Short URL for this page:

[image ALT: Much of my site will be useless to you if you've got the images turned off!]
Bill Thayer

[image ALT: Cliccare qui per una pagina di aiuto in Italiano.]

[Link to a series of help pages]
[Link to the next level up]
[Link to my homepage]

[image ALT: link to previous section]
This webpage reproduces a chapter of
Roman Era in Britain

by John Ward

published by Methuen & Co. Ltd.
36 Essex Street W. C., London

The text, and illustrations except as noted,
are in the public domain.


[image ALT: link to next section]

 p38  Chapter III


The military remains may be divided into 'temporary' and 'permanent.' To the former division belong the various field-works raised during campaigns, whether to hold an army during a halt of a few days or to serve as its winter quarters, or a small detachment of the same charged with keeping open communications with its base, guarding some point of strategic importance, or affording protection to labourers engaged in road-making. To the latter division belong the great legionary centres of York, Chester, and Caerleon, the numerous stations of the garrisons which maintained order and defended the frontiers, and the great frontier lines of Hadrian and Pius; and to these may be added the walls of towns. Broadly speaking, the works of the one set are distinguished by their slight construction, often so slight as to be scarcely discernible, and by the absence of buildings within their defensive lines; whereas those of the other set rank among the most conspicuous and notable remains of Roman Britain.


As the visible remains of Roman entrenched fieldworks are comparatively few and little is known of them, it will be helpful to hear what Roman writers have to say about the art of castrametation.

Two writers, whose works have come down to us, are preeminent  p39 for the fulness of their descriptions of Roman camps: Polybius,​a the friend of the younger Scipio, in the second century before our era, and the author of a treatise, De Munitionibus Castrorum, who is usually called Hyginus, and who flourished about the close of the 2nd century. The camp of Polybius was simple and symmetrical (Fig. 7). The site being selected, the position of the general's tent was fixed, and from this the whole plan was developed. Through this point a line — the decumanus maximus — was drawn, and at a certain distance it was crossed by another — the cardo maximus — at right angles. These served as the base-lines from which the general outline and internal divisions were determined. The resultant figure  p40 was a square, 2150 Roman feet each way, bisected in its 'length' into two equal divisions by the decumanus maximus, and in its 'depth' into two unequal divisions by the transverse cardo maximus. Along these lines ran the two chief thoroughfares, each passing through the rampart. The transverse thoroughfare was known as the via principalis, and the great square of the praetorium, containing the general's tent, occupied the middle of its side next the back of the camp. This square necessarily broke the continuity of the longitudinal thoroughfare, and that portion between it and the front of the camp was known as the via praetoria.º A number of by-ways contributed to divide up the interior into rectangular plots for the tents, and around all, within the rampart, was a clear space or intervallum, 200 ft. wide, to facilitate the drawing up of the troops in marching order. The rampart itself was usually formed of the upcast from the ditch which constituted the outer defensive work.

[image ALT: missingALT]
Fig. 7. — Plan of Polybian Camp

Polybius mentions neither the number nor the names of the gates; but it may be incidentally gathered from Livy and other writers that they were normally four, and were known as the portae principales (dextra and sinistra), the porta praetoria, and the porta decumana or quaestoria. Such a camp as described above was for a consular army consisting of two legions, and if there was need for two of these armies to be encamped within the same lines, Polybius directs that two such camps should be applied back to back with the intervening ramparts suppressed, the result being an oblong enclosure with six gates.

When the treatise attributed to Hyginus was written, some three centuries later, the military system had greatly changed, and, as might be expected, the Hyginian camp reflected the altered conditions. To us, this form of camp is of peculiar interest, as our Roman camps and forts are more akin to it than to that of Polybius. The Hyginian camp (Fig. 8) agreed substantially with that of Polybius. The chief differences were its oblong form with rounded corners, the narrower intervallum, the elongated praetorial space, and the altered disposition of the troops and smaller space they occupied, the last being all the more significant of the altered status of the common soldier under the empire, for while the number of men was nearly  p41 double, the accommodation for the officers had increased three-fold. The two transverse roads divided the Hyginian camp into three segments — the praetentura to the front, the retentura to the back, the middle space being the praetorium and its latera, in which were quartered the general and his chief officers.

[image ALT: missingALT]
Fig. 8. — Plan of Hyginian Camp

In a graphic sketch of a Roman camp, Josephus describes "the towers at regular distances"; the four gates, "one at every side of the circumference," wide enough "the entrance  p42 of the beasts" and "for making excursions if occasion should require"; the rampart like a wall; and "the engines for throwing arrows and darts, and for slinging stones." Within are streets and tents, those of the commanders being in the middle, and in the midst of all the general's own tent, "in the nature of a temple"; a market-place and "place for handicraft trades," and a court of justice. So rapidly and orderly is all accomplished that it is like "a city built on the sudden!"

The value of these literary descriptions will be best appreciated when we consider the remains of our forts, for in the case of our camps it is only in their defensive lines that they can be compared with those of the ancient writers. In some cases the agreement is close: more often it is more or less remote. A few exceed the sizes of Polybius and Hyginus, but the majority are less, and the positions of the gates are often different. The remains of these field-works are unevenly distributed, being of rare occurrence in the lowlands of England and comparatively frequent in the less cultivated regions of the north — a distribution due in some measure at least to the unequal advance of agriculture.

About thirty northern examples were surveyed by General Roy a century and a half ago, and his elaborate plates and notes still remain the chief work on the subject, in spite of some inaccuracies as to dimensions and uncertainties as to details. Of these, eighteen scattered from Aberdeenshire to Northumberland were attributed by him to Agricola, whether rightly so little matters: it is sufficient to observe that they all appear to have the impress of one design and period. These camps are normally oblong in shape, but many are oblique, and some irregular. Their defences are slight, consisting of a small rampart or parapet and ditch. Their entrances are guarded by slight traverses — a characteristic of the Polybian camp, and their number, as shown on the plates, range from one in the smaller camps, to four, five, or six in the larger. But a comparison of their positions leaves little room for doubt that in all, except the smallest, the original number was six, one at each end, and two on each side (A, B, Fig. 9). The six entrances recall the double Polybian camp — the two consular camps  p44 combined in one — but they could not have arisen from the same cause, as some of these camps are vastly smaller than the double Polybian. The sizes vary greatly. Three range from 116 to 130 acres each; one is of 86 acres; seven range from 50 to 58 acres; and the rest from 6 to 42 acres. Since Roy's day more of these 'Agricolan' camps have been noted. There are about eight in the vicinity of the Wall of Hadrian, and two of these near Haltwhistle Burn were trenched in 1908.

