Excavations at the Roman Gask tower of Greenloaning revealed an abnormally large timber tower surrounded by a double ring ditch. The tower had at last two structural periods, but no absolute date for either could be determined.

THE SITE

Greenloaning lies at NN 831072, immediately to the north of the Braco exit of the modern A9 (illus 1) and 2.7km (1.89 R.miles) south of the fort of Ardoch. It stands on the 120m contour, to the south of the Allan Water, with good views in all directions, especially to the north and west, and is in visual contact with Ardoch and the fortlets of Glenbank and Kaims Castle.

The site was discovered from the air in 1985 by the RCAHMS, as an indistinct ring ditch, c.11m south of the Roman Ardoch road, towards which its single entrance break faced (Negative A36934). In the following year the road failed to show, but a much clearer image was obtained (illus 2, Negative A28898) of the site itself, which proved it to have a double ring ditch similar to those of the southern Gask towers of Shielhill North and South (St.Joseph 1976, 22) and Blackhill Wood (St.Joseph, 1977, 135 & fig 8), all of which are also in visual contact. The site thus appeared to be a further member of this series, the southernmost example yet known.

SURVEY WORK

In advance of the excavation, a (40m)2 resistivity survey was conducted to look for features outside the planned excavation area and to gather additional information, especially concerning a number of anomalous aspects of the air photographs. In particular, the photographs appeared to show the inner and outer ring ditches joining on both sides of the entrance, rather than coming to separate butt ends (illus 2). This would be an unprecedented ditch pattern for a Roman tower, but the survey (illus 3) did nothing to dispel this impression. Both ditches produced bands of low resistance around the bulk of their circuits, with the inner ditch showing more clearly than the outer, and the two did indeed show some signs of coming together, especially on the western side of the entrance.

The ditch system registered as c.24-25m in total diameter. But outside the outer ditch, to the south-west, was an additional broad (12-13m wide) band of high resistance, running parallel to the ring ditches. This area was not later excavated, but the feature presumably represents an upcast mound. If so, the total diameter would increase to c.50m, although the survey did not extend far enough to the south and east to confirm that the high readings continued right round the site. St.Joseph (1976, 25) has commented on the lack of such mounds amongst the southern, double ditched Gask towers (although they are well known on the more northerly Gask Ridge sites) but, since all of these sites lie on land that is, or has been, long under the plough, it is hardly surprising that they have not been located in the past. They are easily ploughed down. They do not show well on air photographs. Excavators rarely if ever look outside ring ditch systems and, although they can be revealed by resistivity (e.g. Woolliscroft 1995, 67 and fig 3), so far as the writers are aware, this is the first geophysical survey ever undertaken on the Gask system. Under these circumstances, if confirmation were needed, the presence of an upcast mound here can probably be taken to indicate that all four southern towers had them.

Finally, the air photograph showed a roughly square dark patch, c.11m across, immediately to the north of the eastern side of the entrance break (illus 2), and this was reflected by a similarly shaped area of very low resistance (illus 3). Time did not allow this feature to be excavated, but given the proximity of the Roman road and the fact that the field proved to contain a stratum of gravel aggregate very close to the surface, it probably represents a quarry pit.

THE EXCAVATIONS

The excavations were directed by the first writer and carried out by volunteers, and students of the Universities of Manchester, Freiburg and Warsaw.

The Ditch System

The ring ditches were studied by three full sections through both lines (illus 4 and 5, T's 2,3 and 5), whilst most of the entrance area was opened in plan (T's 4 and 6). In all cases both ditches proved to have a relatively normal Roman "V" profile, although neither showed much sign of the usual "ankle breaker" sump, and both, especially the outer ditch, were rather smaller than is normal on a Roman tower site. This may, however, be a typical feature of the southern Gask area, for the ditches recorded during St.Joseph's excavations at the Shielhill towers (Maxwell 1974, 13) seem to have been similarly slight. As with all of the Gask towers so far examined, no sign of palisades were uncovered, despite a careful search, either inside, between or outside the ditches. The only possible exception was a single tiny square pit feature at the south-west end of T2 (illus 4), which was only 0.16m wide and 0.09m deep and whose very identification as a post/stake hole was highly uncertain.

