Caerhun Excavation.

	The Excavations On The Site of The Roman Fort Kanovium At Caerhun 1926-9 By P.K Baillie Reynolds OBE, CBE.



	Drawing by David Swarbrick

The early 1920’s had been a period of great resurgence for the science of British Archaeology, which had suffered greatly during the First World War of 1914-18, many projects having to be prematurely curtailed, and much of the personnel did not return to resume operations. It is not widely known that the author of the still quoted The Auxilia of The Roman Imperial Army Oxford 1914, G. Cheesman died in action in France. A new period of optimism, however, was apparent, and Caerhun, and the forts at Caernarvon and Capel Curig received attention as a result of this renewed impetus. The Roman fort at Caerhun had been known since Elizabethan times, visited by Camden, Elizabeth’s Chief Antiquary, even at this early date, recognizing it to be the Conovium of the Antonine Itinerary road book. But a crisis now threatened a major portion of the remains. St Mary’s Churchyard was reaching full capacity and it was proposed a graveyard extension should be sited in the area of the S-E quadrant (south pretentura). This can be seen bottom right of drawing above, remains destroyed would have been three barrack blocks, a store, or fabrica, or stable, a corner tower, a building backing onto the S-E rampart used for communal baking, and the east gate. The equivalent to these buildings in the N-E quadrant (north pretentura) had already been lost to Archaeology as it now housed the Parish church of Caerhun, top right on the drawing. The owner of the land, Mrs Captain Gough, while amiable to the extended graveyard, because of pressure from local antiquarians, also sought that the Roman fort be scientifically recorded, in such a manner demonstrated by R.E. Mortimer Wheeler at Caernarvon’s Segontium fort, as such as mentioned above a sign of the changing times for the Science. A lecturer in Ancient History from Aberystwyth University, Wales, P.K Baillie Reynolds was given the task of excavation, the S-E quadrant must have been under some threat as it was here he commenced digging in 1926. Paul Kenneth Baillie Reynolds was born in 1896. Educated at Hertford College Oxford during the years 1915-19, his studies had been interrupted by war service, by serving in the Royal Field Artillery. By 1921 he had become a Pelham student at the British School, at Rome. During his years at Rome he wrote books and articles he is well remembered for, `The Troops Quartered in the Castra Peregrina’ JRS 13 1923 pages 168-87. ‘The Vigiles of Imperial Rome’ Oxford 1926. The Vigiles had been a civilian militia, engaged in fire fighting, inaugurated under the Emperor Augustus. Baillie Reynolds spent much time and effort researching the extant remains of ancient Rome’s aqueduct systems, which had been utilized by the Vigiles in extinguishing conflagrations. Baillie Reynolds’ M.A degree kept him at Rome until 1923, during 1924 Baillie Reynolds became a lecturer on Ancient history at Aberystwyth University. His knowledge of Roman History ensured he was well qualified for the task of excavating Caerhun fort, a task he commenced during July 1926, digging annually for roughly six weeks each summer, from July-September, ending during 1929. His Excavation Report was publish annually in Archaeologica Cambrensis, and finally combined as an unpaginated book in 1938. During these years he published numerous books and reports on British archaeological sites, and beyond, for example P.K. Baillie Reynolds `The Shield Signal At The Battle of Marathon` Journal of Hellenistic Studies 49, 1929 page 100-105. During 1934 he became an Inspector of Ancient Monuments, England. By 1954 he became Chief Inspector, a post he held until 1961, spending much time writing guide books about the Ancient Monuments under the protection of the British Government. An example of what must have been one of his final publications as Chief Inspector was `Chysauster` Cornwall HMSO Guides, London 1960. The booklet about Chysauster was reprinted in 1965, again in 1978 this time published by the Department of the Environment, and again in 1983. Finally his employers received the task of repairing the Aqua Claudia aqueduct, Rome. This ran through the grounds of the British Embassy in Rome. It would seem Baillie Reynolds had been placed in charge of the repair and consolidation of the aqueduct, and as such returned him to affairs that had occupied him forty years earlier. The publishing of the work in the journal Archaeologia Volume 100, 1966, was his last publication, and he died in 1973. So it would seem Baillie Reynolds had a fairly distinguished career, this is reflected in his British Royal Honours, the OBE and CBE. The following is some indication of what he excavated at Caerhun, year by year, his discoveries, and finds made at specific locations, it is made more difficult by the unpaginated nature of the 1938 Excavation Report, so this aside it was as follows. Thanks to Colin Wallace, SAIR Editor, Society of Antiquaries of Scotland for information. Excavation Commences July 1926 It is not relevant to describe in too much detail the recorded dimensions of the excavated features as this has been covered in depth on two other pages in this website. The Fort Plan with the respective building numbers (I - XIX) should be consulted to understand the areas that had been excavated. Also for details on the timber phase fort and for the stone phase fort. During 1925 an Excavation Committee had been established, with such such notables as Lord Mostyn F.S.A as President, Lord Boston F.S.A as Chairman, Willoughby Gardner F.S.A as Hon. Treasurer, Neil Baines F.S.A as Hon. Secretary. Head of Welsh Archaeological studies, National Museum of Wales Cathays Park Cardiff was Dr R.E Mortimer Wheeler MC F.S.A who was appointed Hon. Director of Excavations, and Mr P.K Baillie Reynolds F.S.A as Resident Excavator. The cost amounted to a soon raised £1200, owner Mrs Gough stipulated all relics be deposited at a local museum, aided by his local foreman for three of the four summers, Mr William Roberts, Baillie Reynolds commenced work. Baillie Reynolds was aided in some minor capacity by such notables as R.G Collingwood, and Chris Hawkes, and Wheeler was known to have visited the excavation at least once. Work appears to have commenced in a rather frenetic manner, they only worked for about six weeks, that appeared maximum, and in 1926 we have nine sections of the rampart cut, all in the south east quadrant. The sections seemed cunningly sited A-B sectioned rampart, east gate south guard tower (timber phase only), intervallum