Prehistoric and Roman Westmorland

PREHISTORIC AND ROMAN WESTMORLAND

The difficulties normal to the treatment of a county as a unit of prehistoric or early historic Britain are more acute in the case of Westmorland than in that of many English counties. If to the east the high moors of the Pennines form a natural no-man's land, to the north-west and the south-west the county shares with Cumberland and Lancashire the broadening valley of the Eden and the coastal lowlands of Morecambe Bay. Between these outlets an almost continuous barrier of high fell links the Pennines with the great massif of Cumberland, and divides our county into two distinct parts. The Roman road-makers, it is true, forged some semblance of a link between the two: they anticipated the railway which climbs arduously northwards up the Lune valley, and, with their obstinate capacity for "levelling mountains and draining morasses," even trailed the precipitous 'High Street' into the mists above Haweswater. But in the Middle Ages the real separateness of the two regions was expressed by a divergent history and, incidentally, by their constitution into the respective baronies of Appleby and of Kendal. So, too, in pre-Roman times the no less cogent evidence of archaeological distributions (fn. 1) points constantly to a similar geographically controlled duality.

1. Pre-historic burial-mounds and circles

The earliest relic which comes within the scope of this Inventory is probably the long barrow on Rayseat Pike, in the parish of Crosby Garrett not very far from the upper waters of the Eden. The barrow, or rather cairn, even in the sadly devastated condition in which Greenwell left it in 1875, still looms over the desolate moorland from the 900-ft. contour, which is here the highest on the map. Its excavator records that it was 179 feet long and contained burnt bones in a cremation-trench, instead of the inhumations normal to the long-barrow culture. This unusual feature brings it into line with a well-known Yorkshire series, and, as Professor Collingwood observes, impels us to regard it as a stray from across the Yorkshire border. It thus introduces into the prehistory of Westmorland a problem which has long been familiar to archaeologists but has perhaps been shelved rather than solved by the postulate that longbarrow cremation is a borrowing from round-barrow cremation of the Bronze Age. This postulate implies a remarkably late survival of the long-barrow form, since cremation is as exceptional in the earliest phase of the Bronze Age as it is in the late neolithic, and can scarcely have become sufficiently prevalent to revolutionise the older culture until the approach of the Middle Bronze Age. In other words, if this theory be pressed, the round-barrow 'beaker'burials of Westmorland antedate, perhaps by some centuries, the long barrow of Rayseat Pike, and the latter represents an intrusive revival of Middle Bronze Age date. Without compelling evidence in support of it, this paradoxical conclusion is untenable; and the only safe course is to look elsewhere than to Bronze Age influence for the explanation of long-barrow cremation, even though search is unlikely to discover a final solution until our northern long barrows have been investigated more extensively and by more modern methods than those used by the pioneer-excavators of the 19th century. Everything necessarily depends upon the accurate association of relics; but we are told that the Rayseat Pike long barrow contained "no piece of pottery or flint," and possible associations of sherds with cremations in the kindred long barrows of Yorkshire are, for one reason or another, too uncertain for independent use. (fn. 2)

Elsewhere, however, useful associations are not entirely absent, and the body of evidence suggesting that, in the 'highland zone' cremation was not unknown in the neolithic period proper is gradually increasing. If we put aside for the moment the disputed YorkshireWestmorland long barrows, we may recall that attention has long ago been drawn to the hint of early cremation on the one hand in Derbyshire and on the other hand in Bute, where "the practice of cremation seems to have been the rule" in chambered tombs; whilst in Orkney a chambered tomb included cremations which may have been primary or nearly so. (fn. 3) More recent and reliable evidence is forthcoming from northern Ireland, which has many cultural links with northern Britain. At Dunloy, Co. Antrim, careful excavation has revealed in a horned cairn a fire-trench or 'flue' (recalling Rayseat Pike) and the cremated remains of at least five adults, in association with a leaf-shaped arrowhead and sherds including some of the 'Beacharra' type (a specialised variant of 'Neolithic A'). Again, in a horned cairn at Clady Halliday, Co. Tyrone, the only burial was "a cremation in the inmost chamber," and the objects found included a leaf-shaped arrowhead and "many fragments of round-bottomed pottery of simple shape, undecorated." (fn. 4) And the relevancy of this various evidence to our local problem is emphasised if we return to Yorkshire and reconsider one of Mortimer's barrows (his no. 81) at Garton Slack in the East Riding. (fn. 5) This barrow, though of a round shape that may with probability be taken to imply Bronze Age influence, contains a cremation-trench analogous to that of Rayseat Pike, and a re-examination of the pottery found in the trench shows that it is all 'Neolithic A.' Nor does the evidence end there; for Greenwell records that one end of the cremation-trench was found to have been mutilated by a secondary inhumation-burial associated with a beaker (Early Bronze Age). It is clear therefore that, in Yorkshire, cremation was associated with a neolithic culture, only slightly modified by the (presumably) intrusive round-barrow form, some considerable time before the end of the beaker period, i.e. some time before the normal Early Bronze Age rite of inhumation was succeeded by the Middle Bronze Age rite of cremation. With this evidence, that already cited from the 'cremation' long barrows at Kilburn and Westow (below, note 1) is consistent. The inevitable inference is that, whilst there was, as we might expect, a definite overlap between the 'Neolithic A' or (in this context) long-barrow culture and the beaker (roughly Early Bronze Age) culture, the cremation-trench survived with the former and was not introduced by the latter.

On the available information, therefore, it must be assumed that cremation was already a recognised alternative rite of the late neolithic phase of northern Britain, and that its occurrence there in long or chambered barrows owes nothing to Bronze Age invasion. On this view, the general supersession of inhumation by cremation at the end of the Early Bronze Age, coinciding as it does with the re-emergence from the 'highland zone' of an old neolithic ceramic tradition in the modified form of the food-vessel, is reasonably explained as the parallel triumph of an old native burial-custom of the same northern region reacting against the inhumation-rite of the intrusive beaker culture. If this provisional interpretation be accepted, there is no a priori reason for supposing, in accordance with the conventional view, that the YorkshireWestmorland series is necessarily "much later in date than the true long barrows" and that it represents "the neolithic practice of communal interment in a long barrow, modified by the Bronze Age practice of cremation." Rather does our long barrow take its proper place— whatever that exact place may be—within the limits of variation proper to a substantially intact neolithic tradition, (fn. 6) and it need not therefore be dethroned, as it would otherwise have to be, from its position at the head of the Westmorland monuments.

