The Antonine Wall
1 A great deal has been written about the Antonine Wall already, about its date of construction, how long it remained in use, about the forts and fortlets along its length and the spacing between them, about the so-called expansions found at a few places along the Wall, and so on. The present chapter is concerned with the names of the forts and fortlets along the Wall. A great deal has been written on this subject, too, though in general the conclusions reached by earlier writers have not been very convincing.
2 The clue to the identification of some of those forts and fortlets lies in the list of ten names given by Ravenna and said to be the names of civitates arranged in a line at the point where Britain is narrowest, i.e. the Forth-Clyde isthmus. These ten names are Velunia (191), Volitanio (192), Pexa (193), Begesse (194), Colanica (195), Medionemeton (196), Subdobiadon (197), Litana (198), Cibra (199) and Credigone (200). These names are accepted by most scholars as having been the names of forts or fortlets on the Antonine Wall, as indeed they will have been. But they were not originally the names of Wall forts and fortlets, as will become clear below.
3 Velunia will be the same name as the Veluniate on an inscription found at Carriden (Keppie 1990, 118 and Robertson 1990, 35 and 36), so this group of names starts at Carriden. Medionemeton is half-way through the list of names so we are presumably looking for a place roughly halfway across the isthmus. The name has nothing to do with any Celtic word ‘nemeton’ taken to mean ‘sacred grove’. Indeed nemeton is not even an element of the name. The Celtic name will have been of the form Nedionemedon, comprising the elements Nedion and med, though the first element may possibly have been Sedion, or even Redion since the hill-letter r appears to have arrived in central Scotland before the hill-letter m, as shown by the old-style names Cerma and Cermium. The Ned (or Sed or Red) of the first element of Nedionemedon means ‘hill summit’ and ion was just the ending of the name when it existed in the Ned (or Sed or Red) form. The second element of the name, med (met is impossible as this is an inversion-type element and would not come after an old-style element in the Celtic name), is another old-style element meaning ‘hill summit’. So we are looking for an Iron-Age hillfort on the summit of a hill roughly halfway across the isthmus. It seems quite clear that Nedionemedon was the Iron-Age hillfort on Castle Hill in east Dunbartonshire and that the Romans simply transferred the name to the fort which they built barely two hundred metres away on the summit of Bar Hill. At some stage the initial N of the name changed to M and the d was replaced with t.
4 It follows from the above that there were four named forts between Carriden and Bar Hill. At this point one should take note of the distances as the crow flies (correct to the first decimal place) between selected forts/fortlets in that part of Scotland:
Carriden to Mumrills 11.2 km
Mumrills to Camelon 5.9 km
Camelon to Seabegs 5.6 km
Seabegs to Westerwood 5.6 km
Westerwood to Bar Hill 5.6 km
As noted, all of the above distances are distances as the crow flies. And of course 11.2 km is exactly twice 5.6 km, so there probably was a fort halfway between Carriden and Mumrills, somewhere a little to the east of Nether Kinneil. There would then be seven forts/fortlets arranged in a line from east to west and evenly spaced apart by 5.6 km, except for the spacing between Mumrills and Camelon, which is 5.9 km. This irregular spacing between Mumrills and Camelon probably indicates that the fort at Camelon already existed when the 5.6 km spacing was selected for the other forts. It is clear from the 5.6 km spacing of Camelon and Seabegs that Camelon was an integral part of the fort system discussed above, but we have then seven forts from Carriden to Bar Hill, including Camelon, and only six Ravenna names. The reason for this is quite simple. Although Camelon is clearly included in the fort system discussed above its name is not given here because it is listed later in Ravenna as Poreoclassis. This peculiarity, not listing a name in one group where it could have been included because it is listed later in Ravenna, is seen in several other places in Ravenna, for example in the case of Ypocessa (see Chapter 12 of the Home menu), in the case of Abisson (see Chapter 15) and also in the case of Tagea (see Chapter 16). We thus have six forts (leaving Camelon on one side) and six Ravenna names, so it seems reasonable to conclude that Ravenna does