IV. NOTES ON THE "ROMAN ROADS" OF THE ONE-INCH ORDNANCE MAP OF . Br JAMES MACDONALD, LL.D., F.S.A. SCOT. PLATES III., IV., V. TH . 3 E ROADS. On maps of , such as that in the Monumenta Historica Britannica, a Roman road is represented as having left the Southern Wall, near , and proceeded northwards to Longtown on the Esk, a small market-tow n Cumberlandi n . Thence, takin ga north-westerl y direction t i crosse, e borderth s . Entering Dumfriesshire t i passe, s through Gretna parish into Kirkpatrick-Fleming t I the. n runs into , goes past the well-known fort of Birrens and shortly after- wards skirts the hill of Birrenswark in Hoddam. On leaving Hoddam, t i crosse e south-easth s t par f Tundergarto t h and, stil proceedin. l g north-west, enters . Passine towe nortth th f f no o o ht g e roath , dGallaborryhillo t run n o s , situate centre th f n i deo an extensive holm between the Annan and the Dryfe and a short distance to the north of their confluence. Not far from this, on Tor- wood Moor, there wer e remainRoy'n ei th y sf earthworksda so , described by him as one of Agricola's camps ; and in the neighbourhood there is a large number of other earthworks, some of which have also been called Roman camps. Here the road is divided. What is apparently the main branch pro- ceeds almost due north along the valley of the Annan, keeping to the eas thaf o t t rive l througal r h Applegurt Wamphvayd han . After enter- ing Kirkpatrick-Juxta, it crosses the Annan and thenceforth runs on e leff thath o t t stream, passing Tassiesholm e sitsupposeth e e b o dt " ROMAN ROAD ORDNANCSN "SCOTLANDF O O 9 P 29 MA E .

of another of Agricola's camps. Crossing the Evan, not far from its junction with the and the Annan, it ascends the hill that separates the valleys of the first and the last named of these three rivers, and pro- ceeds in a fairly straight line along it, at first in Kirkpatrick-Juxta and afterward Moffan i s t parisha shor t A t .distanc e north-west froe mth Beef-Tub it enters Lanarkshire. loow kho e bac w o Dryfesdaleholm t kf I branca e se he w strikin, f gof westwards fro maie mth e ndirectio th roae Nit th n i d hf o nvalle d yan going through and Tinwald parishes to the village of Duncow in . Ther t i eturn s northwards and, running through Closeburn and into , proceeds along the pass known Wele th s l a Path. Entering Lanarkshir Clyde th y eeb valley t Tejoini , s the road in the parish of Crawford. north-wese th n I Dumfriesshirf to smala e l portio roaa f nmarkes o di d " Roman" on the Ordnance map. It passes eastward into that county from Altry Hill, in . This isolated fragment must have bee t downse Romas e nbeliea th n fi n that traversin vallee e gth th f yo Cairn it had connected the Ayr and Dalmellington road described in a former paper1 wite Nithsdalth h e road jus t o sucreferreN h . to d road appear othen so r map f Romaso n Britain. It will thus be seen that the Dumfriesshire " Roman " roads may he regarde s formina d g four sections:—(1 e Loweth ) r Annandale) (2 ; the Upper Annandal e Cair e Nithsdaleth th ) n ) e(4 Valle (3 ; d an ;y roads. Of all of them, only detached fragments, as shown on the O.M., now remain. 1. The Lower Annandale Road.—On the O.M. four portions of this roa laie dar d down—on Kirkpatrick-Flemingn ei , about three-quarterf so a mile in length; another in Hoddam, upwards of a mile ; a third, partly in Hoddam and partly in Tundergarth, nearly two miles; and a fourt Dryfesdalehin , abou quartea t milesaia coursbe Its dof r . emay o grean t ta distanceithen ru follow o rt o t e r from o ,Carlisle th , d ean road, which again has guided the engineers of the t I takes. n facti e mos, th , t natural route frowestere mth n

border to the valley of the Clyde. 1 Proc. Soc. of Antiq. of Scotland, vol. xxvii. pp. 417-432. 300 PROCEEDINGS OF THE SOCIETY.

The first of these detached pieces, and the ' only one between the Borde d Birrens t mentionean r no r the o s i , 'd New n Ol eithei d e - th r Statistical Account'of Kirkpatrick-Fleming, and the absence of any notice t oi frof m bot s significanti h s showina , g thae epitheth t t "Roman" greao n d localityte holha th n di . No doubt ther oncd eha e bee roana d there course th , f whiceo h must have bee nSurveyore th pointe o t t dou s as the Annandale " Roman " highway. But almost all traces of it are now obliterated. Itinerariume Th Septenirionale (1727) contain e firsth st referenco et Roman road n Dumfriesshiri s e tha I havt e come across. Neither Camden nor Gordon of Straloch makes any mention even of Roman camp n theii s r notice f thao s t county r RoberSi . t Sibbalds hi n i , Historical Inquiries, speaks doubtfully of there having been a Roman fort at , another at what he calls the "village" of Solway, and also a " Roman port" on the Nith, somewhere below the town of . s onl t hi thes ye Bu referencear 1e o Romat s n remainn i s Dumfriesshire. Yet we know that he had been furnished with accounts no e t" th Shironl e f f "o Dumfrese,yo th Stewarde f o t f o ybu " Anandale " and the " Sheriffdom of Mddisdale " by correspondents who, it may be presumed, wrote, with all the local knowledge of the antiquities of these districts thathes wa tn available.2 The notice of the Lower Annandale Road in the Itinerarium is brief, and not very definite. As far, at least, as Birrenswark, the author infers rather than asserts its existence. He speaks, indeed, of there being "a noble Roman military way" beyond that hill. But he describes the previous part of its course in words that seem to leave a choice between differeno tw t routes e selecth . l Firsal s f Birrenswarko t fooe f th o tt a , whic certaie ar h n earthworks that, owing apparentl o theiyt r forme h , sets dow Romas na n camps d makean , e hil havo th st l e been Agricola's first halting-plac Scotlandein theHe n. speculate the par the of t to sas countr Romae yth n general probably traverse orden di reaco t r h this spot.' Agricola, he implies, must have used some road, which then or afterwards

became a Roman .road and was continued as far as his conquests extended. Sibbald1 , Historical Inquiries, p. 36. 2 Nicolson's Scottish Historical Library, . 19-21pp . " ROMAN ROADS " ON ORDNANCE MAP OF SCOTLAND. 301

