The Roman fort of Bochastle (NN 615078) forms part of a line of sites (fig 1), between Loch Lomond and the Tay, which have come to be known as the 'Highland line' or 'glenblocker' forts.  It is the third fort known from the south-west and lies between Malling and Dalginross (9.3 km (5.8 miles) and 20 km (12.7 miles) away respectively) on an area of low ground beside the River Leny, just west of modern Callander.  It sits, with limited views, at the mouths of two glens: Strathyre and Strath Gartney and, despite being damaged by a disused railway embankment in the south and by the river in the north, significant lengths of its ramparts survive as surface features.  Air photographs show parrot beak ditch breaks at the surviving east and west gates, along with the parch mark of an internal road between the two (fig 2).  The fort appears from the air to be rather smaller and more square in plan than its neighbours, at around 137m (n-s) x 140m (e-w), an area of c 1.9 ha (4.7 acres) over the ramparts, and does not appear to have an annexe.  This latter omission is coming to seem unusual amongst the Flavian and Antonine forts of Scotland, but recent geophysical work by ourselves at the Highland line fort of Drumquhassle, has shown that a series supposed annexes there were probably fairly modern (Woolliscroft & Hoffmann 2006, 38f).  It is noteworthy that both Bochastle and Drumquhassle lie at the smaller end of the local fort size range and it is, thus, possible that only larger forts showed these features.  If so, they might be related to the type of unit in garrison, and in view of the discovery of cavalry related finds at annexed forts such as Cardean (Hoffmann forthcoming) and Strageath (Frere & Wilkes 1989), it may be that cavalry forts were more likely to have them.

The site has been known since the 18th-century and appeared on Stobie's 1783 map of Perthshire.  Nevertheless, it was not universally recognised as Roman until after the Second World War, despite the name 'Roman Camp' which had become attached to it, and despite the discovery, before 1724, of a gold coin of Nero (Anderson 1956, 36f).  Excavations were conducted between 1949 and 1953 by the Glasgow Archaeological Society (fig 3), mostly by means of narrow slit trenches.  The published report (Anderson 1956) is not a model of clarity, not helped by poor illustration, but it does provide useful information, not least the establishment of a firm Flavian date and evidence for two structural phases in the defences.  The ramparts were around 7m wide and consisted of turf work, mostly with an earth and rubble core.  A number of turf and rubble features built against its rear were interpreted as ramps up to a walkway at the top, but could equally be secondary thickenings, and the excavator claimed that the northern rampart angled towards the south, east of the north gate, to occupy a low erosion terrace and thus stay clear of flooding from the river.  Roads were found emerging from the east and west gates, and it was claimed that both then turned to the south (probably in a secondary, perhaps post-Roman, phase), but neither was traced any further to find out where they might be going.

The excavator claimed that the fort had only a single defensive ditch, whereas air photography reveals a more normal two, but the ditch that was found was fairly typical for such a site, being largely V-shaped in profile and 3.66m wide by 1.53m deep.  Anderson also asserted that the fort lies markedly off centre within its ditch system, with the west rampart only 1.8m from its ditch, compared with 22.6m in the east.  The report is very confused at this point, but also mentions occupation levels under and around the road after it emerged from the east gate, raising the possibility that the fort was reduced in width during its occupation, and that a primary rampart thus might be found further east, closer to the ditch.  The ramparts appeared to turn inwards to provide deeply (up to 21m) recessed gates on all four sides.  No signs of towers or other gate structures were located, except for two postholes outside the west gate, but this is hardly surprising since they were looked for in what is almost certainly the wrong place: in the mouths of the rampart re-entrants, rather than their throats.  The published plan shows these re-entrants and, to a lesser extent, the fort corners, to be more angular than is usual on Roman forts.  In truth, however, it is difficult to believe that the real shapes could have been accurately determined by the limited trenching undertaken, and air photography certainly shows more rounded curves at the corners.

Few internal features were revealed and fewer still were properly understood.  Indeed the excavator came to believe that the river had eroded the fort interior at some point, removing all trace of buildings.  A few ovens were found built into the rampart back, one of which was overlain by a later phase of the 12.2m wide intervallum road, after a 0.22m thick layer of earth had had time to form over it.  A second oven appeared to belong with this later phase.  Otherwise little was found inside the fort except for patches of metalling.  This might mean nothing, however, since the few, very narrow, slit trenches dug would have stood little chance of finding intelligible timber building remains, especially if they really had suffered flood erosion.  Moreover, the one reasonable piece of area excavation that was conducted, recorded a floor and a line of posts, which might have been part of a barrack.  The report appears to give this discovery little significance, however and, sadly, did not even trouble to provide a plan.  Likewise other remains towards the north-east corner were brushed over with hardly a word and it is to be regretted that an important site could be treated in such a cavalier manner at a time when archaeological technique was already fairly advanced.

The Fort Survey
In an attempt to learn more, the Roman Gask Project conducted a geophysical survey of the entire surviving area of the fort and its immediate surroundings, in 2006.  Two techniques were used: a 9ha (22.23 acre) area (fig 4) was scanned by resistance survey, using two Geoscan RM15 meters, with readings at 1m intervals and a smaller, 5.58ha (13.78 acres) area was covered by a magnetic survey, using Geoscan FM256 and Bartington type 601 fluxgate gradiometers.

