Roman Remains

Pages 520-527

An Inventory of the Historical Monuments in Dorset, Volume 2, South east. Originally published by Her Majesty's Stationery Office, London, 1970.

This free content was digitised by double rekeying and sponsored by English Heritage. All rights reserved.


In this section

Roman Remains

The Roman roads and the Roman town of Dorchester (Durnovaria) receive separate treatment in the first part of this section of the Inventory; otherwise the remains of the Roman period in the area, being less usefully classifiable, are described in a succeeding gazetteer; in this the monuments are arranged topographically in alphabetical order of parishes.

Where appropriate, a prefatory note to the parish includes the following: crossreferences to monuments described in the Earthworks section of the Inventory which are known to have a Roman phase; sites discovered subsequent to the preparation of the Inventory; discoveries of Roman objects that do not necessarily indicate occupation; sites of which the position is insufficiently known, unless they can be reasonably associated with listed monuments.

Evidence for the prehistoric origin of sites of Roman date, other than those described in the Earthworks section, is considered here; purely prehistoric sites with no earthwork remains, and prehistoric phases of Roman sites where there is apparent discontinuity, are listed elsewhere (pp. 511–12). Maumbury Rings, although an earthwork of Neolithic origin, is described under Roman Dorchester, since its adaptation as an amphitheatre was mainly responsible for its present appearance.

Museum accession numbers are given only for more significant finds where difficulty may arise in identification. For the occasional use of ten-figure map references, see p. 540b, n. 1. In the Introduction that follows, attention is directed mainly to monuments other than those of the Roman town, which has its own Introduction (p. 531).


The area described in this volume fell in Roman times within the territory of the Durotriges who were with little doubt one of the two unnamed tribes that offered a stubborn resistance to the 2nd legion under Vespasian in c. A.D. 44. Evidence for the campaign, and for the subsequent military occupation of short but uncertain duration, comes mainly from the hill-fort of Maiden Castle considered above (Winterborne St. Martin (142), pp. 493–501), the Roman forts at Hod Hill and Waddon Hill, and the Lake Farm site. (fn. 1) The existence of early military sites has been suggested at Dorchester and at Hamworthy on Poole Harbour, and may be suspected at Weymouth, the two latter as links in a coastal series extending from Fishbourne near Chichester perhaps to Topsham near Exeter. (fn. 2) There is little direct evidence for Weymouth either in the site at the head of the Radipole inlet (Weymouth (437)) or in the solitary legionary belt-plate from 'Greenhill' (p. 614) but the alignment of Approach Road 2 at Dorchester may have been a relic of the military phase (p. 533). Military equipment has not yet been found at Hamworthy and the early Roman objects from the known site there (Poole (402)) could perhaps be explained by continuance of the long-established trade attested there by pre-Roman imports, and perhaps also in the hinterland if the Greek coins listed by Milne can be relied upon. (fn. 3) That a military site remains to be located nearby, perhaps closer to the mouth of the Holes Bay channel, (fn. 4) seems certain however from the behaviour, near Badbury Rings, of the road linking the peninsula with the main road from Old Sarum which it seems was first surveyed to run to Poole Harbour rather than Dorchester (see Fig. p. 529). (fn. 5) The military character of the route is confirmed by the presence near it of a Claudian kiln at Corfe Mullen (pp. 525–6) probably making pottery for the army, and of the extensive military site of like date newly discovered beside it in the Stour valley at Lake Farm (see p. 529). The large mill from Hamworthy and part of a donkey-mill of Pompeian type from the Corfe Mullen site, both imports from the Rhineland, are antiquities of the highest rarity in Britain and in their context hardly allow an alternative explanation.

The fort at Hod Hill was evacuated by A.D. 51 at the latest, but military control of the district no doubt continued until the granting of self-government to the civitas, centred on Durnovaria (Dorchester), most probably c. 70 when native occupation of Maiden Castle seems to have ended (see p. 533). Thereafter we find little to indicate political conditions in the region, (fn. 6) although the main tendencies distinguishable in the civil zone of Roman Britain in urban and in rural development, at least in the few known villas, are evident. Mosaic floors in the latter, and at Dorchester in some profusion, are probably largely referable to the 4th century and imply prosperity in the late Roman period at least among landed and perhaps official classes. Two coin hoards of the mid 3rd century, however, at Sterte (Poole (404)) and Jordan Hill (Weymouth (445)), and others at Jordan Hill, Maiden Castle and Dorchester of c. 400, may reflect economic or political insecurity.

