The Roman wall in Scotland

THE ROMAN WALL IN SCOTLAND

PUBLISHED BY

JAMES MACLEHOSE AND SONS, GLASGOW,
publishers to the elmtoersitii.

MACMILLAN AND CO., LTD., LONDON.
Ne-w York, - - The Macntillan Company.
London, • • • Simpkin, Hamilton and Co.
Cambridge. • • Bowes and Bowes.
Edinburgh, - - Douglas and Foulis.
Sydney, • • • Angus and Robertson

MCMXI.

1

4.

I

THE ROMAN WALL
IN SCOTLAND

BY

GEORGE MACDONALD, M.A., LL.D.

WITH MAP, PLANS, AND NUMEROUS PLATES

^ *Vi3

D X / I -
JAMES MACLEHOSE AND SONS V^bl

o <r/ .
GLASGOW ^ ^ Cx . i ( I

PUBLISHERS TO THE UNIVERSITY

\

\

'777
>5

GLASGOW: PRINTED AT THE UNIVERSITY PRESS
BV ROBERT MACLEHOSK AND CO. LTD.

D • M -8

PATRI • OPTIMO

CALEDONIAE- ROMANAE

INDAGATORI • STVD I OS I 881 M O

PRO • PIETATE

FILIVS

PREFACE

THIS book is the direct outcome of an invitation
addressed to me two years ago by the Trustees of
the Lectureship in Archaeology instituted in the
University of Glasgow by the late Mr. J. D. G.
Dalrymple of Woodhead. Being of opinion that the
time was ripe for a fresh survey of the whole body of
evidence relating to the subject, they did me the
honour of requesting me to undertake the task. The
six lectures of which the course consisted were delivered
in the spring of 1910. They form the main strand in
the thread of the following chapters, and it is accord-
ingly the requirements of the ordinary cultivated reader,
rather than those of the specialist, that have deter-
mined the plan of the volume and also the method of
treatment. I am hopeful, however, that even the
specialist will not be sent empty away. At all events,
I have endeavoured to be thorough where thoroughness
seemed desirable. And the mass of relevant material
here brought together is undoubtedly much greater
than anything that has up till now been conveniently
accessible to students of the Antonine Vallum, while
the sifting process has led to deductions which will, I
trust, be accepted by competent judges as representing
a permanent advance.

viii PREFACE

The dedication indicates a debt of which I have
been uniformly conscious. To what is there implied I
would only add that occasionally, and notably in
chapter i., I have been able to utilize my late father's
unpublished papers. The footnotes will show how
extensive are my obligations to other writers and to
friends. In one or two cases a special word of thanks
is due. There are not many of the problems of the
Wall that I have not, at some time or other, had the
advantage of discussing personally with Professor
Haverfield. When the book took visible shape, he at
once agreed to read the proofs, and I cannot speak too
warmly of the pains he has bestowed upon them.
His criticisms have been most helpful throughout, —
always frank and always rich in suggestion and in
stimulus. The sheets of chapters iv., v., and vi.
have been similarly revised by Mr. Alexander Park
and Mr. John M'Intosh, who have accompanied me on
expeditions along various parts of the Roman line, and
whose knowledge of the section between Kirkintilloch
and Croy is peculiarly intimate. From first to last Mr.
Park's kindness in furthering detailed investigation
has been unwearied.

A liberal grant from the Research Fund of the
Carnegie Trust, coupled with a generous loan of
blocks by the Council of the Society of Antiquaries of
Scotland, has enabled illustrations to be provided on a
scale that would otherwise have been altogether im-
possible. Even this assistance would hardly have
been adequate but for the fact that Mr. John Annan
of Glasgow allowed me free access to his extensive
stock of negatives, and also presented me with a large
number of photographs which he was good enough to

PREFACE ix

take expressly for my benefit. These latter include
the frontispiece. Mr. Mungo Buchanan of Falkirk,
besides answering many enquiries regarding the portion
of the Wall with which he is most familiar, readily
permitted me to reproduce his plans of Castlecary and
Rough Castle. The merits of these can only be fully
appreciated by those who know the difficult circum-
stances under which the surveys had to be carried out.
The plan of the guard-house at the exit of the road to
Camelon (Figure 12) is Mr. Buchanan's too, and it
was from him that I obtained the photograph of
the Maiden Castle as it appeared in 1893.

GEORGE MACDONALD.

EDINBURGH,

February, 1911.

CONTENTS

CHAP. PAGE

I. THE LITERARY TRADITION i

II. THE HISTORICAL BACKGROUND : ORGANIZATION OF

THE ROMAN ARMY 42

III. THE HISTORICAL BACKGROUND: ROMAN FRONTIER

POLICY AND FRONTIER POSTS - 64

IV. THE ACTUAL REMAINS: INTRODUCTORY - 85
V. THE LINE OF THE RAMPART AND DITCH - 108

VI. THE FORTS : FROM THE CLYDE TO BAR HILL- - 152

VII. THE FORTS : FROM CROY HILL TO THE FORTH - 203

VIII. MINOR STRUCTURES - - 245

IX. THE LEGIONARY TABLETS - 268

X. OTHER INSCRIPTIONS - 322

XI. MISCELLANEOUS EVIDENCE - 356

XII. CONCLUSIONS - - 383

INDEX - - - - 404

LIST OF PLATES

PLATE PAGE

On Ferguston Muir, 1910 - - Frontispiece

I. A. Coins relating to Britain (Hadrian and Pius);

B. Slab taken out of the Tyne in 1903 - 8

II. Coins relating to Britain (Pius, Commodus and

Severus) - i 6

III. The Danube Frontier in the Second Century A.D. - 72

IV. The Storehouse at Rough Castle - 80
V Culverts beneath the Rampart - 96

VI. Sections through the Rampart- 98

VII. Views of the Ditch — i. East of Castlecary ; 2. West

of Rough Castle - 104

VIII. Views of the Ditch— East and West of Thorn Farm 114

IX. In New Kilpatrick Cemetery - - - 116

X. Remains of the Limes — i. East of Castlecary; 2.

Near Seabegs - 130

XI. Remains of the Limes — i. Cut by the Caledonian

Railway; 2. Near Bonnyside - 132

XII. Section of Rampart and Ditch at Rough Castle,

looking W. - - 136

XIII. i. Above Millhall Burn; 2. The Maiden Castle - 144

XIV. Slab found at Bridgeness in 1868 - - 148
XV. The Outlook from Castle Hill Fort - 160

XVI. i. Castle Hill; 2. Peel of Kirkintilloch - - - 176

xiv LIST OF PLATES

PLATE PAGE

XVII. i. The Military Way on Bar Hill; 2. N.E.

Corner of Bar Hill Fort - 190

XVIII. Section of W. Rampart of Bar Hill Fort - 192

XIX. Pillars from the Principia at Bar Hill - 196

XX. i. Base of Pillar from Castle Hill; 2. Capital

from Bar Hill; 3. Inscribed Slab from Bar Hill 198

XXI. Shoes from Bar Hill - 200

XXII. Plan of Bar Hill Fort - 202

XXIII. Drain through N. Wall of Castlecary Fort 208

XXIV. E. Wall of Castlecary Fort, looking N.W. - 210

XXV. Castlecary Fort— i. Kerb on Inner Side of N.

Wall ; 2. Inner Face of South Wall of Principia 214

XXVI. Storehouse at Castlecary - - 216

XXVII. Plan of Castlecary Fort - 218

XXVIII. Junction of E. Rampart of Rough Castle Fort

with Rampart of Limes - 220

XXIX. West Rampart of Rough Castle Fort looking N. 222

XXX. Rough Castle Fort— i. Stone-lined Pit in

Sacellum; 2. Hypocaust Pillars in the Baths 226

XXXI. Rough Castle — i. Fragment of Statue; 2.

Inscribed Slab from the Principia- - 230

XXXII. Defensive Pits (Lilid) at Rough Castle 232

XXXIII. Plan of Rough Castle Fort - 236

XXXIV. i. Altar from Rough Castle; 2. Tombstone

from Mumrills - 240

XXXV. Beacons on the Danube Frontier in the Second

Century A.D. - 264

XXXVI. The Legionary Tablets (Nos. 1-3) 272

XXXVII. The Legionary Tablets (Nos. 4-6) - 276

XXXVIII. The Legionary Tablets (Nos. 7-9) - - - 280

LIST OF PLATES xv

PAGE

XXXIX. The Legionary Tablets (Nos. 10-12) - - 288

XL. The Legionary Tablets (Nos. 13-15) - 298

XLI. The Legionary Tablets (Nos. 16, 18, 19) - - 302

XLII. The Suovetaurilia on the Bridgeness Tablet - 304

XLIII. i. Fragment with Name of Lollius Urbicus ; 2.
Legionary Tablet from Shirva ; 3. Auxiliary

Tablet from Castlecary - - 314

XLIV. Altars from Castle Hill and Auchendavy - - 336

XLV. Altars from Auchendavy, Bar Hill, and Croy

Hill - 340

XLVI. Altars from Castlecary and Polmont - - 344

XLVII. Tombstones from Shirva - 352

XLVIII. Miscellaneous Sculptures - 358

XLIX. Stone Busts from Bar Hill - - 360

L. The Sepulchral Banquet on Sculptured Reliefs

from Shirva - - 362

LI. Fibulae and Shoes from Castlecary - - 366

LII. Miscellaneous Objects from Castlecary and Rough

Castle - 368

LIII. Stones from Croy Hill - 376

LIV. The Tumulus at Cadder - - 384

LV. Evidences of Reconstruction — i. At Rough Castle;

2. At Castlecary - - 396

MAP.

The Roman Wall from Clyde to Forth At end.

LIST OF ILLUSTRATIONS IN THE TEXT

FIG. PACK

1. Shape of Ditch - - 101

2. Early Fort at Bar Hill - - 189

3. Principia at Bar Hill - - 194
4-7. Capitals of Columns from Bar Hill - - 197

8. Interior Buildings at Castlecary - 212

9. Roy's Plan of the Baths at Castlecary - - 214

10. Road laid across E. Ditch at Castlecary - 218

11. Interior Buildings at Rough Castle - - 225

12. Guard-House, Protecting the Exit of the Road to

Camelon - - 249

13. Sketch-Map to illustrate the 'Find-Spots' of Distance

Slabs Nos. 8 and 9 - - 285

14. Objects of Horn from Bar Hill - 365

15. Stone Lamp from Rough Castle - 367

1 6. Fragment of Decorated 'Samian' Ware from Castle-

cary - - 372

CHAPTER I

THE LITERARY TRADITION

WHEN the dawn of written record begins to break
over the darkness that enshrouds the early history of
Scotland, it is the isthmus of the Forth and Clyde
that receives the first ray of clear and unmistakable
light. Tacitus, in the biography that has given his
father-in-law Agricola an immortality to which his
feats of arms seem hardly to have entitled him,1 thus
describes the campaign of A.D. 81 : "The fourth summer
was spent in securing the districts previously traversed ;
and, were such a thing compatible with the courage of
our armies and the glory of the name of Rome, there
would have been found within the limits of Britain
an ultimate frontier-line. For Clota and Bodotria,
being carried far inland by tides from opposite seas,
are separated by but a narrow strip of land ; steps
were now taken to strengthen this isthmus by fortified
posts and to occupy the entire sweep of country to
the south, so that the enemy were pushed back into
what was virtually another island."2 Here, and here
only in almost the whole course of Tacitus's narrative,

aThe only certain mention of Agricola in any other ancient writer
appears to be a brief reference to his wars in Britain in Xiphilinus's
abridgement of Dio Cassius (Bk. Ixvi. c. 20).

2 Agricola, c. 23.

A

2 THE LITERARY TRADITION

are we on perfectly sure ground. Clota and Bodotria
are the Clyde and the Forth of to-day. We know,
then, that Agricola built a chain of forts between the
estuaries. But there is no evidence to suggest that
any other barrier was erected. The ' Roman Wall,'
in the usually accepted sense, belongs to a later date.
Agricola's further doings in Caledonia, interesting as
they were, do not concern us here. He remained in
Britain three or four years longer, and then Domitian,
who was now emperor, prompted (Tacitus would have
us believe) by envy and by dread of the victorious
general's popularity with the army and the people,
recalled him from the province in the midst of his
triumphs. Returning to Rome with prompt obedience,
he lived in strict retirement until his death nine years
later. We are left to infer that the additions he had
made to the Empire were needlessly sacrificed ; " per-
domita Britannia et statim missa" is the pregnant phrase
in which the effect of the decision is elsewhere summed
up.1 His recall, however, may well have been due,
not to the sinister motives suggested by Tacitus, but
to a conviction on the part of Domitian that the forward
policy of his lieutenant was a failure ; the emperor would
probably have been justified in looking for more sub-
stantial conquests as the fruit of seven years of practi-
cally continuous warfare.2 However this may be, the
garrisons appear to have been almost immediately

1 Tac. Histories, i. 2.

2 Besides, danger threatened at another point. The Dacians had
defeated Oppius Sabinus, and Domitian had to take the field in person.
It seems probable that in A.D. 86 the army of the Danube had to be
reinforced by two legions — the First Adjutrix from Germany, and the
Second Adjutrix from Britain. (^\\tet\vt\<g,Jahreshefte des osterr. archdol.
Instituts, 1904, vol. 7, Beiblatt, pp. 37 f.)

