[NB. The reader will probably find it easier to understand this page if he or she has already acquainted himself or herself with the contents of Chapters 1 and 2. Chapter 1 concerns the basic building-blocks used in Romano-British place-names and Chapter 2 explains how those building blocks were assembled by the Celts to form compound names. The numbers accompanying some names below are those provided by Richmond and Crawford in Richmond and Crawford 1949.]
Olcaclavis to Voran
Roman place-names in Scotland
1 Two of the Roman names in what is now Scotland were referred to towards the end of Chapter 15, but to make the present chapter complete in itself the information is repeated here. Ravenna’s Trimuntium (183), Ptolemy’s Trimontium, was of course the Roman fort at Newstead on the river Tweed, and Ravenna’s Eburocaslum (184) appears to have been the fort at Broomholm, on the river Esk a little to the south of Langholm. The ebur element of the name Eburocaslum may mean ‘high hill’ and caslum appears to be a latinised form of the Celtic compound ucselum, this being a compound in the hill-letters s and l, where the ucs element means ‘steep hill’. The fort at Broomholm was indeed built at the top of a high, steep hill. It should be noted that modern river-names of the Exe, Esk, Axe and Usk family are derived from Romano-British place-names or name-elements of the isca, iscalis, iscaelis, acsula family (this is discussed in detail in Chapter 3: isca-type names), so it is entirely appropriate that the name of the fort at Broomholm, on the river Esk, should include an element of the form ucselum (but see the entry for Eburocaslum in the Alphabetical List for an alternative explanation of the name).
2 We move north now to three forts on the southern side of the Firth of Forth. These forts, Olcaclavis (188), Eiudensca (189) and Rumabo (190), presumably formed part of Agricola's building programme in AD 80, assuming he arrived in what is now Scotlamd in AD 79. Olcaclavis may be identified as the Flavian fort at Elginhaugh (Keppie 1990, 109 and 179) – one can see the derivation of the modern name if one rewrites Olcaclavis as Olgaglanis, i.e. with the common change of c to g, and v to n. The olc and cla elements (respectively using the hill-letters l2 and l1) of Olcaclavis refer to the steep river embankment at the top of which the fort was built. The compiler of Ravenna evidently saw all three names on one and the same map, so Eiudensca must have been a Flavian fort at Inveresk, the sca element of the Roman name corresponding to the river-name Esk. However, Rivet and Smith’s reading of the name, Evidensca, is to be preferred to Richmond and Crawford’s Eiudensca since it is much easier to understand. The correct spelling may have been (L)evidensca, the initial L having been lost at some stage of copying. The v, however, was almost certainly a c originally, the element lec then having exactly the same meaning as the final element sca, namely ‘hill steep’. And the den element refers to a location on top of a hill. The complete name thus refers to a location at the top of a steep hill, a definition which is entirely appropriate for Inveresk. The initial Lec element is of course the same element as the Olc at the beginning of the previous Ravenna name, Olcaclavis, upstream at Elginhaugh. One sees the replacement of c in a Celtic name by v in the latinised form of that name also in Lavobrinta (80), Leviodanum (220) and Levioxava (222). Rumabo was then a Flavian fort on the river Almond, either at Cramond, though no evidence of Flavian occupation has yet been found there, or at the point upstream where Dere Street crossed the river, assuming the road followed an inland route. Rumabo is actually a river-name and the Romans merely adopted the name and applied it to their Flavian fort. The hill-letters corresponding to the r and m of Rumabo are m and n, so it would seem clear that the mond of Cramond is a land-name referring to the location of the Cramond fort at the top of a hill. And in fact the Antonine fort at Cramond was located on the flat ground where Cramond kirk now stands, at the top of the slope leading down to the river. Evidently the land-name of the Antonine fort was of the form Lamond or Almond and this name was simply transferred by the Romans to the river, previously called the Rumabo. Note, however, that it is much more likely that Mond (presumably with some ending), later Lamond or Almond, was actually the name of the hillfort on Craigie Hill, on the north side of the river Almond just north of Edinburgh airport. The Flavian fort will have taken its name, Rumabo, from the river, whereas when the Romans built their Antonine fort at Cramond they transferred the name and presumably the inhabitants of the hillfort to Cramond. The Flavian fort was most probably at Cramond, though not necessarily on the same site as the later, Antonine fort.
