The Earth and Timber Fort at Kanovium

© Gwynedd Archaeological Trust.  This image must not be subject to reproduction, transmission, display, lending or storage in a retrieval system without prior permission of copyright holder.  Thanks to Nina Steele

All of what remains to be seen today of the actual fort earthwork (not vicus, bath-house or other associated features) dates from the early 2nd century A.D, when the fort was rebuilt on a grand scale in stone, though the rampart base and core (invisible) originate from the timber phase.  The sources for the stone used had been numerous, and included Chester, Runcorn, Tattenhall, and Knutsford. And closer to home from Mynydd Hiraethog (often erroneously named as the ‘Denbigh Moors’),  Llangernyw, Llandudno, and of course much shale in the form of boulders and rubble from areas adjacent to the fort.  While individual items such as pot boilers had been made from local dolerites.  Pot boilers were stones heated in a fire and then dropped into a pot of water, either for cooking and also for use in saunas, this type of rock is to be found high on the shoulder of Drum mountain several miles west of the site, or from further away on the Lleyn Peninsula.  Most references in this website are to this stone refurbishing phase, and perhaps it is not so readily known that for forty years prior to this the hillock had been occupied by an only semi permanent ‘field fort’. 

The timber fort would have been constructed from good quality squared (and triangular) section timbers for the four gates, and inner buildings, usually for the 'sleeper beams' which were the foundation blocks for the walls, which were either shingled or wattle and daub (E. Shirley, Building a Roman Legionary Fortress, Tempus 2000). No timbers remained from this phase at Caerhun. The actual fort wall of these forts was usually constructed of turf blocks, roughly 18'' by 6'', neatly laid in courses.  The back of the fort wall would have received a substantial clay packing bank, on average twenty four feet wide, which formed the core of the rampart up to the level of the timber wall walk.  The rampart may also have been constructed in a 'box' manner which was typical of Iron Age hillfort construction, or simply have been re-enforced with the inclusion of timber piles.  The front face of the rampart or 'cheek' would have been pounded and beaten and would have baked hard in a summer's heat, it may also have been strengthened with shingles or a further timber revetment.  The inner face of the rampart at the S-E corner had been gravel washed, with gravel from the nearby river.  The rampart base here was a layer of rough boulders set in clay upon which the turf rampart was built.  Fronting a rampart would be a flat ground of maybe six feet, this was known as the 'berm' often called 'the killing ground' which is probably a romantic misconception, as the berm only served as a strengthening area of ground, as beyond this was the ditch system, if the ditch reached the base of the rampart then the rampart would gradually slide into the ditch.  The earth rampart would have stood to a height of of between ten to fifteen feet, and this would also have included the timber walkway palisade which would have been about four feet higher, possibly castellated, none have survived in Britain to prove this point, this allowed a degree of protection for personnel on the rampart.  It may surprise that the rampart only reached this height and there are two reasons for this, for one the Roman army of the 1st Century A.D were soon to reach their optimum peak of performance and never considered these forts as defensible, and that a taller rampart would have been more likely to be unstable and also occupy more ground area inside the fort.  Certainly many Roman timber military structures could be much taller, as at the German Limes, and Alesia, France, but a build height of this nature was adequate for a new fort, as it is possible that the local inhabitants of newly annexed areas had been suitably subdued prior to the fort building program in newly conquered territory.  This rampart height can be compared to the timber and earth period fort wall discovered at the fortress of Deva, Chester, and illustrated in Graham Webster, The Roman Imperial Army 2nd edition, 1978, here the fort walk way was about six feet high, and the stone rampart here at Caerhun (in the possible original height seen on the S-E corner) was possibly only six feet tall at walkway level.  