THE OLDER AUTHORITIES 3

withdrawn from the line of the Forth and Clyde,1 while
the silence of historians renders it natural to suppose
that, during the remainder of this reign as well as
during the reigns of the succeeding emperors, Nerva
and Trajan, the Romans on the British frontier acted
mainly on the defensive. Doubtless the turbulent
northern tribes from time to time made efforts to drive
the invaders from their advanced positions. One
very suggestive fact is revealed by the inscriptions.
A tablet found at York in 1854 attests the presence of
the Ninth Legion there in A.D. io8.2 Thereafter this
division of the army vanishes entirely from history, the
victim, as Borghesi has with much probability sur-
mised, of some crushing defeat at the hands of the
Brigantes.

Hadrian, the next emperor, personally visited Britain
as well as the other provinces of his vast dominions.
For the events of his reign our main authorities are
his life by Spartian in the Historia Augusta, and
Xiphilinus's abridgement of the sixty-ninth book of
Dio Cassius. The Historia Augusta consists of a
series of biographies of various emperors, princes and
pretenders from Hadrian to Carinus, that is, from
A.D. 117 to A.D. 284. The authors, of whom the most
important for our purpose are called Spartian and Julius
Capitolinus, have taken Suetonius for their model, and
are much more interested in personal gossip than in
history in its larger aspect. Their work professes to
be a compilation to which six different writers have
contributed, and to have been composed partly in the
reign of Diocletian (A.D. 284-305), partly in that of

1 Macdonald and Park, Roman Forts on the Bar ffzll, pp. 14 f,

2 C.I.L. vii. No. 241.

4 THE LITERARY TRADITION

Constantine (A.D. 306-337). As a matter of fact, there
can be little doubt that the names of writers attached
to the " Lives " are in a state of hopeless confusion,
while there is good ground for suspecting that the
real date is much later than the nominal one. Taken
as a whole, the book is of little value ; it has been
stigmatized by Mommsen as " one of the most pitiable
pieces of scribbling that have come down to us from
antiquity."1 It contains, however, strata of varying
quality. Thus it is generally agreed that the sketches
of the reigns of the earlier emperors, Hadrian and
Pius for example, contain not a little genuine historical
material. But no successful effort has yet been made
to separate the wheat from the chaff, or to set the
question of the ultimate sources on a thoroughly
satisfactory basis.

The other authority named above, Dio Cassius
or (more correctly) Cassius Dio Cocceianus, stands in
a different category. Although a Greek by descent
and birth, he was the son of a Roman senator, and had
himself a highly successful career. He came to Rome
in the reign of Commodus, and soon made his mark in
public life, his character and ability winning for him
the confidence of several successive heads of the state.
After filling the usual round of offices, he was ultimately
elected consul or dinarius for A.D. 229. But the strin-
gency of his discipline had made him unpopular with
the army ; and the emperor of the day, Severus
Alexander, while well affected to him personally, was
too weak to shield him from the consequences of the
displeasure of the soldiery. Accordingly he did not
venture to exercise his functions, but, acting on an

1 Hermes, xxv. p. 229.

THE OLDER AUTHORITIES 5

official hint, shook the dust of Rome from off his feet
and withdrew to a voluntary exile in his native Bithynia.
His History, which was divided into eighty books,
covered the whole field from the foundation of the city
down to the accession of Severus Alexander. We
may believe that, had it survived in its entirety, it
would have been of great value, particularly for the
period that fell within the author's own lifetime. The
portion which has come down to us complete is, how-
ever, but small. We know the rest through fragments,
or through epitomes made by later hands — notably the
abridgement of Books xxxvi.-lxxx. by Xiphilinus, a
monk of Constantinople, who lived in the eleventh
century.

We have seen that Hadrian, in the personal survey
he made of his empire, visited Britain. The most
probable date for his crossing from Gaul is the spring
of the year A.D. 122. Coins were struck to celebrate
his arrival. Some show him being welcomed by the
Province, while on others he is represented haranguing
his troops (Plate LA. i). Carrying out the policy he
had adopted elsewhere of seeking to establish an
enduring peace by fortifying securely a selected frontier,
he fixed the limes, or official boundary of the province,
on a line that excluded a very considerable part of the
country Agricola had overrun. To divide the Romans
from the barbarians, he erected (Spartian tells us) a
murus or wall, eighty miles in length.1 This statement
of distance makes it obvious that Hadrian's wall ran
along the neck of land that separates the Solway Firth

1 Brittaniam petit, in qua multa correxit murumque per octoginta
milia passuum primus duxit, qui barbaros Romanesque di-videret ( Vit.
Had. 11, 2).

6 THE LITERARY TRADITION

from the mouth of the Tyne. Whether the barrier
thus set up included both the more imposing stone
wall and the earthen rampart, the remains of which
are still to be seen, or whether it was something
altogether distinct from either, are questions on which
we need not enter. If Spartian is to be trusted,
incidents had occurred in Britain that suggested the
advisability of a special rectification of the frontier.
Speaking of the troubles that faced Hadrian on his
accession, he mentions that " the Britons refused to be
kept under Roman rule."1 A fragment of the lost
history of Cornelius Pronto, De Bello Parthico, hints
at a very serious struggle. Addressing Lucius Verus,
at whose request the work was written, Pronto exclaims :
" What a number of soldiers were slain by the Jews,
what a number by the Britons, in the reign of your
imperial grandfather Hadrian!"2 Juvenal's mention
of the " castella Brigantum " is perhaps a more strictly
contemporary reference.3 These scanty allusions seem
to point to some great outbreak of the northern tribes,
and it is difficult not to accept the suggestion that
connects them with a catastrophe in which the Ninth
Legion was annihilated. Fronto's implied comparison
with the Jewish revolt is extremely significant. There
is some reason to think that in Judaea, too, a
whole legion (the Twenty-Second Deiotariana) was
annihilated.4 The coins of Hadrian struck in honour

1 Brittanni teneri sub Romano, ditione nonpoterant ( Vit. Had. 5, 2).

2 Avo vestro Hadriano imperium obtinente quantum militum ajudaeis,
quantum a Britannis caesum (Ep. de bello Parth. Ed. Naber, pp. 217 f.).

3 Sat. xiv. 196. As is well known, some have thought that the poet had
himself seen service in North Britain. But the evidence for this is at the
best very doubtful.

4 Von Domaszewski, Rangordnung des romischen Heeres, p. 179.

THE OLDER AUTHORITIES 7

of the British victories are the earliest that bear the
type of Britain subdued (Plate I. A. 2).

Hadrian was succeeded in A.D. 138 by Antoninus
Pius. Pausanias, who was a contemporary of both
emperors, tells us incidentally that the latter "deprived
the Brigantes in Britain of most of their territory,
because they too had entered on a war of aggression
by invading the district of Genunia, which is subject
to Rome."1 The district of Genunia is mentioned
nowhere else. But, from the days of Casaubon and
Horsley until quite recently, it has been usual to
identify the operations mentioned by Pausanias with
those of which details are given in the biography of
the emperor, said to be by Julius Capitolinus, in the
Historia Augusta. This sketch is the chief source of
our information regarding the life of Pius, the part of
Dio that related to him having apparently been lost
before Xiphilinus's time. According to Capitolinus,
the earlier portion of the reign was disturbed by insur-
rectionary movements in Britain, Germany, Dacia and
Judaea, not to speak of other parts of the empire.
" Through his legates," we are told, "he carried on
many wars ; for he subdued the Britons through
Lollius Urbicus, a legate, and, after driving back the
barbarians, erected another wall of turf."2

This furnishes a definite landmark. The exact sig-
nificance of the last few words is not, indeed, easy to
determine. But it looks as if Capitolinus, or whoever

1 («r€Te//,€TO Se KCU TWV kv BpiTTavip. Bpiyai'Twi/ TTJV TroXXrjv, on
€7r€o-/2cumv /cat oSroi trvv OTrAois ?jp£av e's TT)V Tevovviav fiotpav virrjKoovs
cP(i>/icu<ov (viii. 43, 4).

2 Per legates suos plurima bella gessit. nam et Brittanos per Lollium
Urbicum vicit legatum alio muro cespiticio summotis barbaris ducto, etc.
(Vit.Ant. />«, 5,4).

8 THE LITERARY TRADITION

the author of the biography may have been, wrote
with Spartian's life of Hadrian or some now lost
history before him, and meant " another wall, also of
turf," implying that he knew of Hadrian's wall and
believed that the new wall and Hadrian's wall were
alike made of turf. He gives nothing to guide us as
to its site, nor does he even indicate its length. But
numerous inscribed stones found between the Forth
and the Clyde prove beyond doubt that here stood the
"wall of turf" erected during the reign of Pius, while
a small fragment of a tablet, containing little more
than the name of Lollius Urbicus as governor of the
province, furnishes conclusive confirmation of the
statement of Capitolinus. It should be noted, how-
ever, that in the inscriptions the barrier is called, not
murus, but vallum, — a designation which, as having
been given to it by its actual builders, should be
preferred to that applied to it by Capitolinus. The
probable date of erection is A.D. 142. Lollius seems to
have been governor of Britain circa A.D. 140-143, and
it was doubtless his victories that led to his master
being for the second time acclaimed " imperator."
Epigraphic evidence enables us to fix the acclamation
as having taken place towards the close of 142 or the
beginning of I43-1 Coins issued about this time show
a full-length figure of Victory, accompanied by the
legend BRITAN. IMPERATOR n. (Plate I. A. 3). Others
bear one of two varieties of the type of Britain
subdued (Plate II. i), a device which, as we saw
above, was originally introduced in the reign of
Hadrian.2

1 Von Rohden in Pauly-Wissowa's Real-Encyclopadie, ii. p. 2506.

2 See supra, p. 7.

PLATK I

A. COINS RKLATINI; TO KKITAIN (HADRIAN AND PIUS)

lMP>ANTOKf
NOAVC'Plop
I LAl'o

THE OLDER AUTHORITIES 9

The line followed by the Vallum lies considerably
to the north of what is usually believed to have been
the extreme limit of Brigantian territory. This latter
does not appear to have extended much beyond the
present border between England and Scotland, — a fact
which has always made it a little difficult to accept
the current view of the passage cited from Pausanias.
Why should aggression on the part of the Brigantes
have had as its immediate consequence the building of a
wall across the Forth and Clyde isthmus ? Within the
last decade epigraphy has come to the aid of history
by suggesting that Pausanias was not referring to the
campaigns of Lollius Urbicus at all, but to grave dis-
turbances that broke out little more than ten years
later. A slab of stone taken from the river Tyne at
Newcastle in 1903 (Plate I. B.) bears a dedicatory
inscription to Antoninus Pius by a draft or drafts of
men sent from Upper and Lower Germany as a special
reinforcement to the three British legions under Julius
Verus, governor of Britain.1 It had not been pre-
viously known that Verus had ever been in command
of the province. But the Newcastle discovery not
only furnished us with this information but also, while
proving that trouble of no ordinary kind was afoot
during his governorship, enabled the scene and the
date of his military activity to be ascertained with some
approach to precision. It was at once pointed out that
he must be the legate Julius, whose name is incomplete
on three other inscriptions — one from Birrens, one

1 The inscription runs : IMP[ERATORI] ANTONINO AVG[vsTO] PIO
P[ATRI] PAT[RIAE] VEXI[L]LATIO LEG[IONIS] n AVG[VSTAE] ET LEG[IONIS]
vi VIC[TRICIS] ET LEG[IONIS] xx V[ALERIAE] V[ICTRICIS] CONTRIBVTI
EX GER[MANIIS] DVOBVS SVB IVLIO VERO LEG[ATO] AVG[VSTI] PR[O]
P[RAETORE]. I have to thank Mr. R. Blair, F.S.A., for a photograph.

io THE LITERARY TRADITION

from Netherby, and one from Brough in Derbyshire,
all forts within the region that certainly belonged to
the Brigantes.1 Of these the Birrens inscription is the
most important. It indicates either the building or the
rebuilding of the fort, and it was set up in the twenty-
first year of the emperor's reign or, in other words, in
A.D. 158. Coins, too, are helpful. Bronze pieces of the
year A.D. 155 have on the reverse a variety of the type
of Britain subdued (Plate II. 2).2 Analogy suggests
that there must have been a special occasion for their
issue.3 Is it not highly probable that it is in the light
of these monuments, rather than of the words of
Capitolinus, that we must interpret the vague state-
ment of Pausanias ?4 The coins may be commemora-
tive of the first decisive success, while the building or
rebuilding of permanent forts marks the final stage in
the long and troublesome task of pacification. So far
no trace of Verus has emerged to the north of Birrens.
But, if the land of the Brigantes was aflame, it is more
than likely that the conflagration spread as far as the
Antonine Vallum.