road (internal rampart cheek communication road) and barrack or workshop building (eventually found to be burnt down). Closely adjacent trench C-D sectioned clay rampart and defensive ditch, moving closer to S-E corner we have E-F, sectioning rampart, intervallum, and workshop building, which is sadly still very visible today, and M-N which sectioned corner tower, rampart, berm and intervallum, locating also a W-E drain exiting at the rounded corner area. Moving around the S-E corner we have fairly massive section G-H, this was 30 metres or 100 feet in length, sectioning the south wall of the workshop, oven floors, the intervallum, rampart, ditch system, running sixty feet or 20 metres south beyond the lip of the ditch, through the annex area to just touch area of site F, see below. Closely adjacent to G-H was trench L-K which was similar to E-F. Possibly frenetic as all these trenches are still visible today and backfilling does not seem to have been as carefully executed as it could have been. Also sampled during 1926 was cookhouse III backed into the rampart, with oven floors, large workshop already mentioned, building VIII, fronting the Via Principalis road, in which was discovered a stone cistern at the north end of the building, and featured bad building practice, barrack ends VII, V, II, which vanish beneath the churchyard, and all the barracks due to be covered by graveyard extension VI, IV, I. It was crucial that traces of VII, V, and II could be located as these were the only traces of the buildings that had stood in the section of fort now covered by the Churchyard. Results here turned out to be slightly disappointing, though enough proof was obtained and this due to the stereotypical nature of Roman forts allowed buildings to be tentatively added to the plan, remains were rather sparse. A bullock was found to have been buried ‘quite recently’ above the remains of II, a 17th or 18th century deposit of stone had been cut into the remains of V, and VII was covered by a sheepfold, a concrete platform of a later building still remains there today. Also at this time the stone phases of the east and south gates are discovered, as are the dimensions of the east and south fort walls in both the timber and stone phases, and also the intervallum road in timber phase, and stone when one didn’t appear to exist in the S-E quadrant of the fort. Much of this work was included in the 1927 season. 1926 indicated the fort wall in its timber phase had a clay bank 24 feet wide at section G-H, composed of blue clay, resting on a kerb at the front of rough river boulders which were also bedded in clay, the eastern rampart back was composed of river gravel, and at section A-B it stood five feet six inches in height, which was possible the original height. The rampart contained river stones and some fragments of Roman pottery. The stone phase wall was found to be six feet thick at the base, cutting back the timber phase rampart two feet, still based on the original boulder kerb, though occasionally overlapping it by two feet at G-H but only one foot at A-B, and only stood on two feet of the boulders. The inner courses of the wall had been rough local stone, not in courses but as rubble, the outer face of the rampart was well cut millstone grit ashlars set in good hard white mortar. In all the sections dug at the rampart in 1926, no continuous wall courses of this millstone grit had been found in situ, all had been robbed away, as was common with angle tower and gate guard rooms, and barrack blocks. The fort ditch was separated from the clay rampart by a berm of five to six feet. At section G-H the ditch was twenty four feet wide, separated by a clay midrib of two feet. Ditch section west of the midrib was eighteen feet wide and eight to nine feet deep, while east was lesser due to the slope of the ground. Much encountered throughout was an earlier occupation layer, prior to the timber fort, in the form of camp fires, thought of as the camp fires of the legion based construction party building the fort, no artifacts were recovered from this layer, which is unusual. 1926 Excavation Finds The most significant find was a small piece of a large dedication slab from outside the east gate. Possibly this could have been the most important object ever discovered at Caerhun as it would have possibly indicated under which Emperor the fort had been rebuilt in stone, and whom was the auxiliary cohort in residence at that time. Timber fort dedication slabs were usually of wood, but not always, so date is a problem. The slab was of slate, one and a half inches thick, worked smooth on the surface which faced outwards, left as split on the inner, it was roughly six by six inches in size and contained two letters in good Roman lettering an A and possibly a O, separated by an oblique stroke. Baillie Reynolds considered the lettering to be of early second century style. More architectural salvage was present at the east gate in the form of a sandstone cornice fragment, of Cheshire stone it likely formed part of a decorative string course at the gate, but was not stratified. Found in the ashes covering the timber workshop south of the east gate was a broken amphora, reconstructed by V.E. Nash-Williams and now in Llandudno Museum, and a bronze horse bell. A flint flake found in the outer portion of the ditch at the east gate was believed to be Roman and was most likely for creating sparks to make fire. Stone phase barrack I also revealed a possible cavalry connection in the form of a copper alloy conical horse boss/strap union, along with the bell found here it is plausible that mounted troops may have been billeted here, but all Roman horses when used by auxilia, legions, or simply as utility stock would have been fitted with bells, strap unions and pendants. Barrack VI provided a very small bronze finger ring, and a much rusted iron finger ring. The cook house III provided a lead hook, with a turned back terminal, this was likely inserted in mortar and was likely for suspension of something, and also a blue paste melon bead from the stone phase layer. In barrack IV a stone spindle whorl was located. Coins had been found at most sections, pottery was common, mostly mortaria, amphora and Samian ware, glass vessel fragments were also encountered in the barrack blocks. 