174 PROCEEDINGS OF THE SOCIETY, MAY 13, 1918.

I. AGR1COL ANTON1NE TH D AAN E WALL PROFESSOY B . R HAVERFIELD, LL.D., F.B.A., HONORARY MEMBER. A brief remar f Tacituso ks biographhi s father-in-lawn i hi , f o y , Agricola (ch. xxiii.), records that that Roman general about A.D con0 .8 - structe a dlin f forto e s acros e isthmussth , som mile5 e3 s wide, which separates the Forth from the Clyde ; and the excavations which Scottish antiquaries carried out some ten to twenty years ago have confirmed the remark, and have provided clues to the actual sites of some of these forts. The object of the excavations was, indeed, to explore the Antonine s defendinit Wal d an l g forts whic e Emperohth r Pius, about A.D. 140-145, built across this isthmus. But, before these excavations had been carried very far, it became clear that some, at least, of the forts of Pius stood on sites whic beed hha n fortifie Agricoly db a sixty years earlier. Agricola, according to Tacitus, had singular skill in choosing strong situation s forthi r s fo s(ch . xxii.). This recogniseswa s t hi onl dno y yb (sometimes partial) biographer, who, after all, neither was nor professed to be a strategist, but also by the expert judgment of military men two generations later, who, for the new series of forts, adopted sites already occupie Agricoladby thu,and s confirme eulogdthe Tacitusyof . Whether all the forts which were built about A.D. 140 along the Antonine Wall stan Agricolan do n sites, canno saide perhapd b t an , ye t s will nevee b r clear. The excavations above mentioned were, indeed, carried out with inadequate appreciation of the chronological value of certain evidences, possibilite th f o d f assigningyo an , with their aide occupatioth , f thino s or that particular site to a particular period. From this point of view, t musi e admitteb t d that som f theseo e excavations ought, whenever circimistances allow, to be repeated, or, at least, to be carried further— though e momentth t a ,e outloo th , f excavationko e nexth tr manfo s y year pooa s si r one s spen ,capitae ha sinc th r t ewa l which might other- wise have increas e gonth o et knowledgef eo . However, historian muce thessow o h t e excavations mucd an , h also ot the researches of our Fellow, Dr G. Macdonald, and of others, who have elicited fro ascertainee mth d facts more than t firs,a t sight, they seemed abl tello et . Thusbees ha nt , i long clear that certain, sites, namely, those of the five following forts (i.-v.) were first selected by Agricola, and that new forts were placed on precisely the same sites sixty years later; some- thing has been learnt, too, about the characteristics of Agricola's forts. t (i.Barhill,A ) abou mile0 1 t se centr th eas f f Glasgowo o et , traces AGRICOLE ANTONINTH D AAN E WALL. 175 of an Agricolan fort were detected in the excavations of 1902-1903, underlying the later structures of a fort datable to Pius. Of the earlier (Agricolan) fort only the ditches were found (fig. 1); its ramparts and its internal buildings—probably earthen ramparts and •wooden buildings, suc s werha e usua Britain i l n durin e Flaviath g n age—ha o doubn d t been levelled away when the. later fort was built, and in 1902 were no

****

^*^ « ******"_^ . ^*0

m-no ? ». «»• so. «. jo, co. TO. 80.90. we. I M I I I I I I I I I I I I

Fig. 1. Early Port Plan at Barhill.