When we leave the Birrens tablet behind, we
plunge once more into the misty region of obscurity
and conjecture. Antoninus Pius was succeeded in
A.D. 161 by Marcus Aurelius, who straightway
associated with himself in the government his
adopted brother, Lucius Aurelius Verus. During the

1 See Haverfield, Proc. Soc. Ant. Scot, xxxviii. (1904), pp. 454 ff.

2 These coins are by no means easy to procure in a good state. I am
indebted to Mr. F. A. Walters for putting at my disposal the specimen
shown on the Plate. It was one of a number found at Croydon in 1905
(see Num. Chron. 1907, pp. 353 ff.).

3 See supra, p. 7 and p. 8.

4 Haverfield, I.e. p. 457.

THE OLDER AUTHORITIES n

joint reign, and probably soon after its commence-
ment, war was threatened in Parthia, Britain, and
several other provinces of the empire. This we learn
from the life of Marcus in the Historia Augusta.
Julius Capitolinus (to retain the traditional name of
the biographer) adds that " Calpurnius Agricola was
sent against the Britons."1 In describing what took
place some eight years afterwards when, on the death
of Verus, Marcus became sole emperor, Capitolinus
notes again the imminence of war in Parthia and in
Britain in almost the same words as before, without,
however, indicating what steps were taken to meet
the danger.2 So confused is the manner in which
the materials Capitolinus had at his disposal are put
together that we cannot be at all certain that he
is referring to two distinct movements. Dio, as
abridged by Xiphilinus, discusses the affairs of the
period at considerable length, but makes no mention
of the campaign or campaigns alluded to by
Capitolinus, although he records the despatch by
Marcus to Britain at a later date (A.D. 173) of
5,500 of the 8,000 auxiliary cavalry which were put
at his disposal when he had secured the submission of
the wild tribes on the Danube.3 The statement of
Capitolinus is, however, fully confirmed by the
inscriptions. Stones found in the north of England
remain to attest the presence of Calpurnius in the
island, and their situation is quite consistent with
the supposition that his troops were operating in the

1 Inminebat etiam Brittanicum helium . . . et adversus Brittanos quidem
Calpurnius Agricola missus est (I/it. M. Ant. Phil. 8, 7 f.).

2 Inminebat et Parthicum et Brittanicum helium (Ibid. 22, i).

3 Dio, Bk. Ixxi. c. 16.

12 THE LITERARY TRADITION

territory between the two isthmuses.1 "It is not
impossible," says Horsley, "that the Caledonians had
broke thro' the wall of Antoninus Pius not long after
it had been erected, which may be the reason we
meet with no inscriptions found there but what belong
to that reign ; at least no other emperor is expressly
mentioned but he, nor any certain date but what
relates to his time." 2

In the reign of Commodus (A.D. 180-192), the
successor of Marcus, there were once more wars and
rumours of wars in various parts of the empire,
including Britain. They are thus briefly recorded
by Lampridius, another of the Scriptores Historiae
Augustae : " the provincials in Britain, Germany, and
Dacia rejected his rule,"3 and again: "the Britons
actually wished to elect a rival emperor."4 A few
further particulars have been transmitted to us from
Dio through Xiphilinus. Writing of the wars in
which Commodus was involved, he says : "By far the
greatest was that waged in Britain. The tribes in the
island having crossed the wall (reF^o?) that separated
them from the cantonments of the Romans, com-
mitted great havoc, and slew a Roman general with
the soldiers under his command. Commodus there-

1 CJ.L. vii. Nos. 225 (Ribchester), 758 (PCarvoran), 773 f. (Carvoran).

2 Britannia Romana, p. 53.

3 In Brittania in Germania et in Dacia imperium ejus recusantibus
•provincialibus, etc. (Vit. Comm. Ant. 13, 5).

4 Appellatus est Commodus etiam Brittanicus ab adulatoribus cum
Brittani etiam imperatorem contra eum diligere voluerint (Ibid. 8, 4).
This is the sort of action for which mutinous legionaries might have been
expected to be responsible. Perhaps, therefore, the allusion is to the
disturbances quelled by Pertinax (Dio, Bk. Ixxii. c. 9), although we
know that the assumption of the title Britannicus was associated with
the suppression of a native rising (see infra).

THE OLDER AUTHORITIES 13

upon in great alarm sent Ulpius Marcellus against
them. . . . Marcellus inflicted heavy punishment on
the barbarians in Britain."1

The name of Ulpius Marcellus, regarding whose
stern and austere habits Dio has much to tell, occurs on
an altar which was found at Benwell, on Hadrian's Wall,
and which seems to belong to the joint reign of Marcus
and Commodus (A.D. i77-i8o).2 As he was also gover-
nor under Commodus alone, we may conclude that he
was in office when Marcus died. The exact date ot
the campaign is fixed by numismatic evidence. Coins
of A.D. 184 refer to victories won in Britain — the type
being a seated figure of Victory, with the legend VICT.
BRIT. (Plate II. 4) — and in the same year the emperor
adds " Britannicus" to his titles. Similar allusions occur
on coins of A.D. 185, when the type of Britain subdued
was again employed (Plate II. 3), and also on coins
of A.D. 1 86, so that there was probably a second and
third series of successful engagements. As the extract
from Xiphilinus shows, the area of disturbance was
once more the north of the province. But we cannot
determine from the language used whether the "wall"
spoken of was that on the lower or that on the upper
isthmus. Nor does the literary evidence make it
clear whether Calpurnius and, after him, Marcellus
succeeded in pacifying the country as far as the
barrier raised by Antoninus Pius. As we shall see,
the results of archaeological enquiry plainly suggest
that one or both of them failed in this task, and that
thereafter the Romans had to be content with securing
against the inroads of the barbarians the territory
near to Hadrian's limes.

1 Dio, Bk. Ixxii. c. 8. 2 C.LL. vii. No. 504.

i4 THE LITERARY TRADITION

Next follow several years of chequered Roman
history, during which, although Britain occasionally
played a part — at one great crisis an important part
— in the imperial drama, the ancient writers are almost
silent regarding the northern tribes. In A.D. 197,
when Septimius Severus had disposed of the last of
his rivals, Virius Lupus became governor of Britain.
About that time Lupus was compelled to purchase
peace, Xiphilinus tells us, from a northern tribe called
the Maeatae, whom the Caledonians, contrary to
pledges they had given, were preparing to assist.1 The
garrison along the frontier can have been barely suffi-
cient for purposes of defence. Nor will this seem a
matter for surprise, if we remember that it was Britain
that supplied the chief part of the great army with
which Clodius Albinus fought with Severus at Lyons
for the mastery of the Roman world. Unless the
numbers of his host are greatly exaggerated, the pro-
vince must have been largely denuded, not only of its
Roman garrison, but also of the flower of its own
youth who were of martial age.

The life of Severus has been narrated at considerable
length by Dio, Herodian and Spartian. Brief notices
of his actions have also been given by Eutropius,
Aurelius Victor and other annalists. Dio was, as we
know, a member of the Roman Senate, and for many
years in close touch with the various departments of
state. Herodian, who was also a Greek, appears to
have spent part of his life in Rome collecting materials
for his work. According to his own preface, he was
a model historian — truthful, impartial, careful. His
History which begins with Commodus, takes us down

1 Dio, Bk. Ixxv. c. 5.

THE OLDER AUTHORITIES 15

to the reign of Gordian III., and thus directly chal-
lenges comparison with that of Dio. The result is not
favourable to Herodian, who was too much of a stylist
to live up to his own professions. But, like Dio,
he was a contemporary of Severus, regarding the
incidents of whose reign he has therefore a special
claim to be heard. Of Spartian we have already
spoken, in discussing the Historia Augusta.

Dio's account of the doings of Severus was con-
tained in the lost books. We are able, however, to
some extent to recover it from Books Ixxiv.-lxxvi. of
Xiphilinus's abridgement. " Among the [still unsub-
dued] Britons," writes Xiphilinus, " the two most
important tribes are the Caledonians and the Maeatae ;
the names of the others have practically been absorbed
in these. The Maeatae dwell close by the wall
(SiaT€i-^ia-fjLa) that divides the island into two parts, the
Caledonians beyond them."1 The passage goes on to
describe the country of both tribes as full of rugged and
waterless mountains, and of barren and marshy plains,
destitute of cities and inhabited by peoples that have
barely reached the pastoral stage of civilization. Their
warriors fight from chariots, and have small swift
horses, while their infantry are very fleet of foot and
very steady in their ranks. Here again, it will be
observed, Xiphilinus speaks, as he did in an earlier
connection, as if there were at this time but one
separating wall. It would appear that Dio, whom he
must be supposed to follow, knew of no more than one
wall, or else that any knowledge that he had of two
was so indefinite that he was unable to distinguish
between them.

1Dio, Bk. Ixxvi. c. 12.

1 6 THE LITERARY TRADITION

It might at first sight seem as if Dio's description of
the physical features of the country occupied by the
Caledonians and the Maeatae, and of the degree of
civilization those tribes had reached, would suit the
territory and the people north of the upper isthmus
better than those between the walls, where the country
is less mountainous and the population had been
directly or indirectly subjected to Roman influences
for about a century. If this were so, the 8iarei-^i(T/j.a of
Dio would be the Vallum of Pius. But it must not be
forgotten that much of the district covered by the
Southern Uplands remains desolate and dreary to this
day. Speaking of the barrier on the lower isthmus,
Professor Haverfield says : " North of that wall, wild
moors and wastes and mosses, trodden to this day
by few but the sportsman or the shepherd, stretch far
into Scotland, and there, on the Cheviots and the
Lammermuir hills, in Ettrick forest, and the mosses
of the south-west is the country which Dio describes
as the home of the Maeatae."1 Even at the worst it
might fairly be argued that Dio had in his mind for
the moment Caledonia and its rugged heights, the
territory of the most formidable tribes that remained
independent, and that he merely extended what was
more particularly characteristic of it to the whole of the
land on the north of Hadrian's limes. There is cer-
tainly nothing in his words to prove that the line of
the Forth and Clyde was in Roman hands at the
accession of Severus.

Herodian applies to the unconquered tribes in the
north of the province the general name of Britons,
not Caledonians. He tells us that the emperor, in

^ Antonine Wall Report, p. 158.

PLATE II

10

COINS RELATING TO BRITAIN . (PIUS, COMMODUS, AND SEVERUS)

THE OLDER AUTHORITIES 17

advancing to attack them, "passed beyond the streams
and mounds of earth that form the defences of the
Roman empire."1 There can hardly be a doubt that
these mounds (x^ara) are identical with the wall
(£ioT«/x<0yta) of Xiphilinus. Herodian also describes
the still unconquered Britons as a warlike and blood-
thirsty race of barbarians, inhabiting a land many parts
of which were so constantly flooded by the tidal action
of the sea as to become marshy and impassable.2 The
two historians agree generally in the account they give
of the northern expedition of Severus. Its immediate
occasion was one of the chronic outbreaks of revolt.
Its special character was due to the emperor's resolve
that punishment should be inflicted on a scale of
unexampled severity. He took the command in
person, and carried the enterprise to a successful con-
clusion in face of tremendous difficulties. On passing
beyond the frontier defences, he found the enemy in
force, but not disposed to risk a pitched battle. When
pursued by the Romans, they retreated to thickets and
marshes, into which it was dangerous to follow them.
By such guerilla tactics on the part of the Britons the
war was far more protracted than had been antici-
• pated.3 " The Romans," says Xiphilinus, "underwent
unspeakable toil in clearing their way through forests,
in levelling heights by digging, in filling up morasses,
in bridging rivers."4 The Caledonians, he avers,

Se TOV (rrparov TO. Trpof3ef3\r)p,fva /aev/xara re nal
'Pw/Acuwv apxys (Herod. Hist. Bk. iii. c. 48). In view of the ' Siarei-
' of Dio, it is very doubtful whether it would be safe to press the
force of the plural 'x**/**70' used by Herodian, and to interpret it as
referring to two walls — the English and the Scottish.

2 Hist. Bk. iii. c. 47. * Ibid. c. 48. 4 Dio, Bk. Ixxvi. c. 13.

B

1 8 THE LITERARY TRADITION

drove sheep and oxen within sight of the Romans for
the express purpose of luring them on to destruction.
Scattered parties were ambuscaded, and the floods
took a heavy toll in human lives. Those who fell out
of the ranks from exhaustion were despatched by their
own comrades to prevent their being left to the mercy
of the enemy. From all these causes the losses of
the expeditionary army were very severe. Indeed,
Xiphilinus reckons them at the incredible number of
50,000 men.

