The 2nd Century Stone Fort of Kanovium

The plan of the stone fort has been drawn by Kanovium Project and is after Llandudno and District Field Club and P.K. Baillie Reynolds et al

This plan was once believed to be the second best auxiliary fort plan to have been recovered from Wales, Gelligaer being the most complete, this plan would have been the premier example if it wasn’t for the fact that the church occupied nearly a quarter and was obviously not available for excavation.  Indeed it was a proposed extension to the graveyard which prompted the excavations to take place.  The early part of the 20th century was a time of great pioneering archaeology, and many local sites received some attention, local celebrity Willoughby Gardner excavated at nearby Pen y Gaer and Pen y Corddyn Mawr, J.P. Hall excavated at the Bryn y Gefeiliau site at nearby Capel Curig.  More in the limelight was R.E.M Wheeler’s excavation at Caernarvon/Segontium, so in this period of great archaeological enlightenment, with the threat of imminent destruction for another quarter of Caerhun, P.K. Baillie Reynolds F.S.A, University of Aberystwyth, was asked to explore the prominent and well preserved remains of the site, becoming ‘Resident Excavator’.  Interestingly Mortimer Wheeler was Hon. Director of Excavations, visiting at least once.

P.K. Baillie Reynolds 1896-1973 had a distinguished career, becoming Chief Inspector of Ancient Monuments England in 1954, but Caerhun fort was earlier in his career, following studies at the Pelham School at Rome. 1923, he became a lecturer in Ancient History at Aberystwyth.  With full support from the landowner,  Mrs Gough of Caerhun Hall, the excavations were put in the hands of ‘The Kanovium Excavation Committee’.  This Committee consisted of Lord Mostyn F.S.A as President, Lord Boston F.S.A as Chairman, Mr Neil Baines F.S.A as Hon. Secretary, and Gardner as Hon. Treasurer. The excavations commenced in July 1926, Baillie Reynolds authorized a foreman from nearby Tal y Cafn, Mr William Roberts, who acted as his deputy for three of the four years of digging, aided by several local labourers to do the spade work. The cost of the excavations was judged to be £1200, which the Committee soon raised.  Interestingly R.G. Collingwood helped out for two days while Baillie Reynolds was called away to ‘some crisis’.  Another additional archaeologist who occasionally helped out during the excavation was Christopher Hawkes, Hawkes became a very eminent archaeologist, working at some premier sites in the near east, it is interesting to think of him digging at Caerhun.  Baillie Reynolds lodged at Llandudno and called in every few days to inspect the remains unearthed.  This may sound a bit different from todays scientific approach, but was  normal for the the time, indeed Wheeler’s following excavations from Segontium at Brecon Gaer were fueled as much by the local fishing as the well preserved remains of the cavalry fort, and it was common to instal a foreman or ‘clerk of works’ to oversee digging work, which was often done more with a pick than a shovel.  When the Roman fort at Capel Curig was excavated during the early 20’s, J.P Hall actually lodged at the house of Dolgam during the excavations, just across the river from the fort of Bryn y Gefeiliau, though a small swing bridge had to be erected for his quick access across the narrow Llugwy river, otherwise he would have had to have gone several miles around to either Pont Cyfyng or Pont Ty Hyll.

Baillie Reynolds excavated for four seasons at Caerhun, in the years 1926-9 usual in mid July-August, for about six weeks, and recovered two Roman forts on the site.  Sadly, crucial (possibly because he was not present to inspect work progress, resulting in very little plans of timber partitions being recovered, particularly from the barrack blocks) he failed to actually discover the foundation date of the first fort.  It is believed that the first fort was constructed in the last main Roman push into North Wales, and was constructed during the years 76-80 AD.  Some of the Welsh forts had been built by Agricola, for one, Segontium, it seems possible that Kanovium was established by Agricola’s predecessor Julius Frontinus a man famed for his literary works on the sewerage system of Rome.

The Roman army could have arrived at the Conwy valley several ways, old thought used to be that they arrived along the North Wales coast from Deva possibly because of Suetonius Paulinus’ A.D. 60 conquest of Anglesey, but now more likely is they reached Conwy by traveling up the Dee valley from Llangollen, via the early fort at Wroxeter, and equally from the area of Llanfor, Bala, during the 1980’s a large supply base being located with similarities (octagonal enclosure wall) to the supply depot for the eastern sector of Hadrian’s Wall, at Corbridge.

So during the years A.D 76 to 80, Roman surveyors will have recognized the site was a fairly good one, these men would had been known as ‘agrimensores’ also Roman soldiers they would have arrived at the site in advance of the main army unit, which at this point was the second legion Adiutrix, a relatively new legion formed in A.D 69 for honours won while as auxiliaries in the Ravenna Fleet by the emperor Titus Flavius Vespasianus, which is proved by 14 tomb stones found at Chester, many books tell you the 20th legion established Deva, this is wrong.  The possibility that Suetonius Paulinus recognized the  value of the siting of Kanovium and is responsible for the first Roman military instillation will never be proven.  Baillie Reynolds hints a Celtic fort or similar occupied the site, which would have been cleared of anything, trees, bogs or Celtic forts, marked out with flags, rampart, principal buildings and barracks marked out and the legion would have been set to work constructing the new fort, one of many around North Wales at this time.  We are unsure even if the cohort of auxiliaries that would garrison the fort were nearby, or if they helped in the work.  The fort was sited to block E-W, and N-S Celtic trade routes, which crossed the Conwy somewere close to the Roman fort. It would have taken maybe 50-100 legionaries about six months to build the earth and timber fort and it was something like this- 