The reason for this was that Roman forts were always protected by a complex ditch system, usually about eight to ten feet deep at the base of the berm, add 10 more feet for rampart and walkway, we have an obstacle totaling nearly 20 feet in height.  Add to this a cunning slot at the base of the ditch, which probably is simply the spade cut to clear out the ditch bottom, or more prosaically to break an attackers ankle, or pointed stakes, thorn bushes or sewerage outfalls (probably attested at Caerhun, S-E corner) then the small fort would not be an attractive proposition to storm.  Add to this the fact the fort could have two, three, four or even five ditches, and that the Roman soldiers will have been actively engaged in the flinging of hot, spiky or just plain dangerous (and cheap)  objects such as stones at the hostiles only goes to re-enforce the usual Roman flair for business like and user friendly building methods for their military installations.  These forts were entered usually by four gates, one at each rampart, at a midpoint, these gates at this stage usually had twin entrance roads, sometimes one for personnel, one for vehicles, each guarded by a timber tower, constructed of stout timbers.  Towers were also to be found at the four fort corners, a post hole found beneath the stone phase tower in the S-E angle is the only trace of the four timber towers that would have stood beneath the stone ones.  Interval towers between gates and corners are also common, but none had been detected at Caerhun.  Shirley considers these forts over engineered, they are a common feature of the Roman army of the 1st century A.D, had a life of possibly about twenty years, the timbers had to be replaced and the turfs replaced if slumping occurred, how do we know this?  In Britain there has been some examples of timber period Roman military reconstructions, Vindolanda Roman fort on Trajan's Stanegate, and also on a greater level at The Lunt site, near Baginton, Coventry.  At Vindolanda it is possible to see a short reconstruction of the earth section of Hadrian's Wall, which was built on a similar design to the field forts, a cobble base with laid turves.  Hadrian’s Wall, initially, west of Harrow's Scar milecastle had been built in this manner, reason not known, but it is surmised there was no limestone available nearby to make mortar.  The example at Vindolanda illustrates both the weathering of the timbers, and the slump of the turf bank.  The Lunt site is a much larger undertaking, the site of a genuine timber fort of the late A.D. 60's, excavation had reached a point by  1970, and some of the fort inner buildings, and rampart(s) and gates were reconstructed on the sites of the original structures, undertaken by a group of soldiers from the British Armies‘ Royal Engineers, using original materials and still serves as the best example in Britain of a Roman timber fort, if just to show how they decline.  The east gate of The Lunt had been planned from evidence discovered during 1966-7 and it took the Royal Engineers three days to build it (prefabricated in army workshops) without using any modern haulage gear.  Later the rampart was reconstructed, initially the turfs had been laid grass up, but the rampart slumped badly in the first year, so it was decided that they must have been inverted, once replaced the rampart stood reasonably upstanding, though by the early 1990s it had slumped beneath the timber walkway by as much as four feet, and looked rather bizarre, add rabbit burrows (not a factor the Roman army would have had to contend with, at least not in Britain) then by 2003 the site is now partially closed due it being in a dangerous condition.  The Roman soldier being more proficient in the construction of its stations, and the fact they usually had plenty of time on their hands would have meant that these forts were kept in better order.  The timbers inserted into the ground, often oak, would receive stone lined pits, but even so after 20 years, rot would have ensured all or most would have to be replaced, this in common too with the sleeper beam foundations for the inner buildings.  The process may have been continuous, and it is likely that during 30 years of occupation that there would have been at least two forts, each on top of its predecessor, and each sealed with a generous spread of clay and gravel.  The ditches, on evidence from the Lunt, would need clearing out once a year, but nine months would be sufficient for the clear section to be blurred by soil and silt wash, usually the destination of the fort drains, they would have been smelly nasty areas, and possibly the source of offensive odour during the summer months. 

