They are broken from their shafts, but they probably all belonged to the same set of columns, since they are comparable in size and in the carving of their mouldings (Fig. 4, nos. 1 and 2). The distinctive feature of the capitals is the short steep cyma mouldings with beads and undercut fillets.29 Dated examples of the type from Caerleon and Kingsweston T.F.C. Blagg, op. cit. in note 2, 339-341. 76 ROMAN ARCHITECTURAL ORNAMENT belong to the late third or early fourth century,30 but the closest parallel for the Richborough capitals' mouldings is an unpublished capital from Chedworth. The bases belong to a type characterised by the very slight curvature of the torus mouldings;31 Chedworth, again, provides a close parallel, as does the Colliton Park house at Dorchester, dated to the first half of the fourth century.32 Stone masons travelled widely in lowland Britain in the third and fourth centuries, and the typological resemblances here are close enough to indicate that the masons who carved these Richborough columns also worked at Chedworth. OTHER SITES The same point may be made with regard to other column fragments from Kentish sites. They are all of similar size, with shafts approximately 0.20 m. in diameter. Where columns of this size survive complete, they rarely exceed 1.20 m. in height,33 and were made to stand on a low wall, usually of a verandah. The column capital from LuUingstone is one of a type distinguished by having very steep moulded profiles, between IT and 80° to the horizontal, and by proportionately tall cyma mouldings, with beads and indented fillets (Fig. 4, no. 3). Capitals from House XIII N at Caerwent and from Towcester are so alike to it in their proportions as to suggest that they were all carved by the same mason.34 A piece of column base from Farningham (Fig. 4, no. 5) has a very tall steep cyma moulding above the upper torus.35 This links it with a group of columns from the Bignor villa, the octagonal temple at Nettleton Shrub, the Westcotes villa, outside Leicester, and Caerleon, datable to the late third or early fourth century.36 That group is also likely to mJRS, xlvi (1956), 122; G.C. Boon, 'The Roman Villa in Kingsweston Park (Lawrence Weston Estate) Gloucestershire', Trans. Bristol and Gloucester Arch. Soc, lxix (1950), 5-58, pp. 11, 19-21. 31 T.F.C. Blagg, op. cit. in note 2, 311-3. Z2R.C.H.M. Dorset (1970), II (South-east, part Hi), 553-8. 33 Blagg, op. cit. (note 14), 140-2. 34 G.W. Meates, The Roman Villa at LuUingstone, Kent, I: The Site (Maidstone 1979), 46 and pi. viii, d; for the comparative illustrations and further references, see T.F.C. Blagg 'Architectural Patronage in the western Provinces of the Roman Empire in the third Century', in (Eds.) A. King and M. Henig, The Roman West in the third Century, (BAR international ser. 109, 1981), 167-88, p. 176 and fig. 11.2. 35 The piece was found in excavations by the Kent Archaeological Rescue Unit: I am grateful to Mr. B.J. Philp for permission to discuss it here in advance of full publication. 36 Blagg, loc. cit. (note 34). 77 T.F.C. BLAGG ^ 7 1 Fig. 4. Column Capitals and Bases: (Scale 1:10.) Nos. 1 and 2, Richborough; no. 3, LuUingstone; no. 4, Eccles; nos. 5 and 6, Farningham; no. 7, Dover. be the work of the same mason, for the columns are closely similar in their capitals and their overall proportions as well as in their base mouldings. In relation to the Farningham base fragment, these additional criteria are absent, so the case for its identity with the group, while plausible, cannot be fully demonstrated. The fragment is, however, very different in type from the small column base found nearby, re-used in the so-called Farningham Villa II (Fig. 4, no. 6).37 This, by contrast, has a low, rather wide-splayed cyma, and torus mouldings which are also very low in relation to their diameter. This profile is similar to that of a base from the villa at Eccles (Fig. 4, no. 4).38 It is by no means usual to find, as has been possible for these "Arch. Cant., Ixxxviii (1973), 11. 38 Arch. Cant., Ixxxiii (1968), 45; A. Detsicas, The Cantiaci (Gloucester 1983), 124; I am grateful to Mr. A.P. Detsicas for inviting me to report on the Eccles column fragments, which also include two pieces of column shaft. 