3 Ravenna now lists ten forts built across the isthmus between the Forth and Clyde, namely Velunia (191), Volitanio (192), Pexa (193), Begesse (194), Colanica (195), Medionemeton (196), Subdobiadon (197), Litana (198), Cibra (199) and Credigone (200). These names are discussed in Chapter 22 (The Antonine Wall) of the Home menu. In the interests of brevity the reader is simply referred to that text. It is merely noted here that Velunia will have been the name of an early Antonine fortlet at Carriden (and later of the Wall fort at that location), Volitanio the name of an early Antonine fortlet just east of Nether Kinneil (and later of the Wall fortlet at Kinneil), Pexa the name of an early Antonine fortlet at Mumrills (and later of the Wall fort at the same site), Begesse the name of an early Antonine fortlet at Seabegs (and later of the Wall fortlet at the same location), Colanica the name of an early Antonine fortlet at Westerwood (and later of the Wall fort at the same site), Medionemeton the name of the early Antonine fortlet at Bar hill (and later of the Wall fort on the same site), Subdobiadon the name of an early Antonine fortlet just east of the Kirkintilloch Wall fort (and later of the Wall fort itself), Litana the name of an early Antonine fortlet around NS 604 725 (and later of the Wall fortlet at Wilderness Plantation), Cibra the name of an early Antonine fortlet just east of the Wall fort at Bearsden (and later of the Wall fort itself) and Credigone the name of the early Antonine fortlet at Duntocher (and later of the immediately adjacent Wall fort).
4 Before discussing Ptolemy’s Pinnata Castra, Tuesis, Devana and Orrea it is helpful to identify the Roman forts along the Highland fault line. These are all listed in Ravenna. The list starts with Lano (201) on the river Leny at Bochastle, Maulion (202) at Malling and Demerosesa (203) at Drumquhassle. After Demerosesa comes Cindocellum (204). It is possible that there should be only one l in the name. Alternatively, since doubled consonants are rare in the earliest forms of Romano-British place-names, the Celtic name may have been Cindoceldum, where both elements cind and celd mean 'steep hill summit'. Cindoceldum will have been an Iron Age hillfort/settlement on the summit of a steep hill, either Dumbarton Rock or Sheep Hill, a little to the east. The Roman fort will then have taken its name from that earlier hillfort/settlement, but is itself most likely to have been built on low ground on the north bank of the river Clyde. A Roman fort adjacent Dumbarton Rock may have guarded a harbour, whereas a fort in the Dumbuck area will have controlled what is believed to have been the lowest fording point on the Clyde. After Cindocellum Ravenna continues with Cerma (205) at Dalginross, Veromo (206) at Fendoch, Matovion (207) at Cargill, Ugrulentum (208) at Cardean, Ravatonium (209) at Stracathro and Iberran (210), somewhere close to the Bervie Water. The name Cerma of the Dalginross fort has presumably survived in the name of Comrie, on the other side of the river. And Veromo was in fact the Celtic name of the Perthshire river now called the Almond – the Romans merely transferred the river-name to their fort at Fendoch. The above identifications for Matovion and Ugrulentum are confirmed by the river-letters corresponding to the hill-letters r, l and n of Ugrulentum – these are s, t and m. The m and t are present in Matovion, which is thus a river-name, and the s is present in the modern river-name Isla, as is the river-letter l applied to minor rivers by those who used the hill-letter n. It seems quite clear that the name Matovion was applied to the stretch of the modern Isla between Cardean and the confluence with the Tay and that a name somewhat of the form Isla was applied to the stretch of the river upstream from Cardean. The Romans then simply took the river-name Matovion and applied it to the fort which they built on the banks of the river at Cargill. Moving on, it seems probable that Ravatonium was originally Rascatonion, where ascat is an inversion-type element meaning 'hill steep high'. The element will refer to the steep, high river embankment at the top of which the Stracathro fort was built (assuming the Celtic settlement was on the same site as the later Roman fort). The asc of ascat has survived in the modern river-name North Esk. At some point the s was dropped from Rascatonion and Rac changed to Rav. Finally, Iberran is just the river-name Iberban (now the Bervie Water) with r b → r r. The name Iberran refers to a Flavian fort built fairly close to the Bervie Water, most probably at a location where it could control movement across the Mounth.