As the stone fort was found to probably closely follow the lines of the  previous earth and timber construction (though the stone rampart projected 13 feet further east than the clay predecessor) it was found that the stone fort was an enclosure 140 x 140 metres or 460 x 460 feet, it covered an area of 1.97 hectares or 4.86 acres inside the rampart.  A small annex which was attached to the south rampart was found to be 0.57 acres, seemingly in connection with the stone fort as the 2nd century south road is clearly seen to avoid the annex ditch, though equally present in the early phases too.  The fort buildings are conventional with the normal auxiliary fort plan, the principia being central, faced east and flanked by a double horrea, north, and praetorium, south.  Structures occupying the front and rear of the fort were found to be of the transverse disposition, known as ‘scamna’.  The interior of the stone fort contained 17 buildings, this includes two buildings inserted into the rampart back, III and XV.  Two long buildings fronting onto the via principalis road are possibly timber and not of the 2nd century stone fort, X, and VIII, but Baillie Reynolds was unsure of this and added them to the plan with reservations.  Building VIII, probably a workshop or fabrica building has a very curious misaligned front wall, Baillie Reynolds commented that it was not as if the fault could easily have been missed (these walls are not substantial, being only a rough boulder foundation which could have only supported walls of wood) and rectified, however the construction party ignored the fault and butted the two walls together.  This fault was caused by work being undertaken by differing construction parties, one not feeling it relevant to correct the former’s mistakes.

The gates are all rather peculiar too, as are the corner towers, the detached corner towers suggest that they preceded the rampart, while the guard chamber of the of the porta decumana gate (west) was 0.7m behind the line of the fort wall, this seemingly built on the post holes of the timber period gate and the stone rampart extending out beyond. As mentioned above the different period ramparts are not on exactly the same alignment, for instance the twin guard towers of the east gate stand on top of the earlier clay rampart but are 13 feet further east.  The detached nature of the corner towers is now easily understood, the timber fort must have been smaller than the final stone one, we know this at least at the east rampart because the stone fort wall was 13 feet further east, and overlay the earlier wall and post holes, in fact the two stone guard towers now stood directly at the centre of this earlier rampart.  We know that probably late in the time of Trajan or early in Hadrian that work commenced on the stone re-build, while the stone rampart was only constructed about A.D 140-50 during Antoninus Pius, it seems likely that the corner towers had been built over the top of the earlier timber towers, a post hole at the south east angle was probably remains of one of these towers, and are therefore on the line of the smaller timber fort.  So these towers had been constructed before the change of plan to increase the diameter of the rampart, so rather than demolish and rebuild 10 feet further out, the builders simply decided to encorpoate the tower superstructure on the clay of the new stone wall which was increased by several feet everywere to compensate for the weight of the new stone wall.  The wall stood on a rough boulder foundation, at the south and west rampart, slate slabs for the north, but the east had no boulder foundation and this was undoubtably to compensate for the slope here which is fairly severe and would make the fort wall and berm slip into the ditch, this was roughly 24 feet wide  including the large packing bank of clay at it’s rearside.

The rampart was surrounded by a defencive ditch, which was 24 feet wide, and 8-9 feet deep deep with a central ‘midrib’ of 2 feet which meant it rather looked like twin ditches, the outer portion had the drainage slot, so typical of Roman forts, while the inner was of round section, which may represent recutting, the inner section of the ditch also had a curious step, next to the berm, attested in most of the trial cuts through the ditch, on three sides of the fort record this.  The outer ditch was not as deep as the inner ditch.  At some  additional time, as an amount of silt had been deposited, but not very long after the stone wall was added this inner ditch had been uniformally packed with a spread of clay.  This clay was found to have been packed as hard as the natural subsoil and must have been put there to cure subsidence.  The berm was obviously not wide enough to support the weight of the new stone wall and was slipping into the ditch.  

Between the ditch and the rampart was a berm 5- 6 feet  wide, the berm is interesting, it has been called the ‘killing ground’.  Roman ditches were of a cunning form they allowed ease of access, but once in, an enemy found the outer face too steep to climb back out so he was forced to scramble up onto the berm, this placed the enemy within easy missile range of the fort garrison.  Excavation of auxiliary forts (South Shields, Arbeia) has shown that stones were the favoured missile, and of a round flat size, similar to the type used by children to bounce over water, the rampart would have had piles of these stones.   You could shoot arrows or throw javelins, but the stone is the oldest and cheapest weapon available.  The enemy, if getting as far as the berm, was a dead man, this on it’s own would not have been easy, as the ditch as well as probably being full of thorn bushes, or even wooden stakes (lilia) also had a channel at the bottom at its outer section, while possibly for drainage, it represents a spade scrape, it could also have been intended to break an enemies ankle.  There is evidence that this fort ditch had the base slot. The ditch was continuous at the south gate that means it had no causeways emerging from the fort gates, but that access was provided by a wooden bridge or drawbridge, and Baillie-Reynolds found evidence of this at the south, or dextra gate, while the east gate (which had no causeway) was not bridged and the ditch was interupted, another strange facet of Kanovium.