The inner ditch varied in width between 2.15m (T2) and 1.54m (T3), averaging 1.91m, whilst its depth ranged between 0.54m (T2) and 0.46m (T3) and averaged 0.52m. Its average outer diameter was 15.54m and it enclosed a central area with an average diameter of 12.5m. Its centre line was a near perfect circle (not a difficult achievement given a stick and a length of string) which is a common pattern on Roman tower sites.

The outer ditch was rather slighter, explaining its fainter appearance on the resistivity plot. Its width varied between 1.74m (T5) and 1.03m (T3) averaging 1.45m, whilst its depth varied from 0.49m (T2) to 0.28m (T3) and averaged 0.4m. Its outer diameter averaged 24.7m and the inter ditch spacing averaged 3.2m, or 4.48m between the ditch centre lines. The outer ditch was rather less circular and its centre lay c.0.5m to the north of that of the inner.

Both ditches showed similar fill patterns (illus 5), although the inner ditch contained rather more fill layers. Both had only a few centimetres of primary silt in their bottoms (illus 5, layers 7, 11 and probably 8), above which were varying numbers of sharply defined and often very different, fairly loose, silt free layers, some of which had undulating tops which could not have been formed by any natural deposition process (e.g. illus 5, layer 10 in T's 3 and 5 and layer 13 in the inner ditch in T3). All of this would suggest that the ditches were backfilled rather than silting normally and, thus, that the site had either been abandoned after only a very short occupation or, perhaps more likely in view of the results from the interior, that the ditches had been kept scrupulously clean whilst in use. This is a broadly similar pattern to that found at the two Shielhill sites (St.Joseph 1976, 22), except that Greenloaning showed no sign of ash or turf in its ditches, except for one small lens at the top of the inner ditch fill in Trench 3 (illus 5, layer 12), which contained charcoal in a grey loamy material which might have been degraded turf.

Both ditches and especially the inner, showed signs that they were initially designed to be larger. For both tended to flare out to become very much wider in their top few centimeters (illus 5, see especially the inner ditch in T2). This is remeniscent of a process recently encountered by the writers at the Antonine Wall tower of Garnhall, Cumbernauld (Woolliscroft and Hoffmann, forthcoming). Only about half of that site's ring ditch had ever been fully completed, but a c.0.15m deep marking out slot the full width of the finished ditch had been dug around at least part of the rest of the circuit. If the widening around the tops of the Greenloaning ditches began as a similar marking out slot, the ditches may originally have been meant to be roughly twice their present size, which would have made them far more typical of Roman tower defences elsewhere. Certainly, the flaring does not seem to have been a product of the ditches being backfilled by having their immediate surroundings spaded into them, as there is little former topsoil in the fill. Most of the backfill (illus 5) consists of varying mixtures of loam, sand and gravel similar to the subsoil strata through which the ditches were dug and so, upcast mound remains not withstanding, they were probably largely filled in with their own original spoil. Why the ditches as actually dug were so small remains a mystery, however. The subsoil profile copnsists of sand and clay, overlying a sand and gravel aggregate and, although the latter is quite hard, it is not particularly difficult to remove with a pick, whereas elsewhere the Romans were quite prepared to dig full sized tower ditches into solid rock (e.g. Woolliscroft and Swain 1991, 22-25).

The entrance arrangements are also somewhat unusual. Despite the air photo indications, the ditches do come to perfectly normal, separate, if rather flattened, butt ends. The inner ditches leave an entrance break 3.6m wide and, in the east, the outer ditch stops, as expected, in line with its counterpart. But, on the western side, the outer ditch stops well short of the inner ditch break (c.42o of arc) to leave a much wider outer entrance 11.1m across (14.7% of the ditch's total circumferance), whose centre line is orientated c.25o anticlockwise (west) of that of the inner ditch gap. It is not unknown for Roman multi-ditched towers to have their entrance breaks slightly askew but, so far as the writers are aware, such a large discrepancy is unique, although air photographs of the nearby tower of Shielhill North (e.g. CUCAP BKE-50) may show a similar pattern there. No sign of any post holes or other gate structures were found.

