Fieldwork
2007 saw the Project temporarily easing back on our excavation program whilst we caught up on the vital publication of past work. But this certainly did not mean a moratorium, and we still managed, amongst other things, an excavation and our largest geophysical survey to date.

Excavation: Innerpeffray
In past seasons, the Project has conducted two different excavations on the Roman Gask road, just east of the point where the frontier crosses the Earn in the vicinity of Innerpeffray. One of these was in Parkneuk Wood, near the Gask tower of the same name, and the other on the huge Roman cutting to the south of Innerpeffray Library. We have, however, long had our eye on another point on the road, roughly half way between these two points.

As the road runs up towards the Gask Ridge proper, it passes two large temporary camps (fig 1), both of which belong to long distance chains. The first is the 63 acre camp of Innerpeffray West, which sits directly across the river from Strageath fort and which, like others in the same series, appears to have an annexe. The second is the still larger, 130 acre camp of Innerpeffray East. The frontier road runs just to the north of the latter, on a near parallel course, and a well preserved length of both it and the camp defences can still be seen in Parkneuk Wood around our old excavation site. It has long been known from air photographic evidence that the road passed through the 63 acre camp, but instead of entering and leaving at a gate, it runs straight through the defences on either side of the north-east corner, suggesting that the two can never have operated together. Neither feature has produced firm dating evidence, even though the road has been studied many times. Nevertheless, the construction sequence had appeared all but certain. It seemed obvious that the road would have been built first, because of its relationship with the late 1st century frontier works, whilst the 63 acre camp group has been thought to belong to the early 3rd century Severan invasion, or possibly to the mid 2nd century Antonine occupation. Nevertheless, it seemed well worthwhile to check, and the Project conducted an excavation and geophysical survey in August.

The results served to confirm the Project's reputation for adding cats to pigeons. The camp ditch was found exactly where our own geophysical work and a RCAHMS air photo rectification said it would be. It proved to be fairly substantial at 3.78m wide and 1.5m deep, and showed signs of having been re-cut or cleaned out, which might suggest more than one period of use, or at least a fairly lengthy single occupation. This degree of occupation complexity was itself somewhat unexpected, although parallels have been emerging on other Scottish temporary camps of late, but the big surprise was the camp's relationship with the road. We had expected the camp ditch to be cut through it, but what we found was the very opposite: for the road overlay the filled in ditch. It thus clearly post-dated the camp, possibly by a significant margin, because the ditch had had time to silt to a considerable depth after its re-cut, before being deliberately backfilled to provide the road builders with a level construction surface. This backfill later settled, causing subsidence of the road's southern edge, but this may have happened after the Roman period, for there were no signs of attempts at repair.

This was a wholly unexpected sequence, which has obvious dating implications, because the assumptions just mentioned are clearly no longer tenable. No datable finds were recovered to provide a clear-cut new time line, but a number of possible scenarios present themselves. The first is that the 63 acre camp series might, after all, date to the 1st century and represent the campaigning phase of the first Roman invasion, whilst the road forms part of later consolidation activities, which included the construction of the Gask frontier itself. This does seem somewhat unlikely, however, for although these camps are undated, work on the nearby example at the Gask site of Ardoch has shown that it came late in the sequence of several overlapping camps to the north of the Roman fort. A second possibility is that the road is Severan, or even post Roman, but this too seems improbable: especially the latter. Unlike the Flavian and Antonine periods, none of the Gask forts are known to have been occupied in Severan times. Instead, the main centre lay away from the Gask, at Carpow on the Tay estuary. There was thus nowhere for the road to be going and it thus seems improbable that a well constructed, all weather highway, with cuttings, embankments and other major engineering works, would have been built here at this time when it had not been earlier. Likewise, few engineered roads of this calibre were built through this region in post-Roman times before the 18th Century; a period from which good documentation survives that would certainly record construction on the scale of the nearby Innerpeffray cutting. Moreover detailed estate maps exist which show that the road was out of use and invisible by the early 19th century, when the lane to Innerpeffray Library was already following its current line, c. 130m further north. The road's structure on the site was slightly unusual in that it consisted only of a rammed gravel and cobble layer, whereas elsewhere on the line it was founded on a layer of larger stones and sometimes a turf mound or agger. Nevertheless, there are parallels on the Gask, including the road bed found in the nearby Innerpeffray cutting, and that passing the aforementioned camps at Ardoch.

