Rome's frontiers in northern England
1 This chapter concerns military developments in northern England from AD69 up to the time of the building of forts on Hadrian’s Wall, a period of some 56 years. The frontier arrangements seem reasonably clear in the Trajanic and Hadrianic periods, less so in the Flavian period, so it might seem easier to begin with the Trajanic period, proceed to the Hadrianic period and then go back to consider the Flavian period. However, in order to ensure that events may be seen in their proper chronological sequence we will in fact start with the Flavian period, sketchy though the evidence for that period is. Unless otherwise stated all Ravenna names listed below are in the form given by Richmond and Crawford in Richmond and Crawford 1949. In general no explanation is given below for the identification of any particular Romano-British place-name – for such explanation the reader should consult Chapter 15: Roman place-names in the North of England.
2 Flavian period
2.1 The argument presented here is based on the assumption that in AD69 or 70 Vespasian gave the order to subjugate the Brigantes and that this order was given after Queen Cartimandua had lost control of the tribe and it had become clear that her successor in power, apparently her former husband Venutius, was hostile to Rome. Tacitus tells us that the Brigantes were the most populous tribe in Britain and Ptolemy indicates that they extended "to both seas", that is to say right across the north of England. The Romans were of course aware that there were other tribes further north than the Brigantes. Ptolemy mentions 18 such tribes, though it is not absolutely clear to what period his information relates. But it is safe to assume that the last thing the Flavians wanted to do was to push the Brigantes to the north where they might coalesce with other tribes and pose a serious threat to Rome's interests at some future date - much better to defeat them in the north of England and bring them under Roman control. But how could the Romans achieve this goal? On the face of it, it could most easily be achieved by moving rapidly up the eastern and western sides of the Pennines, establishing a barrier across the northern end of Brigantian territory and then breaking that territory up into several smaller regions which could then be conquered one at a time. Ravenna may come to our aid here, for whereas in Chapter 15 the writer suggested that the Ravenna names in the north of England could be divided into three groups, it does in fact appear that they can be divided into six groups, as follows:
1) Camulosessa Presidium (Malton) to Alavna (Alnwick);
2) Navione (Brough-on-Noe) to Lagentium (Castleford);
3) Valteris (Brough-under-Stainmore) to Lavaris (Bowes);
4) Cactabactonion (Catterick) to Corielopocarium (South Shields);
5) Brocara (Brougham Castle) to Corda (Leicester); and
6) Serduno (Wallsend) to Maia Fanococidi (Kirkbride).
2.2 It seems possible that the names of groups (1) and (2), referred to in paragraph 2.1 above, document the progress of the Romans up the eastern and western sides of the Pennines and the forts of group (3) not only provided a cordon across the northern end of Brigantian territory but also gave the Romans control of the Stainmore pass. It seems probable that all of the forts of groups (1), (2) and (3) were completed within two seasons of Vespasian giving the order to conquer the Brigantes.
2.2.1 It is worth looking at these groups one at a time, those of group (1) being shown on the maps below (the numbering being that of Richmond and Crawford).