The original rampart was built of clay and rubble, which was laid on a layer of brushwood at the S-W rampart, this suggests the area was waterlogged, we have three water sources in the mid to south area of the fort, a  water cistern in building VIII, and wells inside the principia and outside the praetorium, Baillie Reynolds claimed that as well as the entire spur being levelled that it was entirely capped in clay, a fairly intensive task, not peculiar to Kanovium, it suggests a fairly waterlogged environment, and my thoughts are  this is possibly the site of a Celtic settlement, Baillie Reynolds hints at this, and a La Tene fibula dating to 100 years earlier than the fort has been found down at the river edge.  The clay rampart was 7.0m wide, allowing a walkway height of about 9-13 feet tall (only 6 feet at the south east angle) not counting a palisade probably of timber.  This mid A.D 70’s fort seemed to have needed major re-building work (oak has a life span of 20 years when inserted directly into the ground) around AD 100, some of the internal buildings now being recognizable as the ones on the 2nd century stone phase, though the buildings of the Flavian fort, the first period had looked different (beneath this first fort Baillie Reynolds actually found the fires of the builders of the fort).  The timber phases turned out to not be well defined, much damage being done during the construction in stone, often footings being dug right through the timber phase down to natural sintir.  Finally around A.D120-50 the clay rampart was cut back 6 to 8 feet and a stone wall was laid back into the provided area, which was 6 feet thick on three sides, but 8 feet thick at the west. This wall of Cheshire gritstone featured a decorative cornice, while oddly a shale block has been found near the north rampart and this has a decoration known as a ‘dentil’ which is a projecting square beveled carving on the face of each stone.  Dentils are a common Roman decorative feature and can be seen today in the city walls of Jerusalem and Tarragona (Tarraconensis in Spain) giving a fine architectural decoration to the wall.  This implies that Kanovium’s wall was not all built at the same time, or may represent later patching, the use of local rock is also significant, up to circa 210 AD on the basis of the milestones found on the Carneddau range, had all used the gritstone, but later 3rd century examples of Postumus, Victorinus and Trajan Decius show the use of the gritstone had declined and that local rock was now used, this is odd as the sandstone quarries of Helsby and Frodsham (an area as well as Handbridge which is thought to have been the location of the source of the stone) are still in production, so possibly the psyche behind shipping in stone had now ceased to have been of any significance.  Dr Grace Simpson in her reappraisal of pottery evidence in ‘Britons and The Roman Army’ London 1964, claimed the cookhouse in the south rampart had been constructed about 250 AD so possibly the fort wall received some patching, possibly after damage done by a withdrawing garrison at the end of the 2nd century.

Here basically knowledge of phases ends, finds of third and fourth century pottery and coins suggest a continued sporadic garrison into the fourth century, but the essential remains are still of a 2nd century nature, while at Segontium ‘chalet’ style barracks are to be seen, similar to this period at Housesteads, they overlay the longer earlier type, and are smaller, showing a decline in unit strengths.  Caerhun shows none of this, while the south and west gates also resemble narrow 3rd century types, single tower etc but while Baillie Reynolds did find a little 4th century pottery he found no such structures to suggest the fort was garrisoned at this time.  Possibly attributed to the fact the top layers of the fort had been completely obliterated by ploughing, but equally by continued civilian use of the vicus and bath-house, only further excavation could resolve this difficult issue.

The shape of the fort rampart is also interesting, it can be seen to be square, this square shape is rather old fashioned for the late 1st century, indeed it is more Republican in style, during the late Republic the Roman fort had been square, but by this time the shape had evolved into the more familiar ‘playing card’ shape, which was rectangular with rounded corners.   Caer Gai, Bala shares the square shape, but the legionary fort at Deva,  and other auxiliary forts nearby at Tomen y Mur, Bryn y Gefeiliau (G.A.T Environs Survey) and Segontium, are the playing card variety, surely this can only be caused by personal preference in the design of the surveyors.  The famous forts in Cumbria and Northumberland are all generally the new shape, Hardknott, Birdoswald, Housesteads, Vindolanda, South Shields.  And also some design deficiencies can be viewed at Kanovium, mostly down to faults in the groma, for one, the rampart curves inward from north to south at the west rampart there seems no immediate reason for this, I have been advised that this would normally be down to very large obstructions such as rocks, which doesn’t seem apparent.  Apart from the already mentioned workshop wall, we can see all the walls in the gate towers are different thickness’, the corner towers are not central to the rounded corners, the barrack XVI and the praetorium are constructed so close together that there is a narrow alley between, which is a maximum of four feet and as less as three, this alley was found to have never been used as such and was used as an area to tip rubbish, the amount of discarded Roman pottery was so great that it took Baillie Reynolds ceramic recorders (his wife for one) many weeks to process the finds, there was so much coarse pottery found (at this time the study was not advanced and the bias was toward recording the Samian) that at the 1938 publishing of the book it had not been catalogued.  What became of this pottery is unknown, maybe it found it’s way to the Rapallo House Museum, eitherway it had no examples of 3rd or 4th century date.  A wall in centurial block of building XVII can also be seen to be misaligned, what’s this all about?  Paul Bidwell in his interesting English Heritage Roman Forts in Britain claims this is common Roman military practice, sometimes showing scant knowledge of building practices, now this is certainly the case with villa construction in the late 4th century, for instance towers are built on steep slopes without adequate footings, result? collapse and rebuilding with buttresses.  But you would have the impression that the classic military machine of the late 1st early 2nd century would have had better knowledge than this, Bidwell states that much building work was done above recently filled in pits or ditches, collapse only followed by new work still unconsolidated, this can also be the case with building over water courses, so the legionary builder of this period shows (in Kanovium for one) himself to be surprisingly inept.  Also Bryn y Gefeiliau was not a good place to build a fort, the river plain meadow is less than a metre above the River Llugwy and flooding was a problem during excavation so must have been a nuisance when occupied.  Baillie Reynolds also draws attention to the north end of barrack XVII which has an odd diagonal corner at the centurial quarters of the building, but added that stone robbers could have been the reason behind this.  And, barrack XIX at the south western corner of the building features an odd buttress.

