The Association for the Study and Preservation of Roman Mosaics

About ASPROM News & Events Publications Mosaics in Britain Resources Join ASPROM

The Seasons Mosaic from Silchester and its architectural context

Stephen R. Cosh

Fig. 1. The Seasons Mosaic from Silchester (painting by Stephen R. Cosh).
Fig. 1. The Seasons Mosaic from Silchester (painting by Stephen R. Cosh).

In 1901, during the excavations of House 1 Insula XXVII at Silchester, the fragmentary corner of a mosaic panel, including a bust, was found in Room 16.[1] Very little survived, principally because of the robbing in antiquity of the tiles from the underlying hypocaust. Despite its condition, the mosaic was lifted and is presently in a store belonging to Reading Museum. No illustration appeared in Archaeologia nor, unusually, does it appear to have been drawn by George Fox who drew other mosaics from the building. A tracing of the bust was made by the late George Boon who illustrated it in his Silchester: The Roman Town of Calleva.[2] The illustration here (fig. 1) is the first time the whole fragment has been published.

The surviving part consists of the bust in a square compartment with fragmentary remains of neighbouring compartments to the left and above and a length of straight-tongued double guilloche on one side. The latter had become detached from the main fragment and the author recently identified it as forming part of this mosaic. The fragment is most probably the corner of a scheme comprising a large central square adjacent to lateral rectangles with small squares in the corners, all delineated by dark grey double fillets. The whole is surrounded by another dark grey double fillet and a band of white. The band of double guilloche did not extend all round but formed a straight length on one side and probably two sides originally assuming the coarse borders were the same width to the north, south and west of the panel. Although no part was lifted, the outer border was of coarse red tesserae. The bust in the small square compartment does not demonstrate a high level of craftsmanship, although executed in fine tesserae. Assuming the figure to be female, she has red flowers in her hair and is clad in a grey cloak; over her left shoulder is a twig-like feature perhaps intended to be a long-stemmed flower although no red tesserae representing petals are evident. The excavators believed this figure to be Flora while Boon identified her as Ceres and attributed it to the Corinian School of mosaicists — a claim for which there seems no grounds. However, there seems little doubt that the bust represents a Season, and the poppy-like flowers would indicate Summer or possibly Spring. The neighbouring compartment is more fragmentary and difficult to interpret but appears to contain a foliate scroll with alternating red and pink leaves. Of the large central square panel nothing remained but it presumably contained figured work. It is not known from which part of the panel the fragment comes but, being rectangular, it has to be from the south-east or north-west corner. As the north-west corner was over a junction of hypocaust channels, the south-east corner was probably more likely to have survived.

Fig. 2. Plan of Rooms 16-18, Insula xxvii,1, Silchester (drawing by Stephen R. Cosh).
Fig. 2. Plan of Rooms 16–18, Insula xxvii,1, Silchester (drawing by Stephen R. Cosh).

Room 16, containing the Seasons mosaic, had a broad opening into Room 17, with which it formed a large bipartite room, with Room 18, a large apsidal annex, entered from Room 17. These rooms also had mosaics which, not being over hypocausts, fared rather better; they were also lifted and are in store in Reading.[3] (fig. 2) The mosaic panel in Room 17 was not placed centrally but positioned close to the wide opening into Room 16. The Seasons mosaic is described as 'central' but a dotted line on the plan over the hypocaust channels suggests that this mosaic was also adjacent to the opening. However, the red border around the Seasons mosaic is in marked contrast to the buff surround of Room 17, and the lack of a linking threshold panel is unusual; the gap between the responds was infilled red.[4] It is possible that the Seasons mosaic predates Room 17. The 'nine-panelled scheme' is typical of second-century pavements[5] although the form of the double guilloche is not the 'tight' form prevalent then, whereas the design of Room 17 is matched on fourth-century mosaics at Colliton Park, Dorchester, and Yatton and Lufton in Somerset.[6] Given their relative positions to the opening a contemporary date seems likely. Room 17's pavement was in use long enough to require a repair crudely executed in tile. If Room 17's mosaic was inserted after the conversion of two rooms into one, it would represent a similar process to that seen at Withington, Glos. and Fifehead Neville, Dorset where extensions were built to produce bipartite rooms, with new mosaics added to the older ones.[7] In the Silchester example the division between the two parts of the room is noteworthy. As well as the normal responds, free-standing square foundations suggest the presence of columns supporting an architrave and perhaps an arch (fig. 3).

Fig. 3. Axonometric drawing showing partial reconstruction of Rooms 16-18 (drawing by Stephen R. Cosh).
Fig. 3. Axonometric drawing showing partial reconstruction of Rooms 16–18 (drawing by Stephen R. Cosh).

Similar elaborate architecture including free-standing columns has been noted in a bipartite room at Frampton.[8] The apse (Room 18) is perhaps a later addition and David Johnston has compared its near-circular mosaic to one from Sparsholt, Hants, and attributed it to his Central Southern Group of mosaicists operating in the first half of the fourth century.[9] However, the form and colouring of the double guilloche in Rooms 16 and 18 is identical and contemporaneity cannot be ruled out despite the differing styles of the three panels.