1927 Excavations Baillie Reynolds does not inform us of the date commenced or of the ending of the 1927 excavations, neither does he list his aims, as he does in 1928-9. But it seems obvious that while work planned involves the (interesting and high status) remains of the central sector buildings of the fort, the headquarters (principia) and tribune’s private house (praetorium) much still needed to be done to structures in the S-E quadrant, and much remains unresolved from 1926. The south gate, a complicated structure in the stone phase occupied much time. Baillie Reynolds here revealed the ditch was not bridged by a causeway, but by a timber bridge, boulder courses remaining south of the ditch lip looked likely to be remains of the abutment. The guard tower of the south gate exhibited squatter occupation, which as correctly stated was related to a 17-18 century cottage at the S-W corner, see below. Also barrack blocks, as is the building IX with stone water cistern are again investigated. The cistern which recorded ten by sixteen feet, and two feet deep, of slate slabs lined with blue clay, and mortar, was an unknown commodity, Baillie Reynolds considered it a water trough for animals, it could equally have been a pit containing ‘Fuller’s Earth’ for the washing of fabrics, a brewing vat, or possibly a water supply for blacksmithing work or even a supply for the nearby bath-house. The intervallum, cookhouse, and all barracks here are finally added to the ground plan, along with east and south gates, and work is moved to the highest status building in the fort, the Tribune and his families’ private dwelling, the praetorium, which would have resembled an Italianic town house. It would be cynical here to suggest work was rapidly carried to a high status building, but as the praetorium was sited in the S-E section of the fort it was under threat from the development work soon to commence. Badly preserved in the timber phase, the building XI was crossed obliquely by the church lane, though luckily the lane missed the central range of rooms and mostly crossed later courtyards. Several trees, which were not to be damaged, and the lane meant not as much could be done to the building as would have been needed, so several points such as why was the praetorium so big? and what are the additional four or five courtyards, bath-house, latrines, and an entrance vestibule all about? had to remain unanswered. The work on the building ran into 1928, when possibly some answers were obtained. The internal roads of the fort received some attention in 1927. The Via Principalis (principle street) was sectioned west of cistern building IX (it should be noted here that Baillie Reynolds places building IX at the north end and within building VIII) and in two sections between sleeper beam construction VIII and XI. The metaling has been dealt with elsewhere in this website, but gutters and drains were recorded, of slate (and sometimes tegula roof tiles used to deflect an eaves drips away from walls and down the street) the covered drain from XI ran south to discharge out of the south gate into the fort ditch. 1927 saw a commencement in digging the principia or principle building, next in the central range, it stood roughly in the centre of the fort and was not as such under threat, though the church lane did cover the majority of the front of the building, including L shaped rooms south and north of, and the twenty foot wide entrance door/arch, and a tiny portion of the external courtyard. The principia, which carried on into 1928-9 was complicated, interesting and large, and is dealt with elsewere. Possibly now that the excavation has moved from the development threatened area it can continue at a more orderly pace, and Baillie Reynolds has time to investigate the baths. He plans no excavation, just what was needed to finish a dubious plan already revealed by Samuel Lysons in July 1801, tentative dates for the various phases were included, see Kanovium Bath-House The 1927 Excavation Finds .’The timber principia XII only provided ‘one significant find’ this was a lead letter D, it was 3/8th of an inch thick, and roughly two inches square, found on a floor in room i, backhall, it was considered to be part of an inscription of stone with lead letters. The unpaginated manner of the Report makes it difficult sometimes to assign finds to a certain year so I apologize for any blunders here. Very little finds were present in the timber principia, a ‘Battle of Actium’ denarius of Mark Anthony, and a sestertius of Titus turned up in the gravel of the courtyard and room iv respectively, a surface find was interestingly a 4th century coin, probably a fallen horseman type cententionalis (Fel Temp Reparatio) of Constantius the Second. The well in the N-E angle of the courtyard, was excavated to a depth of eight feet, and completely in 1928, it was found to be nine feet six inches in diameter at the top, but progressively narrowed, lined in clay not masonry. At three feet four inches from the top it was found the well had been filled entirely with loose stones, contained within was a column base and two incomplete columns, sealed with burnt clay and ashes, pieces of the same pottery vessels were found in the clay and within the stones, so it seems this represents the sealing of the well. We must therefore conclude that the additional list of finds, see below, had been recovered in 1928-9. The south gate provided more architectural salvage, this time a plinth fragment, and in Baillie Reynolds’ opinion stone from Bodysgallen, Llandudno. 1928 Excavation Here Baillie Reynolds lists the dates July 9th-September 29th, see below for information in detail on western rampart trench O-P. The aims are as follows and I quote *‘Nearly all the remaining area within the rampart was explored, and the plan of the fort completed so far as possible. Operations included the excavation of* *(1) The defences on the west and north sides.* *(2) The gates on these sides.* *(3) The remainder of the main range of buildings.* *(4) The Retentura* *(5) A little work in the Praetentura.