Smith, Roman Place Names in Lancashire and Cumbria

Some Roman Place-Names in Lancashire and CumbriaAuthor(s): Ian G. SmithSource: Britannia, Vol. 28 (1997), pp. 372-383Published by: Society for the Promotion of Roman StudiesStable URL: http://www.jstor.org/stable/526777 .

Accessed: 11/08/2013 07:41

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Society for the Promotion of Roman Studies is collaborating with JSTOR to digitize, preserve and extendaccess to Britannia.

http://www.jstor.org

This content downloaded from 147.143.2.5 on Sun, 11 Aug 2013 07:41:53 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=sprs

http://www.jstor.org/stable/526777?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


372 NOTES

Some Roman Place-names in Lancashire and Cumbria. Ian G. Smith writes: Despite extensive debate over the past 400 years,140 the identities of several Roman place-names in north-west England have yet to be established unequivocally and the tensions inherent in the current best-fit solutions are fully illustrated in Rivet and Smith's standard work The Place-names of Roman Britain.141 In the face of apparently conflicting evidence, most modern analyses tend, in general, to favour the discipline imposed by the name sequences within sources such as the Antonine Itinerary, Notitia Dignitatum, and even the Ravenna Cosmography, with etymological considerations often playing a subordinate role. This essay, however, places greater emphasis on the intrinsic meaning of each place-name and seeks to reconcile the other evidence with this. The influence of the late Leo Rivet's work on the author's research is respectfully acknowledged.

THE ANTONINE ITINERARY-ROUTE X

Route (Iter) X of the Antonine Itinerary records the names of nine stopping-places along its course of 150 Roman miles142 in north-west England. The problematic northern section of this itinerary, with the generally accepted identifications of Rivet and Smith143 and as adopted by the Ordnance Survey, for example, commences:

Clanoventa-- 18

miles-- Galava-- 12

miles-- Alone-- 19

miles-- Galacunm - 27 miles - Bremetonnaci

Ravenglass Ambleside Watercrook Burrow-in-Lonsdale Ribchester

At Bremetonnaci Route X is securely fixed, by inscription, to the fort at Ribchester144 before continuing southwards to Manchester and Shropshire.

The Cumbrian setting for the route follows the case made by Haverfield in 1915145who could see no alternative course north of Ribchester which would not come into conflict with other established place- names in the region, towards Carlisle or Yorkshire, in directions previously favoured by antiquarians. His identifications follow a straightforward progression through known Roman sites, at appropriate distances, and the association of the northern terminus at Clanoventa with the fort at Ravenglass on the Cumbrian coast is well suited to the most likely derivation of the Roman place-name from *glanno-, 'bank, shore'.146 But, for all that, the chain hinges on the crucial identification of Alone with the fort at Watercrook in Kendal. Alone is almost certainly a river name, probably derived from an Indo-European *al- meaning 'water' as found in the ubiquitous variant Alauna.147 On the other hand, the fort at Watercrook, as its present name reflects, lies closely confined on three sides within a meander of the River Kent which derives from another pre-Roman river name, *Cunetiu.148 In order to accommodate this etymological contradiction, while still maintaining Haverfield's route, Rivet and Smith reluctantly proposed a rare case of a non-fluvial Alauna.149 This seems a somewhat desperate solution, however. Watercrook's intimate topographical relationship with the River Kent basically precludes any association with the place-name Alone and this identification should therefore be rejected.

Rivet (1970)150 had previously sought to circumvent this problem by siting Alone more appropriately on the River Lune at Low Borrow Bridge, the requisite 19 miles north of Burrow-in-Lonsdale

140 W. Camden, Britannia (1586). J. Horsley's comprehensive early study of the subject, Britannia Romana (1732). For a less considered debate see, for example, the acrimonious exchange between Messrs Ferguson and Nicholson in the third issue of the Transactions of the Cumberland and Westmorland Antiquarian and Archaeological Society, CW' iii (1877), 64-94, 167-74, 182-9.

141 A.L.F. Rivet and C. Smith, The Place-names of Roman Britain (1979), 141-2, 170-2, 209-1 I, 220-3, 232-3. 142 Throughout this paper, distances are given in Roman miles; I Roman mile = 1.48 km. 143 Rivet and Smith, op. cit. (note 141), 170-2. 144 I.A. Richmond, 'The Sarmatae, Bremetennacum Veteranorum and the Regio Bremetennacensis', JRS xxxv

(1945), 15-29; RIB 583. 145 F. Haverfield, 'The Romano-British names of Ravenglass and Borrans (Muncaster and Ambleside)', Arch.

Journ. lxxii (1915), 77-84. 146 Rivet and Smith, op. cit. (note 141), 367. 147 ibid., 243-4. W.F.H. Nicolaisen, Scottish Place-names (1976), 186-7. 148 E. Ekwall, English River-names (1928), 226-8. The meaning of the name remains obscure; see also Rivet and

Smith, op. cit. (note 141), 328. 149 ibid., 171. 150 A.L.F. Rivet, 'The British section of the Antonine Itinerary', Britannia i (1970), 54.



NOTES 373

Want Law Birdossinkl a LUX BANNA\ AES,

BownesD-oa-nplw Nedhr I IS VINDOLANDA

Kirk MARA W MPETRIAN1S 1

Beckftoot - -' hbley

P6kscarilas

"/ :::r:n: / I

0-

UXLO / VOHA

....w (.. /w

GABR EN- -./

1

Burunw'/ ( Wails Maide

GABENTM I~ ' \ ILVCascoI ENLANNOVENTA $ I ,I low

Hadma MawBonw Bkl

WaRnao

lop IA

,

ft:

m Drift$

...

M6EDIOBOGIRM

0AlNVAAVATAJ

| uva -

. ,

I

ROMAN CUMBRIA

I r r De..d........ 1/:d2s

Wyre / . .. ...

I : Y : :

1

ROMAN CUMBRIA

MILES (ROMAN) Rbha 4 I 0 5 10 13 20 BRMItNACUM - -

o s 10 15 20 25 30 --

yldsm -___-

'4 - K LOM ETRES

..

i'

FIG. 12. Map of Roman Cumbria. 25



374 NOTES

(Galacum). However, this seems to be a toponymic dead-end on the way to Galava and Clanoventa. Forts further to the north and east are separately named in Routes II and V of the Antonine Itinerary whilst the relatively intractable terrain immediately to the west of the fort does not provide the most obvious of routes into the Lake District. In addition, the distance of at least 20 miles to the next Roman fort at Ambleside (Watercrook again not being an option on etymological grounds) diverges markedly from the 12 miles between Alone and Galava recorded in the Itinerary. Rivet and Smith were also unable to find a Roman road into Lakeland along this route although Wilson, in 1884, traced a possible Roman road up Borrowdale rising at a gradient of 1:6 to cross Whinfell at a height of over 400oo m before heading in the direction of Whinfell Tarn. However, most of his evidence for the construction of this well- engineered, 4.1 m wide track appears to date to the enclosures of the early nineteenth century except for some possible Roman 'diagonal hammer marks' on foundation stones.1"' Even if this route is eventually confirmed, the associated chain of place-name identifications still remains questionable.

