Shielhill North

Including Excavations By The Late Professor J.K. St.Joseph


The Site

Shielhill North was first recorded by Thomas Pennant in a brief account of a trip along the Roman road from Kaims Castle to Ardoch in 1772 (Pennant 1774, 100). At the time, the site was still visible as a surface feature and Pennant describes it as “a small round area, like those on Gaskmoor, but considerably stronger, being surrounded by not fewer than three fosses”. Since Pennant’s day, however, all surface trace has been obliterated and the site was not re-discovered until 1972, when J.K. St.Joseph took two remarkably clear air photographs (CUCAP BKE-50 & 51) which provided a view of itsditch system, the Roman road, which lies just to the south, and even of four internal post holes (fig 1). He immediately identified it as a Roman tower and as part of the Gask frontier (St.Joseph 1973, 218 & 1976, 22).

The site lies at NN 8561 1220 on a small flat topped knoll, just above the 190m contour, near the top of the hill that carries Kaims Castle fortlet at its summit. It also lies c. 90m to the south-west of a small burn which should have provided a reasonably reliable water source. The site is an excellent look out position for, as well as having a good command of its immediate vicinity, it enjoys superb longer range views in every direction except the north and north-west, where it still faces rising ground. Its view to the south is particularly impressive, stretching for many miles right across Strathallan to the Ochil hills beyond, and taking in all of the known Gask installations to its south (i.e. from Shielhill South tower to the fortlet at Glenbank), including the fort of Ardoch, which is in full view at a range of c. 2.75km. Even to the north, the site would have been intervisible with Kaims Castle from the full likely height of a Roman timber tower and, as Kaims is in turn intervisible with the fort of Strageath, the site would have been easily integrated into any frontier signalling system (Woolliscroft 1993, 297 & illus 3).

The Air Photographic Evidence

The Tower
Fig 1 shows a composite transcript of the two St.Joseph discovery air photographs, and a later image taken by the RCAHMS (neg PT/11100). These show that the tower, like all of the Gask installations south of Kaims Castle, had a double ditch (those to the north had only one) and not the three reported by Pennant. The inner ditch is sub-rectangular in plan and c.16m in external diameter, surrounding an internal area of around (13m)2. The outer ditch is c. 24m in diameter and is rather more circular in plan (if still somewhat irregular), which leaves an inter-ditch space of between 2m and a little over 3m.

Both ditches show a single entrance break oriented to the southeast, towards the Roman road. That of the inner ditch is c.3.3m wide, but the outer ditch only covers c.290o of the full site circumference so that, although the southern butt end forms a neat line with that of the inner ditch, to the north of the entrance the outer ditch stops well short of the inner to leave a break of c.11m. Unusually, however, the outer ditch appears to make an outward turn of approximately 137o on this side of the entrance, to form an offset which runs for about another 3.5m before forming a butt end, a feature which has not been recorded anywhere else on the Gask. The post pits for the tower itself would suggest a structure of approximately 4.5m x 3.5m, which would give it an internal area of approximately 15.75m2.

The Road
The Roman road can be followed on air photographs running north-eastwards from NN 8541 1198 to NN 8575 1234. Unusually for the Gask, clear traces of side ditches are visible, albeit intermittently, as are the ubiquitous quarry pits which are the usual mark of the road on this system. The latter appear in distinct groupings at about 50m intervals, most (but not all) of which lie on the northern side of the road. One of these groups lies immediately to the south and east of the tower and it is noteworthy that one fairly large pit is located directly in front of its entrance.

For the most part the road is more or less what one would expect on the Gask at about 7m wide between its ditches and, throughout most of this sector, it follows the Ordnance Survey line almost exactly. It does, however, show one highly unusual feature in that at NN 8559 1215, c. 40m southwest of the tower, it swerves a few metres to the south and forks. One line then continues along much the same heading. It gradually returns to the O.S. line and passes just to the south of the tower entrance, coming to within c. 9.7m of the outer ditch at its closest approach. Immediately east of the tower, however, this branch narrows abruptly (with an inturn of its southern side) to about 4m within the side ditches. It crosses the small stream, mentioned above, immediately south of the modern A822 and there are surface and air photographic indications that it may have done so on a small embankment, although without excavation it is difficult to separate the ancient works from the modern road crossing. It then continues on the same course, apparently heading for Kaims Castle.