in fact indicate the names of those six forts. In other words Velunia was at Carriden, Volitanio a little to the east of Nether Kinneil, Pexa at Mumrills, Begesse at Seabegs, Colanica at Westerwood and Medionemeton at Bar Hill. Note that the Volit element of Volitanio means ‘slope of hill high’ and there is indeed a high slope just east of Nether Kinneil at the exact midpoint between Carriden and Mumrills. Pexa and Begesse will have been Celtic Becsa and Becsesse, which are old-style names in the hill-letter s. These names refer in all probablity to Iron-Age hillforts, and indeed it would appear that Becsa was the name of the Cockleroy hillfort and Becsesse that of the hillfort on nearby Bowden Hill (for further discussion see the entry for Begesse in the Alphabetical List). The Romans simply transferred the names of the hillforts to new forts which they built at Mumrills and Seabegs. The name Colanica presents some difficulty. The old-style Col element means ‘steep hill’, but the anica element is obscure, unless it is just the hill-letter n with ica as an ending (but see the entry for Colanica in the Alphabetical List). There is indeed a steep slope immediately north of the Wall fort at Westerwood, though it is not clear if that is the slope referred to in the Col element. The name is that of a Celtic settlement transferred to a Roman fort and that settlement may have been somewhere else in the vicinity.
5 East of Bar Hill, then, the forts represented by Ravenna’s names are evenly spaced 5.6 km apart, with the sole exception of the Mumrills to Camelon stage, which is 5.9 km long. But west of Bar Hill this 5.6 km spacing appears not to be observed, at least not on the Antonine Wall. If one notes the distance as the crow flies between successive forts/fortlets on the Wall west of Bar Hill nowhere is the spacing 5.6 km. And even if one adds together the spaces between two successive pairs of forts/fortlets, nowhere does the combined spacing equal 5.6 km. It may be, then, that different considerations applied west of Bar Hill. However, it is worth noting that there are four names after Medionemeton in this Ravenna group and worth noting, too, that if one starts at Bar Hill and makes four successive leaps of 5.6 km one can in fact arrive at Duntocher, identified on this website as Credigone, the last of the names in this Ravenna group. Thus, keeping close to the general line of the Wall, a first leap of 5.6 km would take us to around NS 656 741, a little to the east of the Wall fort in Kirkintilloch. A second leap would take us on to around NS 604 725, to the northeast of the Wilderness Plantation Wall fortlet. A third leap of 5.6 km would take us across the Kelvin valley to around NS 549 721, a little to the east of the Bearsden Wall fort. The fourth and final leap, as noted, takes us to the small Wall fort at Duntocher. Robertson tells us that two of the structures on top of the hill at Duntocher predate the Wall (Robertson 1990, 96). It may be, then, that those structures formed the Credigone of Ravenna, but it is equally possible that the original Roman Credigone was only the earlier of those structures, the one surrounded by a small ditch. And we know that there was an early Antonine fortlet on the summit of Bar Hill, on the same site as the later Wall fort (Robertson 1990, 73). It is thus here proposed that the Ravenna names Velunia to Credigone represent early Antonine fortlets arranged in a line across the Forth-Clyde isthmus and evenly spaced apart by 5.6 km, apart from the irregularity between Mumrills and Camelon.
5.1 If the above conclusion is correct Subdobiadon, the first name after Medionemeton in Ravenna, will have been an early Antonine fortlet a little to the east of the Kirkintilloch Wall fort. The Celtic form of this name was almost certainly somewhat like Subredobiladon, a compound in the hill-letters s, r and l, where bred and bilad both mean ‘high hill summit’. Alternatively the end of the name may comprise old-style bil meaning ‘high hill’ together with a duno-type ending referring to the summit of a hill. This will have been the name of an Iron-Age fort/settlement on the summit of a high hill. If such a name is too grand for the location at Kirkintilloch indicated above then Celtic Subredobiladon must have been an Iron-Age hillfort/settlement somewhere in that region, the name then having been transferred by the Romans to their new fortlet east of the Wall fort at Kirkintilloch. If one writes Subredobiladon as Sub[re]dobi[l]adon one can see that by omitting the letters in brackets one arrives at Ravenna’s Subdobiadon.