* Finally time is spent on the north rampart and gate, the N-E section of rampart was not touched in any shape of form, possibly in respect to the church, though berm, and ditch could have been sampled without disturbing the graves. The N-W section of rampart from the north gate right around to the N-W corner angle was stripped back entirely, but no cross sections had been dug, for whatever reason. Interestingly the N-W rampart was found to be founded on flat slate slabs which had been placed directly above the normal boulder base, the foundation was seen to be continuous, and most of the slate slabs remained in situ. The fort wall was found to be slightly irregular in the setting out phase, it curved southward 145 feet west of the north gate, a feature well apparent on the fort plan. At only three points was there any wall facing stones still in place, in the foundation packing was found a Hadrianic Samian form 37 bowl, which could be important in working out a date for the stone wall rebuild, at least at this stretch. The N-W corner tower was located, as yet only the second of the four to be located, see below, the N-E would never be located as it was situated were is now the graveyard. It was seen to be completely detached from the rampart, as was the S-E tower, but was massively ruined, Samuel Lysons’ diagram of the remains in 1801 had shown a tree plantation had stood here. It had been the place of local sport, a rabbit warren was seen to have existed here, and much damage by Terriers and Ferrets (small dogs and weasels) had occurred to the archaeology. Ferrets have a tendency to not want to emerge from burrows so the handler has to dig them out. The north gate, or what was left of it was examined. Excavated before by Lysons, his work was apparent, all Roman levels had been removed, as had been the spina, or the arch springing foundation. Lysons had heard a rumour a ‘flight of stairs had been removed recently’ this must have been the spina. The towers (two of) were seen to be set back from the fort wall by two feet 3 inches, the western tower being 10 feet by 8 feet six inches inside, and the eastern being 9 feet by 10 feet inside. The roadway north which would have been separated by the spina was found to be 23 feet six inches wide, but would have featured two carriageways. A few scraps of Roman pottery remained in the guardrooms, presumably missed by Lysons. Work is then continued in the main range, the praetorium is again sampled, it having not been backfilled since 1927 and a spoil heap had to be shifted before this could be undertaken. However more interesting was what was discovered between praetorium XI and barrack or stable XVI. It would seem that the praetorium grew beyond all control from its earliest form with the result that the west wall almost touched the east wall of XVI see reconstruction the narrow channel between the walls was used as a rubbish dump, a mass of pottery dated in Baillie Reynolds'’ opinion to have no pieces later than the late 2nd century A.D. The alley, which was from four to three feet wide was never used as a pathway. Next dealt with was the principia XII, finally excavated to the natural ground surface, again being left open as excavated in 1927. A cupboard was located in room v, constructed again from slate slabs, it was seen to be heated by a small hearth, formed of roof tiles, tegula and imbrex, like slate this seems to have been a ubiquitous building material at Caerhun fort. The Sacellum, or shrine was found to have been decoratively rendered, and painted in the timber phase and remains of wrought iron indicated decorative metal screens had fronted the shrine, erected above stone pilasters, one of which remained as it had fell complete with fixing points for the wrought iron. The well was finally cleared, at ten feet clean, but shallow water was located, aided by a pump loaned from Dolgarrog Aluminium Works, it was found to be fourteen feet deep, but may have been unfinished. The well contained little domestic refuge, therefore proving it was filled in before full use could be had of it, however it did contain a piece of stone pediment similar to those used on a gable roof, along with the two pieces of column and a pediment. The excavation of this building completed the work moved north to the final building of the central range, the granary or horrea, buildings XIII and XVI, complete with internal yard and typical supporting buttresses, a loading platform was located on the south side of XIII. Building XV was then dealt with, a stone structure, butting up to the N-W rampart close to the west guard room of the north gate. It was a stone building, which had stood at this spot since the timber days of the fort, ash rake outs and the solid build form suggests it was another cookhouse. There was nothing remarkable about the building, but a circular hearth at the west end contained a group of pottery vessels, a Samian form 27 cup, Llandudno Museum, a piece of form 31, part of a form 37, but most significant were three pieces of a white flanged mortaria. This vessel closely resembles a type recovered from Richborough fort on the South Coast, and these are considered 4th century, so this could be a rare example of stratified late Roman activity at Caerhun fort. The retentura was then examined, buildings XVI, XVIII. It has been postulated they had been stables, but a small porch on the east wall of XVIII seems to indicate it was a barrack block, which belonged to a centurion, the thicker ends of these blocks always being officers quarters. XVI also had evidence of late Roman occupation in the form of a white hammerhead mortaria, which sadly was only in the surface soil. Also a pot fragment bore an inscription, recording a Primitivs, surely a garrison member. The two long building in the N-W quadrant were excavated XIX and XVII, described as normal L shaped barrack blocks. Trees remained to make life difficult, but they were a fair match for XVI and XVIII, also the rabbiting damage that had inflicted the N-W corner tower had the same effect on the north ends of these buildings. Patchy areas of intervallum had also been located but not enough to add to the fort plan. Finally four days were spent back in the S-E area of the fort, seemingly the graveyard extension had not yet happened (it never actually did) and by diagonal trenches (a sign of the times) recovered the complete plan of buildings I and IV, but the natural ground surface was not reached, only the tops of the extant walls, as Mr. A. Appleton, the tenant (of Liverpool) wanted his cattle back in the field by nightfall, much coarse pottery and a copper alloy coin of Trajan, number 7 in coin report was found. 