Less certainly a monument of the same general phase is an unexplored cairn of long-barrow form—87 ft. long, 25 ft. wide at the western end and 49 ft. wide at the eastern—in Crosby Ravensworth parish (46), where, again, accessibility from the Yorkshire border might be cited as a controlling factor. For the rest, Westmorland appears to be devoid of vestiges of this kind. The wayfarer approaching the cross-roads nearly a mile E.N.E. of Lowther Castle might eye suspiciously a long mound within the N.W. angle of the crossing but would probably conclude that it is nothing more than an old spoil-dump from the neighbouring road-cutting. 'Pillow-mounds' or 'giants' graves' in Waitby, Mallerstang and (formerly) in Bampton fall into a different, if doubtful, category (see below, p. xxxv). It must be admitted that, structurally at least, the neolithic tradition is poorly represented in our county.

Of the circular mounds or cairns, which may in a majority of cases be ascribed with certainty or probability to the Bronze Age and, in particular, to the second millennium B.C., upwards of 140 are still recognisable within the borders of the county, and might legitimately be increased by the addition of most of the 'stone circles' which, save in rare instances, doubtless stood within or about similar structures. Three or four of the mounds are known to have contained beaker-burials which, on the conventional dating, would be attributed to c. 1800– 1600 B.C., and it is noteworthy that all these examples are situated in the more northerly of our two regions: (fn. 7) i.e., they indicate overland movement from the direction of Northumberland (where beakers are numerous) rather than intrusion from the Cumberland coastline (where they are absent). On the other hand, mounds which have produced Middle Bronze Age relics— food-vessels, collared urns—together with many others of which the contents are unknown are widely distributed in both regions, and even extend into the high fells between Grasmere and Ullswater. Locally, the recognised Bronze Age custom of burial on heights and ridges well above the areas of actual settlement must affect our translation of mound-distribution into terms of settlement-distribution; but, when this allowance has been made, it is still clear that the mounds represent a population which not merely colonised the lower lands but also settled with fair frequency above the 1,000-foot contour. If this Middle Bronze Age culture chose originally any particular avenue of approach into the county, its widespread diffusion has effectively obscured the fact.

Reference has been made to the custom of including a circle of standing stones— occasionally, it seems, two concentric circles—in the structure of a burial mound, (fn. 8) and a majority of the 'stone circles' of the county are of this class. In three or four instances only do the remains court comparison with the free-standing ceremonial circles of which Stonehenge and Avebury are the archetypes. Without entering into the more controversial aspects of these monuments, we may say that none in Britain has been proved to be of a date prior to the Early Bronze Age (say, 1900–1600 B.C.), whilst only in the Scottish Highlands, where the long survival of cultural forms is a familiar fact, are they known to have been constructed as late as the end of the Bronze Age. Beyond the easy guess that the circles were foci of ritual associated with the religious and doubtless also the secular life of their builders, the modern archaeologist does not allow himself to venture, and can only find such satisfaction as he may in classifying them by type and in noting their distribution.

Of the more definite Westmorland examples, the simplest is the small circle (80 ft. in diameter) S. of Shap village, but it has some claim to distinction if nine orthostatic stones strewn over some two miles of country in the vicinity are in fact the vestige of a stone-lined avenue of the Avebury type. The identification is possible but, in the absence of excavation, lacks proof. At Orton a circle of small stones, with a diameter of 138 feet, may be a relic of a monument of the class but, if so, is a singularly poor construction; whilst at Casterton another circle of small stones, with a diameter of only 60 feet, would be dismissed at once as the remnant of a tumulus but for the fact that it is sited carefully on a levelled platform. There remain the two surviving great circles of a former group of three at Eamont Bridge, outside Penrith. Of these, the large circular 'amphitheatre' of Mayburgh, with its 15-foot rampart of piled stones and its series of monoliths, is perhaps unique. It has apparently no ditch and has nothing of the semblance of a fortification; on the other hand, its entrance looks towards the neighbouring 'King Arthur's Round Table' and it is difficult not to associate the two— the one possibly designed for secular assembly, the other for religious ritual. The 'Round Table,' with its ditch, its external bank and its two causeways, has the appearance of a miniature Avebury, but whether it was ever elaborated with a circle or circles of stone or timber is not known. It may be grouped with a not inconsiderable series of somewhat similar circles which include, apart from Avebury itself, the Stripple Stones in Cornwall, Arbor Low in Derbyshire, the circle at Broomend of Crichie in Aberdeenshire, and the Ring of Brodgar and other rings in Orkney. The distribution is a wide one but, in spite of the Aberdeenshire example, has a westerly or Atlantic tendency. This it shares with the megalithic tombs, which are as a class, however, of somewhat earlier date. So far as is known, the Avebury series belongs to the Early Bronze Age, and comparisons have been made between the earthwork-technique of that series and the technique employed by bell-barrow and disc-barrow builders of the Early and Middle Bronze Age. It is indeed tolerably certain that several diverse threads are woven into the 'circle complex.' The use of peristaliths is a common feature of the late neolithic and Early Bronze Age megalithic tombs; the use of timber circles is familiar in the Low Countries on the margin of the megalithic area during the first half of the second millennium B.C., and extended to this island; the use of a more or less elaborate earthen framework seems to be, in the main, a provincial British development of the same general phase. Premature attempts to simplify and to give definite map-direction to a cultural 'complex' such as this are as likely to obscure as to solve its problems, and the involved question of the circles—stone, earth and timber—is not advanced by our present knowledge of the Westmorland series. Without emphasis, it may suffice to note that the series as a whole is linked with the sea by other examples (notably Long Meg and her Daughters) lower down the Eden valley, and that this Atlantic orientation is supported by the existence of several megalithic circles along the Cumberland coastline. (fn. 9)

