The fort wall, towers and gates

Shaw's excavation trench of 1957–58 can be traced across the north-western defences just to the east of the drystone wall. This exposed part of the fort wall; its outer face was badly robbed but the wall had a rubble and clay core encased by regular hammer-dressed blocks derived from thin-bedded coal-measure sandstone.Footnote ³⁵ Shaw thought this stone to be very local, and the precise source has been identified by the authors during wider fieldwork for the Miner-Farmer Landscapes Project. The Roman quarries lie less than 300 m due west of the fort, about 40 m higher up on the slopes of Great Heaplaw, where they survive as a series of broad terraces. Now largely submerged beneath boggy grassland, these quarries are very different in appearance from the more sharply defined and concentrated patterns of pits, shafts and spoil heaps that were created by much later quarrying, by coal mining and by prospection for lead ( fig. 9 ).

The excavation trench revealed that the main body of the wall had been about 1.67 m (5.5 ft) thick at the base, narrowing slightly above the second course.Footnote ³⁶ The earthwork survey and the geophysics suggest a broadly consistent level of preservation along this north-western side, but that there has been even more extensive robbing along the outer face on the south-east and the south-west. Around the eastern corner of the fort there seems to have been wholesale robbing of the wall. The northern half of the north-east wall, in contrast, is marked by a very pronounced bank which is wider than anywhere else around the perimeter. This may be no more than a product of inward collapse, although the earthwork and geophysical plots indicate the robbed outline of a narrow rectangular structure (7) set against the inner face of the wall; this, given its location at the lowest point of the interior of the fort, bears a strong similarity to the position of the latrines at Housesteads, and this may be the best explanation for the features encountered. A further curious aspect of this section of bank is the occurrence of five small mounds along the base of the outer scarp, which suggest a very particular process of demolition at intervals of 3 to 4 m. The spacing could imply the removal of buttresses, but these are not usual features of Roman fort walls, and might only be explained had this particular section been strengthened to prevent collapse. The internal clay rampart that was noted in Shaw's excavation can only be clearly seen on the surface within the southern corner of the fort wall, where it forms a broad slope measuring up to 5 m across and with a fall of between 0.5 and 1.2 m.

There are indications that the fort had four angle-towers, and two interval-towers in the long sides of the retentura. The southern corner marks the highest point on the fort wall (336.7 m OD) and the tower here appears to be the most complete. It is represented by an approximately circular mound of collapsed material, remarkable for the absence of robbing scars. The geophysical survey and the surviving earthworks suggest that a considerable portion of the lower structure may be concealed intact beneath the mound of debris. The western corner-tower is marked by a similar mass, although more material appears to have been removed and the profile has been disturbed by a later field bank, leaving a more angular profile to the rear. The north tower has been quite extensively robbed, but there are indications of good structural survival. The last corner-tower, to the east, is almost completely quarried away, with no part surviving above the height of the adjoining traces of the wall. Thomas Sopwith's plan of the fort ( fig. 3 )Footnote ³⁷ shows a small rectangular, double-celled structure set into the outer scarp at this corner, which is an elaboration of a simple rectangular feature shown on his earlier plan of about 1840.Footnote ³⁸ Bosanquet's suggestionFootnote ³⁹ that this is actually the remains of a corner-tower is untenable. All of the other towers clearly stood within rounded corners, and neither Sopwith himself, nor the authors of the works which contain his plans, mention any exposed fort stonework.Footnote ⁴⁰ In all likelihood, Sopwith was indicating a more recent structure which was distinct from the fort itself. Given the amount of fire-reddened stone exposed in the larger of the two robbing hollows at this point, this structure may have been a small limekiln or, taking into account the distance to the limestone outcrops, it may have been a corn-drying kiln relating to the post-medieval occupation of the fort.Footnote ⁴¹ The surviving remains of the northern corner-tower, and perhaps those of the southern tower also, seem to have presented a curved wall to the interior of the fort. There is thus a possibility that these towers were circular and no more than 8 m in diameter, a response to the awkward angles dictated by the overall plan.

The interval-tower (8) on the north-west wall of the fort is crossed by a field-wall, but is visible as a well-preserved, roughly rectangular mound, measuring 7 by 12 by 0.4 m high, located mid-way between the gate and the western corner-tower. The tower along the south-east wall (9) is similarly positioned, half-way between the gate and the southern corner. Here, although the earthwork evidence is less obvious, the magnetic survey shows the outline as a neat square of collapsed stone set within robbed foundations and a small compound precisely aligned with the perimeter wall.

There is clear evidence for four gates, two located centrally on the shorter sides and two in matching locations on the longer sides, placed approximately — but not exactly — one third of the way along each arm from the north and west corners. The gates, which would have been major sources of reusable stone, have suffered greatly from robbing, more so than any of the wall towers, with the exception of that at the eastern corner. There is very little evidence visible on the ground for the design of any of the gates, and the geophysical evidence is even less forthcoming. The inward curvature of the walls on either side of the south-east gate and to the north of the north-east gate gives the impression that the gates may have been recessed,Footnote ⁴² but this cannot be proved without excavation. The disposition of the gates indicates that the fort faced north-east, towards the Maiden Way. The scale of the opening in the ramparts suggests that the north-eastern gate was provided with a double portal; this may also have been true of those on the north-west and the south-east. The earthworks of the porta praetoria do, however, retain a distinct step which (if not part of the later reuse of the gate — see below) may represent a surviving threshold.

The earthwork remains of the south-west gate are the least distinct. The profiles of the south-west wall and the defences beyond (see below) demonstrate that this entrance was blocked during the Roman occupation of the fort. When sealed, the gate might well have been maintained as a kind of interval-tower, but the earthworks are uninformative and the geophysical data imply that the structures here are disturbed. This is perhaps owing to the immediate proximity of several post-medieval buildings in the retentura (see below), the construction of which may well have entailed plundering the gate towers for stone.

The outer defences

The most distinctive features of Whitley Castle are the complex, multiple defences, and the best-preserved section of these, to the north-west, naturally attracted the attention of the archaeologists in 1957–58. Shaw's excavation across these defences revealed a berm (produced as part of the levelling of the spur that also provided material for the rampart), which extended about 2.1 m (7 ft) from the foot of the fort wall.Footnote ⁴³ On the surface, the berm, varying between 1 and 1.5 m in width, can still be traced around the majority of the circuit, apart from a short section from the east corner to the south-east gate where it has been heavily disturbed by later activity.