At this stage of Caerhun forts development the inner buildings were all of a rather shanty nature, though it seems likely that inner walls may have been rendered to a decent quality, with roofs of good red tile, or maybe thatch initially, though most floors would only have been of beaten clay lined with straw.  Some elements of the fort such as the Principia (headquarters complex), the granaries, cookhouses, and water cisterns, may have been constructed in stone from the outset, and all the roads within the complex were constructed with stone boulder kerbs and a running surface of river gravel, or occasionally cobble stones, again from local rivers.  The fort may only have been required for a year or two, if an area seemed receptive to 'romanisation' the army passed over control of the district to the local worthy and council (tribal chief and ‘ordo’) and moved away in an orderly fashion, taking with them what timber remained in a good condition and possibly torched what remained (possibly attested at Caerhun fort).  If the district did not settle then the Roman army remained, in due course as mentioned above the decline of the timbers, and continued stay in the area meant the fort was constructed from the ground up in stone, it was then considered a permanent installation, this took place in Wales during the early to mid 2nd century A.D, or even much later in the 3rd century (Caernarvon, Caerleon) and many of these early earth fort sites continued to hold Roman garrisons into the 4th century and their names are brought down to us by surviving portions of Roman road books, the Ravenna Cosmography, and the Antonine Itinerary. 
 I thought therefore, it would be useful to give an indication of how the early version of Kanovium fort looked, and the references used here are from Paul Kenneth Baillie Reynolds 1938 Kanovium.
The stone fort had obliterated much trace of the first fort at Caerhun, due to it mostly being sited directly above the former, though the timber fort projected thirteen feet less overall at the eastern rampart.   For instance, the south guard tower of the east gate stood upon what had been the timber phase rampart which had been cut back and flattened.  Many features of the stone fort being cut through the timber phase layers to reach the natural hard clay level, it was also of a similar size, therefore it proved a bit taxing for Baillie Reynolds back in the 1920's at a time when the identification of timber features was in its infancy, so unfortunately he gave us a very limited perception of the earth fort.  There is only one line drawing of the ground plan of this early fort (figure 3), some features correspond to the stone phase, elements of the Principia, and barracks in the retentura, but some of Baillie Reynolds’ buildings don't convince, many are simply on a similar line to the later phases and in this author's opinion have to be disregarded, also some features do not seem to be correctly placed within the fort, so with these points making a summary of the earth and timber phase difficult, this is my idea of what remained on the first fort back in 1926-9.  Post holes  indicated that timber gates had stood on the site of all the stone phase gates, and in the case of the south and west gates may have been larger than in the stone phase.
Baillie Reynolds concluded that there was one timber phase occupation level, which even featured three stone buildings (though possibly these were added during the 20 year refurbishment period) these were the strong room in the Principia XII, a water cistern in VIII, and a  cookhouse XV cut into the N-W rampart inner face.  There was also slight evidence of an earlier level, which may represent occupation prior to the arrival of the Romans, but more likely represented occupation by the first Roman troops employed in the construction of the field fort.  A Colchester type fibula (cloak fastener, or brooch) of a Belgic early 1st century A.D type found at the site casually last century (Bangor Museum) gives some credence to this theory of Iron Age use of the site.  Mostly in the form of small spreads of ash it may also represent the fires of the builders of the fort, Legio 11 Aduitrix, this is evocative, and was found when cutting through all of the Roman levels inside the rampart area.  The rampart was founded on a bed of rough boulders, Baillie Reynolds concluded that these occur naturally at the site, so seemed to have been utilized whenever possible, the entire circuit of the rampart did not receive this boulder base, reason unknown.  However there was also a two foot plinth of these boulders which projected from in front of the rampart  continuously throughout, but it did not always project as boldly, the west rampart plinth projected two feet further out than the at the east rampart, probably due to the east rampart being constructed above a steep ground fall   The rampart was 24 feet thick on average, the inner core of clean blue clay brought up from the River Conwy, the inner face occasionally seemed to be 'pebble dashed' with a covering of the ubiquitous river gravel, used throughout the fort at this stage.  The 24 feet of rampart base suggests the fort would have had a walkway of about 8-10 feet at the most.  Fronting the base plinth was a berm, this was 5-6 feet wide, this led to the single ditch system, this occupied 24 feet of the meadow, therefore the distance of the ditch up to the rampart base was nearly 30 feet.  The ditch was divided with a 'midrib' which as the name suggests was a rib of ground which rose from the base of the ditch up to ground level, and may have been an additional obstacle to attackers.  Kanovium's midrib was built at 18 feet from the berm, it was two feet wide, finally the outer section of ditch was 4 feet wide.  The inner ditch section was of a fairly smooth profile (which may represent later recutting) and was 8-9 feet deep, the outer section at most sections tested had the afore mentioned cleaning slot and was on average five feet deep.  The ditch section of the east rampart facing the river was found to be of lesser dimensions due to the slope of the ground down to the bath-house.
 
