The Roman villa is one of the most interesting in Britain and one of only four known in Devon. (The other three are not far west, at Seaton lower down the Axe valley, Membury higher up it, and Budleigh Salterton along the coast; perhaps Roman culture hardly penetrated past Dartmoor.)
The site is now grassed over, and you can't see it from down in the Cannington lane: it's half way up the hillside to the west, on a shelf of nearly level ground or, one might say, the truncated top of a spur between two side-valleys. (This oddly smooth pocket of ground can be noticed among the contours on the Ordnance Survey map. Perhaps it is part of a geological terrace on the valley side. Looking across from the road up Woodhouse Hill, you can see it and other ledges that might be parts of the terrace. They are roughly level with the foot of the old railway viaduct in the background.)
The villa is nearer to the small side-valley on the south, but is in a field belonging to Higher Holcombe Farm in the larger side-valley to the north, so it is called the Holcombe villa.
There was for centuries along the field's western (upper) edge a hundred-yard-long heap of ruins, overgrown with brushwood and trees, completely enclosed, and avoided as the haunted churchyard or Church Ground. (That it was the ruin of a church was probably suggested by the outline of what turned out to be the villa's bath-house, with its chapel-like apses.) In the early nineteenth century, according to the memory of an old villager named Mansfield, farmer Gay removed hundreds of cartloads of worked stones for building a farmhouse. A visitor, Mr. Bandinel of the Foreign Office, picked up a a brick and roofing stone and suggested the presence of a Roman villa.
In 1850 farmer Bartlett opened a gateway into the enclosure to get at some of the stones and make use of the earth for dressing; and presumably this was when he invited the Rev. George Tucker to investigate. In August of that year Tucker excavated a limited area (publishing the results in 1854). He found the octagonal bath house, and, a dozen yards north of it, an elaborate mosaic pavement. This, left exposed, disappeared except for scattered tesserae; luckily in 1852 William Newbery had made a meticulous water-coloured drawing of a quarter of it, showing many concentric geometric shapes with a few floral ones. According to local newspapers the site for a short time was open to public view, at the trifling charge of 6d. each, then closed owing to some family litigation between the owners of the farm, covered up, but reopened, the fee being now one shilling. Some more investigation in 1855 uncovered the heating flues of what was then thought to be the Roman military station.
In 1870 Captain John Swann observed that the farmer was making a wholesale clearance of the site much I am sure was lost in moving away many cartloads of earth and spreading it over neighbouring fields so he rented what remained of the copse at £5 per annum and, with two or three helpers, dug for nine weeks, clearing an area 64 by 35 feet to a depth of four and a half feet, finding among much else a fine limestone pillar just under three feet high.
For the next century, all above-ground obstruction having been cleared, the field was under the plough, destroying most of the stratified material above floor level, and much of the buried walls and floors too. This is why, when excavation was at last done scientifically, so much had to be deduced from indirect clues the nature of vanished walls and roofs, for instance, from foundations and fallen fragments.
(What does it matter whether material is still stratified? In archaeology, as in detective work, context is everything. You'd be amazed at the detail and sometimes the world-importance of what can be deduced from knowing exactly where a thing was found. Stratification the sequence of layers, and the exact relation of objects to them is the most important component. An ancient bone, pot, or coin may be pretty, but almost all the information it could yield is lost if its context is lost. Never disturb it; leave it for the experts. Even the experts have now learned to leave the largest part of any site for the more expert experts of the future. The major archaeological tragedy in Iraq is not the loss of beautiful statues, all well recorded, but the ongoing plundering of unexcavated layered ancient sites.)
In 1967 part of a tessellated pavement was ploughed up; so archaeologists decided on emergency excavation. Before it could start, the site was ploughed again, and rough grass again grew over it. There were three seasons of work, 1969 to 1971. Farmer L.J. Denning co-operated, and his son Gerald was foreman in the latter two seasons. In the first season the excavators used only manual labour, but after that, to speed the work and get at a larger area, they availed themselves of a mechanical digger to remove the ploughsoil, in which any remains would have become unstratified (jumbled).