e Itinerarium th e tex f th o t n I Gordon inclinee opinioth o t sn that the invader, after subjugating the island of Mona (Anglesey), marched northwards as directly as possible, crossing the upper part of e Solwath yd b tidpushin Firtan eb e t a hg forwar o Birrenswarkt d , wher e restea e timeh takee H r fo d.s into account, howevere th , possibilit e sea-leveth f yo l aroun r coastdou s being highe Agricola'n i r s day than it is now, and admits that the Romans may have been, obliged to cross much furthe . Theiup r r advanc Birrenswarn o e k would then take place alon "e Eomagth "y lainwa d s map eithen dowhi I .n o nr case, Agricola, he believes, encamped on this hill for a longer or shorter period before penetrating into the country beyond. Sufficient proof of this Gordo e nearthwork th find n i s s already mentioned e largeth , r thrown up on its southern, and the smaller on its northern slope. These he at once connected with Agricola. Of a somewhat rectilinear shape, thee eacar y h surrounde a well-marke y b d d rampara ditch d an .t largee Th r inclose a spacs 0 fee 0 fee90 et 60 s tlon y broadha b g t I . three openings or gates on each side and one in each end. A spring of water bubbles up not far from its centre, flowing down- wards in an unfailing stream. A small inclosure in its north-east angl s beeha e n e Prcetorium;supposeth e b o d t therd an e seemo t s have been another, somewhat similar, on the south side of the western entrance, which, however, had not been observed by Gordon. The smaller earthwore samth f e o e larger lengts th i k t onls a bu h,y half o gate tw n eaci s t h bu sided t ha a , e t widthwite i th on hd an ; each end. All the openings in the sides and ends have been pro- tecte moundy db circularw no s s t planned probabla bu ,o s t ,y no raise d in front of them. Another peculiarity is that both earthworks had apparently been joined e easter t th leasa , n e o hillnt y th b sid, f o e a continuous rampart, with intermediate " posts." Accordingly, Gordon regard squareo "e tw s th division r so encampments e formin "s a on t gbu — " the first Koman camp of any to be met with in the South of Scotland, anmose dth t entir besd ean t preserve dI eve r saw." l o milesTw ' south-eas f Birrenswaro t s Middlebi k r Birreno y s Fort, likewise " discovere dGordon y "b e describe inclosurn H a . s a e t si th e 1 Itinerarium Sepientrionale,. 16 . p 302 . PROCEEDINGS OF THE SOCIETY.

interior area of which measured 467 feet by 349 feet. It is of an oblong form, surrounde y foub d r ditche d fivan se ramparts, with four openings intinteriore oplath e givee Th nh .s represent complets a t si e informee ar e tex e th entir d tw n i dan t etha Watee bu • th t Meif o rd nan Haughile th l Burn—two small streams that meet clos it—"havey b e , whe floodn ni , carried awa ygreaa aggee t ditcheth d pare f an ron o t n so side. o subsequenT • " t observers there appeare o havt d e bee s manna y s sevea n rampartx ditchee northersi th d t a san s n endn a wels a s , a l outer entrenchmen r "o t Procestrium e westh e forttn th o "sid f , o e which s crosse"wa d obliquely e o Clydesdale.Eomat bth yy wa n " ^ Gordon conjectures that Birrens ." might have been built by Agricola, by way of an exploratory castellum to the noble camp of Birrenswark." Eoy, on .the other hand, takes Birrens to have been the Eoman head- quarters d regardan , e Birrenswarth s k inclosure s summea s r outposts. He differs, too, from Gordon as to the date of the latter, believing that the mounds in front of the openings are evidence that the " camps " belong to a later era of Eoinan military field-work,—as late, perhaps, as Hadrian. In support of the Eoman origin of all these "camps," Gordon has little adduce—nothingo t , indeed case Birrenswark—excepf th e o n i , n ow s hi t belief that they were constructed by the soldiers of Agricola; At Birrens he saw long hollow square stones, a stone-arched vault, marks of stone buildings and one stone with Eoman letters upon it " but so defaced that it was unintelligible." He does not explicitly say • that these f theo objecty m an wer r se woro eth 'th f ko e Eomans e addsH . , indeed, that several Eoman coins and a gold medal of Constantius Chlorus had been .found theree presenceTh . , however f suco , h portable articles can hardl regardee yb sufficiens da establiso t t s maihhi n proposition. 173n i t 1notabla Bu e discover s mad ywa t Birrensa e , which seemed to justify so far the views Gordon had expressed regarding the fort. This was the sculptured figure of the goddess Brigantia, an altar dedi-

. cate Mercuryo dt , and e inscribeth , d pedesta statua f o lf Mercury—eo : all of which were afterwards deposited in the Penicuik'Collection'of 1 Roy, Military Antiquities, p. 119, and Plate XXIV. Stuart, Caledonia Romana, p. 123. "ROMAN KOADS ORDNANCN O "F SCOTLAND O 3 P 30 MA E .

'Antiquities.1 The circumstances under which they were secured by " Baron Cler besk d relatee s ow"b ha t hi n n wordsi da margina n I . l not o Memoirst ey Life, e writes:—M h f o " About this time (1731) the fine pieces of antiquity now at Penieuik were found near the Roman cam t Midlebiea p . They consis e goddesa statuth f o f to e s Brigantia and two altars inscribed to Mercury. These stood in a little temple which, by age, had fallen down and become a ruinous kind of heap. These ruin se ground th wera poo n i f eo rs e ladySh . caused some stones to be made use of for building a little stable. When I chanced to pass the way I discovered the stones, and gave the poor lady two guineas for them. I consider these antiquities n Britaini e kin w th thereford e chie no df th an , o f I ewrot a eLati n dissertation upon them, tha at least t posteritdespisnot and e may y destroy them." In a subsequent note he describes the spot where they were found as being " on the side of the antient Roman camp at Midlebee."2 Besides these antiquities a numbether e ar ef otheo r r altars and inscribed stones, now in the Scottish National Museum and elsewhere, that were subsequently foun t Birrens.da 3 Gordo s thusnwa , eithe intuitioy b r accidenty b r o n , right whee h n fixed on Birrens as the site of a Roman encampment, although it was probably something more tha ncastraa (estiva subsidiar Birrenswarko yt . Onl ystatioa r forno t couldcircumstancese th n i , , have yielde o mands y lapidary relics of Roman times. We are not, however, to jump to the conclusion that the ramparts of Birrens, as now to be seen, are all a part of the original Roman fort. There is nothing in the classical writer o shot s w thae Romanth t s practised suc a hmod f fortio e - fication. Circumstances may be conceived that would lead them to adopt it in order to meet exigencies new to them. But, before seeking to particularise these exigencies, it is necessary to prove that the defences in question are really what they are assumed to be.

1 These sculptures were presente Nationae th o dt l Museu f Antiquitiemo Scotf so -

lan Georgr 185n Si di y 7b e Clerk, Bart. 2 Clerk's Memoirs of My Life, pp. 138-140 and p. 222. Pennant's Tour in Scotland, vol. iii. pp. 406-410 ; Wilson's Ar&axlogy and 3 Prehist. Annals of Scotland, vol. ii. pp. 68-71. •304 • ' PKOCEEDINGS OF THE ^SOCIETY. " '.'