2. Camps, hut-villages, 'giants' graves'

Structural remains of Bronze Age dwelling-sites have not been identified in the county, and the next group of monuments with which this inventory deals is, on analogy, that of the camps or embanked hill-top towns. These are singularly few in number and small in size. Indeed, the only examples definitely of this class are the small 'fort' enclosing about one-sixth of an acre on the precipitous Castle Crag of Mardale (Bampton 68) and the tiny earthwork of Castlesteads in the parish of Natland (8). The defence of the former is known from excavation to have consisted of a stone wall; but there the evidence ceases, for clay floors associated with charcoal inside the enclosure failed to yield datable relics. Similar small stone forts are found in western Scotland, (fn. 10) and might be expected to occur more freely than they do in the somewhat similar hill-country of Cumberland and Westmorland. Their small size, in contrast with the spacious 'camps' of the English or Scottish lowlands, is doubtless in the main a reflexion of the sparse and scattered character of the upland populations; but it is possible also that they may represent, at any rate in some cases, a social system bearing some slight resemblance to mediæval feudalism. This possibility is suggested in particular by the more regularly and elaborately planned duns and cashels which western Scotland shares with Ireland—small, generally circular structures with massive walls that often contain stairs and chambers and anticipate in a remarkable degree the keep of a mediæval castle. In date, these northern duns seem to belong somewhat vaguely to the first millennium A.D., but they probably derived in part from earlier Iron Age prototypes farther south, in Cornwall and still farther afield along the Atlantic coastline. (fn. 11) Of structures of this relatively sophisticated class there is no certain example in Westmorland, but the wreck of a circular building with an internal diameter of 85 feet on Barton Fell may possibly have been an outlier of the dun-cashel series. (fn. 12)

Whether any of the earthwork-enclosures—all of small size and weakly embanked— that occur here and there in the county (fn. 13) are of prehistoric date is unknown in the absence of excavation. These works can in any case scarcely represent more than lightly protected farmsteads, and in this exposed region a wide range of date might on general grounds be assigned to them. They merge into the hut-villages, which form a large and important class and must indeed be ranked amongst the most striking features of Westmorland archaeology. Over fifty of them are included in the present inventory, and, though the number exceeds any previously recorded total, it cannot be doubted that an exhaustive search of the remoter uplands would extend it. These hut-villages, albeit doubtless of various and even (in some cases) relatively modern date, may best be considered in the present context. Whatever their actual period individually, they represent as a whole a cultural phase substantially prehistoric in kind and environment (see map, p. xxxvii).

With rare exceptions, they approximate to a single type. They are commonly found on upland ledges or plateaux within measurable distance of the 1,000-foot contour, and frequently in the vicinity of a beck. They are never fortified in the full sense of the term, though occasionally they are enclosed by earthwork or stone-walling. The most notable example of the former is Castle Hill in Dufton parish (8), where a shallow ditch with an outer bank encloses the kraal (fn. 14); and as an instance of the latter Crosby Ravensworth (29) may be cited. Often the hutments, or groups of them, melt into a continuous outside wall which thus encloses the main area of habitation—e.g., Asby (21), Crosby Ravensworth (28), and Hugill (13). The huts themselves are of dry stone-walling and, on plan, range from circular to square or oblong, with almost every sort of intermediate form, a squarish plan with rounded corners being a common type. It is evident that the various types of plan were in use contemporaneously, but it is not possible on the evidence available to ascribe specific uses to them. They have indeed every appearance of opportunism, the shape of the building being designed to fit the space available. They were presumably roofed with turves or thatch.

The settlements were doubtless occupied for the most part by pastoral communities who also practised a sort of garden-agriculture. (fn. 15) In some cases strip-lynchets, approximating to the mediæval type though of uncertain date, occur in the vicinity of the hut-groups: notably on Skirsgill Hill in the parish of Askham (43), and in Nateby (10), Waitby (15) and Wharton (8) parishes. The Nateby lynchets vary in width from 12 ft. to 42 ft., and those of Wharton from 12 ft. to 90 ft. This marked irregularity is due only in part to the inequality of the ground, and may in part be due to the incomplete adaptation of an agricultural system essentially foreign to the highland crofters. Unfortunately, the date of the introduction of strip-lynchets into Britain is disputed; and in any case it is possible that the lynchets in question have nothing to do with the adjacent hut-settlements. On the other hand, there can be no hesitation in attributing the small stone-walled fields of Crosby Garrett (8) to the hutments with which they are linked, and this group as a whole, covering 160 acres of fell between the 900 and the 1000-foot contours, is one of the most remarkable of its kind in Britain. The complex of walls and dykes is described and planned in the inventory: here it will suffice to note the square shape of the fields, ranging from less than one to one-and-a-half acres in extent; the farmtracks which can here and there be traced amongst them; and the tendency of the farmsteads themselves to take the form of irregular compartments grouped round a small farmyard. No digging has yet been attempted upon this key-site, but, for all the risk of finding nothing very determinate, the completeness and interest of the plan are more than sufficient to justify a trial-excavation.

The risk of finding nothing on a site of this kind is indeed a real one. The culture of these highland crofters was never exuberant, and local craftsmanship must have expended itself almost exclusively on perishable materials, wood and leather. Only a few Roman knick-knacks, with the astonishing power of penetration that durable Roman things have, reached Ewe Close on the moors of Crosby Ravensworth (25) to witness that man lived there during or soon after Roman times; whilst surer evidence is perhaps to be found in the behaviour thereabouts of the Roman road, which seems to swerve in order to pass Ewe Close and so to imply the preexistence (and co-existence) of that heterogeneous collection of huts and enclosures. A considerable excavation carried out in 1935 on the site of the settlement at Kentmere (18) yielded scarcely anything of import save a fragment of a plain bracelet of clouded white glass and of a type which might occur on any northern site of the Roman period, if not later. (fn. 16) A particle of evidence in the shape of a small bronze bull's head, probably an escutcheon from a second-century bowl, may indicate an approximately contemporary occupation of a hut-settlement near Kirkby Lonsdale (37) (see Fig.) (fn. 17); and this slight piece of evidence is reinforced by the reputed discovery of a Roman 'hipposandal' on the same site. A beehive-quern found in the Castle Hill settlement (Dufton 8) mentioned above is in accordance with the general trend of this evidence. Perhaps account may also be taken of "an old copper pan," some small stone mortars and "about 20 pairs of hand mill-stones" which, from an illustration, were beehivequerns of the Early Iron Age tradition—all found in 1774 during the clearing of foundations from a field called the Quamps and adjacent fields in Dalton Hall demesne S.E. of Burton village. (fn. 18) The foundations discovered during the clearing were 3 yards wide, their faces of undressed limestone and the infilling of "any kind of cobbles, etc." Plan and description are alike vague, but the remains seem to have been those of a hut-settlement which, whatever its actual date, retained something of a late prehistoric culture.

