Romano-British antiquities

ROMANO-BRITISH ANTIQUITIES

The Roman name for York is well established both by epigraphic and literary evidence as Eboracum or Eburacum (fn. 1) and also by a living tradition which preserves in the modern name York a corrupt form of the Roman name. (fn. 2) Itself a latinized form of a Celtic name meaning probably 'the place of Eburos', the Roman name was popularly regarded as meaning 'Boar Town' and on an altar found in Bordeaux a boar appears as the punning badge of York. (fn. 3)

The valley of the Ouse consists of a great tract of level alluvium liable even now to sporadic local flooding and at the time of the Roman invasion may well have been waterlogged. At two points, York and the village of Escrick, six miles to the south, well-marked ridges of boulder clay and gravel, the terminal moraines of ice-sheets, cross the valley. (fn. 4) These ridges provided early man with a natural bridge. (fn. 5) To the west lie the Pennines into and through which the Dales provided routes. To the east is easier ground, but here the Vale of Pickering formed a similar barrier to the Vale of York and could be crossed only at the coast and in the neighbourhood of Malton. North-bound traffic was thus forced to follow the rising ground skirting the Vale of York on either side.

Bearing these considerations in mind, York must have seemed the ideal base to the Roman commander moving his army north out of Lincoln. From York it was possible to dominate east Yorkshire, the north routes and the dales, and the tidal river provided him with water transport. Tactically the site conformed to the normal siting for a Roman fortress and is in this respect the only suitable site on either moraine. A low plateau of boulder clay and gravel lying between the junction of two rivers—the Ouse and the Foss—provided a site for the fortress similar to those chosen at Caerlon and Chester. (fn. 6) Subsequent levelling has obscured the appearance of the site as it was in Roman times, but it is clear from the marked differences that occur in the depth at which the Roman level is found that the contours were then far more emphatic than now. The site chosen for the fortress lies slightly north of where the main ridge of the moraine, running by Heslington Hill and Nunthorpe, crosses the Ouse and is separated from it on the east side by the low-lying ground bordering the River Foss. On the west the ridge broadens northwards to provide ample room opposite the fortress for the later civil town.

The site presented two minor difficulties. The laminated boulder clay, sometimes capped with sand or gravel, which formed the sub-soil under the greater part of the fortress, was not always stable, and foundations occasionally subsided when stone began to be used for building. (fn. 7) Again, on the north-west side of the site, near the gate, was a small valley where the Romans had to conduit a stream. (fn. 8) The subsequent filling in of this valley has led to the preservation here of the Roman defences to a greater height than elsewhere.

It is not known if the site was occupied before the Romans came in 71. Evidence exists for the presence of Bronze-Age man in the neighbourhood (fn. 9) and it seems likely that a site in an accessible and commanding position on a known prehistoric trade route would have attracted settlers. There is as yet no archaeological evidence for a native Iron-Age occupation. The dyke formerly existing across the morainic ridge on Heslington Hill might be linked with the well-known dykes of east Yorkshire for which an Iron-Age date is generally postulated, but both date and purpose of the structure are obscure. (fn. 10)

The evidence for a pre-Flavian Roman occupation cannot be lightly dismissed. Many Claudian coins have been found at York; elsewhere similar finds have been interpreted as evidence of pre-Flavian occupation (fn. 11) but the legionary pay-chest at York may well have contained out-dated coins. The finding of apparently Claudian pottery is harder to explain away. (fn. 12) Historically it is easy to find an occasion for the temporary or permanent basing of Roman troops at York before 71. By 47 or 48 a temporary frontier was established behind the Trent and Severn. An alliance with the Brigantes on their northern flank enabled the Romans to concentrate on attacking Wales and to subdue the revolt of Boudicca without interference from the north. The usefulness of this alliance depended on the strength of the pro-Roman rulers of the Brigantes and it was necessary to prop up their power from time to time. (fn. 13) The speed and efficiency of Roman intervention when needed has prompted the suggestion that there were auxiliary forts or an auxiliary fort in Brigantian territory. (fn. 14) For an army based on Lincoln the obvious approach was across the wolds of Lincolnshire and east Yorkshire and through the friendly territory of the Parisii (fn. 15) and the obvious site for any military post was York. More evidence is required before it can be said with any degree of confidence that such a post existed. By 69 it was clear that the Brigantian alliance was unreliable. The story of the struggle for power within the tribe is well known and has recently been reassessed. (fn. 16) The massive fortifications at Stanwick (N.R.) emphasize the strength of the anti-Roman faction. The only course open was the annexation of the whole territory and accordingly when Rome's own dynastic troubles were over Petillius Cerialis, the new governor of Britain, led his old legion, the Ninth, out from Lincoln into Yorkshire in 71. The result of this campaign, as Tacitus obscurely puts it, was to involve the great part of the Brigantes in war or defeat—magnamque Brigantium partem aut victoria amplexus est aut bello. (fn. 17) Roman arms were carried to the Solway or beyond and an uncertain amount of new territory permanently occupied. Some time in the course of that campaign the legionary fortress was moved from Lincoln to York. S. N. Miller (fn. 18) suggested the existence of a temporary campaigning base succeeded by a more permanent fortress at the end of the fighting in or about 74. Recent excavations (fn. 19) have modified his interpretation of the evidence and although two structural phases have now been observed in the timber defences, they cannot be clearly equated with a temporary base and a permanent fortress.