A Building in the Pretentura S-E Corner 

Excavation early in the season of 1926  located one long building in the south pretentura, while  most of the north pretentura had formally stood in the area now occupied by the parish church and graveyard, and was probably lost to archaeology forever.  The long building, the length of which was not ascertained, commenced south of the via praetoria road at the east gate and ran south, parallel to the east rampart, and indeed it was constructed directly up against the clay packing bank of this rampart.  It ran for at least 120 feet, and was located in the rampart section E-F, it however was not detected at the trenches at the S-E corner, it was found to be probably about 13 feet wide.  It had been found to be destroyed by conflagration, and the deposit of ashes, which was as much as 1ft 9 inches thick, was sealed by the floor of the later barrack block I, which dated from the stone phase of the fort.  The deposit of ash was found to contain ‘considerable pieces of charred oak, a few distinguishable nails, and many lumps of fused iron.’ The building had been constructed directly on the natural ground surface, Baillie Reynolds did not hint at the use, so it may have been a barrack or a workshop, as either building would have left similar deposits in the case of a fire.  A post hole associated with this structure was found 54 feet from its north wall, this was 9 inches square.  40 feet south of this post hole, a trench indicated this building had ended as a causeway was detected, this was 9 feet wide as made, it was partly covered by the ashes of the timber building, and also cut by the later stone barrack I.  Baillie Reynolds concluded this causeway was the early intervallum road, via sagularis, ‘cloak street’ though how it functioned within the east rampart is difficult to imagine as the timber building appeared to occupy the space it would normally be expected to run in.  It was discovered that the floors to these Flavian buildings were made from pounded brick and tile, actually dust, though pounded clay (6 inches thick was also commonplace) this type of floor was particular to the earth and timber fort, and was not noticed in the later phase.  At a later date the timber building was sealed by a layer of clean hard blue clay, and barrack 1 built above this, the clay levelled the area of the fort, which formally had sloped from N-S, and while level at the via praetoria road, became a foot thick toward the S-E corner.  The date for the reconstruction of this sector of the fort is indicated by the finding of the major portion of a buff amphora within the blue clay layer, and it was dated to A.D 80-90, the addenda and corrigenda to the report on the coarse pottery states thus-
'Buff amphora of normal Roman type, unstamped ; found in fragments in the layer of clay covering the ashes of the burnt timber building beneath Building 1.  (AC, p311), cir. 80-90 A.D.  About two-thirds of this amphora were found, including the neck and base.  The restoration was begun by the late Capt.  W.H. Gough of Caerhun, but he was unable to finish it.  After many years the work has been admirably completed by Mr  V.E.  Nash-Williams and his staff at the National Museum of Wales, to whom I am greatly indebted, both for completing the work in time for the inclusion of this note in the final report, and for providing the photograph for reproduction' Thanks to the unremitting energy of Dr.  Willoughby Gardner and the co-operation of the Llandudno Town Council, practically all the objects illustrated in this and the two preceding reports are now housed in the Rapallo House Museum at Llandudno' And to this day there they remain, the amphora is of the squat round bodied type and is very similar to the example found at the Capel Curig Roman fort, to be seen at Bangor Museum.  This shape of amphora may originate from the Tarraconensis (Costa Dorada) area of Spain, and contain fish sauce, olive oil, or wine, Joan P Alcock, Food in Roman Britain, Tempus 2001 indicates an amphora reached Caerhun fort containing cherries.
 
Building IX.