There our meagre evidence ends, and for the rest we are left with generalities. The distribution of the hut-settlements, although occasionally—as at Ewe Close, and beside Maiden Castle in Stainmore parish (1)—coincident with Roman roads, is not generally conditioned by the lines of the Roman occupation. The main groups centre round the clustering streamlets S. of Crosby Ravensworth, and in the upper Eden valley near Kirkby Stephen; but elsewhere these settlements make no sort of pattern on the map, and it is clear that only in the vaguest way was their distribution controlled by the geographical factors upon which in other contexts it has been found necessary to insist. It was in fact as opportunist as was the planning of the settlements themselves, or as the distribution of the Bronze Age round barrows wherewith it almost competes. Barrows and hutments alike reflect a basically primitive and static culture which from age to age varied little in its needs and achievements, and only a little in the details of its equipment. This culture, in the one form as in the other, was not peculiar to the highland zone of Britain; the Westmorland settlements find their counterpart in the hut-villages of which the traces swarm upon the downs of southern England—Rotherley, Woodcuts and the rest— and there represent the peasantry of the Early Iron Age and Roman periods. They stand to-day for the pre-Teutonic or non-Teutonic agricultural traditions which are not even yet entirely obsolete in the 'Celtic fringe' of the British Isles.

It will be convenient in this context to consider a somewhat puzzling series of mounds which occur, in a few cases, in propinquity to these hut-villages. Within 500 yards of the Crosby Garrett settlement, but in the parish of Waitby (14) and on the opposite (eastern) flank of Smardale Gill, is a group of oblong or oval mounds, each mound some 40 ft. by 15 ft. and completely surrounded by a shallow ditch. These mounds are known locally as giants' graves. A further group lies about the same distance to the south of the settlement, within the neighbouring parish of Ravenstonedale (33); and this southern group is included in a straggling system of dykes which frame the settlement itself and can scarcely be dissociated from it. Again, in the parish of Mallerstang four 'giants' graves,' 25–40 ft. long and 15 ft. wide, all surrounded by ditches, are situated only 100 yards from another hut-settlement (17). Similar 'graves' are also found, indeed, apart from settlements of the kind: notably, a group of five mounds, 48–90 ft. long and 21–24 ft. wide, which formerly existed in the parish of Bampton. (fn. 19) But at first glance there would appear to be some reason for associating the mounds with the culture represented by the villages, i.e., with a culture which, in part at least, coincided with the Roman period.

With this possibility in mind, it is necessary to consider the function of the 'graves.' Their name in itself implies a degree of antiquity and oblivion; for it was current by the middle of the last century, when the Ordnance Survey collected the names from local correspondents, and indicates that the original purpose of the mounds—almost certainly not burial—was then forgotten. To-day they are classified generally as pillow-mounds, such as are widely recorded from Kent, Wiltshire, Devon, Wales, and other parts of the country. In many cases these mounds were undoubtedly built as breeding-shelters for rabbits, and were designed to concentrate the animals and so to facilitate the netting of them. (fn. 20) The careful examination of a typical example would probably show whether the Westmorland series was, as general appearance would suggest, of this kind. But if the Westmorland 'giants' graves' are in fact rabbitwarrens, it becomes necessary to reconsider their apparent association with the hut-settlements. There seems to be no evidence for the existence of rabbits in Britain before the Norman Conquest, and the earliest records of them do not go back beyond the 13th century. (fn. 21) We must conclude, therefore, either that the particular hut-settlements in question, in spite of their general uniformity with others of Roman date, are considerably later than the Roman period; or that the association of the huts and the mounds is purely accidental; or that the mounds themselves were not designed as rabbit-warrens.

Something may be said for the last two no less than for the first of these alternatives. Most of the surviving hut-villages are not in fact equipped with the mounds. Similarly, though pillow-mounds do occur, in a few instances, within or close to Early Iron Age camps, (fn. 22) their conjunction with works of this kind is sufficiently rare to compel us to regard it as fortuitous. And, in regard to the last alternative, it was suggested by correspondents of the Ordnance Survey (in 1897) that the Mallerstang "giants' graves" were "merely bracken-stack bottoms" —an explanation which local enquiry has failed to confirm but which, if verified, might presumably apply to the whole of the Westmorland series and might at least remove the daterestriction implied in the rabbit-warren theory. (fn. 23)

There, pending further field-work, the problem may for the moment be left. It is one of sufficient interest to demand local attention, and is probably capable of easy solution.

3. Roman forts and roads

From the hut-villages, with their occasional Roman contacts, we may turn to the Roman occupation in the more conventional sense. As was to be expected from the position and topography of the county, that occupation was of a purely military kind. Roman Westmorland is an arbitrarily chosen slice of the hinterland of the main northern frontier of Roman Britain.

The Roman sites fall into three groups. The first group is threaded by the main road which, like its successor the railway, proceeded northwards along the western fringe of the Pennines to Carlisle. Along this arterial road—the western parallel to the Great North Road— the forts at Overborough (just outside the county), Low Borrow Bridge and Brougham mark normal 15–20 mile stages. The course followed is that of the Lune valley up to Crosby Ravensworth Fell (about 1,200 ft.), and thence in a nearly straight line down to the crossing of the Eamont at Brougham and so on to Old Penrith. The second group is that which lines the important branch-road proceeding eastwards from Brougham to the heights of Stainmore (1,468 ft.) and on via Bowes to join the Roman Great North Road between Catterick and Piercebridge. This transverse route was strategically of importance as the first link of the kind south of the Stanegate and Hadrian's Wall, and as an alternative means of reinforcing the flanks of the main frontier, whilst incidentally serving as a means of policing the high moors immediately to the rear of that frontier. In Westmorland, it was strongly equipped with permanent garrisons at Kirkby Thore and Brough, with a minor post known as Maiden Castle near the top of the pass; but perhaps more striking are the large camps of a more temporary kind at Crackenthorpe and at Reycross just over the Yorkshire border. The third group is posted at the heads of the lakes and valleys that converge directly or indirectly upon the basic coastline of Lancashire and Cumberland. Such are Watercrook (near Kendal), set at the point where the Kent valley narrows into the hill-country, and Ambleside at the head of Windermere, whence the river Leven broadens quickly into the flats of Morecambe Bay.