Although there are still uncertainties about the sequence in the construction of the defences, Shaw's excavation appears to have revealed an entirely straightforward construction technique: the ditches were dug into the subsoil of the hillside, and the upcast was formed into banks along each down-slope side in order to enhance their depth. There was no evidence for ‘punic’ profiles (the angles of the sides being about equal), nor for the narrow ‘ankle-breakers’ or ‘cleaning slots’ occasionally found along the bases of the ditches of other forts.Footnote ⁴⁴

Although the earthworks of the defences are less well preserved between the north-west and north-east gates than they are on the rest of the north-west side, the geomagnetic evidence leaves no doubt that all four ditches continued round to the north-eastern side of the fort (see fig. 7 ).Footnote ⁴⁵ From this evidence and the appearance of the earthworks on the ground, it appears certain that the three outer ditches were filled-in to create the building platform for the bath-house (10) that was excavated in 1810. The outer scarp of this platform retains something of the curvature of the defences and may have been determined by the position of the outermost ditch and bank (IV).

Ploughing, in the form of broad ridge and furrow, has reduced the profile of the outer defences along the north-eastern side of the fort. To the south of the north-east gate, at least two ditches, visible as earthworks and as geophysical anomalies, certainly continued to the eastern corner. That the third did so also seems probable, but the evidence for the fourth is far less certain. A slight outward scarp, much degraded and spread by the plough, hints at the line of the fourth ditch, but the absence of geophysical evidence may imply either that it was never fully dug, or that it has been almost completely destroyed by later ploughing.

Beyond the inner ditch (I), the south-eastern side of the fort has two well-defined lines of defence throughout, and a third and outermost line along part of its length. Although later activities have obscured the details to the north and east of the south-east gate, it is clear that Ditches I and II continued around the eastern corner from the north-eastern side and that they have eroded into broad steps or terraces in the steep slope on this side of the knoll. There are faint indications in the earthworks and in the magnetic survey results to suggest that Ditch III may also have continued around this corner as far as the gate, but ploughing has impacted heavily in this area and the evidence is far from conclusive. The defences are better preserved to the south of the south-east gate where, with the rising height of the knoll, the terraces that represent Ditches I, II and III survive clearly.

The south-western side of the fort has the most impressive defences: here there are seven ditches, each with an outer bank, and a short stretch of what may have been an incomplete eighth ditch and bank (V). The innermost ditch (I) and its associated outer rampart are similar in character to their surviving form on the adjacent sides of the fort. This is also largely true of the second ditch (II) and its bank, although there is a noticeable narrowing of the inner defences on this south-western side, perhaps owing to the level ground here. A slight inward turn in the bank associated with Ditch II may be no more than an adjustment to this narrower plan, or perhaps it reflects a more radical alteration in which the bank between the second and third ditches was divided to increase the number of fortifications on this side. Outside this, the next ditch and bank (IIIa) are clearly part of the additional provision on this side; they separate from the main rampart III at the western angle and barely turn the southern corner of the fort. Farther out again, Ditch IIIb and its bank are short, parallel, and relatively conventional. They are strengthened, however, by an unusual bow-shaped arrangement (IIIc), the ends of which curve outwards before taking up the same alignment, parallel to the inner defences. The difference here is that the ditch — or the space between the banks — (IIIc) is significantly broader, especially in its southern half. Ditch IIId swings round the outside of this bow-shaped arrangement providing continuity with the third ditch along both the north-west and south-east arms of the fort. The only additional feature to note here in its general design is its apparent division into two narrower ditches, divided by a broad berm, in its southern half. Ditch IV is even more regular and seems to have extended at least to the southern corner of the fort. In the re-entrant angle outside the bank of Ditch IV, between the north-west and south-west sides, a short length of another ditch (V) survives, with traces of an associated outer bank.

The complex defences on the south-western side of the fort are not uniform in character. Although they are the least affected by soil movement, or by later cultivation, the earthworks here are unusually fragmented and their condition is extremely varied. There is no doubt that the road (6) from the south-west gate originally ran directly through the middle of the defences. This is obvious from the staggered alignments of the defences themselves, most notably of II, IIIa, b and c, which must originally have been drawn up to either side of a functioning roadway, and by the continuation of this road beyond the final rampart. At some point, however, a decision was made to seal the porta decumana and to close this road. Irregular sections of ditch and bank were driven across it in order to eliminate any weak point that would otherwise have been created. The more substantial infilling of the gap in the outermost bank (IV) contains some large stone blocks which seem to have been simply dumped as obstacles. To the north of the road the earthworks are relatively well preserved, although they are still significantly less substantial than the ditches and banks on the north-western side of the fort. To the south their dimensions are far more modest. On the ground, this disparity suggests the abandonment of an incomplete enterprise rather than subsequent slighting.

The geomagnetic survey detected a striking series of anomalies at the southern corner of the defences ( fig. 7 ) which hint at a reason why the defences at this southern angle have a flattened curve. The most prominent anomalies are two parallel strands of highly magnetic material aligned across the outer edge of the corner from east to west (at 11 on fig. 5 ). The northern strand bears a precise similarity to the terminal of the IIIa bank, and must represent the same feature. The southern strand is partly overlain by a narrow bank of a later period and is barely visible on the surface, but it can hardly be other than the spread remains of the fourth defensive bank. The difficulty with these anomalies, however, is that neither appears simply to terminate at the corner where the earthwork evidence and the topography suggest that they should. Both seem to change angle slightly before descending to the base of the knoll, turning sharply to the north-east, and re-ascending the slope in a diagonal fashion. It has been suggested that these anomalies reflect a short fifth side to the fort's outermost defences, and perhaps even that their peculiar course defined a bulge around a further corner-tower;Footnote ⁴⁶ but these theories cannot be sustained in the face of the local topography, which would place the notional tower at the foot of the knoll, some 10 m lower than the fort wall. A better explanation may be that a significant extramural building, lay just outside the southern corner of the fort and that this building was retained outside the defences, despite the strengthening and expansion of the latter.

The headquarters

At the very centre of the fort, the outline of the headquarters building is clear (12). A combination of collapsed rubble, extant wall lines and robbing hollows, visible as both earthworks and geophysical anomalies, indicates that this was a very substantial stone structure. The building's overall dimensions are approximately 26 by 22 m, well within the range of comparable excavated examples.Footnote ⁴⁷ A low rectangular platform outside the entrance suggests that it may have been approached by a short flight of steps — made necessary by the rising ground — or through a small porch. The internal courtyard was at least 15 m across, and the cross-hall beyond was approximately 6 m wide. Despite being bisected by a standing drystone wall, and subjected to considerable stone-robbing, the remains nonetheless indicate a substantial hall; the walls were about 1 m thick, potentially capable of supporting an upper storey. A small but prominent platform within the north-western corner of the cross-hall may be nothing more than a by-product of wall-robbing, but its position equates to that of the dais.

