THE DISCOVERY OF THE FORT

The area chosen for the first cuttings, in July 1 947, was in Windsor Court (3), just north of Falcon Square and near Bastion 14, the choice being dictated by the absence here of basements of full depth, so that the early deposits would be more completely preserved below the modern floor. In these early days the excavators had yet to learn how thoroughly the super­ ficial promise of a site could be dissipated when the modern floors were removed. In Windsor Court a series of massive concrete foundations drastically reduced the area available for examination. Much of the site was also taken up with the extensive chalk foundations of a mediaeval house, not to mention a number of rubbish-pits and other disturbances of similar or later date. In spite of these handicaps, however, sufficient of the Roman wall remained here and just to the north, behind Bastion 14, to give a first picture of the wall and its related features on this side and to set the in­ vestigators along the road which led to the recognition of the fort. The natural ground surface in Windsor Court was about 6 feet below the

modern street, so that 4-5 feet in depth of the early deposits had survived for examination. Internally the features of the Roman defences followed the normal pattern. Behind the wall there was a bank of clean brickearth, the upcast, no doubt, of the external ditch or ditches: the bank was quite slight compared with those met with in the defences of other Roman towns-here about 12 feet wide and 4-5 feet high. Beneath the bank in Windsor Court and subsequently elsewhere were several mortar-mixing pits (Plate 1) associated with the building of the fort wall, one at Windsor Court retaining its mix intact. Behind the bank was a gravel road, here only partly preserved,

but remaining in places to a depth of about 1 2 inches. The gravel was hardrammed; the road had a width of about 17 feet and was flanked by drainage gullies cut into the natural ground. Study of the wall itself in the southern part of the area was rendered

difficult by destruction and by the fact that the outer face was not at that time available for examination. But these disadvantages were not present behind Bastion 14 (4), where the wall, standing to a height on the outside of just under 4 feet, was well preserved and its outer face readily accessible. The wall behind the bastion was examined as part of the process of

removing the contents of the bastion and this subject belongs to a later section of this account. Two facts immediately presented themselves, how­ ever (Plates 2, 3). In the first place, the wall was in two parts built lengthwise against one another in a straight joint, the outer somewhat narrower and set more deeply than the inner, the base of which did not penetrate through the bank to the natural surface (Fig. 3). These features in themselves provoked a good deal of discussion at the time. But the true explanation of them could not be forthcoming at this stage, for the wall had undergone a late repair which was the possible explanation for the straight joint; and the precise relationship of the two parts could not have been determined with­ out destroying a large part of what still survived. The second and more immediately significant fact was that the external

face of the wall had been given a treatment which had not been met with anywhere else in the Roman city. The Roman town wall of London has been seen at many points on its circuit and its features have been consistent. Though showing a variation in width from about 7 to about 9 feet, its base is always marked by a chamfered external plinth, usually though not invariably of purple sandstone; and at vertical intervals of four to six courses (2-3 feet) in the stonework occur bonding or levelling courses of tiles whose purpose is to provide a new, stable bed for each successive ‘lift’ of the wall (Fig. 9).1 None of these features was present in the wall-face exposed behind

Bastion 14 : there was no plinth to demarcate faced wall from foundation and although the height of the former was sufficient to incorporate the lowest

tile course this also was absent. Whatever the meaning of the unusual features observed on this site, the one clear fact to emerge was that there was some variation in the history of the city wall in this area as compared with elsewhere; and the next objective was therefore to be to determine the extent of the structural differences and to try to date them. It should be noted that throughout this investigation, which was continued at intervals over a period of some months, the face of the wall and the other external elements of the defences were not accessible because of obstructions of one sort or another; and it was not until a quite late stage that the external features were seen anywhere other than behind Bastion 14. The programme therefore took the form of a succession of cuttings

behind the wall and at right angles to it. The work to the north of Bastion 14 can be dismissed briefly. The relative depths of the modern cellars and the degree of rebuilding along the line of the city wall were such that nothing at all had survived. The only exception to this statement was the site of the Barber Surgeons’ Hall near the north end of Monkwell Street (2): there the deposits had survived to modern street-level. The evidence from the Hall did not, however, help with the immediate objectives though it made its contribution in other ways. A detailed account of the features of each of the cuttings to the south of