Perhaps the most likely scenario is that the camp and the Gask road as we see it today, are both Antonine. It still seems probable that the 1st century frontier would have had a road of some kind, and the fact that the entrances of its towers and fortlets are all oriented on the known road would suggest that it followed a similar line. But it is possible that this primary road was little more than a light metalled track that was only replaced by the current superbly engineered line during the mid 2nd century, when the three Gask forts of Ardoch, Strageath and Bertha, came back into use as outposts to the Antonine Wall; possibly for longer than their original occupation. This would have allowed a greater emphasis on wheeled traffic, especially as a number of cuttings and embankments were built to ease the gradients, whereas the 1st century phase may have been designed more for foot and animal traffic (including pack animals). If so, the camp might represent the Antonine re-invasion of the area, whilst the road formed part of the subsequent consolidation. None of this can yet be proven, but the dating of the known Gask road has certainly been questioned before. Indeed the Gask Project pointed out some years ago that a road quarry pit dug through the entrance of the Gask tower of Shielhill North seemed incongruous if the two were supposed to be contemporary. Moreover, this dating would explain two phenomena that have long presented something of a mystery. The first is the fact that the road appears to end at Bertha on the Tay, rather than running on to the legionary fortress of Inchtuthil further north. Inchtuthil was much the most important site built in the Flavian occupation and, for both logistic and strategic reasons, it has always been hard to understand why the road did not go there. The site was not reoccupied in the Antonine period, however, and there would thus have been no reason for a 2nd century road to include a site that had been abandoned for over half a century. Secondly, the apparent lack of a network of similarly well constructed roads to link the Gask with the so called 'Glenblocker' forts to its north has also posed something of a puzzle. Again, however, these had long been abandoned by the Antonine period and only one, Dalginross, has yet shown evidence of reoccupation. Properly constructed stub roads can be seen emerging from some of the gates of these sites (and some of the Strathmore forts) in air photos, which suggests that engineered roads had been planned to link them to both the Gask and each other but, in the event, these were never built. It remains perfectly possible, however, that lighter 1st century tracks were laid down as a temporary expedient, perhaps consisting of little more than hard-core spreads, but these could easily have been ploughed away in later times, to the point of being archaeologically undetectable.

Interestingly, air photo rectifications of the two Innerpeffray camps suggest that the annexe of the 63 acre example might intersect the west ditch of the 130 acre camp, which might some day allow us the opportunity to further flesh out the history of this feature complex, by finding out which of the camps was built first.

Geophysics and remote sensing: Strageath
For the last few years the Project has been conducting a series of very large geophysical surveys, taking in entire Roman forts and their surroundings, and we are eventually hoping to survey most of the forts to the north of the Forth-Clyde line. This year we were able to study the Gask fort of Strageath (NN 898180) via a 13 ha (32 acre) survey, the largest we have yet done. It covered the whole of the Roman fort and its annexes, along with a considerable area to the north. A network of roads has been detected from the air in the latter, which might suggest the presence of external activity, or even a civilian vicus of the kind that has been found outside many Roman forts in northern England, but which is, so far, very rare in Scotland. This was the second time the Project has done survey work on the site: for our member Neil Lockett led a somewhat smaller (3.9 acre) resistance survey on the fort's north-western quadrant in the 1990s, which gave particular attention to the annexes and a large rectangular structure to their west. The 2007 survey was carefully laid out onto Neil's grid base line so that the two sets of data to be directly overlain. This saved us duplicating all of his work, whilst allowing us to compare the data where the two surveys did overlap.

The fort's interior was extensively excavated by Profs. Frere and Wilkes in the 1970s and 80s, and its layout and history are thus reasonably well understood, but very little excavation was done on the defences or annexes, and none at all beyond them. Air photography has, however, shown remarkable complexity in the defences, comparable to (or even exceeding) that still visible on the surface at the fort's southern neighbour: Ardoch. There have been a number of past attempts at illustrating the complete pattern and disentangling the phasing within it, but none have been wholly convincing, not least because certain areas, notably the fort's north-eastern quadrant, have consistently failed to show cropmarks. Fortunately, our own resistivity and magnetometry results have clarified matters considerably and more accurate plans are now in preparation. The data have also allowed us to confirm impressions gained through excavation and air photography, but never before proved. For example, over the last few years the Gask Project's aerial and geophysical work at a number of sites has greatly augmented scant past excavation evidence for fort ramparts turning inwards at the gates, to produce a space in front of them that was exposed to enfilading fire from the protected position of the rampart tops. Indeed, the evidence has grown so strongly that this now shows signs of being the normal Roman fort gate pattern in 1st century central Scotland. It also makes a great deal more sense of the long known so called 'parrot (or eagle) beak' ditch terminals at fort entrances, since these combine with the rampart re-entrants to produce rapidly narrowing funnels to disorganise attacking formations. The Frere and Wilkes excavations had produced signs of another rampart re-entrant, at Strageath's west gate: the gate at which the main frontier road enters the fort and the side on which the fort's annexes lie. There were, however, faint aerial signs that the east gate might also show a re-entrant and so the gates were a particular study target of the survey. In the event, the combined resistance and magnetic data produced evidence for re-entrants at all four gates, which further reinforces the emerging pictures that, in the Flavian period, this tactically powerful arrangement was the standard regional design.