The Buildings of the 2nd Century Stone Fort

As no extant Roman buildings remain in Britain, and it is fair to say that no complete Roman fort exists anywhere today in what was their empire, we cannot really be sure what some fort buildings actually looked like.  Though in Europe many civil Roman buildings do still exist in a complete form so we do have a fair idea of how some of the fort buildings may have looked.  In particular we have no idea what forms the various buildings roofs took, either gabled, hipped or flat, also wall finishes, had they been rendered or simply wattle and daub? or shingled? such evidence is rairly discovered, and reconstructions of the buildings can often be conjectural..  However a ground plan was obtained and the buildings located.

These building images have been drawn from the Roman Frontier in Wales, originally in Baillie Reynold’s site reports they had been drawn for the book by Maureen Manning, and are used here with acnowledgement.  The green areas generally represent areas outside the buildings, while the black lines are walls, and are actual to scale, the yellow areas are rooms  These images are of the buildings of the mid 2nd century stone fort, which overlay the previous earth and timber layers, but generally speaking are on a similar alignment, especially the central range of buildings.

The first image shows the headquarters building or in Latin ‘principia’ which means ‘the principle building (of the fort).  It was the nerve centre of the fort, the administrative building for both military and religious matters for the garrison.  The most commanding building architecturally, we have a fairly good idea how it would have looked because most of these buildings throughout the Empire conformed to the same plan.  Kanovium principia was very large example and measured 30.5 metres in length by 29.8 in breadth, this poses questions on the size of the fort garrison, Roman army expert Mark Hassall comments‘possibly the large size of the administrative buildings at Kanovium points to rather than the garrison being the 500 sized cohort (Ian Richmond) but the 1000 strong regiment’. Also evidence is available from Scotland - The Romans In Scotland Gordon S. Maxwell, Edinburgh, 1989, that should a fort have a principia of this size then it definitely housed the 1000 strong ‘milliary’ cohort, or else was home to two separate cohorts of the 500 ‘quingenary’ regiments, quin - 500 mill - 1000.   Also Maxwell points out that some forts held a strong garrison, while some, only a token force, well it seems likely that Kanovium was a strong fort and must represent the strategic importance of the area, possibly near neighbour Bryn y Gefeiliau was another strong fort? while next along Tomen y Mur was again a strong fort in that the original fort housed 1000 soldiers.  Therefore if the most westerly fort at Segontium was a strong fort there must exist another fort in between, obviously the weak garrisoned type and must be on the coast near Bangor and would have kept an eye on movement up the Nant Francon pass and the east coast of Anglesey.

So what makes our principia large? well it is slightly larger than the one at Chesters cavalry fort, this is odd as Cilurnum featured an ‘ala’ of cavalry, the ala was a fairly high status regiment, especially the 1000 strong ala of which there is only ever one recorded in Britannia, the ala 1 Petriana stationed at Stanwix, the senior regiment on Hadrian’s Wall, and possibly nearly a influential as the legio VI victrix stationed at York.  Cilurnum had the 500 ala, but they had a higher rate of pay than foot soldiers, and again this points to Kanovium garrison being large.  So bigger than Chesters, Chesters is twice the size of Housesteads Vercovicium, this held a 1000 foot regiment, see a theme here? Caerhun principia was a big building.

The view is looking west, and the building can be seen to be composed of three ranges of rooms set around a central courtyard, which was normal Mediterranean design.  The two bottom ‘L’ shaped blocks of rooms surround the gravel courtyard, which featured a well (most principias have this well, once it was believed to be an emergency well in time of siege on the fort, but is now thought to be for religious purposes).  Entered from the roadway the via principalis (principle street) by an ornate entrance arch the ‘L’ shaped rooms and courtyard featured open colonnades surrounded by verandahs.  The next and middle range of rooms (coloured green here) was known as the ‘crosshall’ simply because it crossed the axis of the whole building, this was screened off from the courtyard and entered by a central doorway.  Beyond the crosshall was the final range of rooms, usually five in number.  The ‘chapel of the standards’ was located here, and beneath this often there was an ‘aedes’ a barrel vaulted underground strong room.  The chapel or ‘sacellum’ was the most important part of the fort because it housed the cohort ensigns, which would have been the flag or vexillum and certainly an eagle standard, smaller than the legionary type but showing the importance of these auxiliary cohorts, in that they actually now had Roman status.  Also housed in the sacellum would have been images of the then emperor and his family, which would have been dusted off and used at religious functions, which would have been fairly frequent, and certainly at pay days (three times a year).  Compare this to images of Saddam Hussein, seen in the recent Gulf War of 2003.  The sacellum was surrounded by four rooms and the use of these is unknown, thought of as possibly storerooms or offices for military clerical staff.   Vindolanda principia had the  ornate stone carved screens, Kanovium too had one of these screens still in situ at the front of the sacellum, it is believed that wrought iron railings had been mounted on top of the stone screens (screw holes in the screen illustrating this) and you would have just caught a glimpse of the ensigns and images within the chapel, which would have been lit by oil lamps at night and by a roof light during the day, which would have added to the mystique of it all, finally the aedes was beneath and contained the cohort bank, all pay and booty being stored here, and must have been a secure spot, the sacellum apparently had a double guard at all times night and day.  The principia was certainly an imposing building, the screens are not the only form of decoration featured, two small sandstone eagles have been found (now housed at Caerhun Hall) perhaps there was many more, and most suprisingly a small stone Celtic head was also recovered from the gravel of the courtyard.  Does it record how Celtic the garrison really was? (despite the Roman trapping it was forced to adopt) or is it an example of a minor rebellion amongst the soldiers?  As mentioned the principia was the scene of military religious ritual, possibly the head has been placed there as an antidote to the Roman magic? well we will never ever know.  The building also contained stone dwarf columns and also remains of some fairly large and imposing column bases (well).  The building also contained several hearths, built from of roof tiles and also had a cupboard of slate slabs built into the thickness of a wall.