A series of earthworks, 19 m long, behind the cross-hall, appears to represent four of the usual five rooms of the rear range of a headquarters building; the position of the fifth room, to the north, has been obscured by later robbing. These rooms — largely preserved as upstanding wall lines with sunken or excavated interiors — each measure about 5 m from front to back, and between 3 and 5 m across. The central room, the aedes, is the largest of these. No strongroom can be identified, although there is a curious annexe, measuring 6 by 4 m, attached to the southern end of the rear range; this is connected by a hollow, perhaps a sunken passageway (with steps), containing the identifiable remains of a collapsed barrel vault (13). An alternative explanation of this feature might be that it was a stokehole or furnace room, possibly indicating, as elsewhere, the addition of a hypocaust to the headquarters in the later second or third century. (This, however, seems unlikely in the absence of any significant anomalies indicating intense burning in the geophysical survey.) The evidence is not entirely clear, but this vault may have lain at the southern end of a longer extension superimposed over part of the via quintana at the rear of the headquarters. This putative extension appears to have continued beyond the western corner of the building for approximately 5 m and perhaps included a small annexe attached to the western end of the north wall (14). It is possible that this may have been joined to an irregular series of later buildings at the rear of the granary (see below), but the alignments suggest that any such extension to the headquarters was a later development. This extension was substantially built, to judge from the size of the wall mounds, the size of the robbing trenches and the strength of the geophysical responses. It would have completely blocked the passageway between the headquarters and the granary.

The commanding officer's house

The remains of a courtyard building, apparently the commanding officer's house, lie immediately to the south of the headquarters. The footprint of this building has been quite severely disrupted by the construction and demolition of a post-medieval building and the area doubtless suffered considerable erosion during that later period of occupation (see ‘Castle Nook’ below). As a result, the earlier earthworks are rather slight, but they are still sufficient to identify portions of two building ranges that meet at the western corner of a courtyard (15). The north-western side is approximately 5 m wide and no more than 35 m long, and appears from the geophysical (magnetic) evidence to have been subdivided into a number of regular compartments. The length of the south-west side will depend on the width of the intervallum road here, but it is unlikely to have exceeded 30 m. This side retains evidence of some additional complexity in the form of an inset opening, or a pair of protruding rooms, facing onto the via quintana — possibly a significant addition, not unlike the inclusion of bath-houses at the commander's houses at Mumrills, on the Antonine Wall, and at South Shields.Footnote ⁴⁸ However, although there is a concentration of magnetic material in this area (mostly concentrated on the wall lines of the later building), there is no conclusive evidence for a stokehole or furnace. The remaining two sides of the house, beside the rampart and the via principalis, are no longer visible, and the foundations, if they survive at all, are indistinguishable in the geophysical surveys owing to the extent of later disturbance in this area. It is worth noting, however, that the projected position of the north-eastern side, close to the corner of the drystone wall, overlies an intermittent spring. This now flows eastwards across the surface of the lower terraces at the front of the fort, but might have provided the water supply or part of the drainage system of this building, and that of the postulated latrine (7).

The granary

Shaw's excavations revealed evidence for a granary (16) between the principia and the north-west gate. The short section of the lower courses of this building that was exposed then revealed no buttresses, but the quality of the foundations, together with the presence of internal recesses and sleeper walls for a suspended floor, left Shaw in no doubt about its function.Footnote ⁴⁹ The north-western wall of the granary, exposed in Shaw's trench, was located about 12 m (38 ft) inside the wall of the fort. Given the available space between this wall and the north side of the headquarters (approximately 22 m), and allowing for a narrow intervening road, more than one granary in this location would seem unlikely, although a double-width building of the sort seen at Hardknott (Cumbria)Footnote ⁵⁰ could have been accommodated. The ground in the area of the granary has been significantly altered, perhaps by the robbing of good quality building stone; additionally, as Shaw observed,Footnote ⁵¹ the structural remains are buried by an overburden of mixed soil and rubble reaching depths of at least 0.7 m (2 ft 6 ins). However, although there is scant evidence of this building visible on the surface, the geomagnetic results are more informative ( fig. 7 ). These indicate the outline of a single building, some 10 m in width, extending back from the side of the road on an alignment that fits tolerably well with the foundations revealed in 1957.Footnote ⁵² Although it is far less pronounced, there is some evidence for a third, equidistant and parallel wall closer to the headquarters, which could represent either a separate building phase, or the double-width granary unit postulated above. The granaries found in Roman Britain, with the exception of those in legionary fortresses, rarely exceed 30 m in length. A sharp depression (and a strong anomaly) broadly in line with the rear of the headquarters building suggests that the granary here was no exception.

FIG. 7.

Geomagnetic survey of the fort and surroundings. (Archaeological Services, University of Durham)

Buildings to the rear of the granary

The area immediately to the rear of the headquarters building and of the granary, although rich in relatively minor earthworks and dense geophysical responses, is perhaps one of the most difficult to interpret. This area might be expected to contain either a row of stores and workshops facing onto the via quintana or, if the granary was quite short, a small courtyard building of the kind identified elsewhere as workshops or the garrison hospital. Depending on the precise length of the granary, either possibility could be represented by the fragmentary shallow earthworks — these consist of a variety of narrow channels and slight scarps which suggest individual rooms, perhaps set within a single building measuring at least 10 m square (17). However, these are unlikely to represent the remains of a single building phase, and the pattern here may incorporate elements of an earlier barrack block (see 21 below) and may reflect changes associated with the construction of the apparent annexes attached to the headquarters building, mentioned above.

Buildings in the retentura

The rear portion of the fort, the retentura, is the most elevated and level part of the interior. The surface remains consist of narrow platforms of collapsed buildings and intervening hollows. On the strength of both the earthwork and geophysical evidence this area appears to have contained up to six rectangular blocks (18–23), three on either side of the via decumana, all orientated broadly parallel with the south-western rampart. The most easterly block (18) on the south side of the via decumana lies directly along the via quintana. It measures approximately 34 m in length and some 8 m across, and retains a low mound of wall material forming a longitudinal partition. The southern end of this block lies beyond the present field-wall and is less clearly defined, although it appears to terminate about 2.5 m short of the interval-tower (9). A small cell-like feature overlying the via decumana at the northern end of this block is presumably later. The outline of the middle block (19) has been more thoroughly disrupted by stone-robbing and has mounds of associated upcast, but the general area of raised ground indicates a building of similar width and length, offset by approximately 3 m toward the rampart. These blocks were presumably barracks, not only because they would be expected in this quarter of the fort, but also because the magnetic survey shows that both of them retain details of internal features, with some indications of contubernia, and of the cross-walls that divided off the rear rooms.Footnote ⁵³ There is even a slight suggestion that larger quarters for the centurion lay in the usual position towards the intervallum road, although the presence here of the later drystone wall makes this far from certain. The block beside the south-western rampart (20) is the least visible of the three, obscured by the robbed foundations of a much later, post-medieval building (see Feature E below) and flattened by associated yards and other features; these later intrusions also affect the middle block in this range, albeit to a lesser extent.