[image ALT: missingALT]
Fig. 9. — Plans of Roman Camps. A, Towford; B, Raedykes; C, Chew Green; D, Pigwn; E, Rey Cross; F, Dealgin Ross.
(Approx. 800 feet. to 1 in.)

Leaving out certain small posts, the other camps described by Roy are of smaller sizes than most of the above, more symmetrical and as a rule of stronger construction, but they especially differ in their entrances. In several they are apparently simple unguarded openings; but in most they are covered by curved guards or traverses, joined to the rampart at one end (D and F), the advantage of this arrangement being that the defenders on the traverse were not isolated, but could pass at will from the rampart. Some of the camps have four entrances, and the smaller have three or two. But three are remarkable for their number and distribution, a camp at Rey Cross in Westmorland (E), for instance, having apparently eleven, three on three sides and two on the fourth, and another at Birrenswark, three on one side and one on each of the others. A precisely similar arrangement to the last may be seen in a large camp at Ratby in Leicestershire.

A camp on a well-chosen site was likely to be reoccupied by the army on its return or by another marching along the same line. If, however, the second comers were more numerous or fewer than the first, the general rule was to make a new camp. The smaller of two camps at Pigwn in Breconshire is within the larger, and it is hard to understand why two sides of the larger were not utilized for the smaller, as in A. A curious example is at Ardoch, where two camps intersect one another, and the constructors of the second, whichever it was, did not trouble to level those portions of the first which lay within its lines.

General Roy gives plans of a number of small strongly entrenched posts ranging from about 60 to 160 ft. square, and mostly with one entrance. Several are associated with his 'Agricolan' camps; others appear to be quite isolated. From  p45 their strength — they all have several ditches — it is reasonable to think that they were intended for a more or less protracted occupation. That their use was to keep open communications between the army in the field and its base and to overawe the conquered territory, is equally reasonable.

There is a good example of the Roman adoption and modification of a native camp at Hod Hill in Dorset. The Romans cut off a rectangular portion within the north-west corner, utilizing the old lines for the north and west sides, and completing the enclosure by their own, on the south and east. The remains were partially destroyed many years ago, when many Roman relics were found, including coins ranging from Augustus to Trajan.


We now tread upon firmer ground. The sites of the garrison stations are usually well-defined and easily recognized. The ridges of their ramparts, whether of earth or of built-stone, are frequently conspicuous. The ditches are rarely filled to such a degree that their hollows are not visible. The positions of the gates generally show as breaks in the continuity of the ramparts. If the interiors had not been subjected to the plough, the lines of the chief thoroughfares and the sites of the buildings may often be traced; and now and again these surface-indications may be sufficiently pronounced to admit of plans showing all the salient features. Their distribution is, as stated on page 8, uneven. There is evidence that the forts were not so strongly constructed at first as was customary at a later date. The coast-forts, as a rule, have certain peculiarities, not wholly confined to them, however, which represent a departure from traditional lines and a development in the art of fortification. One of these peculiarities is the presence of bastions, and for this reason we will call them 'bastioned forts,' and distinguish the ordinary type as the 'Hyginian.' This type we will consider first.

Many sites of forts of the latter type have been systematically explored, wholly or in part, during the last quarter of a century. Four of these are notable for the complete plans  p46 they have yielded, — Housesteads one of the Wall forts, Birrens in Dumfriesshire, Gellygaer in Glamorgan, and Newstead in Roxburghshire. These are closely followed by Rough Castle, Castlecary and Bar Hill on the Antonine Wall, Ardoch in Perthshire, Chesters and Great Chesters, two Wall forts, and High Rochester and Haltwhistle in Northumberland. Excavations at Camelon in Stirlingshire, Lyne in Peeblesshire, Birdoswald on the Wall, Hardknott in Cumberland, Ribchester in Lancashire, Castleshaw and Elslack in Yorkshire, Wilderspool in Cheshire, Melandra Castle and Brough in Derbyshire, Caersws in Montgomeryshire, and Coelbren in Glamorgan, have yielded less, but still valuable results. These investigations, as also many on the  p47 Continent, have proved that with the exception of the bastioned forts, the Roman garrison stations were all of one pattern, although differing in details, and that this pattern was substantially that of the Hyginian camp. In fact, we may regard them as translations of that camp into stone or other durable materials, provided they are looked upon as free, and not as literal renderings.

They are symmetrical, usually longer than broad, with the corners rounded off, and four or exceptionally six gates. They have usually one or two ditches, but if one or more sides are more vulnerable than elsewhere, there may be more. Their planning recalls that of the Hyginian camp, presenting two principal streets arranged cross-wise and stretching from gate to gate. The continuity of the longitudinal street is similarly broken by a central building, which has on either side others of diverse shapes, the whole range corresponding with the Hyginian praetorium and its latera, and similarly dividing the rest of the interior into a praetentura and a retentura. The buildings in these divisions are mostly of long and narrow shape, and they recall the strigae of the tents in the camps. The plan of the fort at Gellygaer (Fig. 10) well illustrates these various features, and it is all the more useful for preliminary study, as it is simple and free from confusing alterations and re-buildings.

[image ALT: missingALT]
Fig. 10. — Plan of Roman Fort, Gellygaer. (100 ft. to 1 in.)