The thickness of the walls are a valuable indication into the height of the building, which have very strong footings, this showed the building had two floors, and maybe even three, the third level may even have been a tower, this gave the sacellum a ‘clerestory’ type light effect..  Unusual was that the sacellum walls featured painted wall plaster, sadly in poor condition it was coloured green and yellow and appeared to feature a field of corn, which is no coincidence to what activities the garrison may have been getting up to in this part of Wales. Also found here was remains of a plaster fillet between walls and floor.  These features where not present in the later building, but only in the timber phase.

The stone remains of the principia reflect the massive construction of the building and are therefore the best preserved Roman walls at the site, and still stood to a level of roughly three feet above the footings in 1926-9, hard to believe now as no stone remains inside the fort are to be viewed, but P.K. Baillie Reynolds believed the building to have never been finished, he records some of the dwarf columns may never have reached the site.  The well in the courtyard was found to have been full of demolition rubble (withdrawal of garrison at the mid to late 2nd century AD) and contained more decorative stonework, some with cornices,.  The building was not substantially robbed, much had simply collapsed, including a piece of wall of many courses, it lay were it had fell.  Finally, several examples of local rock being used in the building were recovered and they had arrived from diverse locations, some examples of a very hard volcanic rock which is only found locally at Llangernyw in the Elwy valley, some slate material from Nant Glyn on the Denbigh Moors and some rock which had an effect of marbling and this came from near Deganwy, why the Romans went to this trouble is unknown, along with the stone from the Chester area they were certainly aiming to impress.  A lack of terracotta tile here, and finds of sheet lead, along with a dome headed lead rivet, seems to point to the principia having a lead roof.  At some point in the building’s development it may have had a tiled roof as in emptying the well remains of what looked like an angular roof capping stone was found.

The image left is the ‘praetorium’ or private residence of the commanding officer, I say residence, it took the form of an Italian colonnaded courtyard house, similar to ones found at Roman cities and towns throughout the empire.  Like the principia this example is unusually large for an auxiliary regiment, and is in fact substantially larger than most courtyard houses at Pompeii.  The commanding officer ( a prefect during the late 1st century, but during the 2nd century may have been a senior centurion, or ‘Primi Pili’ as the amount of available polititians to command the cohorts declined) would have been housed along with his family (brought in from just about anywere in the empire) in some considerable luxury, while his comrades slept in draughty sheds with clay floors.

The building featured a private latrine, bath suite and probably gardens, it was abnormally large, but a great deal of ground space is taken up by courtyards, there are four or five and may be flanked by private stables.  It seems that during the long period of Roman occupation at the site that the building received many subsequent additions, in common with the garrison bath-house, more stables, courtyards, gardens and workshops being added at undisclosed periods.  The building however, started off life as the more familiar simple three ranges of rooms arranged around a central courtyard and this original form can be detected on the image as the ranges bottom left which occupies about a quarter of the final building, in this form it is very similar to the praetorium at Segontium, even having a pedestal base in a similar position.  These ranges in common with the principia featured colonnades and verandahs the whole building looked inward and was private from the garrison, away from the dirt and clatter of all the soldiers, excavation failed to discover the use for each room, though did locate a main drain, this drain gave the officer enough water for his latrine and bath suite.  If an aqueduct provided the water is unknown, but the drain ran through the south area of fort to exit at the S-E fort corner and south gate.

The building measured 35 x 35 metres, the size obviously records much later Roman activity, possibly civilian, and possibly connected with the Great Orme copper mines.  M.J. Jarrett claimed that the fort at this time could have actually been an official mint for the usurper emperor Carausius, certainly coins and pottery increase at this time.  I think the mint theory is not possibly provable, but I think Jarrett gets an idea that Kanovium was involved doing something quite important either way, and we do have pottery taking the site down to the late 4th century, which is unusual for an auxiliary fort in Wales.

This is the garrison food storage building known as a granary or horrea, I dislike the term granary, it conjours up an image of Medieval serfs drinking cider, so forage store would be a more appropriate label.  Buildings XIV and XIII on the plan. This image is looking north and features two buttressed buildings with a central yard, use unknown,  Again a larger than usual arrangement, each building measured 20 x 30 metres, the courtyard was actually enclosed and the entire building occupied 337 square metres ground space.  Jarrett thought that maybe there is another granary to be discovered here, so surely the fort was a depot for the collection of local taxes, which as the area had no pre Roman money economy must have been in kind, though if we have the 1000 garrison, if even for a short period then they would have need some feeding.
 

The store was situated near to the north gate of the fort, for easy access to the fort dock just down the vicus, this phenomenon for ease of loading, is noted at other Roman military sites with docks known, Chester and Caerleon.  The building also stands in considerable free space, unlike other buildings which are jammed together badly, this was to allow the large carts used by the Romans easy access, a loading platforms was found, and the causeways out of the fort showed much wear from cart tracks with three re-leveling layers added so obviously much produce was going through the fort.  Produce was of great importance to the Roman army and these stores had to hold one years supply of food for the garrison and their animals, in case of siege, unlikely but possible.  Much of the taxes will obviously have been collected by the Chester legion.

The buildings had eight buttresses either side, the building has very stout walls,  a raised floor similar to a bath-house, or should I say suspended floor, a double thickness wall, louvered) to keep out damp, and a massive stone or terra-cotta tiled roof, with very deep eaves drip to exclude rain water from the walls.  Passageways in the floor and walls allowed small animals access, possibly the Romans used cats, ferrets or terrier dogs to keep the rodent population at bay.  We are unsure of wall finishes, possibly shingled and similar to the reconstructed example at the Lunt fort, Coventry.