The results (figs 5-8) provide a significant supplement to the existing aerial and excavation data, despite some interference from a complex of paleochannels.  They also disprove some of the claims of the Anderson excavation report.  Firstly, they emphatically confirm the aerial indications that the fort had two ditches around all four sides.  This is the normal configuration for a Flavian fort in this area  and it is initially hard to understand how the dig could have found only one.  Interestingly, however, the supposed southward running external road shown on the excavation plan in the west, seems to follow the line of the outer ditch, in which case Anderson may have confused a ditch partly backfilled with stone (in effect an inverted field cairn) with a road, and then not dug deeper to check.  Likewise, there is no geophysical indication that the northern defences made the course change claimed in the report.  Quite the reverse; both the resistance and magnetic data show the eastern defences being abruptly cut off at the erosion terrace mentioned above, indicating that the fort originally had a normal, rounded right angle, north-east corner which has since been eroded by the river.  The survey also confirmed aerial indications that the ditches formed a variant on the normal Flavian 'parrot beak' entrance breaks at the gates, something which has also been seen on air photographs of nearby Malling fort (Woolliscroft & Hoffmann 2006, Plate 4).  In a standard parrot beak, the outer ditch turns inwards to meet the inner on either side of the entrance, but at Malling and Bochastle, both ditches turn inwards before meeting, to create a pattern that more resembles an eagle's beak.  The ditches run c 2.6m apart (lip to lip) and the inner ditch does match the excavated width at c 3.7m, but there are indications that the outer ditch may be slightly narrower: perhaps 3-3.2m.  Outside the ditches a line of higher resistance readings represents the remains of an upcast mound.

In addition to the ditches, the fort ramparts also show well, as a series of high resistance bands c 4m wide.  There was no indication that the east rampart lay at any greater distance from the ditches than the others, despite the excavator's claims.  Instead there was a uniform separation of c 2.6m all round.  Nor was there any trace of a second rampart that could represent a reduction of the fort's original area.  One aspect of the excavation plan was strikingly vindicated, however: the rampart re-entrants, which do indeed occur at all four gates.  As anticipated, these were more rounded than shown in Anderson's plan but, although not completely uniform, they are quite as deep as claimed, ranging from 19m at the east gate to 26m in the west.

Fortunately, three of the fort's four corners have survived, although that in the north-west has only just been spared from river erosion and a only small part of its south-west counterpart projects from beneath the railway embankment.  Nevertheless, this is enough to provide a more accurate size for the site, which now measures 148m (e-w) x 178m (n-s) over the ramparts and 170m x 197m over the outer ditch.  Making allowance for the gate re-entrants, this gives an internal area of c 2.2ha (5.4 acres), which is slightly larger than had been estimated from air photographs, but still quite small by comparison with other Flavian forts in the region, which can be over 60% larger.

There are indications that the fort interior might be much better preserved than the excavator would have us believe.  The road between the east and west gates shows strongly on the resistance plot (fig 5), but the magnetometry (figs 7 & 8) also showed signs of internal (presumably timber) buildings.  In particular, two large courtyard structures are visible immediately south of the east-west road, which even display room divisions.  These look likely to represent part of the fort's main range structures, suggesting that the fort faces north, towards the river, and the configuration of the more easterly of the two suggests a Praetorium (commanding officer's house).  Elsewhere in the fort, the magnetic data show signs of what could be long rectangular buildings which seem likely to be barracks.  However, the resistance plot shows lines of rig and furrow ploughing through the fort in much the same places and it is hard to be sure that these are not the same features.

What might be a faint, 21m diameter, ring feature was detected in the west gate re-entrant which may or may not represent a round house or cist burial.  In addition, the survey scanned a number of relatively modern lime kilns that remain visible at the surface to the fort's west, but without revealing new information.

The Temporary Camp
To the west of the fort is a temporary camp (fig 2).  No gates have been detected on its east and west (long axis) sides, but there are 'Stracathro' style gates in the north and south, a type usually ascribed to the Flavian period.  The site is unusually large for the type, however, for although its exact dimensions require clarification, it appears to measure c 530m (n-s) x  365m (e-w), an area of c. 19.4 ha (47.8 acres).  The difficulty in determining the camp's size derives from an oddity on its west side, where air photographs have shown up to three widely spaced ditches (RCAHMS neg: PT5512).  The easternmost has been traced as far as a corner with the south ditch, but the aerial traces of the north ditch pass further to the west as if to turn onto one of the others.  Only c 70m of the westernmost ditch has been seen from the air, but the other two have been traced, albeit intermittently, along the camp's entire length.  There have also been aerial indications (RCAHMS neg: PT5508) of a small rectangular enclosure attached to the southern end of the east rampart which might be an annexe.  This is not unusual in itself, for there are numerous temporary camps with annexes in Scotland (e.g. Marcus in Strathmore), but these are all even larger sites, c. 26 ha (64 acres) in area, and may be later in date.  They also have titulate, rather than 'Stracathro' type, claviculate gates and at present no other Stracathro camp has produced evidence for such a structure.  The only excavation undertaken on the camp in the past, a series of tests for the north ditch (DES 1998, 93), failed to locate it, despite the strength of the cropmark.              The geophysical survey around the fort scanned the camp's north-east corner area and detected the corner itself, which turns through 100.5 degrees, along with 124m of the north ditch and 258m of the east side, all exactly on the lines predicted by air photography.  No higher resistance bands were detected either inside or outside the ditch to suggest a rampart or upcast mound.  The camp's east ditch runs on a heading that would intersect the south-west corner of the fort but as it approaches it becomes obscured by the fort's upcast mound and may well be overlain by it.