To the north of the fort, the survey confirmed the existence of the road network seen from the air and clarified its layout. It also revealed signs of what might be field enclosures, which would complement aerial indications for a more extensive 'Celtic' type field system to the south of the fort. More interestingly, the magnetometry showed concentrations of possible features in two large areas near the roads, to the north and northeast of the fort. But, although these may well represent signs of extramural activity, perhaps in the form of hearths, this is not quite certain as yet. For they were not sufficiently structured to prove whether or not they represent man made, rather than geological, features, and it is to hoped that we will be eventually able to study them further by excavation.

Air Photography
Thanks to a very wet spring and early summer, 2007 provided one of the worst cropmark seasons of recent times, with virtually nothing visible: not even 'old faithful' sites which almost always show. As a result, we will use our flying budget to concentrate on shadow and soil mark work over the winter, or germination marks in the spring. We have, however, been able to bring the cataloguing and mapping of our existing, 11,000 picture, aerial archive fully up to date and to use that data to extend our knowledge in a number of areas.

Inchtuthil
As outlined in last year's annual report, the Project intends to put a particular emphasis on the legionary fortress of Inchtuthil over the next few years and, more particularly, on its surroundings. For the fortress was the strategic hub of the entire 1st century occupation and it seems particularly important to get a better understanding of its context. Our first priority has thus been to gain an impression of the human and natural environment in which the fortress was established and, to date, we have looked at a c. 450 square kilometre study area stretching 10km to the north, east and west of the site, and a little further to the south to take in the Roman fort of Bertha, which currently marks the north-eastern end of the Gask frontier. The results are shown in the map below (Fig 4).

The area is highly diverse topographically, ranging from the flat lands around the rivers, some of which are barely above sea level, to the Sidlaws in the south-east and the Highland fringe in the north, which rise to over 300m and 450m (1,475') respectively. Naturally, the basic terrain has not changed much over the last 2,000 years. There have, though, been alterations of river courses. Indeed, we still do not know which side of the Tay Inchtuthil was on in Roman times. At present, the fortress lies north of the river, but there is plentiful evidence that the Tay has wandered north of the low plateau on which it stands several times since the end of the last Ice Age, and the presence of a series of oxbow lakes around the north side of the hill, coupled to erosion of the fortress' northern defences would suggest that the last occasion was relatively recent, albeit before the earliest accurate maps of the area were drawn in the 18th century. What has changed most, however, is the amount of land that is well drained and suitable for farming. Even today, there are patches of wetland in the area, but these would have been far more extensive in ancient times before the advent of modern drainage works, and we have been able to plot at least an approximate picture of the natural situation by combining data from old maps, from drainage channels marked on current maps, from land drain patterns revealed through air photography, from geological maps and old estate (and other) records. Furthermore, even today, large areas beside the Tay and Isla and some of their tributaries are liable to periodic flooding, and we have been able to combine flood data from the Scottish Environment Protection Agency, air photographs of past floods held by Perth library and our own aerial archive to map these areas too. This has revealed, amongst other things, that the Inchtuthil plateau, although permanently high and dry in itself, can still become an island when the Tay is in flood and the situation may have been more pronounced in ancient times, before the construction of dams on the Tay above Dunkeld and flood control embankments beside parts of the Tay and Isla.

At present just five permanent Roman sites are known with certainty within the study area: the Inchtuthil fortress itself; the auxiliary forts of Bertha and Cargill; a fortlet, also at Cargill (and probably not contemporary with the fort) and a watchtower at Black Hill. There is also a possible second tower at Woodhead (reported on in last year's annual report) and six temporary camps, two of which lie immediately outside Inchtuthil, and a third close by. Obviously, however, the Romans did not operate in a vacuum and we have also been able to map all of the known likely Iron Age sites onto our environmental map by combining data held in the Sites and Monuments Record of Scotland and adding in the significant number of new discoveries from our own air photography. This has produced a total of 267 sites, many of which involve multiple structures: a total of one known site every 1.65km2 overall, and a much higher density on the driest land. In detail, these comprise one broch, four crannogs, 14 native forts (including hill forts), 57 souterrains (including 23 souterrain settlements), 33 unenclosed settlements, 94 isolated roundhouses and 64 so called 'hut circles' in the hills. Of course the picture so gained can only be the vaguest approximation of the actual situation on the ground whilst the fortress was occupied. For example, although most of these sites can be assigned to the Iron Age with considerable confidence, this was a very long period and there is no way at present of knowing what proportion of the corpus was actually in use during the relatively short Roman occupation. Moreover, at least some of the upland hut circles may not be Iron Age at all, but later shielings and the like. On the other hand, it seems probable that the current site assemblage is just a small part of what once existed and certainly multiple new sites have been discovered in the area virtually every year that we have flown it.