5 Now it is time to deal with the Ptolemy names Pinnata Castra, Tuesis, Devana and possibly Orrea, if this last name is taken to refer to the same place as the Poreoclassis of Ravenna, as is commonly assumed by scholars. The Ravenna names corresponding to Pinnata Castra, Tuesis, Devana and Orrea are then Pinnatis (211), Tuessis (212), Devoni (215) and Poreoclassis (221). In Ravenna these names are of course interspersed with other names, and it is these other names which help locate the names in Ptolemy. The list in Ravenna is Pinnatis (211), Tuessis (212), Lodone (213), Litinomago (214), Devoni (215), Memanturum (216), Decha (217), Bograndium (218), Ugueste (219), Leviodanum (220) and Poreoclassis (221). These names have no connection with the area up near the Moray Firth, as is commonly supposed, quite regardless of Ptolemy’s co-ordinates and his assigning particular names to particular tribes. First of all, to keep the explanation as short as possible, one may take Tuessis to be the name of the fort commonly known as Bertha, at the confluence of the rivers Almond and Tay. The initial Tu will be based on the Celtic name of the river Tay. It is probable that the u of Tuessis is really a v and that the name should actually be Tvessis or Tavessis. The Tv/Tav element is then the Celtic proper name of the river. In addition, endings of the essa-type appear to relate to locations at the top of a steep slope and overlooking a river, without necessarily implying that the raised ground is very high - examples include Abisson overlooking the river Wear at Durham, Camulosessa overlooking the river Derwent at Malton, Demerosesa overlooking the Endrick Water at Drumquhassle, Duabsissis overlooking the river Tweed at Berwick and Ypocessa overlooking the river Yscir at Y Gaer, near Brecon. The Bertha fort stands on a promontory overlooking the confluence of the rivers Almond and Tay, so the name Tuessis appears entirely appropriate. Nonetheless it is much more likely that Celtic Tavessis was an Iron-Age hillfort close to the Tay and that the Romans simply transferred the name to the fort which we now call Bertha. There was a hillfort a little further up the Tay at Broxy Kennels (NGR: NO 091 279) and another downstream at Dow hill (NGR: NO 149 215), the essa-type ending being particularly appropriate for the latter hillfort. As explained in Chapter 4, in names with an essa-type ending the part of the name before the ending may be a land-name with the hill-letter s or a river-name with the corresponding river-letter b. In the case of Tvessis the v is a modified river-letter b. It seems quite clear that the Romans transferred the fort-name to the river, i.e. the Romans called the river Tay the river Tvessis, and so of course the Firth of Tay was the Tvessis estuary. It is thus quite clear that there is an error in the Geography of Ptolemy regarding the location of the Tuesis estuary.
5.1 The Ravenna names after Tuessis, down to Poreoclassis, can then be seen to lie on two routes to the south out of Bertha. The name after Tuessis is Lodone and this name may have survived as Dunning, in Strathearn, just as Lindinis (26) appears to have survived in Dundon (in Somerset) and Stodo(n)ion (158) in Denton (Nether Denton to the east of Carlisle). The name Lodone points to a location on the summit of a hill, or at least on the top of raised ground, so one should look for signs of the fort on the raised ground to the south of the village of Dunning. A suitable location would be on the minor summit just west of the place called Findony on the OS map or, alternatively, on one of the minor summits on the other side of the Dunning Burn. The next most likely location would be a little further south, perhaps in the area called Common of Dunning, though still on the top of raised ground (but see the entry for Lodone in the Alphabetical List for an alternative explanation of this name). The next two names, Litinomago and Devoni, may form a single name of the Newcastle-upon-Tyne or Stoke-on-Trent type. Devoni is the name of the river Devon and Litinomago refers to a place adjacent a steep, high hill, so the complete name would suit a fort on the banks of the river Devon in the Yetts o’ Muckhart area. Rivet and Smith refer to a variant form Lintinomago and this is likely to be closer to the original Celtic name. The latter will then have been Lindinonaco and it so happens that one of the high, steep hills immediately north of Yetts o’ Muckhart is called Lendrick Hill. The next Ravenna name, Memanturum, presents some difficulty. It belongs to a group of names which include the hill-letter m twice and where it appears that one m should be a different letter, most probably l in this case. And the first u in the name was probably a v in the original Celtic name. If one moves the vr to the beginning of the name, and adds a Gaelic dun at the front, one obtains Dunvrmelantum, and it so happens that Dunfermline is recorded as Dunfermelitane in 1128. There seems little doubt, then, that Memanturum was at Dunfermline. The name Melantvrum seems reasonably clear – it appears to be a compound in the hill-letters m, l, n and r, and assuming the t really is a t (and not a modified d) the ant element is an inversion-type element meaning ‘hill high’. The l and m must then also be used in an inversion-type manner, m being the hill-letter most recently added to the compound. The town of Dunfermline does indeed stand on a hill. What is not clear is whether the vr element really is vr, meaning ‘side of a hill’ or was originally the old-style br meaning ‘high hill’.