The camp excavations
Two trenches were excavated on the camp defences to test, respectively, the potential annexe and the possibility of a triple ditch in the west.

Trench 1  was designed to study the northern junction between the camp and the supposed annexe ditch.  No matching southern junction has been detected from the air and so to further study the area and position the trench more precisely, a 1.53ha (3.78 acre) resistance survey was conducted at 1m resolution (figs 4 & 9). The camp ditch showed clearly, exactly on the line predicted by air photographs, as did the feature seen from the air as the annexe north ditch, which approached the camp, as expected, at right angles.  What had seemed to be the annexe east ditch, however, was found to continue further than expected.  It passed to the north of the supposed north ditch and continued right through the surveyed area: a total of 157m, on a slightly (4 degree) converging course with the camp ditch which took it from a separation of 78m in the south to just 67m in the north.

            The trench itself was 10m square (fig 9 for location) and revealed a range of features.  The most conspicuous was the temporary camp ditch (figs 10 & 11, sect A-B, c.7) which, as expected, ran north to south, roughly parallel to the west baulk.  When sectioned, the ditch (c. 57) proved to have a rounded V shaped profile, with a slight, so called 'ankle breaker' sump.  It was 2.26m wide and 1m deep (below the modern surface) and was cut into an orange/brown sand subsoil, containing gravel and larger stones (c.2).  The ditch had a more complex history than might have been anticipated, with at least one re-cut and a number of other features.  The primary fills included an orange/brown silty loam with pebbles in the bottom sump (c. 86) and coarse gravel in a brown sand matrix (c. 137).  The latter was extremely loose, but its upper surface did not seem wholly compatible with a natural angle of rest, so it is possible that the ditch was cleaned out after the layer's formation.  Above these levels was a thicker (0.29m) deposit of orange/brown loam with small to medium stones (c.58), accompanied by a lens of grey/brown silt (c.101), whilst, higher up, c's 87 (red/yellow, gritty clayey loam) and 88 (pale red/brown loose, gritty clay/loam), might also belong to this phase.

The ditch was then re-cut (c. 100) as a shallower (0.78m), more bowl shaped feature.  This was filled with an orange/red, gritty, clay/loam, with black and ochre coloured flecks (c. 80) above which was a grey/orange silty loam with small clay inclusions (c.79), a dark tan/orange clay loam (c. 78), a slightly red/brown clay loam with occasional pebbles (c. 77), an orange/tan clay loam with pebbles (c. 76), and lenses of slightly red, dark brown clay/loam with small clay inclusions (c. 85) and coarse gravel in a mid brown sand matrix (c. 136).  Above these deposits and possibly cut into them, lay a dark brown clay loam with black and ochre flecks (c. 84), an orange/brown loose clay/loam with black and ochre flecks (c. 82) and a small lens of light grey/tan clay loam (c. 83).  Later, a small pit (c. 10) had been cut through c's 76, 77 and 78 and then filled with a slightly orange/brown, loose loam which contained metal working debris (c.7), and a tan/grey sandy loam with occasional pebbles (c. 74).

After the ditch had completely silted, small slots were cut into both lips over the entire length exposed by the trench.  That on the ditch's western (inner) lip (c. 21) was filled by an orange/brown loam with pebbles and charcoal flecks (c. 6).  The eastern example (c. 25) held an almost identical fill (c. 8), with a lens of mixed, loose, grey and orange silty loam (c. 89) and was then overlain by thin layers of brown/tan silty loam with grit (c.90) and yellow/tan silty clay with ochre flecks (c. 91).

No east-west running ditch was detected in the trench, which confirmed the geophysical hints that the supposed annexe was an illusion.  There were, however, a range features to the east of (outside) the camp ditch which may account for the cropmark and low resistance feature which had been suggested as its north ditch.  Firstly, a 2.55m wide layer of mixed grey/brown silty loam (c. 39), up to 0.29m thick was found against the south baulk (fig 10 & fig 11: sect C-D + fig 12: sect's G-H & K-L), partly overlain in the north by a rather darker version of the same material (c. 38).  Both overlay various natural sand deposits: an orange/brown slightly loamy sand (c. 138) in the east and a purer, very gritty orange sand with gravel (c. 117).  All of these layers appeared to be water deposited and filled a cut (c. 67) in the c. 2 natural gravel which might represent a filled in glacial out-wash channel.  Certainly, section G-H (fig 12) showed that the layers were earlier than the c. 25 slot that cut the eastern lip of the temporary camp, along with a number of features to the north, but there was no stratigraphic link between them and the camp ditch itself, as the layer did not extend as far as the only ditch fill layer recorded there: a gritty, mid brown sandy clay with coarse gravel (c. 125).