Standards of Romanization do not seem to have been high outside the capital and the villas; minor urban centres are lacking unless there was one at Wareham (p. 614), (fn. 7) and although the roads shown on the distribution map facing p. 634 can represent only those sufficiently important to be solidly constructed with agger and metalling, and perhaps not all of these, (fn. 8) few settlements show any close relationship to the system of through roads. Local networks of roads associated with a particular settlement, as at Town Hill (Ancient Field Group (1)), may however have been connected with them. As will be seen in the following review of rural settlement, the distribution remained in essence a prehistoric one; such changes as can be discerned, at least in the direction of industrial expansion, seem already in process before the Roman period. The reason may lie in the political, economic and cultural conservatism manifest in what we know of regional history in the preceding five or six centuries, but which it would be inappropriate to enlarge upon here.

Rural settlement. The area of the present survey includes parts of all three major ecological regions of Dorset as defined by Good, (fn. 9) namely (i) a large part of the Poole Basin consisting of Tertiary clays, sands and gravels, (fn. 10) (ii) a relatively small part of the Chalk, mainly to W. and S. of (i), and (iii) most of the Southern Vales sub-region, consisting of older and more varied formations around Weymouth and in Portland and S. Purbeck. The distribution of visible remains of prehistoric and Romano-British settlement, treated in the Earthworks section, is largley an upland one, strikingly in favour of the Chalk and of the limestone areas in S. Purbeck, and almost wholly ignoring the Poole Basin. The pattern is doubtless incomplete and arguably misleading since the uplands have been least subject to destructive agencies, but in the main it is supported by the buried sites treated in this section. These may be considered complementary in that they frequently owe their discovery, directly or indirectly, to the same agencies of destruction.

The distribution of these sites, a third of them perhaps with roots in the pre-Roman Iron Age, does however show that settlement in the valleys was greater than has been supposed; (fn. 11) of 119 sites, (fn. 12) no fewer than thirty-nine fall within the Poole Basin, while of the forty in S. Purbeck, fifteen lie off the limestone plateau on Wealden Beds and Kimmeridge Clay. The extent to which this should be seen as a development preceding the Roman conquest, and how far it should be associated with industry or hunting and fishing rather than with farming, is not wholly clear, but the evidence implies that it was under way in Iron Age 'A' times (fn. 13) around Poole Harbour, e.g. at Wareham, as well as in the Wealden valley and the lower Kimmeridge Clay slopes of S. Purbeck. (fn. 14) The peripheral distribution of sites on Bagshot Beds in the Poole Basin, where ancient field remains are absent despite good conditions for survival, and where the collection of shell-fish and industrial activities such as salt-boiling are evident, suggests that agriculture was desultory in much of this area. Settlements indicated by finds alongside the river Frome, however, can have profited from the more fertile alluvial soils, if adequately drained, (fn. 15) while a number fringing the S. margin of the heathland probably cultivated the strip of downwashed soils at the foot of the Chalk. In S. Purbeck however both Kimmeridge Clay slopes and the Wealden valley show traces of fields (Ancient Field Groups (22, 25–6)).

It will be seen from the distribution map that not many field systems are accompanied by earthworks of settlements, and it is likely that some of these establishments are to be recognized in the sites considered in this section, whether on the uplands as at Bagwood (Bere Regis (120)) near Ancient Field Group (32), or in the valleys. If this should be true of a site like Wilkswood (Langton Matravers (42)) below Field Group (28), we have an indication of the participation of a Romano-British agricultural community in the working perhaps of both shale and Purbeck marble, and we need not regard the many settlements in Purbeck showing evidence of the former as of purely or even mainly industrial character. Evidence from one of the few sites yet excavated systematically, at Eldon Seat (Corfe Castle (240)) near Field Group (22), points in the same direction although it refers largely to Iron Age 'A'. (fn. 16)

The buried sites offer most of the information at present available for a more particular dating of rural settlement in the region; about forty-one can lay some claim to an origin at least as early as Iron Age 'C', and of these at least fourteen can be placed certainly or probably in Iron Age 'A'. There is no reason to doubt continuity on these sites, although pottery characteristic of the regional 'B' phase is uncommon and stratified 'B' deposits are unknown in our area outside the hill-fort of Maiden Castle. (fn. 17) Thereafter most sites not indicated solely by burials seem to range over all or most of the Roman period, but the only examples to show positive signs of late 4th-century occupation are those at West Hill (Corfe Castle (232)), Wareham (85), Jordan Hill (Weymouth (445)), the temple at Maiden Castle (p. 501), and perhaps Whitcombe (26). The settlement at Bagwood (Bere Regis (120)) may have come to an end c. 350.