Taking an alternative tack, Shotter detected a common thread in the names Galacum, Alone, and Galava based on the name of a local god, Ialonus Contrebis, to whom an altar was dedicated near Lancaster.152 The River Lune was seen as a modern derivative, leading to the identification of these place-names with the line of three forts on this river at Lancaster, Burrow-in-Lonsdale (also known as Overborough), and Low Borrow Bridge respectively.153 This alignment then requires that the terminus at Clanoventa should be identified either with Brougham, which is more likely to be Brocavo in Route V in the Antonine Itinerary, or with Ambleside,154 again following the dubious deviation around or over Whinfell Beacon. This alternative route also contains discrepancies in all the stage distances, the most significant being the actual 19 miles between Burrow-in-Lonsdale and Low Borrow Bridge compared with the 12 miles stated in the Itinerary. Furthermore, the association of the names Galacum and Galava with Ialonus is not satisfactory phonologically. All the historical sources, Ptolemy, the Ravenna Cosmography, and the Antonine Itinerary record the initial velar stop g-/c- in the two former place- names which leaves little doubt as to their original forms. They must also have been essentially contemporary with the inscribed form, with its initial vowel or glide i-. This does not bear immediate comparison and prevents any explanation based on historical sound changes.

From this position of apparent stalemate one alternative route remains to be considered. In common with Haverfield's interpretation, the most obvious course for the notably long stage between Bremnetonnaci (Ribchester) and Galacum is northwards along the line of the main west-coast Roman road to the fort at Burrow-in-Lonsdale. At a distance of 30 miles, with no known intervening forts, this link provides an appropriate match for the 27-mile stage registered in the Itinerary. Ptolemy's equivalent place-name Calagum, probably recorded during the early Flavian penetration of Brigantia, may well have been sited along this route and, as there is some limited evidence for Flavian occupation at Burrow- in-Lonsdale,'55 does not contradict the identification proposed. The place-name itself, possibly meaning 'noisy stream', offers no assistance in confirming or denying the location.

Accepting the identification of Galacum with Burrow-in-Lonsdale on mainly geographical grounds, the onward routes to Low Borrow Bridge or to Watercrook then give rise to the problems already stated. The only remaining option is to return southwards to locate Alone at Lancaster. The distance of 14 miles compares with the 19 miles logged for this stage, a discrepancy which is not satisfactory but is no greater than found in the alternative routes proposed and matches other possible five-mile errors in Route X.156 Such an abrupt diversion may seem unacceptable at first sight but it is not an unusual feature of the Antonine Itinerary where many of the recorded British routes undergo sharp changes of direction at some stage in their progress. The Antonine Itinerary was not a comprehensive and systematic

151 Rivet and Smith, op. cit. (note 141), 171. T. Wilson, 'The Roman road over Whinfell', CW1 vii (1884), 90-5. E. Birley,'The Roman fort at Low Borrow Bridge', CW2 xlvii (1948), 16-8, endorses Wilson's route.

152 R.G. Collingwood and R.P. Wright, The Roman Inscriptions of Britain, I: Inscriptions on Stone (1965), RIB 6oo.

153 D.C.A. Shotter, 'The Roman name for Lancaster', in G.D.B. Jones et al., Roman Lancaster, Brigantia Monograph I (1988), 220-2. Repeated in D. Shotter and A. White, The Roman Fort and Town of Lancaster (1990), 12-15.

154 D. Shotter, Romans and Britons in North- West England (1993), 105-9. 155 E.J.W. Hildyard,'Excavations at Burrow in Lonsdale, 1952-53', CW2 liv (I955), 88-9, 93. Also a summary in

D. Shotter and A. White, The Romans in Lunesdale (1995), 40-5. 156 Rivet and Smith, op. cit. (note 141), 172. Camden, op. cit. (note 140), 431-4, also proposed this sequence.



NOTES 375

survey since many of the routes duplicate each other while other important links, along Hadrian's Wall or the Fosse Way for example, are omitted. The document contains routes surveyed throughout the second and third centuries and seems to be a random collection of individual official journeys.'157 In this light, the 120 degree turn to place Alone at Lancaster is considered tenable.

Attempts to equate the Roman place-name Alone with Lancaster and the River Lune, through the restored form Alauna, were eventually abandoned by Ekwall who could find no explanation for the loss of the initial a- or for the unsatisfactory sound change of the originally stressed medial syllable from -au- to -0- , changes which are required to arrive at the twelfth-century English form of the river name, LOn.158 Instead, a local god of the Lune valley, Ialonus Contrebis, may provide the key to the river name, and Lancaster, as Shotter has already suggested.

Contrebis is known from two locations within the Roman Empire, both on the River Lune, at Burrow-in-Lonsdale and at Foley Farm, one mile north of Lancaster (RIB 6oo and 61o). This local duplication of the name together with the community theme in its meaning 'of those who dwell together' and a likely civilian dedication (by Vatta) at Burrow-in-Lonsdale indicate a deity of local provenance.159 There is no great reason to assume a Continental origin although the name has a parallel in the town of Contrebia in Hispania Tarraconensis.160 In the Lancaster dedication, Contrebis then assumes the additional identity of Ialonus and this pairing of names suggests that the latter might also be of local origin. The only other known reference to the god Ialon(us) comes from Nimes in a dedication associated with the goddess Fortuna (CIL xII.3057). In that instance the link with Fortuna, the home- bringer, may indicate that the deity had been imported to a town where the local god Nemausus was widely worshipped. The name does not appear to be associated with any rivers in the region.161

The Itinerary name Alone can be derived from Ialonus without serious difficulty. Both names retain British 6 of the second and third centuries, in preference to earlier au as found in the alternative form Alauna,162 while the initial IA could easily have been read as a single, three-stroke character A (/A) in Old Roman cursive. The case ending -e recorded in the Itinerary version is probably for Classical -ae, locative,163 and indicates that the hypothetical place-name for Lancaster now proposed would have taken the feminine form *Ialona, in contrast to the god in the masculine.