To the north, c's 38 and 39 were cut by a broad (4.1m), but shallow (maximum 0.22m) depression (c. 17) which, in the east, extended beyond the excavated area.  This also cut and thus post dated the c.25 slot (and so the Roman ditch) along with a number of features to the north.  On the other hand, its fills were overlain with thin skins of a number of other materials, including a pale brown, loose sandy loam, with gravel (c's 30 & 33), a dark grey/brown loam with pebbles (c. 34), a similar but darker grey/brown loam with pebbles (c. 35) and a dark brown silty loam (c. 16).  It  was also cut (figs 10 & 12, sect P-Q) by small scoop (c. 50) filled by a pale brown, loose sandy loam with stones (c. 36) along with a tiny pit (fig 12, sect R-S) that might be the stump of a stake hole (c. 53), filled with medium brown loam with gravel (c. 52), and a very dark grey loam with charcoal (c. 51).

The dip was filled largely with a brown, water deposited, loamy silt (c. 12), containing iron fragments, furnace lining, part of an iron pig and fragments of heavily baked clay, but towards the east, it was bottomed with a shallow layer of brown gritty loam with charcoal and turf flecks (fig 12, sect K-M, c. 119), which became wafer thin towards the east baulk (fig 11, sect C-D).  Into this was set the remains of a bowl furnace (fig 13), represented by a roughly rectangular 0.7m (e-w) x 0.92m (n-s) area of deeply burnt red/yellow clay, containing charcoal, metal working debris, iron staining and large inclusions of charcoal and burnt turf (c. 71).  This overlay a layer of burnt, ochre coloured sandy loam, which also contained furnace contamination (c. 70) and it was surrounded by a deposit of grit in a yellow sandy matrix (c. 127) and bands of burnt brown gritty loam with charcoal flecks (c's 120, 121 & 122), which varied only in their stone content.  These may have derived from the furnace's upper structure and c. 121 yielded part of another iron pig.  The metal working debris from c. 12 was concentrated around the furnace area and may have been sitting on the surface of c. 119 when the c. 12 silt was deposited round it.  To the west lay a wider spread of mid brown, very gritty loam with charcoal flecks (c. 123) which might represent a working surface.  If so, the furnace stoke hole may have faced this way and, thus, into the westerly prevailing wind.  The furnace also cut (c. 124) into a series of apparently natural sandy layers (fig 12, sect K-L).  These had been disturbed by a series of mole burrows (c's 64, 68 and 72), but included c. 61, a very fine, burnt ochre coloured sand; c's 62 and 65, a yellow/brown, very sandy loam and c. 66 a brown gritty sandy loam.  No datable finds were recovered from this complex, but carbon samples were taken from c. 71.

To the south of the furnace, a number of more recent features were found projecting from the east baulk (fig 11, sect C-D).  No datable finds were located, but all seem likely to be fairly modern.  One of the earliest features, a small pit, 0.53m wide by 0.32m deep (c. 129), cut both c's 12 and 39 and was filled with a dark brown sandy loam with stones (c. 128) that contained a sheep skull which did not appear to have been in the ground for more than a decade or so.  This feature had been cut by a still smaller pit (c. 134), filled with a similar dark brown sandy loam with stones (c. 130), and this also cut another earlier pit (c. 81), filled by a mid brown, stone free, sandy loam (c. 131).  There was no stratigraphic connection to link this latter feature with the c. 129 pit, so it was not possible to tell which came first, but it too cut the c. 12 fill of the furnace depression.  Finally, the c. 134 pit was cut (c. 135) by a narrow (0.2m), straight sided, 0.25m deep feature, filled by dark brown sandy loam with occasional stones (c. 132), which may have been a posthole.

To the north, the furnace basin cut a series of earlier slots (figs 10 & 11, sect A-B), whose purpose remained uncertain.  The first (c. 111) was filled with a light brown, loose sandy loam with small pebbles (c. 26) and followed a curving course.  This had been cut by two further slots, firstly, a shallow (0.16m), flat bottomed feature (c. 23) filled with brown silty loam with fine gravel (c. 11) and lenses of green/yellow slightly silty sand (c. 27)).  Secondly, a narrower, but deeper, bowl shaped slot (c. 22) filled with medium brown loose loam with gravel and medium sized stones (c. 13), medium-light brown, loose loamy sand with pebbles (c. 14) and a thin skin of dark tan, loose loamy sand with pebbles (c. 97).  The c. 23 slot was cut by another bowl shaped slot (c. 108), filled with medium-dark, brown/orange silty clay with pebbles (c. 92) and a dark grey/brown, loose silty loam with occasional small gravel (c. 93).  The latter underlay a light brown/yellow, loose silty loam (c. 9) which may, have been cut into it (?cut c. 109).  No datable material was found in these features and no stratigraphic link with the temporary camp ditch had survived, except that c. 92 was partially overlain by c. 90 which also overlay the filled in c. 25 slot in the ditch's east lip.  As a result, all that can be said is that they predate the furnace basin.  That said, c. 11 produced more metal working debris, including possible evidence for lead working, which suggests that such activities were already established on the site when the known furnace was constructed.  A section (fig 12, sect N-O) cut to further test these relationships found the furnace depression to be cut into a pale yellow natural sand (c. 60) and with a layer of orange brown sandy loam (c. 59) underlying its usual c. 12 fill.