5.2 So, Ravenna has come down from Bertha via Dunning and Yett’s o’Muckhart to Dunfermline. It would appear that the next place, Decha, was at Inverkeithing (this being presumably another case of element reversal, the keith of Inverkeithing being the Roman name Decha the other way round, the ch having changed to k and the d to th). Presumably the Romans built a harbour at Inverkeithing to provide communication with some location on the other side of the Forth estuary. As to the name itself Decha appears to be a topographical name which has lost its hill-letter. The Celtic name may have been of the form Delca, meaning ‘summit of hill steep’, the development being Delca → Decca → Decha. Now there is a steep hill at Inverkeithing, immediately NW of the Inner Bay, so the best place to look for traces of the fort would be up on top of that hill, though there are other steep slopes in the vicinity. The advantage of such a location is that it would enable the Romans to control the land approaches to the harbour and perhaps also to signal across to Cramond on the other side of the estuary. Having come down from Bertha to Inverkeithing via Dunfermline, Ravenna clearly could not go further in this direction, so it is clear that the next name, Bograndium, lies on another route out of Bertha. Bograndium appears to have been the fort at Ardoch – presumably the name of the nearby village Braco is derived from the Bogra of Bograndium. The part Bogrand of the name is essentially the same compound as the bagland of Gambaglanda at Birdoswald, but using the hill-letter r rather than l. The next name, Ugueste, has the meaning ‘adjacent a steep, high hill’ and no location could be better than Stirling, to the south of Ardoch. The original spelling was probably Muguleste or Mugulesde – compare the structure of the name Ugrulentum (originally Rugulentum). This point is discussed in more detail in Chapters 17 and 19. Ugueste at Stirling is used as a minor node, for the next-following name is Leviodanum at Doune, to the northwest of Stirling, and the next again is Poreoclassis at Camelon, to the southeast of Stirling. The late pre-Roman name of Doune was most probably Lecilodanum, the lec element meaning ‘hill steep’ and dan meaning ‘summit of hill’. The change of the c of Lecilodanum to the v of Leviodanum is seen also in Racatonium → Ravatonium and Leciocsava → Levioxava. Note that the river-letter corresponding to the hill-letter n of Lecilodanum is m (for major rivers). The second l of the place-name is l1, corresponding to the river-letter t, so the river-name corresponding to the lodanum part of Lecilodanum would have a form such as Tem. The initial L of Lecilodanum is l2 and so, as usual, the corresponding river-letter t is placed at the end of the existing river-name, the river-name then being of the form Temit. This form, with the omission of intervocalic m and the change of t to th yields the modern river-name Teith. But Camelon is an odd place for a storehouse of the fleet, the generally accepted meaning of Poreoclassis. However, it is believed that the coastline was much nearer to Camelon in the Roman era than it is now, so it may well have been possible for sea-going ships to enter the lower reaches of the river Carron. The fort at Camelon could then have protected harbour installations receiving supplies by sea and then delivering supplies, as and when necessary, to the naval fleet or to armies on the march. So now, bearing in mind that all the forts along the edge of the Highlands have already been listed, if one looks at the layout of these two routes, Bertha down to Inverkeithing and Bertha down to Camelon, on a map – a modern map, not a Ptolemy map! – and asks oneself where was Pinnatis, the name immediately before Tuessis at Bertha in the Ravenna list, the answer appears very clear - Pinnatis can only have been the legionary fortress at Inchtuthil. It is thus clear that Ptolemy’s Pinnata Castra was really at Inchtuthil, his Tuesis at Bertha and his Devana in the Yetts o’ Muckhart area.