A pair of barrack blocks, situated in the north west corner and looking east, XIX & XVII the buildings are 51.8 metres long, including the centurial blocks which are the thicker ranges of rooms at the end of each block, which may possibly have shared a common roof..  Each block would have housed a century of infantry (80 men) or two turmae of cavalry, the centurion, or decurion and various staff and slaves.  These buildings feature misaligned walls, but are situated at the flattest area of the fort, so maybe belong to senior centurions.

Some of Baillie Reynolds’ plans are rather curious, and non so much as his recording of the barrack blocks, basically he fails to record sub divisions of the blocks but these still show up as parch marks, so why he failed to notice is slightly an enigma.  These plans are missing out many features, the blocks had a central sub division a tiny room either side, usualy identified as space for a tent party or contubernia, eight soldiers, and their kit in the other room, roughly 20 pairs of small rooms.

We have very little information on the finish of these buildings, possibly half timbered, or shingle built on a stone, or wood footings, we have little idea of roof form, though finds of terracotta roof tiles, and roof slates, show the blocks had roofs of either, around barrack I tiles fragments were profuse.  At the earth and timber stage of a fort the roofs may have been thatched.  We have no idea if the walls had been rendered, though we do know that the blocks recieved no imported stone, and had had their footings at least constructed in the local shale, much of which can be viewed in the church and its wall..  Doorways and windows are a mystery, though each barrack room had a small fire, in which the soldiers baked bread.  We do know that the walls had been  2 feet 10 inches to 3 feet thick, on the outside, while internal divisions in stone had been 2 feet six inches, of rough local shale which had been quickly faced on one side, these blocks recieved no good white mortar, as had the fort wall but was simply grouted in clay.  During the timber phase the walls had probably been wattle and daub layed on a sleeper beam foundation, the stone phase had removed most of this evidence.  The stone phase walls had been laid on a boulder foundation of one, two or even three layers, sometimes cutting through all timber occupation layers to bed on the natural clay, though sometimes they rested actually on the timber period floors.  In the south east angle the ground falls from west to east and slightly from north to south, this is very apparent today as the south rampart is clearly seen to step down the slope.  Therefore during the stone phase levelling work to this corner took place.  A large timber building had stood beneath building I and IV of the stone phase, this had suffered a very considerable fire (probably a controlled fire and not a disaster) the evidence took the form of about one foot nine inches of ashes, this was mixed with much fused iron, and also found contained in this level was the greater part of an amphora of Spanish type, fat bellied and used for olive oil or fish sauce.  Over this earlier layer the new blocks had been constructed with floors of greenish yellow clay (which is the natural over all the fort site) one foot thick, at the south eastern corner, to level the fall, the floors decreased to about level toward the east gate to compensate.  The barrack block XVIII had a small verandah of slate slabs still in situ, 12 x 8 by 6 feet wide, these two buildings don’t feature the thick end as do barracks in the retentura, so maybe not even barracks, maybe they are stables ( in which case the mounted soldiers may have shared the accomodation with the animals) or workshops (fabrica).  The porch still contained the entrance threshold and floor, which had been composed of slate slabs.  During the stone phase the Romans seemed to have used slate extensively, the north west rampart had been based on a layer of slate slabs, two inscriptions (north and west gates) are of neatly dressed slate.  All the fort drains, north gate, west gate, and praetorium, had been covered by thin slate slabs, all which had fractured and had collapsed into the waterways.  That slate had been used in the Flavian fort is evident by the two slate slabs found buried in the later clay levelling in the building beneath VIII, representing mixing floors for the water cistern which had stood at the north end of this building.  While elsewere I have commented this cistern could have been to supply the bath-house,, it is just possible these buildings could have been fabricae, the water being used in smithing work, but maybe it was simply a stable with a water tank.

Some evidence of wall finishes for the stone barrack walls can be found from Hadrian’s wall forts, it seems they were limewashed and then surprisingly had small  bricks actually painted on to them in red paint.  Pairs of blocks sometimes shared a common roof, which enclosed a collonaded verandah area between, and this has been noted at Pen Llystyn timber fort near Caernarvon, Gelligaer, South Wales and at Deva.  Believed to allow the men a bit more space in the confined forts, and suprisingly the area may have contained gardens, for growing medicinal plants, or simply for pleasure, this practice was ended by Hadrian at a time of slacking discipline in his armies.  Outside each barrack room would have been a waste bin set in a pit for each contubernia, finds of a game board, in XI show the soldiers sometimes had time to spare, possibly in times of inclemet weather..  The centurial quarters are a small range of rooms, and the eighty men live in very cramped conditions, though perhaps with more room than we think because at any one given time half a contubernia would be out doing duties.

The Fort Gates

The East Gate : Porta Praetoria

The North Gate : Porta Sinistra

The South Gate : Porta Dextra

The West Gate : Porta Decumana

The four fort gates of Kanovium, all probably of slightly differing time periods, the sinistra gate towers recessed into the rampart wall, the praetorian gate however is flush, decumana and dextra are single tower set very far back into the rampart.  This reflects work being done on an ongoing basis, possibly from the time of Trajan to the time of Antoninus Pius, while the sinistra looks Hadrianic, the single towered gates look 3rd century.  The building work shows some discrepancies too, misaligned walls with differing thicknesses, this is down to different work parties at different times who wouldn’t be bothered to rectify someone else’s mistakes, this was common Roman military practice and must have invented the saying ‘thats not my job mate’ the gates are as follows-