A late fourth-century imperial register, the Notitia Dignitatum, includes in its schedule of military commands one which bases the tribune of the cohors III Nerviorum at Alione. The listing of this command lies towards the end of a sequence covering north-west England and immediately precedes the record for Bremetenraco (Ribchester).164 This geographical progression favours Lancaster, more than any other site perhaps, as the location of the command. Maryport, on the River Ellen (Alauna), is often associated with the place-name Alione but this is not supported by the garrison record there165 and

157 Rivet and Smith, op. cit. (note 141), 150-3. 158 E. Ekwall, The Place-names of Lancashire (1922), 167-8. Ekwall, op. cit. (note 148), 270-I. Early forms of the

name are Loin, Lon, Loon, Lonn (twelfth century), Lone (thirteenth century), Loone (fourteenth century) and Lune (sixteenth century). The currently preferred derivation of Lune is from Welsh llawn, equivalent to Old Irish slhn, meaning 'health giving', ibid., 27I. 159 I.A. Richmond, 'Excavations on the site of the Roman fort at Lancaster, 1950', Trans. Historic Soc. Lancs. & Cheshire cv (1953), 22-3, is also of the view that Contrebis refers to the district covering the lower Lune valley.

160 E. Birley,'The Roman site at Burrow in Lonsdale', CW" xlvi (1947), 135-7, considers the possible import of Contrebis from Contrebia via the cohors I Celtiberorum (recorded in Britain by CIL xvi.5i, 69, 93, and possibly at Caersws in RIB 2471.1) but rejects this because of the association of the name with Gallia Narbonensis, through Ialonus, and because of the probable civilian dedication by Vatta. However, in 'The Deities of Roman Britain', Aufstieg undNiedergang der rdmischen Welt II.I8.I (1989), 67, he maintains an open mind on the subject.

161 A.L.F. Rivet, Gallia Narbonensis (1988), 163, 166 (map). The only remotely comparable river name in the region is the Alzon (*Alisos/*Alisone) at the head of the town's aqueduct.

162 The inscription dates to the second century (Birley, op. cit. (note i6o, 1947), 136) or third century (Collingwood and Wright in RIB 6oo). K.H. Jackson, 'The British Language during the Period of the English Settlements', in N.K. Chadwick (ed.), Studies in Early British History (1954), 68-9, proposes British au> o [o:] by the late first century, > a by the late third century. P.Y. Lambert, 'Welsh Caswallawn: The Fate of British *au', in A. Bammesberger and A. Wollman (eds), Britain 4oo00-6oo00, Language and History (1990), 203-15, argues that au> o[3:] and remained as such.

163 Rivet and Smith, op. cit. (note 14I), 244. 164 ibid., 216-25. 165 M.G. Jarrett and G.R. Stephens, 'The Roman garrisons of Maryport', CWR lxxxvii (1987), 65. R.W.

Davies,'Cohors I Hispanorum and the garrisons of Maryport', CW" lxxvii (1977), 7-16. See also n. 171 below.



376 NOTES

its high cliff-top site invites an association with the Notitia Dignitatum's placing of cohors I Hispanorum at Axeloduno. The alternative sites proposed for Alione, at Watercrook or Low Borrow Bridge, were less important than Lancaster in the fourth century and the former, as well as being etymologically objectionable, does not appear to have been garrisoned at this time.166 They must therefore be regarded as less probable locations for the name. In marked contrast, Lancaster was completely rebuilt in A.D. 330-40 as a substantial fort of between 2.2 and 3.6 ha

(5-5-9 acres) based on the 'Saxon Shore' model.167 It seems unlikely that such a key late installation, unique in north-west England, would have been omitted from the Notitia Dignitatum.168 Whilst it is known that Lancaster was garrisoned by cavalry units such as the ala Sebosiana and ala Augusta from the first to third centuries,169 this arrangement may not have survived the major discontinuity brought about by the fort's rebuilding to a radically different plan. An irregular unit of Bargemen, or numerus Barcariorum, is also attested at Lancaster but would have been a subsidiary command if based in the fourth-century fort. Cohors III Nerviorum at Alione can therefore be allocated to Lancaster without necessarily being contradicted by the epigraphic evidence, although it may not have been the sole occupier of this large fort.170

Confirmation that Alione is to be associated with Lancaster is provided by the form of the name itself, with Alione preserving more specifically the local name *Ialona, after metathesis, rather than the more usual but less satisfactory restoration to Alauna.17' This variant also indicates that the initial i- of *Ialona was probably long and non-consonantal in order to facilitate its transposition.

A Hadrianic milestone found at Caton, just over 4 miles to the east of Lancaster, is usually interpreted as 'from L-4 miles' (RIB 2272). However, the right-hand side of the inscription is badly worn and the relevant detail in RIB 2272 must be largely conjectural in its portrayal by Collingwood as

C>,MP1M, presumably in an attempt to provide the semblance of an ansated panel. The cartouche is ambiguous and is now restored to an unorthodox but symmetrical >ILMPIIIII<] in Lancaster City Museum.172 Whatever the case, the initial character 'I' has clearly been formed separately from the enclosing scroll and should be incorporated into the text. Retaining the inverted but symmetrical cartouche gives the restoration LMEPIII, an interpretation which has been adopted more recently by Shotter and White.173 This permits the alternative reading 'from I[a]L[ona] 4 miles'.

The River Lune naturally suggests itself as the modern equivalent of *Ialonal74 and a derivation can be proposed which overcomes Ekwall's two objections to the alternative original, Alauna (see above). The loss of initial, disyllabic ra- can now be explained as a result of the transition of the name into English. It finds a parallel expression in the Old English word for 'river', namely ea, with the dative and presumably locative case being ea but also recorded as re in West Saxon.'75 The lexical coincidence, with the sound substitution of [e:0] for [i:0], would have resulted in the perceived generic ea (Ta-) becoming

166 T.W. Potter, Romans in North-West England: Excavations at the Roman Forts of Ravenglass, Watercrook and Bowness on Solway (1979), 18o.0.

167 Shotter, op. cit. (note 154), 90. 168 An early fourth-century date for the compilation of the Notitia for northern England has been mooted,

prompted by the old-fashioned regimental names recorded there, but this possibility now seems unlikely. The debate is summarized in Rivet and Smith, op. cit. (note 141), 218-25.

169 D.C.A. Shotter, 'The Roman garrisons at Lancaster', in Jones et al., op. cit. (note 153), 212-19. 170 This quingenary unit may have been at Chesterholm in the second century if the largely illegible RIB 1691 is

accepted. One of the unit's lead seals was also found at Newstead; I.A. Richmond, 'Roman leaden sealings from Brough-under-Stainmore', C'2 xxxvi (1936), 125. M.G. Jarrett, 'Non-legionary troops in Roman Britain: Part One, The Units', Britannia xxv (1994), 64. Civilian occupation may also have occurred within the fourth-century fort at Lancaster following destruction of much of the earlier vicus. Shotter and White, op. cit. (note 153), 33-

171 Derivation of Alione from Alauna fails to explain the intrusion of-i- which is regarded as a scribal error by Rivet and Smith, op. cit. (note 141), 245.

172 The absence of the relevant detail is also illustrated in the careless restoration to IPI by W.T. Watkin, Roman Lancashire (1883), 28.