The identification of individual buildings within the range to the north and west of the via decumana is rather more problematic. The pattern of scarps and mounds provides an overall impression of three similar parallel buildings, each approximately 35–40 m in length and 8–12 m wide, and this is echoed in the geophysical results from 2003 and 2009. However, the details are far from clear, perhaps owing to rebuilding within the Roman period, but more certainly as a result of later, post-Roman activity. The 2009 magnetic survey shows these three blocks were orientated similarly, but not identically, with those to the south of the via decumana. The central building on this side (22) is the most clearly defined in the earthworks. It appears to be about 12 m wide and 40 m long, and both superficial indications and geophysical anomalies reflect the division of individual contubernia along much of its length, and a more complex division of rooms at the northern end. To judge from its overall width and complexity, however, this block may consist of partly overlapping buildings of more than one phase.

The two other blocks in this quarter of the fort are less distinct. The easternmost block (21) emerges quite clearly from the 2009 magnetic survey, although most of the earthworks in the central and southern part of this range have been levelled for a yard or cultivated plot, probably during the post-medieval occupation within the fort, leaving only the raised edges around the angle of the viae quintana and decumana to indicate its maximum extent. This building appears to have been in the order of 9 m wide and 35 m long. The earthworks are best preserved at the northern end, close to the rampart, where the end room is clearly defined by narrow mounds and shallow scarps, separated from the buildings to the rear of the granary by the terraced path of the via quintana threading its way towards the interval-tower (8). As in the group to the south, the western building (23) is the least visible of the three in this north-western quarter. It is defined more by the bulk of the platform and the limitations of the available space than by evidence for individual wall-lines. The dominant features in this area relate to the existence and demolition of a further post-medieval range (see Feature F below), which overlay the central and southern parts of the building. However, the evidence of the earthworks implies that this building originally stood, or was rebuilt, closer to the rampart than suggested by the geophysical results, and was slightly foreshortened to account for the base of the corner-tower.

Buildings in the praetentura

The profile of the knoll falls away across the front part of the fort in an overall drop of about 5 m between the via principalis and the north-eastern rampart. This fall was accommodated by two terraces, bisected by the via praetoria which climbed steeply from the west gate, effectively creating four separate platforms. Earthwork and geophysical evidence points to the existence of a single long building on each platform (24–27), although the geophysical results suggest that the northern pair of buildings were different in plan and character from the southern ones.

The two upper platforms each measure about 40 m in length and 15 m in width. The earthwork evidence — again comprising concentrations of collapsed material, robbing trenches and some low, turf-covered foundations — indicates a pair of buildings (24 and 25), each measuring approximately 11 by 35 m, a picture which is echoed in the geomagnetic results. Although many of the superficial earthworks reflect episodes of robbing, there are also indications (particularly across the northern part of 25) of internal subdivisions; the geophysical evidence suggests that these contubernia were slightly broader, perhaps 11 m or 12 m in width, compared with the buildings in the rear of the fort where the equivalent measurement is about 10 m.

The two lower platforms have been more thoroughly disturbed — they were evidently easier for the stone-robbers to reach — and several tracks have eroded the eastern corner of the fort. The northern platform (26) retains some indications, both superficial and geophysical, of a building of similar size to that on the terrace above, although it may have been slightly shortened to allow access around the base of the northern corner-tower. Disturbance within the footprint of this building is more intense towards the northern end, perhaps reflecting the greater quality or quantity of stone which could be reclaimed, and therefore something of the building's hierarchical division. Significant robbing scars between this building and the fort wall may point to the removal of material from another substantial structure, perhaps the postulated latrine (7). Details of the building on the southern platform (27) are sparse, but the rather amorphous pattern of internal rooms visible in the geomagnetic results clearly shows that such a building stood in this location.

The bath-house

The published accounts of the discovery of the bath-house and the plans of the fort made shortly afterwardsFootnote ⁵⁴ leave no doubt that this building (10) stood just outside and to the north of the north-western gate, where the earthwork evidence indicates that a platform was created by levelling part of the defences. Hodgson wrote: ‘The remains of a bath, with a very perfect hypocaustum attached to it, were opened a short time before my visit here in 1810; and the pillars then exposed were black with soot, and, as usual, supporting thin freestone flags, covered, layer after layer with coarse mortar, made of lime and lead mine spar, and lastly with a cement formed of fine gravel and quick lime, pounded limestone, earthenware, and b[r]ick. The mortar, too, in the wall, was much mixed with small mineral coal; and, in turning up the soil here, I twice met with calcareous and fluor spar, mixed with portions of sulphuret of lead. This bath is situated on the north side of the station, and adjoining the north-east corner, on the outside of the outermost earthen rampart, and has a copious and perennial spring nearly close to it.’Footnote ⁵⁵

Hodgson's accounts, although informative, do not include details of the building's plan, but a degree of reconstruction is possible from the earthworks — the hollows which represent the excavated rooms, and the trenches dug to chase the perimeter walls — as well as from the geophysical survey. The bath-house appears to have been broadly rectangular in plan: it measured about 15 m in width and was at least 30 m long on the axis parallel to the fort wall. Banks surviving to the north-west and around its northern corner may define a compound around it. The geophysical evidence suggests that there was a range of rooms, standing roughly parallel to the wall of the fort. The angled robber-trench at the southern end may mark the position of the changing-room or of a frigidarium; the warmer rooms probably lay to the north. The hypocaust that Hodgson mentions is likely to have been revealed in the broad hollow that has been driven into the centre of the building platform. A short stretch of curving wall, visible at the northern end of the earthworks, may be part of the apse of a caldarium, or possibly part of a separate laconicum.