This group starts at Camulosessa Presidium at Malton, proceeds to Abisson at Durham, then goes over via Ebio at Ebchester and Coritiotar at Corbridge to Celovion at Chesters on the North Tyne. Then, amongst other names, it includes Duabsissis at Berwick-on-Tweed, Trimuntium at Newstead and Eburocaslum at Broomholm. The group looks very early. Firstly, Durham, Ebchester, Corbridge and Chesters are in fact aligned with one another, which rather suggests that the sites for the forts were selected and the forts built before the road system was built in that part of the country. Secondly, Coritiotar, a slightly modified form of the Celtic name Coritisotar (see the entry for Coritiotar in the Alphabetical List), was probably the early fort at Red House, a little to the west of the later industrial/military complex at Corbridge. This industrial/military complex became known as Coritisotaroppidum and this form, with omission of some internal letters (a common change in Romano-British place-names) yielded Corstopidum. This last form, with the common d→t change, yielded the Corstopitum of the Antonine Itinerary. The third point is that Celovion would appear to be an unaltered Celtic name including the old-style element celov meaning 'steep hill slope', and was presumably the name of the Iron Age settlement standing on a steep slope immediately east of the village of Wall, on the eastern side of the North Tyne. The name will simply have been transferred by the Romans to their earliest fort in the Chesters area. As explained in Chapter 8 the names from Serduno to Maia Fanococidi in Ravenna define the Trajanic frontier, and this group includes Celunno at Chesters on the North Tyne. It is possible to see development from Celovion to Celunno via Celonion, but difficult to see development from Celunno to Celovion. Instead Celunno appears to have developed into the Cilurno of the ND. Perhaps the early fort had lain empty for a few years after Agricola marched north to Scotland and its name was revived, slightly modified to Celunno, and applied to a Trajanic frontier fort built in that area, probably on the Stanegate and possibly close to the point where the road crossed the North Tyne. Some 20 years later the name will have been transferred to the Hadrian’s Wall fort upstream at Chesters, the name changing at some point to the Cilurno of the Notitia Dignitatum. It thus seems likely that Celovion was a Flavian fort in the Chesters area and Celunno a Trajanic fort in that same area but on the Stanegate. The Hadrian’s Wall fort was probably originally called Celunno but at some point the name was modified to Cilurno. Another point to note regarding the names of group (1) is that the 9th legion was no doubt active up the east side of the Pennines before it moved to York in AD71. It is thus even possible that the fort Camulosessa Presidium at Malton is older than the legionary fortress at York. Indeed the term Presidium may indicate that Malton enjoyed some special status, perhaps as a forward command centre when the legion was still based at Lincoln.
It would be possible for the Romans to build the more northerly forts of group (1) without opposition since the Votadini were on friendly terms with the Romans (assuming we are right on this point) and in any event the forts at Newstead and Broomholm were presumably in the frontier zone between the Votadini and their neighbours to the west and north and so helped protect the Votadini and their territory. The one surprising point is that the names only go as far north as the river Tweed, whereas in the past it has been assumed that the Votadini occupied all of the land between the Tyne and the Forth. Perhaps we need to rethink this point (see Chapter 25, 2.2 and 4.1). The three forts listed after Alavna in Ravenna - Olcaclavis, Evidensca (using the form of the name given by Rivet and Smith) and Rumabo - are indeed on the south side of the Forth estuary, but they are perhaps more likely to have formed part of Agricola's activity in AD80, when he built posts across the Forth-Clyde isthmus. If those three forts really belonged to group (1) one would expect them to be listed after Trimuntium or Duabsissis rather than after Alavna.
One last point about the names in this group - Duabsissis may possibly have been on the Tweedmouth side of the river Tweed, at the end of Margary road 87, rather than at Berwick itself. The name refers to a location at the top of a steep slope and overlooking the river Duabs, i.e. the Tweed, and there are indeed suitable locations on the Tweedmouth side of the river. But Duabsissis is a Celtic name transferred to a Roman fort/harbour, so it is quite likely that Celtic Duabsissis was at the top of the steep slopes on the Berwick side of the river and that Roman Duabsissis was at Tweedmouth.
2.2.2 Looking now at the names of group (3) the names appear to form a ring of forts around a substantial piece of territory at the northern end of the Pennines, as is clear from the map below.
The one strange point is that Gabaglanda at Birdoswald sticks out to the north of that ring. This appears to suggest that the fort at Birdoswald formed a link to that at Broomholm, founded earlier, so that the troops in both forts could co-operate in patrolling the frontier zone between the Votadini and the Selgovae.
So far as dates are concerned it is believed on the basis of dendrochronological evidence that Flavian Carlisle was built in the early 70s AD, perhaps AD72 (see, for example, Britannia 21 (1990), 320). And it is likely that the compiler of Ravenna saw the names of group (3) on one map, quite divorced from other names - a map in an end-of-campaign or end-of-season report, as it were. It is therefore likely that all of the forts in group (3) were built in the early 70s AD.
It can be seen that the forts Lagubalium to Lincovigla of group (3), together with Abisson to Celovion from group (1), effectively formed a barrier across the northern end of the territory of the Brigantes, thus not only preventing tribes living further north from coming to the aid of the Brigantes but also, and just as important, preventing the Brigantes from moving north - they were now penned in, ready to be brought under Roman control. But there was a weakness in the Roman position, and that weakness lay in northwest England. It was still possible that northern tribes might cross the Solway to come to the assistance of the Brigantes, or that some at least of the Brigantes could slip away over the Solway to what is now southwest Scotland in order to escape Roman domination. It is precisely for this reason that the forts of group (2) are likely to be early, to date from the early 70s AD, around the time the forts of groups (1) and (3) were built.