The director of the excavation was Sheila Pollard (who also in 1969 worked with Henrietta Miles in excavating the Honeyditches Roman site at Seaton). Her report of over a hundred pages, A Late Iron Age Settlement and a Romano-British Villa at Holcombe near Uplyme, Devon, was published in Proceedings of the Devon Archaeological Society, No. 32, 1974, and reprinted by the Philpot Museum of Lyme Regis.
The farm now belongs to John Duffin, who kindly escorted us to the field, even lending us Wellington boots since we had come unprepared for the muddy ascent.
Excavation is like any probing into the past: you discover the clues from the top down, to a story that happened from the bottom up. Back in the late Iron Age prehistoric period (called by archaeologists Iron Age C), people presumably of the Durotrigian tribe built two circular huts, each about 25 feet wide, with walls consisting of double lines of posts, probably wattle-and-daub between them, and probably thatched roofs; two had narrow twisting porches to keep the weather out. These were succeeded by another two round huts, in the middle of a roughly rectangular area defended by a ditch more than five feet deep and eight feet wide, perhaps originally with a bank on its inner side. A gated causeway across the eastern ditch led to an outer enclosure, probably for livestock, with a slighter ditch. Three troughs cut in the ground contained slag, charcoal, and heat-reddened soil, and must have been furnaces for roasting iron ore. Potsherds of known Durotrigian types, found in the ditches and three pits, date from the latest pre-Roman period, 20 to 44 A.D. It is not quite certain that the settlement was pre-Roman some native pottery continued in use after the conquest but the defensive ditch would probably not have been needed, or allowed, under Roman rule.
At the bottom of one of the pits was a mirror: a nearly circular thin sheet of bronze ten inches wide, with a rim and a five-inch handle, the whole weighing 800 grams (nearly four pounds). In its somewhat cloudy surface (polished probably by rubbing with wood ash) a Briton lady will have seen, according to the Roman poets, a painted face and hair dyed blue. On the back, almost indiscernible till it was carefully restored, was engraved an intricate, symmetrical, wonderfully flowing design of two lyres, one inside the other, embellished with crescents, scrolls, and peltae (three-curved shields), and filled with two kinds of hatching. The accuracy of the curves implies that they were first scratched with a compass. The handle was cast in the form of several rings and collars, and joined to the mirror with a complex mount which, if the mirror is hung from the handle, suggests a feline face, with two red copper studs for eyes, and the two trumpet scrolls around them forming ears and muzzle. At least sixteen decorated mirrors of this general style, peculiar to southern Britain, are known; the Holcombe Mirror (bought by the British Museum and there is a copy in the Lyme Regis museum) is one of the two or three best. It seems that quite rich Celtic farming squires lived in humble houses. It could hardly have been an accident that the mirror (possibly wrapped in material that later decayed) was at the bottom of the pit, which was then seemingly filled in one action; a theory is that it was hidden as the Romans approached and the inhabitants fled to one of the hill forts.
After the conquest, the site was apparently unoccupied for a while; the next dateable pottery was from about 70-85 A.D. Post-holes (traced with much difficulty) show that four small rectangular timber-framed houses were built, in no certain order except that one succeeded another on the same spot. The ditch around them was allowed to silt up, and the rubbish in it, including Durotrigian pottery, fine samian (imported) pottery, bronze brooches, a nail cleaner and tweezers, suggests some prosperity. Perhaps a Durotrigian family continued to occupy this farm.
Around 180-200 A.D. the little houses were replaced by a house 56 feet long, with stone foundations (packed flints). Two north-south rows of square posts divided the main room into aisles, and probably supported a pitched roof (of which some fallen tiles survived); a smaller room was partitioned off at the south end. Refuse trapped when new clay floors were laid included a coin of Marcus Aurelius (A.D. 170). Eastward from this house was added a large irregular outbuilding, with its north side and east end positioned over the old ditch. It may have been kitchen, farm workers' quarters, stable, barn, or all of these. It too was divided up by internal posts, but it had no stone foundation. It was enlarged several times but rather soon abandoned. No later outbuildings were found; perhaps, after this experience of trying to maintain such relatively flimsy structures on the exposed hillside, they were sited down in the valley.