The air photographic indications that the inner and outer ditches joined on each side of the entrance were probably caused by a number of later features. In the east, the outer ditch butt end was all but obscured by a small SSW-NNE running ditch which cut into the backfilled tower ditch and ran roughly parallel to the Roman road (illus 4, T'6, ditch 1 and illus 5, section I-J). This cannot have been the cause of the pit-like feature described above, which should lie just outside the excavated area and, with hindsight, it is separately, if faintly, visible on the resistivity plot (illus 3). But the optical illusion of a join in the tower ditches here seems to have been caused by a modern land drain slot, which runs slightly west of due north past both butt ends, cutting both the later ditch and the outer ring ditch. Ditch 1 had filled to about a third of its depth with a sand and gravel mix (illus 5, layer 8) which is probably a weathering product of the sand and gravel aggregate (illus 5, layer 9) into which it was dug, but the rest of the feature was filled up with a mixture of charcoal and sand (illus 5, layer 17) which resembled fire waste.

On the western side of the entrance, the outer ditch butt end is crossed by two later ditches, both with a "U" shaped profile, which actually cross at this point (illus 4, T4 and illus 5, section G-H). The earlier (illus 4, ditch 2) runs essentially SW-NE, narrowing and growing more shallow to the east and, as it almost reaches the inner ditch butt end, it is probably this feature that caused the Air signs that the ring ditches joined on this side. The final ditch (ditch 3) runs almost due north, but could not be followed to the northern limit of the trench. Each of these ditches contained only a single fill layer. In the case of ditch 2 this was a sand and gravel mix (illus 5, layer 18), which was probably a weathering product from the layers through which it was cut, but ditch 3 contained nothing but a clean rich brown loam indistiguishable to the base of the modern plough soil.

No closely datable material was found in the ditches, but a total of seven body sherds (2 in T3, 4 in T4 and 1 in T6) of a red coarse ware were located in the plough soil. These were, however, badly abraded and Dr J. P. Wild, who kindly examined them, was only able to describe them as "possibly but not definitely Roman". Environmental sampling by Dr S.Ramsay of the University of Glasgow found no organic material surviving in the ditches' free draining backfill and silt layers and no large enough fragments of charcoal survived to attempt either a species analysis or a C14 date.

The Interior

Most of the internal area was examined (illus 6) and proved to have been surfaced with a layer of rammed brown clay, mixed with fine gravel and a few larger stones. Much of this layer had been badly plough damaged, but considerable patches did survive and here its thickness varied between 30 and 70mm with an occasional area, up to 170mm thick, where the layer seemed to have been used to level slight undulations in the ground. Two patches in particular had survived in near perfect condition: one immediately to the south of feature 4 (illus 6, pit) and the second at the northern end of Trench 5. The eastern end of Trench 3 also showed it in good preservation and in both Trenches 3 and 5 the inner ditch proved to have been dug through this layer. Both the clay and the gravel used are native to the site, although not in combination and this surfacing seemed to have been laid onto subsoil, rather than onto the original topsoil. The southwest edge of Trench 1 also produced a small pile of small boulders, but these did not form a recognisable feature.

In all, six post holes were discovered, four of whose post pipes survived. These fell into in three distinct pairs (illus 6 and 7, PH's 1, 2 and 3) which had clearly formed three of the four corners of a two period rectangular structure. The fourth (north-east) corner had appaently been destroyed by a pit (illus 6, feature 4) as, indeed, had much of the opposite south west corner (PH 3). This structure was set slightly south-west of centre, with its northern side 3.5m inside the entrance. It, thus, sat well to the rear of the internal area, with its south-western corner only 1.15m from the inner ditch. There was no sign of any feature akin to the "ladder platform" found at the nearby site of Westerton (Hanson and Friell 1995).

As to the post pits themselves, all four were oval in plan, rather than rectangular as at Shielhill North (DES 1974, 51 & St.Joseph 1973, 218) and, where it was possible to tell, all had held circular posts rather than the square timbers reported at Shielhill. The secondary posts would appear to have been somewhat wider and shallower than their primary counterparts but the features are otherwise best dealt with individually. It should, incidentally, be noted that, as the post pipes were initially rather difficult to see in plan, thanks to plough damage to the tops of the post pits, none of the sections in illus 7 quite cut through their exact centre lines, so that their true diameters were only determined whilst the sections were being dug.