In spite of these difficulties and of his own growing
physical weakness, Severus succeeded in forcing his
way to the extremity of the island. According to
Xiphilinus, he ultimately compelled the Caledonians
to come to terms, and to cede a considerable portion
of their territory. The peace so sealed was, however,
but short-lived. The Maeatae took up arms again
almost immediately. Severus despatched troops against
them, with instructions to adopt the sternest measures
of repression. This was to be a war of extermination.
In his address to his soldiers he quoted the lines of
Homer, in which Agamemnon urged his brother to
give no quarter to the Trojans.1 But the rising was
not to be so easily quelled. The Caledonians threw in
their lot with their kinsmen, and the emperor realized
that he must once more take the field himself. While
he was busy with preparations for a second campaign,
he was carried off by disease in February A.D. 2ii.2
His son Caracalla promptly gave up the struggle
and withdrew from the enemy's country, abandoning
certain fortified posts which had been established
within it.3

1 Iliad, vi. 57 ff. 2 Dio, Bk. Ixxvi. c. 15. 3Dio, Bk. Ixxvii. c. i.

THE OLDER AUTHORITIES 19

So far we have been following Xiphilinus, and
therefore presumably Dio. Herodian knows nothing
of any cession of territory or of any need for a second
campaign. He writes almost as if he believed that
Severus died in Scotland and not at York. Of the
two versions, that of Dio, abridged though it be, is
the more circumstantial and (it can hardly be doubted)
the more trustworthy. The evidence of coins confirms
his statement that a first campaign ended in a definite
success. On pieces struck in A.D. 211 Severus
assumes the title " Britannicus," while the number of
varieties that ring the changes on VICTORIAE BRITTAN-
NICAE leaves no doubt as to the importance which he
wished to attach to his achievements (Plate II. 5-10).
At the best, however, the sum of our knowledge
amounts to little. That the objective of Severus's
expedition was Caledonia, and that he penetrated
far into its mountainous and marshy wilds before
he began to retreat, is nearly all the positive infor-
mation regarding it we get from either historian.
Neither makes any allusion to the building of a
wall.

Spartian's references to the actions of Severus in
Britain are few and brief. But one of the statements
he does make is the much discussed assertion that, in
order to defend the province, the emperor built a wall
(murus) across the island, that this wall ran from ocean
to ocean, and that it was the greatest glory of his
reign.1 In another passage we are told that, after
subduing the hostile tribes in Britain, Severus died at

1 Brittaniam.) quod maximum ejus imperil decus est, muro per trans-
•versam insulam ducto utrimque ad finem Oceani munivit ( Vit. Sev.
18, 2).

20 THE LITERARY TRADITION

York in A.D. 211, worn out by disease.1 Again, in nar-
rating some omens that gave warning of the approach-
ing death of Severus, Spartian uses the expression
"post maurum apud vallum missum in Britannia"*
It is more than likely that this contains some allusion
to a wall in Britain, but the passage is corrupt, and its
full meaning unintelligible. In the De Caesaribus of
Sextus Aurelius Victor, written, it is supposed, about
A.D. 360, we find the information that Severus built a
wall across Britain repeated in language virtually
identical with that employed by Spartian.8 We are
also told in an epitome (by an unknown hand) of
another work ascribed to Aurelius Victor that Severus
drew a rampart (vallum) in Britain for thirty-two miles
from sea to sea.4 Almost the same words are found
in Eutropius, who wrote shortly after Victor, but the
MSS. of his History vary as to the length of the wall
or rampart, some giving it as xxxii, others as cxxxii
miles.6 Later chroniclers, like the ecclesiastical writers
Jerome (continuing Eusebius) and Paulus Orosius, as
well as the statesman Cassiodorus, notice this 'vallum
of Severus, and all give it a length of cxxxii miles.
But, as the words they use are almost the same as those
of Eutropius and the epitome of Aurelius Victor, it is
plain that the story must have been taken either from
one or other of these writers, or from some common
source. And such obvious borrowings can have no

1 Vit. Sev. 19, i. 2 Ibid. 22, 4.

3 Britanniam^ quae ad ea utilis erat, pulsis hostibus^ muro munimt, per
transversam insulam ducto, utrimque ad finem Oceani (De Caesar.
c. xx. 4).

* Hie in Britannia vallum per triginta duo passuum millia a mart ad
mare deduxit (Epitome^ c. 20).

6 Hist. Rom. Bk. viii. c. 19.

LATE CLASSICAL WRITERS 21

value as independent evidence. As to the statement
itself, various explanations of it have been given. It
was long interpreted as referring to the stone wall on
the lower isthmus, and this is a theory that has lately
been revived on a new and more solid substructure of
archaeological fact. Dr. W. F. Skene1 held that it
applied to repairs made on the turf wall of the upper
isthmus, a view which was independently advocated
by Mommsen when he wrote his Roman Provinces?

For seventy years or so after the death of Severus,
the Roman historians preserve an almost unbroken
silence regarding Britain. Their attention was again
directed to it by the romantic adventure of Carausius,
who seized the island and proclaimed himself emperor
in A.D. 286. After a prosperous reign of seven years
he was assassinated by Allectus, one of his principal
officers. The usurper maintained his independence for
three years ; but he was finally defeated and slain by
the forces of Constantius Chlorus who, under the
arrangement made by Diocletian, regarded Britain as
part of his dominions. Constantius died at York in
A.D. 306, on his return from an expedition against the
Picts (as the great enemies of Roman rule in northern
Britain are henceforth called), leaving his share of the

1 Celtic Scotland, vol. i. 89 ff. footnote ; The Historians of Scotland, vol.
iv. pp. 391 f.

2 Eng. Transl. i. p. 187, footnote. Both Skene and Mommsen rely
mainly on the statement of distance. They assume xxxii to be the
correct figure, and argue that, while this might suit the Scottish Wall, it
is out of the question to try and apply it to the English one. But xxxii
is not very appropriate even for the Vallum of Pius. On the other hand,
it is quite conceivable that the original from which all these passages are
ultimately derived may have had Ixxxii, which would be very suitable for
the Tyne and Solway isthmus ; see Horsley, Brit. Rom. p. 62, and
Haverfield, Appendix to E.T. of Mommsen's Roman Provinces, ii. p. 352.

22 THE LITERARY TRADITION

government to his son, famous in history as Constantine
the Great. For the next half century the annals of
the island as a whole are virtually a blank. Of the
relations that subsisted with the frontier tribes we
know absolutely nothing that is certain. We are safe,
however, in assuming that the pressure steadily in-
creased. The empire was tottering to its fall and,
after a few fitful efforts to defend its British subjects,
made during the latter half of the fourth century and
once or twice renewed afterwards, it was compelled to
abandon them to their fate.1

Some of the events of this period are related by
Ammianus Marcellinus, a contemporary writer, with
whom the line of Roman historians may be said to
end. He tells us that in A.D. 360 Julian, the Emperor
of the West, sent Lupicinus to Britain to drive back
the Picts, with whom were now allied the Scots from
the west.2 According to Ammianus,3 the Picts were
divided into two sections, — the Dicalidonae and the
Vecturiones. Besides being harassed by the Picts
and Scots, the Romans were assailed by a warlike,
perhaps cannibal,4 tribe called the Attacotti, and also

1 Regarding the real nature of the ' abandonment ' see, however,
Oman, England before the Norman Conquest, pp. 174 ff.

2 Cum Scotorum Pictorumque gentium ferarum excursus, rupta quiete
condicta, loca limitibus vicina vastarent (Am m. Marc. Bk. xx. c. i, i).

3 Bk. xxvii. c. 8, 5.

4 Gibbon {Decline and Fall, ch. xxv.), following Jerome, says : "When
they hunted the woods for prey, it is said that they attacked the shepherd
rather than his flock ; and that they curiously selected the most delicate
and brawny parts both of males and females, which they prepared for
their horrid repasts." His comment is quaintly characteristic : "If, in
the neighbourhood of the commercial and literary town of Glasgow, a
race of cannibals has really existed, we may contemplate, in the period of
Scottish history, the opposite extremes of savage and civilized life. Such
reflections tend to enlarge the circle of our ideas ; and to encourage the

LATE CLASSICAL WRITERS 23

by the Franks and Saxons who were already attempt-
ing, more or less successfully, to take possession of the
eastern and southern shores of the island. Lupicinus
had been recalled before accomplishing much ; and
Valentinian, who had become emperor in A.D. 364,
sent his ablest general, Theodosius, to take his place.
So completely was the greater part of Britain by this
time at the mercy of the enemy that the safety of even
London — described as "an ancient town that has
since been named Augusta " — is said to have been
gravely menaced.1

Landing at Richborough with reinforcements, the
new commander at once assumed the offensive and drove
back the marauders, restoring cities and fortresses, and
establishing stations and outposts on the frontiers,2 —
" tile Caledoniis posuit qui castra pruinis" to quote the
eulogy of Claudian.3 Ammianus Marcellinus adds
that the province again brought under the authority of
its legitimate rulers was from that time forth called
Valentia, by desire of the emperor.* Influenced
perhaps by Claudian' s picturesque phraseology, some
have argued that the cities and fortresses of which
Ammianus speaks were on the lower, and the stations
and outposts on the upper isthmus.5 That, however,

pleasing hope, that New Zealand may produce, in some future age, the
Hume of the Southern Hemisphere."

1Amm. Marc. Bk. xxvii. c. 8, 6.

* Instaurabat urbes et praesidiaria, ut diximus, castra^ limitesque
•vigiliis tuebatur et praetenturis (Amm. Marc. Bk. xxviii. c. 3, 7).

3 De IV. Cons. Hon. 1. 26.

4 Bk. xxviii. c. 3, 7.

6 It will be noted that in two of the passages quoted above Ammianus
uses the plural limites. That, however, lends itself even less readily to a
forced interpretation than does the xw/xara of Herodian (see supra, p. 17) ;
limites were not necessarily frontier-lines (see chapter iii. infra).

24 THE LITERARY TRADITION

is pure conjecture. Even if it be accepted as a probable
explanation of the historian's words, it does not warrant
the assumption, first made by Gale,1 then repeated
by Father Innes,2 and in Bertram's spurious De Situ
Brttanniae, and so received as an article of popular
belief, that the province or division of the island which
now received the name of Valentia was the territory
lying between the two walls.

The end is approaching rapidly. Honorius, son
of the Emperor Theodosius and grandson of the
Theodosius mentioned above, became ruler of the
West in A.D. 395. Stilicho, the able adviser of his
father, was appointed his chief minister, and earned
lasting fame by a vigorous, and for a time successful,
attempt to stem the tide of barbarism that was rolling
in from almost every quarter. We owe our knowledge
of his achievements largely to the panegyrics of the
poet Claudian, who throws round them a veil of
imagery that effectually conceals the real truth. But it
seems at least to be certain that Stilicho did something,
or caused something to be done, which brought a
measure of relief to the Romanized inhabitants of
Britain. The respite, however, was brief. A legion
was withdrawn in A.D. 402. Four years later the
remaining legions revolted, and proclaimed as emperors
one general after another — Marcus, Gratianus, and
finally Constantine, " the tyrant." The last-named
did not find Britain large enough for his ambition.
Prompted by the desire to make himself master of
Gaul and Spain as well, he crossed to the Continent

1 Horsley was more cautious : see Brii. Rom. p. 73.

2 Critical Essay on the Inhabitants of the Northern Parts of Britain
(1729), i. p. 21.

LATE CLASSICAL WRITERS 25

in A.D. 407, taking with him all the troops he could
muster, and leaving the provincials to the tender
mercies of their northern foes. From motives of
policy Honorius had acknowledged Constantine as a
colleague. But in A.D. 410 war broke out between
them, and in 411 the latter was defeated and put to
death in captivity. Honorius, finding himself unable
to give the Britons all the assistance they required to
enable them to repel the attacks of the Picts and
Scots, did not even make a pretence of sending help.
The garrison, withdrawn by Constantine, was never
subsequently replaced. Britain was thus among the
first of the great provinces to be abandoned by the
Romans.1 It may have been some little time before
the people fully realized what had happened. But from
that epoch the island ceased to be, except nominally,
a part of the Roman Empire.

Another work must be glanced at before we leave
the older literary authorities behind. The Antonine
Itinerary (Itinerarium Antonini Augiisti) is a list of
the principal roads and cross-roads throughout the
whole empire. These are indicated by means of
stations and places situated on them, the distances
being recorded in Roman miles. It is generally
thought to have been first compiled under Pius or
Caracalla, but it bears evidence of having undergone
various recensions down to the reign of Diocletian
(A.D. 284-305). Two of the fifteen roads in Britain
reach the lower isthmus. The second Iter, which is

1 Dacia had gone about 270, Northern Gaul, etc., in 406. According to
Zosimus (Bk. vi. c. 6) what happened in our own island was that the
Britons revolted, expelled the Roman officials, and established a form of
government of their own, instead of proclaiming another emperor. See
Oman, England before the Norman Conquest, p. 174.

26 THE LITERARY TRADITION

by far the longest of these, runs in zig-zag fashion
from Rutupiae (Richborough, in Kent) to Luguvallium
(Carlisle), from which it is continued for 12 miles to
Castra Exploratorum (usually identified with Nether-
by), and thence for 12 miles more to Blatobulgium
(apparently Birrens), where it stops. The first Iter
approaches the southern wall at Corstopitum (Cor-
bridge-on-Tyne), whence it proceeds for 20 miles
to Bremenium (High Rochester) and is not carried
further. The testimony of the Itinerary, therefore,
like that of the great majority of the authors whom we
have passed in review, is purely negative, so far as our
immediate purpose is concerned. If we eliminate
Tacitus's brief reference to the forts erected by
Agricola, and the bare statement of Capitolinus as to
Lollius Urbicus having built a wall after driving back
the barbarians, there remains literally nothing that can
be definitely and unmistakably associated with the
Forth and Clyde isthmus.