Mr Baillie Reynolds curiously called this a 'building' which it clearly was not, it was a sunken pit, and was located at the north end of what would become building VIII (a probable sleeper beamed timber walled, long building, with a misaligned front wall, facing the via principalis roadway).  The structure was actually in the S-E angle of the via principalis/via praetoria roadway, the pit was found to be as follows - identified as a timber phase feature, it was recognized as a cistern, and it measured 16 ft 6" east to west by 10 ft by 10" north to south. It was sunk into the ground about 12 inches, and the walls were preserved to a height of about 2ft 6 inches. Built of local grey slate slabs set in clay the inside of the pit was originally plastered in a cement which contained a large quantity of brick dust, this brick dust having much in common with the composition of the floors used in the Flavian period of the fort. The base of the cistern was pounded clean blue clay, though its base was not regular and varied from 9 inches thick in the south part, to 2ft thick in the north section. 2ft  south of this cistern was a pile of this blue clay resting upon two slate slabs. Also around this area was a small heap of cement and the brick dust, and it is likely that this represents the mixing floors used for the construction of the cistern and buildings walls. This cistern was not part of the original Flavian fort and in common with the underground strong room of the timber principia, is likely to represent a transitional period between the better defined main periods of the earth and stone layers respectively. This is indicated by the fact this cistern appears to not correspond to building VIII of the stone phase. What is this cistern? Baillie Reynolds tentatively claimed it may have been a water supply for horses and therefore represented a stable, it could equally have been a water supply for a fabrica (fort workshop). It could also possibly have been utilized for the bath-house directly beneath, east, it could also possibly have been a pit used for fulling as Roman towns in Southern England have revealed similar cisterns containing traces of Fuller’s Earth.
 
The  South Intervallum

This was excavated in 1926 and between the S-E angle and south gate there was found cookhouses. These were between the rampart and intervallum road, these were perceived as timber occupation period levels, two levels were located, the earlier level in common with most areas of the fort representing temporary construction levels in the form of slight spreads of fire ash which must have survived from the original Roman troops who built the fort.  Baillie Reynolds' section G-H sliced through a circular oven floor 12ft in diameter made from water worn stones from the Conwy river circled by large boulders based in clay, this actually cut into the back of the rampart. This was accompanied by much charcoal and clay burnt very hard representing long use of this oven. This oven had a post hole 5" in diameter, in common with an oven located next to the well in the south intervallum, it therefore seems likely that these ovens/ kilns had been covered by a timber structure. Further east of this oven there are possibly three more. Between these ovens was a clay occupation level covered in 9" thick of ash and must surely represent the rake-out of these kilns.
 
The Commanders Private Residence- Praetorum, Building XI.

This was investigated in 1927 but sadly no trace of the timber building that must have stood on this spot prior to the stone building, remained. But there was evidence of timber period occupation whenever trenches were put through the second century stone building, mostly in the form of pottery and coins.
The natural surface of the ground was found to be 2ft 6 to 3ft beneath the modern meadow and as elsewhere thin floors or fires representing the troops building the first fort were plain to see. The floor was sometimes of a slightly more substantial nature and was uniformly laid in the characteristic Flavian red brick and tile dust.
Also discovered here were gravel paths within this building, which had been very thinly surfaced, Baillie Reynolds concludes by saying that 'In general the evidence for the timber period (Praetorium) was disappointing'.
 
