Parkneuk – Roman Road

D.J. Woolliscroft and M.H. Davies
with contribution by S.Ramsay

A section of the Roman Road in Parkneuk Wood, Perthshire was partially re-opened to clarify earlier work by Mrs. D. M. Lye and obtain new pollen samples, confirmed the existence of a number of turf layers underlying the road foundation stones, presumably placed to improve the camber and drainage. No dateable finds were recovered, but the pollen analysis suggested an open landscape in Roman times

The Site
Parkneuk Wood stands on the western shoulder of the Gask Ridge c.25m above and 1500m east of the Roman crossing of the River Earn at Strageath. The wood contains a number of Roman antiquities, notably the Gask series tower of Parkneuk (the westernmost on the Ridge), part of the north rampart and ditch of the 130 acre Innerpeffray East temporary camp and, immediately to the north of these two and the subject of this report, one of the best preserved sections of the Roman Gask road anywhere on the system

In 1967 the Perthshire Society of Natural Science under Mr J.K. Thomson and the late Mrs D.M. Lye sectioned the Roman road at NN 915185, c.140m to the west of Parkneuk tower, and followed this in 1969 with a watching brief during the construction of a forestry track through the road c.112m (367′) further east. Sadly, although brief notes did appear in print (DES 1967, 28f, DES 1969, 38 & RNFS 1 1970, 25f), Mrs Lye, who held the field records, died before a full report could be produced. As part of the Roman Gask Project’s program of disseminating previous research on the system, the first writer thus obtained Mr Thomson’s permission to publish this work and was fortunately able to almost totally reconstruct the original records from material in the possession of Mr Thomson himself, Perth Museum and the RCAHMS.

The results were largely straightforward (fig. 1). The road was c.19′ (5.79m) wide and consisted of a 6″ – 9″ (152-228mm) cambered layer of gravel metalling (L’2), overlying what was described as an 18′ (5.48m) wide and 6″ – 8″ (152-203mm) thick layer of, again cambered, compacted red/brown clayey loam (L’3). In its central 12′ (3.66m), a central spine had been formed by embedding fairly closely packed rubble into this layer, which acted both as a foundation and to further build up the camber. There were no stone kerbs and, although the trench was later extended up to 23′ 11″ (7.29m) to the north of the road edge, to give a total trench length of 47′ (11.27m), no sign of a side ditch was encountered, despite the fact that the gleyed sub soil uncovered beneath the road provides evidence for prolonged water logging of the site. The results from the 1969 section to the east were broadly similar, except that here L’3 was missing.

Perhaps the most interesting aspect of the Road, however, was its substructure, for, although this was only studied closely in the 1967 trench, there were signs here that, although much of the road had been built directly onto the original ground surface, the central area had been further built up on a low platform of laid turf, perhaps to improve its camber or drainage. Two small areas of the original trench, designated Rectangle’s A and B, were reopened some months after the main excavation specifically to allow Mr B.M. Shipley of the Macauley Institute for Soil Research, Aberdeen, to study these layers. His opinion was that two rows of laid turves were present (in addition to the natural surface turf), with the bottommost deposited grass side down and the upper layer grass side up. This had produced two black carbonisation streaks: a thicker lower stripe, which also extended over the original surface further out, and a thinner upper streak which directly underlay the road foundation, although there was speculation that this latter layer may have been discontinuous and may only have represented attempts to fill in hollows and produce an even building surface. Samples of this turf were later subjected to a pollen analysis by Dr S.E. Durno

The 1997 Excavations
Once the 1967 records had been assembled a number of problems became apparent which made it difficult to produce a thorough and confident report:

1.The pollen samples taken in 1967 were potentially very important, since they derived from the contemporary ground surface sealed by the Roman road and were thus both firmly datable and representative of the local environment immediately prior to the road’s construction. Unfortunately, although a copy of Dr Durno’s raw pollen data had survived, his interpretation had not and the data sheets were thus sent to Dr S. Ramsay (University of Glasgow) for re-examination. Her analysis of the species represented suggests that the site lay in reasonably open country, but she also suggested that the original samples may only have been fairly hurriedly examined so that only pollen from a few of the most common species present had been detected, where one would also have expected to see many more “trace species” which would have enabled a more detailed evaluation. Dr Durno can hardly be blamed for such a situation, since he seems to have made his analysis un-paid as a personal favour, but as the original samples appear to have vanished and so could not be re-examined, it seemed desirable to try to obtain new ones