Bastion 14 is not called for here. Suffice it to say that the wall retained its double construction throughout, with a straight joint defining the two parts and the inner consistently let into the artificial bank and not penetrat­ ing to the underlying natural surface. In the cellar of the building which was originally no. 33 Noble Street (8), however, the investigators met with a surprise. No. 33 was just north of the angle of the Aldersgate re-entrant, the point at which, from a number of earlier observations, the city wall is known to have turned westwards in its course for the second north-western angle at Newgate (p. 52); and the wall now being examined might have been expected to show some signs of accommodating itself to this change. In fact, it was found to behave in a completely contradictory fashion, depart­ ing from its original straight course and curving inwards, to the east, and towards the interior of the city (Plate 4). The meaning of this conduct became clear in the next basement south­

wards, that of no. 34 (p). Here there had been much disturbance in the eighteenth and nineteenth centuries, but a good fragment of the outer ‘half’

of the double wall had survived and was found to be continuing the eastward curve of the wall in the basement of no. 33. The really significant feature of the cellar, however, was the presence of a small sub-rectangular turret, built against the inner face of the wall on the crest of the curve. Taken in con­ junction with the rest, it was immediately recognisable as the quite typical corner turret of a Roman fort (Plate 4). Before continuing the account of this investigation three points require

to be noticed for future reference. First, the curve of the Roman wall in no. 33 Noble Street was sufficient to bring the face into the basement, clear by about 2 feet of the modern party wall, and therefore to expose the junc­ tion with it of Bastion 15 (8). Secondly, the inner ‘half’ of the double wall was not continued southwards into no. 34 Noble Street but stopped short of the corner-turret. And thirdly, a break in the inner ‘half’, where it had been cut by a modern wall, made it possible for the first time to see the true relationships of the different parts-and this without any destruction of the already much-reduced remains. From the section thus exposed it could be seen that the outer ‘half’ of the wall was the earlier. This, which must now be called the fort wall, had a partly-battered inner face with an offset founda­ tion. The internal bank, of clayey brickearth, overlay the offset and rested against the inner face of the wall: wall and bank were clearly of the same period. The inner ‘half’ of the double wall had on the other hand been let into the bank, as previous cuttings had suggested. The inner wall, or more truly foundation (for as seen today it is (with one exception: p. 48) buried in the bank), is therefore an addition to the fort wall. Its date and significance are better discussed later. To eyes opened by these discoveries it was now easy to see the shape that

this unexpected fort might take; and since so much of the area was devas­ tated by bombing the pursuit of its main elements was in theory at least a simple matter. It has long been recognised that the Cripplegate angle of London Wall, by St. Giles Cripplegate Church (i), is a right angle; and although the wall itself has been destroyed the line of its modern successors must reflect the original and is typically that of a fort corner. The existing north-east side of the Aldersgate re-entrant, therefore, a wall some 260 yards long, must represent the west wall of the fort; its north wall must have been some part of the city wall that runs eastwards along the south side of St. Giles Cripplegate churchyard and beyond. Here two features are

significant: the gate of Cripplegate itself; and a change in direction made by the wall on the east side, somewhere in the neighbourhood of Alder­ manbury Postern (id). As to the former, Cripplegate, though its remains have never been seen,

has long been accepted as a Roman gate on the strength of a mention in the Laws of Ethelred of c. a.d. iooo.1 A gate at this position does not fit into the recognised pattern of Roman roads approaching the city, but need not have done so if it was part of the standard layout of a fort.2 As to the wall-alignment: the course of the wall is fixed by the St. Alphage stretch to the west of Aldermanbury Postern (17), and by the London Wall stretch to the east (25). Observations made in 1857 show that the London Wall alignment must have been carried as far as Aldermanbury Postern.3 The change in direction, therefore, though variously shown on plans of the Roman city, must have taken place within the space of a few yards on the west side of Aldermanbury Postern, about 230 yards east of the Cripplegate corner. The final significant point in this preliminary survey of the possibilities is that Cripplegate itself is midway along this dimension, occupying therefore the appropriate position for a gate if this should prove to be the shorter side of the ‘playing-card’ that is the normal outline for a Roman fort. This tentative conclusion carries one further corollary. Wood Street, which passes through Cripplegate, is approximately parallel with the west wall and divides the 260 by 230-yard enclosure just defined along its main axis. A feature of the internal layout of most Roman forts is their main streets linking opposed gates, which are prevented from intersecting at the centre only by the presence of the headquarters building, athwart the long axis, with its main frontage on the short axis, or via principalis. Wood Street therefore has claims to Roman ancestry; and at or near the point at which it crossed the south wall of the fort, yet to be found at this stage, should be the remains of the south gate. In fact, the likely position for this gate was to the west of Wood Street (12), for as with most ancient roads Wood Street had not been able to retain its original straight course throughout its history and as it went southwards had taken on a curve to the east. It will be seen presently that these prognostications followed fairly

closely the results obtained by excavation. But they are set out here neither as the outcome of a species of second sight nor as a demonstration of wisdom

after the event, but as an example of the kind of ‘field-work’ that every excavator is obliged to do before beginning to explore his site. The problem presented by a continuously occupied area like the City of London is that later activities mask (when they do not actually destroy) the earlier. Yet the earlier features have frequently left their mark (p. 55) and the difficulty that confronts the archaeologist working in such conditions is to recognise the signs and appreciate their possibilities. Working, then, from the hypothesis that the course and extent of the