The second branch appears to maintain the original width throughout and follows a slightly more southerly track to reach a maximum separation from the northern branch of c.28m. With current air photographic evidence, it can only be traced as far as the stream, and it is possible that it may indeed end there, but there are slight indications that it is starting to turn northwards again, as the Air indications end, as if it was intending to loop back to the northern branch.

The Geophysical Survey
In 1996, the Roman Gask Project conducted a resistivity survey of the site to act as a control to settle differences between the air photographic evidence and the St.Joseph excavation data discussed below. The results, as shown by Fig 2, correspond well with the picture gained from the air, for they provide a clear trace of a sub-rectangular inner ditch and almost exactly confirm the dimensions given above. The outer ditch shows far less clearly, however, which would suggest that it may have been rather less substantial. In particular, the offset at the northern outer ditch entrance could not be detected and the ditch’s overall shape cannot be plotted with any confidence. Again, there was no sign of Pennant’s third ditch, which thus seems unlikely to have existed, but intermittent areas of high resistance immediately outside the ditch system may represent the ploughed out remains of an upcast mound.

St.Joseph’s Excavations
In October 1972, shortly after re-discovering the site, St.Joseph undertook a limited program of trenching. The work was accompanied by rather larger scale excavations at the neighbouring sites of Shielhill South and Blackhill Wood, which were at the time the only other members of the southern double ditched tower group known, but, although some notes appeared in print shortly thereafter (St.Joseph 1973, 218 & 1976, 22), this work is only now being fully published (for Shielhill South: Woolliscroft & Hoffmann 1998, for Blackhill Wood: Glendinning & Dunwell forthcoming). The excavation consisted of just two trenches. The first was a long (22m x 0.8m) slot running from east to west across the site (fig 3). This sectioned both ditches on the western side of the tower and explored one of the four internal post holes, before bending a little towards the south to reveal the inner ditch on the eastern side, this time, apparently, only in plan. The second was a much smaller (3.1m x 0.8m) cut and exposed just the inner ditch on the southern side of the tower, again, apparently, only in plan. No attempt was made to search for Pennant’s third ditch outside the two visible through air photography. St.Joseph’s drawn plan (fig 3) is probably best regarded as a schematic representation of the site (albeit to scale) to show its essential anatomy and the relationship of the excavation trenches to it, rather than as a strictly accurate representation. Both ditches are shown as perfectly circular which, as St.Joseph’s own air photographs had demonstrated, is clearly not the case, and the tower is shown as a perfect (3.34m) square, which again contradicts the Air evidence. This was St.Joseph’s usual practice on the Gask, for the ring ditches at Shielhill South (Woolliscroft & Hoffmann 1998, illus 2) and Blackhill Wood (Dunwell & Glendinning, forthcoming) were also shown as perfect circles even though more recent excavations have shown that again both are clearly sub-rectangular or sub-circular. Shielhill North had produced by far the best Air evidence of the three towers studied and so this approach might be thought to be rather more surprising here, but it is probably justified given the fact that St.Joseph’s excavations were on such a small scale when compared to his work at the other two sites.