1928 Excavation Finds. Found at the west gate, a moulded sandstone plinth block, said to be from the Tattenhall area of Cheshire, and a piece of a slate possibly a border to the dedication slab, chamfered on both edges. A small stone head from the 2nd century gravel courtyard of the principia, of Cheshire stone, its origin is likely Celtic. A portion of a slate gaming board found in building XI. Many melon beads, counters, gaming pieces, spindle whorls found at most areas of the fort, but buildings XI and XII in particular. The forts equestrian connection is strengthened by more conical horse bosses/strap unions being located around buildings XVII (two of) an another close by in the road surface west of XIX. Other items of harness had turned up in the granary XIV, and XIX in the form of thick bronze rings, a common Roman military find. XII sacellum contain a small solid (almost pure) silver pointed bowled spoon, likely involved in the rituals of the shrine. XII gravel courtyard supplied a pennanular brooch, in room ii was found what Baillie Reynolds thought was a plate brooch (though he worried about the pin mounts) it was in fact a leather stud in the form of an enameled eye. West of X was found an iron spear butt ferrule, this could have originated from the javelin, the ‘pila’ or the auxiliary thrusting spear, the ‘hasta’. In the robbed out south wall of granary XIV was a bronze key, looped and with an inscribed St Andrew’s cross. A flint arrowhead was found in XVIII, considered Romano-British, a flint scraper of similar date was found nearby between XVI and XVIII. Many iron objects found throughout, eleven nails, knives, a pick axehead (dolabra), cauldron fragments at building X, trowel tang XVII. Much lead, a billet, weights, one weighing 10 unciae, counters, rivet, and a plummet. Baillie Reynolds had the billet, or lead ingot examined by Mr. G. C. Whittick, Lecturer in Classics and Ancient History at Armstrong College, Newcastle-upon-Tyne, and Dr. J. A. Smythe, Lecturer in Metallurgy, also from Armstrong College. Whittick reported the lead piece was indeed the upper portion of a wedge shaped ingot, cast in a mould. This shape of mould was unknown at the time and Whittick claimed it was basically a new method of casting, but this ingot had failed in that it had been removed from the mould before suitably cooled. Dr Smythe comments ‘Assay of the lead showed a silver content of 1 oz. 9.1 dwt to the ton of lead. The average silver-content of the extant pigs of Roman lead derived from Flintshire, all of which have been de-silverised, is only 14 dwt’. The Romans had been very adept at removing silver from lead, and this ingot therefore has had more removed than ingots found near Deeside now at Grosvenor Museum Chester, it is interesting conjecture that this ingot was actually processed at Caerhun. From where would the lead have been obtained? more analysis may have proved this, but supplies would have been common, large deposits existed near Llanrwst, north of Capel Curig fort, though if exploited at this time remains unknown, and also on the north coast at Llanddulas, and doubtless on the Great Orme’s Head. The 1929 Excavations. ‘By the continued kindness of Mrs Gough of Caerhun and of Mr. A. Appleton, the tenant, a fourth season of work on the Roman Fort of Kanovium was undertaken in 1929. Work began on July 22nd, and the filling in was completed on September 24’. Summary :- *(1) A little work was done to the fort itself to complete the plan.* *(2) The main effort was concentrated on the area outside the fort to the south, where it was hoped that a civil settlement might be found.* *(3) A little exploration was done with a view to ascertaining, if possible the course of the roads in the vicinity of the fort.* So it seemed the hard work had been done, the plan was almost complete, time could now be spent beyond the fort earthworks, though judging by the amount of time he had left, and the amount of sections still being dug it wasn’t a time of leisure. Once more he digs up the tarmac of the lane (having done this previously in excavating the north gate) ‘thanks again to Conway Rural District Council’ this time he seeks a building to correspond to XVII, he digs up the modern car parking area locates the well preserved long building, calls it X and moves on. He then moves off right to the opposite side of the fort to attempt to locate the S-W corner tower. He reckons he did, but admits it was dubious as the road was above it, and also a tree, enough was found (small piece of walling) to add it to the plan however. He now moves to the S-W rampart, which is west of south gate, this area being the focus of 1929, and adds a very late section Q-R, this sampled the rampart, the clay backing bank, and the intervallum. The rampart etc here was similar to other sections dug previously, though a feature of the boulder foundation was that it contained brushwood, which was usually used for stabilizing foundations in wet conditions. A well was located, but it had been overlapped by the road metaling of the 2nd century intervallum, giving some credibility to it being part of the timber phase fort. The well was a rough circle, cut into the clay natural, it featured no stonework, but showed evidence of a patchy clay lining, it was six feet six inches wide, and had been filled to the top and sealed with rubble. Filled with clay that contained stones, much Roman pottery of the rough coarse types, an olla in good quality ‘rusticated’ grey ware, and the handle and neck of an amphora that usually contained wine or olive oil. The well was excavated, at around sixteen feet boulders had been flung in, these removed, water was seen at nineteen feet four inches. More boulders were encountered, a stake of wood protruded, being one foot nine inches above the water line. It turned out to be a camp palisade stake, once removed with the aid of a ‘crow bar’ it was found to be six and a half feet long, judging the well to be at least twenty five feet deep, Baillie Reynolds gave up, calling in local quarry workers had been his only option, which was not warranted as he believed he would only find more rubble. A second trench E-T forty five feet west, sampled the intervallum, it uncovered another timber fort feature, a circular hearth, post holes showing it had once been roofed, the hearth contained a bed of clay ten inches thick, which contained a mostly complete amphora in pieces, the hearth had then been formed of broken tegula (large, flat, and with useful flanges) roof tiles, in the ash above this was found 30-40 sling shots (ancient missiles) of baked clay, Llandudno Museum, clearly the timber fort had been evacuated as the hearth was found to all be overlaid by the intervallum road of the stone phase fort. Another section of ditch was flung in at S-T, it revealed that as elsewhere the stone wall had started to slip into the ditch due to an insufficient sized berm, seen first in 1926, and the section of ditch in bound of the midrib had been capped with a clay packing, thereby reducing the ditch size, silt revealed the ditch had been open for a while prior to the soldiers realizing the fort wall was due to collapse. Next the excavation team return to the east gate in search of a road down to the baths, (and also the slate dedication slab recovered partially in 1926 was also sought) a road was seen for 22 yards, but then faded out, presumably curving north or following the fort ditch, numerous trial trenches located nothing, and it was concluded a