north and west walls and the position of the north gate were known, the course of the south wall was pursued on the assumption that it would run parallel with the north wall, beginning from the south-west corner in Noble Street (p). As has been said, practically all the buildings along the supposed line of the south wall had been destroyed and in theory at least the investiga­ tion should have been simple enough. In practice there were the inevitable complications springing partly from disturbance of the ground, partly from low cellar levels, which were frequently only a few inches above the original surface, so that practically all the accumulated deposits had already been removed. While therefore the main objective was the wall (or, more usually, its foundation) too often it had been destroyed or was too mutilated for its behaviour to be determined. It was then necessary to concentrate on the ditch as a pointer to the line and direction of the wall, with which it would have run more or less parallel. Returning now to the account of the remains, the south-west corner was

completed by the discovery of the wall on the opposite side of Noble Street in what was then Honeypot Lane (10).1 Here, having turned through a right angle, the wall began its straight run eastwards. It survived to a height of three courses beneath a seventeenth-century floor: stone steps relating to the floor, which was perhaps that of a cellar, rested directly upon it. From this point eastwards across Staining Lane and for some distance the results were not satisfactory, but the line was confirmed by the finding of a short length of foundation-trench in a much-mutilated area on the east side of Staining Lane (11). The course set by these features showed only too clearly that if the south

gate had indeed stood where expected on the west side of what is now Wood Street, little of it was likely to have survived. As it neared Wood Street the Roman wall converged at a very acute angle from the north on a massive nineteenth-century wall-foundation (12); and while it was likely that in due course it would emerge on the south side the levels and the scale of the modern foundation were such that nothing of the Roman work could be expected to survive for some distance. These forebodings were only too thoroughly ful­ filled when the area came to be excavated. By some extraordinary chance a small fragment, less than a foot square, of the foundation of one gate-jamb had survived on the south side of the modern wall: that it was the east jamb of an opening was shown by the fact that it ended on the west against solid brickearth; its Roman date was established by the characteristic use of a sticky grey-yellow clay as a bonding material. But the degree of disturbance and destruction on the north side in particular was such that all other traces of the masonry of the gateway had vanished apart from a few stones, no longer in situ, some of which still had sticky clay adhering to them. But if the masonry had gone, other features relating to the gate had

survived. Within the enceinte of the fort-to the north, that is, of the walltraces of the north-to-south road remained as a thin much-broken gravel layer immediately beneath the cellar floor. More important, outside the wall not only was there a further piece of the road, here a very compact gravel surface, but the ditch across the gate-approach had survived undamaged for a length of about 19 feet (Plate 5). The ditch was a remarkably slight affair, V-sectioned and a mere 5 feet wide by 4 feet deep, but it had been necessary to carry the road across it on a bridge which had been provided with timber supports let into the sides of the ditch. Not enough had survived to enable the complete arrangement of these timbers to be worked out. They occurred sometimes singly, sometimes in groups of three, and most of them had been at least roughly squared. An interesting feature was the use of one or two obliquely-set struts which gave support to the middle of the bridge from the underside. Immediately to the east of Wood Street a short stretch of the wall foun­

dation was recognisable (13). From this point on the line was masked by obstructions. Fortunately the crucial area, that of the cellars on the west side of Aldermanbury, was clear; for here, if the preliminary reading of the evidence was correct, should be the south-eastern angle of the fort, where

the wall should turn northwards, parallel with the already established west side. The relevant cellars were those of nos. 7 0 A and 7 1 Aldermanbury, 7 0 A being to the north (14). Once again, the wall in 110. 70A, though not completely destroyed, was

seriously mutilated. Both masonry and foundation had survived, but no­ where sufficiently to enable the actual line to be made out. Recourse had therefore to be had to the ditch; and since the purpose of this particular investigation was to establish the course of the ditch and not merely its existence the whole cellar was cleared to enable it to be seen in plan as well as in section. The removal of the cellar floor and its modern make-up re­ vealed an area of completely artificial deposits which was obviously com­ posed of a number of separate elements, mainly pits of mediaeval date. The identification and removal of these features along the lines already described, however, revealed the Roman fort ditch as it were between and amongst them (Plates 6, 7). The ditch was a good deal larger than further west, 12 feet wide and nearly 6 feet deep with well-cut convex sides; and a real obstacle. Its important feature, however, was its course. For some feet it ran straight; but as it approached the eastern side of the cellar to pass under the modern street it began to take on a pronounced northward curve, anticipating unmistakably the presence of the south-east corner of the fort beneath the roadway of Aldermanbury. At this stage, then, excavation can be said to have confirmed in a general