When the bath-house stood exposed in 1810, Hodgson noted that ‘many pillars of the hypocaust were standing, covered by large thin slabs of freestone, and a strong calcareous cement’. Some twenty years later, however, Sopwith wrote that many of these pillars had been removed for building materials.Footnote ⁵⁶ The hypocaust is no longer visible, although a single stone pillar, narrow and roughly finished, now rests on its side in the base of the large excavation hollow. No dating evidence was recorded from the nineteenth-century exploration, but the fact that the mortar contained ‘lead mine spar’ may be seen as a further indication, if such were needed, that this building was not put up in the primary phase of the site (when such a material might not have been available).

The source of the water for the bath-house has not been positively established. The obvious source is the Lort Burn, and Frere and St Joseph suggested that the water may have been brought to the baths by means of an aqueduct that they identified across the intervening hillside;Footnote ⁵⁷ both the earthworks and contours of the ground make it clear, however, that this is a misinterpretation of a post-medieval bank and ditch (Feature D). A possible explanation is that a modest supply was diverted from the much smaller stream course that now runs within a natural crease in the hillside, parallel to and just below the north-west defences. The water may have been taken down the outer ditch of the defences, since it is clear that these were considered to be largely redundant by the time that the bath-house was built. If wooden launders were used — as seems to have been the case for part of the course of the aqueduct that served the fort at Great ChestersFootnote ⁵⁸ — it would have been easy to tap the much more plentiful supplies of water in the Lort Burn, but no trace of this would have been left in the landscape.

The midden

Just to the west of the bath-house, a substantial Roman rubbish dump was discovered and exploited by the proprietor of Castle Nook in 1825, and was subsequently explored by Hodgson.Footnote ⁵⁹ It contained a large quantity of leather shoes (of which Hodgson provided considerable details of their designs), a bronze pin and spoon, fragments of red tile and of figured samian and other pottery (one with the stamp SAMO or SAMIO), lead pipe, glass bottles, a comb, querns, objects of jet, and a ‘huge battle-axe’. Hodgson added that ‘much hazel-wood was in the heap, and stakes that had been pointed with an ax, still standing upright. Also moss, of the species called hypnum squarrosum, and the straw of some species of grain, both used as the bedding of horses’.

The midden had been exposed by the excavation of a drain beside the north-western side of the fort, and was marked by the letter ‘A’ on Sopwith's 1853 plan of the fort ( fig. 3 ). On the ground, this point broadly coincides with a large and very distinct hollow (28), with a flattened base and a cart exit to the north-east, dug into the base of the slope below the bath-house. Other hollows in the vicinity may have had a similar purpose — to retrieve organic soils which provided a useful dressing for adjoining land. These hollows include two excavations into the outer face of the platform of the bath-house, both of which reveal dark deposits which suggest that organic material and rubbish may have been a significant component of the bulk used to level the defences.

Mr Henderson's drain, the cutting of which revealed the midden, was probably intended to improve the natural water-course beside the defences at the foot of the spur. This and other attempts to drain the ground to the north of the fort may have altered the saturated conditions that enabled the remarkable assemblage of shoes and other organic artefacts to survive. However, reports of further leather fragments unearthed when the drain was recut in quite recent times suggest that the waterlogged conditions may persist.Footnote ⁶⁰

Temples and cemeteries

The spring mentioned by Hodgson and illustrated on Sopwith's plansFootnote ⁶¹ (probably marked by the letter ‘B’ on fig. 3 ) lies close to the corner of the field above the large modern byre, and now supplies the farm through a buried metal pipe which is clearly visible on the magnetic survey plan ( fig. 7 ). In this area around the spring — now a large boggy hollow crossed by the drystone wall beyond the northern corner of the fort — two major religious inscriptions were discovered. The Hercules pedestalFootnote ⁶² is reported to have been found before 1810 in a swampy spot ‘a little to the south-east [sic; presumably north-east was meant] of the baths and the north-east corner of the station’, fitted to a socket stone and accompanied by the fragmentary remains of a statue of Hercules,Footnote ⁶³ circumstances which suggest that it was discovered in, or very close to, its original location.

The Apollo altar, which depicts the god in four different guises (Citharoedus, Sol, Mithras and Maponus),Footnote ⁶⁴ came from this same area when it was drained in 1837.Footnote ⁶⁵ Descriptions of its recovery clearly show that it was discovered in situ — the altar was dislodged but otherwise complete, and the pedestal on which it stood was still mounted on four pillars (each one capped by a single coin) in a paved area, over which, it seemed, water once flowed.Footnote ⁶⁶ Piecing together the information from the accounts of the altar's discovery, it would seem likely that it originated within a shrine sited on a spring,Footnote ⁶⁷ 100 m or so from the corner of the fort, probably within the modern enclosure to the west of the byre at Castle Nook, close to a point where the channel widens out near the drystone wall and there is a slight change of angle in the north-west-facing slope. However, it has been impossible to prove this assumption, either by earthwork analysis or geophysical survey, because of modern earth-moving and the quantity of relatively recent farming detritus spread across this field. Nothing of archaeological interest was found by examining the various hollows that had been dug to unblock or repair the modern pipes in this area. Another strong possibility is that the shrine lay immediately to the west of the drystone wall, about 80 m north of the corner of the fort, where the position of a building is indicated by a distinctive terraced platform (29), measuring some 17 by 8 m, beside (and partly truncated by) the stream. The geophysical results are, however, quite ambivalent in this area owing to the proximity of the drain and the buried pipe from the spring. This platform (29) is approached by a hollow-way (Tr3) from the direction of the north-west gate, which is one of the more substantially-worn routes across this area of the hillside.

According to Hodgson this area, close to the Maiden Way, was known as the ‘The Burial Ground’, and he was told that the Rector, Mr Kirby, had ‘curious stones taken from it in his possession’.Footnote ⁶⁸ There is no other record of Mr Kirby's collection, but it is possible that some of these pieces were tombstones, and that there was a cemetery here, in the traditional position beside the Roman road.