Only a 1.5m wide strip was uncovered beyond the ditch's inner lip which was probably not enough to reach the rampart.  Certainly no slumped material from it was found, and the evidence from Trench 2, further west, would suggest that it has probably been ploughed away.  Nothing else was found in the area which could confidently be assigned to the camp, but a number of other features were detected.  In the trench north-west corner was a small pit  (c. 18) cut into a natural yellow/brown sand (c. 4), and filled with dark orange/brown silty loam with gravel (c. 1).  This had been disturbed by an animal burrow (c. 105) filled with a yellow/grey silty loam (c. 99), that cut its relationship with another small pit (c. 24) which was filled with orange/brown loose sandy clay/loam (c. 3).

Further south, both the temporary camp ditch (with a fill of orange/brown loam with small to medium stones (c. 58)) and the slot in its west lip had been cut (c. 19) by an irregular pit (figs 10 & 12, sect's E-F &  I-J).  This was filled largely with a medium/brown clayey loam (c. 5) containing a sherd of Victorian china, overlying thin layers of orange/brown loam (c. 55) and orange/tan sand (c.56).  This feature had been cut by a narrow, shallower pit (c. 116), filled with what appeared to be fire or furnace waste.  A lower layer of absolutely black loam (c. 113) was predominantly made up of soot, whilst the top fill was mostly grey/brown ash (c. 114).  This had in turn been cut (c. 40) by a shallow scoop filled with dark, slightly orange/brown loam (c. 29).

Trench 2  studied the three possible camp west ditches seen from the air.  To ensure accurate trench location and further investigate the area, a 90m (e-w) x 30m (n-s) resistance survey was conducted (figs 4 & 14), before work commenced.  This identified all three ditches as N-S running low resistance bands, but found them to lie a few meters to the east of the lines predicted by existing air photo rectifications (fig 2 shows the true position).  In addition, a narrow, E-W running, high resistance anomaly was found which seemed likely to represent a stone feature such as a former field wall.

The trench was 70m (e-w) by only 1m (n-s) (figs 14 & 15).  It revealed a range of natural materials, all of which appeared to be of glacial origin, including (from E-W) an orange/brown gritty clay with stones (c. 503), a yellow/brown sandy clay (c. 506), a beige/yellow coarse gritty sand with pebbles (c. 511), a buff yellow sand (c. 512), a dark ochre/yellow grit, with small stones (c. 515), a red/brown clayey sand with grit and small stones (c. 516), a buff clayey sand (c. 527), an orange sandy clay (c. 524) and a yellow ochre coloured gritty sand (c. 523).  The three ditches were revealed exactly where the resistance survey had predicted, but the area had been plough damaged to the point that there were no remaining stratigraphic connections that might have made it possible to determine the order in which they were dug.  Likewise, no corresponding rampart or upcast material had survived.  All three ditches had V shaped profiles that would fit well into a Roman military context, but no carbon or other datable material was found to shed greater light.  The westernmost example (figs 15 & 16, sect T-U), however, may have reused an earlier ditch, perhaps connected with the nearby Dunmore hillfort.  Indeed, most of the ditches showed more complex histories than anticipated.

The primary cut (c. 567) of the western ditch was flat bottomed in profile, with asymmetrical sides which were near vertical in the west, but more gently sloping in the east.  If this was a camp ditch, the vertical face would be on the outer side and we might thus interpret it as a Roman fossa punica, but the (1.22m wide) flat bottom might sit better in a prehistoric context.  Only a 2.3m width of this phase had survived, but an extrapolation of the surviving sides would suggest that the original width would have been greater (perhaps c 2.75m).  It was 0.97m deep from the modern surface and was filled with what was almost certainly a single layer of orange/yellow sandy clay (c's 563 & 565).  This had, however, been cut (c. 569) by a slightly deeper (1.04m), V shaped ditch with an 'ankle breaker' sump.  Again, the 3.32m width that survived was not its full original extent (perhaps c 3.6m) and the feature had filled with a single layer of orange/yellow sandy clay with grit (c. 564).  The layers above do not easily fit with a natural silting process and it seems likely that the ditch was then cleaned out (c. 568) to produce a roughly V shaped ditch, only 0.91m deep and at least 2.1m wide.  This had filled with layers of red/brown sandy clay (c. 562), burnt black sandy clay (c. 561), mid brown sandy clay with orange streaks which may have been degraded turf (c. 566), orange/brown sandy clay (c. 560) and orange sandy clay with stones (c. 559).  These layers were then truncated, probably by plough action on what would, by then, have been a shallow surface depression.  The resulting soil, a buff/red sandy clay with charcoal flecks (c. 558), contained quantities of fire debris and had, in turn, been overlain by a red/brown sandy clay (c. 557) and the modern ploughsoil (c's 0 & 528).