The subject matter and schemes of the mosaics are typical of bipartite rooms. Seasons appear in no less than nine such rooms in Britain.[10] However, this may not have any special significance for, as Professor Ling has pointed out,[11] the fact that there are four seasons makes them eminently suitable to occupy the corners of figured panels. The use of the scheme based on tangent circles in Room 17 is also disproportionately common in such rooms. Its occurrence also at Frampton, Hinton St Mary, Fifehead Neville and Lufton[12] is so frequent as to suggest that the design may have reflected that of a vaulted ceiling.

The large half-heated bipartite room with apse in House 1 Insula XXVII may seem unusual but actually House 1 Insula VIII has a very similar room (fig. 4, b).[13] Again a hypocaust occupies one half, but here the floor had been totally destroyed. Two more examples from Silchester are suspected (fig 4, e–g),[14] but it is uncertain whether they were single rooms since wall divisions are shown which may be relics of an earlier phase as at Whatley, Somerset (fig. 4, f). A bipartite room with an apse to one side can be noted on at least three villa sites in south-west Britain. The Frampton example is well-known, and similar rooms were found at Whatley and Banwell (Riverside) in Somerset (fig. 4, e–g),[15] only the last being partially heated. It is perhaps significant that, apart from Banwell, all the apses have southerly aspects and the exterior is curved and not squared off as is often the case. Presumably both were chosen for maximum lighting at any time of day. The rounded exterior suggests a series of windows as at Pliny's apsed room with the same aspect in his villa at Laurentum; he used this as a reading room.[16] The rooms all seem to be at the end of the porticus/corridor. Admittance was gained to the room with the Seasons Mosaic in Silchester from the plain buff tessellated porticus/corridor through a wide entrance, probably with double doors.

The question arises whether the many similarities between these rooms indicates the same function, especially as their final forms date, where known, to probably the second quarter of the fourth century. It is tempting to associate them with the increasing trend in dining rooms for the stibadium, a semicircular couch. The wide borders in the Silchester apse could easily accommodate such furniture and the wide borders around three sides of the Seasons Mosaic can be interpreted as positions for three couches before the addition of the apse although the position of the door may preclude this. In that case, if they are to be regarded as more than convenient filling of corner compartments, the Seasons may refer to the provision of the meal in the same way as the frequent occurrence of canthari in British triclinia could relate to the liquid refreshment.[17] In his study of sixteen apsed rooms with mosaics[18] (there are actually over twice that number known in Britain), Dr Ellis has argued that apses generally indicate the former presence of stibadia even though the furniture would obscure the design, as at Frampton where the cantharus would probably have disappeared from view. Ellis suggests this may '...reflect a more flexible use of furniture in the room..' and that '...designs are always liable to be occluded by furniture.'[19]

He suggests that the subject matter and inscription at Frampton, comparable with Mediterranean examples, supports its identification as a triclinium — or, perhaps more properly, a cenatio. On the other hand, the same mosaic at Frampton, especially because of the inclusion of a chi-rho and the similarity to the 'Christ' mosaic at Hinton St Mary, led Black to write: 'such elaborately composed, and explicitly Christian, floors can hardly have belonged to anything but chapels or house-churches.'[20] Similar interpretations both as triclinia and chapels, albeit Orphic rather than Christian, have been proposed for Whatley (fig. 4, f) and the triconch room at Littlecote.[21]

It is probably wrong to assume that a similar architectural design implies identical function, especially when comparing town and country situations. Function can change even, under the same ownership: Pliny, for instance, describes one of his rooms as either a bedroom or dining room (vel cubiculum grande vel modica cenatio).[22] On balance, and without the benefit of knowing the probable figured contents of the centres of the Seasons Mosaic and apse, the identification of the Silchester room as a triclinium/cenatio seems the most likely — perhaps doubling up as an audience chamber as suggested for the large apsed room at Box, Wilts.[23] It was certainly the grandest room in the house and no other room is an obvious candidate. This would be suitable for winter as well as summer use, even though it did not enjoy the panoramic views of more obvious summer triclinia at Folkestone, Abinger, Downton, Rockbourne, etc.[24] where square mosaic panels are set forward in central unheated rooms to allow three couches to be positioned without obscuring fine mosaic. An 'all year round' dining-room was probably more suitable in a town house than a country 'mansion' where lavish entertainment during winter months was less likely, and the view in a town situation was probably of much less importance than internal appearance. This is not to say that winter triclinia did not exist in country villas — for instance, the heated Room 4 at Dewlish with a fine mosaic and a three-sided apse simply paved in coarse tesserae.[25] The unheated rooms at Frampton, Whatley and Littlecote are not suitable for winter use, but neither do they make much sense as summer dining-rooms — they lack the typical broad southern or eastern entrances, such as at the Keynsham, Abinger and Rockbourne triclinia,[26] where, possibly, folding doors (valvae), known to have been used throughout the Roman period,[27] could be pulled back for diners to enjoy the view. The Silchester example is of great interest, and its unusual design and heating arrangement may hint at a multi-purpose, but clearly impressive, room. It serves to show that it is important in the study of mosaics not to divorce them from their architectural contexts.