Turning next to the earliest native historians, we
find that what they have to tell us regarding the Roman
occupation is curious rather than valuable. The oldest
of them, who is also the most ancient British author
known to our own or to mediaeval times, is Gildas,
surnamed the Wise. Bede and Alcuin call him Gildus.
A few facts regarding his personal history can be
gleaned from his writings.1 He was apparently a
native of what is now Wales, and he seems to have
been in priest's orders. His date is fixed by a well-
known, if somewhat obscure, passage, in which he

1 See Mommsen's Monumenta Germaniae Historica : Chronica Minor a,
vol. iii. pp. 4 ff.

GILDAS 27

speaks of himself as writing forty-four years after "the
year of the siege of Mount Badon," which (he adds) was
also the year of his own birth.1 As Maglocunus, one
of the five kings against whom his denunciations are
levelled,2 is known from other sources to have died in
547, it follows that Gildas must have been born before
504. We may safely place his floruit about 540.
Several lives of him have come down to us. Even
the oldest of these, however, was not written till the
beginning of the eleventh century. It is therefore
hardly surprising to find in them statements so contra-
dictory that some have supposed them to be biographies
of different individuals to whom chance had given the
same name.

The book that has made him famous is the De
Excidio et Conquestu Britanniae, the subject of which
had occupied his thoughts for ten years. It is usually
divided into a hundred and ten chapters, of which the
first two are introductory, and the next twenty-four
historical, while the remainder form one long and
uninstructive wail, mainly in Scriptural language, over
the sins and follies of the Britons, five of whose kings
or chiefs he singles out for special invective. There
being, as he tells us, no native records in existence, or
at least none left within the island, Gildas was obliged
to draw upon foreign writers, whose notices of Britain
he found fragmentary and far from clear.3 It is not
•easy to imagine who the foreign writers were. Beyond
some loose but obvious reminiscences of Virgil, he

1 De ExtidiO) etc. c. 26. 2 Ibid. c. 33.

3 Non tarn ex scriptis patriae scriptorumve monimentis . . . quant
transmarina relatione^ quae crebris inrupta intercapedinibus non satis
claret (De Excid. etc. c. 4).

28 THE LITERARY TRADITION

betrays small knowledge of any classical authors,
although it seems possible that he was familiar with
Juvenal, Persius, Martial and Claudian. His state-
ment of the latitude and longitude of Britain may have
been taken from Orosius, and it is certain that he had
read the Ecclesiastical History of Eusebius in the
Latin translation of Rufinus, as well as the Chronicles
of St. Jerome.1 His references to these works are,
however, few and unimportant. The one book from
which he does quote freely is the Bible. The style of
the De Excidio is turgid and often difficult, and its
tone throughout fully warrants the use of the name
that has been applied to it — The Book Querulous.

It is during the reign of Maximus that Gildas first
brings his readers into touch with North Britain. This
emperor, when setting out on the continental campaign
that was to prove his ruin (A.D. 383), took with him
not only the Roman soldiery, but the flower of the
native youth — a circumstance to which the subsequent
miseries of the islanders are to be traced. The en-
feebled remnant was an easy prey to the cruelty of
two foreign nations, " the Scots from the north-west
and the Picts from the north." They sent ambassadors
to Rome, protesting continued loyalty to the empire
and begging assistance. A legion was despatched
in response to this appeal, and the invaders were
speedily expelled from British territory. With a view
to future defence the Britons, at the suggestion of their
deliverers, built a wall that stretched from sea to sea
across the island.2 Owing, however, to the absence of
any competent supervision, they foolishly made it

1 See Mommsen's Chronica Minora, Hi. pp. 6f.

2 Inter duo maria trans insulam murum (De Excid. etc. c. 15).

GILDAS 29

of turf instead of stone, so that it proved of little use.
On the withdrawal of the legion, the Picts and Scots
renewed their predatory incursions, breaking through
the boundaries and overrunning the whole country.
Again the Britons implored the Romans for help, and
again their request was granted. A Roman army and
fleet quickly drove Picts and Scots beyond the seas
across which they were wont to transport the plunder
seized in their annual inroads.

On this second occasion the Romans, before quit-
ting the country, informed the Britons that they were
not prepared to render them further aid, as the strength
of the empire was not to be wasted on distant expedi-
tions. They constructed, however, with the assistance
of native labour, a wall of stone that ran " in a straight
line from sea to sea between cities that happened to
have been built at that place through fear of the
enemy." l Towers were also erected at regular
intervals along the south coast, and then the Romans
bade Britain a final adieu. Learning of their departure,
the northern marauders renewed their attacks. In
assailing the wall they used weapons provided with
hooks, by which the unhappy defenders were dragged
down and hurled to the ground. Finally wall and
"cities" were alike abandoned, and the Britons
scattered in flight. Once more an appeal was made
to Rome. The letter ran thus : " To Agitius, consul
for the third time, the groans of the Britons. . . . The
barbarians drive us to the sea ; the sea throws us back

1 Murum non ut alterum, sumpto publico privatoque adjunctis secum
miserabilibus indigenis, solito structurae more, tramite a mart usque ad
mare inter urbes, quae ibidem forte ob metum hostium collocatae fueranty
directo librant (De Excid. etc. c. 1 8).

30 THE LITERARY TRADITION

on the barbarians ; between these two kinds of death
we are either slain or drowned." * Even this bitter
cry failed to move the Romans from their resolution.
Nerved by the courage of despair, the Britons faced
their foes unaided, and gained some measure of
success. The respite was but temporary. Ere long
they were forced to invite the assistance of the Saxons,
and so put themselves at the mercy of a still more
ruthless foe. The " Agitius " of the letter is evidently
the famous Flavius Aetius who from 430 to 454 was
the real master of the Western Empire. As Maximus
was killed in 388 and Ae'tius enjoyed his third consul-
ship in 446, Gildas must have compressed into his
short narrative the events of fifty-eight years.

No doubt the story as given above is, in a very
general way, a true enough account of the series of
events that led to the Saxon conquest ; the Romanized
Britons were crushed between the upper and the nether
millstones. The details, however, are utterly untrust-
worthy. Even when allowance is made for the fact
that Gildas wrote more than a hundred years after the
events he professes to describe, his ignorance is so
surprising as to suggest a violent break in the tradi-
tion.2 By his own admission he could find no native
authorities to guide him. The foreign sources of
which he speaks must have been unsatisfactory at
the best. Practically none of them can now be re-
cognized. Possibly they are lost. Possibly he trusted
to his memory, and did not therefore reproduce the
exact words of such as he may have quoted. Not
that he is likely to have found his narrative of the

1 De Excid. etc. c. 20.

2 See Haverfield, Romanization o Roman Britain, pp. 31 f.

GILDAS AND BEDE 31

building of the walls in any foreign writer. Tales
spring up quickly round monuments that are con-
spicuous landmarks. It is not improbable that the
Roman soldiers who wrought at the walls of Hadrian
and Pius may have employed natives to fetch and
carry for them at their task. It is barely conceivable
that a tradition of this may have lingered long enough
to grow into the story Gildas reproduces. But such
an hypothesis, even if it be admissible, is not necessary
to account for the phenomenon. After all, the legend
he sets forth is a less strange perversion than that
which represented the Tower of London as having
been built by Julius Caesar.1

The next witness who demands a hearing is the
Venerable Bede. According to his own simple record
of his uneventful life, he was born about A.D. 672
within the " territory" of the monastery of Wearmouth
and Jarrow. At the age of seven he was placed in
charge of the abbot to be educated. " Since that
time," he proceeds, " I have spent the whole of my
life within the said monastery, giving all my mind to-
the study of the Scriptures ; and, amid the observance
of the discipline of the order and the daily care of
singing in church, I have ever found delight in learning,
in teaching, in writing."2 This modest sentence sums
up years of deep and earnest study. Bede had the
instincts of a scholar, and the list of authors whom he
quotes shows the width of his learning.3 For his
knowledge of the later doings of the Romans in
Britain he depends largely upon Gildas. For the

1 Shakespeare, Richard HI. iii. I, 69 ff., and Richard II. v. I, 2.

2 Hist. Eccles. Bk. v. c. 24.

3 See Plummet, Baedae Opera Historica, vol. i. pp. 1 ff.

32 THE LITERARY TRADITION

earlier period he follows Eutropius and Orosius.
Tacitus he had apparently never heard of. Of Dio
and Herodian, who wrote in Greek, he naturally knew
nothing.

Two of Bede's works contain passages that concern
us. In the earlier of these, the Chronica, we find
little regarding the walls save a repetition of the
ipsissima verba of Orosius and Gildas ; in the later,
the Historia Ecclesiastica Gentis Anglorum, written
about 731, we have some interesting modifications.1
In recounting the actions of Severus he practically
confines himself to the statement that that emperor
41 drew from sea to sea a great ditch and a very strong
rampart, fortified at frequent intervals by towers." 2
The length is given in the Chronica 3 as cxxxii
miles. This detail is omitted in the Historia, where
the original version of Orosius is amplified by an
explanation (identical, as Dr. George Neilson has
pointed out,4 with that of Vegetius) as to the distinc-
tion between a vallum, or rampart of sods, and a
murus, or wall of stone,6 and where it is clearly
indicated that the rampart in question is the 'Vallum'
which to-day constitutes one of the chief problems of
the series of works on the lower isthmus.6 In another
chapter of the same book, in speaking of the final
severance of Britain from the empire in 410, he tells

1 Attention is drawn to the significance of these in The Antonine Wall
Report^ pp. 20 ff.

2 The words are identical with those of Paulus Orosius (Hist. vii. 17).

3 345 (Mommsen's Chronica Minora, vol. iii. p. 289).

4 Antonine Wall Report, p. 22. The identity does not necessarily
imply direct borrowing. It rather suggests a glossary.

*ffist. Eccles. Bk. i. c. 5. * Ibid. c. 12.

BEDE 33

us that it was only the country south of the barrier of
Severus that the Romans had occupied in any real
sense, although they exercised a right of suzerainty
over the rest of the mainland and also over the adjacent
islands.1

The chief point in which Bede differs from Gildas,
in his description of the help subsequently given to the
Britons by the Romans, is that he definitely localizes
the two walls built across the island for defence. He
adopts from his authority the epithet " foreign " (trans-
marinas] as applied to the hostile tribes, the Scots
from the north-west and the Picts from the north. He
adds, however, that they were called "foreign," not
because they did not live in Britain, but because they
were separated from the Britons by two arms of the
sea running far into the land, the one from the eastern,
the other from the western ocean, so as to leave a
small isthmus between them. "In the midst of the
eastern firth is the city of Giudi, while the western
has above it, on its right bank, the city of Alcluith,
which means in their tongue the rock of Cluith, that
being the name of the river near which it stands."2
The Forth and Clyde are easily recognizable. Where
Giudi was, no man knows ; it may have been a settle-
ment on Inchkeith or on Cramond Island, although
George Buchanan would have it that it was Camelon.3
Alcluith is generally assumed to have been Dumbarton.
It was here, according to Bede, that the Britons
erected their first wall, that built of turf. " Of this
work erected there," he continues, " that is, of a
rampart of great breadth and height, there are very
evident traces to be seen to this day. It begins about

1 Ibid. c II. *Ibid. c. 12. * Rerum Scot. Hist. Bk. i. c. 21.

C

34 THE LITERARY TRADITION

two miles west of the monastery of Aebbercurnig, at a
place called in the Pictish language Peanfahel, but in
English Penneltun, and running westward ends near
the city of Alcluith."1 It is plain, then, that Bede
identified the useless wall of Gildas with the remains
of the Vallum of Antoninus Pius. Similarly, he found
the more substantial wall of stone in Northumberland
and Cumberland, for it is to the English Wall that his
description obviously applies. " This still famous and
conspicuous wall [the Romans] erected by the expen-
diture of public and private funds, a body of Britons
lending their assistance. It is eight feet in breadth
and twelve in height, in a direct line from east to west,
as beholders can see to this day."2

Bede's estimate of the dimensions of the English
wall is interesting. It seems to have been in much
better preservation in his day than it is now. And
we can hardly doubt but that he had seen it. We
must not take too literally what he says about spend-
ing his life within the walls of the monastery. He
must at least have journeyed frequently between
Wearmouth and Jarrow, and it is certain that he
had visited Lindisfarne and York.3 Whether he had
ever travelled as far as the Forth and Clyde isthmus,
it is impossible to say. The language he employs
regarding the wall there is vague, and may embody

1 Hist. Eccles. Bk. i. c. 12. The passage was made so much use of
by later writers that it may be well to quote it in full : Cujus operis
ibidem facti^ id est valli latissimi et altissimi, usque hodie certissima
vestigia cernere licet. Incipit autem duorum ferme milium spatio a
monasterio Aebbercurnig ad ocddentem in loco, qui sermons Pictorum
Peanfahel^ lingua autem Anglorum Penneltun appellatur; et tendens
contra ocddentem terminatur juxta urbem Alcluith. Aebbercurnig is,
of course, Abercorn on the Forth. Peanfahel is Kinneil ; see infra p. 147.