The Fortress

Excavations undertaken between 1925 and 1927 (fn. 20) and in 1955 and 1956 (fn. 21) have revealed details of the earthwork and timber defences of the first fortress at York. It occupied the site on the north-east bank of the Ouse that was to remain in legionary occupation until the end of Roman rule. In spite of S. N. Miller's arguments for a reduction in the size of the fortress in the 4th century (fn. 22) there can now be no doubt that the familiar outline of the 4th-century fortress was also that of the preceding ones. The 1st-century defences have now been found on three sides and there is strong presumptive evidence that they also existed on the fourth. (fn. 23) The similarity of the almost-contemporary fortresses at York and Caerleon in size, shape, and siting is then most striking. (fn. 24) The York fortress took the form of a rectangular enclosure with rounded corners containing just under 50 acres. Its long axis ran from north-east to south-west with the main gate (porta praetoria) central to the south-west side and facing the river crossing. It can be assumed that the gates on the other sides occupied the same positions as in the later stone fortress, and that the two lateral gates were on the sites of the later Bootham Bar and King's Square, south-west of the centres of their sides. The length of the sides is 1,580 ft. (north-west and southeast) and 1,350 ft. (north-east and south-west).

The defences of the first period excavated in Coney Street (fn. 25) consisted of a wooden palisade strengthened behind with a bank of clay; elsewhere sand may have been used with turf to strengthen it. The bank was 16 ft. wide and rested on a strapping of narrow oak boughs between 1 and 2 ft. apart. It has not been possible to make an accurate estimate of the height of the defences but the slightness of the existing remains suggests that it did not exceed 10 ft.

At an unknown date but still within the Flavian period this first rampart was rebuilt. The new rampart, of the same width as the first and at least 7 ft. high, was placed above the levelled remains of the first and separated from it by a corduroy of closely laid oak planks. This new bank, like its predecessor, was built of clay (although in places, as on the northwest side, the material was probably sand) strengthened at back and front by courses of turf. It may have supported a timber palisade but evidence for this has not been revealed. Remains of a timber interval tower have been found on the south-west side of the fortress. Sufficient details have not been recovered to reveal its plan entirely. Stout wooden posts had been erected in a thick bed of turf. In the turf was a considerable quantity of early Flavian pottery. This would be consistent with an Agricolan date for the tower. The tower belonged entirely to the second period of the defences; those of the earlier period had been completely removed from its site. An Agricolan date would explain the renewal of the defences as a replacement—after a decade of use—of the rough green wood of the first defences by seasoned and carpentered timbers. (fn. 26) The post holes found by S. N. Miller at the east corner are better explained as belonging to a timber corner tower than to a timber palisade. (fn. 27) Outside the ramparts was a double, possibly in some places a treble, ditch of a maximum depth of 7 ft. 6 in. and an overall width of 41 ft. (fn. 28) Traces of timber buildings of this period have been found at one or two points inside the fortress enclosure but remains are too fragmentary and ill recorded for much to be deduced from them. (fn. 29)

Outside the fortress must have been the amphitheatre, public baths, and the canabae or extramural settlement that accompanied every fortress to provide for the non-military needs of the soldiers. In the absence of systematic excavation the canabae cannot be traced in detail. Early pottery associated with substantial timber buildings was discovered in 1939 near the site of the Old Railway Station on the south-west side of the river (fn. 30) and similar finds have been made north-west of the fortress in the Museum gardens. (fn. 31) These two widely separated sites need not imply a settlement as extensive as the later town. The evidence of early burials in the Micklegate area (fn. 32) underlying the later town shows that on the southwest side of the river, where later was to be the main urban centre, development began with cemeteries along the main road to Calcaria (Tadcaster). This road, which divided, beyond Tadcaster, for Chester, the Pennines, and the south through Castleford, left the porta praetoria, crossed the river at the site of the medieval Guildhall and ran approximately along the line of Toft Green, Blossom Street, and The Mount and thence along the morainic ridge. (fn. 33) It has been shown to be of Flavian date. (fn. 34) The tombstone of Duccius Rufus from Vienne (Isère), a standard bearer of the Ninth, (fn. 35) and other early tombstones and cremations (fn. 36) have been found along this road. An altar dedicated to Silvanus by a cornicen of the Ninth shows that there were shrines among the graves. (fn. 37)

There was ample room for expansion on the east side of the river in two areas: south-west of the fortress in the well-protected area between the two rivers, and on the north-west. In the first of these we should naturally expect the first civil site, all the more so because the primary approach to the fortress from Lincoln via Petuaria (Brough on Humber) ran this way. A road has been traced along the Hull road to Walmgate Bar, whence its course was probably directed along the south-west side of the fortress to the river crossing. First-century burials along this road have been found at Lamel Hill, (fn. 38) Walmgate Bar, (fn. 39) and in the Fishergate area. (fn. 40) Insufficient evidence has been obtained from excavation to reveal the nature of habitation or other occupation within the area although later civil development is recorded from Castlegate and Ousegate. On the north-west side of the fortress a road approached from Clifton and made direct for the river crossing. A second road left the north-west gate and may eventually have joined the other. (fn. 41) Early timber buildings from the Museum gardens and early burials from Bootham and Clifton (fn. 42) suggest the possibility of a 1stcentury date for these roads. Sporadic 2nd-century occupation material (fn. 43) and fragments of streets (fn. 44) indicate that by that time a more systematic expansion may have begun in this direction. This development was not maintained and from the 3rd century onwards the whole area beyond St. Mary's was given over to cemeteries.