simple path, and not a road, headed east down the short slope to the bath-house. The dedication slab sadly remained illusive and no more trace of it could be found. The S-W corner had long since interested Baillie Reynolds as it showed parchmarks of a long building, thinking it possibly Roman, a days work revealed it was a Welsh Longhouse of 17th-18th century date, cut into the S-W fort wall, damaging it badly, the entire area from the S-W corner east to the south gate contained its yard, outbuildings and garden complete with a surviving pear tree (now gone). Two roomed with one plastered room with a hearth. Much pottery of that date, Buckley Ware, black and brown glazed slipwares and coins of George 11 and George 111. It looked likely to be one of three dwellings reported to be standing by St Mary’s Church by Edward Lhwyd in 1699. Slightly disappointed by this rather a waste of time, Baillie Reynolds was cheered to find Roman occupation debry beneath the site of the longhouse, including a complete Samian form 37 bowl, Llandudno Museum, which appeared to be deposited in a pit by the 17th century folk! Now work is moved to the area outside and east of the south gate. An area (obviously vicus) could be (and can be) seen outside the gate, crossed by a very clear Roman causeway (road) heading S-W towards Dolgarrog, I get the impression Baillie Reynolds could have done without the complexity of the remains located, though little was evident in the Geophysics Survey by Gwynedd Archaeological Trust of 2002, there seemed as many as six layers of Roman activity discovered, a separate annex containing six layers inside a round house, a branch road into it, which is shown on the featured reconstruction, and two separate period roads out of the fort partly overlaying each other. Any section dug revealed intensive Roman occupation, and he listed these sub sites as sites A, B, C, D, E, F, H, K, site F being the annex. It would be pointless here conjecturing the use of these sub sites, they simply represent the south vicus, and are represented by copious pottery deposits, boulders, hearths, and gravel occupation layers. The annex however was of more interest, a ditch was located 30 feet south of the south gate (east of) commencing south of what appeared to be a branch road into it. Twelve sections proved it ran south for 110 feet, before curving east, before turning north to join with the fort wall at the N-E corner. The course of the ditch was located by the twelve sections, and considered to be unfinished, section 1 - 6 showed it to be three feet six inches deep, while at sections 10-12 it became a shallow gutter. No rampart was located, though a stone foundation close to the branch road may have been a preparation for a rampart bonding it to the S-E fort wall, it never happened. Inside the annex was site F, an area of artificial hollow ground, it contained 16 hearths at one level alone, and many others close by this was reported as showing intense occupation, it was also considered a ‘Pit Dwelling’ which today would be taken as meaning Iron Age roundhouse. Roundhouses inside Roman fort areas are rare but not unknown, the fort of Vindolanda near Hadrian’s Wall had a number inside an annex rampart in a early 3rd century Severan context : all the hearths showed evidence of intense metalworking. The west road is then located, the 1st and 2nd century roads seemed on differing alignments, the metaling differs in width, and gravel type, it seems the road(s) have long since been removed, but headed in a westerly direction once crossing the vicus. The west road is sought, leading out toward what is now Caerhun Hall gardens, field number 502b had had a trench cut across the area of the west gate, nothing was located other than Roman period occupation material. Finally a massive trench is cut right across the field but no road is located, air shots have now clearly revealed the road. Finally the Caerhun Fort Excavation is concluded by three more trenches in the area north of the fort, which today would be known as the northern vicus. At this late point Baillie Reynolds ceased to label his sections, in search of the north road (pointed at by burials close to Caerhun Cottage, holiday cottage) he digs a section in field 1069, 250 yards north of the forts north gate and locates a gravel, clay, and cobbled road two feet below the modern surface, he infers the Roman date and moves on to the enigmatic dock east of the north road next to the river. The dock which is today regarded as an unknown commodity had a section cut which is still visible, containing six layers, it was believed the bottom four were Roman work, the top Roman layer contained gravel similar to that used in the north road, and brushwood similar to that seen in rampart cutting E-T, south west corner. The dock and associated features are dealt with here at Caerhun History. The final trench was done in a deep cutting just south west of the dock, in field 1069, behind a ruined boathouse. It was believed that it was the Roman dock road, but had been confused by being used for a trolly bed to transport stone when Caerhun Hall was being built. A trench indicated it had received some type of gravel metaling, with stones protruding out, but this was likely to be the trackbed, or it was believed it could have been used to transport bricks from the now defunct brickworks across the river. Air shots (as do the 1880 Ordnance Survey map) show there to indeed be some track from this cutting up to the Roman fort north gate, though it could simply be an old footpath or possibly it was a modern footpath on the old Roman line. 1929 Excavation Finds Rarest find is possibly the oak palisade stake recovered from south intervallum well, preserved in the waterlogged conditions of the well. Baillie Reynolds calls it a Vallus. The collection of clay sling shots from the hearth nearby. A collection of many iron objects from the site F in annex, possibly indicating a manufacturing area. Blue paste melon bead south of fort, while attempting to locate south road. A tile fragment, Llandudno Museum, bearing the imprint of a child’s footprint, possibly made at Holt on Dee it is of the type used in bath-houses, and possibly represents the dismantling of the tribune’s residence. It was found ‘deep’ in the filling of the intervallum well, and it was thought it represented a pulling down of the timber fort prior to it being rebuilt in stone. From level 2 of site F Iron Age hut came the majority of a form 37 Samian bowl which was recovered with Roman window glass. This complex topic can be explained in more depth by a detailed look at western rampart trench O-P below.