The extramural settlement to the north of the fort

Recalling his notes from 1817, Hodgson later wrote that ‘no traces of suburbs excepting the bath have been found, nor indeed are apparent’.Footnote ⁶⁹ This view remained unchallenged for more than a century. Shaw's site plan shows a vague impression of scarps in the area between the fort and the Lort Burn, but these were not mentioned in the report.Footnote ⁷⁰ In 1981 Farrar recorded the more pronounced elements of the earthworks to the north of the fort on his incomplete plan, but the first published suggestion that this area might contain a civilian settlement or vicus emerged a few years later, arising from Frere and St Joseph's study of aerial photographs.Footnote ⁷¹ It was their contention, supported by the discovery of fragments of pottery and building stone exposed by land drains across this area in the early 1980s, that led to the scheduled area being expanded northwards in 1985.Footnote ⁷² Further evidence of buildings was provided by the 2006 geophysical survey, when the foundations of walls were located on two of the larger terraces ( fig. 4 , 58), beyond the western corner of the fort, that had previously been identified by Farrar.Footnote ⁷³

The earthwork survey indicates that the area between the Lort Burn and the north-western defences of the fort contains a pattern of small terraced enclosures, linked by a complex network of minor tracks and lanes which clearly extended around the western side of the fort (see below). On the basis of the earthwork evidence, the remains on this side of the fort can be divided into several distinct areas or zones. The northern zone consists of a series of terraces or lynchets, defined by a number of broad scarps (34–38). These are overlain by an orderly arrangement of three parallel long rectangular plots (30, 31 and 32), each about 20 m in width, which extend for 60 to 70 m down the slope at right angles to the road (5) that emerges from the north-western gate. There are some indications of a fourth rectangular plot (33), adjacent to the Lort Burn, although a combination of severe vehicle erosion and an overburden of rubble spread to improve farm access has largely obscured this area. For similar reasons it is now impossible to determine how far this pattern may have continued down the slope towards Castle Nook farm. The most probable explanation is that — rather than being part of a settlement, as such — the terraces were fields which developed earlier in the sequence of the Roman occupation, or even beforehand (see below), and which were subsequently modified to form smaller compartments.

The most convincing settlement remains in this northern area are found on the north side of the small stream that runs immediately below the bath-house — here there are adjacent terraced platforms (the terraces perhaps echoing the earlier lynchets) that clearly seem to indicate the positions of buildings (39, 40). The terraces on which they stood appear to have extended across the stream towards the fort's outer defence works, where one at least (39) is partly overlain by the spread of material around the base of the platform of the bath-house. The third terraced platform in this row (29) is the suggested position for the origin of the Apollo altar. Curiously, the resistivity survey provides little supporting evidence for structures within these plots,Footnote ⁷⁴ and the magnetic survey has produced none whatsoever. There may, of course, have been timber buildings here, rather than stone ones (although more anomalies, derived from occupation debris, might still have been expected in the geophysics if that were the case), or the negative readings may be the result of the amount of modern disturbance.Footnote ⁷⁵

The pattern of enclosures on the slope above and to the west of the road emerging from the north-western gate of the fort differs in several respects from the pattern below and to the east, most notably in that there is a less orderly, perhaps more ‘organic’ development of hollow-ways and minor tracks. The most prominent feature here is a broad hollow-way (Tr4), which climbs the hillside for 25 m from a well-defined junction with the road before dividing into two more sinuous routes (Tr5 and Tr6). A ditch and the footings of a later field wall (D) cut across the hollow-way, and this area is heavily disturbed by modern tracks focused on the field gate to the north. A substantial terrace running parallel to the road has traces (again invisible in geophysical data) of an individual building platform set beside a small sunken yard within a wider compound (41); below this there is a further, lesser scarp (42) which might, with other parallel scarps (43–46), belong to the pattern of earlier lynchets (34–38) mentioned above. As was the case to the north-east of Road 5, the better-preserved building platforms here lie close to the foot of the ramparts. Two of these (47 and 48) contain clear evidence of buildings, visible on the ground (if not through the geophysical survey results), giving something of the appearance of a row of quite regular tofts, with structures set beside the hollow-way and small enclosures extending back towards the fort. Whether or not these are contemporary with the occupation of the fort is uncertain, although each plot either abuts or overlies the bank of the outermost rampart, and therefore all must post-date that phase in the fort's development. The outline of a further building (49) measuring some 9 by 7 m, and accompanied by a more irregular sunken yard, can be demonstrated from the earthworks, and from magnetic readings, within the angle defined by the fork in the hollow-ways Tr5 and Tr6, partly overlain by the drystone wall that extends from the fort towards the curving intake boundary.

A series of narrow banks (50–57), parallel to one another and similarly aligned to those overlapping the defences, has been recorded at intervals of between 12 and 36 m up to and beyond the western corner of the fort. These seem to be a continuation of a regular arrangement of field boundaries (see below). On the higher slopes, either respected by or superimposed over these plots (the distinction is unclear), are the remains of a striking building complex (58) overlooking the fort from this side. This is most clearly visible in the geophysical surveys of 2006Footnote ⁷⁶ as well as those from 2009 ( fig. 7 ), although some elements are also visible as well-preserved earthworks. Taken as a whole, the evidence implies a standard of construction which is notably absent from the rest of the settlement remains on this side of the fort. Traces of buildings, boundaries, and pits, are neatly contained within a rectangular area measuring about 70 m in length and between 40 and 60 m in width. The pattern of buried remains seen in the geophysical survey results is divided by the branch of the well-marked hollow-way (Tr6) which ascends the slope from the fork to the north-east. The resulting triangular terrace to the north is the setting for a building, about 7 m square, with smaller detached elements to north and south. Another building, similarly aligned and measuring about 12 by 10 m, which seems to have contained two principal rooms, lies to the south-east of the hollow-way, within a complex of further wall lines, some visible as slight banks defining a compound stretching toward the fort, perhaps 30 m long and 20 m broad. It is conceivable that such an elaborate arrangement could be interpreted as a guest house (mansio) or as the house of an important official, complete with stables and with its own baths.

Further settlement activity has been recorded on the north-east side of the fort, to the north of the road leading to the porta praetoria. The superficial evidence in this area is entirely dominated by medieval ridge and furrow ploughing and modern drainage patterns, yet beneath the surface, the geophysical survey (magnetic) has revealed a rectilinear complex (59) measuring some 35 by 50 m set within the angle between the ramparts and the road. The details are uncertain owing to the extent of later ploughing, but the geophysical data suggest the presence of at least one small building, and possibly others, set within a sub-divided compound of some kind.Footnote ⁷⁷ Leading due north from the porta praetoria, a gently curving level terrace can be traced for approximately 35 m before it is lost amidst the later ridge and furrow. Although this terrace was undoubtedly utilised within a post-Roman settlement set around this gate (J, below), the geophysical evidence suggests that it began as a minor road leading into the compound 59 — an indication, perhaps, that this complex, like that to the west (58), served some official function. Further enclosures have been postulated to the east, on the basis of the magnetic survey, clustered along the far side of the Maiden Way.Footnote ⁷⁸ These, however, are not convincing as being of Roman date since their principal orientations relate to the ridge and furrow and to later drainage, rather than to the Roman road.