2 Ibid. 3 Plumtner, Baedae Opera Historica, vol. i. p. xvi.

BEDE AND NENNIUS 35

the results of enquiry rather than of personal observa-
tion. In any case, such value as attaches to Bede's
references to the two walls depends entirely upon his
knowledge, direct or indirect, of their actual condition
in his own day. What he says as to their history is
merely copied from Gildas. The legend doubtless
seemed to him to account for the archaeological data ;
he accepted it as a plausible explanation of the exist-
ence of the two ruined lines of defence, and of the
circumstance that the one lying to the north was built
of turf and not of stone.

The Historia Brittonum, which passes under the
name of Nennius, has been the subject of much keen
controversy among scholars. Into the complicated
questions that the study of the text has suggested, it
is unnecessary to enter here.1 The book is apparently
a composite work. In its original form it probably
goes back to a date earlier than Bede. Subsequent
recensions, including that by Nennius, who is respon-
sible for a prologue, are considerably later. The
account given of the doings of the Romans in Britain is
brief. A passing notice is taken of the achievements of
Julius Caesar and Claudius. The third emperor who
crossed the strait to Britain, we are to believe, was
Severus. In order to secure certain territory which
he recovered, he " drew a wall and a mound from sea
to sea through the whole breadth of Britain, that is,
for cxxxii miles, and it is called in the British tongue
Guaul."2

1 For a full discussion of them see H. Zimmer, Nennius Vindicatus
(Berlin, 1893), a°d Mommsen, Chronica Minora, iii. pp. 114 ff.

- Murum et aggerem a mari usque ad mare per latitudinem Brittanniae,
id esl per cxxxii milia passuum deduxit, et vocatur Brittanico sermone
Guaul (Hist. Britt. c. 23). ' Gttau/' is Cymric for ' Vallum!

36 THE LITERARY TRADITION

This, it will be seen, is little more than a repetition
of a statement with which we are already familiar.
Only a single wall is mentioned, and no hint as to its
situation is given, unless the mention of a mound
(aggerem) can be held to connect it with the double
lines on the lower isthmus. In some manuscripts,
however, there is a marginal addition which identifies
the wall of Severus with the Forth and Clyde rampart,
and asserts that it was rebuilt and fortified subsequently
by ' Carutius,' who is obviously Carausius.1 The
author of the Irish version of Nennius knows of two
walls, or at least two aggeres, but he attributes both
to the energy of Severus.2 Regarding the fortunes of
the Romans in Britain after the death of Maximus, the
Historia Brittonum has practically nothing to tell us
except what we have already found in Gildas.

It seemed advisable to deal in some detail with the
earliest British historians, because there was, at least
in the case of Gildas, a possibility that the information
given might be of value as embodying the current
traditions of a time sufficiently near the events them-
selves to have preserved a lingering reflection of the
truth. Although the result has been disappointing, it

1 The gloss deserves quotation in full : Per cxxx vero miliaria /., id
est a Penguaul, quae villa Scottice Cenail^ Anglice vero Peneltun dicitur,
usque ad ostium fluminis Cluth et Cair Pentaloch, quo murus ille finitur
rustico opere, Severus ille praedictus construxit^ set nihil profuit. Carutius
•bostea imperator reedificavit et vii castellis muni-vit inter utraque ostia
domumque rotundam politis lapidibus super ripam fluminis Carun, quoa
a suo nomine nomen accepit, fornicem in victoriae memoriam erigens
construxit (Mommsen, Chronica Mtnora, iii. p. 165, footnote 2). The
influence of Bede is manifest. Cair Pentaloch is Kirkintilloch, and the
fact that the wall is said to end there shows the worthlessness of the
whole extract. The allusion to 'Arthur's O'on' is the oldest extant
reference to that monument. Cenail is, of course, Kinneil.

2 See Zimmer's Latin translation, apud Mommsen, I.e.

JOHN OF FORDUN 37

is not unimportant, as laying bare the groundwork on
which the mediaeval chroniclers raised their imposing
superstructure of fable. We may take John of Fordun,
who lived in the fourteenth century, as a representative
of his class. The emperors who play a prominent part
in his narrative are precisely those a memory of whose
connection with our island had filtered down from late
classical sources through Gildas, Bede, and Nennius.
The main facts, too, are the same, embellished though
they be to an extent that renders them almost unrecog-
nizable. Julius Caesar, according to this veracious
record, not only invaded Britain, but actually reached
the banks of the Carron, where he left an abiding
monument behind him in the shape of the building
known as "Arthur's O'on."1 Claudius is mentioned
shortly, while Hadrian is passed over almost in silence.
Severus, as we might expect, figures as the architect
of a wall.2 Not content, however, with making him
die a natural death at York, the chronicler provides us
with a lively description of how he was slain outside
the city by Fulgentius, the leader of a besieging host of
Picts and Scots.3 These examples will suffice to show
how fruitless it would be to seek truth in such a
tissue of phantasy. Yet there is one point of real

1 Scotichronicon, Bk. ii. c. 16. The Chronicler adds an alternative
explanation of the "O'on" to the effect that "Julius Caesar had this
chamber carried about with him by his troops, with each stone separate,
and built up again from day to day, wherever they halted, that he might
rest therein more safely than in a tent ; but that, when he was in a hurry
to return to Gaul, he left it behind." This is interesting as being probably
a reminiscence of Suetonius, who mentions a story that Caesar ' in expedi-
tionibus tessellata et sectilia pavimenta circumtulisse* (Divus Julius, 46).

2 Scotichronicon, Bk. ii. c. 32. Following Bede, Fordun regarded the
' Vallum ' on the lower isthmus as the wall built by Severus.

3 Scotichronicon, Bk. ii. c. 33.

3 8 THE LITERARY TRADITION

interest to light up the general dullness of Fordun's
narrative. While adopting Bede's account of the
origin and history of the Scottish Wall,1 he tells us that
it was called ' Grymisdyke,' because it was destroyed
by Gryme, the grandfather of King Eugenius.2 The
popular name of ' Graham's Dyke,' not yet entirely
obsolete, is thus older than 1400. What its real
significance may be, we must leave it to philologists
to determine. None of the explanations hitherto put
forward seems to be satisfactory.

An even stranger literary compound than the Scoti-
chonicon is the Scotorum Historiae of Hector Boece,
the first Principal of King's College, Aberdeen, who
died in 1536. Boece was a man of real learning, full
of the new knowledge of the Renaissance ; but he
implicitly accepted the tales told by his predecessors
in the field of Scottish history, and his book is an
elaborate attempt to weave the two into a homogeneous
whole. He gives a prominent place to the various
Roman invasions of North Britain, devoting to them
almost the whole of his third, fourth, and fifth books,
as well as some parts of his sixth and seventh. Amid
much that is as wildly extravagant as anything in John
of Fordun, it is always possible to trace the influence
of his superior mastery of authorities. Hadrian, for

1 Scotichronicon^ Bk. iii. cc. 3 and 4. He adds some fresh details, such
as that the Wall was " finished off at enormous expense, by strengthening
it with towers, at intervals, such that the sound of a trumpet could reach
from one to the other." This last expression was doubtless responsible
for the ridiculous story, mentioned by Nimmo, "that there was a hollow
in [the Wall], through which the sound of a trumpet, blown at one
end, could be conveyed to the other" (Hist, of Stirlingshire, Ed. 1777,
P- 38).

2 Scotichronicon, Bk. iii. c 5.

BOECE AND BUCHANAN 39

instance, comes into his own again as a wall-builder.1
Agricola, too, reappears after being lost for centuries,
and his son-in-law's sketch of his life is made the basis
of a very detailed account of his campaigns and
marches. Boece evidently knew his Caesar and
his Tacitus well. He rejects the legend that Julius
had invaded Scotland, on the ground that such an
expedition is not mentioned in the Commentaries or
in any other work by a reliable Roman historian.2
At the same time he does not hesitate to make
Agricola fortify Stirling Castle.8 Thus madly do
romance and fact chase one another through his
pages.

George Buchanan's Rerum Scoticarum Historia is a
work of a very different character. Buchanan does
homage, it is true, to the mythical kings of Hector
Boece ; it may be that even his independent spirit
shrank from depriving James VI. of so many genera-
tions of royal ancestors. But he shows greater caution
in his account of their exploits. This is especially the
case as regards the period of the Roman occupation.
There is a refreshing sanity about his treatment of his
authorities, the various writers being marshalled in
order, and the appropriate passages quoted in full.
The manner in which he moves over the ground bears
witness to a thorough grasp of the subject-matter, and
leaves a stronger impression of Buchanan's real great-
ness than does almost any other portion of his history.
He was, of course, aware of the existence of the two
walls. The one in England he attributed to Hadrian,

1 Scotichronicori) Bk. v. c. 4. In the quaint words of Bellenden's trans-
lation, the wall was built " of fail and devait."

2 Bk. iii. c. 4. Bk. iv. c. II.

40 THE LITERARY TRADITION

the one in Scotland to Severus.1 The theory — bor-
rowed by Bede from Gildas — that both were of British
origin is tacitly set aside, the inscriptions being inci-
dentally cited as a proof that the Scottish barrier at
least was Roman. At the same time the framework
in which this theory was set is so far adopted. The
Romans are represented as returning from Gaul in
response to an urgent appeal from the hard-pressed
Britons. The invaders were defeated in a bloody
battle on the banks of the Carron, and the Romans
repaired the "vallum Severi" as Buchanan calls it,
before quitting the island again.2 Hardly were they
gone, when the Picts and Scots surged up once more
against it. Their chief champion was Graeme, who
was able, by using sea-transport, to plant a force in the
rear of the wall and break it down so that his comrades
could pass through. Yet another call for help was
addressed to the Romans. A legion accordingly
recrossed the channel. The success of the relieving
force was less pronounced than it had been before.
But the Picts and Scots were ultimately driven back,
and the "vallum Severi" rebuilt, — this time in stone,
8 feet broad and 12 feet high.3 Even so great an
effort proved of no avail. Graeme and his followers
pressed on again, and when peace was made the Picts
and Scots were able to insist on the frontier being
pushed forward to the lower isthmus, — "limes utrisque
esset Adriani vallum." *

1 Rerum Scot. Hist. Bk. i. 22, iv. 29, 30, 37.

2 Rerum Scot. Hist. Bk. v. c. 2.

3 Rerum Scot. Hist. Bk. v. c. 6. Buchanan has lifted this description
bodily from what Bede says about the English Wall, and has transferred
it to the Scottish one.

4 Rerum Scot. Hist. Bk. v. c. 7.

GEORGE BUCHANAN 4r

As a historian of the Roman Wall, then, George
Buchanan has nothing to say that is at once new and
true. But, looked at from another side, his work
marks an epoch in its study. He is the first writer
to show any appreciation of the value of the inscrip-
tions. Unfortunately he does not quote examples.
But he tells us that many have been found, some
being dedications by military officers, others sepulchral
slabs.1 Here is the germ of the archaeological method.
And only a few years after Buchanan's death Scotland
was visited by certain German scholars who were
making a collection of epigraphic material. Cadder and
Kilsyth seem to have been included in their itinerary,
with the result that Scaliger in his Thesaurus Tem-
porum, published at Amsterdam in 1606, was able to
cite three inscriptions from the Vallum of Pius. One of
this advance-guard of the Berlin Corpus Inscriptionum
came, appropriately enough, from Prussia. Another
was a Silesian. The names of these two were Gericke
and Reichel.2

1 Rerum Scot. Hist. Bk. i. c. 22. One of those which he had in view
was doubtless the Dunottar stone : see infra, chap. x. No. 14.

2 Latinized they became Crispinus Gericius and Servatius Rihelius.
The ' Gericius ' of C.I.L. vii. is, however, now known not to have been
Gericke, but possibly Reichel ; see Haverfield's forthcoming supplement
to C.I.L. vii. in Ephem. Epigraph.

CHAPTER II

THE HISTORICAL BACKGROUND: ORGANIZATION
OF THE ROMAN ARMY

AT least one fact has emerged clearly as a result of
our survey of the somewhat confused mass of literary
testimony. The occupation of North Britain by the
Romans must have been mainly military. Even if we
were to set the historians altogether aside, the inscrip-
tions would prove conclusively that the Wall with which
we are dealing was a military work, a defensive line
erected and manned by the troops of the Emperor
Antoninus Pius. At the same time its nature and
situation combine to show that its purpose was the
protection of the northern frontier of the empire. We
cannot, therefore, hope to study its remains intelli-
gently unless we have some knowledge of the manner
in which the Roman army was organized in imperial
times, and some idea of the leading principles on which
Roman frontier policy was based, particularly in the
second century A.D. There are, of course, many more
or less important matters of detail, regarding which a
final agreement has not yet been reached among
scholars. But excavation and epigraphy are day by
-day bringing us nearer to certainty. And already the
anain points stand out with sufficient distinctness.