The central ditch (figs 15 & 16, sect V-W) lay 17.3m from the western ditch (lip to lip).  It began as a V shaped cut (c. 572), 1.27m deep and with an original width of perhaps c 0.75m greater than the 2.43m that survives.  It was cut through a range of natural deposits, including layers of yellow gritty sand (c. 520), beige/brown coarse sand with dark striations (c. 537), red/brown grit with pebbles (c. 539), beige/brown coarse sand (c. 538) and beige/yellow sand (c. 540).  The cut had filled with layers of beige/grey silty sand and grit (c. 542), gravel (c. 545) and beige/yellow gritty sand (c. 541).  It was then re-cut (c. 553) as a slightly shallower V shaped ditch, 1m deep and at least 2.7m wide, with fills of orange/red gritty sand with charcoal flecks and iron slag (c. 547), and orange/brown sandy clay with pebbles (c. 548).  Once the ditch had filled almost completely, a number of other features were cut into it.  Towards the centre was a square sectioned slot (c. 554) filled with orange/red sand and grit with charcoal flecks and slag (c. 546).  Just to the west of this, but in an uncertain stratigraphic relationship with it, was a possible bowl furnace (c. 556) made up of mixed yellow and brown burnt gritty sand with small stones (c. 544), overlain with dark red/brown gritty sand with charcoal and pebbles (c. 543).  Meanwhile, a second bowl furnace had been cut (c. 555) into the former ditch's eastern lip and consisted of layers of black charcoal and iron working debris, including furnace lining fragments (c. 550), singed orange/yellow sand with small stones (c. 549) and burnt brown/red loam with stones (c. 551).  These were then covered by the modern ploughsoil (c's 0 & 528), which here yielded more metal working debris, including slag, furnace lining and pieces of iron ore.

The eastern ditch (figs 15 & 16, sect X-Y) lay 27.8m from the middle ditch (lip to lip) and was both smaller and stratigraphically less complex than the others.  It was a V shaped feature, 1.02m deep and 2.33m wide, cut (c. 501) into the c. 503 natural, and filled with layers of orange/brown silty clay with small stones (c. 531), orange gritty clay with stones (c. 532), light grey/brown sandy clay with gravel (c. 530), grey/brown sandy clay with gravel (c. 529) and brown gritty sand with gravel (c. 502).  The form of c. 531 seemed difficult to reconcile with a natural silting process, however, and it is possible that the ditch was cleaned out after its deposition (?cut c. 535).

Nothing was found within the trench which could represent the sort of internal structures (pits, ovens etc.) that have been found inside Roman temporary camps elsewhere.  Nevertheless, a number of other features were detected, although most proved to be of little archaeological interest.  In the west, a curved slot (c. 525), filled with dark brown sandy clay with gravel (c. 526), proved to be an animal burrow.  14m further east, a spread of flat stones (c. 573) was encountered where the trench crossed the high resistance feature mentioned above.  No in situ structure survived, but the stones were very much of the size and type used in local field walls.  To the east of the central ditch, a 0.4m diameter, circular area of dark red/brown sandy clay (c. 509), was initially interpreted as a posthole, but further investigation showed it to be a shallow depression in the subsoil, probably where a boulder had been ploughed out.  Immediately to the east of this was a curving slot (c. 507) that initially appeared as a possible roundhouse foundation but, when emptied, it proved to be an unusually deep plough scar.  Three more such slots were located running parallel, a little to the west of the easternmost ditch (c's 519, 521 & 522).  Each was 0.29m wide and, at first, they resembled a pattern of sleeper beams, but these too proved to be large plough scars.  Finally, the westernmost plough scar cut through a 1.9m wide band of rough metalling (c. 504) which might be an old field track.

Conclusions
The survey showed that Bochastle fort had a more normal Flavian configuration and a much better preserved interior than the Anderson excavation report had alleged.  One of its most helpful results, however, was the confirmation of the Glasgow excavation's claimed gate re-entrants.  For similar features have been found at a number of other 1st century forts in the north, with Cardean (Hoffmann forthcoming), Drumquhassle (Woolliscroft & Hoffmann 2006, 38) and Strageath (Frere & Wilkes 1989, fig 66) all showing gates of a similar design, albeit not on all four sides.  It is possible that gates of this type might prove to be more common than is currently suspected, and may eventually come to be seen as another diagnostic late 1st-century feature.  Tactically, they fit well with the parrot/eagle beak ditch breaks, to create what amount to funnels that could be guaranteed to cause confusion amongst even an ordered rush on the gates.  As the outer ditch turns in to meet the inner at a parrot beak gate, an attacking force would have been able to pass through the outer ditch on quite a broad, c 25m wide, front.  It would then find itself rapidly compressed as the outer ditch ran towards an inner ditch break only c 15m wide.  Some of its outermost members might even be pushed into this ditch, but the others would be forced inwards, causing confusion in the ranks at a time when they were already under fire from the fort.  They would then pass down the rampart re-entrant, becoming still more crushed and disordered, to an entrance just c 6m wide.  There they would be faced with the barrier of the gate itself, whilst under enfilading fire from the rampart funnel, and with their escape route blocked by their own comrades forging on from behind.  The result would be a killing ground with the funnel acting as a trap.