For, whatever purpose the multiplication of ramparts was intended to serve, it was an expedient frequently resorted .to by the builders of n hild othean low r fortsou . '• These hill fort'peculiaritied ha s f theiso r own. The ramparts • rose one above another round the summit of the hill; the spaces between them were less uniform in breadth than the representations of Birrens given by Gordon and Roy show its lines to have been; and the shape of the whole entrenchment was oval or cir- •cular instead of rectilinear. There were probably: differences of even. •greater importance e principlth e t doubleth Bu f e.o , triple quadruplf o , e agge ditcr o r theree s hhili Th l . for f Barniekyno t Aherdeenshirn i e e may be cited as an example.1 . In different parts of the country 'are small square "redoubts," surrounde doubly db e rampart trenchesd an s . On f theeo m describe r ChristisoD y b d n is situate e downwarth n do d slop f Laverhao e y Heigh n Wamphrai t y Glene spo s Th desolati t . e enough now, but in its neighbourhood there was, three or four cen- turies ago considerabla , e population. A-rill wit hScandinaviana name flows pas , withiyards it norttw it fe f sho na side. Jn appearancs i t ei a miniature Birrens, with fewer mounds and ditches but, as Birrens seems havo t Ches e a bee d wn an Green and Ardoch certainly were, "directly connected, r Christison'D e us o t " s words, " wit n extensiva h e system of rectilinear works.'1" Triple rampart d ditchesan s , drawn acrose th s headland, with a gate near the centre, defended the only approach e forth t o t that once occupied e extremitth , f Burgheao y 'thn o de Moray Firth—known to have been held first by the natives and 8 afterwards by the'Norsemen. A similar series of mounds with interr .vening fosses had been thrown up outside the peninsular fort ,of Dunbeg in the county of Donegal.4 There must have been • conditions in the warfar f otheo e r days, common probabl o moryt e thae raceon n , under s thoughwhicwa t i h t thae assaulth t f ao nt enem a-strongholn o y d could be most, successfully resisted by adding to the' number pf its

• Anhceologia Scotica, vol. ii. pp. 324-327. l 2 Proc. Soc. o/Antiq. of Scotland, vol. xxv. .pp. 228-229; Koy, Military Antiquities, Plates XXIV., XXII., XX.. 3 Proc. Soc.'of Antiq.-of Si-otland, vol. iv. 321-369>pp . 4 Archaeological Journal, vol. xv. pp. .2-5. • _ • • '--- • "KOMAN ROADS" ON OKDNANCE MAP OF SCOTLAND. 305

lines of defence. Further research may, therefore, yet furnish an answe e questioth o t r n sometimes put f , I Birren t Romanno s i s , what is it 1 Nor ought we to lay undue stress on the finding of Roman antiquities there. "Wit e exceptioth h f thoso n e onc t Penicuikea e th , exact spot where the sculptures were found has unfortunately not been noted. Sir John Clerk's, however, were certainly located out- side the mounds and ditches. What connection, if any, they ever witd presene ha hth t fort determinede cannoh w no t . Roy, aware that they were not found inside the ramparts, thought he saw traces of other mounds within which they had been placed, and which he calls a Procestrium or addition to the fort. But, instead of being of the same age as the fort, they may have been cleared away from their original position to make room for it, and afterwards protected as Sir John Clerk suggests. For, according to another marginal note to the Memoirs, the building in which they stood was not of Roman con- struction I " doub. t not, e "Baron"th " remarks, "but some great me n e i (were?Englannar o Jwh d lover f antiquito s yr fa hav o s e reverenced the heathen religion as to have built a temple for the sake of this statue. It is certain, as Tacitus writes in his Agricola, that the Brigantes were the most powerful nation in Britain, and that with probability this statue was like another Diana of Ephesus." * The Romans y suppose, ma and e e Romaniseth ,w , d natives after their 1 Memoirs o f addee myb y dLife, . 157ma p tha r Joht I Si .t n Clerk's references to the exact spot where the sculptures were found are not altogether consistent. In shora t notic themf eo writte,e saib o dt himy nb , whic gives i h n in " Additiond san Corrections " to the Itinerariv/m (1732), we read :—"These three stones were found last July in an ancient fortification or camp call'd The Birrens." Here there is no s Latimentiohi n a templei f t dissertationno Bu . , writte n 1743ni , thougt hno printed till 1750 have ,w a fullee r descriptio t i tha f o nn e margioe nth th f no Memoirs, with the possibility of the temple having been a Roman shrine put forth n alternative:—"Anna s a o MDCCXXXI n provincii , a Annaudiana, apud Scoto- Britannos detectae fuerunt antiqui aedificii ruinae, prope castra ea Romana stativa ex lapide quadrato, quae in parochia Middelbea conspichmtur. Ex forma ejus brcvi apparuit, templu delubrau mse m Romanu mfuissed i , dec Mercuri t deaoe e Brigautiae consecratum : nain in ejus area disjecta fracta jacebant monumenta, de quibus mine disserendum estlapidx E . e quadrato constructum erat, longum trigint pedesx ase , latum circiter duodecim ; hau d inelegaus insignioribud se , s arohitecturae ornamentis vacuum locn I o. paulnlum humido situin erat extra castrorum munimenta, quasi VOL. XXVIII. U 306 PEOCEEDINGS CM? THE SOCIETY.

example or on their advice, selected the position of their stations with such care that succeeding peoples would readily fix their settlements e samth en i localities o BomaT . n camps (castra) e mostsomth f o e important towns of the of to-day owe, if not their origin, at least the beginning of their greatness. It seems to follow that, in the present state of our knowledge, it is d nationalitan e e unsafth ag f conclusioy e y o com o an th t e o o et t s na builders of the present Birrens fort, or its connection with the antiquities that have heen found in it or near it. What a careful and thorough excavatio e wholth f eo n site migh n sayca t. e revealUnlesson o n , , however, this coul e carriedb d t unde.ou r proper supervision, Birrens had better be left as it is. Unskilful hands might destroy the mounds, and leav problee eth mt presenti s unsolved. It is natural to suppose that if the Romans . were established at Birrens, they would avail themselves, for the purpose of observation, of vere th y remarkable hil f Birrenswarko lt higheass i no f t yo i s , i t I . greao n t ascentta distances i t t commandi i , d an , extensivn sa e prospect of the country in all directions. In the event of their having done so, there was probably a road between the two places, of which one or bite f botroath so f n Hoddadhi o d Timdergartman h markee th n o d e remainsO.Mth t s e a Romanleasa . b r y theo y t, ma yoccupma , y the track. But no clear evidence of a Roman occupation exists here, e cas f th r o Birrensd e thiothefo n an i s d rs a an reason; e th s formation of the road by the Romans becomes doubtful. The two rectilinear in closures constitute the chief claim of Birrenswark to be the site of a Roman camp. It has, however, yet to be shown tha beliefe th t prevaleno s , t since Gordon wrote, which ascribe l thessal e and similar works to the Romans, has any real foundation.1 Last July

nullis humanis praesiiiiis egeret, sed deorum Romanornm curae atque providentiae committeretur." (Dissertatio de Monumentis quibusdame Th . Romanis,6) , 5 . pp marginal Notes were probably written later than the Latin dissertation. Bishop Pocoke, who visited Birrenswark in 1750, remarks :—"The people say tha1t (the soutcampe mads th ho Kinwa y t ) e b g Charle armys sthehi d an ,y certainly encamd di it.n po " Following Gordo Horsleyd an n adde h , s that doubthero n s ei t thoy are both Roman works, and supposed to be the castra aestiva of Blatum Bulgium. Z'ours Scotland,n i . 7 . p " EOMAN EOAD OEDNANCSN SCOTLAND"F O O 7 P 30 MA E .