The Timber Principia. Headquarters Building XII

It was found that beneath the later stone Principia there had been a timber one, this had at an undisclosed period a stone lined ‘sacellum’ inserted in the middle room of the backhall.  It was thought that this new room represented similar period occupation to the stone lined cistern discussed above.  It is rare but not uncommon to have a stone pit in a timber principia.  Baillie Reynolds described the phenomenon as follows  ‘A stone Sacellum in a timber fort is unusual but not unparalleled.  The earlier fort at Newstead appears to have had this arrangement, and analogies may also be quoted from the German Limes at Feldburg the Sacellum and one other room, and at Kapersburg the Sacellum and its flanking rooms were of stone, the rest of the Headquarters being of timber in both instances' this was during 1927 and the timber principia was again examined in 1928, as this initial search only took the form of trial trenches.  As usual the early timber phase was preceded by an earlier occupation layer, again presumably the construction layer.
The principia posed problems as it was conjectured that the sacellum was inserted in an earlier building, possibly indicated by two postholes, the unpaginated nature of the site report making interpretation from my point of view difficult as sometimes Baillie Reynolds has differing opinions which are not always easy to interpret.  If further reading in this website produces inconstancies in the descriptions of the building phases then this is the reason.  Baillie Reynolds concludes that in the first instance the principia was of timber, as with the vast majority of the other buildings, but later in the Flavian period a stone sacellum was inserted and decorated elaborately with painted wall plaster, see below, and that wall to floor joints were fitted with a rounded plaster fillet.
The decoration of the sacellum is interesting, the painted wall plaster was of a cream background bordered in black and green, and upon it was a design or depiction composed of red and yellow stripes, one portion in particular had traces of what appeared to be depicting green leaves with yellow stalks, this plaster had the typical keying marks applied to the reverse, this allowing it to bond with the walls, very little of this was found in situ.  In association with this plaster remains of a quarter round plaster fillet was discovered, and beneath all this the floor was clay, a foot thick, laid above the natural ground level.  At an undisclosed period between the additional work done to the timber structure and then the final stone building a sunken pit was dug in the floor of the sacellum, on the north wall of the room, and was likely to either contain the regimental pay chest, or even religious imagery.  The latter in view of a similar positioned pit in the Principia at the Bewcastle outpost fort at Hadrian's Wall, Fanum Cocidii, the pit was found to contain two lead plaques depicting the mysterious local Celtic Deity, Cocidius.  Cocidius was probably a local hunter god, of a possible dark nature, he is depicted on the plaques as a amalgamation of the Roman legionary soldier and the Celtic wheel god Tauranus, it is odd but not unparalleled that the Roman troops would not wish to offend Cocidius, but to actually place them inside the sacellum, a very 'Roman' place is strange.  During the 2nd century at Kanovium principia an undisclosed individual buried a small gritstone (Cheshire sandstone) Celtic head in the gravel of the principia yard, so the similarity in appeasing local deities may be similar.
The elaborate sacellum was fairly short lived, we are unsure to the exact date of the commencing of the final stone phase, but possibly late in the reign of Trajan this commenced, the sacellum was demolished as the later Principia stone was bigger throughout by several metres, but before demolition the painted wall plaster was hacked off and tipped into the pit.  The denuded walls of the sacellum were sealed with clay and a small amount of plaster and fillet remained in situ to describe the original design.  I feel we see a period of no occupation between these two buildings as the pit not only contains wall plaster but also it 'abounded in fragments of roof tiles (a tile roof in the timber period, but a lead roof in the stone?), and ‘in shells of oysters, cockles and mussels’, but there were only ‘half a dozen potsherds' It would seem this area had become strewn with domestic refuse, but not many pots showing people were not living in the building, it is unlikely a Flavian principia would display such shoddy house keeping? certainly in the later Roman periods but not at this stage?  Surely this could indicate a reduction in garrison even as early as A.D 80? this is in common with the nearby fort at Caer Gai, Bala, Caer Gai seems to show a change in unit around A.D 86, most definitely linked with the evacuation from the province of the Chester legion Legio 11 Adiutrix.  As here the control was passed to the new legion Legio XX Valeria Victrix, who brought new cohorts of auxiliary regiments with them, the former North Welsh auxiliary garrison withdrawing with Leg 11 Ad.