5.2 Litana will then have been an early Antonine fortlet somewhere around NS 604 725, to the northeast of the Wilderness Plantation Wall fortlet. The Lit of the name means ‘hill high’, presumably referring to the hill on which the fortlet stood.
5.3 Cibra, again if the above reasoning is correct, will have been an early Antonine fortlet a little to the east of the Bearsden Wall fort. The old-style element Cibr means ‘steep, high hill’. There are of course several steep high hills in Bearsden. The name will have been that of a Celtic settlement transferred to the Roman fortlet.
5.4 The cred element of Credigone means ‘steep hill summit’ and gon means ‘steep hill’. An old-style compound such as this almost certainly refers to a hillfort, so if there had been no hillfort on top of the hill at Duntocher then there must have been a hillfort on the summit of some other steep hill in that region, the name then having been transferred by the Romans to Duntocher. A possible derivation of the tocher element of Duntocher may be noted in passing. It may be that the d of Cred moved to the front of the element and that Decr/Docr later changed to tocher. There are other examples of this process in Ravenna. For example the t of the Macat element of Macatonion may have been moved to the front of the name and Tamac/Tomac may have become modern Dymock in Gloucestershire. Likewise it is at least possible that the s of the Becs element of Becsesse (Celtic form of Begesse) moved to the front of the element and Sebec became modern Seabegs.
6 In support of the proposition that Ravenna’s names Velunia to Credigone represent early Antonine fortlets it may be noted that Ravenna’s list does not include a name for Castlecary, yet enough 1st century material (in the form of glass, samian ware and bronze coins – see Breeze 1982, 46) has been found there to make one think that Castlecary must have been the site of one of Agricola’s forts. Ravenna’s names Velunia to Credigone cannot therefore represent the forts Agricola built across the Forth-Clyde isthmus, as the present writer at one time thought. And yet they cannot represent Antonine Wall forts/fortlets either, for why should Ravenna give the names of only ten of the Wall forts/fortlets, and why should it name the fortlet at Seabegs but not the forts at Castlecary, Croy Hill, Cadder and Balmuildy? It might be thought that Ravenna’s names represent a short-lived late first-century frontier when the forts north of the Forth-Clyde line were abandoned, but there is little evidence to support that view and Robertson, who excavated Duntocher, seemed clear that the earliest structure there, the small square structure surrounded by a ditch, was early Antonine. It might be thought that Ravenna’s ten names represent Antonine forts which survived into the late Antonine period, perhaps when the Wall itself had fallen out of use or was about to be abandoned. But again Ravenna’s list does not include a name for Castlecary and there is apparently evidence of a temple being constructed there in the 170s or 180s (Breeze 1982, 124), suggesting that the fort was still occupied at that time, so Ravenna’s names probably do not belong to that period. The writer thus feels forced to the view that the names Velunia to Credigone do indeed represent fortlets of the early Antonine period, before the Wall was built.