In detail priority must, as knowledge stands at present, be given to the great camp of Crackenthorpe, in the second group. With its light defences, its numerous entrances screened by traverses, and its large size, it is still, in its reduced condition, a noteworthy example of the type of earthwork which a Roman brigade would throw up as a temporary base during a campaign or during the preliminary organisation of a countryside. It has been calculated that the closely similar and clearly related camp, some 20 acres in extent, at Reycross on Stainmore would hold a legion, (fn. 24) and the Crackenthorpe camp—2 or 3 acres larger, to allow for the gulley that divides it—has much the same capacity. But whether the two camps, sixteen miles apart, "mark the passage of early Roman invaders of Cumbria" at two points separated by "a respectable distance for an army on campaign to cover" is less easy to say. They may rather have been the base-camps for legionary working-parties engaged upon the task of roadmaking after the completion of the actual campaign. Certainly at Reycross the Roman road has been observed to bend as it enters the camp, thus indicating the priority of the latter; whilst at Crackenthorpe, though the camp is not actually crossed by the road, the dependence of the one upon the other is emphasised by the presence of four gateways in the rampart facing the road and apparently only one gateway in the rampart furthest from the road. (fn. 25) If therefore at Reycross the road is not prior to the camp, at Crackenthorpe the camp can scarcely be prior to the road or at least to its intended course; so that the construction of the road would appear to be bracketed by that of the camps and must be associated with them. This inference helps to suggest a close date for the combined work: for on the one hand the road cannot have been built before the conquest of the Brigantes between the years 69 and 74, and on the other hand it is very unlikely to be later than the construction of the more northerly Stanegate which, in some form or other, must be ascribed to Agricola in or shortly after the year 80. A date within a very few years of A.D. 80 is thus indicated for the Reycross and Crackenthorpe earthworks.

Of the permanent forts which eventually sprang up along the new road, the first of any magnitude encountered by the traveller descending from Stainmore occupied the commanding site where the mediæval castle of Brough now rises above the Swindale Beck. But near the head of the pass the Roman road had already proceeded through or by a small stone-walled enclosure (about 150 ft. by 120 ft.), to-day known as Maiden Castle, whence Roman objects of late 2nd-4th-century date have been recovered. As a shelter for patrols, and as a signalstation situated at the furthest point conveniently visible up the road from Brough, this fortlet is self-explanatory (fn. 26); though how, if at all, this outpost communicated eastwards with the next fort at Bowes is less clear. A further signal-stage may indeed be represented by a small square platform, 30 ft. by 37 ft., within the Reycross camp. (fn. 27)

The fort at Brough—the Verterae of the Antonine Itinerary—has been too extensively mutilated to suggest more than that it was of relatively small size, apparently with a single ditch. Its internal buildings are almost unknown to us, but fragments indicate that they included at some period stone barrack-blocks and the usual colonnades of provincial Doric. Two sherds of Flavian pottery serve as a slight further hint that the Crackenthorpe road was already in being before the end of the 1st century; an inscription dates from the time of Septimius Severus; and both pottery and coins suggest that the site was still held in the last twenty years of the 4th century; otherwise the history of the place still awaits methodical excavation. But the evidence already available from haphazard digging is not without interest on the cultural side. A funerary inscription, now in the Fitzwilliam, is one of the comparatively few Greek inscriptions from Britain and, like the Greek gnostic charm from Caernarvon, brings home the essentially cosmopolitan character of remote frontier-posts as late as the 3rd or 4th century. A series of inscribed leaden seals from the fort or its vicinity has recently provided Mr. I. A. Richmond with the text for a discussion of the local disposition of troops in that area. (fn. 28) And, not least, an unusually extensive series of brooches from the site includes a number of the trumpetheaded and sometimes enamelled type which is familiar as illustrating the late flowering of a Romanised Celtic craftsmanship in Northumbria during the half-century (c. A.D. 100–150) following the pacification of the region. It seems likely that within the purlieus of the fort flourished one of the centres of this individual and attractive industry.

Twelve miles north-westwards from Brough, at Kirkby Thore, stood the next permanent fort; but it is possible that the interval was broken by one or more wayside posts of the type already noted at Maiden Castle. On the hill known, significantly perhaps, as Castrigg in the parish of Long Marton (13), nine miles from Brough and three from Kirkby Thore, the faint traces of an earthwork some 240 ft. long and of less certain width have been identified within 100 yards of the Roman road. (fn. 29) The ditch seems to have been about 40 ft. wide, measured from the centre of the bank to the outer lip, and a slight excavation in 1932 showed that it was upwards of 7 ft. deep below the present surface. No relics have yet been found here, but the position and the form of the work (so far as it can be traced on the surface) point to a Roman origin.