2.2.3 The forts in group (2) are shown on the maps below.
|Gabrocentio||=||Hard Knott||Zerdotalia||=||Melandra Castle|
It is believed on the basis of dendrochronological evidence that Bresnetenaci at Ribchester was built in the early 70s AD (see, for example, Britannia 21 (1990), 328) and it seems to be generally agreed amongst scholars that Lagentium at Castleford dates from the same period. Now, if all of the Ravenna names for the north of England had appeared on one map it is highly unlikely that the compiler of Ravenna would have picked out the above names for inclusion in one group. It is much more likely that he saw the names of group (2) on one map, quite divorced from other names. That map may have formed part of an end-of-season report documenting the progess of Roman troops up the western side of the Pennines and so it is probable that all of the forts of group (2) were founded in the early 70s AD. This is of course contrary to conventional wisdom, which has it that the forts along the Cumbrian coast date from the 90s AD. A question to ask is why the Romans should wish to build forts on the coast at Ravenglass, Moresby, Workington and Maryport. Such forts might be built to protect harbours, but the Romans surely had no need for four harbours on that part of the Cumbrian coast. Alternatively, the forts might be built to protect the province against a perceived danger of invasion by sea, but then one might ask at what time such a danger is most likely to have been perceived. There can surely have been no need for such forts from the time of Agricola's invasion of Scotland up till the time Trajan formed his frontier between the Tyne and Solway (see below), for throughout that period the Romans had troops in the north of England and the south of Scotland and could have sent a force sufficiently large to deal with any possible invaders before they reached the northern shore of the Solway. The most likely time for building the Cumbrian coastal forts would thus be in or before AD78 or after AD103. And even this latter date can be moved forward, since although Trajan had his troops in forts spaced apart across the Tyne-Solway isthmus, some of them will have been mounted units which no doubt patrolled the region to the north of the frontier to gather intelligence, so that if any trouble were brewing in the south of Scotland the Roman commander could send a force to deal with it. It was different after Hadrian's Wall was built, for in building the Wall Hadrian was saying that the tribes north of the Wall were outside the empire and it was his intention that they should remain outside the empire. Hadrian therefore had to guard against the possibility that at some future date a tribe or tribes hostile to Rome might seek to circumvent the Wall by sailing across the Solway to attack Cumbria. It follows that the most likely time for building the Cumbrian coastal forts would be in or before AD78 or after AD123. In recent excavations at Maryport traces were apparently found of an earlier fort under the visible fort, but so far as the writer is aware no date was assigned to the earlier fort. It is also clear that there were two phases of occupation at Ravenglass, the earlier fort being called Alavna (the Romans simply transferring to the fort the name of the river on which the fort stood) and the later having a name of the isca-family, this name later being transferred by the Romans to the river, now the Esk. But again this does not tell us when the earlier fort was built, only that there was one. And we can be sure that there were two phases of occupation at Ambleside, the earlier fort being the Iuliocenon of Ravenna and the later the Tunnocelo of the ND. It is easy to see how the name changed - the initial i of Iuliocenon changed to t and the l and first n swapped places so that the name became Tuniocelon, this then becoming the Tunnocelo of the ND. Note that Iuliocenon will not itself be the original Celtic name - the iu at the beginnning is suspect and there may originally have been a consonant between the i and the o. The rivers at Ambleside come to our aid here - these are the Rothay and the Brathay. These river names are normally considered to be Old Norse, but it is just as likely that they are Celtic river-names using the river-letters b, r and t (later changed to th) corresponding to the hill-letters s, m and l. The Celtic form of Iuliocenon might then have been somewhat like Isulimocenon, or Sigulimocenon if the hill-letter s were qualified by the letter g meaning 'steep', just as the hill-letter m is qualified by the c meaning 'steep' - there are steep slopes to the east and northwest of the Ambleside fort. But again, this does not tell us when the earlier fort was founded, just that there was one. It may be noted in passing that the Notitia Dignitatum indicates that Tunnocelo was commanded by a tribunus cohortis primae Aeliae classica. If this designation of the unit - a naval unit or at least a unit raised from naval personnel - was still relevant at the date of compilation of the ND, perhaps it was not inappropriate that the unit should be stationed at Ambleside, at the northern end of Lake Windermere. One further point to note about the names of group (2) is that by building forts at Skipton, Ilkley and Castleford the Romans gained control of the Aire and Wharfe valleys, thus dividing up the Brigantian territory into various smaller regions which could be picked off one at a time.