The topography of 1st-century York may be summarized in this way. On the north-east bank of the Ouse lay the legionary fortress enclosed by earth and timber defences. In front of the main gate of the fortress was a river crossing upon which converged roads from Chester, Lincoln, and the north. Along these roads lay cemeteries and, nearer the fortress, the beginnings of urban development. The nucleus of this civil settlement has not yet been discovered: it need not necessarily have been the same as that of the later town on the south-west bank of the river and may in fact have been in the sheltered area southeast of the fortress. Fortress and town are likely to have contained buildings constructed almost entirely of wood.

There is further, though slight, evidence from epigraphic sources for this period. The presence of the Ninth Legion is vouched for by tomb inscriptions and by stamped tiles. The origines of two soldiers are given on their tombs, Vienne (Isere) in south Gaul (fn. 45) and Novara near Milan. (fn. 46) There is also a record of the unusual presence in a western legion of a soldier from the eastern half of the empire. (fn. 47) The suggestion has been made that he was brought over by the legionary commander, Caristanius Fronto, who came from Pisidian Antioch and had held previous command in the east. There are also two stamped bronze plates in Greek dedicated by a secretary, called Demetrius. He has been identified with the Demetrius mentioned by Plutarch who met him at Delphi in 83 or 84, newly returned from intelligence work in northern Britain. (fn. 48)

The beginning of the 2nd century was marked by a firm consolidation of the military area in Roman Britain under Trajan. Earth and timber defences were replaced by stone at Caerleon in 99 (fn. 49) and at Chester in 102. (fn. 50) At York this rebuilding began in 108 and 109 as is known from the surviving portion of the fine building inscription from the south-east gate in King's Square. (fn. 51) The suggestion has been made that rebuilding may at first have been limited to the towers and gates and that a stone curtain was not built until Hadrian's reign. (fn. 52) A fragment of a building inscription with part of that emperor's titles, shows that some building in stone took place in his reign but it is not certain that this stone originally lay in the fortress.

Another change occurred in the early 2nd century: the garrison changed from the Ninth to the Sixth Legion. Both the precise date and significance of the change are uncertain. The Ninth was in York in 108–9; the Sixth had certainly replaced it by the middle of the 2nd century as is known from Ptolemy (fn. 53) and from early legionary burials on Baile Hill. (fn. 54) The orthodox view is that the Sixth was brought over by Hadrian in 121–2 to replace the Ninth which had suffered so heavily in the serious troubles that followed the end of Trajan's reign as to necessitate its removal from the Roman army. This view has been challenged and a later date suggested for its departure from York, (fn. 55) but the arguments advanced have not found general acceptance. They do, however, prove that the legion was not, as has been suggested, 'annihilated' (fn. 56) and they show the lack of evidence for the suggestion that the disaster took place at York. (fn. 57) The slow promotion of some surviving officers reflects the disgrace which the legion suffered and supports the suggestion of a recent writer that it was 'cashiered . . . after an ignominious defeat'. (fn. 58)

The fortress that the Sixth Legion took over is only imperfectly known. As has been said, its defences may still have been those of the earlier earth and timber fortress with the addition of stone towers and gates. Nothing is known about the gateways but they were presumably faced with magnesian limestone like the block on which the Hadrianic inscription was carved. More is known about the towers. The foundations of the east corner tower and of the interval tower immediately to the north-west have been recovered; both were internal. The foundations of the corner tower were of clay and cobble, 1½ ft. deep and 5 ft. wide; they were not rectangular in plan but had sides splayed outwards so that the tower was narrower at the back than the front. Such foundations could have accommodated a tower with sides and back nearly 20 ft. long. The superstructure had been completely replaced by a succeeding tower. (fn. 59) The interval tower lay about 125 ft. north-west of the angle tower, an interval which implies the existence of six towers on the short sides, although they may not have been exactly evenly spaced. As with the angle tower, the superstructure had been replaced and only the foundations remained. These, like those of the corner tower, were of clay and cobble but in this case were 4 ft. deep; they would have accommodated a tower 17 ft. wide with 12 ft. sides. (fn. 60)

The stone wall which was associated with these towers was found by S. N. Miller near and on the east corner. (fn. 61) In the Museum gardens on this front there is evidence for the filling in, during this period, of the inner 1st-century ditch, an operation which could have been preparatory to the erection of a wall. (fn. 62) It seems safe to assume that this fortress was of the same extent as its predecessor and successors, although no other structural remains of it have yet been found despite careful excavations on the southwest front. The wall found by Miller had been built immediately in front of the 1st-century earth rampart. It was about 5 ft. wide, of unknown height and was composed of concrete and rubble faced inside and out with neat courses of magnesian limestone blocks. Its foundations were weak and consisted of cobbles run with mortar; the clay beneath it was strengthened where necessary with piles. In one place there had been considerable settlement and it is probably this weakness of the foundations rather than a deliberate under-cutting by the Maeatae (fn. 63) which led to the complete replacement of these walls at the end of the century. At the corner, the foundations have a slightly different angle of curve from that of the later wall, and both here and where they have been found on the south-east side, they have been left in front of the later wall to give it additional support.