2.Although the 1967 plan and section drawings had survived, these did not include the turf substructure layers and Mrs Lye’s correspondence makes it clear that they were not drawn. They also appear not to have been photographed. A rough sketch by Mr Shipley has been found, but neither this nor the interpretation he based upon it appear wholly consistent with the excavators’ written descriptions of these layers. For example, the carbonisation streaks were drawn as two neat, flat and parallel lines, whilst a letter from Lye to Thomson describes them as irregular. This is hardly surprising since Shipley drew his sketch to explain his ideas and it was thus presumably meant as a schematic, rather than an archaeologically true representation. But, more seriously, the written descriptions also make the entire turf deposit sound rather too narrow to fit Shipley’s hypothesis (even allowing for almost two thousand years of settlement and the compression caused by traffic that may have continued to use the road long after the Roman occupation), for the entire central, three turf layer, sequence is said to have been only half an inch (13mm) thick. Once again, therefore, it seemed desirable that these layers should be re-examined

3.Aspects of the 1967 plan and section did not appear to match either each other or the excavation photographs, although the latter were generally taken at rather oblique angles and with the sections somewhat under exposed so that they were difficult to evaluate fully

In an attempt to clarify these difficulties, part of the trench was reopened in September 1997 under the direction of the second writer. Little trace of the 1967 work was visible at the surface, but the trench was relocated without difficulty from the original survey data, which was thus proven accurate. As the original work was conducted in imperial measurements and Scheduled Monuments Consent had been granted on condition that the minimum of further disturbance took place, the re-excavation was obliged to follow suit.

The 1997 trench (fig. 2) was designed to include the 1967 areas: Rectangle A and B, in which the turf substructure had been most fully examined. It was thus 4′ (1.22m) wide (e-w) x 11′ 2″ (3.4m) long (n-s) and located so that its northern end lay 13′ (3.96m) south of the 1967 base line, and 23′ (7m) south of the northern end of the original trench. The backfill was removed and the plan and section redrawn and photographed. The limits of the 1967 excavation, including Rectangles A and B, were clearly evident and, on the whole, the 1997 results matched those of the earlier dig. It was, however, discovered that Rectangle A had extended 1′ (0.3m) to the west of the main 1967 trench and so a 1′ (e-w) x 3′ (0.914m) (n-s) wing was added 1′ from the southern end of the 1997 trench to relocate the original section.

As expected, the modern topsoil (L’1) overlay an 8-20cm thick (slightly narrower than recorded in 1967) layer of gravel metalling, set in an orange/brown loamy matrix (L’2). There were no visible signs of repair or resurfacing, but considerable disturbance by root action may have destroyed the evidence. This overlay the road foundation stones which, as expected, ended c.60cm from the northern edge of the trench. These were set in a compacted pink clay/loam (L’3), as stated in 1967, to form a layer up to 24cm thick (somewhat deeper than described in 1967).

Only Rectangle A was excavated to natural in 1967 and the 1997 dig followed suit. Here at least, however, a complex stratigraphic pattern was uncovered beneath the principle road materials. The natural subsoil (L’10) is a grey, water logged, clayey loam identified by B.M. Shipley as a surface-water gley soil, probably originating as a red boulder clay with loam. The water logged environment had produced reducing conditions so that “The bright coloured ferric iron compounds present are reduced to dull and greyish coloured ferrous iron compounds hence the predominantly grey colour”. As expected, a number of turf layers were encountered between this deposit and the road foundations, and these do conform better to Lye’s written description as “irregular” than to the neat layering of Shipley’s 1967 sketch, but neither is wholly accurate. For the most part, two carbonisation layers (L’s 5 &, 8) were present, but these were neither flat nor parallel and nor was the lower of the two uniformly thicker than the upper, being thinner than it towards the centre of Rectangle A. The actual sequence seems to be as follows: the subsoil (L’10) is overlain by a turf layer (L’9). This is essentially homogenous, and appears to represent the original turf and topsoil, but was interrupted by a narrow (c.8cm) slanting gap filled with material identical to L’10 which resembles a tool cut, but might be an animal burrow or the result of a sapling being extracted when the site was cleared. These layers were completely covered by the lower of the two black lines of carbonised grass (L’8), which continues right across the gap in L’9 just mentioned, although it becomes noticeably thinner at this point. As this gap does seem to represent a genuine breach in the top soil at the time the road was built, Shipley’s theory that the lower line was a double feature made up of both the original surface grass and that from a layer of turves laid grass side down upon it, has been confirmed by a thin section examination by J.C.C. Romans (pers com). Above L’8 are two more turfy deposits: L’6 in the southern half of Rectangle A and L’7 in the north, separated by a brief, c.18cm, area where the upper carbonisation line (L’5) becomes so thick and smudged that the two black streaks become merged. L’s 6 and 7 are almost identical, both to each other and to L’9, except that L’6 contained a number of small yellow patches in its upper half which Shipley’s interpretation would suggest represent areas where air had penetrated the gleyed soil and allowed localised re-oxidation to take place. The fact that these patches only occur in the upper half of L’6 might support the 1967 conclusion that the overall layer made up by L’s 6 and 7 are made up of two layers of laid turves, with the upper set grass side up and the lower grass down.