General Pitt-Rivers communicated to the Wiltshire Archasological and Natural History Society an account of the excavation of a squarish, intrenchmen Rushmorat t e Park side Its somewha, sare t irregularand , not in straight lines " as," he remarks, " a Roman camp would probably s it s t havsidesn openinit a Bu f es o . been.ha e g t on I onl "n i y general resemblance to some of the camps in North Britain that have been set down as Roman is striking enough. In the course of the excavation numerou se Bronzrelicth were f so eAg e found, which, from their position, proved beyond question that this camp belonge thao dt t period. There were also Roman relics, especially fragments of pottery, mostly Romano-British o othen t r Bu inferenc. e coul drawe b d n from their position relative Britisth o t eh relics than thae camth t p had, subsequen s constructionit o t , been occupie Romane timth a r y eb dfo s or, more probably, frequente Romanisey b d d Britons.1 Col. Lane-Fox, Roovesmor e papes th hi n n o ri e For r Ratho t , County Cork, remarks:— " Besides these circular raths, some of which have double and triple banks and ditches, oval intrenchments of the same kind are sometimes, though rarely centrae th , southerd foundn i an l d an ;n part Munsterf so , rectilineal works of nearly the same superficial area are distributed in belts ove mose th r t fertile country.e partth f so " 2 e fortsa papeth n n ,I o r camp d motean sf Lanarkshireo s , read before this Society some years ago r Christiso,D stateds ha n , brieflt ybu clearly, the difficulties that surround the subject.3 It has been said, an doubo dn t wit hconsiderabla e degre f trutho e , thae rectilineath t r work primitiva f so e generally racma e distinguishee yb d from thosf o e one more advanced by the regularity or irregularity of their lines, the former having been laid down by the eye, the latter by the measuring- r lino e d usero r thi fo de s th purposs wa t e e th rodBu onl y .b y Romans ? Besides, \ve must make sure that the plans from which we for r opinioe campmou th e accuratf ar o ns e before accepting thes ma sufficient evidence. The necessity of this in the case of Birrenswark is inclosureo showa comparisotw y b s ne planit th f s o f sgiveo n y b n

1 Wiltshire Arclueol. and Nat. Hist. Magazine, vol. xxvii. p. 206. Archceological Journal, vol. xxiv. p. 139. 2 3 Proc. Soc. of Antiq. of Scotland, vol. xxiv. p. 293. 8 30 PROCEEDING E SOCIETYTH F O S .

'Gordo d Boan ny respectively. The e representear y y Gordonb d s a , shown on a somewhat reduced scale in Plate III. But in Boy's plate they assum e for eth whic n mi h they appear slightl,a alsn o y reduced scale, in Plate IV. No one would suppose on first looking at these two Plates that the same set of earth-works was delineated on each of them. "The geometrical precision of the one set contrasts strongly with the marked irregularities of the other. Gordon seems to have thought himself entitled to represent " Boman camps " not as he saw them, but as he thought thebeen,d y ha r ough o havo t t e been, where newly made. Nor was he singular in this respect among the antiquaries of last century. r frofa ms faultlesswa Eve y nRo . It must not be forgotten, when we are investigating the origin of such remain e sideseee stile b th Birrenswarf t Birrennar o sa n o t l o s sa d san k Hill, that we know far too little about the military methods and social arrangements of the different peoples that inhabited the country before the Norman Conquest, and even for centuries later, to entitle us to ignore their presence in the country. All inclosures of the kind were, from some poin f viewo t , defensiv thein ei r object somd an ,f the eo m wer doubo en t camps—it may be, Roman camps. But there are grounds for believing that in others there had stood the dwelling of the petty lord of the dis- trict, built perhaps of wood, while the huts of his chief dependents were withi e samth n e ramparts. Tha e Romanth t s protected their tem- porar wels ya theis a l r stationary camps wit a hvallua ditch d s ma an , rectilinear in form, as the nature of the ground would admit, is no proof that this particular mod fortificatiof eo peculias nwa themo t r . Besides, it bas just been pointed out that camps of this form are found which cannot be the work of the Romans, whose example, however, would- be followed by those with whom they came into hostile contact. Caesar inform s thau s t e tribesomth f f Gauseo o l that took p armu s against him began after a time to fortify their camps in imitation of the Romans. It is but reasonable to suppose that other peoples to whom the Roman mode of warfare had become familiar would act as the Gauls did. Some other evidence, therefore, than the shape of an ancient earth-work is necessary witbeforn ca h t i ecertaint classee yb Romans da beford an , t ei Proc. Soc.Antiq. Scot. Vol.XXVIII. Plate III.

GORDONS PLAF O N THE "ROMAN CAMP" OF BIRRENSWARK. Plate IV.

ROY'S PLA F BIRRENSWARO N K HIL N ANNANDALI L E WITH "ROMAN S &P 9 M "CA "ROMAN ROADS" ON ORDNANCE MAP OF SCOTLAND. 309

hele b prov o dt n ca e tha a roaworte th f d k s o pasi r leadin o , , it t it o gt Romanse th . Whe considee nw r that Lower Annandal possesn i s ewa - sio f Gaelo n , Roman, Scot, Saxo d Danan n e successively surpriso n , e need be felt if it affords examples of more than one variety of ancient camp or fort. Whether we shall ever be able, with the aid of the data now left us, to distinguish the intrenchments of different nationalities or periods, it is hard to say. But if we cannot untie the knot, we should . it t nocu t While, however, so far as the evidence goes, the present Birrenswark " camps" may or may not be Roman, it is conceivable, as has been already observed, that the Romans may have used this prominent hill as a post of observation in connection with Birrens. If they did, it woul e ras db o affirt h o piecemtw f roa thao se dth t running through the parishes of Hoddam and Tundergarth—the one for about a mile east, and the other for two miles west of it—were not roads or track-ways at the time of the Roman occupation. Little appearance of a road, however, is now to be seen where they are marked on the O.M. But recently, beneath a few inches of turf on the track of the western one, nea e far th f rCoursteinm o a s roaexposed, wa d 8 fee1 , t wide, with a rise in the centre of about 6 inches. Small flattish stones lay embedded in its surface, which had almost no other appearance of having been artificially formed. These stones were more carefully laid on the level sides than in the centre. There were neither kerb-stones, ditches nor mounds. The place where it crosses the river Milk used to be knowe Drovth s a ne Ford1—an appellation that indicatedy ma e w , e sure b e chieth , f e timpurposon e r t whica fo e e road th hha d been used. In neither of the Old Statistical Accounts of the two d accounol e f Hoddamo th tparishes n i r s theri ,referency no , an e e Romaa o t more nth road en I f Tundergarthreceno . e on t , howevere w , read of one, "discovered a few years ago," of which the following account is given:—"The road was covered with about 9 inches of earth. It is formed of broad, flat stones, well packed together, with watee interstices sand th s an n raboui i d t I fee8 t . t wide. f thiI " 2 s 1 Singer's General View of Agriculture Countye (h n i of Dumfries, . 639p . 2 Kew Stat. Ace. of Dumfriesshire, . 197p . 310- PROCEEDINGS OF THE SOCIETY.