Nevertheless, despite the limitations of the site plan and a complete lack of datable finds, valuable data was produced by the excavation. The section (fig 4) produced two shallow “V” shaped ditches, with no “ankle breaker” bottom slots. In confirmation of the resistivity data, the outer ditch was extremely slight, at 1.31m wide. The drawn section shows it as being just 0.2m deep, below the modern plough soil, but St.Joseph’s notes record a maximum depth of 0.37m so the bottom may have varied within the trench. The inner ditch was slightly more substantial at 1.9m wide and 0.43m deep, and the inter-ditch spacing, as measured between the ditch lips, was 2.4m. No signs of palisading were uncovered either inside, outside or between the ditches. St.Joseph’s notes offer no detailed description of the outer ditch fill, but fig 4 marks it as being filled with peat. The inner ditch contained a number of small boulders at its bottom and the ditch fills were otherwise described in St.Joseph’s field notes as follows:

“The inner ditch revealed up to 10″ (254mm) of ash, burnt clay and possibly daub, with a whitish, even textured silty earth (possibly turf) filling the rest of the ditch, rather as if the fill had been thrown in in shovelfuls. There was no silt beneath this. The fill for the last 2′ (0.6m) on the interior side of the ditch consisted of more contaminated burnt clay/ash, etc as if this represented a throwing in of scrapings or tailings from the backfilling process. This mixed fill overlaps the inner lip, suggesting that it was thrown in from the interior of the enclosure, and was overlain by a band of peaty soil of a dull matte colour, possibly an in situ natural growth of peat”

In the interior, the north-eastern post pit was drawn in section and recorded in plan (Fig 5) at a depth of 0.37m below its surviving top, at which point the post pipe itself became visible, showing that the tower here had stood on a rectangular sectioned timber of 0.3 x 0.27m. A 0.44m depth of post pipe had survived, suggesting that the timber had originally extended to a full depth of at least 0.72m below the modern plough soil. The uppermost 0.28m had, however, been destroyed by a pit, presumably, a demolition feature dug to allow the timber to be removed when the site was abandoned. The post pit measured 0.97m x 0.8m at the depth at which it was planned. The post had been chocked with a fill of clean gravel (Layer a), rather than with stones as might have been expected. The post pipe itself had filled with layers of fine silty earth (L’s b and c), whilst the demolition pit was filled with gravel similar to that in L’a but mixed with “dark earth”, charcoal and daub specks (L’ d). No other internal structures were located except for a narrow linear feature, which was probably a modern land drain and, in particular, St.Joseph’s records make no mention of any trace of metalling in the interior, or of an internal rampart. No attempt was made to look for a gate structure at the entrance, but these are anyway, so far, universally absent on the Gask towers.


The Tower
Despite the absence of dating evidence there seems no reason to doubt St.Joseph’s original identification of the site as one of the Roman Gask series towers. As such, it is broadly consistent with the other three known double ditched towers at the southern end of the line: Shielhill South (Woolliscroft & Hoffmann 1998), Blackhill Wood (Glendinning & Dunwell, forthcoming) and Greenloaning (Woolliscroft & Hoffmann 1997). For example, the slight nature of the ditches can now be paralleled at all four southern towers. The average inner ditch width and depth amongst these sites is 1.83m x 0.51m, as compared to Shielhill North’s 1.9m x 0.43, whilst the group’s outer ditches average 1.67m wide x 0.36m deep, as opposed to Shielhill North’s 1.31m x 0.2m. The closest parallel, not surprisingly, is the site’s nearest neighbour, Shielhill South, whose inner and outer ditches average 1.16m x 0.47m and 0.99 x 0.36m respectively.

The overall diameters of the site’s two ditches are also comparable with the rest of the southern tower group, as is the shallow “V” shaped profile and the frequent (but not universal) lack of an “ankle breaker” sump. St.Joseph’s single ditch section does not show the unusual flared top profile recorded at all three of the other southern towers (e.g. Woolliscroft & Hoffmann 1997, 569, illus 5) but, as this feature appears only somewhat intermittently within the other sites’ ditch circuits, it remains possible that further work might find it at Shielhill North as well. The asymmetrical outer ditch entrance break can be paralleled at Greenloaning (Woolliscroft & Hoffmann 1997, 568, illus 4), although no comparable offset was uncovered there. The apparent absence of silt in the ditches can also be seen elsewhere on the Gask and need not necessarily imply an unusually brief occupation, since the fortlet of Midgate appears to have been abandoned part way through a ditch re-cut (Woolliscroft 1993, 305ff). The lack of palisading is universal on the Gask, although palisades are know from Roman towers elsewhere (e.g. Woolliscroft & Swain 1991), and the markedly sub-recangular/sub-circular site plan has now been proved on all of the southern Gask tower group except Greenloaning, which does appear to be at least roughly circular.