Contrary to its description in 1967, the upper of the black carbonisation lines (L’5) did not directly underlie the clay/rubble road foundation. Instead a further thin layer of turfy material (L’4) lay between the two, which appeared identical to L’s 6, 7 and 9. This might be interpreted as a layer of trample formed over the laid turf platform during the construction of the road, but the layer was so clean that it seems more likely to represent a fourth layer of turf, laid grass side down. Certainly, this layer is approximately half as thick as the supposedly two turf thick L’s 6 and 7 and there is further corroboration in the fact that at one point, close to the northern end of Rectangle A, L’5 splits in two, briefly dividing L’7 into two roughly equal halves. This might suggest that one turf from the upper half of L’7 was accidentally laid the wrong way up (grass down) and if so, the fact that L’5 splits, rather than simply dipping, would imply that a second grass layer was available, which can only have been provided by L’4. Again contrary to the 1967 description, the total thickness of the turf layers is c.10cm to the original surface and c.29cm to the subsoil (making the road’s total thickness average c.70cm). This is considerably greater than the 1/2 inch (13mm) quoted by Lye and seems far more compatible with the idea of a low, settled, turf stack, although it remains unknown how consistent this turf deposit might be along the line of the road.
Samples were taken from all of the archaeological strata and subjected to pollen analysis by Dr S.Ramsay (see below)

Finally, Mr J.C.C. Romans, also formerly of the Macauley Institute for Soil Research, has kindly made available the results of a “Bright ring” and soil thin section analysis conducted on the 1967 turf samples collected by Mr Shipley. In conjunction with samples taken from the 1969 Forestry Commission section these reveal a similar, although much smaller scale, infield/outfield pattern to that discovered by Mr Romans at Strageath (Romans Pers Com and Romans and Robertson 1983, 139) using the same technique, with the 1967 section in the outfield area and the 1969 section in the infield, where the presence of some scattered soil pores with oriented clay suggested shallow, possibly hoe, cultivation in the pre-Roman period. This agricultural pattern may have been in force on the site 800 years or more before the construction of the Roman road. The presence of unbroken bright rings in the original surface material, however, and a thin undamaged layer of wind blown diatoms at the original surface, suggested that the site had been used for grazing rather than cultivation for quite some time, perhaps 200 years + 100 before the road was built, which matches well with the results of the pollen analysis (see below)

The width of the road, at 19′ (5.79m) is slightly small by the standards of the Gask as a whole, although not dramatically so. Young (PSAN 1898, 99) reported it as 20′ (6.09m) wide a little to the east in 1897, as does the New Statistical Account (p282) for the Parish of Gask. A section excavated by St.Joseph (reported in Glendinning & Dunwell forthcoming) at Blackhill Wood, just to the north of Ardoch (in 1974), found it to be around c.24′ (7.4m) wide, as did Pennant (reported in Christison 1898, 429) who measured the road as a surface feature at the eastern end of the Gask Ridge, near Dupplin. A 1971 watching brief of a section cut for drainage purposes at Kirkhill, near the centre of the Gask Ridge (DES 1971, 57), found it to be 25′ (7.61m) wide and Christison (1898, 432) reported it as 26′ (7.92m) wide as it passed Ardoch.

As regards the road’s construction, a similar turf substructure may also have been present, in the 1969 Forestry Commission section 367′ (112m) to the east of the 1967/97 trench, although this is less than certain as the road foundation is simply said to have been “laid on turf” (DES 1969, 38). But nothing of the kind was reported at any of the other three sections through the Gask road for which information has survived: Blackhill Wood, Kirkhill and a series of rather poorly located sections dug by Young between Raith and Gask House in 1897 (PSAN 1898, 99), although a single carbonisation line was found beneath the road at Kirkhill which presumably represented the original ground surface. The turf work may, thus have been a purely localised response to waterlogged ground in the immediate vicinity, by lifting the road foundation slightly above its surroundings. Similar turf work may have existed occasionally beneath Roman roads in similar conditions elsewhere in Britain, notably on the road between Drumburgh and Kirkbride in Cumbria, which is built up on a low mound of peat sods, where it crosses moss country (Bellhouse 1952, 41ff and pers com) and possible on parts of Dere Street in Northumberland, although here absolute proof is lacking (Snape and Speak 1995, 26ff).