The Ploughed out Western Rampart, P.K Baillie Reynolds’ Trench O-P 1928 in Detail : The West Gate & Rampart



© Kanovium Project

This is the western fort rampart, this image is taken looking north and I am standing in the very silted up defensive ditch. I do not have the space to feature all the sections of the rampart, but I thought it would be interesting and necessary to reveal what Baillie Reynolds actually discovered when he tested sections of the earthwork at this area. This area became available for excavation during 1928 and is featured in Baillie Reynold’s third report concerning the site, he thanks the landowner Mrs Gough, and her tenant Mr Appleton of Liverpool for continued support in the venture. He only dug one section at the western fort wall and ditch, and this was immediately in front of the camera roughly in the direction of the tree, on his plans he referred to it as cutting O-P. This trench sectioned the entire rampart, including the ditch, the berm, the stone wall, and the clay packing bank to the rear of the rampart. O-P revealed that the make up of the fort wall was broadly similar to sections dug in the south fortification, extensively diminished first by natural decay ((initially most of the high quality dressed sandstone wall facings had simply fell from the fort wall into the ditch and lay above a layer of natural silt) followed by widespread robbing of the rampart and gate, followed by ploughing, presumably from the medieval period through to the time between World Wars one and two. Decay, robbing and farming has reduced the 30 feet or so of berm and rampart, and an even wider area for the ditch to the blurred earthwork visible today, which is to the left of the church lane as you enter the fort earthwork. Below reproduced is a description of what Baillie Reynolds discovered remaining in 1928 (it may not remain today) of the western defences. Section O-P The Earth and Clay Rampart The original form of the clay block bank was very much reduced from how it would have stood originally, this was however, not due to post Roman damage, but due to the damage done when the Romans rebuild the rampart with the stone revetment in the early second third of the 2nd century. This reconstruction of the rampart had damaged the original boulder core, in that at the south rampart this core projected beyond the rampart face (unevenly) but here it appeared to have been removed or never projected at all, but enough remained to show that the clay wall had stood on a footing of rough river boulders. No artifacts to date the original fort wall had been found in section O-P The Stone Wall This was 8 feet thick at the base, a full 2 feet wider than the south and east ramparts. No stone facings of the fort wall remained insitu at section O-P, the wall had been largely reduced to rubble, which still however, stood 2 feet above the footings, and was very close to the surface of the meadow. A portion of a good quality millstone was found amongst this rubble of the wall. The fort wall was built on a foundation of a single course of boulders, this in places had been dug through the previous earth and timber phase and was bedded directly upon the natural sintir of clay. The only object found was a single small piece of Samian pottery which proved undateable. Beyond the remains of the fort wall was a spread of clean blue clay, this represent the denuded remains of the packing bank which was always present at the rear of a Roman fort wall. The Fort Defensive Ditch This proved to be of the same design of ditch located in sections at the south and east areas, being as usual a single type with central midrib. As at these other portions of ditch the inner section had been deliberately filled with a hard packing of clean green clay (packed as hard as the natural clay ground level). Beneath this clay packing was a layer of natural silt, this indicates that the inner ditch had remained open for some time, before it was found necessary to reduce the depth of the ditch, which can only indicate the berm was not sufficient to support the weight of the new stone fort wall which must have been suffering from subsidence. The inner face of the inner ditch also had a curious step dug into it, this was also noticed in a similar position at the south ditch, this surely cannot have helped in a situation of the fort wall subsiding into the ditch? The base of the inner ditch was found to be 7 feet 6 inches below the base of the fort wall, while the ditch base still contained water ( a culvert located near to the west gate guard room apparently still being capable of drainage) above this water was a layer of natural silt and this was found to contain burnt wood and bones and a single dateable fragment of a Samian DR 37 bowl which was dateable to the early 2nd century. The outer ditch, as with other sections at varying areas of the rampart, was found to not be as deep as the inner, and was only 6 feet deep, it also lacked the distinctive ‘cleaning slot’ which was located elsewere. The slot has been a topic of argument for many years, first postulated by Richmond and Webster to be a trap to break an intruders ankle, it could also clearly simply be evidence of cleaning with a spade, which would be the only possible angle to remove silt in the cramped location at the bottom of the ditch. The bottom contained the usual spread of silt, above this, numerous well dressed sandstone facings lay where they had simply fell from the fort wall, above this was a small amount of modern plough soil. This concludes the evidence found at section O-P, Baillie Reynolds then turned his attention to the west fort wall in general, he found that the wall foundation could be traced unbroken from the north-west to south-west fort wall corners, he excavated three trial pits at varying sections of the west rampart and found that the facing course of the lowest levels were still insitu in each pit. Next, the west gate was examined. The West Gate Earth and Timber Period Of the original Flavian fort gate nothing remained