The fact that our existing evidence is thus only a sample, coupled to the fact that so much of it derives from air photography, also means that the data is likely to be somewhat biased. Firstly, the upland sites are likely to be over-represented as a proportion of the whole, because much of the hill country has never been ploughed and they are thus far more likely than lowland sites to survive as easily detectable surface features. The only real exceptions are the large areas of forestry plantation in the hills (particularly around Dunkeld), which obscure any surface features from the air. For, although there are also significant plantations on the lower ground, most of these occupy actual or former wetland, which is likely to have been more sparsely settled. The over-representation is particularly true of large sites such as hill forts. For, although there are a few cropmark hill forts, most are well preserved earthworks, which are most unlikely to escape notice, which means that the chances are that virtually all of the hill forts in the study area are already known and so will again be over-represented in the site corpus. On the other hand, the lowland sites occupy better farmland and are thus far more likely to have been ploughed out. As a result, they are likely to be under-represented in the corpus, probably dramatically so. Some may have been destroyed beyond all hope of discovery, either by agriculture or through the construction of modern roads, buildings and railways. Others may well produce cropmarks, given time, but this is a notoriously capricious process. After all, we are still finding many new sites a year despite the fact that the Gask area has been subject to regular archaeological flying for 50 years, and it may take many more decades before this torrent begins to dry up. Moreover, there are probably many sites whose remains are too superficial to ever produce cropmarks and these may only be found serendipitously, during excavations or geophysical work. As a result we may have barely a tithe of the lowland Iron Age sites that once existed.

Yet despite such reservations, the general pattern revealed by our existing data is probably sufficiently representative to allow general conclusions to be drawn. In particular, as might have been predicted, the native sites tend to cluster on the best land, around the rivers, with the lower Isla and the reach of the Tay between Bertha and Stanley being particularly popular. Perhaps more significantly, there is an exact reverse correlation between the sites and the flood/wetland data. The 20th century saw a growing tendency to build in flood planes and there has been considerable controversy over the all too predictable flood damage that resulted. Earlier periods had more sense, however, and with only a few exceptions our mapped wetland and flood-prone areas are almost totally devoid of Iron Age occupation. Interestingly, we did a similar survey in 2003 on a slightly smaller, c. 300km2, area around the Roman fort of Cardean, during Birgitta's work to prepare her own geophysical work on the fort and the late Prof A.S. Robertson's 1960s and 70s excavations for publication: and the results were remarkably similar. We have been keeping the resulting map up to date ever since, as every year has produced additional features from the air, and we now have 135 sites in the sector, an increase of over 28% in just 4 flying seasons. The same reservations apply here too as regards the likely completeness of (and biases in) the data, but again the overwhelming majority of the sites lie away from ancient wetlands or areas liable to flood, but instead cluster on good, dry land. The two surveys overlap to some degree and can thus be plotted together to provide a wider picture of around 700km2 of lower Strathmore and Strathtay. These maps have demonstrated the value of our own intensive air photographic flying of the area. For, despite the fact that this area has been flown for over half a century, whilst we only started in 2000, between 20 and 25% of the current site corpus are our own discoveries.

We have also been able to begin a comparison between Inchtuthil and other legionary fortresses in Britain and abroad, in concert with scholars elsewhere in the former empire, and it is hoped that this will eventually give us a broader picture of the form and role of these installations. Amongst other things this has shown us that Inchtuthil's semi island position is far from unusual, with the fortress at Bonn in Germany providing a particularly striking parallel. Studies of other fortresses also give us some idea of the sort of additional features that we might need to look out for at Inchtuthil, as they seem to be all but standard elsewhere. These include riverside harbour works and external settlements, market places, cemeteries, baths and even amphitheatres. We would not expect all of these features to be present at Inchtuthil, which was only in use fairly briefly, but those that do might shed light on the Roman's priorities, because these will be the facilities that appeared early and thus, presumably, those deemed to be the most urgent. It has also proved to be extremely common for legions, which were themselves heavy infantry units, to have a cavalry regiment stationed in a fort nearby, and here Inchtuthil might already be fitting the pattern, because our geophysical and metal detecting work in 2005 at the closest auxiliary fort, Cargill (less than 5km away), produced finds of cavalry equipment.

Collaborations: representing Perth in Aschaffenburg
2007 saw a deepening of our existing collaboration with the museum of Aschaffenburg in Germany, as part of their twin town arrangement with Perth. The twinning is particularly apt because Aschaffenburg also lies on a Roman frontier: in this case the so called 'wet sector' of the German Limes (pronounced 'Leemase'), which is only a little later than the Gask line in date and runs along the bank of the River Main for some 70km between Groß Krotzenburg and Miltenberg. The relationship with Perth thus provides us with the opportunity to conduct comparative studies of the two systems, to study the differences and similarities between them and so, hopefully, shed greater light on both. This June Aschaffenburg funded us to spend 10 days on the line, to familiarise ourselves with its sites and terrain, and conduct a study of it's likely signalling system, similar to those we have already done on a number of other Roman frontiers. The latter have included the Gask, Hadrian's Wall and the Antonine Wall in Britain, and two other 60km+ sections of the German Limes: one on the early Wetterau Limes, in the Roman province of Upper Germany, to the north of the Main sector (Fig 6, sector 1) and the second well to the south-east of the Main, on the later, mid 2nd century 'Outer Limes', on the border between the provinces of Upper Germany and Raetia (Fig 6, Sector 2). The work was done in exceptionally hot conditions, but we were still able to achieve all of our planned objectives.