The ramparts vary considerably. In the 'earth forts,' of which Birrens and Ardoch are notable examples, they are usually of great size and width; but the term 'earthwork' fails to express their intricate construction. There is usually a pavement-like foundation of stones, or, as at Coelbren, a 'corduroy' of logs. The rampart itself is more or less stratified, seams of clay, earth, gravel, and decayed sods being of common occurrence, as also bonding-courses of branches or brushwood. In several instances, and perhaps in all, the face was of clay. Ramparts of sods and turves laid in definite courses are not uncommon, and may be regarded as a connecting-link between earthwork and masonry. Those of Rough Castle and Bar Hill are good examples,  p48 and like that of Birrens, rest upon stone bottomings. The Antonine Wall is of the same construction, only on a larger scale.

In the 'stone forts' the face at least is of masonry, serving as a strong retaining-wall for an earth-bank behind. Gellygaer furnishes a good example of one of these 'composite' ramparts. The wall is from 3 to 4 ft. in thickness, and the bank behind, including a thinner retaining-wall at the feet of its slope, makes up a total rampart-width of about 20 ft. The material of the bank is derived from the ditch and the trenches for the foundations of these walls. The rampart of Caerwent, which, however, was a town, not a castellum, is similar, but on a larger scale. Its wall has a thickness of nearly 11 ft. at the base, and in one place, where it remains to the height of 24 ft., this is reduced to 6 ft. 6 ins. at the summit by off-sets on the back, the front being, as is usual in Roman work, vertical. The bank attains almost the same level. In most other 'stone forts' of the Hyginian type, the bank is less conspicuous or is apparently absent. At Housesteads, the wall is somewhat thicker than at Gellygaer, and remains to a greater height; and there are scarcely any perceptible traces of a bank. But here, as also at Chesters and Great Chesters, are indications that the bank was reduced or removed in Roman times. At Caerwent, there is evidence that the rampart was originally of earthwork only, the wall being a late addition, and this may have been frequently the case;  p49 at Gellygaer, on the other hand, the masonry appears to have immediately followed the throwing up of the bank. As time went on, more reliance seems to have been placed in walls of masonry — thicker, loftier, and stronger; and in some of the later bastioned forts, the wall alone appears to have intervened between defender and assailant, there being apparently neither ditch in front nor bank behind.

[image ALT: missingALT]
Fig. 11. — Section of Ditch and Rampart at Gellygaer, showing restoration of latter. (15 ft. to 1 in.)

The ditches are almost invariably of an open V‑shaped section, with an average width of 19 or 20 ft. and depth of 6 or 7 ft., and there is always an interval or berm between the ditch and the rampart of a few feet or more, the chief use of which was to ensure the safety of the latter by giving it a firmer foothold. A single ditch was often deemed sufficient, and perhaps as often there were two. Where the defences were naturally weaker than elsewhere, there were sometimes more; for instance, the more assailable end at Birrens is sheathed with five additional ditches, and the corresponding end at Ardoch presents a remarkable complex of intricate ditches and ravelins. The upcast from a ditch was sometimes partly or wholly used to form a low glacis-like mound along the outer side, the object of which was apparently to increase the height of the counterscarp, but never to such an extent as to afford cover for the enemy. In the walls of Antoninus and Hadrian the whole of this upcast was so utilized.

 p50  Turrets were a usual feature of the 'stone forts' and probably also of the 'earth forts.' The remains of their basements are well seen on the Gellygaer plan, attached to the inner side of the wall, and with doorways to the intervallum; but in many forts they were confined to the corners. Ancient writers, as Josephus, for instance (page 41), refer to them in connection with fortifications; and they are represented on the Column of Trajan, two of which are shown in Fig. 12, the one within the rounded corner of a fort and apparently constructed of wood and roofed, and the other a stone one with a flat top. Both stand a storey above their respective rampart-walls and are entered from the parapet walks by doorways. No traces of kindred structures have been found in the 'earth forts,' but if, as is probable, they were of timber their remains might easily escape detection.

[image ALT: missingALT]
Fig. 12. — Fortification Turrets on Column of Trajan

The gates varied considerably: some were of stone, some of timber, and a few of both materials; some had a single passage, others two; and the larger were provided with guard-chambers. The gates of the 'stone forts' of the Hyginian type were normally double ones, that is with two passages each, side by side, and between two guard-chambers which did not project beyond the face of the rampart. The gates at Housesteads and Birdoswald (Fig. 13) are typical examples. The portals were arched and  p51 were provided with doors of two leaves, the lower pivots of which turned in iron-sheathed stone sockets in the angles between the jambs and the side walls of the passages. Those with a single passage were of similar arrangement and construction, but they rarely had guard-chambers. The smaller gates at Birdoswald and Chesters were of this character, and simpler ones may be seen in several of the mile-castles of the Wall of Hadrian. It is noteworthy, however, that the gates of the Scottish forts seem invariably to have been single-passage ones with or without guard-chambers, and the great castellum of Newstead is no exception in this respect.

[image ALT: missingALT]
Fig. 13. — Plans of Gates, Housesteads and Birdoswald. (30 ft. to 1 in.)

These structures are in too ruinous a condition to supply definite information as to their upper work, but the sculptures of the Column of Trajan will again be helpful. In Fig. 14 are shown four examples of gates therefrom. The first two, it will be noticed, have no upper chambers, and in the second of these  p52 is shown the wooden parapet of the continuation of the rampart-walk over the gate. The second two have upper chambers with window-like openings, and lateral doorways from the rampart. The arched openings of the fourth example imply a stone structure. There are several Continental Roman gates — notably at Rome,​c Turin, Verona, Autun and Treves — which still retain their superstructures, and their façades present one or two storeys of arched openings of considerable size over their portals.

[image ALT: missingALT]
Fig. 14. — Fortification Turrets on Column of Trajan

A gate-building on a mosaic in the Avignon Museum so elucidatively fits in with the remains of the double gates described above, that we reproduce a sketch of it by the late Mr. C. Roach Smith (Fig. 15).