6 It seems quite clear that Ravenna has started at Inchtuthil, come south to Bertha and then given names on two different routes south out of Bertha, the first route going by way of Dunning, Yetts o’ Muckhart and Dunfermline to Inverkeithing, and the second going to Camelon via Ardoch and Stirling. One can see that there is a name missing here – there is no name for the fort at Strageath, a fort some 3.5 acres (1.4 ha) in extent and believed to have had two occupation phases in the Flavian period. It would therefore be reasonable to expect the name of that fort to be included in the Ravenna list – but it is not there. And this is precisely the clue needed to locate the final six names in Ravenna, namely Levioxava (222), Cermium (223), Victorie (224), Marcotaxon (225), Tagea (226) and Voran (227), for where in Ravenna the forts along a particular route are being listed and a name is missing which it would be reasonable to expect would be included, this is usually because that name has already been listed (as one of several names on another route passing through the same place, perhaps even taken from a different map) or is mentioned later. Compare, for example, the Ravenna names coming up from Decuaria at Brough on Humber via Devovicia at Stamford Bridge, Dixio at Thirsk and Lugunduno at Dinsdale Park on the Tees to Coganges at Chester-le-Street. This route passes through Durham but Durham is not mentioned at this point. Instead it is listed later as Abisson on a route from Camulosessa Presidium at Malton to Celovion at Chesters. Now, the name of the Strageath fort is certainly not mentioned earlier than the names from Bertha down to Camelon, so it must be included later. In other words the names from Levioxava to Voran define another route passing through Strageath, and the only sensible route – indeed apparently the only possible route - is one going along the valley of the Earn. The name of the Strageath fort is therefore included in the group Levioxava to Voran, by far the most likely candidate being Tagea, the final geath element of the modern name presumably being derived from Tagea (most likely just another case of element reversal). There are four names before Tagea and one after, so the most sensible arrangement is to place Voran generally to the west of Strageath and the other four names on a route stretching away to the east along the valley of the Earn. If one then assumes that the spacing of the forts to the east of Strageath was the same as that between Strageath and Ardoch, and assumes further that the name Victorie seen on a map merely indicated the site of the battle of Mons Graupius, then this would place Marcotaxon a little to the east of the Ruthven Water, Cermium a little to the east of the Water of May, and Levioxava at the mouth of the river Earn. However, bearing in mind that the forts would need a good supply of running water, it seems sensible to pull all three names back a little to the west and place Marcotaxon actually on the Ruthven Water, Cermium on the Water of May and Levioxava at Abernethy. The names Cermium and Levioxava refer to locations adjacent a steep hill, and in the case of Marcotaxon the hill is considered high as well as steep, so the meanings of all three names fit the locations. In addition, the identifications of Marcotaxon and Levioxava are confirmed by the local river-names. The hill-letters in Marcotaxon are m, r and s (x = cs), the corresponding river-letters being r, s and b. It would thus appear that the v of Ruthven is a modified river-letter b and that the s of the Celtic river-name changed to th, presumably at some date later than the Roman occupation. The Celtic form of Levioxava appears to have been Lecimocsava, where the l is l2, the river-name corresponding to the place-name thus having a form such as Arbet or Arbete, which with the change of b to v and then v to n would become Arnet or Arnete. Presumably later settlers coined the name Aberarnete, referring to the location of their settlement at the mouth of the river Arnete. Aberarnete may then have been shortened to Abernete, this later becoming the modern Abernethy. But for some reason it was the initial part Arne of the river-name which survived as the modern river-name Earn. At some stage the intervocalic m of Lecimocsava was dropped and the c changed to v, this yielding Leviocsava, appearing as Levioxava in Ravenna. The c → v change is seen also in Lavobrinta, where the original Lac element referred to the steep slopes on three sides of the hill on which the fort at Caer Gai stands. Note, finally, that if the above reasoning is correct then the nethy of Abernethy has no connection with any Pictish king called Nectan or Nechtan, as is commonly supposed.