In the interior, the shape and dimensions of the tower itself are also comparable with the rest of the southern group and, although square towers are known elsewhere on the system, all four double ditched sites are now known to have rectangular towers. The tower’s ground area of c.15.75m2 compares well with the group’s average of 15.8m2 and, although this figure does conceal a fairly broad range for the group, from 12m2 at Shielhill South, to c.22.6m2 at Greenloaning (one of the largest Roman timber towers known), these sites are almost universally larger than the towers on the single ditched northern Gask sites, which average just 9.74m2.

St.Joseph records no signs of metalling in the interior and, as this contrasts markedly with the other three towers in the southern sector, it might be a hint that a certain amount of the original surface has been ploughed away. Given the similarity of the ditch depths at all four towers, however, and the fact that Shielhill North’s tower post holes are slightly deeper than the average for the group, the loss has probably not been significant. Whatever the case, some additional evidence for the interior can probably be inferred from the ditch contents. For example, St.Joseph does not record any signs of an internal turf rampart to parallel those found further north on the towers of the Gask Ridge proper (Christison 1901 and Robertson 1973). But the probable turf found in what is clearly a demolition deposit in the inner ditch fill may well have come from such a structure, all in situ trace of which may have been removed by a combination of Roman demolition and modern ploughing. In the past, (e.g. Woolliscroft & Hoffmann 1997, 573) the writer has tended to doubt whether the small interiors of the double ditched southern Gask towers would have had enough room to contain such ramparts, despite the presence of a certain amount of, albeit less clearly stratified, turf in the inner ditch fills at Greenloaning and Shielhill South, but the recent discovery of a rampart at Blackhill Wood (Glendinning & Dunwell forthcoming) has obviously rendered this view untenable. Another objection to internal ramparts, the fact that the southern corner posts of the tower at Greenloaning were set so close to the inner ditch lip that again there would not have been room for a rampart (Woolliscroft & Hoffmann 1997, 571, illus 6), has also now been answered. For the northernmost and most recently excavated Gask site, Huntingtower (Woolliscroft forthcoming), proved to have a similar configuration, but with clear surviving evidence that the tower posts had been set into the rampart body, thus removing the space problem. This is a rare configuration in Roman towers, but it is common enough in forts and, given the evidence from Shielhill North, it is probably not unlikely now that all four southern group towers had ramparts.

Likewise, the presence of daub in both the ditch fill and the demolition layers of the excavated post pit (which might suggest that the tower was fitted with wattle and daub side cladding) is also consistent with the rest of the southern group (and especially with Shielhill South which yielded hazel twigs which might have derived from wattling (Woolliscroft & Hoffmann 1998 forthcoming)), as is the evidence for systematic demolition and the in situ burning of demolition materials. Indeed, the entire Gask line shows a similar picture of careful demolition (e.g. Woolliscroft 1993, 307 and Woolliscroft & Hoffmann 1997, 573) with or without signs of burning.

Finally, recent excavations at the four Gask towers of Greenloaning (Woolliscroft & Hoffmann 1997), Shielhill South (Woolliscroft & Hoffmann 1998, forthcoming), Huntingtower (Woolliscroft forthcoming) and Blackhill Wood (Glendinning & Dunwell forthcoming), have all produced signs of multiple structural phases and, occasionally of ditch re-cutting. Work in the 1980’s at the tower of Westerton (Hanson & Friell 1995, 504ff) also produced (albeit more equivocal) evidence for the same process and it has begun to seem likely that the Gask Frontier may have remained in use for appreciably longer than the brief, season or two, span that had previously been envisaged (e.g. Breeze 1982, 65). St.Joseph’s Shielhill North excavations record evidence for only one tower and ditch phase, however, but for a number of reasons this cannot, by itself, be regarded as conclusive. For a start, it does have to be said that the same excavator did not recognise evidence for phasing at both Shielhill South and Blackhill Wood, which has now been deciphered by more extensive work. Moreover, St.Joseph’s excavation at shielhill North took in only a small sample of the site and it is perfectly possible, in this one ditch section and post hole, that secondary features may have completely destroyed any earlier evidence. Under the circumstances, therefore, whilst bearing in mind the possibility that the site really may have had just this single structural phase, it might be best to allow the matter to remain open until more extensive work can be conducted.