[image ALT: missingALT]
Fig. 15. — Gate of Town or Fort, from Mosaic in the Avignon Museum

The gates of the Scottish earth forts were wholly or mostly of timber, and their remains are consequently slight and often indefinite. In most cases they appear to have had single passages, and masonry, if used, was confined to their sides, perhaps more for the purpose of retaining the ends of the rampart than anything else.

The approaches to the gates varied. The ditch was either continued in front of the gate and was crossed by a bridge, or it stopped short on either side, leaving a causeway-like approach. Gellygaer provides an example of each. The sides of the ditch in front of the south-west gate are stepped out evidently to receive two rows of supports of a timber bridge; while in front of the south-east gate the ditch is simply discontinued. The causeway approaches were usually simple and direct; but in some of the Scottish forts they were devious, and at Ardoch  p53 there is evidence of timber palisades along their sides and transverse structures to prevent the entrances being rushed.

The bastioned forts differ from those of the type described above, not only in having bastions or external towers, but in their thicker and loftier walls, and their gates being fewer, more strongly defended and of a single passage each — these apparently never exceeding two in number, any additional entrances being posterns. These forts also show a decided tendency to depart from the traditional rectangular form. They certainly indicate a change in the principles of defence. The above modifications had a twofold effect: they increased the passive resistance against attack by the greater strength of structure and the restriction of entrances, and they increased the active resistance by providing means of enfilading both walls and gates by the introduction of bastions. The remains of castella of this type may be seen at Burgh Castle, Bradwell-juxta‑Mare, Richborough (Fig. 16), Lympne,º Pevensey, Porchester, Bitterne, and Cardiff — all coast-forts, the first six or seven belonging to the series which about the close of the 4th century was under the control of the 'Count of the Saxon Shore.' The bastions vary in shape and projection. At Cardiff and Richborough they are of slight  p54 projection, those of the former being polygonal, and of the latter rectangular, with circular ones capping the corners. At Burgh Castle, Lympne,º Porchester, and Pevensey, they boldly project and have rounded fronts and straight sides. These castella are undoubtedly late, and can hardly be assigned to an earlier date than the latter part of the 3rd century.

[image ALT: missingALT]
Fig. 16. — Plans of Roman Forts, Porchester and Richborough. (300 ft. to 1 in.)

The walls of some inland forts and towns had bastions — the multangular tower at York is a well-known example — but in some of these, as at Caerwent, the bastions were added to work of an earlier period.


The chief building in a Roman fort was a central one, which is generally known as the praetorium, also as the forum from its forum-like planning. There is no evidence, however, that either was its ancient name. It is probable that it was known as the principia; for inscriptions recording the erection or restoration of principia have been found on the site of the central building at Rough Castle, and its vicinity at Lanchester. In any case, it can safely be called the headquarters, for such it certainly was.

[image ALT: missingALT]
Fig. 17. — Plan of Headquarters, Chesters.
(50 ft. to 1 in.)
The plan (Fig. 17) of the headquarters at Chesters is typical of the larger buildings of the kind. A wide doorway gave access to its yard, nearly 50 ft. square, paved, and surrounded with a stone gutter, while in one corner was a well. On each side of this yard was a wide portico supported on square piers, and next the street, a narrower one or passage with openings to the yard, probably arched. The pavement of the porticoes was slightly higher than the yard. Along the back was the front wall of a second main division, having five openings, all probably arched, the end ones being somewhat smaller and providing direct access from the side porticoes. The transverse space behind was also paved, but it lacked a marginal gutter. It had on its nearer side a portico or aisle supported on four piers, and at each end of this was a side-door into the building. On the farther or opposite side of the space were five rooms or offices, of which the middle was the largest, and this and the two adjoining rooms had wide openings, all probably arched. The end rooms were entered from  p55 the contiguous ones by doors in the intervening walls. In the middle room were steps descending into an arched vault under the room on the left; while the corresponding room on the right had a central square of flagstones.

The headquarters at Gellygaer was smaller but simpler, the chief differences being the absence of a portico or aisle in the transverse space behind the yard, of side entrances, and of a vault at the back. In the first two differences, the present example is typical of most of the smaller buildings of the kind.

We can in some measure reconstruct one of these buildings. The yard was certainly open to the sky, and it usually contained a well. The porticoes had pentice roofs sloping into the yard, and the gutter below caught the rain-water from the eaves. The transverse space behind is generally regarded as an inner courtyard; but there are reasons for thinking that, in this country at least, it was roofed, one being that in no instance has it a marginal gutter. The offices at the back were normally five. The middle room was the most important, and when its remains are sufficiently intact, they invariably show a wide opening to the cross-hall. This opening at Housesteads retains its sill, which is chased to receive the bottom of stone or timber parapet or screen with a central gate or door. In most forts, this room has a vault or other under­ground receptacle as at Chesters, and there is good evidence that this structure was of late introduction. The two adjoining rooms, when sufficiently defined, have also wide openings, too wide to have been fitted with doors, and probably they had screens or parapets. The end rooms, on the other hand, were entered by narrow doors from the contiguous rooms or from the cross-hall.

 p56  The resemblance of these headquarters' buildings to a common type of forum-group in the towns, of which that of Silchester is a good example, is noteworthy. The yard with its porticoes is the counterpart of the forum-proper; the cross-hall, that of the basilica; and in both there is a range of rooms at the back, the basilica, however, usually having one at each end as well. It is well known that some of these rooms were fenced off from the hall by screens (cancelli) of wood or other material, and that they were used as courts for the administration of justice and for other public purposes; this is also a common arrangement for shrines or sacella. There is practically no doubt that the corresponding rooms in the forts were used for administrative purposes, or, as we should say, were orderly rooms, the middle one being the sacellum, the place where honours were paid to the genius of the emperor and of the regiment, and to the standards. The shrine would, according to ancient usage, be the treasury — hence the vault or other under­ground receptacle found in most of them. It is interesting to note that almost invariably these under­ground structures are of late work, and from this it would seem that in the declining days of the empire, the growing lawlessness necessitated a stronger protection for the treasure (probably kept in a chest in earlier times) than that afforded by the sanctity of the spot.