Extramural settlement to the south and west

There is clear evidence for a pattern of radiating fields and a well-defined track (both described in greater detail below) extending around the western corner of the fort, but no obvious signs of the continuation of the northern settlement into this area. This may in part be owing to the bog outside the western angle of the fort. Several field banks run into this area, but leave no trace on the surface and disappear from the geophysical results, so the possibility that other features are similarly masked cannot be ignored. To the south, however, on the narrow ridge that connects the lower slopes of Whitlow Common with the knoll on which the fort is located, the geomagnetic survey has revealed the characteristic signature of further Roman settlement (60) clustered along the continuation of the via decumana outside the defences.

The 2006 geophysical survey provided hints of settlement activity in this area,Footnote ⁷⁹ as did the earthwork survey during which two small, plot-like divisions along the north side of the road were identified, along with fragments of colour-coated pottery and glass vessels in mole hills on either side of the road. However, the 2009 geophysical survey provided the first unequivocal evidence for a vicus along the line of the via decumana. Footnote ⁸⁰ The magnetic survey ( fig. 7 ) shows a very distinct pattern of linear anomalies running parallel and perpendicular to the buried ditches which flank the agger of the road (6). This pattern is only reflected by two shallow scarps on the surface, and has no relation to the pattern of minor cultivation earthworks in this area. The features, buried ditches and wall foundations, indicate a narrow arrangement of tightly packed and solidly-built enclosures, some containing traces of small buildings, set gable-end onto the road in the manner common to many military vici recorded at forts on the Wall and elsewhere.Footnote ⁸¹ The settlement appears to stretch westwards beyond the defences for perhaps 100 m, to the point where Road 6 divides into various tracks on the slope above the fort. The width of the settlement is unclear, but at present it can be traced for about 20 m on either side of the road. However, given the masking effects of the bog on the north side of the saddle and similar conditions to the south, the area covered by buildings may be more extensive than has so far been revealed.

The acute angle between the easternmost plot seen as a geophysical anomaly on the south side of the via decumana and the line of the fourth rampart bank (IV), suggests that the expansion of the south-western defences overlay a portion of this settlement; and the subsequent closure of the road through the defences would tend to support the conclusion that these alterations ultimately led to the demise of the settlement on this side of the fort. A further element which appears to be connected with this area of the vicus has been identified just outside the southern corner of the fort's defences (61). The earthwork evidence indicates a simple platform, perhaps bounded by slight banks, but the geophysical evidence (magnetic) depicts a very pronounced rectangular enclosure apparently protruding from beneath the earthworks, with internal anomalies that suggest the remains of a substantial structure; one part of this, adjacent to a projecting annexe on the south-western side, has a particularly strong magnetic signature. This appears to have been a substantial building, swept away by the expansion of the south-western defences. Its form and function are unknown, but if it was a bath-house (as would be plausible here), it is conceivable that this building was ultimately replaced by the baths (10) that overlie the defences at the northern corner of the fort.

The geophysical evidence (magnetic) gives the impression of fragmented ditched plots (62), aligned with the Maiden Way to the south-east of the fort, east of Holymire barn.Footnote ⁸² However, the field evidence in this area suggests that these may be no more than by-products of the intersection of the road with the medieval cultivation pattern and with the upcast from a seasonal drainage channel. In general, the southern side of the fort exhibits a conspicuous absence of settlement evidence. This is somewhat puzzling, given the opportunities offered by the expanse of comparatively sheltered, well-watered and level ground to the west of the road to the south-east gate. The area has been extensively ploughed leaving bold patterns of ridge and furrow on the surface, and it seems unlikely that earlier earthworks would have survived this process. Yet the geophysical surveys (see fig. 7 ) indicate that the pattern of later cultivation conceals a single significant feature — an embanked and ditched square, some 12 m in width, set below the southern corner of the ramparts (63). A suggestion, to which we shall return in the discussion, is that this area (i.e. the wider area to the south-east of the fort) may have been reserved for specific military activities.

The principal Roman roads

The Maiden Way ( figs 2 and 4, 1) runs across the slope below and to the east of the fort some 80 to 110 m from its wall, where it appears as a broad but slight terrace, which is rather irregular owing to later disturbance. In the area surveyed, it is overlain by a pattern of broad ridge and furrow ploughing, but this is, in turn, fragmented by the later development of a narrow and more sinuous track, along the line of the earlier road, which is shown on the first edition of the Ordnance Survey (1865) 6-inch scale map between Castle Nook and the farms at Whitlow, to the south of Holymire barn.

The front gate of the fort, the porta praetoria, faces north-eastwards directly towards the Maiden Way, and the road that issued from this gate is clearly visible in the magnetic survey ( fig. 7 ). A slight hollowing (3) beneath the ridge and furrow reflects the same alignment, although the junction between the two roads is almost completely obscured by a modern field boundary at this point. There is clear evidence that the fort was also connected to the Maiden Way by a well-defined spur road (4) which ran broadly parallel to the Roman main road and which may have carried the majority of the traffic to and from the south-east gate of the fort. This route separated from the Maiden Way some 300 m to the south of the fort, at a point now lost amidst the activity associated with the hamlet at Whitlow. Within the survey area, the earthworks of the road emerge at the eastern end of the barn at Holymire and extend as a narrow hollow-way across the adjacent level ground (where it is cut by two seasonal streams), before rising sharply to the south-east gate. On the far side of the fort, after descending the slope beyond the north-west gate, the course of the road lies along the contour on a broad terrace (5) leading towards the Lort Burn. There is a slight suggestion that the route turned fractionally to the west before arriving at the burn, but no clear evidence could be found of a crossing point, although one might be inferred where the old Alston to Brampton packhorse road, visible as a terraced hollow-way, traverses the slopes on either side. To the north of the burn, the carriageway of the former packhorse road is well preserved, and it is possible that this road may reuse the line of a Roman spur road similar to the one to the south.

The fourth road from the fort, mentioned above (6), extends the line of the via decumana beyond the south-west gate. In passing through the defences, this road ran at right-angles to the south-western wall, but beyond the defences it swung slightly to the south where the remains of a straight and well-defined agger, some 8 m in width and 0.5 m high, continue across the centre of the saddle between the knoll and the foot of Whitley Common. The magnetic survey indicates flanking ditches either side of the agger which can be traced as far as the later intake wall, and perhaps slightly beyond. Further from the fort, the road appears to divide into a variety of minor tracks leading up and across the slope. The most pronounced of these tracks that can be ascribed to the Roman period is an embanked and deeply-incised hollow-way (Tr1) which runs up the sharp gradient outside the intake wall for some 20 m before turning north and extending towards the settlement areas beyond the north-western defences.