refers to the road of the O.M., their difference in width at two places, which must have been near one another, is not easily accounted for. The fourth and smallest portion now left of the Lower Annandale Eoad goe frof e Carlisloi s mth Glasgod ean w Eoa milenorto o s dtw r so h of Lockerbie, and leads up to and past the manse of Dryfesdale. It is in daily use, and in no way differs from an old parish road. A short distanc e Torwooth e esitf th o esouth-wes s di Moot i f ro t "camp," mentioned above, whic y considereEo h e Eomab connected o t dan n d wit roade hth v dissertatioa n I anciene th n no t syste Eomaf mo n castrametation, Eoy, chiefle authoritth n yo f Polybiusyo , explain e earliesth r forEomaa f mo n temporary camp and afterwards identifies as such this Torwood Moor inclosure and a number of other earth-works, more or less entire in his day, whic e describeh d figuresan s . Mos f theseo t e thinksh , e ar , Eoman camps which were actually occupied by Agricola during his campaigns in North Britain. This conclusion he reaches on the ground that a different system of castrametation, generally known • as the Eyginian, is supposed to have come into use among the Eomans shortly after Agricola's day. Polybian camps in Scotland, therefore, which he asserts are of two kinds, a greater and a smaller, must mar course kth f Agricolao e . Three smalleth f f themo o e rl al kind, , e Weste founth h n i ,d betwee e Antonine Bordeth th nd an r o Wall —Torwood and Tassiesholm in Annandale, and Cleghorn in Lanark- shire. Of Torwood he thus writes :—" The first camp we meet with on the Western coast (besides thos t Birrenswara e k Hill, which, beina f o g different construction, seem not to have been the work of Agricola) is situated on Torwood Moor, near Lockerby, in Annandale. The greatest part of one side, with its two complete gates, and likewise a part of each end, remain entire." This camp Eoy estimates would contain an army of from ten to eleven thousand men.1 Its gates had been protected by " straight traverses " instead of quasi-circular mounds as at Birrenswark, s forapparentls it md wa an y somewhat irregular Eomao N . n antiquities 1 Mil. Antiq. vfthe Romans in Britain, p. 61. Proc. Soc.Antitj. Scot. Vol.XXVIII. Plate V.

ROY'S PLA "AGRICOLA'F NO S CAMP"ON TORWOOD MOOR.°"A"""°" " ROMAN EOADS " ON ORDNANCE MAP OF SCOTLAND. 311

have been found near it. It was entirely on account of its shape that it was supposed to be Eoman. Eoy's plan of it is reproduced, on a smaller scale, on Plate V. Statisticald AccordinOl e th o gt Account, n thi i ther se ar epar f o t Dryfesdale " evident remains of eight camps or forts, some square or Eoman, others circular or British. . . . The most remarkable are two —the one British and the other Eoman—facing each other, and separated by a narrow morass. They are built on two hills east of the village of Bengali, the name alluding to the forts in the hills, signifying the Hill of the Gauls.1 Old pieces of armour and warlike weapons have frequently been found in them, and not many years since the skeleton e intervenin founs th a cair wa n n i o dnn ma oga f morass, thougho t t have been laid there for some ages, and some fragments of his dress or accoutrements wer ea curiosity carries a f of d , particularl s sandalsyhi , curiously wrought bound an , d aroun e mouthdth s with leathern thongs, were conveyed into Englan Oxfore th r dfo d Museum. Ther anothers ei , a Eoman fort, beautifully situated upon a large eminence, in the centre of the united and extensive holm of Dryfe and Annan, called the Gallaberry e burg th e e d ter Gaulsstatior th th an ho m ,r f o fo n; f Saxoo berry s i n origin, signifyin ' burghg , mansion r strength.o , ' The most entire is a British one at Dryfesdale gate, occupying about two acre f groundo s , commandin a mosg t extensive prospects it : counterpart is a large Eoman one, about half a mile due east, inter- ruptea moor y b d, bein e placth g e celebrate e bloodth r yfo d battle betwee e armth nf Juliuo y s e Agricolforceth f d Corbreduo s an a s

vers i t 1yI doubtfu e Romanth f i l s were ever spoke thy b e f nativeno f Nortso h Britain as Gauls. But, as Dr Anderson informs me, the Norwegian aud Danish invader Vikiu'e th f so g period (i.e., eighte elevente th th o ht h century generalle ar ) y referred to by the Irish chroniclers as "The Gaill." Now the chief centre of Norwegian-Danish place-names in Scotland is in Dumfriesshire. The Scandinavians, migrating from Cumberland, where at one time their numbers were considerable, crosse e Liddeth d Esk d an l , spreading themselve e Niths th westward s .a r fa s a s Everywhere throughout this district evidence is still borne to the fact by such names as Middleby, Applegarth, Tinwald, Torbeck, Thorn iewhat, &e. Bengali and Gallaberr campe hole swely th th Dryfesdallistf e md yo n ma si th l an , e are, perhaps, other memorial same th f eso perio Annandalf do e history . Worsaae'Cp . s Danesd an Norwegians England,n i Scotland ,d an . 203p . 312 PROCEEDINGS OF THE SOCIETY.