Near the headquarters was another important building, in some of the larger forts, two. These buildings varied considerably, but all of them had a house-like plan, and for this reason they have been identified as the residences of the commandants of their respective forts, and may have included rooms for the chief members of their staffs. They usually consisted of a number of rooms gathered round a small courtyard, an arrangement which is well seen on the Gellygaer plan. In some of the larger forts, one or more of their rooms were heated by hypocausts, and in one or two instances baths were attached.

Another essential was one or more oblong buildings of very distinctive character and remarkable for their thick and buttressed walls. Two at Gellygaer, one near each lateral gate, will be easily recognized on the plan. Two was the usual number, but occasionally there was only one in a fort. Rarer still there  p57 were more than two — at Birrens there were three, and at High Rochester, four. They were built singly or in pairs. They varied in length considerably, from 54 ft. at Hardknott to 130 ft. at Newstead, but rarely exceeded the limits of 22 ft. and 25 ft. in width. A comparison of the remains shows that these buildings had two features in common — a raised or suspended floor supported on dwarf-walls or pillars, and openings in the side walls below their floor-levels; but the actual floors have disappeared except at Corbridge, where they are of flagstones spanning the intervals between the dwarf-walls. The remains of doorways have been found in several instances, and invariably at one or both ends of the buildings. The latter was the case at Gellygaer, where their positions are indicated by the remains of porches; but in most, these appendages were lacking.

Little can be inferred as to the superstructures. There is no evidence that the buildings were divided into rooms above the floor-level, or that they had a second floor, although the buttressed walls could have carried one. It is probable that the buttresses were carried up to the roof, and that the intervening walls were pierced with openings for the admission of light and air similar to those below the floor. The object of these lower openings was to keep the floor dry by the free circulation of air under it. That these curious buildings were granaries can hardly be questioned. On the site of one of the pair at Corbridge was an altar dedicated by the praepositus of the horreum. Inscriptions have been found in Roman forts — one in this country at Great Chesters — recording the restoration of horrea, and on the site of the building of this type at Camelon was a large quantity of charred wheat.

On the plan of Gellygaer will be noticed six long L‑shaped buildings, four in the praetentura and two in the retentura. They were approximately 145 ft. long, and their recessed sides had verandas or porticoes supported on timber posts. Several buildings of similar shape have been partially uncovered at Chesters, which differed, however, in having stone columns instead of posts, and in being divided into rooms by stone walls — the 'heads' into several of unequal size, and the 'limbs' into a series of equal size, each with a door to the portico. At Housesteads  p58 and Birrens, the corresponding buildings were of a long oblong shape divided into eleven or twelve rooms of equal sizes, and at the former place there are indications that each was subdivided into a front and a back room. At Newstead, these buildings are represented by rows of eleven huts, each row being about 190 ft. long. There is little doubt that the Gellygaer buildings were divided into rooms as at Chesters, but by timber partitions.

These buildings, whether L‑shaped or oblong, were certainly barracks. They recall the arrangement of the tents in the Hyginian camp. There, to each century, which at the time consisted of eighty men with a centurion and petty officers, was allotted a row of tents — ten for the men, and two, or a space equal to two, for the officers, the total length of the row being 120 ft. Usually two of these rows were placed face to face with a space between, the whole forming a striga; while a single row constituted a hemistrigium. At Gellygaer, there were two of the former and two of the latter. At Chesters and Housesteads, the number of rooms in each block, and at Newstead, the number of huts in each row, approximate to the number of tents in the hemistrigium; and in the first it is reasonable to think that the centurion had his quarters in the 'head,' and that some of the rooms there, were offices and one possibly the shrine of the century. Assuming that each block at Gellygaer provided accommodation for a century, the six would represent an ordinary cohort, cohors quingenaria. At Housesteads, there were ten blocks which can be reasonably identified as barracks, and we know that its garrison — the First Cohort of Tungrians — was one of those entitled miliaria, nominally a thousand strong, and consisting of ten centuries. At Birrens, the plans of the buildings are less perfect, and this may be feasibly explained by the fact that the garrison — the Second Cohort of Tungrians — was not only miliaria but equitata, that is, it included a small detachment of horsemen, for whose use stabling as well as additional barracks would be required.

On all the more complete plans of forts may be noticed other structures which cannot be classed with those described  p59 above, but we can only conjecture their uses. Each fort was the scene of many necessary operations — the cornº had to be ground and the daily food prepared, and there must have been repairing shops of various kinds, as smithies, armouries, joineries, and so forth, and most if not all of these operations would require suitable buildings. In cavalry forts, and those containing both infantry and cavalry, the stables must have been an important element; and perhaps in most of the infantry forts, a few horses were kept for scouting and dispatches — and horses imply the storage of fodder. Among the minor structures would be latrines, cisterns for the storage of water, ovens and other cooking arrangements, wells, drains, etc.

Of the arrangements for the preparation and cooking of the food for the garrisons, little is known. At Birrens, the remains of a row of four oven-like structures were found on the inner side of the rampart near the east gate. They may have been ovens, or they may have been fitted with cauldrons for boiling purposes. Similar structures have been noticed at Newstead, Housesteads, Great Chesters, and Birdoswald.

The streets of the forts were mostly of gravel; less frequently they were paved with cobbles. The larger usually had a gutter on either side; the smaller, sometimes only one along the centre. The drainage was always well considered and carried out. The larger drains had built sides and were covered with large slabs, the floors being often paved, and the smaller were often constructed of flagstones.

A plentiful supply of water for drinking and cleansing purposes was one of the first considerations. A well has been found in most of the excavated forts, in each case in the headquarters' yard, but this seems to have been a precaution to ensure a supply of water in time of stress, the normal supply being derived from without. At Great Chesters, for instance, there are the remains of a small canal or aqueduct which conveyed water from Haltwhistle Burn, five miles away; and at Birdoswald a culvert brought the water of a spring some hundreds of yards away to a large cistern near the centre of the fort. At South Shields and Chesters are inscriptions recording the construction of aqueducts. As at Birdoswald, so in several  p60 other forts, large and well-constructed stone cisterns or tanks to receive water have been found. Remains of latrines have been uncovered at Housesteads, Castlecary, Bar Hill, and Gellygaer, in such positions that they could intercept waste water and street drainage for flushing purposes.