One somewhat unexpected result of the survey was to give us a very close parallel with the so called 'Glenblocker' line of forts in Scotland. These are deployed to the north of the Gask line, over a long arc from Drumquhassle, near Loch Lomond, to Fendoch, near Crieff and ultimately, Inchtuthil and, as their name suggests, they sit in or near the mouths of the main Highland glens. In the past, two main strategic explanations have been offered for their existence. Some say that they were simply defences against raiding hoards of highlanders swarming out into the Roman occupied lowlands. Others see them as springboards intended to act as bases for future Roman advances which, in the event, never materialised. Neither has seemed wholly satisfactory, however. For example, there is a problem with the putative invading armies of hill people: because archaeology has not been able to find them. As stated above, the Highlands are ideal for site preservation because so much of the land has never been ploughed and so the chances are that we know of a much higher proportion of the original total of Iron Age habitation sites than we do for the more intensively farmed lowlands. Yet, although aerial and surface surveys do find Iron Age sites in the hills, they are relatively thin on the ground and would certainly not suggest a population large enough to provide much of a hoard. On the other hand, the springboard theory has its own problems. Firstly, without people, there was little to invade in the Highlands. The land itself would have been of little value and for large areas there was nothing beyond it but sea. At the same time, only Inchtuthil had sufficient manpower to mount an invasion and, although greater force could have been brought to bear at other sites, one wonders how much use a mere auxiliary fort would have been as an invasion base. Perhaps the most important clue to the system's real rational lies in the nature of the Highland glens. A particular weakness of the springboard model for the Glenblockers is the simple fact that most of the glens do not really lead anywhere. They do, however, interconnect: so much so, in fact, that it is possible to go up virtually any of the glen mouths between the Clyde and the Tay and emerge from any of the others. They do, however, also connect with two routes that penetrate much further to the north, and in particular Moray, where in Roman times, as today, there was rich agricultural land which did support a significant population. To the north of the Tay there are two potential invasion routes into and out of northern Scotland. The main route follows Strathmore. Most post-Roman campaigns into (and from) the north of Scotland have followed this route, as does much of the modern transport infrastructure. A less popular route, however, follows Strathtay through modern Pitlochry, Glen Garry and Strathspey, and then forks to head for either Elgin or Inverness in the manner of the modern A9. Such an interconnection network would obviously have undermined the Gask frontier, because it would have made it dangerously easy to outflank. For example, one could slip down the Glen Garry route as far as Blair Athol, then turn west towards Loch Tummel or Loch Tay, and come out virtually anywhere between Crieff and the Clyde. Alternatively, without Inchtuthil, one could have moved down Strathmore, turned up Strathtay and again entered this maze of glens to outflank the Gask. The forts are, thus blocking the glens, but not necessarily against forces which originated in those glens. Indeed, for this very reason, it would not be surprising to find one more Highland line fort somewhere between Loch Lomond and the Clyde: possibly at Dumbarton, which was the head of navigation until the demolition of the Dumbuck shoal in the late 18th century.

The strategic position of the Main sector tuned out to be remarkably similar. For the forts face a range of hills known as the Spessart, across the river to the east. These range up to a little over 500m in height, which is broadly comparable to the Highland fringe, and they are also bisected by a series of deep, steep sided, interconnecting valleys, very like the Highland glens. They too link up to longer distance, potential invasion routes, including an ancient ridgeway called the Eselsweg, but were themselves historically something of a no man's land. Indeed, they were notoriously lawless until well into the 19th century, when the unification of Germany finally allowed the authorities to deal in a concerted fashion with the robber bands that haunted them. It is perhaps not surprising, therefore, that most of the Roman forts on the Main face the mouths of these valleys, in exactly the same way as the Glenblockers. Some also watch what were the sites of fords across the river, until the artificial navigation improvements of relatively recent times. The line would thus have been a perfectly sensible place to have run the frontier, even had the Main not existed, but the presence of such a major navigable river is still a significant difference between the 'Wet Limes' and the Glenblockers. Large rivers can be something of a double edged sword for a defence system. They can present formidable barriers - and the Main is over 200m wide in parts of this sector - but they are also important highways. The latter role can be a great boon for the defenders' logistics, but it also adds an obvious element of vulnerability. That said, the Main may only have been navigable on a more seasonal basis in ancient times. At present a system of locks and other regulatory measures makes the whole of this part of the river passable year round by large cruise vessels and cargo barges in all but the worst of droughts. But historical records make it clear that the Main could sink to little more than a trickle in a really dry summer before these improvements were installed, so that at times the river could effectively, if temporarily, cease to exist as a strategic factor.