173 Shotter and White, op. cit. (note 155), o101. 174 For the importance of river deities, with British examples, see J.P. Alcock,'Celtic water cults in Roman Britain', Arch. Journ. cxxii (1965), I-I12. For some place-name examples see also G. Webster, The British Celts and their Gods under Rome (1986), 72-3.

175 A. Campbell, Old English Grammar (1959), 254.



NOTES 377

detached from the, by then, meaningless proper name to give ea Lona176 probably in the seventh-century when the region first came under Anglo-Saxon domination.'77 Concerning the evolution of the medial vowel -0-, Jackson identifies a British long close 0 [o:] only, this becoming a in the late third-century and adopted subsequently as u, i or y in Old English.178 This, together with the reduction of the vowel arising from the English accent shift to the first syllable,179 would have rendered Ialona incomparable with Lon > Lune. However, Lambert has argued for the existence of a neo-Britonnic long open 0 [Z:] <au18o and, with the generic ea detached to leave the original stress, takes Ialona directly to the medieval form. Finally, the inflexional suffix -a, if it had not already been dropped in the British original, would have fallen into disuse as the name evolved into Middle English. The word ea itself was also lost from the English vocabulary at this later period. Thus *Ialauna > Ialona > ea Lona > Lon > Lune.

The meaning of Ialonus is given by Richmond as 'god of the meadowland', derived from Gaulish ialo- meaning an open space, clearing, or meadow, together with an -n suffix of association.'8"' Ekwall suggests that the same word may be incorporated in compound form in the names of the Wiltshire rivers Deverill and Fonthill although deriving from consonantal j and with long -a-.182 In these cases, the sense of river is also dependent on the initial part of the compound. Phonological objections apart, it is now clear that the name is more directly associated with an important river and an alternative derivation must be sought.

The river *Ialona has a namesake in the Iala, an unidentified tributary in the upper reaches of the River Po recorded in the fourth-century Peutinger Table,183 and this comparison suggests that both might share a common root ral- combined with a grammatical suffix -a/-ona. The suffix -ona is not to be confused with the -ona which occurs commonly in river namesl84 but is rather to be found, with &o<au, as the second element in the name Alauna/Alaunus.s85 The root itself is probably to be broken down further to T-al based on Indo-European *al- meaning 'water', again as found in Alauna.186 The initial T- perhaps had an emphatic function'87 to distinguish the name from the more common alauna/*alaventa names which appear to apply to lesser watercourses and tributaries.'88 The river *Ialona may therefore be regarded as a variant of Alauna but with a stronger meaning such as 'the main stream'.

Having established Alone (*Ialona) of the Antonine Itinerary at Lancaster would suggest that the initial sequence of Route X should then be retraced further southwards, leading to a potential but unidentified location for Galava at a distance of I2 miles in the vicinity of Garstang, bridging the Wyre. Clanoventa might then be located a further 18 miles beyond, perhaps on the Fylde coast. However, the place-name Galava, probably meaning 'vigorous stream'189 is incompatible with both the meandering character and the ancient name of the River Wyre, and its tributary the Calder, which extend over the

176 In compound names, ea usually, but not always, forms the last element as in Mersey, Waveney, Sweeney and Twineham. A contrasting form is Yeadon/Eaton.

177 K.H. Jackson, Language and History in Early Britain (1953), 195 (sound substitution), 214-18 (Anglo-Saxon ascendancy).

178 ibid., 305-17. British a was retained in Anglo-Saxon as a in the fifth-sixth centuries but was rendered as either u or i in words adopted in the sixth-seventh centuries and as u or y in the seventh-eighth centuries.

179 Campbell, op. cit. (note 175), 148. Alne and Allan, from Alauna, are obvious examples of this process. 180 Lambert, op. cit. (note 162), 203-15. The Notitia name Alione, dating to the end of the fourth century, unless a

fossilized form, undermines Jackson's third-century 0 > a and supports Lambert's case for a retained O[3:]. 181 Richmond, op. cit. (note 159), 22. Derivation based on G. Dottin, La langue gauloise (1920), III, 262.

Examples in A. Holder, Alt-Celtischer Sprachschatz n (1904), 7. 182 Ekwall, op. cit. (note 148), 124, 161; Jackson, op. cit. (note 177), 345. 183 K. Miller, Itineraria Romana (1916), cols 29-30 (map), 387. G. Wissowa and W. Kroll (eds), Paulys Real-

Encyclopddie Ix.i (1914), col. 545, equate the Iala with Pliny's lactus (Natural History III.I I18). 184 Nicolaisen, op. cit. (note 147), 177-9, for this formation with numerous examples. C. Watkins,' "River" in Celtic and Indo-European', Eriu xxiv (1973), 8o, for an alternative explanation.

185 Nicolaisen, op. cit. (note 147), 186, proposes a -na extension to a ud- formation. Rivet and Smith, op. cit. (note 14I), 243, Lambert, op. cit. (note 162), 214, and E.P. Hamp, 'Alauno-, -a. Linguistic change and proper names', Beitrage zur Namenforschung2 x (1975), 173-8, discuss alternative etymologies.

186 Nicolaisen, op. cit. (note 147), 186-7. See also Rivet and Smith, op. cit. (note 14I), 243-4, for a discussion of alternative derivations.

187 J. Pokorny, Indogermanisches Etymologisches Wdrterbuch I (1959), 285. 188 See, for example, the distribution map of British Alauna names in Rivet and Smith, op. cit. (note 14I1), 243. 189 Jackson in Rivet, op. cit. (note 150o), 74.



378 NOTES

relevant area. Garstang itself is not noteworthy as a Roman find-spot.190 This line of inquiry must therefore be abandoned.

The only alternative is to consider a return northwards from Lancaster to Carnforth and into the Lake District. Whilst the course of a Roman road in this direction has not yet been established, except for the initial section near Lancaster,9' it seems most likely that there would have been a direct road towards Kendal on the course followed by medieval travellers and by the general alignment of the present A6, rather than the diversion via Burton (A6070) used in more modern times.192

This route would place Galava on the River Bela, at either Beetham or Milnthorpe, at a distance of approximately 12-14 miles from Lancaster. The Bela's name is of Scandinavian origin193 and therefore raises no difficulties whilst the local topography, where the river falls 5 m over a limestone fault and cuts through a ridge of glacial deposits to reach the sea, to some extent reflects the definition of Galava as a 'forceful or vigorous stream'. At Beetham there have been numerous coin finds from the second to fourth centuries indicating a potential Roman military site.194 The village boasts a Saxon church foundation, while the cross-roads, bridging-point, and old stone buildings also provide a promising context. Although this hypothetical site is a short 8 miles from the next Roman fort at Watercrook, it may have been better placed to control the local coastline and routes crossing the Kent estuary into the southern Lake District. The Roman road heading north-west from Burrow-in-Lonsdale by Farleton Fell'95 may also have run to Beetham via Elmsfield rather than to the assumed destination at Watercrook. Altogether, there would appear to be sufficient grounds for investigating Beetham as the site of the Roman fort, Galava.