In the vicinity of many Roman forts may be seen the remains of buildings and other indications of an human occupancy. Housesteads is a notable example. To the east and south are the foundations of streets and houses which were more conspicuous a century ago. Altars, statues, columns, and carved stones have been turned up from time to time, and tell of temples (one of which, a mithraeum,º has been excavated), shrines, and other goodly structures. There is no doubt that the suburbs of Borcovicus were of considerable extent, and sheltered a considerable population, which presumably would consist largely of the soldiers' wives and families, time-expired soldiers, traders, and other civilians who served the garrison in various capacities. In the vicinity of other forts along the line of the Wall of Hadrian may also be discerned the indications of buildings. Hard by those of Chesters and Great Chesters are the ruins of extensive baths; and similar remains may be seen, or have been revealed by the spade, close by the Roman forts at Camelon, Newstead, Slack, Binchester, Gellygaer, and elsewhere.

Some of the forts had attached to them enclosed spaces or annexes, fortified, but, as a rule, less strongly so than the forts themselves. Rough Castle, Castlecary, and Gellygaer, had one each, defended by a ditch and a rampart. At Lyne, there were two, one on each side, like two wings. At Camelon, there were also two, a smaller, the original annexe, and a larger which exceeded the fort itself in area, and was protected by a large rampart and several ditches. At Newstead, there were three. The spade will undoubtedly bring to light more annexes, but it is almost certain that many forts lacked them; no trace of one, for instance, has been noticed along the Wall of Hadrian.

Little is known of the contents of these enclosures. That  p61 at Castlecary contains the remains of the baths. The larger annexe at Camelon was traversed by two streets, and contains the remains of two large buildings, one apparently baths, and traces of others. At Gellygaer, the baths were also in the annexe, and in addition two large enclosures (one containing furnaces) and several small structures. Several annexes on the Continent, notably at Heddernheim and Saalberg, have been found to contain the 'civil settlements,' but whether this is the case in our country further exploration alone can prove. There is certainly no room for such a settlement in the Gellygaer annexe, but the spade may discover yet another annexe, or what is equally likely, prove that the suburb was not enclosed as at Housesteads.


Few Roman remains in Europe have attracted more attention than the two Walls, the lower stretching from the mouth of the Tyne to the Solway, and the upper across the narrower isthmus between the indents of the Forth and the Clyde. The term Wall does not convey an adequate idea of these great works. Each was a complex of forts, continuous rampart and ditch, military roads and outlying posts, planned with consummate skill and on an imperial scale; but in addition, the southern line has enigmatical features which have long been the subject of controversy.

Both lines appear to owe their inception to the military genius of Agricola. The strategic advantages of the upper isthmus were certainly recognized by him, for he held it by a number of posts; and it is probable that some of the forts upon or near the Solway-Tyne line were also due to him. His immediate successors lacked his energy, and during the period of border unrest which followed, the Caledonians made at least one serious inroad into the Province. To remedy this dangerous state of affairs, Hadrian, in accordance with his policy of consolidation rather than expansion, constituted the lower isthmus the frontier in A.D. 120. It is almost certain that the Agricolan posts of the upper isthmus had already long been abandoned; but twenty-five years after Hadrian's visit, and in consequence of further border trouble,  p62 Antoninus Pius fortified that isthmus with a 'wall.' This may have been dictated by a return to the 'forward' policy of Agricola, the intention being the conquest of North Britain by successive stages; or the object may have been to place the natives of the intervening country under a protectorate and thus create a friendly buffer-state between the Province and the Caledonians. Under any circumstance, the barrier of the lower isthmus continued to be held, and in fact served as the base whence detachments were drafted to man the upper line. This duplication of frontier lines, however, was of short duration and there is reason to think that the upper wall was abandoned at the time of the great Caledonian inrush of A.D. 180. The lower wall, on the other hand, continued to be the recognized frontier to the close of the Roman era.


[image ALT: missingALT]
Fig. 18. — Map of the Antonine Wall.
The known sites of Forts shown as solid squares; the uncertain or conjectural, as hollow squares.
(6 miles to 1 in.)

This structure was about 36⅓ miles in length, and for most of this distance its rampart and ditch are still visible. Less conspicuous is an irregular mound or glacis on the northern side of the ditch; and at a varying distance behind the rampart is the stony ridge of the military way. "The work is thus in its entirety a quadruple line, which, instinct with Roman greatness of design and thoroughness of execution, undulates across the isthmus with a course as direct as the strategic requirements of strength would admit. It skilfully takes advantage of high ground, commanding throughout almost its entire course a valley or low-lying ground in front."​2 Add to this 'quadruple line' the remains of a dozen or more garrison stations and the traces of 'periodic expansions' at the rear of the rampart, and the reader will have a general idea of the Antonine Wall.

Excavations between 1890 and 1893 proved that the rampart was constructed of turves or sods laid in definite courses resting upon a spread of rough stones between two kerbs of squared stones. The width of this foundation, averaging between 14 and 15 ft., indicates the original width of the rampart, which has spread under its own weight and the disintegrating effects  p64 of the weather. This discovery confirms the statements of Julius Capitolinus, who, writing about the close of the 3rd century, relates how Antoninus Pius conquered the Britons and built a murus cespiticius. And the Welsh and English chroniclers, Gildas, Nennius, and Bede, tell obscurely of a turf and stone wall between the Picts and Scots of the north and the civilized population of the south.