Signalling
Our previous work on frontier signalling in Britain and Germany had produced a remarkably consistent pattern in which all, or virtually all, of the minor sites (watch towers, fortlets etc) had a clear line of sight to a fort, and so a direct connection via the visual signalling techniques in use in the Roman period. At the same time, the forts themselves could be linked together to form long distance strategic chains, either by again being inter-visible or, more frequently, by means of intervening relay sites. Other scholars have since produced similar results in Judea and Dacia, so that it begins to appear possible that this was a universal aspect of the design of fortified Limes systems. The basic architecture of the Main line is rather different from that of a purely land frontier, however, and in particular very few minor installations are known on the system. It was thus interesting to see whether we could still reconstruct signalling arrangements for the line. As usual with such surveys, our program of work was simply to visit each of the known sites to see what each could see, using a camera on a tall collapsible pole to mimic the likely full original height of fort gate towers and the like. Long stretches of the river banks are today lined with tall trees, which blocked views and so caused a certain amount of inconvenience, but the fieldwork was still relatively straightforward to accomplish.

The results were certainly promising (Fig 9), but more work will be needed before a definitive account will be possible. For example, in the north, the forts of Groß Krotzenburg, Hainstadt and Seligenstadt are all mutually inter-visible. Further south, Obernburg fort can see Wörth, which can itself see Trennfurt, whilst both Obernburg and Wörth can see two independent watchtowers that lie roughly half way between them and supervise a loop of the river at a point where the south side of its valley is unusually steep. Finally, in the extreme south, the two forts of Miltenberg Altstadt and Miltenberg Ost are also inter-visible. This still leaves quite long gaps, however, but there is a certain amount of evidence that points to the existence of additional sites that would fill at least some of these. For example, a surface feature has long been known at Nilkheim, between the forts of Stockstadt and Niedernberg, which is suspected, if not yet proven, to be a Roman fortlet. If the identification is correct, then the site is in visual contact with both forts, and could thus act as a relay between them. Likewise, there is a rather long gap between the forts of Niedernberg and Obernburg, which are not themselves inter-visible. But there has been speculation in the past that there might be an intervening fort, guarding an old ford, roughly half way between the two at Großwallstadt. This would certainly be an ideal spot and it is in full view of both Niedernberg and Obernburg, but until now there had never been any evidence for a Roman presence except, perhaps, for a suspiciously regular street pattern in the current Medieval village centre. During our visits to the site, however, we were able to pick up Roman pottery from a construction site close to the likely fort area, which would be a strong indicator of a Limes position, albeit it might also be a pointer to just a civilian presence.

If Großwallstadt and Nilkheim do prove to be Roman military sites, only two gaps remain in the system. The first lies between the forts of Seligenstadt and Stockstadt. These are over 10km apart, which is far further than the usual fort separation on the line and it seems highly likely that we have a fort missing on this sector, although as yet there is no material evidence for it. The best position might be close to the modern village of Mainflingen, where it would oversee a long loop in the river. This position is certainly within sight of the two known forts, as are long stretches of the river bank to either side, but for the moment that is all that can be said. The second gap lies between the forts of Trennfurt and Miltenberg Altstadt. This too is a rather long separation (c. 8km), but the Main passes through a virtual gorge in this stretch, which leaves little or no room for another fort on the river bank. There is, however, a spot known as the Martinsblick on a hillside above the modern village of Laudenbach, which used to be famous for its superb views up and down the river. It now lies in a dense forestry plantation, but there is an open field just to the north from which Trennfurt and Miltenberg Altstadt are in full view, along with the final fort on the sector: Miltenberg Ost. Once again, as far as we know, there is currently no evidence for a Roman site here, but it would certainly be an ideal spot, and so in principle there is very real potential for a comprehensive signalling system on the Main, albeit we cannot yet claim to have proved it in full.

In early September we returned to Aschaffenburg to take part in a Roman festival and during that time we were able to conduct another signals experiment. We have always accepted the accounts of Roman writers, that long range signalling was done visually. But the discovery of what might be a trumpet mouthpiece on a Limes tower has led some German scholars to believe that signalling on the German frontier might instead have used acoustic methods. Certainly we know that trumpets were used at fairly close range for battlefield communications, as they have been in many other times, but we have always doubted whether they could have been used over the sorts of ranges (six miles or so) needed for signalling on Roman frontiers. The Roman festival gave us the chance to find out.