Galdus, the twenty-first king of the Scots, about the end of the first century."1 This extract shows on what slender grounds certain inclosures and forts were in past days classified either as British or as Eoman camps, as the case might be. No search-was made for antiquities near or within their lines. The absence of these might have been deemed immaterial had any satisfactory evidence of another kind been offered. But nothing of the sort was adduced. -Thus an inclosure supposed, without further inquiry, to be a Eoman camp, was held to vouch for e origia piecth f f disuseo no e d road nea . it rSometimes , againe th , proces reverseds road e existencol swa th d n d vouchea e an ,f th o e r dfo Koman origin of a" camp." 2. The Upper Annandale Road.—From Gallaberry Hill the principal " Eoman road " goes northwards, keeping along the east side of the Annan through Applegarth A Eoma" . n road,w Ne " saye e writeth th s f o r Statistical Account,." also traverse e paris northerlth a s n i h y direction: tracee b y ninn di ma e t i place severar sfo l hundred yards abov levee eth l of the adjoining ground." The road then enters "Wamphray, in the Account f whico e reaw hd :—" There appear o havt s e bee Eomaa n n mile5 r o s 4 roathroug r dfo h this parish, from nort o southt h , neao t r d roaol d e froth m Glasgo o Carlislewt , whic stils hi turnpika l e road, :d wherean n somi , e places, pavement stone e occasionallar s, up g du y especiall e loweth n ryo grounds, where therfea we ear houses , forminga small village, called Gilga r Newtono l . Along this line, too severae ,ar l large stones, about 5 or 6 feet in height, fixed firmly and perpendicularly in the ground, and supposed to have been placed there by the Eomans, probably for measuring distances. Several traces of camps, supposed by some to be Eoman, by others British, are also visible in this parish." Little evidence of a positive nature can be said to be contained in these extracts; but s nothin,a g stronge s beeha r n brought forwarn i d favou f thio r s portioe roath df o nbein g Eoman s beeha nt i , thought faiquoto t r e theml tha e Wamphrawritee Al th th t. f o r y notice feels himself warranted to say is, that "there appears " to have been a Eoman Stat.d 1Ol Ace, of Scotland, . 425 volalsp e . oSe .ix New Stat. Ace. of Dryfesdale, p. 453. "ROMAN EOADS" ON ORDNANCE MAP OP SCOTLAND. 313

road there. The reference to the " milestones," which is taken from Statisticald Ol e th Account, t giveno s ni quite accurately. According to the earlier writer, they are " nearly at the distance of a Scotch mile froe anothermon d thereforan , e suppose somy b d havo et e been mile- stones." Very likel s successo ye hi incongruit th w sa r f makinyo e gth Romans measur f theio e distance th y er an road n y "lano sb s g Scots mile thoughsd "an ; t judiciousi t , when expressin t verno yga confident belief that they were placed there by that people, to avoid weakening s cashi e f stilo o l furthetw r r o the statiny e fa aparte b ryar w On gho . these stones I believe, , still remain. Crossing "Wamphray Water d afterwardan , e Annanth s t "Burna , - foo f Kirkpatrico t k " accordin Boyo t groae th , d proceeds thence along the west branch of the latter river and passes at Tassiesholm, on the farm of Milton, what Eoy took for the second of Agricola's camps. He writes , howeverit f o , , with considerable doubt o gat. n es " A exists," he says, " and only a very small part of the intrenchment, the vestiges mus consideree b t sligho to enabls o t dta pronounco t s eu e absoluteln yo the head." Besides the only angle of the inclosure that remained, Roy shows in his plate, quite close to it, " a square redoubt." No e anglremainsw tracth no f t ei eveo e f i r, existed e " redoubt,Th . " Dr Christison was informed, is known locally as the " moteknowe." s bee ha levelleo s n t I ploughiny b d g that, thoug s outlinee hit b n ca s distinctly traced, the class of works to which it belonged cannot be e noteb identifiedy d ma tha t a similaI t . r " redoubt s showi " n o n Roy's plate nea Torwooe th r d Moo seee Platb n rno y . campeV ma s a , Dr Christison adds :— ' Roma "A n road markes i ' de eas closth t o t e of it on the O.M., but I could see no trace of it. Roy marks his Roma e Annane banth e nth edg f th f o kroa o e n , o dwher e thers i e now a farm road." This difference between authorities as to the exact course of our Roman roads is by no means uncommon. Statisticald Ol e e writeth Th f o Accountr states tha e madh t e some excavations at Tassiesholm, or Tatius-holm, as he writes it. " Upon digging, I could only find some earthenware in fragments, very strong and coarse. I have- procured from the people who have tilled the adjoining field bitsw f greeafe so party-coloured an n d glass smald an , l 4 31 PROCEEDING SOCIETYE TH F O S .

piecea substanc f o s e resembling marble, about haln inca f h 'broad,; round, smooth, and flat on one side."1 An examination of the place so incomplete cannot be held as conclusive; but, as far as' it goes, it -tends to show that the holm was the site of a native settlement rather than of a Roman camp. The belief, however, that it was a " camp," and that a Eoman road ran near it, is prevalent in the locality; and one of the

proofs by which it is attempted to support the popular view is: amusing. A little to the west are a few houses, often called collectively Hop- pertuttie d thian s; has. been gravely supposea corruptio e b o t df o n oppidum Tatius,d an "th Agricola'f o e e namon f eo s generals" otherwise unknown to fame ! Shortly after crossin e Evagth n the road ascend e Coateth s s Hill, and soon becomes visible. Its course may be.traced by being in some places slightly raised above the surface of the ground, and, as the season advances lightee th y b ,r .herbagcoloue th surfaces it f o rn t eo I . runs all along the ridge' between the two rivers, not exactly on the summit shora dow y Annat e th wa tbu n, n frono sidemt s i trac It .s ki almost straight till it enters Lanarkshire, there being no obstacle in its y deviationo reasoan n r wayd fo nan , . Wit e exceptioth he th f o n Chapel farm, wher t i disappearse e grounth o par f n s undei o ,d t r cultivation. Trenches were cut across it last summer on several parts of the ridge. These showed tha e modth t f constructioeo r frofa ms uniformnwa d an , precluded the idea that it had been made according to any fixed plan. It was first opened on the Chapel Hill, a short distance north from the cross roa Evao dt nture Waterth f n beinO . g removed surface th , s wa e found to be slightly raised in the centre, with a slope to each side, "and coveree b o t d wit hlayea f smalo r l stone inches4 s deep. Beneath this was another layer of larger stones, 11 inches deep, mixed with and f clay o e largesd th , restinbe a t n stonego s bottome beinth t ga . There was no appearance of pavement, or even of a causeway; but on the west side, for a breadth of 6 or 7 feet, were large stones, placed with their flat sides uppermost, and large interstices between them. They spoiled the original contour of the roadway, and were probably an after^ 1 Old Slat. Ace. of Scotland, vol. iv. p. 522. "ROMAN EOADS ORDNANCN O "F SCOTLAND O 5 P 31 MA E .