The ditch is normally V‑shaped, with an average width of 40 ft. and depth of 12 ft. Its distance from the rampart varies from 18 ft. to 112 ft., but usually it does not overstep the limits of 24 and 30 ft. The outer mound consists of the soil from the ditch. It is very irregular in form, being sometimes flat, sometimes heaped up, and its use seems to have been to give greater height to the counterscarp of the ditch, as the ground generally slopes to the north. But it is nowhere so pronounced as to interfere with the 'command' of the rampart or to afford cover to the enemy. Here and there at the back of the rampart are remains of 'periodic expansions,' rounded bulges, so to speak, of the same construction as the former. Their use is uncertain; the most plausible theory is that they were the platforms — ballistaria — for military engines.

The garrisons were stationed in forts of the usual Roman form, of which the sites of ten are known and those of six or seven more are surmised. The known sites, starting from the east, are Rough Castle, Castlecary, Westerwood, Bar Hill, Auchindavy, Kirkintilloch, Balmuildy, New Kilpatrick, Castlehill, and Duntocher. These are on the actual line; but a little north of it, near Rough Castle, is the fort at Camelon which may be regarded as an advanced post. The stations appear to have been tolerably evenly distributed, the shortest interval being about 1¾ miles, and the longest 3¾ miles. Normally, they were applied, like the mile-castles of the lower isthmus, to the Wall, its rampart forming their northern defence; but that at Bar Hill, and perhaps that at Kirkintilloch, were slightly set back from its line.


[image ALT: missingALT]
Fig. 19. — Map of the Wall of Hadrian.
The Wall shown as a thick line; the Vallum, as a thin line;
the Forts, as large black squares; and the Mile-Castles, as small squares.
(6 miles to 1 in.)

This grand barrier extends from Bowness on the Solway to Wallsend on the Tyne, and is 73½ miles in length. Like the Antonine line, it has a similar succession of ditch with glacis-like outer mound, a wall set back so as to leave an intervening berm-like space, and a military road behind; also, at intervals, stations for the accommodation of the garrisons. But, unlike it, the wall is built of stone; and the stations are in two series, one of greater and the other of lesser size, which may be distinguished respectively as forts and mile-castles. The most striking point of difference, however, is a ditch between two banks in the rear of the military road, and known as the Vallum. The lower barrier thus resolves itself into two sets of works, the Wall with its appendages and the Vallum (Fig. 20).

[image ALT: missingALT]
Fig. 20. — Diagrammatic Section of Wall of Hadrian.
A, Ditch; B, Wall; C, Road; D, Vallum

These two lines pass from sea to sea in close companion­ship as a rule, running parallel some 60 or 80 yds. apart for miles on the stretch along the lowlands of the eastern and western thirds of their course; but in the intervening rugged region they seem at first sight to pursue independent courses, drifting apart here and there to the extent of half a mile or more. These divergencies in the middle third are due to the configuration of the country. The Vallum pursues the more direct course, while the Wall forsakes its companion for the higher grounds. In this region, where the hills have gentle dip-slopes to the south and craggy precipices to the north, the normal position of the latter is the crest; that of the former, the slope behind. Between these two great works, the Wall and the Vallum, the military road in the more hilly regions pursues a path which is parallel to neither, but which was determined with a view to the easiest possible route from point to point.

As already stated, the wall was of stone. Where best preserved, it remains to the height of 5 or 6 ft.; but in those districts where the land has been long under cultivation, it is more often reduced to a mere ridge of foundation rubble or has so completely disappeared that only the ditch remains to indicate its line. Where ascertainable, the thickness varies from 6 to 9½ ft. The  p67 ditch varies considerably in dimensions, but an average width of 36 ft. and depth of 15 ft. may be accepted as fairly correct. It accompanies the wall throughout its course, except along the edges of cliffs where it would be of no practical use, and for a mile or two west of Carlisle where the Eden takes its place. The upcast from it was used to form the glacis-like mound or spread as in the Antonine Wall.

Along the actual line, or in its vicinity, are the remains of the garrison stations. Of these, about nineteen are known, some still imposing though in ruins, others reduced to the barest traces. Their distances apart fall, as a rule, within the limits of 3 and 5 miles. Their plans, so far as they are known, are those of typical Roman forts. Some, including the three or more detached stations, were apparently constructed, not only before the wall, but before it was contemplated, and were subsequently woven into the mural scheme; the majority, however, were undoubtedly part of the scheme. The mile-castles were smaller than the stations, and were more numerous. They were all, so far as is known, of similar size and shape, and distributed at tolerable even distances apart. The sites of about fifty have been identified, but their remains are for the most part extremely slight, but in the wilder middle region some still present conspicuous ruins. These fortlets averaged 60 ft. by 50 ft., and were attached to the wall, that structure forming the northern side, the remaining three sides being of similar thickness and bonded into it. The free corners were rounded, and each fortlet had two gates, one to the north and the other to the south. In the lowlands they appear to have been as nearly as possible a Roman mile apart; but in the hilly region they are irregularly spaced, the engineers here relaxing their rule in order to select  p68 advantageous positions for them. The original number of these structures was about eighty.

The remains of only a few turrets are known, but there is reason to think that they were numerous, and were placed where look-outs were needed. They were small rectangular structures recessed into the back of the wall, with a narrow doorway to the south.

The mural road provided communication between the stations and the mile-castles. Its stony ridge is best preserved in the hilly districts; elsewhere it is mostly obliterated or buried. Its usual position is from 60 to 100 ft. behind or south of the wall; but here and there it recedes when it is necessary to do so in order to gain gentle gradients. In serving the stations and mile-castles it necessarily clung to the wall, and thus participated in much degree in its sinuosities. Hence in the hilly region where the wall zigzagged and curved considerably to the north, it was neither a direct nor an easy means of communication between distant points, and here it was augmented by a more direct route from lowland to lowland — the road now known as the Stane gate.

The great earthwork, known as the Vallum, consists of a flat-bottomed ditch, about 30 ft. across at the top, from 10 to 12 ft. across the bottom, and about 7 ft. deep, between two mounds formed of its upcast, each set back about 25 ft. from the brink of the ditch. Where best preserved the mounds are still 6 or 7 ft. high. Besides these there is, here and there, a smaller mound usually cresting the south brink, but occasionally the north one. In some places the Vallum is a conspicuous and imposing feature forming a great triple band of a total width of some 130 ft. The small mound seems to be always on the side of the ditch which from the natural slope of the ground would be the lower, and its object is apparently to level it up to the height of the opposite side. As the slope of the ground is nearly always to the south, this explains its usual position.