The absence of side ditches is far from unusual on the Gask. On the many occasions the road has appeared on air photographs it is usually flanked by roughly parallel lines of quarry pits, but almost never by ditches. St.Joseph’s section at Blackhill Wood found none, despite extending more than the road’s own width to the south and nor did the 1969 Forestry Commission section in Parkneuk Wood. Indeed, the only work so far that has produced ditches is Young’s 1897 section at Gask House (PSAN 1898, 206), which found them to be 36′ (10.97m) apart, and the watching brief at Kirkhill, where they are reported as being “V” shaped in section and 2′ (61cm) deep. But as both these observations come from the Gask Ridge itself, where much of the road has remained in use to the present day, there is no guarantee that these are original features. As already stated, however, the fact that in Parkneuk Wood the road may have been passing through damp ground might have been expected to encourage the provision of drainage and, for exactly this reason, the forestry road whose construction produced the 1969 section is flanked by significant ditches. It is possible that the 1967 trench simply wasn’t long enough to intercept a ditch running some way from the road, but this does seem unlikely. For the trench extended 23′ 11″ (7.29m) from the road whereas Young’s 1897 ditches ran only 8′ (2.44m) from the kerbs, whilst the Kirkhill ditches are simply said to have been “close to the road”. One possibility is that the turf stripped areas which must have resulted from the central turf platform might have functioned as a shallow ditch as well, perhaps, as a marking out line for the construction of the road itself. The laying of three courses of turf on the original ground surface would obviously have necessitated the stripping of an area three times the size of that covered, in other words, an area 3 x c.12′ = c.36′ (c.11m) wide, or a band c.5.5m wide on either side of the road. Such an operation would obviously have only produced a hollow a few centimetres deep, but this may have had some drainage effect and may not have been detected by the 1967 dig. I should be noted, however that the pollen analysis (below) contains suggestions that not all of the turf used may have come from the same source and so it is possible that the material may not have come from the site.

As to the structure of the road proper, a cambered surface of rammed gravel laid on a foundation bed of larger stones is hardly unusual for a Roman road and there was a plentiful supply of both available close by in the bed of the River Earn. Indeed almost all of the stones used in the road’s construction appear water worn. Details in construction vary considerably along the line of the Gask, however. For example, the red clay into which the 1967 foundation stones were set, was already absent in the 1969 section only 112m to the east and has not been reported in any other known section. But the stone bottoming itself is a normal feature, although St.Joseph at Blackhill Wood, reports that the road there had “no special bottoming layer” but consisted entirely of compacted gravel with a few larger cobbles. Elsewhere, however, it is the gravel layer that is missing. The New Statistical Account for Gask has the road consist simply of “rough stones, closely laid together”. Walker and Maxwell report its absence at Kirkhill, and Young is uncertain, saying: “large stones were found all over the road but I do not think the centre had ever been paved but had been of gravel”. This might suggest that in any given spot the Romans simply used whatever materials came to hand, rather than following any set design, but again these reports come from areas where the road has remained in use into modern times and it is perfectly possible that the original Roman surfacing may simply have been worn away. CertainlyChristison (1898, 432) at Ardoch revealed a fine gravel surface which he described as being “tightly compacted…..slightly arched, free from ruts and as smooth as a cyclist could wish”.

Finally, Thomson and Lye specifically say that the road showed no sign of kerb stones in either of the Parkneuk Wood sections and St.Joseph also found none at Blackhill Wood. Again none are mentioned at Kirkhill and Ardoch, but Young does report his road to have “very large stones” at its edges, so there may have been kerbs in places.

The variations within the Gask road make it possible that different sections were built by different building teams, something which also seems to be true of the towers. As yet, however, not enough data is available to prove this conclusively, let alone to suggest where the construction sectors might begin and end.



Susan Ramsay


During the re-excavation of the Roman Road at Parkneuk Wood, the excavators sampled many of the contexts for subsequent pollen analysis, in particular the turf layers and origi

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