2.3 The significance of group (4) is not entirely clear. The names of this group take us down from Cactabactonion at Catterick via Isurium at Aldborough and Eburacum at York to Decuaria, normally assumed to be the same name as Petuaria at Brough-on-Humber. They then take us back north again via Devovicia at Stamford Bridge, Dixio (most likely at Thirsk), Lugunduno on the north bank of the Tees at Dinsdale Park and Coganges at Chester-le-Street to Corielopocarium at South Shields (Dixio and Lugunduno are shown in italics above since they are actually given as one name Dixiolugunduno in Ravenna). Perhaps the names from Cactabactonion down to Decuaria merely represent a consolidation of the Roman position in Yorkshire, moving the frontier further west, closer to the Pennines, whereas those from Lugunduno to Corielopocarium link Catterick to a new supply depot at South Shields on the Tyne. It may well be the case then that group (4) dates from AD71, assuming this was indeed the year in which the 9th legion moved to York. On the other hand it may be that this group of names documents the completion of the road from Catterick down to Brough-on-Humber via York and that from Brough up to South Shields via Stamford Bridge and Chester-le-Street. If this is correct then group (4) may date from a period a year or two later than AD71.
2.4 And now perhaps we can see more clearly the contribution of Agricola, as governor of Britain, to the conquest of the Brigantes. Late in the season of AD77 (assuming he came to Britain as governor in that year) he led his troops into the mountains of North Wales and defeated the Ordovices (almost wiped them out, if Tacitus is to be believed). Perhaps there was a similar task awaiting him in the North of England - some pockets of resistance in the higher, more remote parts of the Pennines which needed to be eliminated. Note how vague Tacitus is regarding the military achievements of Agricola in his second season - he doesn't tell us in which part of the island Agricola was active, nor which tribe or tribes he was fighting against - and Tacitus is not normally sparing in his praise for Agricola's genius, wisdom and bravery. And indeed it would appear that almost all the work involved in conquering the Brigantes had already been done. Agricola had a small, but important task to carry out - and it is the names of group (5) which appear to document the campaign of his second season. The Ravenna names of this group are shown on the maps below.
|Alitacenon||=||probably Adel||Loxa||=||Exley Head, Keighley|
|Cambroianna||=||Slack, west of Huddersfield||Maporiton||=||Bramham|
But the names of group (5) should perhaps be considered in the order opposite to that given in Ravenna. Agricola may have gathered his troops at Leicester (the ones who had almost wiped out the Ordovices at the end of the previous campaigning season) and marched north to Chesterfield, then operated in the high ground to the northwest, extending up to the Aire valley, which the Romans apparently already controlled. He presumably founded Smetri somewhere to the west of Sheffield and Cambrolanda (Ravenna's Cambroianna) at Slack, to the west of Huddersfield, and ended this particular campaign at Locatreve at Trawden. He then made for Tadoriton at Tadcaster via Loxa (Exley Head, Keighley), Alitacenon (probably Adel) and Maporiton (Bramham). What is not clear is whether Agricola founded Locatreve, Loxa, Alitacenon, Maporiton and Tadoriton or whether these forts had been built earlier, around the time that the forts at Skipton, Ilkley and Castleford were built - the latter seems more likely. After Tadoriton at Tadcaster Agricola headed northwest via Carbantium (Harrogate) and Clindum (Clint) into the high Pennines, in the region between the Wharfe valley and the Stainmore pass. He presumably founded Smetriadum at Bainbridge, and ended this particular campaign at Brocara (Brougham Castle), though this may have existed already. He then operated in the high ground between the Stainmore pass and the Tyne-Solway isthmus, ending this campaign at Stodoion (Nether Denton). This fort may have been built earlier, but Agricola probably founded Croucingo at Croglin, though no fort has yet been found there, so far as the present writer is aware. That was quite enough activity for one season, but with the last pockets of resistance wiped out the military power of the Brigantes was finally broken. And so, thanks to the network of forts and the roads built by his predecessors as governor Agricola was free to invade what is now Scotland in the following season with no need to fear trouble to his rear from the Brigantes.