The fort at Kirkby Thore was the Braboniacum of the Itinerary and the Brauniacum of a Numidian inscription to which attention has recently been drawn. (fn. 30) To-day there is scarcely a vestige of fortification to be seen, though Roman stones are built plentifully into the modern field-walls, and buildings were dug up here in the 17th century, whilst the quarry from which their stones were presumably derived is known from the inscriptions that it formerly bore. (fn. 31) An outlying stretch of escarpment to the west of the main sites faintly suggests the possibility of a fortified annexe, or even of the former existence of a large camp of the type of the neighbouring Crackenthorpe; but, here too, all is uncertainty without digging. The "finds" are slightly more illuminating. Coins range from the latter part of the 1st century to the beginning of the 3rd, and their apparent deficiency in the 4th century is made good by a reference to the place in the Notitia Dignitatum. In general, therefore, the evidence tallies, as we should expect, with that from Brough. Unlike Brough, however, which can scarcely have held more than 500 infantrymen, Kirkby Thore is known at some period to have been garrisoned by a cavalry regiment or detachment, though its name is not preserved. The Numidian inscription already cited tells us that during its sojourn here this regiment was at one time under the command of a Publius Licinius Agathopus, who was a native of Gadiaufala in Numidia, where he died: a casual reference which suddenly and almost dramatically lifts the horizon of this remote Westmorland hill-top. Within the same enlarged horizon stands also a fragmentary altar dedicated to Jupiter Serapis, now preserved at Lowther Castle; but the more provincial and local aspects of the frontier-post are equally emphasised in the same collection by an inscription to a native deity Belatucadrus, and by tombstones barbarously and grotesquely carved with the traditional classical subject of a horseman spearing a foe, or with the equally traditional funeralfeast—the last on the tombstone of an officer's daughter who must have lived in the vicus within the environs of the fort (pl. 3). Only in one fragmentary equestrian tombstone is the workmanship of a higher order, not incomparable, in its accomplished if stilted and overprecise style, with the better classical tomb-sculptures of the 4th century (pl. 3). On the whole, as at Brough, the local craftsmen reached a higher achievement in the working of metal, and the certainty that metal was actually worked here is proved by lumps of fused ore now in the Kendal Museum. Numerous bronze brooches, preserved in the British, Tullie House (Carlisle) and Kendal Museums and in private collections at Barton and Morland, include several of the Romano-Celtic type already noticed at Brough, and in the same context may be added a remarkable bronze sword-pommel (Tullie House) ornamented with the bean-shaped swellings of mature Celtic art. An analogous pommel (British Museum) from Warton, Lancashire, stresses the northerly origin of this Celtic phase, which flourished, it seems, between the end of the 1st and the middle of the 2nd century A.D. Amongst the Kirkby Thore relics in Miss Cumpston's collection at Barton Hall are a dragonesque brooch of a familiar northern 2nd-century type, and two small bronze dogs in relief. The latter belong to an extensive series of similar figures which were sometimes (as at Lydney in Gloucestershire) used as votive offerings; but one of them, a cowering whippet looking backwards over its shoulder, is exceptional in the quality of its workmanship.

The location of a fort at Kirkby Thore was determined by the fact that hereabouts a road branched northwards to the forts of Whitley Castle and Carvoran. The point was thus a focal one, from which cavalry could readily control a considerable tract of the frontier zone in three directions. Apart from this special strategic factor the proximity of the site to the fort at Brougham (the Brocavum of the Itinerary), only 7 miles away, would have been puzzling. But the Brougham site itself requires no searching explanation; it conforms obviously and completely with the conventions of the Roman war-office. The fort lies beside a main road at a confluence of two rivers and at a road-junction. To this nodal site a large fort is appropriate, and the vestiges beside and partly under Brougham Castle are such as to imply an enclosure of about 4½ acres and the presence here of a garrison approaching a thousand strong. The unit may, at any rate for a time, (fn. 32) have been the obscure numerus equitum Stratonicianorum mentioned on an altar now at Brougham Hall; but a regiment of more regular type, a cohors equitata, is referred to on another inscription, whilst a third stone, a dedication to the Germanic mothergoddesses in the form Deabus matribus tramarinis, was set up by a vexillatio Germanorum, implicitly soon after its arrival from overseas. These and other inscriptions from Brougham have recently been collated by Mr. Birley (fn. 33) and need not be extensively cited here. It will suffice to observe that they indicate the presence of a vicus or civilian suburb to the east, along the flanks of the Kirkby Thore road, with a cemetery on the outer fringe; and that a number of the personal names mentioned on stones from the area of the vicus—Vidaris, Lunaris, Annamoris, Audagus— are non-Roman in character, suggesting an essentially barbarian element which is emphasised by four dedications to the native deity Belatucadrus. The cult of this deity, already noted at Kirkby Thore, is, as Mr. Birley points out, restricted to north-western Britain and seems, from the nature of the dedications, to have been of a somewhat plebeian kind, with an approximation to the cult of Mars. Altogether, the evidence is clearer here than at most forts that the opulation assembled outside the lines was predominantly "barbarian" in origin, comprising native elements variously modified by foreign (especially German) elements and united by a more or less uniform veneer of Roman culture.

As to date, the pottery found in late years on the site during reparations to the mediæval castle by H.M. Office of Works is partly of the 2nd century and partly of the 4th and even of the late 4th century; but so important a site can scarcely have been unprotected at the time of the lay-out of the two principal converging roads in the Flavian period, and an Agricolan fort on or near the site of the surviving work is a safe postulate. (fn. 34)

Either to Brougham or, less probably, to Kirkby Thore may be related an inscription, apparently of early 3rd-century date, commemorating the restoration of a bath-building (balneum). The inscription was found in 1886 during a restoration of the church of Cliburn, midway between the two sites, and is now built into the church-porch. (fn. 35)

From Brougham, after crossing the central highlands of the county, the road descends into the Lune valley, and presently passes the site of the fort of Low Borrow Bridge, set characteristically in the angle where the Borrow Beck joins the main stream. Nothing is known of the history of the site, although slight excavations have been carried out there. Its stone wall and projecting E. gateway, comparable with gates at Caerleon, the Brecon Gaer and elsewhere, presumably date from the 2nd century; but the site is of sufficient strategic value to suggest that, as at Brougham, the occupation may have begun in the Flavian period. Once more, a little careful excavation may some day settle this and other matters without great difficulty.

So much for the camps and forts of the first and second of our groups. The general course of the roads which they controlled is recoverable on the ground and even on the map, though only at a few points is the actual line both clear and at the same time free from modern metalling. The existing traces may be tabulated as follows:—

A. From Low Borrow Bridge to Brougham.

(i) About 1½ miles S.W. of Orton church, on Low Moor, is a length of road in good preservation, consisting of a slightly raised causeway about 20 ft. wide, with a ditch 3 ft. deep on each side.

(ii) A length of 2¼ miles, from Coalpit Hill to Dalebanks, running past the settlement of Ewe Close (above, p. xxxiii). Three sections dug across it by Mr. W. G. Collingwood during his excavation of Ewe Close showed that the metalling averaged 20 ft. in width between kerbs (C. & W. Trans., n.s., VIII, 357).

(iii) Thereafter, the road appears to have turned slightly westwards through the parishes of Sleagill, Newby and Great Strickland; a stretch 1¼ miles long in this area is still used and bears the name of "the street."

B. From Brougham to Brough and Bowes.

(iv) From Brougham through Kirkby Thore the Roman road coincides approximately with the present road, parting company from it near the ninth milestone from Penrith, E. of Crackenthorpe camp. From this point for just over 2 miles it runs through hedgerows as a kind of drove-road, and takes the form of a firm green lane averaging 20 ft. in width.