7 But if the nine fortlets Velunia to Cibra were all as small as the earliest structure at Duntocher, said by Robertson to be some 18 metres square, then the ten fortlets together would seem to provide a rather weak frontier. But it is probable that they were not themselves intended to constitute a frontier. It is much more likely that they were built as part of the preparatory work for the construction of the Antonine Wall. It is possible, for example, that they housed teams of surveyors, each team charged with surveying the land in the area for which it was responsible so as to identify a suitable line for the Wall, taking into account the local topography and the firmness of the ground, and also perhaps to assess the availablity of suitable building material in that area. It is possible that the nerve-centre of the operation was the fort at Camelon, where staff would record and co-ordinate the work of the different teams of surveyors. There was a fortlet at Carriden on the Forth but no fortlet on the Clyde, so presumably each team of surveyors was charged with the task of determining the best line for the Wall between its own fortlet and the next fortlet to the west, the team at Duntocher having to find the best line for the Wall from Duntocher itself to the Clyde. This would seem to imply that the ten fortlets were themselves placed on the line which it was originally intended the Wall should take. It would appear, then, that it was originally envisaged that the Wall would go via Camelon, but for some reason those in charge changed their minds and the Wall when built took a more direct line between Mumrills and Seabegs. In addition the location of the fortlet to the northeast of Wilderness Plantation suggests that the original plan was to take the Wall more or less due west from the Cadder area to cross the Kelvin somewhere to the east of Summerston. Presumably as a result of the survey work carried out it was decided that the ground near the river Kelvin in that area was too soft or perhaps too vulnerable to flooding, so it was decided instead to take the Wall southwest from the Cadder area to cross the river at Balmuildy. But after Balmuildy the Wall goes smartly back up north to Summerston to get back on the originally planned line. And when the survey work had been completed, the plans approved and work on the Wall itself started, some of the ten fortlets may have been integrated into the Wall, as at Duntocher. Others were probably dismantled.
8 But it is probable that the names of the ten fortlets discussed above were simply transferred to the respective nearest forts or fortlets on the Wall when these were built. Thus the fort at Carriden would retain the name Velunia (changed at some stage to Veluniate) of the earlier fortlet at that site, the Wall fortlet at Kinneil would take the name Volitanio and the fort at Mumrills would retain the name Pexa of the early fortlet which had been located there. The Wall fortlet at Seabegs would take the name Begesse, the Wall fort at Westerwood would be called Colanica, and the Wall fort at Bar Hill would retain the name Medionemeton of the early fortlet which had stood on the same site. The Wall fort at Kirkintilloch would take the name Subdobiadon. The name Litana of the early fortlet around NS 604 725 was probably transferred to the Wall fortlet at Wilderness Plantation. The Wall fort at Bearsden would take the name Cibra and the small Wall fort at Duntocher would retain the name Credigone of the immediately adjacent earlier fortlet.
9 We know the locations of the early Antonine fortlets at Duntocher and Bar Hill. As to locating the others it should be possible to locate the fortlet at Carriden, the one to the east of Nether Kinneil and the one at Mumrills. As to Seabegs Robertson seems quite clear that the Wall fortlet there was of one build with the Wall, so the earlier fortlet must either have been very close to the later Wall fortlet or actually on the same site, having been dismantled to make way for the Wall fortlet. It should be possible to locate the fortlet at Westerwood. It will either have been on the same site as the later Wall fort or very close to it. It is probably not now possible to locate the fortlet to the east of the Kirkintilloch Wall fort since it may lie under modern buildings or a road. But it should be possible to locate the fortlet to the northeast of the Wilderness Plantation Wall fortlet since there are no buildings or roads in that area. Finally it is probably not now possible to locate the early fortlet to the east of the Wall fort at Bearsden. Like the fortlet to the east of the Kirkintilloch Wall fort it probably lies under modern buildings or a road.
10 One assumes that the fortlets Velunia to Credigone would need to be able to communicate with one another and with Camelon. To that end there probably were intermediate signalling towers between successives ones of Velunia to Credigone. One such intermediate signalling tower may possibly have been located on the platform at the summit of Castle Hill, between Duntocher and Bearsden (Robertson 1990, 93).
11 One final point should be mentioned. There is known to have been a fortlet, now thought to be of Antonine date, on the same site as the later Wall fort at Croy, but this does not fit into the 5.6 km spacing of the ten fortlets discussed above. This does not require us to abandon the ideas set out above. It may simply indicate that special problems were encountered in that area and so the powers that be decided to install an extra team to complete the survey work faster. Robertson, on the other hand, thought that this fortlet may have housed a construction party engaged in building the Wall fortlet a little to the west of the Croy Wall fort (Robertson 1990, 70).
[This page was last modified on 16 May 2021]