but the river gravel roadway. As the causeway was of twin roadways, 8 inches apart this implies the original gateway had been of twin toweres/roadway form. The northern roadway was 6 feet 6 inches wide, while the southerly road was 7 feet 6 inches wide, unsure as to whether there had been two entrances into the fort here, Baillie Reynolds adds that the northern road may indeed have not been such, but the remaining gravel may represent the floor of an earlier gate tower. The Stone Phase Gate This gate was very badly stone robbed, most of the superstructure had been removed, leaving only the boulder footings, and even this had vanished in places too, the plan only being obtained by tracing the depressions left in the natural clay. The pauce remains of this gate had been found to be mostly less than a foot below the meadow. One sandstone block was found with an edge moulding, and a fragment of moulded slate slab, a dedication plate, suggests that the gate received the same ornate decorative stone work seen previously in the east and south gates. Only one sandstone wall facing block remained in place, this was in the east wall of the guard room. This tower was to the north of the roadway. This stone gate which was 8 x 10 feet internally was larger than its timber predecessor, and as at the east gate the rampart had to be dug back, therefore the north wall of the guard tower used the rampart boulder footings as its own foundation level. Roman sandstone blocks had been found in both the floor make up of the tower and in the roadway. A curious feature was located on the north edge of the roadway, for 7 feet 3 inches the road is kerbed by large irregular boulders, which then suddenly end, no similar feature was found to the south edge of the road, but such curious lines of boulders have been noted elsewere in the fort., near to barracks at the via Praetoria. The south wall of the tower cut right through the previous timber occupation layers and rested directly on the natural clay. The tower was found to be of misaligned construction, a common feature at Kanovium, particulaly at its north wall. The tower floor was found to not be original, though of rammed clay, Baillie Reynolds considered it incomplete, and found no occupational evidence. Though in undiagnostic positions three pieces of course ware and a single sherd of Samian DR 18 plate with the stamp SACIRIOF had been found. The front wall of the guard tower was found to be eight feet back from the fort wall and the gap had been filled, at least as high as the roadway with rough boulders. R.G Collingwood considered these gates to be of similar form to Hardknott at Eskdale, he considers both forts to be Trajan-Hadrian. The west gate, incidentally seemed to be the latest built, it exhibited ‘no sign whatsoever’ of later blocking walls, as had been seen at the south gate, or any alterations to suggest reductions in later Roman periods. The roadway was found to be very well preserved, still exhibiting its camber, it was found to be constructed in quarried gravel, in difference to the timber phase, which used river gravel. This layer of gravel also contained large stones, Baillie Reynolds considered the gravel to be of a similar type to ‘gravel being still quarried today a mile away at Tal y Cafn’ this stone and gravel layer was found to be a foot thick, it was separated from the earlier road by eight inches of hard clay. Not mentioned in the 1928 report was the apparent lack of causeway continuing for any distance across the area west of the Roman fort, this area was then not available for excavation. The following year however the field did become available, and Baillie Reynolds declared that just as there was no east road, there was also no west road either. This is most unusual, while terrain such as encountered at the east gate may be explained, this is not so here, and the absence of a causeway is hard to explain. The lack of evidence for a road was illustrated by a very lengthy trench across the whole field which must now be under the small golf course attached to Caerhun Hall, no road was present. Two trial trenches between this trench and the west gate also failed to locate the road, Roman occupation was however present, several pot sherds and debry being found in the two small trenches. So while Baillie Reynolds concluded on no road, it has possibly been ploughed away, no trace remaining. This flat area of the fort, unless the site of the parade ground (a level area for the garrison to drill and undertake religious ceremonies) must have had some vicus occupation, and a road for access to the fort would have been essential. In agreement for this I must point out that at the west, Decumana gate, a roadway is unlikely as it is barred by the back wall of the principia, and the area between the four long buildings here is not large enough to support roads for any other use than at best double files of soldiers. We can conclude here with the fact farming has seriously reduced this western portion of Kanovium, and much must either have been lost, or is still present beneath the gardens of Caerhun Hall, gardening activity can even render archaeology difficult to locate by Geophysical survey.

The source for this page was Excavations on The Site of the Roman Fort Kanovium, At, Caerhun, Caernarvonshire, Kanovium Excavation Committe and P.K. Baillie Reynolds FSA CBE OBE, Cardiff 1938. After Caerhun excavation Baillie Reynolds moved on to greater things, as did R.E Mortimer Wheeler after his Caernarvon and Brecon Gaer excavations. The excavation finds were as Mrs Gough wished deposited at the then Rapallo House Museum at Llandudno, now the Llandudno Museum, by her late husband Captain Gough, here they remain. The 1926-9 work is mostly still all we have of reference, today we can still view many of the rampart sections, which I suppose has now become history in themselves. I sincerely hope one day more excavation can happen at Kanovium, the founding date, and information on the strange west and south gates would be very welcome.