Imperfectly as they are known some of these sites yield direct evidence for agriculture in the shape of storage-pits for grain, as at Corfe Mullen. The stone-lined 'bee-hive' chambers, peculiar to the limestone of Portland and a sign perhaps of an insularity that has persisted to the present day, call for notice in this connection; carbonized grain was recovered from one of a series at King Barrow (Portland (101)) and their form suggests that they were intended for storage under sealed conditions. Instances of laterally and vertically conjoined pairs at King Barrow (Fig. p. 606) and of a subterranean passage leading to a chamber at Coombefield (105) imply storage of other commodities. They are presumably pre-Roman in origin.

There is little evidence in the area for the high farming associated with villa estates; only five recorded buildings can be regarded as villas, and their plans and history are largely unknown. One on the Chalk at Olga Road, Dorchester (212), lay unusually close to the Roman town; two near Weymouth (439, 447) suggest increased exploitation of clayey soils in the vales. The two remaining examples, at E. Creech (Church Knowle (57)) and Brenscombe Farm (Corfe Castle (229)), were situated some three miles apart on Bagshot Beds at the S. edge of the Purbeck heaths, and here the closely adjacent parts of the Chalk ridge, where conditions for survival have been excellent, show no trace of ancient cultivation. If these areas were within the domains of the villas they were probably used for pasture; the villas' arable will have been the narrow strip of downwashed soils at the foot of the Chalk worked by the existing farms (fn. 18) The fact that these villas seem to turn their backs on soils preferred by early farmers suggests that their owners were interested in the industrial potential of the heathlands in pottery and salt, while the Creech villa, like a few other heathland sites, may have been concerned in the shale industry.

Apart from the villas the most romanized settlement yet known, at Woodhouse Hill (Studland (46)), lies towards the broader E. end of this arable strip, about half a mile N. of Ancient Field Group (29). The site is highly complex with little stratification and with its earlier phases much disturbed by later building. Its interpretation in detail is open to doubt, but the community, established in round hut perhaps shortly before the Roman conquest, seems to have practised mixed farming and to have formed by the 4th century a hamlet or small group of rectangular cottages in which a workshop and perhaps quarters for a few favoured animals could be housed under the same roof as the human occupants. Rectangular buildings of less certain use are known or implied at a few sites by rough footings of local stone or remains of floors, most clearly at Encombe (Corfe Castle (235)) where a small building seems to have stood alone and out of conformity with the 'Celtic' fields in which it lay. At Fitzworth (Corfe Castle (226)) a small round hut seems to have been occupied as late as the 4th century.

Industries. The notable concentration of sites in Portland and S. Purbeck suggests that the presence of raw materials enabled these areas to support a larger population than might have subsisted by farming alone, although modern exploitation of both these regions, frequently disclosing antiquities, and in Purbeck a long tradition of archaeological fieldwork, doubtless overstress the contrast. Of the industries involved—stone, Kimmeridge shale, pottery and salt—the first three at least were of more than regional importance.