3 Trajanic period
3.1 In the last thirteen years or so of the first century AD the Romans appear to have given up most of their gains in Scotland but still held some forts in the south - Trimuntium and the fort at Dalswinton are believed to have been held until around AD103. It was around this time that Trajan built a string of forts stretching from one side of the country to the other, substantially on the line later to be taken by Hadrian's Wall - but there was no fort at Newcastle and to the west of Carlisle the frontier was represented by the forts at Burgh-by-Sands III and Kirkbride. The names of the frontier forts listed by Ravenna are Serduno, Condecor, Vindovala, Onno, Celunno, Brocoliti, Velurcion, Esica, Banna, Uxelludamo, Avalava and Maia Fanococidi. Magnis at Carvoran, Gabaglanda at Birdoswald and Lagubalium at Carlisle will also have been on the frontier but are not listed at this point in Ravenna because they were listed earlier within a group of Flavian forts (group (3) discussed above). An explanation of the above names is given in Chapter 8: The Trajanic frontier from Tyne to Solway. The explanation will not be repeated here, but a few points are worth emphasising. Ravenna's list, Serduno to Maia Fanococidi:
1) does not include Hadrianic Pons Aeli at Newcastle;
2) includes Celunno at Chesters on the North Tyne, this name being derived from the Flavian name for the same site - Celovion (discussed above) - and itself developing further to yield the Cilurno of the Notitia Dignitatum;
3) includes Banna, which is topographically appropriate for the fortlet at Throp but not for Birdoswald (Banna was originally Banva meaning 'high hill slope'): the fortlet at Throp is known to have been Trajanic and to have been dismantled around the time work started on Hadrian's Wall; and
4) includes Avalava, which is topographically appropriate for Burgh-by-Sands III but not for the Wall fort at Burgh-by-Sands: Burgh-by-Sands III will have been the earliest fort in that area, being replaced by Burgh-by-Sands I which stood to the south of the later Wall and will have formed part of Hadrian's original plan for the frontier: Burgh-by-Sands III will thus have been a Trajanic fort.
3.2 But whilst it may be true to say that Trajan built a string of forts substantially on the line later taken by Hadrian's Wall, this statement needs to be modified in one important respect. The problem concerns Ravenna's names Brocoliti and Esica. There can be no doubt that the corresponding ND names, Procolitia and Aesica, refer to the Wall forts at Carrawburgh and Great Chesters respectively. But Ravenna's names are earlier, they are Trajanic, and they may not be appropriate for Carrawburgh and Great Chesters. Brocoliti is a topographical name comprising the transitional element broc meaning 'high hill steep' and the inversion-type element lit meaning 'hill high', the name clearly referring to a location adjacent a steep, high hill. Now, there is a slope to the south and west of the Carrawburgh fort, but it is not clear whether it is reasonable to describe that slope as being steep and high. Collingwood Bruce tells us that the slope west of the fort was covered with buildings in the Roman period (Collingwood Bruce 1947, 101), so it is probable that the slope is not all that steep. And it is difficult to tell from topographical maps whether that slope can sensibly be described as steep and high. There is thus a question mark as to whether the name Brocoliti is really appropriate for Carrawburgh. But the name is entirely appropriate for the fortlet at Newbrough on the Stanegate - there is a steep, high hill immediately west of the fortlet. And Esica is clearly a member of the isca-family of names, where isca is an inversion-type element meaning 'hill steep', and usually at a place with a name of the isca-family the slope is very steep, sometimes nearly vertical, for example at Isca at Exeter and Evidensca at Inveresk. Now there is a slope to the south of Great Chesters, though it is not particularly steep. There is also a slope to the north of the fort, though the writer has never seen it described as steep and one cannot really say on the basis of topographical maps whether it is steep. There is thus a question mark as to whether the name Esica is really appropriate for Great Chesters. But it is most certainly appropriate for the fortlet at Haltwhistle Burn. At Collingwood Bruce 1947, 145, we read of the fortlet at Haltwhistle Burn: "the fort is perfectly visible, being surrounded by deep ditches and standing sixty-five feet above the stream on an almost precipitous bluff". Now that is an Esica! As for Ravenna's Velurcion, corresponding to the ND's Borcovicio at Housesteads, if that Trajanic fort had been on exactly the same site as the Hadrianic Wall fort one might have expected to see the letter d meaning 'summit' in the name. But there is no d in Velurcion. The name