Post hole 1 consisted of a pit c.0.9m wide and 0.38m deep containing a post (illus 7, PH1, layer 2) 0.42m wide and 0.52m deep, packed in place with various clearly defined and different fill layers (illus 7, layers 3, 4a, 6 & 7). Once the post had been installed, the rammed clay and gravel surfacing, refered to above (layer 1) had been laid around it, covering the rest of the pit. Both this pit and its post were secondary, or even tertiary features, however, for the pit had been dug through a layer of pink clay (layer 4), which might represent part of an earlier pit and which had itself been cut through another layer of clay and gravel surfacing (layer 1a). This in turn overlay what appeared to be the remains of a still older pit (layer 9), although no post pipes survived in assosiation with either of the earlier pits.

Post hole 2 was the only feature to produce two substantially intact post pipes. Sadly, the line of the section cut through these two proved not to be ideal, but a combination of the plan and section in illustrations 6and 7makes the situation relatively clear. The initial post (illus 7, PH 2a, layer 7) was 0.37m wide and 0.89m deep and had been set in a pit at least 0.87m wide and 0.89m deep. It had been packed in place with a mixture of clay and gravel (layer 12). The same clay and gravel surfacing had then been laid around the installed post (layer 1), although this layer was thinner and less well preserved here than at PH1. Later, this feature had been cut by a second post hole (illus 6 and illus 7, PH2b, layer 2), suffering damage in the process which somewhat distorted its shape. The secondary post showed clear signs of a surrounding post pit at the surface (layer 14), but, at a lower level, although the gravel aggregate (layer 11) into which it had been cut was noticable looser in the immediate vicinity of the post, the pit's exact dimentions could not be traced. The post itself, however, was 0.45m wide and 0.66m deep. Once again clay and gravel surfacing had been laid around the installed post (layer 1), but this was in a poor state of presevation. Lastly, PH 2a, had also been damaged by a small hollow (layer 13) filled with brown clay and charcoal which apeared to have been a product of the removal of the second period post.

Post hole 3 was the worst preserved of the three. Here what appeared to be a primary 0.3m wide post pipe (illus 7, PH 3, layer 3) had been all but destroyed by a later pit (Layer 6) which may represent part of a secondary post pit. This, in turn, had been almost totally obliterated by a still later, much larger, but shallower pit (layer 10 and illus 6 feature 3), so that only a c.0.16m segment of the feature survived in plan. No post pit could be traced around the surviving post pipe, although, again, the sub soil (layer 8) was looser in its vicinity, and no trace of the internal clay and gravel surfacing had survived on the line cut by the section.

Thanks to an accident of survival, it is not possible to give exact dimensions for structures founded on these posts in either of the two building periods. For, although PH 2 did produce a post pipe from both structures (illus 6), PH 1 (illus 6) yielded only a second period timber, and PH 3 only a first period post. This means that we have only a long axis side length for the first building period and a diagonal for the second. Nevertheless, as the post pits intersect, both designs seem to have been much the same size, which would put both at about 5.25m (n-s) x 4.25m (e-w). Both structures were, thus, unusually large for free standing Roman timber watch towers. Indeed, so far as the writers are aware, Greenloaning is the largest four post tower ever found. It is certainly much larger than the (3.35m)2 tower found by St.Joseph (1976, 22) at Shielhill South and is even larger than some 12 post towers, such as Johnson's Plain in Cumbria (Woolliscroft and Swain 1991, 25-27). Both structure also seem certain to have been far from perfect rectangles. The angle of the corner at PH 3 is about 76 degrees, suggesting a parallelogram with opposing corners of 76 and 104 degrees.