THE AUGUSTAN REFORMS 43

The Roman army as a permanent force was called
into existence by Augustus. Originally, soldiers had
been enlisted, not for a definite term of service, but for
the purposes of a particular campaign. The fighting
over, they were disbanded, to return to civil life until
the state should once more need their services. A
system so simple was well enough suited for the
exigencies of home defence. It was altogether unfit
to stand the strain of that task of universal conquest
which destiny had imposed upon the energies of Rome.
If the gates of the temple of Janus were never closed,
it was inevitable that the life of the soldier should
become more and more professional. So far had the
process of transformation advanced by the beginning
of the first century B.C. that from the time of Marius
onwards the term of enlistment stood fixed at sixteen
years. With the advent of the empire the final step
was taken. In constitutional theory no small part of
the powers of the emperor, and that by far the most
important part, derived from his possession of the
imperium — in other words, from his position not merely
as chief magistrate, but also as head of a body of
troops ; and, theory apart, a standing army had
become a necessity, if the provinces that the preceding
century had seen brought under Roman rule were to
be efficiently administered and secured. It was thus
natural that there should be called into being a great
military organization having its centre in the person of
the emperor himself. He was the imperator of every
man who joined the ranks. It was to him, and not to
the state, that the oath of allegiance was taken. In
whatever country the soldiers might be serving, and
under whatever leadership, the personality of the

44 ORGANIZATION OF ROMAN ARMY

officer who exercised direct command was over-
shadowed by that of the central figure whose image
had its place in the shrine of the regimental standards.
Sometimes, indeed, when the subordinate was excep-
tionally capable, or exceptionally fortunate, the sub-
stance might grow more impressive than the shadow.
That meant civil war, so soon at least as the fateful
secret was discovered that emperors could be made
elsewhere than at Rome.

In the army as remodelled by Augustus1 there were
two great classes of troops, the legionaries and the
auxiliaries. The legions were the direct descendants
of the bodies of armed citizens who in early days had
been wont to respond to the call of consul or dictator.
Possession of full burgher rights continued, therefore,
to be a distinguishing mark of the soldiers who com-
posed them. The years of turmoil and strife that
preceded the establishment of the empire had involved
the maintenance on a war footing of a much larger
body of trained men than was necessary merely to hold
foreign foes in check. When Augustus finally rid him-
self of his rivals, he found that he had at his command,
in addition to the seasoned warriors he had himself
enrolled, numbers who had originally sworn allegiance
to Lepidus or to Antony. He had, in fact, three
armies from which to draw the material for one.
Having decided on the size of his permanent force, he
disbanded certain of the legions and retained the rest.
Fully alive to the value, from a military point of view,,
of the sentiment and the associations that cling to the
time-honoured name of a regiment, he made no change

1 In re militari et commutavit multa et instituit, atque etiam ad anti-
quum morem nonnulla revocavit (Suetonius, Divus Augustus, c. 24).

THE LEGIONS 45

in the official numbering of the legions whose existence
he continued. Each of the three armies spoken of
contributed a "Third" legion to the new combination,
which also contained two " Fourth " legions, two
" Fifth," two " Sixth," and two " Tenth." l This con-
stituted a precedent, and there are many instances of
similar duplication among the legions that were subse-
quently raised. Confusion was avoided by the custom
of appending special titles to the numbers, these titles
being derived sometimes from the name of the place
where the legion had been levied or of a country
where it had gained distinction, sometimes from the
name of a commander or emperor closely associated
with its history, sometimes from an accidental circum-
stance more or less obvious. In all official records
number and title were used together. Thus, the
legions which at one time or another formed part
of the army of occupation in Britain were the //
Augusta, the XIV Gemina Martia Victrix, the ill-
starred IX Hispana? the XX Valeria Vicirix, the
// Adjutrix, and the VI Victrix. Of these the first,
the fourth, and the last were all actively engaged in
the building of the Antonine Vallum.

Under Marius the legion had numbered 6000 men.
In imperial times its full strength was probably 5600.
The subdivision into 10 cohorts still remained, but
there were now only 59 centuries.3 On special occasions

1 Mommsen, Res gestae Divi Augusti, pp. 73 f. 2 See supra, p. 3.

3 The first cohort contained 5 ' centuries,' the first of these being 400
strong, the second 200, the third and fourth 150 each, and the fifth 100,
giving a total of 1000 men in all. Each of the remaining nine cohorts
consisted of about 500 men, divided into six 'centuries.' See Mommsen,
Ephem. Epigr. iv. pp. 228 ff., and v. Domaszewski, Rangordnung des
romischen f/eeres, pp. 28 f.

46 ORGANIZATION OF ROMAN ARMY

detachments of legions might take the field as separate
units. Such a detachment was called a vexillatio, and
seems usually to have contained about 1000 men.1
The legion was, of course, a body of heavy-armed
infantry, but it generally included a certain proportion
of cavalry. Originally, no doubt, the horsemen were
Roman citizens, just as were their comrades. As early
as the second Punic War, however, Spaniards and
Numidians began to be recruited on account of their
peculiar aptitude for work of the kind. A century or
so later the mounted contingents had become entirely
non-Roman.2 Their size appears to have varied.
During the civil wars, when large armies were manoeu-
vring against one another in the field, they had of
necessity developed into an important arm of the
service. For a force whose main occupation was to
be garrison duty, cavalry was clearly less useful.
Henceforward a mere handful of horsemen, 120 in all,
was definitely attached to each legion. The remainder
of the mounted troops were relegated to the inferior
class of auxilia.

One of the notable changes initiated under the later
Republic, and definitely accepted by Augustus as a
fundamental principle, was to place the command of
the whole legion, as well as of the accompanying body

1 C.I.L. viii. No. 2482 and x. No. 5829 ; Tacitus, Ann. xiv. 26, xv. 10 ; etc.
But there was no fixed limit to the number that might be thus temporarily
attached to a vexillum — that is, a special set of colours. Tacitus, Ann.
i. 49 and xiii. 38 indicate vexillations 3000 strong, while Hist. ii. 57 and
83 point to contingents of 2600. Again, vexillations of 2000 are men-
tioned in Hist. ii. n.

2 Julius Caesar's cavalry, for instance, consisted entirely of foreigners —
Gauls, Spaniards, and Germans : see Rice Holmes, Caesars Conquest of
Gaul, pp. 583 ff.

THE PRINCIPAL OFFICERS 47

of auxilia, in the hands of a senator, known as the
legatus Unionist The position of this officer corre-
sponded roughly to that of a modern general of
division. In earlier days there had been no single
authority supreme within the legion, the control being
exercised in turn by six military tribunes, young men
of good social standing who frequently made the office
a starting-point, not for a military, but for a political,
career. Under the new system the tribuni militant
became subordinate to the legatus legionis, and were
thus shorn of some of their dignity. Another appoint-
ment brought into prominence by the altered cir-
cumstances was that of the special officer who had
to take charge of the camp in which the troops were
henceforth to be permanently quartered.2 The prae-
fectus castrorum was not, originally, associated with
any particular legion ; he rather belonged to the
camp where his duties were fulfilled. In the reign
of Domitian, however, the custom was introduced of
keeping each legion in a camp of its own, — a tolerably
obvious application of the maxim, 'divide et impera?^
— and thereafter the praefectus castrorum naturally
tended to become more and more of a divisional officer.
From the third century onwards he supplants the

Marquardt, Rdmische Siaatsverwaltung, ii.2 p. 457, footnote 4,
for references.

2 Erat etiam castrorum praefectus . . , ad quern castrorum positio, valli
et fossae aestimatio pertinebat, Tabernacula vel casae militum cum
impedimentis omnibus nutu ipsius curabantur, etc. (Vegetius, Bk. ii.
c. 10).

3 Geminari legionum castra prohibuit, nee plus quam mille nummos a
quoquam ad signa deponi; quod L. Antonius apud duarum legionum
hiberna res novas moliens fiduciam cepisse etiam ex depositorum summit
videbatur (Suetonius, Domittanus, c. 7).

48 ORGANIZATION OF ROMAN ARMY

legatus, and is usually called praefectus legionis? He
was invariably a soldier of tried capacity and experi-
ence,2 generally one who had gained the confidence of
his superiors as a centurion.8

The centurions, sixty or so in number, were the real
backbone of the legion. As a rule, they were men who
had won their position by sheer merit. Their force of
character, and the fact that they had been schooled in
•a long series of campaigns, combined to give them an
influence to which officers who were nominally their
superiors could often make no pretence. On the other
hand, the strictness of their discipline was sometimes
bitterly resented. In the memorable pictures Tacitus
has sketched of the mutinies that broke out in Pannonia
and in Germany on the accession of Tiberius,4 nothing
is more striking than the fierce determination to exact
vengeance for the cruelty of the centurions — ' saevitia
is the word that is used.5 There were various grades
among them, and advancement from one grade to
another followed a well-understood course.6 A select
few stood on a different plane from their fellows,7 and
were invited to take part in the councils of war along
with the tribuni militum. The senior centurion was
the man who was in command of the first century of

1 See v. Domaszewski, Rangordnung, p. 120.

2 Is post longam probatamque militiam peritissimus omnium legebatur,
at recte doceret alios quod ipse cum laude fecisset (Vegetius, Bk. ii. c. 10).

3 Cf. Tac. Ann. xiii. 9, 3 with 39, 2.

4 Annals, Bk. i. c. 16-30, and 31-45.

6 Op. tit. c. 1.7. Cf. c. 32 : Ea vetustissima militaribus odiis materies et
saeviendi printipium.

6 For details, see v. Domaszewski, Rangordnung, pp. 92 ff.

7 These were ihtprimi ordines, who appear to have been the centurions
of the first cohort (v. Domaszewski, op. tit. p. 94).

THE CENTURIONS 49

the first cohort. He had the title of primus pilus or
centurio primi pili> and on completing his term of
service became known as a primipilaris. The primi-
pilares were eligible for promotion to various higher
offices.1 As a rule, the others quitted the legion on
retirement into civil life. Occasionally centurions were
employed on special duties. This happened even in
the first century A.D. It became increasingly frequent
under Hadrian and his successors, — that is, during the
period with which we are more immediately concerned.
They might, for instance, be given the command of an
auxiliary regiment of foot or horse, or they might be
put in charge of a fort upon the frontier.2

As has been already remarked, the legions were
recruited entirely from those who possessed the
rights of Roman citizenship. The free extension of
this privilege to the provincials opened up a wide
field, and no difficulty was found in enlisting annually
a sufficient number of men to meet the normal
wastage. Probably not more than 20,000 were
required in any ordinary year. Mommsen has shown
from an analysis of the inscriptions that there were
three distinct stages in the evolution of the system
of levy.3 Under Augustus, Italy and the Latinized
West supplied the material for the legions stationed
in the western provinces, while those stationed in

1 See v. Domaszewski, Rangordnung, pp. ii6f., where instances from
Tacitus and from the inscriptions are collected.

2 See Miiller, ' Abkommandierte Centurionen,' in Philologus^ 1882,
pp. 482 ff., and v. Domaszewski, Rangordnung^ pp. io6f. Particular
examples are discussed in chap. x. infra, Nos. 32-36, and 45.

3< Die Conscriptionsordnung der romischen Kaiserzeit,' in Hermes^ xix.
pp. i ff. See also O. Seeck, ' Die Zusammensetzung der Kaiserlegionen,'
in Rhein. Mus. xlviii. pp. 602 ff.

D

50 ORGANIZATION OF ROMAN ARMY

the eastern provinces had their ranks filled from the
Hellenized East. The enlistment of Italians was
forbidden (or, at all events, discountenanced) by a
later emperor, apparently Vespasian, but the Augustan
system was in other respects maintained. Finally,
local conscription was introduced by Hadrian. Two
points should, however, be noted as so far modifying
these general conclusions. In the first place, the
centurions were frequently Italians, long after Italy
ceased to be a recruiting ground,1 while even private
soldiers were occasionally enlisted there at least as
late as the reign of Marcus Aurelius.2 In the second
place, it must not be assumed that the British legions
were necessarily raised locally in the age of the
Antonines, when Southern Scotland was occupied ;
every rule has its exceptions, and the area of the
Romanized part of the province was small compared
with the size of the force that the military situation
demanded.

Service nominally lasted at first for twenty and
afterwards for twenty-five years.3 Frequently, how-
ever, it was prolonged far beyond this limit. There
are inscriptions which tell us of centurions who were
on the active list for forty or fifty years.4 And during
the first century time-expired men were liable to do

1 Mommsen, I.e. p. 39, and v. Domaszewski, Rangordnung, p. 83.

2 Mommsen, I.e. pp. 20 f. See also chap. x. infra, No. 42.

3 Augustus originally retained the Marian limit of sixteen years, but in
6 A.D. he extended the period of service to twenty years with the colours
and five in the reserve. Cf. the demand of the mutineers for a return to the
older arrangement (Tac. Ann. i. 17 and 36). On the subsequent extension
to twenty-five years see Mommsen, Arch, epigr. Mitt. 8, pp. 189 f.