Two strands of evidence indicate that the majority of the tracks surrounding the fort, notably Tr(s)1, 3–8, 10–13, are directly associated with the fort, the vicus and related enclosures beyond the northern and western sides of the defences. Firstly, it is apparent that these routes frame the enclosures and provide lines of communication between them. Secondly, both ends of this interconnecting system lead into roads that serve the gates of the fort. The relationship between Tr1 and the approach to the south-western gate is particularly significant, since the gate became redundant within the lifetime of the fort providing a terminus ante quem for this part of the network. Given the absence of evidence for any farmsteads or similar structures on this side of the fort there is no plausible context for the development of this network of tracks in a later period. All the later settlement identified is located either within the fort itself or to its south and east, and where separate arrangements for access and communication are readily visible.

Field-systems

Some of the remains of the field-systems that may be associated with the fort and vicus have already been alluded to. They can be divided broadly into five groups, although certain areas overlapped and perhaps merged over time. The first group (mentioned above) comprises the series of low, irregular terraces (34–38), at intervals of about 15 m, which seem to underlie the linear enclosures (30–33) running down the slope in the area between the fort and the Lort Burn. The second group consists of a pattern of similar boundaries, although larger and more sinuous in form, to the west of Road 5 and to the north of Tr4/Tr6. The four scarps here (42–45) contribute to the definition of the putative building platforms 41 and 49, but like those on the gentler slope to the east they may have initially developed through a process of stone clearance and cultivation.

Moving anti-clockwise around the fort the third group of fields is located near the western corner and continues around the south-western flank. The boundaries here comprise a series of narrow, stony banks and scarps (50–57) running down the south-east-facing slope. They can be traced, in varying states of preservation, on both side of track Tr1 extending down into the boggy ground bordering the western defences. Enough survive to indicate that this entire slope was probably divided into a series of strip-like fields radiating from the fort. The form and orientation is similar to the banks noted beside the building platforms 47 and 48, and this part of the settlement may well relate directly to the development of these fields. In this respect it is worth noting that none of the banks in this pattern encroaches on the tracks that serve this area.

The fourth group comprises lynchets of a sinuous form to the south-west of the fort, extending in fragmentary form beyond the survey area into the boggy ground on the higher slopes beyond. The pattern here consists of two main terraces, one above and one below the line of Tr1 and Tr13. Despite the imposition of several later tracks and hollows beside the intake wall, it remains apparent that the terraces to the south were divided into at least four, or possibly five, smaller plots flanking Tr13. It is also evident that this pattern evolved through time, since one small plot (64) lies awkwardly across Tr13, and must have been added or extended after this route had fallen out of use. In terms of proximity and ease of access along the connecting tracks, there seems little doubt that these plots or paddocks were closely related to the fort and probably to the settlement area outside the porta decumana. Changes in the fortunes of this settlement, linked to the expansion of the ramparts and the closure of the gate, may be reflected in the alterations to the plots themselves.

The fifth and last group of fields is located at the extreme south-west of the survey area (65), and is best preserved to the east of the intake wall. Some of the very low earthworks here are part of a much more extensive complex of settlement and fields associated with the medieval and post-medieval hamlet of Whitlow to the south, as well as with a small, later farmstead close to the intake wall to the north (M). Within this group, however, several scarps align with a sinuous, forking track (Tr2) which appears to lead toward Tr1 and the south-western gate of the fort, perhaps signifying an origin in the Roman period.

The earlier Castle Nook

The third building in the interior of the fort (G) stood quite apart and on a completely different alignment from the other two ( figs 5 and 8). The wholesale demolition and robbing of this building has left a well-defined rectangular hollowed terrace, measuring approximately 25 m long, 10 m wide and up to 0.6 m deep, bisected diagonally by a later drystone wall. A narrow sunken yard flanks the south-western side, and contains the impression of a small square structure (H), which may have abutted the southern corner of the main building. A second smaller building (I), located on and aligned with the track along the via principalis to the north-east, may also have been part of this complex, the greater part of which overlies the presumed location of the commanding officer's house (15). The adjacent section of the former rampart, south of the main building and yard, appears to have been levelled and perhaps subdivided into a series of small pens set against the line of the fort wall.

To judge from its overall dimensions, the main building was quite unlike the extended bastles nearby. It was much closer in scale to the linear form of farmstead that developed locally in the more peaceful circumstances of the later eighteenth and nineteenth centuries, and may, therefore, have superseded the other late structures within the fort. It may be possible to equate this house with the documented existence of a named building within the fort — the house in which the Northumbrian historian John Wallis was born. Hodgson believed that Wallis' birthplace, termed ‘the Castle-nook’ in the Kirkhaugh baptism record for December 1714, stood ‘just at the south entry to the station’, and that it was in this same house — the home of the later proprietor of Whitley Castle, Henry Wallace — that Horsley recorded a fragmentary Roman inscription in 1728.Footnote ⁸⁹ By 1840, however, when Hodgson published Sopwith's first map of the fort, this building was long gone and the name had been transferred to the present farmstead to the north of the fort.Footnote ⁹⁰

Later structures set into the gates and ramparts

At some point after the fort ceased to function, the approach to the north-eastern gate was occupied and blocked. The earthworks here define a narrow L-shaped building platform just outside the former gate, linked to a series of irregular terraces which descend between and beyond the terminals of the outer defences (J). Narrow banks and stone foundations indicate that some parts of the defences were enhanced to form pens and enclosures, suggesting the presence of a small farmstead. The only clue to the date of this arrangement lies in the adjacent pattern of broad ridge and furrow which, in stopping short of these enclosures, suggests that they were in use at the time of the ploughing. A medieval date is implied, although it is far from certain.

Another rectangular structure (K), evidently later than the fort, stood on a sheltered terrace overlying the defences between the south-east gate and the eastern corner, immediately above two scooped pens or yards, each with level bases, which were cut into the line of the innermost ditch and enclosed by a stone wall. This appears to have been a small farmstead with pens or sheepfolds, presumably long disused when Sopwith's plan of the fort (which also shows a small building, or corn-drying kiln inserted in the eastern corner) was published in 1853 ( fig. 3 ). The foundations of a further walled enclosure (L) extend across the north-western gateway and overlie the collapsed remains of the gatehouse. The track through the fort, the old via principalis, cannot have been much used at the time when this feature — a common form of sheep pen — was built.

Within the area of the survey there are four, possibly five, further locations adjacent to the fort where there is earthwork evidence of other late medieval or post-medieval farm buildings. These — M, N, O, P and Q — are described in the archive report, together with the pattern of enclosure that developed in the nineteenth century.