No Roman stone-workings are known, but quarrying of outcrops rather than collection of weathered stone began in Portland probably by the first half of the 2nd century A.D., on the evidence of an undated sarcophagus of hewn stone found in a cemetery apparently of that date at N. Common (Portland (99)). Quarrying of Portland stone may be assumed to have continued into the late 3rd or 4th century from like evidence at Poundbury (Dorchester (225e) and p. 537), but its place in the economic history of Roman Britain is difficult to estimate since stone of this type is not peculiar to Portland; it does not seem to have been preferred to Ham Hill stone at Dorchester. (fn. 19) Purbeck stone was probably quarried extensively at Upwey for the walls of Durnovaria (see p. 543) but the outcrop extends from Portesham on the W. to Durlston Bay, Swanage, and it is not possible to determine the exact source lithologically. The thin bands of the so-called marble of the Upper Purbeck, however, are distinctive and outcrop in a narrow belt on the lower slopes from Peveril Point to Worbarrow Tout, Tyneham, beyond which there are only negligible traces at Lulworth. This fossiliferous limestone was highly valued for monumental inscriptions, architectural veneers and decoration found as far N. as Chester and Lincoln, (fn. 20) and was used most notably in the pre-Flavian or early Flavian Cogidubnus inscription at Chichester and for the dedication slab of the forum at Verulamium in 79. Its use at Camulodunum before A.D. 60 indicates a promptness of exploitation all the more remarkable in that there appears to have been no pre-Roman industry to attract attention. Early prospecting, perhaps official, might well explain the relatively sophisticated pottery at Wilkswood (Langton Matravers (42)), situated on the marble outcrop and furnishing the only instance of worked marble of Roman date yet found in Purbeck. Other sites within yards of the outcrop, besides one newly discovered at Dunshay (p. 620b) and another unlocated at Lynch (p. 596b), are at Blashenwell (Corfe Castle (234)), and Worbarrow Bay (Tyneham (40)); this site has yielded an unfinished mortar of 'burr-stone' from the Broken Shell Limestone stratum closely associated with the marble, but in this sector the latter is perhaps too thin to be usefully worked. Dunning has suggested (fn. 21) that the latest evidence for the industry is to be found in the small, lugged mortars in marble or 'burr-stone', with straight sides, dated to the 4th century and reaching Corbridge (Northumberland); the column capital from Jordan Hill (Weymouth (445)), however, if not reused, should indicate that the industry was still capable of supplying stone for major monuments at this late date.

The manufacture of armlets from the bituminous Kimmeridge shale in the Iron Age and Roman periods has been studied by J. B. Calkin, (fn. 22) who has shown that the material, though liable to fracture, was very tough and intractable when freshly quarried, and better worked on the lathe with hafted flint chisels than with metal tools; the ornaments were finished by grinding and perhaps polished with way to resemble jet. Sites of the earlier hand-cut armlet industry, originating apparently in Iron Age 'A', (fn. 23) are betrayed mainly by the presence of roughed-out rings and discs, and unspecialized or semi-specialized flint flakes; those of the later lathe-turning industry mainly by specialized flint flakes and waste shale cores, of several well-defined types, formerly known as 'coal-money'. (fn. 24) It is evident that usually two armlets of different sizes were turned from each prepared piece of shale.

Armlets from Maiden Castle, and from Glastonbury, Meare and Cheddar in Somerset, suggest that the origin of the lathe industry should be sought in Iron Age 'B' or 'C', although no lathe-working of certainly pre-Roman age has yet been identified in Purbeck or elsewhere. Prototypes of the standardized flint lathe-tools, however, have been found at Green Island (Corfe Castle (224)) in association with a typologically early and rare variety of lathe-core (class A), which may belong there and at Gallows Gore (Worth Matravers (40)) to Iron Age 'C'. Turned armlets, moreover, appear at Wilkswood (Langton Matravers (42)) in an early post-conquest deposit. Survival of the older technique in some places in the 'C' phase is suggested by waste showing use of iron chisels, as at Green Island, while the essential continuity of the industry is shown by the number of sites where both rough-outs in the Iron Age 'A' tradition and lathe-cores occur. The growth of the industry in the Roman period, especially perhaps from the early 2nd century, is marked by the vast preponderance of lathe-cores of class C, and it clearly lasted into the 4th century at least, together with its specialized flint industry.

There is little evidence for the organization and equipment of the workshops, although, as has been remarked above, there is reason to suppose that shale-working was not the sole activity of the communities involved. At Gallows Gore in the Roman period low stone blocks and perhaps also lengths of drystone wall may have served as bases for pole-lathes, while at Kimmeridge Bay ('Gaulter', Steeple (31)) similar walls were found in a secondary Iron Age 'A' horizon, although less certainly associated with shale; at Encombe (Corfe Castle (235)) a rectangular 'cottage' standing amidst 'Celtic' fields seems to have been occupied by shale-workers. There is no sign of any substantial return in the form of money or imperishable luxuries. The sites yielding waste from the armlet industry, of which over thirty are now known, (fn. 25) are not confined to the vicinity of the outcrop, and include Green Island and Hamworthy (Poole (402)) in Poole Harbour, as well as a number of sites on the S. fringe of the Tertiary heaths where, as has been suggested, it is likely that villa owners were to some extent concerned with the industry. The raw material in block or roughed-out form may well have been exported to be worked up further afield, as perhaps at Maiden Castle, Hengistbury Head (Hants.), Glastonbury and Meare. It has been assumed that the best 'coal' or 'blackstone', outcropping as a thin layer only in the cliffs near Kimmeridge Bay, was alone exploited, but it is likely that lower quality Kimmeridge shale occurring there and formerly accessible at Portland was also used; characteristic flint lathe-tools, though not the shale, have recently been found at Abbotsbury in W. Dorset. (fn. 26) Other superficially similar materials, such as lignite and cannel-coal, were also widely used in Britain in Roman and earlier times, and increase the difficulty of assessing the full importance of the Purbeck industry for the Roman province, at least in satisfying the demand for ornaments and spindle-whorls.