The Road
The Air evidence for the Roman road at Shielhill North raises two matters, both of which ultimately concern dating. The first is the fork immediately to the southwest of the tower. This occurs as the road nears a small stream crossing and it is, of course, far from rare for such crossings to migrate over time, especially where a Roman road has remained in use into later times. For these may have lacked the ability and/or inclination to maintain bridges, culverts or fords and so sought easier ground. Alternatively, riverine changes, differing usage patterns or damage caused by simple weight of traffic may have rendered the original crossing unsuitable. On this occasion, however, both prongs of the fork do seem likely to be Roman, since both appear to be of similar size (at least initially) and construction and, in particular, both show associated quarry pits, which form such a strong diagnostic feature of the Roman road on the Gask. Of course, this does not necessarily mean that both branches are exactly contemporary and there may well have been modifications made either within or between the three main Roman incursions into the area but, even so, the branch does appear more likely to be a Roman, rather than a post-Roman feature. Certainly, there is one, albeit tenuous, piece of evidence that neither branch remained in use into at least relatively modern times. For, although Pennant was able to follow the line in 1772, and although the road was still marked, albeit as a dotted line, on Stobie’s map of Perthshire in 1783, the only gate in the dry stone field wall at the eastern side of the tower field lies well to the south of both road branches, whereas elsewhere on the Gask the field gates often lie right on the road, reflecting its frequent later use as a field track. This is only of moderate help, however, and it quite possible that the road may have stayed in use for centuries before the present field system came into being.

It remains far from proven that the southern branch does loop all the way back to the northern, as suggested above, and the answer to this question must await excavation or better air photographic coverage. If it does, however, it might be thought to form a by-pass loop. This is a common enough feature at Roman forts, notably on the nearby Antonine Wall (e.g.MacDonald 1934, figs 17, 31 and 42) but, as far as the writer is aware, it is unprecedented at a tower, where it would seem to serve little purpose, since the main road line at the towers does not pass through the installation as it does at the forts, which means that there is really nothing to bypass.