The internal buildings of the fortress were also rebuilt in stone in the early 2nd century. A long range of buildings of this period built into the rampart on the north-west side was excavated by Miller and were probably store buildings. (fn. 64) A similar range on the north-east side may also have been built at this time. (fn. 65) In St. Sampson's Square were found the stables belonging to the small cavalry detachment that formed part of every legion. (fn. 66) These buildings had walls 2 ft. to 2 ft. 6 in. thick and were roofed with the usual red tiles. Nothing is known of the sites of the other main buildings but their position may be guessed from what is known of the typical plans of other legionary fortresses. The streetplan is fairly certain. From the south-west gate— the porta praetoria—the via praetoria followed approximately the line of Stonegate: the metalling found there (fn. 67) cannot be dated but the street-plan of the fortress was probably the same throughout its history. The via principia, running at right angles, followed the line of Petergate and again its metalling has been found. (fn. 68) North-east of their junction are likely to have stood the central administrative buildings of the fortress. Massive column bases found near the west end of St. Michael-le-Belfrey Church (fn. 69) and a solitary base found under the minster are all that remain. (fn. 70) Behind the central buildings, the via decumana carried on the line of the via praetoria and led to the porta decumana or north-east gate. Part of its metalling and of a colonnade on its south-east side have been found under the Treasurer's House. (fn. 71) The intervallum road, following the inside of the defences, has been discovered in Davygate; adjacent to it have been found the remains of a large structure built into the back of the defences and the end walls of three barrack blocks of which the long axes are parallel to the long axis of the fortress. (fn. 72) The barracks belong to a later period but must have had predecessors on this site: walls found adjacent to them on the other side of Davygate are of two periods. (fn. 73) Fragments of sewers and water-mains have also been found. (fn. 74)

The military activity in the north under Antoninus Pius and the disturbances in the opening years of the Emperor Commodus probably had their repercussions at York but they do not seem to have left archaeological trace. The withdrawal of troops from Britain in 196 by the governor Albinus to support his claims to the imperial throne led to the disastrous incursion of the Maeatae from the north. At York it was necessary to rebuild the fortress defences completely and this work was probably finished by the time Severus arrived in 208.

Considerably more is known about the Severan defences than about their predecessors. It is to this period that the remains exposed at the east corner of the fortress, adjacent to the Merchant Tailors' Hall, belong. They survive to the base of the parapet and it is possible therefore to reconstruct their whole arrangement. (fn. 75) Wherever found, the new curtain wall has been cut back into the 1st-century earth bank, and the early-2nd-century stone wall completely removed, except where, as has been said, its foundations were allowed to remain in front of those of this new wall. The wall was 6 ft. wide at its base decreasing to 5 ft. above, and stood to a height of 16 ft. above the foundations. It was faced externally with neat blocks of magnesian limestone with narrow joints; internally, the facing was of dressed stone but less neat. At the base of the wall was a chamfered plinth and a projecting cornice ran along the top below the parapet. The core was a strong concrete of rubble and cement. The difference in width between the bottom and the top of the wall was in places accomplished by a series of narrow off-sets on the inside, but elsewhere (fn. 76) there are rough, sloping footings internally above the concrete foundation and at the base of the dressed face. The work was done in sections by separate working parties: S. N. Miller was fortunate enough to locate the junction of two such sections north-west of the east corner. (fn. 77) An inscription on one of the facing stones at the east corner records the work of the Tenth Cohort. (fn. 78)

On top of the surviving wall there was almost certainly a parapet with a platform behind it. If a width of 2 ft. is allowed for the parapet the platform would have a maximum width of 3 ft. This would be a little narrow for a rampart walk and it was therefore supplemented by a cobbled walk running about 6 ft. below the parapet platform, along the top of the strengthened earthwork defences of the 1st century. This cobbled walk survives and can be seen adjacent to the north-east corner tower.

The east corner tower rested on the foundations of its predecessor and, like it, was internal. The slightly different curve of the new curtain wall placed the tower walls on a different line from the earlier tower foundations on which they stand: but the 5-ft.-wide foundations of that earlier tower could easily accommodate the 2 ft. 6 in. walls of the later tower. The lower part of the new tower walls, covered by the 1st-century rampart bank, consists of a core faced internally and externally with rough undressed blocks of oolitic limestone. Above the level of the rampart bank the external face is dressed. The side walls are 14 ft. and the rear wall 22 ft. long. The tower was hollow except for the bottom 3 ft. 6 in. where the 1st-century rampart bank was left intact. The basement was floored with concrete, was only 5 ft. high, and was approached by a manhole at the back. There was an entrance to the tower above from the rampart walk but insufficient details have survived to say anything about the arrangements of the upper part of the tower. The basement was subsequently filled in, possibly to support the heavier artillery of the later empire. (fn. 79) An interval tower rested on the foundations of the earlier interval tower 125 ft. north-west of the corner. (fn. 80)