Other less well known products of the shale industry may have been equally important. The occurrence in S.E. England (fn. 27) of the graceful Belgic lathe-turned pedestal urns presumed to be of Kimmeridge shale may show export of the raw material, since neither they nor their equivalent in pottery are known in Dorset, but fragments of turned shale bowls or dishes, rarely figured in the literature, are common in Roman deposits in the region. Shale furniture has been better studied in the whole or fragmentary carved table legs found mostly in and near Dorchester but also as far away as Verulamium, Rothley (Leics.) and Caerleon. (fn. 28) The distribution of these objects, and of the bevelled oblong plaques or tablets incised with geometric decoration at centre and edges, and sometimes identified as wall or furniture panels, (fn. 29) entitles us to assume that the material thus widely valued was shale from the Kimmeridge formation. The sites of manufacture have not been traced, although Norden (Corfe Castle (230)) was perhaps one. There was also a limited use of shale, if not exactly for building at least for lining or covering graves, for mosaic cubes, and square floor-tiles as at the Preston villa (Weymouth (447)) and Bignor (Sussex).

The pottery and salt industries are to some extent related. Site finds suggest that the regional pottery industry was concerned to a very large extent with the production of vessels of black burnished heatresistant kitchen ware. (fn. 30) Only two kilns are known, at East End (Corfe Mullen (24)) and near Cleavel Point, Ower (Corfe Castle (227)), the latter of unusual horizontal-draught type. Pottery-making, however, is possible or likely at several other known sites on Bagshot Beds in the parishes of Arne and Corfe Castle, especially perhaps at Stoborough (Arne (50)) where indications, in addition to what may be a clay-puddling basin, seem extensive. (fn. 31) The sites now visible are known mainly from surface debris consisting of oxidized brown or reddish sherds from vessels of burnished kitchen ware intended to be reduced in firing to a black colour; the sherds are regarded therefore as probably belonging to wasters, although, like those from the known kiln at Corfe Mullen, which are also oxidized, they do not seem to include the contorted or fused examples often found beside kilns. Several of these sites, however, also yield evidence for salt-boiling, implying combustion on a large scale probably in open conditions, which could have caused oxidization of sherds of domestic vessels lying about in the vicinity; the distribution map therefore acknowledges only proven kilns, although it is clear that at Ower, as in some Romano-British saltings outside the county, both activities were carried on.

Although there is evidence nearby for the production of black burnished vessels in the late 3rd or 4th centuries, the kiln at Corfe Mullen belongs to the mid 1st century A.D. It is anomalous in that its products fit only partially into the conspectus of regional wares. The flagons and mortaria are romanized types while tnote ids f Durotrigian shape (classes C and E) are small, relatively delicate wheel-thrown or wheel finished versions of the native types. The kiln was doubtless worked by a native potter, but it is probable that production was instigated by the army during the early years of the conquest; it stood by an early military route, and its products have not been identified on occupation sites save one evidently of military character recently discovered near by at Lake Farm (see p. 529). The Ower kiln produced vessels of normal black burnished fabric, but since types in the Durotrigian Iron Age 'C' tradition were stratified here and in the surrounding area with types current in the 2nd and 3rd centuries and even the 4th century, it is likely that more than one kiln was involved. The absence of any indication of manufacture nearer Dorchester is remarkable, although there is indirect evidence there for a tile-kiln (see p. 537).