6.1 Going on from the above, it is quite clear from the order of the names in Ravenna that Victorie appeared on a map at a point somewhere between the Water of May and the Ruthven Water. Thus, provided no other hill in that relatively small area suits the name Mons Graupius and better fits Tacitus’ description of the battlefield, it is clear that the Ravenna Cosmography does in fact provide direct confirmation of the view of Ian Smith that Craig Rossie is the Mons Graupius of Tacitus (Smith 1987, 38 and 39.) Smith indicates that in 1970 Richard Feachem, following a hint made as long ago as 1926 by W.J. Watson, identified Duncrub – on the low ground at the foot of Craig Rossie at a point a little to the west of Dunning – as Mons Graupius (Smith 1987, 2, 38, 39, referring to Feachem 1970 and Watson 1926, 56). However, common sense would seem to dictate that the term mons can hardly be applied directly to a location on the low ground. At any rate, whilst Smith and Feachem may differ as to the precise identification of Mons Graupius, with the former identifying Craig Rossie and the latter identifying Duncrub, they must surely be in agreement that the battle, following the description of Tacitus, was fought on the lower slopes of Craig Rossie and on the low ground at the foot of the hill. This view, that the battle of Mons Graupius was fought at the foot of Craig Rossie, has not been taken seriously by all other scholars over the years, but it is clear from the Ravenna Cosmography that Smith and Feachem were in fact right.
6.2 The one point giving cause for concern is that if Strageath was Tagea then Voran (227) must have been generally to the west of Strageath, but Ravenna has already listed Cerma (205), which appears to have been the fort at Dalginross. It appears, then, that Voran is simply an alternative form of Veromo at Fendoch. Note that the Fendoch fort is quite different in design from the forts at Dalginross, Malling and Drumquhassle (Maxwell 1989, 99), and so was probably built in a later season.
7 Almost as a footnote the writer points out that he is still not entirely happy with the place-names in central Scotland. The problem concerns the name (L)evidensca (189)(using Rivet and Smith’s spelling). This would appear to be the only name in the British section of the Ravenna Cosmography in which the element den, meaning ‘summit of a hill’, appears in the middle of the name instead of at the end or, as in a few cases, at the beginning. There would appear to be no such name in the British section of the Antonine Itinerary either. This may suggest that the Celtic name of Inveresk was in fact only Levidanum (Lecidanum originally) and that the Romans themselves added isca. This might indicate that the Legio II Augusta, which had earlier been stationed at Isca at Exeter and which had later simply applied the same place-name to their new fortress at Caerleon, built and/or occupied the fort at Inveresk. It seems best to assume that the Romans added isca to Levidanum at Inveresk to distinguish this fort more clearly from the Leviodanum at Doune, and that Levidanisca somehow turned into the (L)evidensca of the Ravenna Cosmography. Of course, if there was a perceived need to distinguish between Levidanum at Inveresk and Leviodanum at Doune this necessarily implies that Levidanum at Inveresk was indeed Flavian, since the fort at Doune is known to have been occupied only during the Flavian period (Maxwell 1984).
8 The above discussion covers all of the Ravenna names in the territory which is now Scotland. Ravenna’s river-names, together with the river-names of Ptolemy, are discussed in Chapter 19: The rivers of Roman Britain. Ptolemy’s Vindogara appears to have been a Roman fort/harbour at Irvine, North Ayrshire, and his Rerigonium a fort/harbour at Girvan in South Ayrshire, at the mouth of the Rerigonion river, i.e. the Water of Girvan. No traces of a fort or of harbour installations have been found at Irvine or Girvan, so far as the writer is aware. Ptolemy's Tarvedum and Virvedrum appear to be the names of promontory forts which have been transferred to the promontories on which the forts stood. Thus Tarvedum, probably originally Carvedunum, appears to have been the promontory fort on Holborn Head, just north of Scrabster in Caithness. Virvedrum, possibly originally Birvedunum, but more probably Birvedurum, appears to have been the promontory fort on St. John's Point, a little to the west of Duncansby Head in Caithness.