To the Severan period or the early 2nd century belongs the gatehouse in St. Helen's Square whose plan can be reconstructed from remains found at various dates from the 18th century onwards. (fn. 81) It was the main gate of the fortress, the porta praetoria, set centrally in the south-west side and was built as a separate structure independent of the walls. Only half the gate has been observed but it may be assumed that it was symmetrical. Massive foundations, 11 ft. 6 in. wide, lay under the front and rear of the gatehouse which projected 2 ft. 6 in. in front of the curtain wall; the cross walls rested on foundations 5 ft. wide. The building was about 85 ft. long and 40 ft. wide. There were two arched entrances 15 ft. wide opening on to passages 4 ft. wider. On either side were guard chambers with internal dimensions of 12 by 18 ft. Later alterations have complicated and obscured the plan, making it difficult to interpret. It is clear from what was discovered that at a subsequent period the gate projected much further beyond the line of the curtain wall and it is natural to link this with the 4th-century reconstruction of the river front with its projecting interval and corner towers.

Nothing is known about Severan alterations to the interior of the fortress. The domus palatina referred to in accounts of the emperor's death was presumably something more than the normal legionary commander's residence and may have been outside the fortress. York became, after Severus's division of Britain into two provinces, the capital of Britannia Inferior and the domus palatina perhaps became the official residence of the governor. Outside the fortress, a wall of military character has been found which continues the line of the south-west wall of the fortress at least another 100 yds. to the north-west. (fn. 82) It is possible that it may be the wall of an annexe to the fortress designed to provide some of the extra military accommodation required during the emperor's stay at York.

The military and administrative arrangements of Severus and his successor Caracalla guaranteed peace to the north until the late 3rd century. In 287 Carausius, while in charge of a fleet to suppress Saxon piracy in the channel, revolted and assumed sovereignty in Britain. His successor, Allectus, in order to maintain his position, withdrew troops from Britain. Although Allectus was defeated by Constantius Chlorus in 296 and Britain was reunited with the empire, the north had meanwhile been invaded and wasted. The damage was made good and the fortress at York rebuilt. The curtain wall was rebuilt on the old foundations from the north-east gate to the south-east gate. On the river front projecting interval and corner towers of new design replaced the earlier internal ones, but on the northwest, north-east, and presumably also the southeast sides, the towers were either rebuilt on the old plan or the old towers remained in use.

The new curtain wall, large parts of which survive particularly at the west corner, was higher than its predecessor: it was 17–18 ft. high excluding most of the parapet. Behind it the old earthen bank was increased in width and height to provide the base for a rampart walk at a height 4½ ft. below the parapet platform. The hidden, inner face of the wall was faced with rough, undressed stone, very unevenly coursed. Externally the facing was of dressed magnesian limestone blocks similar to those of the Severan wall but not so neatly built. A band of five courses of brick, 7–8 ft. above the foundations, enhanced the appearance of the wall and helped to bond the facing to the core. It is probable that similar brick courses ornamented the parapet. Between the two faces of the wall was a core similar to that of the Severan wall but containing a considerable amount of debris from destroyed buildings. (fn. 83)

On the south-west or river front, towers were added of entirely new design. Of these the 'Multangular Tower' at the west corner is the best known; it survives to a height of 19 ft. (see plate facing p. 332). (fn. 84) The corresponding tower at the south corner is preserved in the basement of a large store in Feasegate. (fn. 85) These two large towers project well beyond the line of the curtain walls: in plan each forms ten sides of a regular 13-sided figure, interrupting the curve of the curtain wall at the corner. The maximum internal width of the west tower is 35 ft. 6 in. Between its sides and the external face of the wall are sharp re-entrant angles overlooked by two embrasures in the tower wall. Internally, it was divided by a cross wall which extended into the interior of the fortress; behind the tower, inside the fortress, was a forecourt or ante-chamber. The curtain wall was not carried across the entrance of the tower but was bonded with the side walls. Across the 21-ft. gap at the entrance the rampart walk was probably carried on a double arch supported on the cross wall. Externally the tower is of similar masonry to the curtain walls and the band of brick is continued round it; internally, unlike the walls, it has a neat facing of dressed limestone. The tower is considerably thickened at the base, decreasing from 6 ft. 9 in. at ground level to 3 ft. at a height of 17 ft. Its original height is uncertain. The south corner tower is of similar plan and construction; its walls survive only to a height of 8 ft. but are much better preserved, particularly in the details of the external plinth. Reused material on the inner face includes a centurial stone built in upside down.

Remains of two projecting interval towers have been found. There were six towers symmetrically spaced three on either side of the porta praetoria. In plan these towers have six equal sides projecting 17 ft. from the wall. The greatest width of the towers was 32 ft. One of these towers was, in 1958, exposed in the Museum gardens, 125 ft. south-east of the Multangular Tower; (fn. 86) the other, 125 ft. farther to the south-east, lay beneath a shop in Lendal. (fn. 87) The masonry of the towers was similar to that of the walls; in the Museum gardens the facing of the tower is bonded with the curtain wall to form a continuous face. The foundations, however, were built against the existing Severan foundations used by the curtain wall.