(v) Where it crosses the parish of Appleby it is used as a by-road and bears the name of High Street.

(vi) S.E. of Appleby there is a further stretch, two-thirds of a mile long, across cultivated ground before it rejoins the present road.

(vii) It does not reappear until it approaches Maiden Castle, in the vicinity of which it can be clearly seen. Almost certainly of Flavian date, it may have antedated Maiden Castle by at least a century and must originally have traversed the site of the fortification. When or after the latter was built, however, there is evidence that the road was looped round its northern side, leaving only a comparatively narrow passage through the fort itself. Thence it proceeded straight to the Reycross camp on Stainmore.

C. From Kirkby Thore to Carvoran.

(viii) A road, known as the Maydengathe or Maiden Way as long ago as the 12th century, (fn. 36) branched northwards across the moors from Kirkby Thore to Whitley Castle and Carvoran, and served as a median link between the transverse highways of Stainmore and Stanegate. Its exact course at Kirkby Thore has not been verified, but the general indication of its line would carry it down on the western side of the fort. Only 3 miles of the road lie within the limits of the county.

D. From Brougham towards Ambleside, called High Street.

(ix) Another mountain-road climbed on to the fells S.W. of Brougham and ultimately descended to the valley of the Troutbeck, 3 miles S.E. of Ambleside, (fn. 37) presumably joining there an unverified road or track from Ambleside to Watercrook (for which sites, see below). Its identifiable course is for the most part sufficiently marked on the Ordnance Survey maps. From the north it leaves the present main road at Tirril, 2½ miles S.W. of Brougham, and ascends to Burton Fell, whence it passes Loadpot Hill, Whether Hill and High Raise, and reaches the height of 2,600 ft. on High Street Fell. Its course is more circuitous than that of the normal Roman road, but rarely more so than the difficult condition of the ground dictates. Its specific identification as a Roman road goes back only to J. Hodgson. (fn. 38) It is clearly identical, however, with the Brethstrett, Brethstrede or Brethstrette which is named as a land-boundary in a 15th-century copy of a grant drawn up between 1220 and 1247 (fn. 39); and this name implies, on the one hand, that the track was at that time already of sufficient antiquity to be attributed vaguely to the 'Brettas' or Britons, (fn. 40) and on the other hand that it was already a metalled way (street) rather than a mere fell-track. The latter inference is important, since a metalled road at that time and in that place can only have been of Roman origin. Nor does the presence of metalling depend solely upon place-name evidence, for at four points the actual metalling is recorded to have been seen. In the southern stretches of the street, Cornelius Nicholson, the Kendal antiquary, "laid bare the Roman pavement in two or three places, a foot beneath the turf that now covers it" (fn. 41); and in Blue Gill Prof. Collingwood has observed "the remains of the metal and bottoming, about 8 ft. wide" at more than one point. (fn. 42) But the most extensive examination was carried out in 1898, when Dr. G. B. Grundy and Mr. W. H. Parkin cut sections across the Street at two points north of Loadpot Hill. Prof. F. J. Haverfield's summary of their results may be quoted in full (fn. 43) :—

"Two sets of trenches were dug. One was across a straight embankment, 40 yards long, by which the road traverses a slight depression, a little to the north of Loadpot Hill. This embankment is about 32 ins. above the depression and is just on a level with the surrounding surface: it is about 10 ft. wide at top, 15–16 ft. wide at base. When trenched, it was found to consist of four layers, first the surface and a layer of vegetable matter such as often occurs under old thick grass: then, a layer of gravel, like river gravel save that the stones were not rounded, 9 ins. thick on the crown of the road but tapering off at the side: thirdly, a layer of peat, 8 ins. thick, and lastly a layer of larger stones, rough but obviously quarried and more than two feet thick. Apparently this was the foundation, which was coated with peat and then received the actual gravel of the road. No trace of curbstones was detected anywhere at this spot.

"The other spot trenched was about two miles further north and very nearly 1,200 yards south of the point where the Ordnance Map takes the road over the Elder Beck. Here curbs were visible on the surface, stones roughly square on top and sides but not below, measuring on an average 9 by 9 by 5 ins. The curbs were 10–11 ft. apart on each side of the road: the road itself had been unfortunately much worn by packhorses and its exact construction could not be determined. The stones which lay about were mostly flakes 6 or 7 ins. long and 1 or 2 ins. thick: these are the natural cleavages of the rock. I have never noticed them on Roman roads in the North of England, but they correspond very closely to stones found by Mr. Grundy and myself last October in the lowest stratum of a Roman road in Blenheim Park, near Oxford.

"The quarried stones visible at both spots correspond exactly in character with the stone in a quarry called Loadpot Hole, on the north face of Loadpot Hill as one descends to Swarth Fell. This quarry, 1,800 feet above the sea and far removed from any house or work of men except High Street, seems almost inevitably to be the quarry from which, at some time or other, metalling was taken for this mountain road."

The evidence thus points to the building or adaptation of this mountain-road by the Romans, in spite of its high altitude and its constant shroud of mist. Whether its line was already that of a mountain-track in pre-Roman times cannot be said; it takes no obvious place in the topography of pre-Roman Westmorland. In post-Roman times it remained in use, (fn. 44) and in places it has been worn into a hollow way and augmented by subsidiary or alternative paths. In Hodgson's time the straight stretch on High Street Fell was used annually on July 10th for horse-races, on the occasion when "everyone brings the sheep that have strayed from their heathing, for their owners to challenge." (fn. 45) The summit is still known as Racecourse Hill, and the fell is hereabouts approached, not only by the tracks which represent the Roman road itself, but also by others from Haweswater, Patterdale, Helton and Martindale. These various uses have served partly to obscure and partly to perpetuate the vestiges of one of the most adventurous roads of Roman Britain.