One of the highlights of the festival was the arrival of a reconstructed Roman warship, the 'Regina', built by the University of Regensburg. The ship was to approach the town from the south (upstream) under oars, doing around 4 knots and we manned a beacon signal on a hill known as the Ludwigshöhe, c. 3km to the south of the town, which would sight the ship around half an hour before she docked. The hill was visible from the docking quay and the idea was simply to light the beacon as soon as the ship came into view to let the town know that she was coming. This much was just a bit of theatre to demonstrate how beacons worked, but the real academic interest lay in a race between the beacon and a trumpet signal produced by pupils from a local high school: the Kronberg Gymnasium. Thanks to their teacher, Dr Ludolph, the children had put a lot of preparatory work into the exercise and one of their first discoveries was that it was absolutely impossible to hear a trumpet blown on the Ludwigshöhe from the Main bank in Aschaffenburg; so they were forced to organise a relay system. The only question then was how many relays would be needed: and the answer, to even our surprise, was six. In other words, to get a trumpet signal a mere 3km took seven trumpets in all: the originator and six relays. Of course it might be argued that this was not quite a fair test because the final stage of the link went through Aschaffenburg itself, where noise from modern motor vehicles and the like might act to jam acoustic signals. But we should not forget that Roman forts would also have been very noisy places, as were their external vici, because as well as the noise of the hundreds of people present, the ubiquitous metal workshops would have generated a constant din of hammering. To make matters worse, the fort would have been generating internal trumpet signals and all of this would have raised a powerful jamming field. Not surprisingly, therefore, the beacon beat the trumpets hands down for speed, and took only a single unskilled person to operate (David with a cigarette lighter), whilst the acoustic signal needed seven trumpeters, all of which had needed the training necessary to know how to play. In other words, visual signals proved themselves to be faster and more efficient as a means of long distance communication and we are now still more doubtful whether acoustic signals would ever have been of much use at range. Our traditional assumptions would thus seem to have been vindicated. The trumpeters did, though, beat the school's other attempt to take on our beacon: a relay of runners. These tried gamely in pouring rain and did pass on their message with impressive speed, but both signals overtook them with ease.

Ironically, the wet weather on the day of the test, though hardly welcome to the festival organisers, made the experiment even more valuable to us, because the signal still got through despite the fact that the visibility between the Ludwigshöhe and Aschaffenburg was dramatically reduced. Knowing that a system will work in ideal conditions is one thing, but knowing that it will still work when faced with real adversity is much better, especially as the beacon itself was much less powerful than we would have chosen, thanks to the need to minimise the fire risk on a largely forested hill.

Publications, outreach and publicity
Our outreach program this year spanned both sides of the Perth/Aschaffenburg twinning, providing opportunities for hands on experience or, for the less active, just the chance to witness archaeological work in progress. Our Innerpeffray dig was entirely manned by local volunteers and also received many visitors, including an organised trip arranged by our sponsors, the Perth & Kinross Heritage Trust. We also took volunteers on the Strageath survey to supplement our members and students. These made a substantial contribution to the geophysics, but we were also lucky to have another opportunity of working with Paul Smith and his network of metal detectorists, who made a number of useful finds from outside the fort itself, which when fully analysed should help us to form a better picture of activities around the fort.

Across the North Sea, our participation in the Aschaffenburg Roman festival has already been mentioned, but our signals experiment was not our only activity there. Birgitta had spent time at the Kronberg Gymnasium earlier in the year to explain a number of Roman signalling techniques to the pupils and, under Dr Ludolph's supervision, teams had been formed to practice using them with a view to a public demonstration during the festival. We took with us full scale reconstructions of original Roman equipment and, on a sadly rather damp morning, the children were able to transmit signals across the Main near the Museum, whilst a second team, c. 200m away on the opposite bank, successfully read them, despite the fact that they were in Latin and the teams had not been told in advance what the texts would be. We also had another opportunity to work with Scottish Roman army re-enactment group: the Antonine Guard. The Guard gave numerous demonstrations during the festival, with Birgitta giving a German voice-over, and the locals took them to their hearts, especially the children who received a certain amount of basic Roman army training. Indeed one young lad turned up for almost every demonstration, acquiring his own home made Roman uniform in the process. Our Roman signals equipment also drew large crowds, especially once the children discovered that they could use it to signal their names. From our point of view, however, the culmination of the festival was probably the trip we took down the Main from Aschaffenburg to the Roman fort town of Stockstadt on the full size Roman warship reconstruction, Regina. With the Antonine Guard aboard, alongside Regina's own crew of rowers, she looked superb. Indeed, we were a little conspicuous as the only two present wearing modern day clothes, but it was a real experience to see the capabilities of the vessel at first hand and to watch the skill with which she was handled by her crew in what is a very busy, and surprisingly fast flowing commercial water way. All in all, the festival was a great success which brought a great deal of good will to the town twinning, as well as real educational benefits, and it would be nice to think that something similar might be organised in Perth. It is a shame that it would probably be too difficult to get Regina across the North Sea, for the sight of her coming up the Tay Estuary would certainly be a sight to behold. But we would be happy to work with Perth schools in the way that we did with the Kronberg Gymnasium.