On the north-west side of the fortress the towers were internal as in the earlier period. One, measuring 13 ft. square internally, survived to a considerable height and was discovered and destroyed during the construction of St. Leonard's Place. The upper part of this tower, rising above the curtain wall, had two openings in the front which showed signs of the wear caused by the artillery the tower had housed. There were side openings on both sides grooved for shuttering to protect those in the tower. (fn. 88) The tower lay 125 ft. south-west from the north-west gate of the fortress. Remains of a similar tower were found 125 ft. farther south-west. (fn. 89) These intervals, the same as those on the south-west side, suggest the existence of three towers between the gate and the west corner. North-east of the gate, however, the position of yet another internal interval tower found by S. N. Miller (fn. 90) implies that different intervals were maintained between the gate and north corner.

Remains have been found of the north-west gate, nearly on the site of Bootham Bar. There appears to have been a double-arched entrance through the centre of a gate-house, with guard chambers on either side. The gate-house measured externally 76 ft. by 30 ft., with 7-ft.-wide passages; the guard chambers were 20 ft. square internally. The whole structure projected little at all beyond the line of the wall and was built with large blocks of dark gritstone. (fn. 91) Fragments of several decorative carvings have survived, including a deity driving a quadriga and a triton blowing a conch-shell trumpet. (fn. 92)

The remains of the south-west gate suggest that it was modified at this time. The major part of this structure has been described; (fn. 93) some walls continuing forward beyond the main mass of the gate suggest that it may at this period have been rebuilt to conform with the new towers on this front and have projected beyond the curtain wall.

Inside the fortress considerable alterations were probably made. Part of a large bath-house has been found in St. Sampson's Square. It is unusual to find a bath within a fortress; to accommodate it, earlier buildings, including the stables already referred to, (fn. 94) had been cleared away. The barracks, of which the end walls have been discovered, seem to have been renewed at this date. (fn. 95) The store buildings on the north-west side of the fortress had now fallen into disuse, and indeed were buried under the newly extended rampart bank. (fn. 96)

On the death of Constantius Chlorus, his son, Constantine the Great, was proclaimed emperor at York. A fine head from a greater than life-size statue of this emperor is now in the Yorkshire Museum: it perhaps stood in front of the headquarters building (see plate facing p. 332). (fn. 97)

The Roman garrison remained at York for another century but there is little evidence of structural alterations during this period. The large bath-house became disused and was adapted for other purposes. (fn. 98) The building of the chamber, which remains behind the Public Library and which was probably the undercroft of a tower, belongs to a period when the wall was already ruinous and breached. (fn. 99)

The garrison of the fortress continued from the early 2nd century until the end of Roman rule to be the Sixth Legion. Epigraphic evidence at York for this legion is naturally more abundant than for the Ninth. There is a fine series of tile stamps giving, in addition to the legion's regular title of Victrix pia fidelis, those of Gordiana and Severiana. (fn. 100) Claudius Hieronymianus, legate of the Sixth, built a temple to Serapis. (fn. 101) Q. Antonius Isauricus, also a legate, almost certainly commanded the Sixth; his wife set up an altar to Fortune. (fn. 102) Centurions include Aurelius Super, (fn. 103) Flavius Flavinus, (fn. 104) Felicius Simplex (fn. 105), Attius Severus, C. Aprilis, (fn. 106) and Septimius Lupianus. (fn. 107) The last was a centurion ex evocatis, that is to say that after service in the ranks of the praetorian guard at Rome he had been chosen for promotion to the legionary centurionate. The fine coffin he provided for his wife, found near that of Aurelius Super in the Castle Yard, stood originally in a tomb; it was re-used for an interment which, though still in the Roman tradition, may have occurred at a time when the centurionate no longer existed at York to protect its tombs from violation. On centurial stones on the fortress wall are recorded the names of two more centurions: Antonius (fn. 108) and Iullinus. (fn. 109) Two veterans of the legion are recorded, Caeresius (fn. 110) and Manlius Cresces (fn. 111) and two soldiers, L. Baebius Crescens of Augusta Vindelicium (Augst in Austria) (fn. 112) and M. Minicius Mude. The last set up an altar to the mother goddesses of Italy, Africa, and Gaul, presumably the three areas which then provided recruits for the legion. (fn. 113) Vespasian is sometimes said to have excluded Italians from the legions but another altar, from Hadrian's Wall, dedicated by cives italici et norici serving in the Sixth, confirms their service in that legion. (fn. 114)

The Colonia

Out of the canabae developed a thriving town which by the 3rd century was dignified by the title and privileges of a colonia. The name of the colonia is given in abbreviated form on three inscriptions and may be expanded colonia Eboracensium. (fn. 115) The date of its creation is not known. The name is mentioned in connexion with the date 237, on an altar found at Bordeaux. (fn. 116) A Severan date has been suggested (fn. 117) and in view of Severus's sojourn there, together with his court, his creation of York as the capital of the new province of Britannia Inferior, (fn. 118) and his known interest in the military towns that had grown up alongside fortresses, (fn. 119) it seems likely that he was responsible for the elevation of the town. An alternative suggestion links the creation of the colony with a reference by Pausanias to a punitive curtailment of Brigantian territory by Antoninus Pius and the use of the confiscated land for this purpose. (fn. 120)