E. From Ambleside to Hardknott (?).

(x) Reference may here be made to a 'corduroy' road identified in 1900 by excavation at a point some 500 yards N. of the N.E. angle of the Ambleside fort, and running northwards on the line of Borrans road, whence it presumably turned westwards to the fort at Hardknott. (fn. 46)

Four stones demand notice in connection with these roads. The first, bearing a dedication to Constantine I, was dug up in 1602, as Camden tell us, (fn. 47) "at the confluence of Loder and Eimot," i.e. at Brougham, and was probably a milestone or roadstone; a second roadstone, bearing a dedication to Philip (A.D. 244–9), was "dug out of the military way 1694" at a point unspecified but doubtless near the farmhouse of Hangingshaw, in Long Marton parish, where it was preserved until its removal to Tullie House Museum about 1915 (fn. 48); and another milestone bearing the inscription M.P. liii—possibly indicating, with sufficient exactness, the distance from Carlisle—has been erected near the main road in the parish of Middleton, where it was dug up in 1836. (fn. 49) The main road hereabouts closely represents the Roman road running from Low Borrow Bridge southwards to Overborough. A fourth stone which, from its form and position, was probably a milestone though it now shows no traces of an inscription, stands beside the main road in the parish of Temple Sowerby, N.W. of Appleby.

The two forts of our third group—Watercrook (near Kendal) and Ambleside—are reserved to the end because they command no arterial traffic-routes, and owe their existence to other causes. They are as spear-points thrust inland from the coastline, thrust as far indeed into the massif of the lake-country as was possible without loss of grip. Here if anywhere we are on a Roman North-west Frontier, amidst hill-beleaguered Chitrals beyond the bounds of the normal life and movement of the province.

The Watercrook site is now a meadow in a bend of the river Kent, a mile S. of Kendal. Vestiges of the fort are slight, but indicate an area of 2½–3 acres, and some hint of the normal street-plan has been obtained from the sun-dried grass. Excavation has shown that the rampart had at some time a stone wall about 5 ft. thick, and that the N.E. gate had the usual double entry and square guardrooms. Outside the fort the bath-building was found in the 17th century, apparently on the site of Watercrook farm house, and other buildings have been identified vaguely to the N. and W. of the fort, where the civil settlement evidently stood. A tombstone bearing the name of a P. Aelius Bassus was re-used in one of the structures. Pottery from the site ranges from the time of Hadrian to the middle of the 4th century or later, with a faint hint (a Samian sherd signed Albini m.) of a Flavian occupation. If the Xth Antonine Iter is correctly identified with a route starting from Ravenglass and joining the arterial road to the north either at Overborough or at Ribchester, then Watercrook may be identified with Alone; but the equation of the route and the Iter is not free from difficulty, and the identification falls short of certainty. (fn. 50) If it be correct, the further identification with the Alione of the Notitia follows, with the added information that at the time of the compilation of the Notitia (c. 4th century) the fort was garrisoned by the Cohors III Nerviorum.

If Watercrook was Alone or Alione, then Ambleside was Galava. This doubtful detail apart, the history of the Ambleside fort is clearer than that of any Roman site in Westmorland. The results of its excavation by Professor R. G. Collingwood in 1913 and the following years have entered into the familiar fabric of Romano-British archaeology, and the summary of the structural evidence, given below in the Inventory (Ambleside 1), need not be anticipated here. The site was fortified in the first flush of Agricolan victory, c. A.D. 80, and, as often, the sure strategic instinct of Agricola is witnessed by the re-occupation or reinforcement of the spot when the northern frontier was consolidated under Trajan and his successors. The early fort is marked from others of its age by its unconventional use of a rocky knoll as a natural cornerbastion—a piece of bold adaptiveness refreshingly rare in the works of the Roman war-office. And yet scarcely less impressive, perhaps, is the almost insolent indifference with which the later fort ignores the rock, passing beneath it as though its impending proximity were of no moment to the imperium of 2nd-century Rome. Only the uncharitable observer might prefer to accuse these later engineers of an unpliable and mechanical adherence to the war-office pattern-book, with a thick-headed obliviousness to local circumstance. Be that as it may, the two forts of Ambleside stand out amongst the scraps and pieces of Roman Westmorland as something coherent and intelligible and even human; and with the intrusion of humanity our catalogue may timely cease.

POSTSCRIPT

Attention may here be drawn to a number of small and anomalous structures, not referred to in the preceding pages and difficult at present to classify. They may all be of mediæval or even later date, but an earlier origin has been suggested for some of them. They are as follows.

1. A small oblong building with rounded corners, together with other foundations possibly representing a cattle-yard, impinges upon the plan of the hut-settlement of Ewe Close (Crosby Ravensworth 25) and has been compared with "early Teutonic houses of well-known types." The elementary character of the plan robs this or other comparisons of any determining value, and, beyond the likelihood that the buildings are those of a small 'hafod' or 'shieling bothy' or upland cot occupied by cattle-tenders during the summer months, conjecture is at present unlikely to be profitable.

2. A small oblong building with a flat three-sided 'apse' at one end and a porch or annexe at the other is represented by foundations at Cow Green in the same parish (Crosby Ravensworth 36). Remains of similar buildings are visible in the vicinity. Here again Scandinavian analogies have been cited, but excavation is obviously desirable before further discussion. For 1 and 2, see C. & W. Trans., n.s., XXXIII (1933), 206 and 211. See plan, p. xlvi.

3. A similar building can be identified in Glencoyndale (Patterdale 65). See plan, p. xlvi.

4 and 5. In the parish of Troutbeck (35) are foundations of two structures, within 100 yards of each other, situated like the preceding examples on the desolate uplands. Other fragmentary foundations can be traced in the vicinity. The larger of the two main groups appears to represent three small rooms or pens with a central passage. The other is a small oblong building, about 40 ft. by 16 ft., with a double partition, presumably representing a transverse passage, slightly nearer one end than the other. The simple character of these structures is again a bar to their closer identification from the superficial remains. A central opening or passage is characteristic of certain small farmhouse-sites which, in Glamorgan, have been ascribed to the sub-Roman 'dark ages,' (fn. 51) and the feature recurs in a building at Tintagel, Cornwall, attributed provisionally to the 5th century (fn. 52); whilst yet another building, found long ago on Trewortha Marsh, near Tresillian in the same county, with Roman or sub-Roman sherds, appears to have been of a comparable kind. (fn. 53) These possible analogies, however, do not define the date of the Troutbeck structure in the absence of excavation. Attention was drawn to the foundations in C. & W. Trans., n.s., I (1901), 133. See plans, p. xlvi.

R. E. M. WHEELER.