From there the final or, strictly speaking, initial leg of Route X must reach into the heart of the Lake District resulting in Clanoventa being placed at Ambleside, approximately 20-22 miles from Beetham compared with the 18 miles recorded in the Itinerary. The most likely meaning of the place-name is 'shore market, town, or field' but it need not be located on the sea coast (at Ravenglass) as Rivet insists.196 Shotter also locates Clanoventa at Ambleside, via the Whinfell route, but prefers a derivation from clan meaning 'clear' rather than the preferable and more obvious 'shore' connection.197 Ambleside fort was regularly occupied, possibly from as early as the Flavian period through to the fourth century, and had a sizeable civilian vicus attached. It lies only yards from the edge of Lake Windermere where there is evidence of a possible landing-stage (undated) providing ready access to the shores of the I I-mile long waterway.198 The fort also lies at the hub of several landward routes and the name could not be better suited to the location. It is certainly more apt than Galava which is difficult to associate in meaning with this flat and normally placid lakeside site.199 A coastal site may have provided a more logical terminus in a systematic survey but, as stated above, this was not the rationale behind the Antonine Itinerary. Ambleside seems a totally appropriate terminus for Route X of the Itinerary and this, in turn, reinforces the case for locating Galava at Beetham.

THE NOTITIA DIGNITATUM AND THE CUMBRIAN COAST

The terminal fort on Hadrian's Wall at Bowness is universally identified with the Roman place-name Maia which begins a sequence of western Wall names on two souvenir bowls, the Rudge Cup and the

190 A Roman shield was apparently found near the village in i8oo. C. Rothwell, Garstang Town Trail (1983), 2. 191 Shotter and White, op. cit. (note 153), 50. 192 B.P. Hindle, Roads and Trackways of The Lake District (1984), 49-61 (medieval), 62-79 (modern). 193 Ekwall, op. cit. (note 148), 31. 194 D. Shotter, 'Recent finds of Roman Coins in Cumbria', C'2 xciv (1994), 293. D. Shotter, Roman Coins from

North-West England: First Supplement 1995 (1995), 77-81. 195 F. Villy, 'A Roman road north-west from Overborough', C'2 xxxvii (1937), 49-51. 196 Jackson, op. cit. (note 177), 70; Rivet, op. cit. (note 150o), 53-4. 197 D.C.A. Shotter, 'Watercrook and Ravenglass: the names and the garrisons', in Potter, op. cit. (note 166),

316-17. Shotter subsequently accepts the 'aquatic' relationship in P. Graystone, Walking Roman Roads in East Cumbria (1994), 5-

198 R.G. Collingwood, 'The exploration of the Roman fort at Ambleside: Report on the second year's work

(1914)', C'2 xv (I915), 2-62. R.H. Leech, 'The Roman fort and vicus at Ambleside: Archaeological research in

1982', CW2 xciii (1993), 51-74. Shotter, op. cit. (note 154), 35. 199 I.A. Richmond and O.G.S. Crawford, 'The British section of the Ravenna Cosmography', Archaeologia xciii

(1949), 35, contemplate the River Rothay 'bursting' into Lake Windermere to justify the identification of Galava with Ambleside.



NOTES 379

Amiens Patera. Contradicting this identification is the place-name Tunnocelo which is listed towards the end of the sequence of Hadrian's Wall forts in the Notitia Dignitatum. This name also occurs in more garbled form as Iuliocenon in the Ravenna Cosmography of about A.D. 700 and can be restored to the correct form Itunocelum with the meaning 'Eden Point'. Bowness, at the mouth of the Solway (Eden) Firth, is the most obvious site for this name which can hardly be located elsewhere in north-west England without ignoring its clear etymological message.200 Locations further down the Cumbrian coast, beyond the Cardurnock peninsula and the inlet at Moricambe, do not bear such a close geographical connection with the River Eden whilst an alternative identification with the River Ehen seems based more on hope than a fully argued case.20' This latter location is also unlikely to have featured a landmark to match the name, even allowing for coastal changes. A lead sealing found at Ravenglass and bearing the stamp of the known garrison at Itunocelum, the cohors I Aelia classica, has led Shotter to equate the two places although the site excavator cautions against any such comparison given the widespread distribution of these artefacts, usually at a destination rather than the point of origin.202 However, it may possibly be more than coincidence that fragments of a military diploma (A.D. 158), issued to a veteran of the same unit, 'Hadrian's Own Marines',203 have also been found on the shore at Ravenglass204 although these artefacts are similarly foot-loose. What is certain is that the location, beyond St Bees Head, bears no geographical relationship whatsoever with the Solway Eden while the probable British origin of the local river names (Esk, Mite, and Irt) would also appear to preclude the existence of a second Eden at Ravenglass.205

The combined place-name sequence illustrated in the Rudge Cup and Amiens Patera-A MAIS ABALLAVA VXELODV(NV)M CAMBOGLANS BANNA ESICA-represents a contiguous line of forts on Hadrian's Wall and these have for the most part been satisfactorily identified as the respective sites of Bowness (or Drumburgh), Burgh-by-Sands, Stanwix, Castlesteads, Birdoswald, and Great Chesters.206 A significant exception is the omission of one of the two forts at the western end of the Wall and this is usually assumed to be the smaller intermediate fort of o.8 ha (1.95 acres) at Drumburgh rather than the larger terminal fort of 2.4 ha

(5.9 acres) at Bowness. The place-name Maia, possibly meaning 'larger (ones)',207 offers no specific topographic clue to support a location at either Bowness or Drumburgh. Bowness was previously thought to have been larger, at 2.8 ha (7 acres), and occupied by a cohors milliaria which would have ruled out occupation by the cohors I Aelia classica at Tunnocelo. However, the fort has now been confirmed as being of standard size.208 suitable for a cohors quingenaria, which leaves this option open.209

200 The location was first proposed by Horsley in 1732, op. cit. (note 140), 102-4. Rivet and Smith, op. cit.

(notex41), 380-1, reject the identification out of hand. 201 ibid., 381. 202 Shotter, op. cit. (note 154), 107. Potter, op. cit. (note 166), 42. RIB 2411.94. 203 A.R. Burn, The Romans in Britain:. an Anthology ofInscriptions (1932), I1 I, for the translation. 204 M.W.C. Hassall and R.S.O. Tomlin, Britannia xxvi (1995), 389-90. This army unit, located in a frontier coastal

outpost, may have retained its naval tradition although similar units in Aquitaine (Villeneuve-sur-Lot), Lower Germany (Cologne), and Syria (archers at Beirut) appear to have been regular army regiments. P.A. Holder, Studies in the Auxilia of the Roman Army from Augustus to Trajan, BAR Int. Ser. 70 (1980), 67, 238, 333-4. For a discussion of this issue, see J.G.F. Hind, 'Agricola's fleet and Portus Trucculensis', Britannia v (1974), 287 (n.14). If cohors I Aelia classica had operated around the Cumbrian coast in a naval capacity, this might possibly explain the distribution of the two artefacts discussed. Another diploma recording the unit's name has been found at Chesters (RIB 2401. 1io/ CIL xvI.93).