longer distinguishable, save that the lines, etc., of the ditches indicate plainly those of the ramparts. This Agricolan fortl was apparently (fig. 1) a very small rectangular castellum, with an internal area of about 143 feet—no0 18 y b t quite two-third n acre a d barel f an ,o s y one-sixth of the area of the fort of Pius; it was defended by easily traceable and somewhat intricate ditches, which sugges d rount thaha t largi t di t e and complicated ramparts and ravelins, like those of Rough Castle or •of Ardoch. 1 Full report in Proceedings, vol. xl. pp. 403-546, from which fig. 1 is reproduced above. 176 PROCEEDING E , SOCIETY191813 TH :F Y O S MA , (ii.) At Rough Castle, near , excavations in 19031 showed, just north of the fort, a group of small pits (fig. 2), each pit about 7 feet long, fee3 t wide 2d £ feean , t dee e group th ; p covere l one-fiftal acrn n di a f eo h feet)(50 pit0e 20 fee th ;s y werb t e clearly mean obstacles a t enemn a o t s y charge. As I ventured to point out when they were found, they recall the " lilia " with which B.CCaesa, n i , 52 . r (de Bella Gallico, strengthenevii) 73 . d his blockading lines round Alesia in Central Gaul. Like those, they were . probably equipped inside with sharpened stake impalo st e fallen enemies —very similar pits were used in 1904 in the siege of Port Arthur. It is true that no actual traces of stakes were noted by the excavators, hut the wood may well have perished in eighteen centuries—or its vestiges may have been overlooked in excavation. Even without stakes, the pits would serve their purpose well. " Camouflaged coverina by " stragof w f heathero r o , like Bruce's pit t Bannockburnsa , they would effectively brea impetuosite kth impacd yan Highlana f o t d charge, which s mana , y remain Scotlann si northern i d dan n England show Romane ,th s especially dreaded.2 In the Roman age the barbarian offensives naturally took the for f heavmo y massed attacks e d owinrecklesth an , o t g s couragd ean e vasth t number e barbariath f o s n assailants d owinan , e g th als o t o scanty knowledg f chemistro e y possesse e wholth y eb d ancient world, barbarian massed attacks had real chance of success. From Barhill, we can see (as, indeed, can be guessed from continental evidence) that some of Agricola's forts were, probably, much smaller than the forts used forty year se Wal lateth f Hadrianr o l rfo , which itselbuils wa ft about A.D. 120, or for the Wall of Pius, built twenty years later still (A.D. 140).3 We may even conceive " journalist sHadrian'n "i s reign looking bac Agricolao kt , and styling the larger forts of their own time " super-castella." We may infer that the Agricolan garrisons were correspondingly small, and though t Barhilla s ,a , thesmald yha l area defendo st shord an , t ramparts mano t , they non lese eth s needed special defences when confronted with vastle th y superior force barbariaf so n assailants. Suc hspeciaa l defence was the " lily-bed" at Rough Castle: No remains as yet found inside its pits have yielde s precis it cluy o dean t e date, beyon face dth t s thai t i t 1 Full report in Proceedings, vol. xxxix. pp. 442-449. 1 The "multiple ditches" which defended many Roman forts in and in north England uncommoe ar southern ni nContinent—ae Englanth n o d dan t leastsame th equaen n i , i for d lman elaboration. It is fair to suppose that they were devised against some peculiarly dangerous attacks, such as Highland charges have throughout history been. Romae 1Th n castell secone th n ai d century varie gooda d deainevitables arean i wa l s a , , when garrisone th s varied fro mminimua infantr0 50 mf o 100 o yt 0 hors t Newsteae(a d perhaps more). acre 5 rouga 3- s sr a tako On hy 4 averageema 3- . Nineteen fort 'n o s s Wall (the fort a t Newcastle is wholly doubtful) seem to have averaged in area 3| acres. The areas of the forts along the Antonine Wall are still very imperfectly known, but seem to have been in general somewhat smaller than those on Hadrian's Wall; owing to the fact that they were mostly pro- vided with annexes, their internal areas would anyhow be difficult to calculate with precision.

178 PROCEEDINGS OF THE SOCIETY, MAY 13, 1918. Roman work. Perhaps fresh digging will some day, at trifling cost, furnis needfue hth l evidence tha excavatorsw e th tno , , know better what looo t r thaknows kfo nwa 1903n ni . However r MacdonalD , d (Roman s prettha ) f. y 1 welWall,23 . l p proved that these Rough Castle pits date from Agrieola, whil n inscriptioa e n found there connect e adjacensth t fort with the reign of Pius (Eph. epigr., ix. 1241). (iii. t ,A ) near Falkirk, excavation 189n si 9 yielded potsherds of Agricolan date, although those who reported on these diggings in our Proceedings were unduly sceptical.1 One might- connect with these Flavian potsherds a Camelon inscription mentioning the legio ii Adiutrix , but I believe this to be a modern forgery.2 (iv.) At Castle Gary, near the middle of the isthmus, pottery was found3 whic f Agrieolao he belongag e t combinei ;th o st s with other indications note t thida s fort suggeso t , t that Agrieol firse th t s buildeawa for a f tro on this hill-top thad an ,t Pius reconstructe t (Macdonalddi , Roman Wall, p. 374). The site, near the middle of the isthmus, is of strategic value as wel tacticalls a l y strong t wouli d an ,d naturally attract attention when the isthmus was being fortified, either by Agrieola or by Pius. Thu t i appears s tha considerabla t e proportioe th e fort th n o sf no Wall, which have been lately investigated -with the spade, were first established by Agrieola, and the remark of Tacitus is confirmed that the isthmus between Forth and Clyde was by him garrisoned by a row of detached forts, forming almos a wallt d veran ,y probably connected together by a road, as the Wall of Pius was. (v. I )desir quoto et more eon e piec f evidenceeo , which, thougt hno unpublished t beeno ns publisheo ha ,t w ne thin di e b s countryy ma d ,an most readers of our Proceedings. I came upon it after Dr Macdonald's Roman Wallissueds wa t addI .fore sf th Gadder o te lis > f placeth o t o t s wher footprinte eth detectede b f Agrieola ages so n hi ca ,f o . r o ,Accord - ing to Camden, there existed in Scotland, apparently between 1607 and the year of his death, 1633, an altar of which nothing has since been seen inscribed:— DEO SILVANO TANICVL S VERVS 5. PRAEF VSLLM