clearly refers to a location on the side of a steep hill. Of course the Trajanic fort may have been built a little below the summit at Housesteads or near the foot of the hill - the slope between those two locations is surely too steep for building a fort. Or the fort might have been built on the slope going down to the Krag Burn. But turning to the Stanegate, at a point roughly half-way between Newbrough and Chesterholm there is indeed a location which is eminently suitable for the name Velurcion. This is the point where the Stanegate crosses over the southern flanks of Grindon Hill. A fortlet built almost anywhere on that section of the Stanegate would be appropriate for the name Velurcion. We are thus in the situation where the names Brocoliti and Esica are doubtfully appropriate for Carrawburgh and Great Chesters respectively, but are clearly eminently suitable for Newbrough and Haltwhistle Burn. In addition the name Velurcion could be appropriate for a fort in the vicinity of the Wall fort at Housesteads, but equally appropriate for a fortlet on the Stanegate at Grindon Hill. Under these circumstances, on the basis of the meanings of the names and the topography, it seems only sensible to conclude that Ravenna does in fact identify the Stanegate as being the Trajanic frontier, at least between Newbrough and Throp. The forts at Chesterholm and Carvoran would also be on the frontier but are not listed at this point in Ravenna since they had been listed earlier within a group of Flavian forts.
3.3 But there is a little puzzle concerning Castlesteads, Old Church Brampton and Boothby. Trajan will surely not have needed troops based in two forts spaced only two kilometres apart - Castlesteads and Old Church Brampton - so these forts are not likely to have been occupied simultaneously. There is a simple and obvious solution to this puzzle, and such solutions are generally to be preferred to more complicated ones. The original Uxelludamo will have been the fort at Old Church Brampton, so that the Trajanic frontier went from Kirkbride → Burgh-by-Sands III → Carlisle → Old Church Brampton → Birdoswald → Throp → Carvoran → Haltwhistle Burn and so on to the east, but with no fort at Newcastle. Note that the Old Church Brampton fort stood on the top of high ground with a steep drop down to the river Irthing, so that the name Uxelludamo is entirely appropriate (Uxelludamo = Ucselludamo, where ucs means 'steep hill' and dam means 'summit of hill'). But this fort was actually a little to the south of the Stanegate and this must have seemed anomalous, so the fort was relocated within the Trajanic period, while retaining its name, to a promontory on the other side of the Irthing, to the place we now know as Castlesteads. The fortlet at Boothby will have been built at this time to act as supply depot for Castlesteads and the fort at Old Church Brampton will have been dismantled, either at that time or a little later, once the Romans had decided that it was of no further use to them. This would not be the only case where a fort was relocated while retaining its name. At Burgh-by-Sands the original fort will have been Burgh-by-Sands III built on a slope a little to the SW of the later Wall fort, so the name Avalava of Ravenna is topographically correct (the val part of the name means 'side of hill, slope'). In the original plan for Hadrian's Wall this fort was replaced by a new fort, Burgh-by-Sands I, built on high ground to the south of the Wall - the name was simply transferred to the new fort, though with the initial v changed to b, in itself a common change in Romano-British place-names. Then, with the change of plan for the Hadrianic frontier this name was simply transferred, as Aballava, to the new fort built astride the Wall at Burgh-by-Sands village, this new fort being known today as Burgh-by-Sands II. And at Carlisle it is most likely that the name of the fort - Lagubalium - was simply transferred in the Hadrianic period to the new fort built on the other side of the river Eden at the place now called Stanwix. The Stanwix fort was the fort at Carlisle, at Lagubalium. The fort had simply been moved across to the other side of the river to ensure that the troops were on the military side of the Vallum.This new fort is the one called Petrianis in the ND, because the ala Petrianae had been based there for many years. In the same way York is referred to as Sextae in the ND, because the 6th legion had been based there for very many years. And of course once Trajanic Castlesteads had been built the pre-Hadrianic tower at Pike Hill to the west of Birdoswald would serve for communication between Trajanic Birdoswald and Castlesteads. The most likely time for the minor change from Uxelludamo to Uxelodunum will have been the date of relocation of the fort to Castlesteads. The dunum ending was more common than damo and the Romans will have been more familiar with it.