205 Ptolemy's Ituna estuary, from its relative position (Geography 11.3.3), must refer to the Solway Eden and not to the smaller inlet at Ravenglass.

206 D.J. Breeze and B. Dobson, Hadrian's Wall (1976), 272- 3, for a typical summary of the evidence. 207 Rivet and Smith, op. cit. (note 141), 408--9. 208 P.S. Austen, 'How big was the second largest fort on Hadrian's Wall at Bowness-on-Solway', in V.A. Maxfield

and M.J. Dobson (eds), Roman Frontier Studies 1989-Proceedings of the XVth International Congress of Roman Frontier Studies (1991), 6-8. R.L. Bellhouse, 'Hadrian's Wall: the forts at Drumburgh', CW2 lxxxix (1989), 35, gives a slightly smaller figure of 2.2 ha (5.46 acres).

209 A tribune, Sulpicius Secundianus, is recorded at the fort in A.D. 251 -3 (RIB 2057, 2058), but by this date the title may not be indicative of a milliary cohort command. A naval tribune is recorded in CIL 11.2224.



380 NOTES

A relative dating sequence has been established for the Rudge Cup which enabled it to be dated to about A.D. 150.210 The Amiens Patera has an almost identical date, being intermediate in style between the Rudge Cup and a piece known as the Hildburgh Fragment, both of which may have been made by the same hand.21' This post-Hadrianic date is widely accepted212 and, as such, securely ties the first place- name Maia to Bowness at the western terminus of the Wall. However, it must be noted that the dating assessment lay within a range extending to a couple of decades later or a 'generation earlier'213 thereby encompassing the whole Hadrianic-Antonine period. An Antonine date, when the Wall had been largely abandoned, lacks credibility and the substance of the text dictates that the cups should be Hadrianic. The most obvious moment for the production of such commemorative artefacts would surely have been to celebrate a significant completion of the Wall by about A.D. 130. Furthermore, freeing the name sequence from its supposedly later date permits more flexibility in its interpretation.

Returning to the Notitia Dignitatum, the universally agreed equation of Maia with Bowness presents numerous difficulties.214 Not only is this place-name, and hence a significant fort,215 seemingly omitted from the Notitia but Tunnocelo then assumes a peripatetic role, being assigned by various authors to most of the sites on the Cumbrian coast or even to a postulated fort at an unlocated Eden Head.216 Congavata, following Aballaba (= Burgh-by-Sands) in the Notitia listing, is usually allocated to Drumburgh in order to provide this remaining Wall fort with a place-name but this erroneously assumes that the geographical sequence in the Notitia is reliable. The small fort would only have been manned by a detachment of infantry and would not have merited independent listing in the Notitia. This identification also requires a convoluted argument to explain the omission of Congavata from the Rudge Cup.217 All these problems are immediately overcome, however, by locating Tunnocelo at Bowness and Maia at Drumburgh. In other words, Drumburgh was the terminus of Hadrian's Wall at the time the Amiens Patera and Rudge Cup were produced.

The archaeological evidence to support this proposition is mixed although the direct evidence is generally affirmative. At Bowness, a sealed quarry-pit preceding the first phase of building work contained both coarse-ware and Samian pottery thought to be of late Hadrianic/early Antonine date and the excavator considered that all the finds from this early feature would fit more comfortably into a later Hadrianic date rather than to a mid-I20s fort foundation.218 A similarly late date is also apparent at the immediately preceding Milecastle 79 where the site excavators did not find a single sherd of coarse ware (out of a sample of 750 fragments) typical of primary deposits elsewhere on Hadrian's Wall and made a telling comparison with the pottery assemblage from Milecastle 50 (High House).219

The establishment of the fort at Bowness is likely to have coincided with the abandonment of its Trajanic neighbour at Kirkbride. Unfortunately, the limited dating evidence from Kirkbride fails to provide a specific terminus post quem for this although finds of Black-Burnished Ware may indicate a lingering presence there well into the Hadrianic period.220

210 J.D. Cowen and I.A. Richmond, 'The Rudge cup', Arch. Ael.4 xii (1935), 310-42. 211 ibid., 324. J. Heurgon, 'The Amiens Patera', JRS xli (1951), 22. 212 Rivet and Smith, op. cit. (note 141), 233, as just one example among many. 213 Cowen and Richmond, op. cit. (note 210), 319 (n. 17), 331, 333. 214 The on-going confusion is more recently illustrated in N. Hodgson, 'The Notitia Dignitatum and the later

Roman garrison of Britain', in Maxfield and Dobson, op. cit. (note 208), 90-1. 215 Bowness appears to have been occupied into the late fourth century but possibly only with a reduced garrison.

Potter, op. cit. (note 166), 323. 216 J.C. Mann, 'Birdoswald to Ravenglass', Britannia xx (1989), 76. It is clear, from the largely intact chain of

Roman defences around the north Cumbrian coast, that any coastal changes are unlikely to have been substantial enough to allow for such a lost feature.

217 Rivet and Smith, op. cit. (note 141), 315. 218 Potter, op. cit. (note 166), 333-4, 337-9, 344-5. 219 I.A. Richmond and J.P. Gillam, 'Report of the Cumberland Excavation Committee for 1947-49: 3. Milecastle

79 (Solway)', CW'2 lii (1953), 28-37. They report an absence of 'carinated bowls, handled mugs and debased rustic ware' at the MC79 site.

220 R.L. Bellhouse and G.G.S. Richardson,' The Roman site at Kirkbride, Cumberland', CW2 lxxv (1975), 79-89.



NOTES 38 1

These later dates are also supported by slight evidence that the fort at Bowness was built across the line of the Vallum221 indicating that it may post-date the Vallum's construction in about A.D. 125-7.222 However, at Burgh-by-Sands, the Wall fort also appears to impinge on the Vallum,223 suggesting that the whole western stretch of Hadrian's Wall was the last section to be completed. Burgh-by-Sands nevertheless conforms to the layout of earlier forts, lying astride the Wall, a characteristic which was possibly abandoned in about A.D. 126,224 and the two dates seem to provide a narrow time range for this fort's construction. In comparison, the forts at Drumburgh and Bowness were both built flush with the Wall, according to the later design, but this may have been due to their waterfront locations. Drumburgh is usually regarded as quite a late addition to the Turf Wall and certainly it was constructed separately from the Wall, as was often the case.225 The west ditch of this fort stops 12 m short of the Wall, and is therefore later than the planned or actual Wall, but it is not possible to conclude whether this dating gap is significant.