1 Vol. xxxv. pp. 329-427 ; see esp. pp. 380, 392. 2 Eph. epigr., ix. 1367; Classical Review, 1904, p. 461. There is no reason to doubt that this legion cam Britaio et n removes abouwa d t an dA.D , abou70 .inscription e Th t A.D . 85 . f genuine,,i woullateo n e rdb than Agrieola could an , d hardl earliee yb r than about A.D because, 80 . , before that time Romao n , n troop reached sha neighbourhooe dth Falkirkf do . ' Report of excavations, Proceediiigs,\o\. xxxvii. pp. 271-346; the pottery was not noticed till later. AGKICOLE ANTONINTH D 9 AAN 17 E WALL. d Silvanusgo e . th TanicuL , o "T s Verus, prefec a f (o t , pre- sumably), pay s vow.hi s " generalls ha t I y been recognised that this cop probabls yi itee mon yn i incorrect e prefecte namth th ; f . eo Tanicus L , , musmisreadinga e b t ; otherwis e inscriptioth e n migh.thf y o b e e p tmanu on pas t s se a ys soldiers on the Wall of Pius, which, as must be confessed, add little to our knowledge. However, the Bodleian Library in Oxford contains a

numbe f paperso r , mostly relatin Camdeno gt , know "s n a MSS . Smith." J

I. •V£J{VS F V-$ J-L.M

Fig. 3. Inscription from . In autograph of Camden,

Britannia, . 1607 ed . 699 p , , Bodl. Lib. 2 Nothe.Smitn i .1 h collection (MS. Smitcopa s i f Camden' y) o h 1 s (ed. 1607), with marginas l noteha e Camdenh y sb 9 margie 69 th . p n f o ;n o written in his own hand a copy of our inscription; he does not give any specific provenance for the stone, but his note is so placed that it can only refe Caddero t r texe Th t .whic sete t differh sou onl lettey e b sy on r from the commonly received text (fig. 3). It has " L. Tanicius " instead of " L. Tanicus." Small as the variation is, it is important. Tanicius is an exceedingly rare Roman nomen, whicf o hI fin d onl othee yon r example. 1 Dr Thomas Smith, 1638-1710, Fello f wMagdaleo n College, Oxford, edited Camden's Corre- spondenc 1691n ei . 2 Camden seems to have written first "Tanisius," and then corrected that to "Tanicius." "Tanisius" is a very rare nonwn, no doubt a by-form of the much commoner "Tanusius." But pretts i t i y plain that here Camden meant dowse o nt " Tanicius." 180 PROCEEDINGS OF THE SOCIETY, MAY 13, 1918.