way the outline of the fort as suggested by the evidence above ground; for the position of the south-east angle beneath Aldermanbury corresponds neatly with a north-east angle beside Aldermanbury Postern (16). But the exact location of the east wall of the fort remained a matter of some un­ certainty. All the indications were that it should lie just within the cellars on the west side of Aldermanbury, north of Addle Street. An attempt to find it here (15a) came up against the problem of levels: all trace both of wall and of ditch had vanished and it must be assumed that they were removed when the basements were dug. At the north-east corner itself there was a similar state of affairs, except that where the early features had not been destroyed by deep floors they had been replaced by massive modern walls or founda­ tions. About 30 feet west of Aldermanbury Postern, however, in much disturbed ground (16), were the slight remains of the fort ditch curving in from the south to merge with the U-sectioned ditch that is characteristic of

the City defences (p. 15). Here, therefore, must have been the north-east angle of the fort-a fact which can now be seen to be reflected in yet another surface detail. The angle in the frontage of London Wall1 where it is-or was (for this area is now completely altered)—entered by a small court which is nameless on recent maps is not an accident but owes its existence to the disposition of the Roman defences of London 1,700 years or more ago (Plate 8). Since the above was written further discoveries at the south-eastern cor­

ner have established the position of the east wall of the fort. Here on the east side of Aldermanbury the demolition of buildings as a preliminary to the re-development of the Guildhall area exposed ground which had not pre­ viously been available for examination. In 1965 a group of volunteers under Messrs. Peter Marsden and N. Farrant exposed a 40-foot length of the foundation-trench of the wall with its accompanying ditch. The foundationtrench, which had been almost completely robbed lay just within the western margin of the area available; the ditch as it passed under the road had begun to turn westwards to make the corner, which as already noted must lie under the street, but somewhat further to the east than was originally thought from the evidence of the ditch at 71 Aldermanbury.2 With this valuable information the overall dimensions of the fort can now

be fairly accurately stated as about 760 feet north-south to 7 10 feet eastwest-‘fairly’ accurately because in the conditions it has not been possible to relate individual structures directly to one another for surveying purposes, particularly when (as here) discoveries are made over long periods of time. As plotted, also, the outline is not quite a true rect­ angle, but the variation is very slight and again may be due to the same factor. Before 1965 there was little to say of the north wall of the fort. Of the

part west of Cripplegate a considerable stretch towards the corner had been completely destroyed and replaced by modern brick: it was tested in the cellars on its inside in a couple of places in 1947-8. Further to the east, more or less opposite the east end of St. Giles Cripplegate church a small frag­ ment of post-Roman wall was recognisable above the modern surface but

1 The original street of that name, of course, not the new London Wall (Route 11) . 2 Thanks are due to Messrs. Marsden and Farrant for permission to incorporate their discovery

did not appear to be extensive. Since the cellars here contained much bombrubble no attempt was made to pursue it. The lowering of the level of the churchyard as part of the Barbican de­

velopment and the removal of obstructions on the south side showed that appearances on this site (ia) had been deceptive. As this is being written much work remains to be done, but the base of the fort wall appears to sur­ vive (and serves as a foundation for its multi-period successor) over a dis­ tance of something over 100 feet. Here as elsewhere the face remains to a maximum height of only 3-4 feet on the outside; internally in the one place where it can be clearly seen the face survives to 6-7 feet and displays the usual batter in its lower part. The depth of the cellars against the back of the wall was such that the ‘city thickening’ had completely disappeared. At the extreme western end of the surviving wall, about 175 feet from the centre­ line of Cripplegate Buildings (the site of the north gate of the fort, recog­ nisable by the hump that it makes in the modern road) a fortunate chance has preserved the stump of the east wall of an internal turret. This is not only a valuable addition to the plan: its siting seems to help to account for that of the turret on the west wall (7) which is further discussed below

(p-34)- Another completely unexpected find on this site is that of a bastion which

does not appear to figure in any of the records. More is said of this in the appropriate place later (pp. 7 1-5). East of Cripplegate the fragment in St. Alphage Churchyard (17) is

noteworthy for the later periods; but the base and foundation of the wall are Roman, and although the outer facing-stones have gone the back shows all the features met with on the west side of the fort. The inner face of the fort wall is battered, and built against it but at a higher level is the internal addi­ tion, the inner ‘half’ of the double wall. The evidence from this site is valuable in that it shows that on the north as on the west there had been a double wall. Apart from the very recent discoveries on the north wall which have been

fitted into their place above, the investigation of the Cripplegate Fort has been described so far in the order in which work proceeded, though with other demands on the excavators’ attention the individual excavations did not always immediately follow one on the other. In establishing the outline of the fort, with its four angles at St. Giles Cripplegate, St. Anne and St.