The stratigraphy between the two buildings was relatively clear cut (illus 7). The corner posts of both periods probably occupied much the same positions. For, although only the north-western corner (illus 6, PH 2) preserved traces of both posts, the secondary post pipe here actually cuts the primary. This is not particularly surprising, since there was little sign of any prolonged interval, or abandonment phase between the two periods. There are no separate demolition pits for the removal of the primary posts, unless layer 4 in PH 1 is interpreted as a filled in demolition feature, and nor are there signs that the posts either burned or rotted in situ. Indeed, the secondary post pits may initially have acted as the primary structure's demolition pits. Moreover, PH1 (illus 7, layers 1 and 1a) preserved evidence that the internal clay and gravel surfacing was repaired once the second period post was installed and, although there was a thin layer of a brown gritty loam (layer 3) between the primary and secondary surfacing, this type of material seems more likely to have come from the digging of the secondary pit then from trample or natural deposition. It would thus seem more likely to represent a demolition/re-building layer rather than a first period occupation or abandonment deposit. No traces of burning could be found in association with the demolition of the primary structure.

As with the first period, there was little evidence for the length of occupation of the secondary structure, since plough damage had removed all traces of occupation deposits from the internal surfacing. There was, however, one possible hint that it may have been more than transient. For layer 4 in PH1 (illus 7) provides tenuous evidence for a third structural period. This feature had clearly been cut through the primary surfacing and through layer 9, which is all that remains at this point of the prosumed primary post pit. But it was itself cut by the final post pit, which might, therefore, as already stated, be tertiary. Since the layer consists of clean, deposited clay, it certainly seems more likely to belong to a construction rather than a demolition feature but, alternatively, as no parallel evidence was found at the other two surviving corners and only two layers of internal surfacing were found, it may be that this particular post simply required re-setting for some reason during the construction of the secondary building or, perhaps less probably, maintenance at some period during its subsequent occupation.

The evidence for the demolition of the secondary structure is less ambiguous. The secondary post pit at PH3 had been all but destroyed by what is presumably a demolition pit (illus 7, layer 10), and the pit shown as feature 4 (illus 6), probably marks the destruction of the north-eastern corner, although it seems somewhat large for the purpose. The latter is certainly in the right place, but time did not allow it to be completely emptied to allow a search for residual features beneath it. Posts 1 and 2b, on the other hand show flared tops which suggest that they were rocked out of position (illus 7).

Once the posts had been removed, the remains of the building appear to have been burned. The entire internal area bore traces of charcoal and scorching, most notably towards the south-eastern corner of Trench 1 and in the area around PH3 and the demolition pits themselves were filled in, largely with sand (illus 7, layers 10, 17 and 18), probably before the burning took place. Such a demolition pattern is normal on the Gask system, although there were no traces of burned daub at Greenloaning to parallel those found at Shielhill North (St.Joseph 1976, 23).

Finally, no trace was found of an internal rampart to parallel those found at a number of the northern single ditched Gask towers, but this is hardly surprising since, with the tower so close to the inner ditch, there would not have been room for one. Another fragment of the same red coarse pottery was found just south of PH2, which again could not be closely dated.

CONCLUSIONS AND PARALLELS

Despite the lack of dating evidence, there seems no reason to doubt that this is a Roman tower of the Gask series but, as ever with this system, its exact date is open to question. So far, outside of the forts, only two firmly datable finds have ever been made on the Gask: a fragment of Flavian mortarium found unstratified at Gask House tower (Robertson 1974, 20f) and a sherd of Flavian Samian ware from a beam slot at the road side enclosure of Cuiltburn, c.600m to the south of Strageath (Woolliscroft forthcoming 1997). The only other possible dating evidence is a piece of what may or may not be first century mortarium, from an unpublished context at Westerton (Hanson and Friell, 1995, 506), and the fact that Blackhill Wood tower, just north of Ardoch, is earlier than a 130 acre temporary camp (the latest of the series of camps around the fort and usually seen as Severan) whose rampart partially overlies its south-eastern defences (St.Joseph 1977, 135-138 and fig 8).

Two scraps of datable pottery, from dubious contexts, is not much with which to date an entire military system at least 38km long, and Greenloaning can only add a further eight, still more questionable, sherds. So, although the present consensus is that the system is Flavian, this cannot yet be regarded as proven beyond reasonable doubt. It should not be forgotten that the forts on the system, whilst Flavian foundations, show clear signs of Antonine re-occupation, and the fortlets, none of which have produced published datings, are of a type that is more often seen in the second century than the first. Timber towers, it is true, do tend to be more of a first than a secondary century phenomenon. But this is not an absolute, for there are a number of known Antonine timber sites, such as Garnhall (Woolliscroft and Hoffmann, forthcoming) on the Antonine Wall. For the moment, we cannot even guarantee that the southern twin ditched Gask towers are contemporary with the single ditched group further north and it might, thus, be prudent to withhold judgement until more evidence can be obtained.