Also noted in the timber principia was a zig zag line of post holes they represented the walls of the backhall, ie the central sacellum and two flanking rooms either side, this would be replicated in the stone building, and there were 17 of these.
The building had the usual gravel courtyard, it has been quoted that this courtyard, which was in front of the backhall but between the fronting rooms facing the via principia, was large enough to contain the entire regiment, in times of speeches or ritual.  A pit was located in the courtyard, of a rectangular shape it was possibly the robbed out pit for a stone block which housed a 'tribunal' this was a dais that the commanding officer stood on while addressing the assembled garrison, and is roughly in the correct spot comparing examples at Chesters, Vindolanda, and Housesteads.  A drain was located along the south wall, adjacent to were the timber Praetorium stood.
Further evidence for the timber phase it not profuse, other long buildings can be seen in both the retentura and the pretentura but these largely coincide with the later stone barracks, stables and workshops so description is not required, though some also supplied evidence of the Flavian floor type of pounded orange brick, and also hearths.  There was an interesting hearth located in the south intervallum, covered by a wooden structure illustrated by three postholes, this was sealed by the later stone phase road, it contained 30-40 clay sling shots and it is a graphic illustration of activity inside the fort, did they leave in a hurry? as they never managed to retrieve the missiles.  Also located near this oven base was a well, it contained timber phase demolition material, and was sealed with clay and rubble during the stone phase refurbishment.
All the roads within and without the fort in the timber period were the distinctive local river gravel, of all sized stones including some small boulders, this would be obtained easily from the many local brooks, and indeed for the foreshore of the River Conwy, around twelve feet wide for exit roads and about six to nine feet wide inside the fort, usually about 8 inches thick always with a pronounced camber, and bounded by boulders some quite massive and not always on both sides.  It was seen around the join of the via praetoria and via principalis that a huge boulder had been placed midway at the join of the roads, this was to deflect water away from the building VIII, and down the road and out of the east gate.  Also nearby was a curious drain composed of flat roof tiles, this to take the eves drip away from the same building, during this stage the tiles (tegula) were used for almost anything, latrines, drains, cupboards, and hearths, and possibly as Baillie Reynolds thought, confirms the building was only of half timbered construction.
The building XV west of the north gate abutting the N-W rampart was a stone cookhouse and would appear to have been built as such from the outset of the fort, but in view of the nature of the use it is not surprising.  It continued in the same position throughout the Roman occupation and may have continued to have been used into the post Roman period, a possible flat roof being added, and there was evidence of fires, long sealing the old Roman levels.  It also had the distinction of being the only building that definitely provided 3rd century occupation material, a series of pots dating from 180-220 A.D. found within a roughly circular hearth composed of boulders stolen from the same building.  The earlier Roman levels had been well sealed by a thick layer of ash, and indeed this was equally sealed by the ash from this hearth.  They comprised of a fragmented Samian 27 of Flavian date, piece of a Samian form 31 of Antonine date, a piece of form 37 rim 2nd century date, three pieces of white flanged mortaria, this seemed to have affinities with examples found at Richborough on the South coast and the Richborough examples dated to the 4th century, these have also been discovered at nearby Wroxeter and also at Caerwent.  Baillie Reynolds concluded that though these mortaria had a long life it was highly likely that they date later rather than earlier.
Finally, the annex to the south of the fort appeared to contain some timber period features.  The area excavated roughly in the middle of the annex enclosure Site F. was believed to be contemporary, and was rather odd.  Taking the form of an irregular hollow, cut two feet deep into the natural clay, the cut back facing the rampart was lined with many hearths, and indeed the whole space was filled with hearths, around 18.  Following a curious pattern of the times perpetuated by Augustus Pitt Rivers and R.E. Mortimer Wheeler, Baillie Reynolds called it a ‘native pit dwelling' believing the Britons lived in subterranean houses, such beliefs have long been confined to the archeological rubbish bin.  Occupation in the hollow was profuse, not only in the timber phase but also throughout, no less than six levels being detected, reaching a peak in the Antonine period, A.D 140, which is at variance with evidence from inside the fort.  Other timber phase features detected in the annex was the south road, partly overlaid by the later 2nd century road, and a timber bridge over the ditch at the south gate.  The fort bath-house was also likely established soon after the founding of the fort in the mid to late A.D 70’s