As ever, the Directors have continued to give lectures to a variety of academic, student and amateur bodies in addition to our University teaching. As always, we made particular efforts to speak in Perthshire, with talks this year in Perth, Pitlochry and Meigle, and a guided tour of Ardoch Roman fort. Elsewhere, we gave lectures in Callander, Melrose and Manchester, and both directors gave papers to the annual Roman Army School in Durham. Media activity also continued. We took part in a BBC Radio documentary called, 'In the footsteps of Calgacus', about the Roman invasion of Scotland and discussions have begun with 'Time Team' and the 'History Channel' with a view to future TV appearances. Our activities at Strageath and Innerpeffray both made the papers and the Aschaffenburg Roman festival attracted Radio, TV and double page spreads in the German press for several days running.

As always, 2007 has seen a number of Gask Project publications. Birgitta continues to be editor of the 'Hadrianic Bulletin' and both of us published papers in the journal. David's being a summary of our geophysical and aerial surveys on Roman forts to the north of the Forth and Clyde, and Birgitta's a study of the light that finds analyses are shedding on Roman/native relationships in Scotland. She also published a number of other finds papers including a section in the multi-volume 'Corpus of Roman finds in European barbaricum'. Likewise, we prepared a number of pieces for future publication. The long delayed final report of our five season excavation on the Roman tower and ancillary structures at Garnhall has been submitted to PSAS. The Innerpeffray excavation report is finished apart from its pollen analysis. Birgitta reported on a bead found at the Black Spouts homestead, Pitlochry. The Strageath report is in progress, and we have started work on the report on our excavations at the fortlet of Glenbank.

Finally, Birgitta received a well-deserved professional honour during the year when she was elected as a Fellow of the Society of Antiquaries.

Sponsorship and Acknowledgements
The Project continued to be sponsored by the Perth & Kinross Heritage Trust, whose support has, as always, been indispensable and is very much appreciated. In 2007 the Trust funds went into our Innerpeffray excavation and Strageath survey and the residue will go into winter air photographic flying. The Trust also made a one off grant to fund Birgitta's report on the Black Spouts bead. We are also grateful to our long-standing corporate sponsor (which continues to insist on anonymity). Historic Scotland funded the preparation of the Garnhall report and were also most helpful during our application for Scheduled Monument consent for Innerpeffray. We have received a number of smaller donations from speaking engagements, voluntary bodies and private individuals and, as usual, the royalties from our books have gone to the Project, with last year's new volume "Rome's First Frontier' doing better than expected.

The Project continues to owe thanks to the farmers and land owners who have allowed us access to sites and to Peter Green who has continued to do a wonderful job of maintaining our web site. We thank the staff of the Perth J.K. Bell Library, Dr Ludolph and the pupils of the Kronberg Gymnasium, Dr Marcus Marquhardt of Aschaffenburg Museum, for organising and funding our German work, Dr Bernd Steidl of the Prähistorische Staatssammlung Munich along with Dr Marcus Scholz and Dr Barbara Pferdehirt of the Römisch-Germanisches Zentralmuseum in Mainz, for information on the Main Limes and discussion of our preliminary results, and Messers Bill Fuller and Dan Boddice for their work on cataloguing our air photos, finds and excavation archives. As always, we are grateful to our many field volunteers, especially our long-standing geophysics team: David and Susie Hodgson (who were married earlier in the year), Rachel Hunt and Paul Murdoch and to Dr Peter Morris and Paul Smith (our metal detector liaison) for their help at both Innerpeffray and Strageath. Finally, we would like to say a particular thank you for the way our field team and local friends rallied round after the serious car accident we suffered during the Strageath survey. As well as the personal support we received, work on the site itself never missed a beat.

The Future
2008 should see another in our series of whole fort geophysical surveys and, at present, we have our eye on Bertha, just upstream of Perth on the Tay. We had in fact hoped to work there in 2007, but substituted Strageath for agricultural reasons. Our air photographic program will continue, hopefully with better crop conditions, and we hope to excavate one of two candidate sites for new Gask towers. We also plan to conduct some preliminary trenching outside Strageath to augment and assist in interpreting our geophysical data.

Out of the field, we plan to finish the Glenbank report, prepare German and English reports on our Aschaffenburg work and prepare the final excavation report for our five season excavation on the Roman period native settlement of East Coldoch. The Directors will continue to give public lectures where invited. Six society talks have already been booked, along with further papers to the Roman Army School. We will both be participating in a conference on the 1st Century invasion, to be held at Perth Museum. Birgitta will be addressing the Classical Association Conference in March, and other dates will no doubt be scheduled as the year progresses. We also plan to continue our work on the hinterland of the Inchtuthil legionary fortress. All in all it should be a busy year and we can hardly wait to get going.

D.J. Woolliscroft and B. Hoffmann
Directors: The Roman Gask Project
SACE
University of Liverpool.