<strong>and</strong> so may not be detectable without excavation or geophysical survey. O<strong>the</strong>r forts may be grouped at present amongst <strong>the</strong> sub-rectangular enclosures of <strong>the</strong> county. These sub-rectangular earthworks, assumed to be a variant of rounds, were first summarised by Dudley (1954); known numbers have been recently increased by study of existing air photographs (Johnson <strong>and</strong> Rose, 1982). These enclosures are scattered throughout <strong>the</strong> county, <strong>and</strong> occur in Devon, but with concentrations around <strong>the</strong> Helford River, <strong>the</strong> upper Fal valley <strong>and</strong> in north <strong>Cornwall</strong>. Trevinnick, near Wadebridge, was excavated by Fox <strong>and</strong> Ravenhill (1969) to check whe<strong>the</strong>r it was a Roman military site, but proved to be an enclosed civilian settlement of <strong>the</strong> Roman period, not closely dateable. Grambla, near Wendron, a virtually square earthwork enclosing c. 0.35 ha in <strong>the</strong> Helford River area, was partly excavated in 1972 (Saunders, 1972). This also had been a civilian settlement, occupied from perhaps <strong>the</strong> 2nd century AD until <strong>the</strong> 6th; it contained distinctive oval houses. Its ditch was 3m deep <strong>and</strong> its entrance had a quadrilateral setting of post-holes suggesting a gate tower. The 2 ha rectangular site at Carvossa, near Probus in <strong>the</strong> Fal valley, was excavated from 1968 to 1970 (Douch <strong>and</strong> Beard, 1970; information from P.M. Carlyon). It produced a rich range of finds, starting with South-Western Decorated pottery. The samian sequence began with a single Tiberio-Claudian sherd, increased in <strong>the</strong> Neronian period, <strong>and</strong> continued with some fluctuation until <strong>the</strong> mid-second century Antonine period. The coin list <strong>and</strong> o<strong>the</strong>r imports reinforce <strong>the</strong> pattern of <strong>the</strong> samian, suggesting that occupation still flourished in <strong>the</strong> Hadrianic period <strong>and</strong> subsequently spanned <strong>the</strong> 2nd <strong>and</strong> 3rd centuries AD. The range of finds for <strong>the</strong> Neronian — Early Flavian period would be entirely appropriate for a military site. The stratigraphy was not clear <strong>and</strong> <strong>the</strong>re is some evidence for at least two earthwork phases. The main 2 ha earthwork had a ditch 4.5m deep which was not of military type, nor was <strong>the</strong> excavated east entrance. Features found in <strong>the</strong> interior, such as curvilinear gullies, suggest civilian occupation. The ditch had silted by at least 1.5m beneath <strong>the</strong> lowest samian sherd. The rectilinear earthwork may be suggested to be of pre-Roman date, perhaps as early as <strong>the</strong> 1st century BC. The finds indicate that activity continued throughout <strong>the</strong> suggested period of military occupation of <strong>Cornwall</strong>. A fort may have been established in part of <strong>the</strong> earthwork complex, in which case most of <strong>the</strong> artefacts would derive from an external vicus. Perhaps some trusted chief was encouraged to develop at Carvossa a focus for trade up <strong>the</strong> Fal valley. Alternatively, <strong>the</strong> finds may indicate a fort in <strong>the</strong> vicinity, perhaps within <strong>the</strong> substantial earth-work of Golden one mile to <strong>the</strong> south. If Roman military occupation in <strong>Cornwall</strong> is demonstrated in future to be slight compared to that in Devon, a network of trusted local chiefs could be <strong>the</strong> reason. Ano<strong>the</strong>r local chief was allowed to continue occupation at St Mawgan-in-Pydar, where <strong>the</strong> range of small finds, imports <strong>and</strong> elaborate gabbroic pottery is second only to Carvossa. Carvossa certainly attracted traded artefacts. It also produced a rich range of gabbroic pottery, Cordoned wares <strong>and</strong> innovative pieces copying items such as a Dr 29 samian bowl <strong>and</strong> a bronze h<strong>and</strong>led patera. There was also evidence for metal-working, especially iron smelting in <strong>the</strong> ditch. If Carvossa was at one stage a military post, it will have been h<strong>and</strong>ed over to a local group c. AD 75. Here would be happening, on a small scale, <strong>the</strong> regular practice of turning over military bases to serve as local centres of administration <strong>and</strong> trade under Roman supervision. At Exeter this process led to <strong>the</strong> establishment of a cantonal capital. Isca Dumnoniorum, which administered both Devon <strong>and</strong> <strong>Cornwall</strong> (Bidwell, 1980, 56). There must have been a series of small centres throughout <strong>the</strong> South West. As no formal Roman building, apart from <strong>the</strong> Magor villa known since 1931, has been found in <strong>Cornwall</strong>, it may be presumed that such centres were generally developed in <strong>the</strong> local tradition. In <strong>Cornwall</strong> <strong>the</strong>se may be looked for amongst <strong>the</strong> more substantial small enclosures, where <strong>the</strong> descendants of chiefs — 122
whose ancestors had occupied hillforts — administered <strong>the</strong> fur<strong>the</strong>r parts of Dumnonia. Such centres would have been useful as local tax collection points, as <strong>the</strong> capital of <strong>the</strong> responsible civitas at Exeter was so distant. Trethurgy Al Trethurgy T2 Ft 10 20 310 5 10 Chysauster 5 Castle Gotha Fig 3 Trethurgy Houses Al <strong>and</strong> 72, late 2nd century AD; Chysauster Courtyard House 5, 3rd century AD (after Hencken); Castle Gotha Oval Hut, 2nd century AD (after Harris). 123