In the next phase (the archaeologists' Roman phase III), the rather small main house was extended south, across the old ditch. The first two small rooms had to be built on top of the ditch, and one of them was probably just a corridor, leading to the larger third and fourth rooms. The flints of the foundations were now set in yellow mortar, and there were no wooden pillars. It seems the family moved into these better-built bedrooms and gave over the house's earlier area to farm uses, such as a grain-receptacle and another ore-roasting pit.
Later in the third century (phase IV) there was much addition at the other end. The north wall of the old house was demolished, and four new rooms extended northward to the edge of the old ditch, the second being a cross-passage but the third becoming the house's principal room; then on both sides were long rows of narrow rooms perhaps partitioned corridors or verandahs extending along the sides of the old house also. The foundations were made with flints not mortared but rather carefully shaped and laid, broad ends outward. Now there were floor tiles of buff-pink brick (with knobs underneath to key them into the ground); wall plaster painted bright pink with white, black, and maroon stippling; heavy hexagonal roof slates (some still had nails through them); greenish window glass. Also limestone steps; the base of a limestone pillar or pedestal; a probable altar base; a gully of unknown purpose running from room to room under the walls; a burial of a newborn baby. Also shells, a lead steelyard weight, an iron bucket handle, hobnails, hairpins, and coins from 260-273 A.D. (as well as older ones still in circulation).
Phase V, in the fourth century, was when the villa became as grand as any in southern Britain. Beyond the new rooms at the south, two more were added, the first being probably the one described by Tucker as floored with square red tiles; the second and larger was almost certainly the one with Tucker's great mosaic. A corridor or verandah was built along the east side of these; later another stretch, to connect this with the earlier (and narrower) corridor, so that the whole now ran entirely along the eastern face of the building. Foundations were now deeper and wider, though unmortared, so the walls above must still have been timbered. Fragments of plaster show that some walls were painted plain deep maroon, others with fine white stippling on a lighter red ground. There were hexagonal roof tiles of a hard grey lias probably quarried nearby at Uplyme; and softer limestone ridge tiles with an angle (130 degrees) indicating a low-pitched roof.
The side corridor projected southward past the new rooms to the most luxurious feature: the octagonal bath, or rather bath-house cluster, since it was surrounded by eight rooms.
It was entered through one of them, which must have been the apodyterium where bathers undressed. (A coin of Constantius II (330-335) had fallen into the yellow mortar foundation of this room's floor while it was still wet.) The next room to the right, dug three feet lower, was the furnace, as shown by heat-reddened clay and remains of charcoal. From it led under-floor hot-air flues to the other rooms (and fallen box tiles indicate flues running up the walls also). Going on counter-clockwise, the next rwo rooms had apses (semicircular bays) and would have been the calidarium and tepidarium, hotter and cooler steam rooms. The floors (with red tiles) were hypocaust: that is, supported over the hot-air space by pilae or squat pillars (made of fragments mortared together, including re-used roof tiles from earlier periods). Next, at the south, came a rectangular room; next, a small irregular one, probably the latrine, through which the flint-lined drain from the central bath ran to a sand-filled soakaway outside, then to the main storm drain that came around the building from the furnace. Next, a large room at the east. This, onto which the eighth and last room had been joined by removing a wall, also had an apse; and through the base of this was an opening to a pit outside. Apparently the intention, since this area was so far from the furnace, had been to make in the pit a second furnace; but this was never done, and the pit was still open (as shown by roof tiles fallen into it) at the end of the building's life. The three eastern rooms had been spared by the earlier excavators, and so more of their foundations had also lain below the destructive reach of ploughing.