Evidence of the crystallization of salt from sea-water in prehistoric and Roman Dorset comes entirely from coastal areas within the region studied in the present volume, and has been more fully discussed elsewhere. (fn. 32) The sites, indicated by the presence of 'briquetage' (hand-moulded clay props and pieces of crudely made containers, the latter necessarily fired before use), occur (i) on the Tertiary heaths around Poole Harbour, and (ii) as a small group, mainly on Kimmeridge Clay, from Kimmeridge Bay to Tyneham Cap, in which shale was used for fuel. One site, which probably used shale from Portland, is known further W. on the Fleet at Wyke Regis (p. 512), and two more are suspected, at Jordan Hill near Weymouth, and Osmington. There are no mounds of burnt earth such as are elsewhere associated with saltings, but mounds of shale and ash are known at two adjoining sites in Tyneham parish (41–2).

In view of the presence of a pottery-kiln at the salting in the Ower peninsula, Corfe Castle, evidence for structures of burnt or fired clay in connection with the industry has to be treated with caution, but a floor of fired or fire-hardened clay at Shipstal (Arne (54)) and rounded ovens at Hamworthy, of Roman date, may belong to it. A narrow rectangular stone-lined hearth below present high water mark on the foreshore of the Fleet, Wyke Regis, seems to show salt-boiling with shale fuel late in Iron Age 'B' or early in the 'C' phase. A small quantity of'briquetage' was found, but domestic jars of a uniform type may have been used as containers for drying and marketing. (fn. 33) Elsewhere the only positive evidence for the pre-Roman industry is at Gaulter, Kimmeridge Bay (Steeple (31)), where it belonged to Iron Age 'A' and was associated with small semi-cylindrical clay troughs and small bowls. Some of the Romano-British saltings may nevertheless have had their origin in Iron Age 'C'. Here the evidence is for two mutually exclusive varieties of container: heavy sub-rectangular flat trays, confined with the exception of Hamworthy to the Kimmeridge group, and what seem to have been large thin-walled semi-cylindrical troughs, peculiar to the heathland group; hand-moulded props are common to both.

The sites on Kimmeridge Clay, though coastal, are nearly all on the cliffs at 50 ft. or more above sea level. The heathland sites are near the water's edge, except at Godlingston Heath (Studland (45)), 1½ miles inland at 300 ft.; here the most reasonable explanation would be the manufacture of boiling-vessels, or perhaps the refining of impure salt by boiling in fresh water.

Temples, shrines and cemeteries. The persistence of native tradition apparent in material aspects of life is no less so in religious and burial ritual, as far as these can be revealed archaeologically. The late 4th-century pagan temple at Maiden Castle, and the building of like date interpreted as another temple at Jordan Hill (Weymouth (445)), share with the villas an architecture of competent provincial type; but at Maiden Castle the 4th-century round hut, built within the largest prehistoric hut known there and reasonably interpreted by Wheeler as a primitive shrine, suggests the continuity of a pre-Roman cult despite an apparent gap of three centuries. Jordan Hill affords a prime example of the ritual shaft or pit common amongst the Celts (fn. 34) and associated there either with the supposed temple or the cemetery; a like significance may be suspected for simpler pits and cists containing animal bones and ashes in this cemetery, and in others at The Grove (Portland (102)) and Dorchester (218b, d). Devotions of a more Roman kind, presumably at domestic shrines, are implied by the niche in the town house at Colliton Park (Dorchester (182)) and perhaps by a small uninscribed altar of Purbeck stone ploughed up at Kingston (Corfe Castle (238)). The relief from Whitcombe (26), however, if not a tombstone, would suggest an E. European pagan cult of the mounted hero or demigod, acclimatized in its rendering in Portland stone but not otherwise very clearly attested in Britain. Evidence for Christianity in the region here surveyed is confined to Dorchester (p. 536).