9 One last name should perhaps be mentioned here, since it may be a place-name. This is Loxa, given as a river-name in Ptolemy, the name of the river now called the Lossie. There would appear to be three possibilities. The first possibility is that there was an important native settlement called Locsa (Celtic form of Loxa) on the banks of the Lossie and the Romans were well aware of its existence. They may have spoken of the river at Locsa and in due course the name came to be applied to the river itself. This would be unusual. Indeed no example of it comes to mind. The second possibilty is that Loxa was indeed the name of the river. It would be a name of the kind comprising a river-element, here the river-letter l, used as a prefix to a place-name employing one or more hill-letters. The place-name in this case would be the old-style element ocsa meaning ‘steep hill’. There are several river-names of this kind, for example Abravannus, Traxula and Velox. The third possibility again involves a native settlement and there was indeed an important settlement close to the Lossie. This was the Iron-Age settlement at Birnie, some 6 kilometres south of Elgin. This settlement was immediately north of a hill on the eastern bank of the Lossie. The sides of the hill are steep, especially on the western side, adjacent the river. The name Locsa would thus be appropriate, and indicates that the settlement had been in the hands of people who used the hill-letter s but was taken over by people who used the hill-letter l. The Romans might then have built a fort in the vicinity of that settlement and simply transferred the name of the settlement to their fort. They then transferred the name of their fort to the nearby river, this now being the Locsa, or Loxa as given by Ptolemy. Now there is in fact a rectangular enclosure about one kilometre south of Birnie, at Thomshill (NGR: NJ 211 574), though many scholars appear reluctant to accept that it might have been a Roman fort. However, the enclosure is rectangular and is surrounded by a V-shaped ditch with a drainage channel at the bottom, and the ditch has rounded corners (see, for example, the entry for Thomshill on the Canmore website of Historic Environment Scotland). The enclosure is much smaller than the known Roman marching camps in the north of Scotland but with an internal area of about 4.3 acres it has a size fairly common for Roman forts. And there are a number of examples of the Romans transferring the name of a Celtic settlement to a new fort and then transferring the name of their fort to a river, for example Lemana and Tuessis. This third possibility seems to the present writer the one most likely to be closest to the truth, to the facts. This is because the river-letters t, changed to d, and b, changed to v, appear to be present in the river-names Deveron and Divie (a tributary of the river Findhorn) and this, coupled with the presence of the hill-letter s in Locsa, appears to indicate that much of the region south of the Moray Firth was at one time occupied by a people who used the hill-letter s and river-letter b, but was taken over by a people who used the hill-letter l (actually l1) and river-letter t. This makes it much more likely that Locsa is a land-name in the hill-letters s and l than that it is a river-name of the kind in which the river-letter l is used as a prefix to the land-name osca. But note that Loxa will have been the Roman name of the river. The Celtic name most probably included the river-letters t and b corresponding to the hill-letters l and s in Locsa. And since the name was transferred from the fort to the river, then the very fact that it is the Roman river-name Loxa which is included in the Geography of Ptolemy, believed to date from around AD140, necessarily implies that the Roman fort at Thomshill dates from an earlier period, indeed from the Flavian period.
10 The map below shows the locations of the Roman forts discussed in this chapter. The forms and numbering of the place-names are those given by Richmond and Crawford. The identifications are as follows:
188 Olcaclavis = Elginhaugh
189 Eiudensca = Inveresk
190 Rumabo = Cramond (or at the point where Dere Street crossed the river Almond)
191 Velunia = Carriden
192 Volitanio = Kinneil
193 Pexa = Mumrills
194 Begesse = Seabegs
195 Colanica = Westerwood
196 Medionemeton = Bar Hill
197 Subdobiadon = Kirkintilloch
198 Litana = northeast of Wilderness Plantation fortlet
199 Cibra = Bearsden
200 Credigone = Duntocher
201 Lano = Bochastle
202 Maulion = Malling
203 Demerosesa = Drumquhassle
204 Cindocellum = Dumbuck or Dumbarton
205 Cerma = Dalginross
206 Veromo = Fendoch
207 Matovion = Cargill
208 Ugrulentum = Cardean
209 Ravatonium = Stracathro
210 Iberran = location on or close to the Bervie Water in the Mondynes area
211 Pinnatis = Inchtuthil
212 Tuessis = Bertha
213 Lodone = Dunning
214, 215 Litinomago Devoni = Yetts o'Muckhart area
216 Memanturum = Dunfermline
217 Decha = Inverkeithing
218 Bograndium = Ardoch
219 Ugueste = Stirling
220 Leviodanum = Doune
221 Poreoclassis = Camelon
222 Levioxava = Abernethy
223 Cermium = location on the Water of May
225 Marcotaxon = location on Ruthven Water
226 Tagea = Strageath
227 Voran = Fendoch
The map also indicates with Victorie the site of the battle of Mons Graupius, between Cermium (223) and Marcotaxon (225). Locsa at Thomshill, south of Elgin, is not shown on the map. Nor are Ptolemy's Tarvedum at Holborn Head and his Virvedrum at St. John's Point, both in Caithness.
[This page was last modified on 08 March 2021]