The fort has long been well dated, but as no datable material has ever been recovered from the temporary camp, the Stracathro gates are the only available clue.  Nevertheless, an extra hint might be derived from the fact that the camp ditch becomes obscured by the fort's upcast mound.  It is always dangerous to look for stratigraphical answers in remote sensing data, but this might imply that the camp predates the fort and, although this did not fall under the remit of the present investigation, the issued could be now answered for certain by a single small trench.  A number of other northern forts have camps outside them, and one model for their function might be that they housed the soldiers who built the forts.  At other sites, the fort and camp remain totally separate (although Stracathro fort's annexe crosses its camp defenses), which makes it difficult to check this model.  Nevertheless, the camps have always seemed rather large for such a function and the fact that it seems unlikely that Bochastle fort's builders would deliberately dig a fort ditch through their own camp, appears to make the scenario less likely still.

The excavation successfully addressed the question of the camp annexe, which was firmly shown not to exist, but considerable ambiguity still surrounds the western ditches.  None of the three seen from the air could be safely discounted as Roman but, at the same time, none can be proven to be so.  Moreover, even the relatively short known length of the westernmost ditch is too large to be an external feature, such as a titulum, which would anyway sit badly with the Stracathro gates to the north and south.  As a result it is perfectly possible that all three represent Roman camp ditches.  If so, the site may have seen a prolonged occupation during which its size needed to be adjusted several times, or successive re-occupations by forces of differing size.  The evidence for cleaning and/or recutting of the ditches suggests further complexity, but the present evidence does not allow the site's occupation history to be reconstructed in greater detail with any confidence.  It is possible that there were three separate occupations, each of which was long enough for silting to make cleaning or recutting necessary.  Alternatively, a greater number of shorter occupations could be envisaged, some of which saw the digging of wholly new ditches, whilst others involved only the reconditioning of old ones.  Whatever the case, however, temporary camps do not, as a rule have fort style, multiple ditch systems and so, assuming that at least some of the phases in all three ditches are Roman, it is still unlikely that two or more were ever in active use together.

The site produced plentiful evidence for post-Roman, probably early modern, industrial activity, with metal working furnaces in both excavation trenches and surface indications of lime kilns, to the west of the fort beside the Leny.  The surrounding hills hold a range of metal ores, including iron, along with other useful minerals such as limestone and slate.  The 1799 Old statistical Account for Callander (vol XI, 586f) mentions lime burning and quarrying in operation at that time, with demand outstripping supply.  It also mentions the discovery of a vein of richly argentiferous lead, which yielded 20 ounces of silver per hundredweight of ore.  The latter might be a pointer to the date of the possible lead working evidence found in Trench 1.  Indeed, the fact that large pieces of iron pig were found in association with its furnace might even be a pointer to cupellation activity, rather than straightforward iron working.  For in the latter case one would not have expected the iron to be abandoned, whereas it can be used in silver cupellation, as an expendable material for chemically breaking down galena ore and absorbing impurities such as sulphur (Goldenberg 1996, 34ff).  Metal extraction may have ended in the early 19th century, however, for although the 1845 New Statistical Account (vol X , 352f) still records local lime and slate working, metals are no longer mentioned, perhaps thanks to the growing industrial revolution and the availability of cheaper materials produced elsewhere.

Interestingly, the prospect of metal, and especially precious metal, resources may have held attractions in the Roman period too.  Indeed Rome seems to have exploited gold and silver deposits in Britain so quickly after taking over the areas that contained them that one suspects that prior intelligence of their location was given some priority.  The prime reason for the fort's location was undoubtedly strategic: blocking glens that allowed egress from the Highland massif, but the silver in particular, could have provided an added attraction if the vein was known about at the time.  Certainly, this is a consideration that future excavations might bear in mind.  Nevertheless, the lack of an annexe might be a hint that the resource was not being exploited, at least on any scale, since these are often the sites of Roman military industrial activity.  Likewise, nothing clearly identifiable as a furnace was detected around the fort by the magnetic survey.

 D.J.Woolliscroft & B.Hoffmann
University of Liverpool

Bibliography
Anderson, W. A.  (1956)  'The Roman fort at Bochastle', Trans Glasgow Arch Soc (New ser 14), 35-63.

Frere, S. S. & Wilkes, J. J.  (1989)  Strageath, excavations within the Roman fort 1973-86, Britannia Monograph Series No.9, London.

Goldenberg, G   (1996)   'Archäometallurgische Untersuchungen zur Entwicklung des Metallhüttenwesens im Schwarzwald. Blei- Silber- und Kupfergewinnung von der Frühgeschichte bis zum 19.Jahrhundert'. in G. Goldenberg, J.Otto and H.Steuer (eds), Archäometallurgische Untersuchungen zum Metallhüttenwesen im Schwarzwald. Sigmaringen, pp.9-275.

Hoffmann, B. (forthcoming)  Excavations at Cardean 1967-75, SAIR, Edinburgh.

Woolliscroft, D J and Hoffmann, B    (2006)    Rome's first frontier, the Flavian occupation of northern Scotland, Stroud.