In the middle was the plunge bath itself, where the bathers finished with a cold dip. For the 1967 excavators little was left but the fawn-colored tesselated floor (which sloped gently to the drain) and a few tesserae up the sides of the bath. From the earlier descriptions, and Newbery's 1852 drawing, it is known that the bath was three and a half feet deep and twelve feet wide, with the width reduced on all of the eight sides except one or two (there is discrepancy here between the 1852 and 1974 accounts) by a step 7 inches high and 2 feet 10 inches high or a foot wide and two feet high capped with neat slabs of blue lias. The ambulatory around the bath, paved with a fairly complex mosaic, was only two and a half feet wide, so there would hardly have been space for seats. The debris in the bath house included thousands of shells, chiefly of the common snail and the garden snail. Evidently these were the esteemed delicacy; more abundant in the refuse of the kitchen were the shells of the humble winkle.
Though the aborted attempt to modify the bath house shows that it may have been too ambitious for its builders' skill, parallels to it in Roman Britain are few and partial (some octagonal pools, a hexagonal bath house). As comments Mrs. Pollard, such a structure would have been notable even in a large town of the period. Bathing was a lengthy process in Roman times, and the bath house can be visualised as being the social centre of the district. Here the proud owner would entertain his friends, who would sit around chatting whilst bathing and snacking on snails.
The villa had grown to be a 250-foot-long and relatively narrow range of about 28 rooms (including those of the bath complex). Later in the fourth century (phase V c), all the older parts were remodeled. Along the west side was built a new outer wall; along the east, the corridors were replaced by an open portico. The original house got a new mortar floor, and from it (as shown by postholes and flint foundation) a double door 44 inches wide opened to the portico; two limestone blocks now in a field wall lower down the hill may have formed the doorstep. In the extensions at the north, partition walls were removed, to make one large kitchen. Into its clay floors were cut (not all at the same time) nine cooking hearths with low walls around them, two ovens (one being a large jar laid on its side with a dome built around it), a small furnace, two rubbish pits, and another infant grave. Hearths, pits, the rake-out aprons around ovens, and a tip outside the north wall provided the archaeologists with shells, sheep and goat bones, a whetstone, some coins dating from 323 to about 370 A.D., and copious pottery (some black-burnished Durotrigian, some from more distant sources such as Oxford and the New Forest) pushing the dating into the fifth century. Outside the portico on the east was a rectangular pit about 8 by 6 feet and 18 inches deep, with sides and floor so smooth that the excavater said the fill peeled off them: it might have contained a metal water tank for kitchen use, filled by rain or hand (no pipes were found) and removed for its value when the villa was abandoned: the pit must have been still open, since debris from the final collapse was inside it.
There was no sign of fire or violence, but for thirteen centuries the ruins were left to decay. There was no evidence of later dwellings among them, but at uncertain times people re-used the furnace room of the bath for smelting lead and iron; dug iron-working hearths at two other points; dug five small cooking hearths into a heap of debris outside the building on the east which for some reason escaped the nineteenth-century clearance; and no doubt carried away many of the stones for building.
There are records from the early nineteenth century, tantalizingly incomplete, of other Romano-British finds in the district. In a field beside the road in Holcombe Bottom there was a large heap of stones, provincially called a stone barrow and thought to be a fortification to the villa; a man employed to remove it discovered an earthen vessel containing a large number of Roman coins, which he took to Exeter and sold. In 1817 a labourer digging a hole for a gate-post in a field near Gore Lane found a rod of pure flexible gold, about fourteen inches long, an eighth of an inch thick but swelling at the ends, weighing two ounces; he sold it to an Axminster watch-maker, who melted it down. It was perhaps a British ornament called a torque, worn about the neck or for supporting part of the dress; or it could have been a wand or verge used by Druid priests. In 1818 about half a mile east of Hunter's Lodge on the Bridport Road near Greenway Head, under a heap of stones, were found twenty-two silver coins of the first and second centuries A.D.