The regional burial rite in the Iron Age was inhumation, and this continued scarcely affected by Belgic or Roman practice of cremation, even in the capital; a cremation at Cogdean (Corfe Mullen (25)), the sole instance in our area outside Dorchester, where too it was exceedingly rare, (fn. 35) was probably of the early military phase. Known burials of Iron Age 'A' and 'B' are too few to determine a rule, but, as in the 'C' phase, the body seems normally to have been placed on its side in an oval grave, with limbs contracted or flexed. Practice in Iron Age 'C' is best shown, despite the special circumstances, in the 'war cemetery' of c. A.D. 44 at Maiden Castle, where food offerings and vessels of Durotrigian black burnished ware accompanied the dead. In the Roman period the body was usually fully extended on the back, with arms crossed or by the sides, but flexed posture was not uncommon and at Weymouth (438) and (441) could apparently be late in the period. Personal possessions are not often recorded, (fn. 36) and imperishable grave-goods, where present, consist normally of one or occasionally several black ware vessels. In the 1st century at least these would mostly be standard Durotrigian types (fn. 37) and will not alone suffice to fix a date before or after the Roman conquest, as the graves at Whitcombe (26) most clearly show. The rarity of 3rd and 4th-century vessels with burials must imply a decline in offerings. Unaccompanied burials in extended posture, occurring so often amongst known Roman graves, have been treated as Roman in the Inventory; some few may be Saxon, before burial in churchyards became normal, but doubtless more really belong to British communities spanning the period of some 250 years between the end of Roman rule and the Saxon settlement of the region in the 7th century.

Wooden coffins, doubtless introduced in the Roman period, are quite common; the earliest known, at Maiden Castle, with a female skeleton with hob-nails at feet, was probably buried before c. A.D. 70. (fn. 38) Cist graves, lined and covered with stone slabs or occasionally shale, could enclose wooden coffins and are common except on the Chalk, where graves could be cut in the rock; they need not therefore reflect a distinct tradition. It is reasonable to suppose that short cists containing flexed skeletons can be pre-Roman, but there are no proven examples here, either short or long, for the Iron Age. Sarcophagi, of Portland stone with ridged lids, occur in Portland and at Dorchester and neighbourhood where, however, most are flatlidded and of Ham Hill stone; they may contain wooden or plain lead coffins but none is reliably reported as having grave goods.

There is no evidence for walled cemeteries except possibly at Jordan Hill (Weymouth (445)). Groups of burials rarely display system, although there are exceptions at Bare Cross (Church Knowle (59)), Charborough Park (Morden (57)) and Afflington (Corfe Castle, p. 443), as well as at Dorchester; (fn. 35) the Afflington Barrow furnished one of several examples of intrusive burials of the period in Bronze Age barrows. Orientation of individual graves in an E.-W. direction seems to occur in about two cases in every three, with head as often at one end as at the other; a few graves contain more than one individual, usually side by side, and the heads could be at opposite ends. Notable instances of the careful interment of bodies with lower jaw or whole skull removed and placed separately are recorded at Kimmeridge (Steeple (31)) and St. Nicholas's Church (Studland (47)).