Evidence for the corporate life of the town is provided by two sepulchral inscriptions. The stone coffin of Flavius Bellator, a decurion of the colony, was found in the Roman cemetery that occupied the site of the new railway station and sidings. It contained his skeleton, still wearing his official gold ring set with a ruby. Nearby was found the coffin of Marcus Verecundius, a sevir of the colony, who came from Bourges (Cher), (fn. 121) and that of his wife, Julia Fortunata, from Sardinia. (fn. 122)

The site occupied by the colonia seems from the preponderance of the remains to have been southwest of the Ouse where higher land provided a capacious and well-drained site above flood level. As already mentioned, timber buildings of the 1st century show that this site was already used then, but the evidence of burials (fn. 123) implies that the main development on this side of the river did not occur until well into the 2nd century. The destruction during the disturbances that followed Albinus's withdrawal of troops in 195 will almost certainly have caused as much damage in the civil town as in the fortress. It is therefore probable, although no archaeological evidence has yet been found, that the town was almost wholly rebuilt under Severus. The cessation of civil occupation along Bootham and Clifton on the north-west side of the fortress at the same period (fn. 124) is consistent with a greater concentration of the town closer to the fortress on the south-west bank of the river. New building at the turn of the 2nd and 3rd centuries is attested only by the dedication of a new temple to Serapis by Claudius Hieronymianus, the legionary commandant. (fn. 125)

The area of the colonia on the south-west side of the river can be defined first by the actual area in which buildings have been found. This evidence is inconclusive because it depends on chance finds made during building or drainage. Secondly, the area may be defined by burials because the Romans enjoined that these should be made outside the town; but this evidence is difficult to interpret because the later colonia extended over earlier burials and the date at which the main extension took place is only approximately known.

These two criteria, however, may be used to give the following approximate boundaries. The buildings are concentrated in the Micklegate and Toft Green areas; near the old railway station they extend right up to the medieval wall in at least two places. (fn. 126) Burials have been found immediately outside the medieval wall at this point (fn. 127) and it may safely be assumed that the wall here represents the boundary of the colonia. Near Micklegate Bar, however, there is evidence for two buildings outside the medieval wall; (fn. 128) inside the Bar they extended right up to it. (fn. 129) The town, therefore, may have extended a little beyond the medieval wall here, but the junction of the Aldborough and Tadcaster Roman roads found near Blossom Street was certainly outside the town. (fn. 130)

On the south-east side of the colonia the boundary is less certain. The few structural finds are a considerable distance from the medieval walls; and if some of the pottery found in Priory Street (fn. 131) is sepulchral, as may well be the case, then it seems that the medieval town was here larger than the Roman. The farthest south-east at which structures have been found is near the churches of St. Mary, Bishophill, Junior (fn. 132) and Senior. (fn. 133) A mosaic pavement found on Cherry Hill, (fn. 134) some distance to the south-east, is separated from the other two finds by burials (fn. 135) and seems to be suburban. These burials appear to belong to a small cemetery which lay mainly outside the walls in the Clementhorpe area.

The conclusion seems to be that the Roman town south-west of the Ouse extended from the river as far as the walls of the medieval town north of Micklegate; to the south there is no evidence that it reached the medieval limits. The built-up area on this side of the river covered at least 50 acres. The colonia was probably defended by a wall at least in the 4th century. On the north-west side the medieval wall has in three places been found to overlie an earlier wall (fn. 136) but this has not yet been shown to be Roman.

The axis of the colonia was the main Calcaria road. In relation to later topography the road crossed the medieval wall at a point 130 ft. north-west of the centre of Micklegate Bar, passed through the gardens behind houses in Micklegate, forming, as it were, a chord to the curve of Toft Green and Tanner Row; it crossed Tanner Row just above the junction with Rougier Street and reached the river nearly opposite the Guildhall. (fn. 137) Another street ran parallel to this through Micklegate Bar, following approximately the line of Micklegate until that road curves away towards Ouse Bridge. Its pavement was found under the front of three houses at the corner of Barker Lane. (fn. 138) There is some slight evidence for a grid of streets, later distorted to form the basis of the medieval street plan. (fn. 139)

The colonia had a public water system; a street fountain, protected by a vaulted roof, was found outside the church of St. Mary, Bishophill, Junior and the cistern is now preserved in the Museum gardens. (fn. 140) The remains of several buildings have been found by accident in the area of the colonia but are inadequately recorded. Among the more important are the following. In Railway Street near the junction with Tanner Row were found the bases of a double colonnade with massive columns 3 ft. in diameter and set only 6 ft. apart. (fn. 141) Other fragments of a colonnade or portico were found in Micklegate at the other end of Railway Street. (fn. 142) Various finds made during the erection of the new railway offices included a building with a heavy gritstone façade facing the main Calcaria road; behind it lay fragments of other buildings and building debris, including three stone capitals, one enriched with acanthus leaves. (fn. 143)