The behaviour of the Vallum to the stations has an important bearing on the question of its origin and use. It is indistinct, or even obliterated, in the vicinity of the stations; and this has given rise to the belief, reasonable enough, that whatever its  p69 purpose may have been, it fell out of use at an early date, and was intentionally levelled or allowed to be obliterated by the gradual process of agriculture at these places, before the close of the era. This obliteration is responsible for some wrong impressions as to its relation to the stations that stand across its line or otherwise seem to touch it. There are stations that lie beyond its extremities, as Wallsend and Newcastle in the east, and Drumburgh and Bowness in the west; and there are intervening stations that are entirely off its line, as Housesteads, Great Chesters, Carvoran, and Chesterholm. Of the residue, Benwell, Rutchester, Halton, and Chesters are so placed that their southern ramparts appear to be in line with the Vallum; while Carrawburgh and Birdoswald stand across it. The old view assumed that these two stations were in actual contact with it. Excavations in 1896‑97, however, have shown that in their case the Vallum curiously and purposely avoids them by skirting around them, and there is also reason to think that the mile-castles were similarly avoided.

The purpose of the Vallum has long been a subject of controversy. It has been regarded as a great pre-Roman barrier; as a Roman defence against the south, and particularly against the Brigantes; and as a sunken and fortified road. But none of these is consistent with the facts. The curious manner in which it deliberately goes out of its way to avoid the stations which it otherwise would strike shows that it is part of the mural scheme, and this tells equally against its being a pre-Roman defence or a road, and, apart from this, excavations on its site have failed to yield any evidence of a road either in its ditch or elsewhere between the mounds. With regard to the second theory, high military authorities have pronounced against its being a defensive barrier of any kind. Professor Mommsen suggested that it was a civil boundary — "that the Vallum marks the southern or inside edge of the limes or 'frontier strip' of the empire,' the two works, Vallum and Wall, being regarded as contemporary, but the one a legal, and the other a military line. Dr. Haverfield is of a similar opinion, but considers that its purpose "was forgotten or ignored even in Roman times," and in support of this, he instanced the evidence of the  p70 early filling up of the ditch "where its presence may well have been inconvenient, as near a fort." The reader must draw his own conclusion as to the mention of this "strange earthwork, the inscrutable Vallum"; but it is safe to predict that his verdict will be that the last word has not been said upon it.

The discovery of the remains of a turf wall in 1895, "introduced," as Dr. Haverfield puts it, "a new factor into the whole Mural problem." It has long been observed that for about a mile west of Birdoswald, a ditch runs parallel to the Vallum at about 90 ft. to the north; and this was usually regarded as a supplementary defence to that work. But a series of trenches disclosed the remarkable fact that it appertains, not to the Vallum, but to a former turf walk, from 10 to 15 ft. wide, which appeared to have been purposely destroyed, and evidence was forthcoming to prove that this work represented the original line of the Wall hereabouts. No trace of a turf-wall has been found elsewhere along the line; but the discovery is strongly suggestive that the Wall of Hadrian was, like the Antonine Wall, originally constructed of turves, and was subsequently replaced by a stone wall, the builders of this stone wall finding it necessary for some reason or other to deviate from the old line in the vicinity of Birdoswald.

Enough has been said to show that the Wall embodies works and modifications of different times, all Roman, of course. The first emperor whose name appears in connection with it is Hadrian. In four of the mile-castles have been found inscribed tablets in his honour, and presumably similar tablets were placed in the others. Some of the stations may have been such of Agricola's camps as happened to be in the line of the projected wall; but we cannot imagine a prior existence for the mile-castles — they are integral parts of the wall itself. If Hadrian erected these, that structure must have been already determined upon. You true that no contemporary writer mentions his building a wall in Britain, but a century and a half later, Spartian states that "Hadrian went to Britain and put straight many things which were crooked therein, and was the first to draw a wall eighty-thousand paces, to divide the barbarians from the Romans."

But the same writer tells us that Severus, more than eighty  p71 years later, also built a wall — "The greatest glory of his reign is that he fortified Britain by a wall drawn across the island and ending on both sides with the ocean" — and this is reiterated by subsequent writers. But, as in the case of Hadrian, no contemporary writer records such a work on his part; still more remarkable is it that both Dion Cassius, writing a few years after his death, and Herodian a little later, should describe his Caledonian campaigns in graphic terms,​b yet make no allusion to his wall-building.

That Severus had something to do with the barrier of the lower isthmus is, however, beyond question. It is true that no inscription to him has been found along the Wall; but his name is inscribed upon Cumberland quarries, and upon slabs at Hexham, Risingham, and Old Carlisle. Between Hadrian's day and that of Severus there had been troublous times; and it is likely enough that the second emperor found the Wall in a ruined condition, and that he not only restored, but strengthened it. If we accept this view of the part played by Severus, there will be little difficulty in also accepting as literally true the statement that it was Hadrian who "first drew a wall, etc."; in other words, in assigning to this great emperor the initiation of the general scheme of wall, forts, mile-castles, and vallum.

Thayer's Note: For a very thorough, comprehensive site on Hadrian's Wall, see Roman-Britain.Org (archived).

The Author's Notes:

1 For more particulars, see Romano-British Buildings and Earthworks, chap. i.

2 The Antonine Wall Report, p2.

Thayer's Notes:

a The Greek historian devotes a long section of Book VI to the Roman camp: it is online here.

b Cassius Dio: LXXVII.11 ff.; Herodian: III.14.

c Among them (see the respective pages, with photos) the Porta Appia, the Porta Asinaria, the Porta Latina, the Porta Ostiensis, the Porta Pinciana, the Porta Praenestina, and the Porta Tiburtina.

[image ALT: Valid HTML 4.01.]

Page updated: 14 Dec 20