Background papers

Aerial photographs

NOTES ON OTHER FEATURES
Birgitta Hoffmann

 Although outside the scope of the Gask Project survey and excavation, notes were taken on two other features inside or in close proximity of the Bochastle Roman sites.  These were a group of lime  and a large, raised, stone walled oval enclosure which seemed to have been cut into during the construction of the outbuildings of Bochastle farm.

The early modern lime kilns
Four lime kilns were discovered to the south of the River Leny, 60m west of the Roman fort's north-west corner and to the east of the farm road.  They appear to have been dug out of the river terrace from the south.

 From west to east:

[KILN 3] The field wall and adjoining field track appear to have partially destroyed a lime kiln.  The only surviving evidence consists of visible stone packing, similar to the sides of the next two kilns on the east side of the track.

[KILN 2] This, the best preserved kiln, appears to have been lined with stone, especially river cobbles, many of which are still visible in the edge of the bowl. Diameter: 4.80m, max. depth of bowl: 1.02 m. Above the level of the terrace bank, the structure originally continued as a 0.6m wide bank of stone and earth.  After use, the structure appears to have been broken open from the south side and emptied, leaving an c. 0.4m high step on the south side, which is c. 0.8m above the  ground of the adjoining field.  The gap in the wall of the kiln on this side is about 1.2m wide at the bottom, but increases towards the top of the preserved remains, where it takes up about a third of the kiln's circumference.

[KILN 1] Part of the original dome is preserved above the level of the terrace (c. 0.6m high).  Diameter: of  dome: c. 3.6m. Height of structure above the field c. 1.8m.  Bottom of the structure c.1.2 m above general field level.  The structure is similar to kiln 2, but only partially clear.  A pile of river cobbles and limestone fragments (now pink with burning) lay on the bottom of the structure.  Size of limestone blocks c.35 cm x 30 cm.

[KILN 4] On the  edge of the terrace to the east of  kilns 1 and 2 is a nearly circular bowl shaped depression that appears to have entrances from north and south on the edge of the terrace.  The structure is markedly shallower than the two better preserved kilns and the bottom of the bowl is barely 0.3m above the field level: diameter: 4.8 m  There are no stones visible in the east edge of the bowl, but a typical stone lining  is clearly visible on the western side.

A large number of nettles to the south of the lime kilns suggests increased fertility, implying that the burnt lime was mainly extracted on this side.  Oral history suggests that lime from this estate was used in the construction of the Free Presbyterian Church in Callander.

The large, raised, oval enclosure
Due to the presence of a series of occupied wooden kennels on the south side of the structure, the survey in that area was only cursorily.  The feature consists of a stone revetted platform with a circular stone structure in its centre.

The platform's long axis diameter is c. 50m.  It is near level in the centre and curves down slightly towards the sides.  It is surrounded by a heavy dry-stone revetment, consisting of large boulders of head-size and above.  The revetment is fairly well preserved on the south side, but there is evidence of collapse and potential stone robbing to the west and especially on the north side.   A substantial holly tree and the root systems of at least three further trees attest that the feature was at some point in the
past two hundred years edged with bushes or trees, which were growing at the approximate current level of the revetment wall top.

The farm building is clearly cut into the structure (the cut is plainly visible on the north side) and has destroyed the eastern third of the feature (perhaps more).  Where still preserved, the revetment is faced with flattish stones, leaning slightly inward (c. 0.15m at 1m above ground).

To the west, facing a small stream valley about 100m away, is a gap in the revetment that appears to be intentional, since the stones appear to come to a straight edge on both its north and south sides, with no indication of a continuation of the revetment line in the gap.  This suggests a gate: c. 2.6m wide in its centre.  The revetment at the gate's north-side is 1.35m and c. 1.4m on the less well preserved south side.  The distance from the centre line of the revetment wall, at the gate, to the western edge of the internal round stone structure (see below) is 16.5m

Throughout the interior a probing survey with a quarter inch soft steel probe encountered stones/packed gravel about 10-15 cm below the surface, suggesting a stone/gravel raft underlying the current grass sward.

Originally, the internal circular stone structure was probably in or near the centre of the large oval platform, but it now almost touches the back wall of the farm stable building.  It is built largely of river-rounded boulders and some broken stone rubble.  Its wall appears to consist of two mortared stone faces with a white mortar core.  The walls are 0.6-0.7m wide and the structure's overall diameter is 8.65m.  The structure still stands to a height of 1.1m on the south side, with indications of an overgrown foundation offset of about 0.4m on the outside. This could not be investigated further, however, because of the dense undergrowth.

The barn/stable west wall shows indications of a blocked door originally opening into this structure suggesting that the farm has used the structure at some point but, although the two structures appear to be joined by a small patch of small rounded stones in mortar at the surviving north side, nothing can be seen in the south.  There is no indication that the round structure was bonded with the barn in the way that other 18^th century farms in Stirling are bonded with contemporary horse mills. The round structure’s size would, however, have been too cramped for use as a horse mill and is markedly smaller than known horse mills in the area, for example on Blairdrummond Moss. Likewise, although it is not absolutely clear whether the farm building and circular structure were designed together, the use of different materials (large stones in the rounded structure, rubble in the farm building) suggests otherwise.