Extensive remains of important public baths were found during the erection of the old railway station and again in the same area in 1939; they included what has been described as the largest caldarium in Britain. (fn. 144) Claudius Hieronymianus's shrine of Serapis comes from the same area, (fn. 145) as does a statue dedicated to the god Arimanes. (fn. 146) A Mithraeum lay nearer the river: its reredos was found opposite St. Martin's Church in Micklegate and bears upon it a notable scene of the bull-killing. (fn. 147) From Toft Green come several mosaic pavements, including one lying under the mound of the medieval wall near Bar Lane with the unusual decoration of haunches of venison. (fn. 148) Numerous building finds have been made under Micklegate (fn. 149) and a colonnade was found in Trinity Lane. (fn. 150) Domestic baths come from Fetter Lane, south of Micklegate. (fn. 151) During the rebuilding of the medieval bridge across the Ouse many Roman architectural fragments were found that had belonged to structures with some pretensions to grandeur. (fn. 152) Most of the walls found were of stone but there is some evidence that some at least of the buildings, although based on stone walls, may have borne half-timbered superstructures. (fn. 153)

East of the Ouse, evidence of buildings comes mainly from the triangle of land between the Ouse and the Foss in the lee of the fortress. It is in this area, on the banks of the Foss, that indications of an extensive wharfage system have been found. (fn. 154) Finds of carbonized wheat in Market Street suggest the presence of granaries nearby. (fn. 155) Fragments of a building have been found in Parliament Street. (fn. 156) Buildings of a non-commercial nature have been found farther west in the Ousegate-Castlegate area. Part of a tessellated pavement was found under a buttress of St. Mary's Church in Castlegate; (fn. 157) at the corner of Nessgate and Ousegate were found various architectural fragments including two building inscriptions, one for a temple of Hercules and the other for a temple devoted to emperor worship; (fn. 158) and at the corner of Ousegate and Spurriergate various successive structures have been found alongside the main road to the south-west gate of the fortress, including part of a large bath house. (fn. 159)

North-west of the fortress there is clear evidence from the Museum gardens of occupation from the 1st to the 4th centuries, but there is no evidence north of Marygate for any use of the land after the 2nd century except for burial purposes. (fn. 160)

Suburban development was mainly limited to the cemeteries. At first haphazardly developing along the main approach roads, these came later to be concentrated into three main areas: along the Calcaria road, on the site of the new railway station, and northwest of the fortress between St. Mary's and Clifton. There were other, smaller cemeteries, notably in the Castle Yard and Fishergate areas.

These cemeteries have produced a wealth of grave goods and are remarkable for the variety of burial methods. Cremation in the early burials was replaced during the 3rd and 4th centuries by inhumation in large stone coffins, (fn. 161) stone or tile-built cists, (fn. 162) wooden iron-bound coffins, (fn. 163) lead coffins, (fn. 164) or simply in graves without coffins. (fn. 165) Perhaps the most remarkable burials are those where gypsum was poured in and has produced casts of the body, preserving, in some cases, clothes and hair. (fn. 166) A remarkable range of ornaments, toilet, and trinket sets with remains of the bronze-bound dressing cases that contained them, (fn. 167) a folding fan, (fn. 168) and parts of a parasol, (fn. 169) both in ivory, are among the grave goods. A Christian burial from Sycamore Terrace with an open-work motto in bone, SOROR AVE VIVAS IN DEO, was equally lavishly equipped with jewellery and glass vessels for food and drink. (fn. 170) Some of the jewellery from the graves is made of jet and objects ranging from plain bangles to elaborate necklaces and carved pendants were made in this material. (fn. 171) Some of the stone coffins were inscribed and remains exist of at least two that were elaborately carved as, for example, that of Julia Victorina. (fn. 172) Coffins such as these were probably associated with fine tombs, but almost all of these have perished. On The Mount is preserved the only example of a vaulted tomb-chamber. (fn. 173) Fragments of sculpture survive from tomb monuments: a female head in 3rd-century coiffure from Fishergate, (fn. 174) the shaft of a monumental candelabrum, (fn. 175) a fragment of a stone chair, (fn. 176) and a noteworthy figure of a harpy (see plate facing this page). (fn. 177) One other fine sculpture which is not funerary should be mentioned; it is a life-size statue of the youthful Mars in panoply and comes from a mixed cache of stones in Blossom Street. (fn. 178)

The inscriptions on the tombs of two local magistrates, the one a decurion and the other a sevir Augustalis have already been mentioned. These seviri were drawn from the freedmen class and were charged with the local emperor worship. Their wealth was mainly based on commerce and their cosmopolitanism is illustrated by the birthplaces, Bourges and Sardinia, of Verecundius Diogenes and his wife. The Bordeaux inscription of Marcus Aurelius Lunaris, who was a sevir Augustalis both of York and Lincoln, and who travelled to Aquitania, presumably in connexion with the wine trade, illustrates the trade connexions of York. Lunaris was perhaps an army contractor with civilian clients in both coloniae. (fn. 179) Nicomedes, who erected a statue to Britannia, now lost, was a freedman of the emperor and probably a minor civil servant. (fn. 180)

The army no doubt provided York with the bulk of its wealth but the town must also have served the civil population as a commercial centre. There were minor local industries such as those which produced the jet and bone ornaments, (fn. 181) and coarser pottery was most probably made locally; but scarcely any evidence has survived of the sites of these industries. (fn. 182)