Comparison of the morphological characteristics of the Turf Wall milecastles and Cumbrian coast milefortlets either side of Bowness indicates that they were constructed as two separate systems, the former with landward and probably seaward gates, as at Milecastle 79 (Solway), the latter with seaward gates only and enclosed by a ditch, as at Biglands Milefortlet I and the somewhat anomalous Milefortlet 5 at Cardurnock. The coastal fortlets were also set within a linear zone of parallel ditches with fronting palisade.226 On the lower Cumbrian coast, at least, the chain of milefortlets and towers appears to have formed an integral part of the initial Hadrianic frontier system, being more or less contemporary with the foundation of the fort at Maryport in about A.D. 122-3.227

Evidence for the possible continuation of this early open coastal system back from the apparent pivotal point at Bowness to Drumburgh is sparse. Biglands Milefortlet I is situated 1,675 m from the presumed Bowness Milecastle 8o which is further than the standard 1,480 m, although not uniquely so.228 This might suggest a setting-out point based on either the corner of a pre-existing fort at Bowness or, alternatively, a terminus at Drumburgh. However, excavation at Bowness in 1976 and at Milecastle 79 provided no signs of an eastward extension to the double ditch and milefortlet system,229 although this earlier chain may have taken a more forward line hugging the coast. This was suspected by Collingwood who detected potential fortlet sites at Knock's Cross (also known as Grey Havens or Old Police House) and at Fisher's Cross (Port Carlisle) which is associated with Roman coin finds and a nearby inscription.230 Higham and Jones also doubted that Bowness was the most logical location for a switch

221 R.L. Bellhouse, 'Roman sites on the Cumberland Coast: Hadrian's Wall. The fort at Bowness-on-Solway, a reappraisal', CW2 lxxxviii (1988), 33-53. F.G. Simpson, I.A. Richmond and J.M. McIntyre, 'The Stone Wall, Turf Wall and Vallum west of Burgh-by-Sands', CW2 xxxv (1935), 213-20.

222 Accepting the timetable proposed by Breeze and Dobson, op. cit. (note 206), 77. 223 P.S. Austen, 'Recent excavations on Hadrian's Wall, Burgh-by-Sands', CW2 xciv (1994), 49-53. R.G.

Collingwood, 'Exploration at the Roman fort of Burgh-by-Sands', CW2 xxiii (1923), 2 (map). 224 Breeze and Dobson, op. cit. (note 206), 68-9. 225 ibid., 47. F.G. Simpson and I.A. Richmond, 'Report of the Cumberland Excavation Committee for 1947-49: 1.

The Roman fort at Drumburgh', CW2 lii (1953), 13. 226 T. Potter, 'The Biglands milefortlet and the Cumberland coastal defences', Britannia viii (1977), 149-83.

Richmond and Gillam, op. cit. (note 219), 17-37. F.G. Simpson, K.S. Hodgson et al., 'The coastal mile-fortlet at Cardurnock', C'2 xlvii (1948), 78-127. G.D.B. Jones, 'The western extension of Hadrian's Wall: Bowness to Cardurnock', Britannia vii (1976), 236-43.

227 Jarrett and Stephens, op. cit. (note 165), 61, date the foundation of Maryport to about A.D. 122-3. R.L. Bellhouse, 'Roman sites on the Cumberland Coast, 1968-1969', CW2 lxx (1970), 40-7, established that the coastal chain was set-out from Maryport at the same or later date. R.L. Bellhouse (M.G. Jarrett), 'Roman sites on the Cumberland coast, 1954', CW2 liv (1955), 40, 49-50, reports the find of a coin in near mint condition and dated to A.D. 119-21 which appears to have been placed in the foundations of Tower 13a and this suggests that the coastal system and fort were contemporary.

228 R.L. Bellhouse, 'Roman sites on the Cumberland Coast 1966-1967', CW2 lxix (1969), 69-74. G.D.B. Jones, 'The Solway frontier: Interim report 1976-8 I', Britannia xiii (1982), 296.

229 Potter, op. cit. (note 166), 324-35; Richmond and Gillam, op. cit. (note 219), 28-37. 230 R.G. Collingwood,'Roman signal-stations on the Cumberland coast', CW2 xxix (1929), 149-50. The spacing

of Collingwood's sites is not consistent, however, with a Wall terminus at Drumburgh nor with an extension of either the milecastle or milefortlet systems. Sites some 200 m west of Milecastles 77-80 would be more suited to a potential Drumburgh-Biglands milefortlet arrangement.



382 NOTES

from Wall to coastal defence systems and noted unspecified anomalies in Vallum features to the east of Bowness.231 If the double-ditch system had followed a similar line to the Vallum, the slight ditches might have been obliterated by it. However, no signs of a second parallel ditch were noted, at least on the south side of the Vallum, near Kirkland (Milecastle 78).232

Drumburgh's small size and close spacing relative to its neighbours at Bowness and Burgh-by-Sands is anachronistic. However, there does not appear to be any weakness or specific threat to the Wall in this location, of all places, and the justification for a later insertion is hard to see. The unique circumstances may be better explained by regarding Drumburgh as the original terminus of Hadrian's Wall, at the point where it first reached the coast and the Solway Firth, rather than at Bowness, isolated on the Cardurnock peninsula. If the Turf Wall had terminated at Drumburgh, it may have run down to the shore immediately to the north-west of the fort, possibly along the line of one of the present field boundaries. Richmond reports the find of squared sandstone blocks, with Roman tooling, close to the shore in this direction, conceivably marking a harbour installation or a turret.233 Although a larger fort at Drumburgh would be more pleasing symbolically, its small size is perfectly compatible with a terminal site fronting the sea, marsh and inter-tidal sandflats. The situation is also mirrored at the east end of the Wall where the original terminus at Newcastle now seems to have ended in a relatively small fort of approximately 1.25 ha (3 acres).234

Great Chesters (Esica), omitted from the Rudge Cup but added to the later Amiens Patera, is also regarded as a later Wall fort since its northern face is aligned with the Wall. It may be too neat to suggest that the Rudge Cup represents Hadrian's Wall as completed to Drumburgh by about A.D. 128 with the Amiens Patera, including Great Chesters, representing the situation a year or two later, but with both cases preceding the final completion of the Wall to Bowness in the mid-1I30s.