That is in a long and well-attested inscription of a soldier with precisely the same names as our prsefect—L. Tanicius Verus; I imagine one may conclud eithes e sam thae wa (th n rth e teh e ma most natural assumption) or a close relative, and a man of the same period. Now, this other Tanicius Verus, accordin s inscriptiohi o gt n (foun Uppen i d r Egypts wa ) born at Vienna in Gaul, now Vienne, on the Rhone, a little south of Lyon, reached an e randth f centurio ko "n ni Legi i Cyrenaicoii a " (GIL. iii. 34). t furtheI r record stationes s thawa e h t Uppen di r Egypt thered an , , like many Romans in the Empire, went to hear the sunrise song of the vocal statu Memnonf eo , muc modera s ha n soldie t Bulfora r d might last June, 23rd-24th, have gone to see sunrise at ; indeed, so impressed was Verus, that, while in Upper Egypt, he went to hear it thirteen times. The stone gives the dates of his thirteen visits—the first on 7th November 80 A.D., the last on 2nd June 81; after that, the visits stopped, and we may assume that he was transferred from Egypt, or at any rate from anywhere nea e statuth r f Memnoeo t Carnacn(a e Egyptiath , n Thebes, in Upper Egypt) t woulI . d seem that actuall transferres wa e yh Norto dt h Britain and to the command of an auxiliary cohort there.1 As he was at Thebes in June 81, he would reach Britain towards the second half of Agricola's governorship, just after Domitian had (in September 81) succeeded to Titus. It was about that time that the building of the forts on the Clyde-Forth isthmus began (Agr., ch. xxii.).2 One of these cause fortse doubto t et sno clearlo )d Gadder s CamdeI s ywa (a s f i , nwa right in assigning the altar to that place. I note in this connection that it seems to have been the middle of the isthmus of which Agricola took firmest hold. The easternmost of the five forts in question was Camelon; there Agricola seized the point from which by far the best natural route led north from the isthmus to the Highlands. By way of the rock of Stirling, which old Hector Boece not unnaturally thought Agricola to have fortified, an easy route from the Forth leads along the banks of Allan Water and the north face of Ochile th s int Eare oth n Pertho vallet d y.an From that strategic centre, further valleys lead north-eastwards through Strathmore, betweee nth foothill e easterth f o sn Grampian e Sidlawsth d o Forfaan o st t , d an r ease th t coas f Scotlano t d near Montrose; henc intermittenn ea t coastal strip of lowland leads on to and even beyond Aberdeen, whilst the

1 According to A. von Domaszewski (Rangardniing ties romischen Heeres, p. 108) the promotion oa "centurif o legkmis e pos f th "prsefectuo t o "t s cohortis thn i e s earlie"wa r empire quite ordinary. So we put Verus in Egypt first, and in Britain afterwards, not vice versa. * Or in 80. It is, uncertain whether Agricola came to Britain in 77 or 78. I rather incline to purposbuildin our greathe fort ho poin the so, 77 eFor the of ts gif of ; moment is begat 80. nin . But if it began in 81, we might ascribe the idea of fort-building to Domitian's initiative, which Tacitus would naturall t emphasiseyno . AGRICOLANTONINTHE AAND 181 E WALL. valleys which reach the coast near that city afford a not difficult means of entry thence intinterionorte d th oan f ho r uplands. e routThith s ei s by whic n army-leadera h , wishin o deagt l wit e Grampiath h n massif, must inevitably advance from a base on or near the Forth, and it is likely that Agricola took this line, however far he finally penetrated into Scotland. But that raises problems as difficult as the Grampians to momene negotiate time th th t I have a t entanglo d t t no ean , e myself in them. f illustrationo y I addwa y b , , another cashomonyma f eo altarn A . , found abou t Croyhill yeara 0 o 10 t sag , name officen sa r " Fabius Liberalis " (Macdonald, Roman Watt, p. 341). An officer of the same name occurs on an altar found in a fort on the " " in Germany, at Stockstadt near Aschaffenburg e Germath t Bu n .alta s ascribe i re lates th y tb d German writer on it to the third century A.D. He seems to have thought

that there may be som1 e identity of men. But he has overlooked the fact tha Wale th t f Pius abandoneo l swa e Romanth y db s about A.D. 180o s , that the Croyhill Fabius belonged practically to the generation before the Stockstadt Fabius. There is thus an appreciable difference in time betwee e Stockstade date th th e n Croyhilf th o s d an tl officers, which makes their identity a priori unlikely. Had the names "Fabius" and " Liberalis " been less common, probabilities woul f coursdo differente eb , but there seems to be known only one other Tanicius in the wide Roman world, while wit n "Fabii,e cognomehme th d "an n "Liberalis" abound. The case, therefore r identifyinfo , r Taniciuou g s Verus wit e onlhth y other known Tanicius, who is also Verus, is far stronger than the case for identifying the two (if they be two) men called "Fabius Liberalis." Moreover lattee ,th r identification involve I hav s s(a e said) plaia , n diffi- cult datesn yi , which doe arist sno relatio n ei Taniciuo nt s Verus. 1 Drexel, ORL., xxxiii. (1910), p. 96.