East, Praetorian gate -  Named ‘praetorian’ because it is located on the street the via praetoria, of typical early 2nd century construction date the walls of the towers are built up to the rampart face.  A monumentally large twin towered structure with central ‘spina’ which is still in situ, from which the arches of the roadway sprung, the whole structure is 60 feet in length (the previous timber gate had been 42 feet wide) and is the biggest of the four gates, rather peculiar as it is basically a folly, meaning really it doesn’t need to exist at all, a Roman camp was always built to the same design,  there is no roadway emerging from this gate as the bath-house would be in the way, and also there is only the river beyond.  So basically this gate didn’t need to be built, but the Roman soldier always ‘built by the book’ so the biggest gate goes nowhere.  It was seen that the original turf rampart had to be cut back and levelled before the guard towers could be built. This gate yielded the remains of the dedication slab which would have been set up above the roadways to record the emperor, and the garrison that worked on the fort, only two letters remained,  an O and an A, a shame as it would have told us the name of one of a garrisoning unit, either way it is the first recorded use of Snowdonia slate.  The previous timber gate had been proved by the finding of six postholes and a causeway of river gravel, which ran up to, but not out of this gate.  The considerably larger 2nd century gate not only overlaid the earlier gate but it also, along with the east fort wall projected 13 feet further east than the timber fort, this meant the stone guard towers now stood dead centre of the earlier earth rampart.  During the timber phase an open drain exited the fort at the wall of the south guard tower, during the stone phase the drain was recut and used the metaling of the timber phase roadway as the base (levels had actually risen that high in the 30 or 40 years) this was now covered by slate slabs, its destination could not be traced due to the fact outside the fort it was destroyed.  The walls of all drains at Kanovium used local stone and boulders, these are about three feet deep.

The east gate was  found to be tragically robbed, though with all of its footings though still present, most of the superstructures, including the fort wall had been absolutely removed down to ground level.  The north guard tower could not be completely excavated as it ran beneath the graveyard wall.  Interestingly Baillie Reynolds records the walls here vanishing beneath to still be four courses high, therefore the fort was radically robbed by the time this the modern wall was built (when? well certainly before 1720 when the lychgate was added).  So this gives us a valuable indication of what was remaining three or four hundred years ago, when it must have looked in the same condition we see Segontium fort at Caernarvon today.  Most sad to me was the description of the height of the Roman level and what remains today, if you view the rampart here it still stands to a respectable 5 or 6 feet but the floors of the guard towers actually lay at a height higher than the remains of the wall today.  So it is difficult to image this 60 foot gate here, with the twin towers rising to a height of 20-5 feet.  A find of a block bearing a cornice here proves the gate was subject to some decoration, the block is either Cheshire sandstone, or Bodysgallen stone, which is from Llandudno.  The stone of Llandudno appears to be mostly limestone so the place this rock was quarried is unknown today..

 The use of stone from widespread sources is also another Kanovium enigma : stone has been quarried in the following areas and used here, mostly in the high status buildings, but also fort wall, from Llangernwy, a small village east of Llanrwst, quarry unknown, but the yellow volcanic rock is used in some farm houses of the village, and an old resident once informed me his house was built of this ‘special stone’ but he neither knew the quarry or any connection with the Romans.  Nant Glyn, on the Denbigh Moors, this was 20 miles from the fort and no records of any roads exist over the Denbigh Moors, was found to be a hard shaley slate and may have been used in place of material from nearby Snowdonia, unlike the fort at Capel Curig.  Knutsford, Tattenhall in Cheshire, a white sandstone used  for a plinth of the west gate.  The small carved head found was of Cheshire stone.  Lleyn peninsular or Foel Fras Mountain, provided a micro dolerite used for corn grinding querns. Local granite for pot boilers, possibly from Penmaemawr which is close to the west road at Bwlch y Ddeufaen  (these stones are heated in a fire and then added to water to heat it up). And finally  sandstone from Runcorn, Helsby and Frodsham, the quarries now overlooking the modern M56 Motorway

The South Dextra Gate - Features a narrow roadway with single setback guard tower, and is odd for a 2nd century fort, you would expect much traffic at this time, with twin roadways, but this gate seems to indicate a decrease in garrison, with resulting decrease in traffic.  Baillie Reynold excavated the floor of the tower to only find 2nd century Samian fragments, absolutely no third century material, but he does record two layers of road metal through the gate, a considerable rise in floor levels from the 1st century fort, and the gate tower incidentally is two metres or 6 feet set back from the rampart, proving the stone towers are built on the site of the timber towers of the earlier fort, the stone rampart wall then projecting beyond the towers.  The history of this gate is as follows, in the initial timber form it comprised of twin entrances but no guard towers, but absolutely no trace of this gate remained.  It was found to have a road 15 feet wide, while the whole gate was only 17, which was of river gravel, this rested on a clay foundation which sloped as it left the gate to compensate for the drop southward.  The road had a pronounced camber, on the western side, and the later 2nd century re-lay copied the camber exactly, both showed evidence of considerable wear from wheeled vehicles, in the form of grooves.  A drain ran out through the western portal to exit into the ditch.  The ditch was continuous here and was crossed by a wooden bridge.  At the later stage the stone gate was planned on the same dimensions as the earlier, having twin entrances of 6ft 6 inches on the west, and 8ft 6 inches on the east.  The gate spina remained, providing evidence for a double arch structure, but it seems a change of plan occurred and this stone gate was never completed.  Firstly at the western portal the ditch was filled in and a causeway built, but the eastern still retained the bridge.  Then the western entrance  was removed and a blocking wall was inserted, and the east entrance became a guard tower, its southern wall was 8 feet thick and this was because it had to support the wall walk, and also maybe a decorative arch.  During the 1929 excavations a fragment of a sandstone plinth was found here, so this gate too was decorated.  At some later undisclosed date the gate was subject to some re-occupation in the form of rough walls, added badly to the ruined Roman walls, these may simply be 17th -18th century out buildings associated with the post Medieval cottage that had stood at the south western angle, excavated in 1929.