thought, laid down to form a kind of pathway for the use of passengers t weatherwe n vacane i th , t spaces having been fille n witi d h eartr o h gravel. There was also a single row of large stones near the edge of the east side, but not placed in close or regular order. All the stones were undressed, and had teen got near the spot. The breadth of this part of the road was 21 feet, and its depth, as far as the bed of clay, 17 inches in the centre, and 12 to 14 at the sides. Mr John T. Johnstone f muco d tooks h, an a wa greaMoffat servic e o m t wh o ,t e interes excavationse th n i t f opinioo s ,i n tha foundatioe th t f clao yd nbe had been brought to the spot from some distance. He believes it to differ from the ordinary till which covers the ridge, and to be such as is found beneat shalloe hth w layer neighbourhood e mosf so th n si t a s wa I . first disposed to think the same. It was difficult, however, to see what object coul e gaineb dn distanc y o e claygoiny an th b d r o an t g; efo cutting into the hill close by, clay of a similar light colour was found in abundance I suspect. , therefore, bottoe tha laye e situ.th n i th tt a ms ri n otheI r cuttings mad t shorea t distancee s on nort e soutd th h an f ho now described appearancee th , s were similarroae shallowes th d wa t bu , r and less material had been used in its construction. Here and there were some very large stones embedded in it, almost boulders. It was also laid open near the footpath over Coates Hill, a mile south of the Chapel Hill cutting t Meikleholmsida d an , e Hill northe mileo th ,tw o st . In the first of these cuttings there was at the bottom, instead of clay, a layer of moss, the stones throughout were smaller and there was no clay mixed with them thin I s. plac roae onls eth d ywa 15 feet wide. There were some stones that might have been e mean footpatha th r n o fo t t bu , east instead of the west side. At Meikleholmside it was of the same widt t Chapea s ha l Hill, without, however, stone eithen so s r wa side t I . quite flat on the surface. Here the stones in the roadway were mostly small ,botto e thosth t ea m being mixed with l alonclay Al s coursg.it e smal f whic seeno e b t l y pithou , materialsma d beeha s n take mako nt e or repair the road. "We appea havo t r e very early mentio f roado n n Annandalei s r M . George Neilso a lecture n i n 1 delivere a fed w year o beforag e s th e 1 Annandale under the Unices,. 7 . p S16 ' PROCEEDINGS OF THE SOCIETY.

Glasgow Annandale Association, directed attention to a charter of David I., stil n existencei l , granting the land f Anuandalo s e ("the valley of Anant"), in forest, to Eobert de Brus, sometime after 1124, and containing a clause that forbade any one else to hunt in or go throug e saidth h , forest, "e straigh excepth y b tt road appointed": Et ne ullus eat per predictum forestum nisi recta via nominata.1 Mr Neilson inform e tham s t recta via—a phras eo mean n her y b se free of difficulties—is found elsewhere in mediaeval writings in the indubit- able sense of an actual road, and that he is strongly inclined to believe that ther s littli e e difference betwee "d highway.an t i n anothen I " r charte rea e e grea rf w "(magna,y dth o twa via) which goes from Blawath to the Church of Eeinpatrick " (in Gretna parish, Lower Annandale).2 , 3. The Nithsdale Road.—One portion of this road appears on the Map a little to the north-west of Lochmaben, and another in Tinwald parish, not far from Amisfield. Near the former is Wood Castle, a circular fort "y , Roman.calleEo y b d " "But," remark r Christison,sD . who examine t i dcarefully , "although unusually massive t i doet , no s see diffemo t r essentially fro e ordinarmth ye country. fortth f o s " 3 Both portions of road have the appearance of old parish roads and are still partially open. roae Th d next appearAYele th s le a nortsPat th f Durisdeer,n hho i , intermediate th es cours it par f o t e being e lef conjectureo th t ts a r fa o s , O.M concerneds i . Wele Th l. Pat vers hi y much wha s namit t e implies —a narrow roadwa f vero y y primitive construction overgroww no , n with gras passabld san e onl footn yo . Cairne Th . 4 Valley Road.—The small piec f "Romaneo " road that enters the north-west of Dumfriesshire . from Kirkcudbright and is unknow otheo nt r map f Romao s n Britaie s nplacth it musn e i e ow t O.M. to the communication made to George Chalmers by Joseph Train, who represente a Romad n roa s "a d traversin g Dumfriesshire through the vale of the Cairn .Water, by Conrig, to the top of Glencairn."4

National MSS. of Scotland, vol. i., Plate xx. 1 2 Lib. de Metros, pp. 668-9. 3 Proc. Son. of Antiq. of Scotland, vol. xxv . 252.p . 4 Caledonia, vol. iii. p. 448. "ROMAN EOAD3" ON ORDNANCE MAP OF SCOTLAND. 317

r SingerD speak a roa f do s that turne de wes th " oft o t ffro e th m Mthsdale road, crossin e Nitd passinth gan h g through Tynroy b n Scarvvater " and thence doubtless to . Being accepted by some local historians, such a "Eoniau" road became part of the popular beliefe situatioTh . f thino s isolated fragmen loftys i t , exposed and bleak, overlookin sourcee Kene g th Skarre th f th , s o Shinnee th , d an l the Dalwhat. It is a path rather than a road. The till that covers the hill top has been levelled and slightly hollowed for a width of about feet8 . Some eart stoned han s have bee r elso ne, it afterwardlai n do s use repairsr ds difficulfo i wha e t se I .t o t purpost t coulei d have served, though it must have had one. Within sight of it the positions of similar roads were pointed out by an intelligent shepherd who knew the ground well, some of them, he said, much better made. He thought he could tracmosw f theeho o t frod mle m these lonely wilds int plaine oth s below. But where the one we stood on came from, or whither it went, he was unabl sayo et . t unlikelno s i t yI tha e roaone th tth e d e spokeb Trai y y b f nma o n mentione a charte n i d r (temp. Beg. Alex. II.), whico t attentioy m h n has been drawn by Mr Neilson. It is described as the "King's high- way " (regia via) leading from Dercongal (i.e., Holywood, near Dumfries) oGlencairo nt d thencnan passage "o et th e acros marse sth Atheuf ho - weran."1 The conjectural Eskdale "Eoman" Road of the Monument a map e sak th f maycompleteness eo r fo , alse b , o mentioned sups i here t -I . pose havo dt d fro le em Netherb r somyfo e distance e righupth t bank of the Esk, and then to have gone northwards to Trimontium (Eildon Hills). Another branch of it is said to have continued its course o Castlt k eEs alon O'ed Raeburnfoote an th gr . Accordine th o t g New Statistical Account, the manse of Kirkmichael occupies the site oa castelluf m s thaapproachewa t o a Eoma tw y b dr o n e roadOn . other e noticear s y varioub d s writers r SingeD . r states r thaSi t John Clerk took great pains to trace all the supposed Eoman roads in Dumfriesshire. Apart froe Birrenth m s sculptures, almos o undoubten t d Eoman 1 Lib. de Mclros, p. 183. 318 PKOCEEDINGS OF THE SOCIETY. antiquities have been found within Dumfriesshire, excepcoinsw fe a t; for urns classed as " Roman" by the older authorities are certainly British. The Roman origin of the " camps" and roads north of Birrens is thus left to rest almost entirely on any evidence that can be supplie themselvesy db . Believing tha a Polybiat n camp f founi , n Norti d h Britain, muse b t ascribed to Agricola, and a Hyginian to a later date, Roy saw the importance of endeavouring to ascertain the particular system according to whic hr " Romaeacou f ho n camp beesd "ha n planned. "Warrantable inferences might then be drawn as to the history of the, Roman occupa- tion. Accordingly, after givin elaboratn ga e accoune differenceth f o t s between the two systems, he proceeds to classify the camps under one or other of them. The. large majority he proves to his own satisfaction .te Polybianob d thereforan , e wor th ef Agricola o k s reasoninHi . s gi fair, with much appearanc f beino e g scientifice enteres h hi t n Bu do . investigations wit a hmin d greatly prejudice n favouri d f Melville'o . s assumption t musti sd ;I ,an lthink , strik unbiassey an e d reader than i t his hands irregularities accommodate themselves too readily to the normal Polybian type. Yet it is chiefly on their agreement in shape and size wit a Polybiah n cams base ha e Romap th dy tha Ro tn o origis f o n -many of our northern earthworks. Roy himself, after he had put the finishing touches to his Military Antiquities, seems to have had his suspicions aroused as to the sound- ness of some of his arguments. This we may gather from the second of supplementaro thetw yLI.d platean ) t . attache s(L no wors t hi kbu o dt firse texte f thesreferreo Th th t .n ei (Plato dt ) representeL. scama p at Rae-Dykes of TJry, near Stonehaven. In his account of Agricola's cam- paigns, Roy, by means of a supposed series of Polybian camps, which he point wit o t utmose s th h t confidenc s havina e g containe troope th d f so -Agricola, trace e coursth s f thaeo t genera r nortfa s Strathmors ha a l n ei Forfarshire e decisivTh . e battle betwee e e RomanCaleth th n d - an s donian believee sh havo st e been fought " about Fettercairn, Montboddo, or still nearer Stonehaven." d expresseAfteha e h r d this opinione h , received from some quarter the plan of Rae-Dykes, which, in spite of its