3.4 It is possible that Maia and Fanococidi are in fact two quite separate names. In this case Fanococidi will have been the westernmost fort on the frontier, that at Kirkbride, and Maia will have been somewhere between Kirkbride and Burgh-by-Sands III. Presumably there was a shrine to the Celtic god Cocidius at Kirkbride and the Romans simply called their fort Fanum Cocidi, appearing in Ravenna as Fanococidi. The most likely location for Maia in this case will have been up on Fingland ridge. The fort is presumably that which is reported to have existed at NY273575 (see, for example, the Pastscape website of Historic England). This opens up various possibilities as to the original form of the Celtic name. It may have been Maga or Magia, where mag is an inversion-type element meaning 'hill steep'. Alternatively, it may have been Mada or Madia, where mad is an old-style element meaning 'hill summit'. It is also possible that the name was simply Mala or Malia, where the m and l are the hill-letters m and l. Note, too, the present writer's suggestion in Chapter 8 that the names on the Rudge cup and the Amiens patera are those of new forts built to form part of the Trajanic frontier. If this is correct then Maia on Fingland ridge was a Trajanic foundation, whereas Fanococidi (which is not listed on the cup or patera) must have been occupied at that time and was simply incorporated into the new frontier. In other words, Kirkbride was a Flavian foundation.
3.5 Leaving Birdoswald on one side it is clear that the Trajanic frontier forts lay a little to the south of the line later to be taken by Hadrian's Wall, all the way over from Kirkbride to Newbrough. It is thus likely that this was also true of Birdoswald as well. It is evident that there was considerably more land to the south of Hadrianic Birdoswald in Roman times than there is today. There is very little space between the south wall of the fort and the Vallum, especially at the corners of the fort, but Simpson and Richmond indicated (see Simpson and Richmond 1933) that there was in fact a primary Vallum crossing at that point, complete with arched gateway. Now, if the edge of the cliff had been as close to the Vallum in Hadrianic times as it is today it would have been pointless, senseless really, to build that crossing. It thus seems clear that a considerable area of land to the south of the Hadrianic fort has been lost to river erosion and that the Trajanic fort (and probably the Flavian fort before it), was on that land, i.e. to the south of the turf Wall at Birdoswald.
3.6 And now one can understand more easily the inscriptions on the Rudge cup and Amiens patera. As noted above the cup and patera appear to commemorate the building of new forts forming part of the Trajanic frontier. They do not therefore list Fanococidi at Kirkbride, but do list Trajanic Maia on Fingland ridge and Trajanic Avalava at Burgh-by-Sands III. They do not list Lagubalium at Carlisle since that was an operational Flavian fort which was simply incorporated into the Trajanic frontier. They then list the new Trajanic fort at Castlesteads and go on to list the fort at Birdoswald. This presumably implies that Flavian Gabaglanda (originally Cambaglanda) at Birdoswald was abandoned when Agricola marched his troops north into Scotland and had to be rebuilt to form part of the Trajanic frontier. Both cup and patera then list Trajanic Banna at Throp and the Amiens patera on its own then lists Esica - this was of course Ravenna's Esica at Haltwhistle Burn and not the ND's Aesica at Great Chesters. It thus seems quite clear that the Rudge cup and Amiens patera are older than has hitherto been thought - they pre-date Hadrian's Wall and are thus not souvenirs of it.