78 ROMAN ARCHITECTURAL ORNAMENT Kentish column parts, typological correspondences so close as to indicate the hand of individual masons. Their recognition depends on the recurrence of exaggerated or abnormal features in the mouldings or proportions which, in comparison with the normal range of variation, can be considered as trade marks. The great majority of Romano-British plain-moulded columns cannot be classified so closely: so, for example, the capital from the Classis Britannica fort at Dover (Fig. 4, no. 7), one of a widespread southern British type, with standard cyma and cavetto mouldings, current from the early second to the fourth century.39 The quantity of these columns which survives from Kent is small; nearly all those known to the writer have been discussed here.40 The area lacks building stone suitable for this sort of architectural decoration, and therefore the basis for the development of a local school of carving. On the occasions when that decoration was required, it seems that the masons were brought in from elsewhere. It is unusual, and fortunate, that the activities of several of them can be attested in other areas. Furthermore, Kent's geographical location was favourable to masons coming to work there from across the Channel. The best evidence for this is the Corinthian capital which crowned the votive column in the temple precinct at Springhead.4' It is of a different type from that current in most of lowland Britain, of which the Canterbury capital (above) is an example, though that type, too, has Gaulish antecedents. The Springhead capital is unique in Britain, but it is very closely related to east Gaulish capitals of the late second and early third centuries. The Richborough Corinthian capital fragments mentioned above, which are also unparalleled in Britain, might also be explained as the work of a Gaulish mason, but the possibility requires further research. A continental origin, too, perhaps in one of the legionary fortresses of the Rhine, may be suggested for the carver of a fine moulding decorated with foliage, recently found at Dover.42 39 T.F.C. Blagg, 'The Column Capital', in B.J. Philp, The Excavation of the Roman Forts of the Classis Britannica at Dover, 1970-1977 (Dover 1981), 177. 40 The others comprise a base from Dover, a base and a capital from Richborough, and part of a column from Springhead; the writer would be glad to hear from readers who know of other examples not mentioned in this paper. 41 T.F.C. Blagg, 'The Votive Column from the Roman Temple Precinct at Springhead', Arch. Cant., xcv (1979), 233-9. 42 Found in excavation of the Market Hall site in 1982 and currently displayed in the Dover Painted House Museum. I am grateful to Mr. B.J. Philp for permission to cite the discovery here. An ampler note on it, pending full publication, will appear in the Kent Archaeological Review. 79 T.F.C. BLAGG CONCLUSION Roman architectural ornament in Kent has survived the destruction of later generations in fragmentary form, but the chance survivals show a considerable variety and richness. In Canterbury, and in the villas, architectural connexions can be established between the Cantiaci and other civitates of lowland Britain. Unlike the latter, however, Roman Kent retained military establishments for most of the period, and ports of entry from the continental Empire. It was not so self-contained as, for example, the architecturally rich civitas of the Dobunni. That compensated to some extent for the lack of the freestone which the Dobunni could use so abundantly. Kent attracted patronage, and this produced a rich diversity in its architecture; notably, an imperial monument, in the Richborough Arch, and classical temples and a theatre, of which there are very few known from Roman Britain. The evidence may be fragmentary; but a fragment of a Corinthian capital or a decorated cornice is evidence for the former existence of a complete and elaborate building. Some day, excavation may reveal more of them. Meanwhile, since Caesar said that in his day Kent was the most civilized part of Britain, it is pleasant to record, in the hundredth volume of its Archaeological Society, that there are good architectural reasons for thinking that it remained so after his successors had conquered it. 80