As to the site itself, a number of areas require comment, in particular, the discovery of more than one structural period. It is not, of course, difficult to think of reasons, perhaps an accident or poor initial construction, why one individual site might have required re-building in service whilst its neighbours did not, although fire can be ruled out as the cause of destruction of the primary tower. Greenloaning need not necessarily, therefore, have any connotations for the rest of the system. That said, however, the site is not now as unparalleled as it might have seemed a few years ago, for a closer inspection does reveal signs of re-building elsewhere. At Midgate, for example, a tower was replaced by a fortlet (or vice versa), and the first writer has recently suggested that air photographs show a similar sequence at Raith (Woolliscroft 1993, 302-308 and 297-299). Lastly, at Moss Side, Christison (1901, 29 and fig 7) found beam slots linking the tower's corner posts. These could, of course, be interpreted as bracing to strengthen the tower by locking its uprights together, and/or as foundations for wattle and daub side cladding. But, although the latter would be consistent with the burnt daub found at Shielhill North, no such foundations have been ever found anywhere else on the Gask. Moreover, the excavator specifically says that at least one of the corner post holes had been backfilled to make room for these beams, which means that this tower too may have been replaced, presumably this time by a beam founded structure. Certainly Christison's plan shows the beam slots to be unusually large (c.0.6m wide) to support simple side panels and, although they were not set deep enough to have carried a tower by themselves, they would have made a perfectly adequate foundation for a lower structure, such as a block house.

The near total absence of small finds on the Gask towers has hitherto been taken as denoting a very brief service life. But such finds are notoriously rare on Roman timber towers everywhere, and these growing signs of structural complexity may require us to re-asses this impression. Again, further evidence is needed before we can tell if such re-building was a more general, let alone systematic, phenomenon. But, for now, it would be dangerous to argue a negative from the silence of the other excavated sites on the line, since almost all of the towers so far published have only been dug in a fairly peremptory manner. The phasing at Greenloaning was only confirmed by the removal of parts of the interior surfacing, and the sectioning (sometimes more than once) of all of the features discovered and, so far as the writers are aware, no tower on this system has ever been examined in such detail before.

Another aspect of the site is also without parallel on the Gask: the surfacing of the interior area. But, again, this may be due to the way in which timber towers have generally been dug in the past. For the clay and gravel layer was badly damaged and fragile and could easily have been removed un-noticed during excavations elsewhere. Partial parallels can, however, be found on other sites in northern Britain. For example, at Garnhall the tower stood on a slightly raised rectangular, clay and gravel platform (Woolliscroft and Hoffmann, forthcoming), whilst at Beattock Summit (Maxwell 1976, 36 and fig 2), at least the floor and doorway of the tower itself had been surfaced. But the writers have yet to locate an exact parallel for a tower having its entire central area surfaced, with no distinction being made between the tower itself and its surroundings.

The internal surfacing also produces another possible problem, for the inner ring ditch was cut through it. We have already seen that the two ditches have slightly different centres and so it is possible to argue that the two may not have been dug at the same time, but that the inner ditch is somewhat later. As well as explaining why a ditch should have damaged laid surfacing, such a scenario would also have other advantages. For example, the outer ditch has much the same diameter as the single ditches of the more northerly Gask towers, suggesting that in the double ditched examples it was the inner and not the outer ditch that was the addition. An early single ditched period would also have had room for a northern style internal rampart and, if this was later removed, it might explain the strong geophysical traces for an upcast mound, when the ditches seems to have been largely backfilled with their own original spoil. There is also, however, contradictory evidence. For a start, the outer ditch seems far too slight and to have had too wide an entrance break ever to have stood alone. There are also no traces of either turf or gravelling between the two ditches. Once again, the explanation may come from the tower at Garnhall (Woolliscroft and Hoffmann, forthcoming). For this installation produced (albeit tenuous) evidence that Roman towers may sometimes have had their interiors constructed before the ditches were dug. This seems an eminently sensible arrangement, given that access would then have been very much easier and, if Greenloaning was built in this order, it would be less than surprising if the ditch diggers clipped some internal surfacing. Again, it is best to withhold judgement, for there is not enough evidence either way. Time did not allow an examination of the upcast mound and all traces of stratigraphy between and immediately outside the ditches had been ploughed away. On balance, the ditches do seem more likely to be contemporary, but the writers cannot guarantee it.