The second issue raised by the road is the presence of what are assumed to be quarry pits (although St.Joseph’s excavation records do not mention the nature of the subsoil) immediately outside the tower’s outer ring ditch, including one that actually blocks the entrance. It is perfectly possible that this pit may have been backfilled, perhaps using spoil dug from the ditches, although this might have led to subsidence problems and would require excavation to test. But, at the very least, the positioning of this pit does seem rather odd for if it was open during the tower’s occupation, it would have been inconvenient and would certainly have made the site’s interior inaccessible to wheeled traffic. An alternative might be to suggest that this particular feature is actually a demolition pit dug to conceal abandoned materials from an enemy, albeit such a feature would be unprecedented on a Gask tower. But, although this could be tested by excavation, it would appear to require a certain amount of special pleading given the fact that the pit seems to form such a natural part of a whole series of what do seem to be perfectly normal quarry pits along the road side, especially as elsewhere on the site so much obvious demolition material was found dumped indiscriminately into the ditches. Of course the pit might have been retained as an additional defence, but it does raise the possibility that the road, at least in its final well surfaced form, might not have been contemporary with the towers. If so, the pit’s position would suggest that the road as we have it might have been significantly later than the tower occupation for, if the road was built first, it would have been perfectly easy to build the tower a few metres further to the west to avoid any pre-existing pits. Alternatively, had the tower been in occupation, or at least planned, at the time the final version of the road was constructed, it would have been a simple matter to have sited the pits slightly differently. Indeed, it might even be argued that if the tower and road had been exactly contemporary there would have been no need for separate quarry pits at this point because the act of digging the ring ditches should have yielded adequate material. It might thus be worth speculating whether the road might be Antonine or even Severan. Against this hypothesis, however, is the fact that there is no evidence that any of the quarry pits cut or were cut by the tower ditches, although it is worth noting that, at Greenloaning, the tower’s outer ditch was cut by a running feature which may have been a side ditch for the road (Woolliscroft & Hoffmann 1997, 570 & 576). There is also the fact that, elsewhere, quarry pits have been found underlying structures built very shortly afterwards, probably as part of the same scheme, for example beneath the Antonine Wall expansion of Bonnyside East (Steer 1957, 164ff). Indeed, the wide outer ditch entrance break could be interpreted as an attempt to avoid a number of quarry pits, which might thus pre-date the tower. Moreover, the possible by-pass loop, if it existed, seems even more difficult to explain in the absence of the tower. It could be argued that the narrow stretch of road to the east of the site might represent an original Flavian track, with the broader road of the southern branch being later, but this would not explain why the south-westernmost section of the northern branch was also built at the same broad gauge, unless one was to argue that this might show that the towers were returned to use in a later period, perhaps manned from Ardoch (which lies to the southwest), and for this there is absolutely no evidence. It is also worth noting that no sign of an earlier trackway has ever been found in any section cut through the Gask road, although admittedly it is possible that such a feature could have been destroyed by later road building. At the very least, however, this situation does act to highlight the fact that, at present, we still have no dating evidence whatever for the construction and history of the Roman Gask road and so a Flavian date contemporary with the towers cannot simply be taken for granted.

The writer would like to thank Ardoch Farming Co Ltd, their then factor, Mr R.D. Baird, and their tenant Mrs M. Rimmer for allowing the resistivity survey. Thanks are also due to Mr G.S. Maxwell and the RCAHMS for allowing access to Prof St.Joseph’s field records and for providing copies of his drawings. The Roman Gask Project is sponsored by The Perth and Kinross Heritage Trust.


Breeze, D J 1982 The Northern Frontiers of Roman Scotland, London.

Christison, D J 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.

Glendinning, B D and Dunwell, A J Forthcoming “Excavations of the Gask Frontier Tower and Temporary Camp at Blackhill Wood, Ardoch, Perthshire & Kinross”, Britannia.

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

Macdonald, G 1934 The Roman Wall in Scotland, (2nd ed), Oxford.

Pennant, T A Tour in Scotland: 1772, 1774

Robertson, A S 1973 “Roman ‘Signal Stations’ on the Gask Ridge”, Transactions of the Perthshire Society of Natural Science, (Special Issue), 14-29.

St.Joseph, J K “Air Reconnaissance in Britain, 1969-1972”, JRS, 63, 214-46.

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

Steer, K A 1957 “The Nature and Purpose of the Expansions on the Antonine Wall”, Proc Soc Antiq Scot, 90, 161-169.

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

Woolliscroft, D J forthcoming “The Roman Gask Series Tower at West Mains of Huntingtower, Perth & Kinross” Proc Soc Antiq Scot.

Woolliscroft, D J & Hoffmann, B 1997 “The Roman Gask System Tower at Greenloaning, Perth & Kinross”, Proc Soc Antiq Scot, 127, 563-576.

Woolliscroft D J & Hoffmann, B 1998 “The Roman Gask Series Tower at Shielhill South, Perth & Kinross. Excavations in 1973 and 1996”, Proc Soc Antiq Scot, 128, forthcoming.

Woolliscroft, D J and Swain, S A M 1991 “The Roman ‘Signal’ Tower at Johnson’s Plain Cumbria”, C&W(2), 91, 19-30

A long term research project to study the Romans north of the Antonine Wall