3.7 As noted above, the Trajanic forts stood a little to the south of the later Wall, all the way from Kirkbride over to Newbrough, so it may be that this was also true for the forts to the east of Newbrough. The fort after Brocoliti heading east was Celunno at Chesters on the North Tyne. This Trajanic fort was probably south of the later Wall and Wall fort, close to the point where the Stanegate crossed the North Tyne. The next fort to the east was Onno. This may have been the old-style name Convo originally, the name meaning 'steep hill slope'. And there are indeed steep slopes to the south of the Wall at Halton Chesters, around the villlage of Halton itself. Ravenna's Onno may have been in that area. Then comes Vindovala at Rudchester. This name may originally have been Bindobala, an old-style name meaning 'high hill (bal) called Bind', where Bind is itself an old-style element meaning 'high hill summit'. There may earlier have been a native settlement called Bind on the summit of the hill and this name came to be associated with the hill itself, hence the later name 'high hill called Bind'. But Vindovala may originally have been Vintovala (see the entry for Vindovala in the Alphabetical List). The Trajanic fort may well have been a little to the south of the Wall at that point - there is sufficient space for a fort before the ground falls away down to the Tyne. There is less space to the south of the next fort, Condecor at Benwell. Again this is an old-style name and means 'steep hill (cor) called Cond', where Cond is itself an old-style element meaning 'steep hill summit'. There may have been a native settlement called Cond on top of the hill and this name came to be associated with the hill itself, hence the later name 'steep hill called Cond'. But here there is very little space to the south of the Hadrianic fort before the ground falls away steeply to the Tyne, so the Trajanic fort must either have been on the same site as the Hadrianic fort or very close to it on its southern side.
3.8 In conclusion, the Trajanic frontier ran from Serduno at Wallsend via Condecor at Benwell, Vindovala, just south of Rudchester, and Onno, somewhere in the vicinity of Halton, to Celunno, south of Chesters on the North Tyne and perhaps at the point where the Stanegate crossed the river. From there the frontier ran over to Brocoliti at Newbrough and then along the Stanegate via Velurcion at Grindon Hill, Vindolande at Chesterholm, Esica at Haltwhistle Burn and Magnis at Carvoran to Banna at Throp. The frontier then ran over to Gabaglanda (originally Cambaglanda) at Birdoswald and then via Uxelodunum at Castlesteads (originally Uxelludamo at Old Church Brampton) to Lagubalium at Carlisle. West of Carlisle the frontier went via Avalava at Burgh-by-Sands III and Maia on Fingland ridge to Fanococidi at Kirkbride.
4 Hadrianic period
And now we can move on to the time of Hadrian. The intentions of the new emperor are not entirely clear but it appears that he built a fort - Pons Aeli - at Newcastle to protect his new bridge over the Tyne, and planned a continuous Wall from that point, and a little forward of the Trajanic frontier posts, westwards all the way to Bowness-on-Solway. The part of the Wall from Newcastle to Wallsend was presumably an afterthought since it was a narrow Wall in contrast to the broad Wall west of Newcastle. The significant change in the original plan was the decision to build forts on the Wall itself, for if Hadrian had decided to dig the Vallum so as to create a military zone behind the Wall then he could have dug the Vallum just south of the Trajanic forts, which were in most cases close to the Wall, and so it would not have been necessary to build forts on the Wall at all. Presumably there was a military problem to the north of the Wall and it was feared that the troops could not be deployed quickly enough through the milecastle gates. So new forts were built on the Wall itself, some of them actually protruding north of the Wall, and most of the Trajanic forts were dismantled. The forts at Chesterholm and Carvoran were presumably retained because they served other functions associated with the Stanegate and the Maiden Way. It may be that the perceived military threat was greatest in the west, so priority was given to building that section of the Wall as fast as possible - it was therefore built of turf. And at Castlesteads the Hadrianic fort will probably have been built on the site of the old Trajanic fort for a very simple reason - the visibility to the NE, N and NW from the Wall in that area is rather poor (higher ground close to the Wall), so for security the fort was built on the promontory and the Vallum was diverted to the south to embrace the fort. Note that there were probably no civilians living around Carrawburgh, Housesteads and Great Chesters at the time Hadrian's Wall was built, so there was no-one available to give the Romans the Celtic names of those places. The Romans thus simply transferred to each of their new forts at Carrawburgh, Housesteads and Great Chesters the name of the respective nearest fortlet on the Stanegate - and then they dismantled the fortlets. Leaving Bowness-on-Solway on one side, there was no problem regarding the naming of the other forts on Hadrian's Wall since there had been Trajanic forts at all of the locations concerned (other than Newcastle, which was given a Latin name), and of course the Romans were were well aware of the Celtic names. In the case of Bowness perhaps there had been a Celtic settlement there when the Trajanic frontier was formed, and the Romans were aware of the name of that settlement (Concanata).
[This page was last modified on 23 March 2021]