Finally, some comment is needed on the three later ditches towards the northern end of the site, all of which were dug after the tower ditch had been backfilled. No particular function could be assigned to the two in Trench 4 (2 and 3), although they may be connected with drainage, but the example at the north end of Trench 6 (ditch 1) appears to run parallel to the Roman road (at a range of 10-11m). The excavation had hoped to run a trench further north at this point, to examine the road and quarry pit and to look for a parallel for a rectangular timber building found immediately outside the entrance at Garnhall. Time pressure caused this part of the work to be cancelled, but conditions elsewhere on the site would suggest that it was anyway unlikely that a stratigraphic link with the road would have survived. Nevertheless, it is at least possible that ditch 1 was dug as a side ditch for a later refurbishment of the road. If so, and if we assume that the Gask towers are in fact Flavian, the ditch would presumably be Antonine or even Severan in date.

University of Manchester
University of Freiburg

ACKNOWLEDGEMENTS

The writers are grateful to the Society of Antiquaries of Scotland for their generous funding and to Blackford Farms Ltd and their Factor, Mr R.D.Baird, for allowing access to the land. We would also like to thank Prof A.S.Robertson and the site's discoverer, Mr G.S.Maxwell, for their advice and encouragement, the RCAHMS for permission to use illus 2, and Dr S.Ramsay of the University of Glasgow for her environmental work.

BIBLIOGRAPHY

Christison, D 1901, 'Excavations Undertaken by the Society of Antiquaries of Scotland of Earthworks Adjoining the "Roman Road" Between Ardoch and Dupplin Perthshire', Proc Soc Antiq Scot, 35, 16-43.

Grew, F O 1981, 'Roman Britain in 1980', Britannia, XII, 313-368.

Hanson, W S & Friell, J P G 1995, 'Westerton: a Roman Watchtower on the Gask Frontier', Proc Soc Antiq Scot, 125, 499-520.

Maxwell, G S 1974, 'The New "Signal Stations" North of Ardoch', Roman North Frontier Seminar, 9, 12-14.

Maxwell, G S 1976, 'A Roman Tower at Beattock Summit, Lanarkshire', Britannia, VII, 33-38.

Robertson, A S 1974, 'Roman "Signal Stations" on the Gask Ridge', Transactions of the Perthshire Society of Natural Science, (Special Issue), 14-29.

St Joseph, J K 1973, 'Air Reconnaissance in Roman Britain 1969-72' JRS, 63, 214-246.

St Joseph, J K 1976, 'Air Reconnaissance of Roman Scotland 1939-75', Glasgow Archaeol J, 4, 1-28.

St Joseph, J K 1977, 'Air Reconnaissance in Roman Britain 1973-76' JRS, 67, 125-161.

Woolliscroft, D J 1993, 'Signalling and the Design of the Gask Ridge System', Proc Soc Antiq Scot, 123, 291-314.

Woolliscroft , D J 1995, 'The Ring Ditch Site Above Middleton Hall, Cumbria: Surface and Resistivity Surveys', Trans of the Cumberland and Westmorland Antiq and Archaeol Soc, 2 ser, 95, 61-71.

Woolliscroft, D J Forthcoming 1997, 'Excavations at Cuiltburn on the Gask Ridge 1995: Interim Report', Manchester Archaeological Bulletin, 10.

Woolliscroft, D J and Hoffmann, B Forthcoming, 'Sites on the Antonine Wall at Garnhall, Cumbernauld'.

Woolliscroft, D J and Swain, S A M 1991, 'The Roman "Signal Tower" at Johnson's Plain, Cumbria', Trans of the Cumberland and Westmorland Antiq and Archaeol Soc, 2 ser, 91, (1991), 19-30.

Background papers

Aerial photographs