The closing sequence in the list of commands recorded in the Notitia Dignitatum as coming under the control of the Duke of the Britains can now be completed for north-west England. The relevant posts, with preferred identifications, are: ...Aballaba (Burgh-by-Sands): Congavata: Axeloduno: Gabrosenti (Moresby): Tunnocelo (Bowness): Glannibanta (Ambleside): Alione (Lancaster): Bremetenraco (Ribchester)... Of the remaining commands to be discussed, all probably on the Cumbrian coast, Aballaba is securely tied by inscription and the Rudge Cup to Burgh-by-Sands235 while Gabrosenti, home of the cohors HII Thracum, can be fairly safely identified with Moresby on epigraphic grounds.236 Maryport, one of the earliest forts in the region and garrisoned throughout the second to fourth centuries, would have been a key site for monitoring the western sea approaches from Ireland and Scotland. It must have merited inclusion in the Notitia and can probably be identified with Axeloduno on epigraphic and topographic grounds.237 Finally, Congavata, probably Latin for 'curved', must have been one of the remaining coastal forts at Burrow Walls, Beckfoot, or Ravenglass or is, alternatively, an unknown site. Burrow Walls appears to have been founded in the late fourth century and may have been occupied by an irregular auxiliary unit, or numerus, rather than the cohors HII Lingonum recorded at Congavata. Beckfoot has not yet provided evidence of late occupation.238 As the place-name description

231 N.J. Higham and G.D.B. Jones, 'Frontier, forts and farmers-Cumbrian aerial survey 1974-5', Arch. Journ. cxxxii (1975), 20-3.

232 Simpson, Richmond and McIntyre, op. cit. (note 221), 214-15. 233 Simpson and Richmond, op. cit. (note 225), 14. Some stones are still located at NY 2625 6027. 234 S.S. Frere, 'Roman Britain in 1985: I. Sites explored', Britannia xvii (1986), 376-8. The conjectural fort outline

encloses about o.85 ha (2 acres) but should probably be larger. 235 E. Birley, 'The Beaumont inscription, the Notitia Dignitatum, and the garrison of Hadrian's Wall', CW2 xxxix

(1939), 190-4. 236 E. Birley, 'The Roman fort at Moresby', CW2 xxxxviii (1949), 55-6. RIB 797, 803, 804. 237 Accepting Davies, op. cit. (note 165), 7-16. Jarrett and Stephens, op. cit. (note 165), 6I1-6, are inconclusive and

depend on dubious identifications for other forts in the region. Rivet and Smith, op. cit. (note 14I), 221, regard the Notitia name Axeloduno as a copying error and rather arbitrarily correct it to Mais (= Bowness). This is based on the need to find a place for Bowness in the list of Notitia commands and to retain the Rudge Cup name for Stanwix as Uxelodunum while arguing that Petrianis, also located by the Notitia Dignitatum at Stanwix, is a ghost name. Mann, op. cit. (note 216), 78 (n. 13), makes a persuasive case for rejecting Petrianis as a ghost name and an alternative explanation for this problem, requiring no re-writing of the Notitia, may be that both Maryport and Stanwix were initially named Uxelodunum. However, the proximity of the two forts and the resulting confusion may have led to the latter being renamed after its garrison which was the premier auxiliary unit in Britain, the ala Petriana.

238 R.L. Bellhouse, 'The Roman fort at Burrow Walls, near Workington', CW2 lv (1956), 40.



NOTES 383

perhaps favours the coastal configuration at Ravenglass, its identification with Congavata is provisionally proposed.239

THE RAVENNA COSMOGRAPHY A FINAL WORD

The early eighth-century Ravenna Cosmography lists approximately 300 place-names from the Roman period in Britain. However, it is hopelessly lax in its methodology with wholesale corruption of the name forms. Richmond and Crawford have attempted to establish a logical pattern in the sequence of names but a rather startling map in An Atlas of Roman Britain illustrates the full extent of the problem and the unreliability of this source.240 With this proviso, likely Cumbrian place-names can be discerned within the partial sequence ...Aluna (Alauna): Camulo duno (Camulodunum): Calunio/Gallunio (Galava): Medibogdo (Mediobogdo): Cantiventi (Glannoventa): Iuliocenon (Itunocelum): Gabrocentio (Gabrosentum): Alauna: Brihra (Bibra): Maio (Maia).. .241

Most of this section coincides with the place-names already discussed with Aluna (variant Alicinca), perhaps duplicated by Alauna, being *Ialona/Alione/Alone at Lancaster. Camulodunum is a site recorded by Ptolemy in Brigantia and is probably a Flavian site in West Yorkshire. Similarly, Bibra cannot be identified precisely but presumably lies near Drumburgh on the Solway Plain. Mediobogdo, listed between Galava and Glannoventa, is usually associated with the Roman fort at Hardknott midway between Ambleside and Ravenglass.242 However, given the locations now proposed for these two place- names, at Beetham and Ambleside respectively, Mediobogdo, meaning 'place in the middle of the curve', may be sited more appropriately at Watercrook.

36 Featherhall Avenue, Edinburgh

239 Ptolemy's estuary at Moricambe, meaning 'curved or crooked sea' and probably located on the Cumbrian coast, might provide a comparable native name. Rivet and Smith, op. cit. (note 141), 420-I.

240 Richmond and Crawford, op. cit. (note 199), 1-50. B. Jones and D. Mattingly, An Atlas of Roman Britain (1990), 31 (Map 2:14). See also L. Dilleman, 'Observations on Chapter V.31, Britannia, in the Ravenna Cosmography', Archaeologia cvi (1979), 61-73.

241 Names as restored by Rivet and Smith, op. cit. (note 141) and numbered IIO-20 in Richmond and Crawford, op. cit. (note 199), 18.

242 Richmond and Crawford, op. cit. (note 199), 40, see the sweeping landscape there reflecting the name. However, Shotter, op. cit. (note 154), o107, favours the more meaningful association of Mediobogdo with Watercrook.

A Roman Lead Sealing from Goodnestone near Canterbury. Michael Still writes: In 1995 a lead sealing was found by metal-detector in a field at Goodnestone near Canterbury (FIG. 13; PL. XXXIVB). It was one of the many objects, including military items and pre-Roman coins, located on this site.243 The various artefacts span the period from the first century through to the fourth. The physical details of the sealing in question are as follows: Size of flan (width by height by thickness): 23 by 20 by 8 mm Size of impression: 19 by 15-5 mm Direction of thread hole (given as hours of the clock): 3.00oo-8.00oo Size of thread hole (given as width by height while sealing is lying flat): 1.5 by 2 mm.

The sealing bears what was intended to be a circular impression on the obverse while the blank reverse is hemispherical. The sealing was badly impressed when formed, resulting in a certain amount of distortion and damage. The impression shows a figure on horseback riding towards the right and holding a spear, all on a pearl groundline. There appears to be a strange shape behind the figure's head. Unfortunately the highest points of the relief detail have been worn away. There are already 18 sealings of this type known although poor states of preservation together with the vagaries of stamping molten lead and the differing quality of past illustrations make it extremely difficult to say which impressions are from the same dies.

The details of the parallels are as follows:

243 The site will be the subject of a forthcoming report by the Canterbury Archaeological Trust which will be published in Archaeologia Cantiana, the journal of the Kent Archaeological Society.



Smith, Roman Place Names in Lancashire and Cumbria

Documents