The author wishes to thank David Pearson of Reading Museum for the help rendered in this study of the 'Seasons' and other Silchester mosaics, and Pat Witts and David Neal for their helpful comments; any errors, though, are the responsibility of the author.


1W. H. St. John Hope, 'Excavations on the site of the Roman city of Silchester, Hants, in 1901', Archaeologia 57 pt. 1 (1902) 21, pl. II (plan).
2G. C. Boon, Silchester: The Roman Town of Calleva (2nd ed., 1974), 216, pl. 31.
3Room 17: G. C. Boon, op. cit., fig. 30; Room 18: W. H. St. John Hope, op. cit., pl. III.
4W. H. St. John Hope op. cit., 21.
5D. J. Smith, 'Mosaics before the fourth century' in H. Stern (ed.), La Mosaïque Greco-Romaine I (1965), 95–116.
6Colliton Park: RCHM Dorset II 3 (1970), pl. 220; Yatton: PSANHS 31 (1886) pl. facing p. 70; Lufton: PSANHS 116 (1972) pl. V.
7Withington: RCHM Glos. (1976), pl. 16; Fifehead Neville: RCHM Dorset III 1 (xxxx), pl. 133, plan p. 94.
8S. R. Cosh, 'A possible Achilles at Frampton', Mosaic 23 (1996) 13, fig. 1.
9D. E. Johnston, 'The Central Southern Group of Romano-British mosaics' in J. Mundy & M. Henig (eds), Roman Life and Art in Britain, BAR 41 (1977) 195–215, pl. 7.I.
10Room 26, Bignor, Sussex; Room 12, Brading, Isle of Wight; House VII, Room 6–7, Caerwent; Room 5, Chedworth; Littlecote, Wilts; Lullingstone, Kent; Malton, Yorks; Pitney, Somerset; Rudston, Yorks; and possibly Winterton, Lincs.
11R. J. Ling, 'The Seasons in Romano-British mosaic pavements', Britannia 14 (1983) 19–20.
12Frampton & Hinton St Mary: D. J. Smith 'The mosaic pavements' in A. L. F. Rivet, The Roman Villa in Britain (1969), pl. 3.27 & 3.29; Fifehead Neville: see note 7; Lufton: see note 6.
13Archaeologia 54 (1894) pl. 18.
14Archaeologia 52 pt. 2 (1890) pl. 29; 60 pt.2 (1907) 434–5 fig. 1, pl. XL.
15Frampton: the architectural nature of this room is discussed by B. Walters 'Exotic structures in 4th-century Britain' in P. Johnson (ed.), Architecture in Roman Britain (1996), 157–8, fig. 13.6, but does not link it to the other examples shown on fig. 4 here; Whatley: VCH Somerset I (1906), fig. 77, PSANHS 114 (1970), 37–47; Banwell: JRS 58 (1968) 199 (reference only).
16Pliny, Letters 2.17, 34.
17 However, for an opposite Empire-wide view, see R.J. Ling, 'The decoration of Roman triclinia' in O. Murray & M. Tecusan (eds), In vino veritas (1995), 239–51, British School at Rome.
18S. P. Ellis, 'Classical reception rooms in Romano-British houses', Britannia 26 (1995) 169–78
19Ibid., 171
20E. W . Black, 'Christian and Pagan hopes of salvation in Romano-British mosaics' in M. Henig & A. King (eds), Pagan Gods and Shrines of the Roman Empire (1986), 150, fig. 1.
21Summary: Current Archaeology 80, vol. VII no 9, (December 1981) 264–8.
22Pliny, Letters 2.17, 43.
23H. Hirst et al., 'Excavations at Box Villa 1967–8', Wiltshire Archaeological Magazine 81 (1987) 31.
24Folkestone (Room 40): VCH Kent 3 (1932), pl. XXI; Abinger: Mosaic 23 (1996) 12; Downton (Room I): WAM 58 (1963) 307, fig. 2; Rockbourne (Room 33): Arch.J. 140 (1983) 147, fig. 2.
25Dewlish: PDNHAS 93 (1971) 157–60, fig. 20, 21 (plan). Very similar rooms with the same form of channelled hypocausts and identical aspect were found at Wigginton, Oxon, and Colerne, Wilts.
26Keynsham: Archaeologia 75 (1924–5), 125–6, fig. 6. Abinger and Rockbourne: see notes 24 & 23.
27Pliny, Letters 2.17, 22; Sidonius, Poems XXII, 20.

© Stephen R. Cosh 1997. This article was originally published in Mosaic 24 (1997) 8–12.
This page is maintained by Ruth Westgate.