Preliminare Se y Remarks . 48-5pp , 1 (supra). 1 "KOMAN EOADS" ON OEBNANCE MAP OF SCOTLAND. SI 9

great irregularity, he had no difficulty in including among the number of Agricola's camps. Subsequently, tha f Re-Dykeo t r Glenmaileno s , nea Ythane e sourcth th r f o e, reache interpretatioo n dn o him. s A 1 n of Tacitus' narrative could Agricola hav er north fa beemusy o s nRo , t have been perplexed representatioA . s givei t n Plati o n f o ne LI.; but, though it is in every way nearer the form of a true Polybian camp tha nt assigne thano t Urys a ti Agricolo t t di , r eveo a n called Roman. In some letterpress printed on the Plate, the gates are said to be " covered with traverse e samth en i s manne e campth s a sr whiche ar supposed Romane e soutb th o t n he Grampians.o th sid f o e " Either s leda fina s wa a , l y conclusionRo o entertait , n doubt o Torwoot s a s d Moo d othean r r earthworks being Roman after all , moror , e probably, s insertee e EditoMilitaryth wa th Plat f . y o b drLI e Antiquities on responsibilityn ow s hi also owh , marke e cams th a dn Roy' o p ma s Oast. Rom., not Agricolae. Those who may now entertain similar doubts can hardly be regarded as showing a groundless distrust of Roy's conclusions r eveo , ndispositioa looo nt criticao t thea kto mn i l a light.2 Simpl e morth o refe t yl e al northerr f theso n e "campsa o t " subsequent e Romaf perioonth o ef d o dnri t occupatioge o t s i n

"We distinctly learn this from Sinclair's Statistical Account f Scotland,o vol. 1 xii. p. 315, note, where it is said he at first put it down on his map as Castra Agricolae. 2 Sinc above writtencopieMilitaryes th e o th etw wa f se o th , Antiquities. MS lef n i t by General Roy have been obligingly examined at my request—that in Burlington House by Mr "W. H. St John Hope, and that in the King's Library by Mr W. de Gray Birch. In neither of them is there an original of Plate LI. The Hon. Secretary Perte oth f h Literar Antiquariad yan n Society also inform thae sm t Colonel Shand's archiveGlenmailee e papeth th n n o i r tha f sw o tno n Society t camno s t pseemi I . s never to have been returned by Whitaker, to whom it had been sent for perusal (Proc. Soc. f Antiq.o f Scotland,o vol. vii . 29) p .e drawin Th . f thgo e camd pan particulare th s Plate giveth en n o wer e doubtless furnishes wa e Shandy h db t Bu . evidently too proud of his discovery to insinuate the doubt as to its being a Roman camp, expressed in the words I have italicised. The insertion of this doubt must be due to the Editor or Editors of Roy's posthumous work, who had clearly seen that the existence of this "camp " so far north all but upset the reasoning by which Roy sought to identify the "camps" north of the Grampians with Agricola and Romanse th . 320 PEOCEEDINGS OF THE SOCIETY..

difficulty by substituting another for it. In order to satisfy the pre^ sumed requirements of Eoman history they have .been transferred to Urbicus or Severus, and consequently to the later instead of the earlier syste Romaf mo n casfcrametation, without any referenc e considerath o t e - e conclusiotion'ty th o Eo s that d onlle no tn y that theye th mus e b t work of Agricola but that they are Eoman and not native fieldworks. The matter, be the transference right or wrong, can hardly he settled n thai t off-hand fashion. Nor ought Torwood e separateMooh o t r d from Gleiimailen, as a comparison of Eoy's Plates will show. In their characteristice identicalar o tw .s e othehi th Theys l ral wels a .s a l " camps " and " stations " may yet be proved to be Eoman. But the review already given of the evidence hitherto advanced in favour of this opinion appear o indicatt s e very clearly that more convincing evidence and further investigations are required before the popular belief can be held to be scientifically tenable. n callinI g Birrens also arid earth-work a simila f o s r type Eoman, and in seeking to identify them with " towns " . mentioned by Ptolemy e anpseudo-Richardth d s quit wa n accori e y Eo d, wite belieth h f prevalent in his day. He has made no attempt, however, to assign to the approximaty man e o accoun e t silencdate th r r f Eomano ,o fo et n military writersubjecthe workson of t s constructe samthe e on pland . Yet at Birrens at least there was undoubtedly a Eomau or Bomaiio- British settlemen f somo t e kind e sai r Horsley'b d muc fo dn an ; ca h s conjecture. Moreover e facth , t tha o mans t y sculptured stones have been found near the pre sent mounds, might go far to demonstrate that the latter are Roman too, could a connection in time be established betwee e sculpture nth e moundsth t merd Bu an se . proximito tw f yo objects of antiquity is not sufficient proof of their belonging to the same people and age. To sum up: of the Annaudale and other Dumfriesshire " Roman roads " of the O.M. it may meantime be said that, since they camo into notice along with the " Eoman camps," their presence in the future will depend largely on the ultimate fate of the latter. If the camps are Eoman e roadremainy th e ,camp ma th s t e explodesb le ; t wil i d l an d be difficul preveno t t e roadth t s from disappearing with them..