It seems the reduction in the size of this gate must date to around A.D 125, while the north and east had been completed on the classic twin portal, twin roadway design, a movement of troops northward for Hadrian’s Wall meant the large gates now became superflous to the amount of soldiers garrisoning the fort, so these smaller gates became the normal.  So it shows that a concious re-build in stone had indeed started very late in Trajan, but Baillie Reynold seemed to imply it was done certainly slowly, but these gates show it was still being done maybe 8 years later, and here a change of plan is evident with the reduction of this gate.  Finally the west gate is completed, and then the stone fort wall is added circa A.D 120-50, which gives a total of about 28-30 years to fully re-build the fort in stone.  What has happened is really there was no concious plan to re-build in stone, the fort had long since been considered important and permanent and buildings are just replaced piecemeal when repair was needed.  Around the time of the change at the south gate the work on the principia was halted and it remained incomplete.  But most of the fort was now completed on a grand scale, but the days of such ostentation were declining, what exactly happened in the last years of the 2nd century is unknown, by A.D 200 it was torched and was left derelict.

The West Decumana Gate - This gate stands now facing Caerhun Hall and is similar to the south gate, narrow single track road, and again a set back single guard tower.  Surviving now as a slight break in the badly ploughed west rampart this structure was so badly stone robbed as to be very difficult to work out a plan, in fact even some of the footings had been robbed and the excavators had to trace depressions left by the stones in the clay floor packing.  The original timber gate had entirely vanished and was only represented by the gravel causeway which was still present, this was 10 inches thick. Featuring twin roadways 8 inches apart, the north one was 6 ft 6 inches wide while the southern one was at least 7 ft 6 inches wide. Rampart evidence proved that in the original form it had either one or no guard chambers.

The stone gate consisted of a single entrance with a guard tower on the North side. Less than a foot beneath the ground, very little remained. The sandstone blocks lying nearby showed that this gate had been constructed of this material, and a fragment with a corner moulding suggests that there was some achitectural decoration. A piece of a slate slab two inches thick with a moulded border is remains of a dedication slab which must have been displayed above the arch of this gate. The stone gateway was bigger than the previous timber gate, some of the earlier rampart having been removed prior to building the north guard tower, being in common with the east gate. Most of the foundations of this gate had been constructed throught the timber phase into the natural. The road of this second century period was still in good order and consisted of quarry gravel in difference to the earlier period fort which had only used river gravel. All the gravel of this second century fort was this quarried material, its source is unknown but the Excavation Report claims it is still quarried a mile away at Tal y Cafn. The road still showed its camber which was over a foot thick and separated from the first century road by eight inches of puddled clay.

It seems evident that this gate was the final fort gate to be built here, while it exhibits similarities with the south gate, it exhibits no evidence of reconstruction or design change so evident in the south gate. The front of the guard room was eight foot back from the line of the rampart. This gate according to R.G. Collingwood points out in the Excavation Report that these two gates are very similar to the gates at Hardknot fort which at that time they believed to be contemporary. Baillie Reynolds records absolutely no blocking walls at this gate which would indicate third century activity. For a page concerned entirely with the archaeological excavation at the west gate please follow this link.

The North Sinistra Gate - Finally the sinistra gate, literally the ‘sinister’ bad luck to leave by gate, a more normal 2nd century gate in that again like the east it has twin towers and roadway out of the fort rampart, but differing in that the towers are recessed into the rampart, the twin towers seem built on differing alignments with the usual misaligned walls.  This was the gate excavated by Samuel Lysons in 1801, which he described as ‘twin sepulcres which contained human remains ’, the sepulcres have now vanished beneath the modern car parking area, and absolutely nothing is to be seen, though possibly footings will remain beneath the fence and stile, P.K. Baillie Reynolds records the remains of this gate to be seen beneath a hedge, but now the hedge has also vanished. Bailiie Reynolds found that all the Roman floor levels had been removed by Samuel Lysons and the best he could do was obtain the simplest plan. But he couldn’t even get this in its entirety, while Conway District Rural Council allowed him to dig up their road it seemed the church authorities were loathed to allow him to remove a yew hedge which at that time covered most of the east guard room. The spina to the gate was not present but there is a local story of ‘a flight of steps’ (which must be the spina) being removed by a horse and cart many years before. This gate features set back guard towers which are as much as two feet three inches back from the line of the fort wall. It is in the authors opinion that this gate was probably the first to be re-build in stone at the end of the reign of Trajan. The east gate following after around 125 AD then the south gate, and finally the west gate.

So what did these gates look like?  it is thought they would be three levels high  with a pitched roof, and narrow arched windows devoid of glass, many well preserved examples of these gates can be viewed at Hadrian’s Wall, some like gates at Birdoswald and Housesteads still standing to the height of the arch springers, also fairly good examples can be seen at the Caernarvon fort.  There is recent evidence from Egypt in the form of graffiti that indicates these gates had four floor levels, Roman Warfare, Adrian Goldsworthy, Cassell, 2000. There was no access from the towers from groundfloor level, access usually being from the rampart walkway, possibly with castellations, the towers are occasionally solid for the mounting of spring guns (as at Caernarvon) this is a later innovation when the Romans are on the backfoot, so Caerhun is mostly indeed 2nd century, the threat of enemies attacking forts not yet being a possibility.  It seems obviously that the majority of these four gates ended up being the material to build Caerhun church many hundreds of years ago, one door jam stone can be seen high up in the bell tower, right side eight course above crucifixion carving.  The well cut ashlars must have been very appealing to the church builders.