  • 1. For Hod Hill see Dorset III, forthcoming, and Sir Ian Richmond, Hod Hill II (1968); for Waddon, see Dorset Procs. LXXXII (1960), 88–108, and LXXXVI (1964), 135–49; for Lake Farm see below, p. 529.
  • 2. Arch. J. cxv (1958), 57, 95; G. Webster and D. R. Dudley, The Roman Conquest of Britain (1965), 103–4, 109.
  • 3. J. G. Milne, Finds of Greek Coins in the British Isles (1948).
  • 4. It could alternatively have been sited near the Wareham Channel on the S. side of the peninsula, in which connection see p. 531, n. 1.
  • 5. A contrary view is discussed below (p. 528).
  • 6. For indications of a division of the tribe, at least in the 3rd or 4th century, between Dorchester and Ilchester, the only other walled town of the Durotriges, see pp. 531–2.
  • 7. There is evidence for the survival of a British community in this area in the early Christian inscriptions at the church of Lady St. Mary (see p. xlix).
  • 8. A road connecting Dorchester with the Isle of Purbeck is the most obvious omission.
  • 9. R. Good, A Geographical Handbook of the Dorset Flora (1948), 9–12.
  • 10. The region is slightly more extensive than the area of surviving heath indicated in the distribution map facing p. 634.
  • 11. The suggestion has been made for other regions of 'Wessex'; see Charles Thomas (ed.), Rural Settlement in Roman Britain (C.B.A. Research Report no. 7, 1966), 54–6.
  • 12. The total includes the Olga Road 'villa', Dorchester (p. 570), the temple complex at Maiden Castle (pp. 500–1) and a newly discovered site at Broadmayne (p. 594). If some 23 sites indicated solely by burials are excluded the ratios are not significantly affected.
  • 13. Bronze Age settlement on the heaths, although largely in different areas, is suggested above (p. xxxvi). Two sites of Iron Age 'A' in S. Purbeck show pottery with Late Bronze Age antecedents (Eldon Seat, Corfe Castle (240), and Sheepsleights, Worth Matravers (p. 512), the former on Kimmeridge Clay).
  • 14. At sites where the earliest known material is in the Durotrigian tradition of Iron Age 'C' the evidence is often insufficient to determine a date before rather than after the Roman conquest.
  • 15. The same should be true of the Stour valley (the storage of Spelt wheat is attested at East End (Corfe Mullen (24)) on Valley Gravel).
  • 16. Cf. Sheepsleights, Worth Matravers (p. 512), where occupation involving grain-storage and shale-working seems, however, to have ceased before Roman times.
  • 17. Iron Age 'C' occupation has been assumed, unless there are reasons to the contrary, wherever sites of native character appear to have existed already in the early Roman period; for Iron Age 'A', only sites with stratified levels or with unambiguous 'A' pottery have been included. For the currently accepted dating of Iron Age cultural phases in the region, see p. 495.
  • 18. Cf. L. E. Tavener, Report of the Land Utilization Survey of Britain, part 88, Dorset (1940), 283.
  • 19. The use of other local stones, and of Ham Hill stone from Somerset, is discussed on pp. xxxix-xl.
  • 20. The industry has been studied by G. C. Dunning who has kindly made his distribution map available (see Archaeological News Letter I, no. 11 (1949), 14–15).
  • 21. B. W. Cunliffe (ed.), Richborough V (1968), 110–11.
  • 22. Dorset Procs. LXX (1948), 37–40, 55; LXXV (1953), 43–71; LXXXVIII (1966), 114–15.
  • 23. The shale was also used for ornaments in the Neolithic and Bronze Ages.
  • 24. The latter could be hoarded, perhaps for supposed talismanic qualities held in common with jet and other substances, and are not alone sufficient to prove manufacture on the site. Old records of the ritual use of shale in Purbeck lack confirmation.
  • 25. Several have been added to Calkin's appendix 1 in Dorset Procs. LXXV: the settlement on Knowle Hill (Earthworks, Settlements, Church Knowle (54)), where working in Iron Age 'A' is proved, Corfe Castle (238–9) and Worth Matravers (42–3). Hamworthy (Poole (402)) and E. Creech (Church Knowle (57)) should probably be included; Calkin's site 15, Whitecliff Farm, Swanage, where evidence for ancient occupation seems unfounded, should be omitted.
  • 26. Dorset Procs. LXXXIX (1967), 121–3.
  • 27. As at Old Warden, Bedfordshire (Later Prehistoric Antiquities of the British Isles (B.M., 1953), 71. pl. XXIII, 3).
  • 28. J. Liversidge, Furniture in Roman Britain (1955), 37–47, 51. A curved slab of Kimmeridge shale from Bagber, Milton Abbas, in D.C.M., is doubtless part of the circular top of one of these three-legged tables.
  • 29. Fragments occur at rural sites such as Bagwood (Bere Regis (120)), as well as in more romanized centres (e.g. in London). They are known to have served as trays for vessels in burials at Jordan Hill (Weymouth (445)) and Winchester (Ant. J. XLVII (1967), 230–4, 248–50, where they are called trenchers).
  • 30. Some early types must be regarded as at least the progenitors of one of the most important strains of kitchen ware in Roman Britain, most fully studied on northern military sites in deposits dating from c. 120 by J. P. Gillam (Arch. Ael. xxxv (1957), 180–251; Procs. Society of Antiquaries of Scotland XCIV (1960–1), 126–9).
  • 31. One formerly accepted kiln site, Hamworthy (Poole (402)), has had to be removed from this category.
  • 32. Dorset Procs. LXXXIV (1962), 137–144. Turlin Moor (Poole (403)) may perhaps be added to the sites listed therein, while new evidence from Gallows Gore (Worth Matravers (39)) suggests at least trade in salt.
  • 33. Ibid. 132–6.
  • 34. Cf. Anne Ross, Pagan Celtic Britain (1967), 24–8.
  • 35. For an analysis of burials in the town cemeteries, see p. 571.
  • 36. Notable exceptions are at Weymouth (446), and at Whitcombe (26) where the armed warrior must surely have been buried before the Roman occupation.
  • 37. The commoner types are listed by J. W. Brailsford, P.P.S. XXIV (1958), 103.
  • 38. Similar indications of footwear are not common but have a wide distribution in the region.