4 Extreme instances are C.I.L. hi. No. 11031 (fifty-eight years) and No.
1 3360 (fifty-five years) ; cf. v. Domaszewski, Rangordnung, p. 83, foot-
note, for others.

51

duty as veterans (vexilla veteranorum) for five years
beyond the stipulated twenty before they were
allowed to enter into their heritage as colonists.
Even the five years' limit was frequently exceeded.
This was one of the chief grievances of the mutinous
legionaries whose vindictive conduct towards their
centurions was alluded to above.1 The deserving
private had a fairly attractive career before him.
There were open to him a number of more or less
important posts, which brought with them increase
of prestige and of pay.2 If he chose to devote him-
self to drill and tactics, he might hope ultimately to
become a centurion. If his bent was practical or
administrative, he had ample opportunities of proving
his quality in a self-contained hive of industry like
the legion.

Both for soldiers and for officers camp life was
strenuous. Even when no actual fighting had to be
done, the troops had to be kept in proper trim by
exercise of all sorts, — by drill, by regular route march-
ing, and the like.3 Hard manual work was exacted
from them too.4 They raised with their own hands

1 See Tac. Ann. i. 17 and 35. The complaints there made as to
' tricena aut quadragena stipendia ' for the private soldier are fully borne
out by the inscriptions ; cf. Mommsen, CJ,L. iii, p. 282.

2 For a list see Marquardt, Rom. Staatsverwaltung, ii.2 pp. 544 flf.

3 Seneca, Ep. 18, 6 : Miles in media pace decurrit sine ullo hoste, vallum
jacit et superuacuo labore lassatur, ut sufficere necessario possit ; Tertullian,
Ad Martyr. 3 : Etiam in pace labore et incommodis bellum pati jam
ediscunt, in armis deambulando, campum decurrendo, fossam moliendo.
Cf. Veget. i, 27, and other passages cited by Marquardt, Rom. Staatsver-
ivaltung, ii.2 p. 567.

4 Often merely to secure that they should be kept actively employed —
' ut miles otiutn exueret' (Tac. Ann. xi. 20 ; cf. xiii. 53). See the passages
quoted by Marquardt, op. cit. p. 568.

52 ORGANIZATION OF ROMAN ARMY

the permanent buildings in which they were housed,
quarrying stones, felling trees, and manufacturing
bricks and tiles as they were required.1 They laid
the roads that ran between the strategical points
of the district that they occupied. When the
need arose, they piled up special defences like the
English and Scottish Walls, structures which are
therefore monuments, not only of the engineering
skill of the Romans, but also of the discipline and
energy of their legionaries. Amid all their severe
exertions, however, the soldiers appear to have been
well fed Even the mutineers, whose grievances
Tacitus sets forth so fully, have no complaint to make
about their rations. In spite of the length of the term
of service, celibacy was nominally compulsory for all
ranks save the very highest. But some of the many
inscriptions mentioning a conjunx or uxor (not neces-
sarily a fully wedded wife) must be at least as early
as the third century. And such unions, though not
strictly legal, would be permanent. An important con-
cession attributed to Severus by Herodian2 probably
refers to facilities for domestic life. At a still later date
every soldier was made free to marry, provided he
applied for and obtained a formal permit.3 The change
was significant of a general movement towards easier
conditions.

The organization of the auxiliary forces was more
complex and varied than that of the legions. Thus,

JCf. Vegetius, Bk. ii. c. ri : Habet praeterea legio fabros tignarios
structores carpentarios ferrarios pictores etc. a Hist. iii. 8.

3 The whole question regarding the conubium of the legionaries is still
somewhat obscure. See Marquardt, Rom, Staatsverwaltung, ii.1 pp.
560 ff., founding upon Mommsen in CJ.L. iii. pp. 905 if. ; and cf. Cagnat,
Larmte romaine (fAfnque^ pp. 444-454.

53

there were more than thirty cohorts recruited from
those natives of Italy who were anxious to follow a
military career but were debarred from enlisting in
the legions after the order had gone forth that these
latter were to be raised in the provinces. Such
volunteers, however, — cohors voluntariorum was a very
usual title — need not concern us here, for none of
them, so far as we know, ever saw service in Scotland.
It is otherwise with the regular auxiliary regiments.
These were sometimes composed wholly or mainly
of infantry. Sometimes they consisted entirely of
mounted men. In the former case they were called
cohortes. The cohorts fell into two great classes —
cohortes quingenariae, each containing 500 men under
the command of a praefectus, and cohortes miliariae,
each containing 1000 men under the command of
a tribunus?- Where they included a proportion of
horse-soldiers, the descriptive epithet ' equitata ' was
added. Apart from a very doubtful mention of a
cohors Batavorum? the following is a list of the
cohortes auxiliariae whose presence along the line of
the Scottish Wall is attested by inscriptions : — Cohors
I Baetasiorum, civium Romanorum ob virtutem
(appellata] ; Cohors IIII Gallorum equitata ; Cohors
I Hamioi'um sagittariorum; Cohors VI Nervioruw,
Cohors II J^hracum equitata ; Cohors I Tungrorum
miliaria ; Cohors I fida Vardullorum, civium

1 See Grotefend in Banner Jahrbiicher, xxxii, pp. 6if. as to this difference
in the rank of the commanding officers. There are two notable
exceptions, both in regiments that formed part of the garrison of Britain.
The Cohors I and Cohors II Tungrorum, although 'miltariae' were
commanded by praefecti. The reason is unknown. Cf. also infra,
chap. x. No. 43.

2 C.I.L. vii. No. iioi. The stone is lost ; and, at the best, it was a mere
fragment.

54 ORGANIZATION OF ROMAN ARMY

Romanorum, equitata, miliaria. The national name
obviously indicates the people from among whom the
corps was originally raised. The Hamian archers, for
instance, hailed from Syria ; the Baetasii, Nervii, and
Tungri from the Lower Rhine ; and the Vardulli from
Hispania Tarraconensis. The title ' fida' recalls the
Loyal North Lancashire Regiment' That of ' civium
Romanorum ' denotes that on some occasion the gift of
citizenship had been bestowed on the cohort in a
body, as a reward for conspicuous gallantry. The
privilege was, of course, a personal one, and did not
extend to new recruits. Its memory was, however,
perpetuated by being allowed to survive in the official
designation.

An auxiliary regiment of cavalry was known as an
ala, so called because the wing of an army was the
natural station for mounted troops. The alae, like
the cohortes, were divided into two classes according
to their size. An ala quingenaria mustered 480
strong, with 544 horses, and was divided into sixteen
squadrons or turmae, each containing 30 men and 34
horses. An ala miliaria consisted of twenty-four
turmae, each with 42 men and 46 horses, giving a
total of 1008 men and 1 104 horses.1 In both cases the
commander was a praefectus, this being the highest
point in a cavalry officer's career. Further, each turma
was under the charge of a decurio, for whose use two
spare horses were reserved. In status the decurio
ranked above the centurion of the auxiliary cohort
and immediately beneath the centurion of the legion,
promotion from one to the other of the two latter

1 See v. Domaszewski, Hy^ini Gromatici Liber de Munitionibus
Castrorum, p. 52.

COHORTES AND ALAE 55

grades being not uncommon.1 Within the turma his
subordinates were the duplicarius and the sesquipli-
carius? each of whom had a spare horse at his disposal.
The only ala that can as yet be associated with the
Scottish Wall is the ala I Tungrorum.

Speaking generally, one may say that the official
designations of the alae are analogous to those borne
by the cohortes. Some of the older ones, however,
have titles that preserve the names of individuals
almost or altogether unknown to history. It seems
highly probable that these may have been called after
the officers who were responsible for raising them,3 and
that some of them may have been in existence before
the reign of Augustus. In any event the Augustan
reforms marked an epoch in the history of the auxiliary
forces. Whatever may be the case in regard to the
mounted branch, it was then that the cohortes of infantry
were first systematically organized. The fact that they
were cheaper would be a sufficient reason for bringing
them into being, for the great expense of maintaining
the army rendered rigid economy desirable. Before
long we find another motive coming into play.4 The
forcible enlistment of auxiliaries came to be regarded
as a recognized method of pacifying districts that were

1 v. Domaszewski, Rangordnung, p. 53.

'2 These non-commissioned officers were so called because they received
respectively double and one and a half times the ordinary amount of
rations (Varro, De Ling. Lat., v. 90). This enabled them to keep, or com-
bine with others in keeping, body servants (Cagnat, Uarmte romaine
d'Afrique, p. 436). The terms are, however, better known as applied to
legionary soldiers, on whom similar privileges had been conferred.

3 See Cichorius in Pauly-Wissowa's Real-Encyclopddie^ i. p. 1225.

4 See Ritterling, 'Zur Geschichte des romischen Heeres in Gallien
unter Augustus,' in Banner Jahrbilcher, cxiv.-cxv. pp. 183 f.

56 ORGANIZATION OF ROMAN ARMY

prone to disturbance, the regiments so formed being
permanently quartered in an entirely different portion
of the empire. This gave an immense impetus to the
development of the auxiliary forces, and that just at a
time when the need of light infantry was beginning to
be strongly felt. After the great revolt of Civilis, for
example, quite a number of new corps make their
appearance, drawn from the region of the Lower
Rhine. They were seemingly deported immediately
upon embodiment, some of them being despatched to
strengthen the garrison of Britain. That their acquies-
cence was often reluctant may fairly be inferred from
the remarkable story which Tacitus tells in the Agricola?-
of the adventures of the mutinous cohort of Usipi, a
regiment whose enrolment, as Mommsen points out,2
was almost certainly the result of a successful campaign
of Domitian.

The same plan was adopted in our own island. It
is the subject of bitter complaint in the speech which
Tacitus puts into the mouth of Calgacus before the
battle of Mons Graupius.3 And the strongest justifi-
cation of the complaint is furnished by the inscriptions;
while the British frontier was garrisoned by detach-
ments from Germany, Gaul, Spain, Thrace, Syria, and
elsewhere, it is in Central Europe that we have to
look for the memorials of the cokortes Britannicae and
cohortes Brittonum. The majority of these seem to
have been levied in the latter part of the first century
A.D., — that is, during the forward movement that was
ended by the recall of Agricola ; there were several

1 C. 32.

2 Roman Provinces, i. p. 150, footnote.

8 Agricola, c. 31 : Per dilectus alibi servituri auferuntur.

FORCED LEVIES 57

in existence as early as A.D. 6g.1 One or two are with
more probability assigned to the reigns of Trajan and
Hadrian. But, so far as our present purpose is con-
cerned, the deportation to which most interest attaches
is that which is believed to have taken place in con-
nection with the operations that culminated in the
building of the Scottish Wall.2 In the reign of Pius
sundry ' numeri Brittonum ' are for the first time
encountered in the forts on the southern section of
the German Limes. There are indications that they
required to be carefully watched. Independent evidence
suggests circa 145 A.D. as the date of their appear-
ance. It is a reasonable hypothesis that they represent
the living fruits of the victories of Lollius Urbicus.3

It will not have escaped notice that the corps whose
origin we have just been discussing are styled numeri.
This means that they belonged to a class of troops that
retained more of their national characteristics than was
permissible in the case of the regular cohortes and alae.^
Except for the lack of the civitas (which they did not
as a rule receive until their discharge), the latter were
to all intents and purposes Roman soldiers. Outside
the regiments of a special character — the Syrian
bowmen, for example, — all were armed and equipped
in Roman fashion. Their officers, including centurions
and decuriones, had to be Roman citizens. Latin was

1 See, for instance, Tacitus, Histories, i. 6, 43, and 70, and iii. 41.

2 See E. Fabricius, Ein Limesproblem (Freiburg, 1902), pp. 18 ff.

3 Fabricius (I.e.} suggests that this lends a new meaning to the " summotis
barbaris" of Julius Capitolinus (see sitpra, p. 7, footnote). But it may be
doubted whether Capitolinus was doing more than repeating the words or
Tacitus, who uses the same phrase in speaking of the action taken by
Agricola in A.D. 81 (see supra, p. i).

4 See Mommsen, Hermes, xix. pp. 219 ff., and x\ii. pp. 547 ff.

58 ORGANIZATION OF ROMAN ARMY

the language of command. Latin, indeed, was in all
likelihood the ordinary medium of intercourse. It
must often have been the only possible one, for the
tombstones show that, national names notwithstanding,
the composition of the individual regiments was very
miscellaneous. Gaul and the banks of the Lower
Rhine were evidently the chief recruiting grounds for
those stationed in Britain. Renegade Britons, too,
were occasionally enlisted. The denunciations of
Calgacus suggest that a few such had enrolled them-
selves in Agricola's Batavian and Tungrian cohorts.1
And somewhere near the eastern end of the Scottish
Wall there sleeps a Brigantian soldier who served for
nine years, against his Caledonian kinsmen, in the
Second Cohort of Thracians.2

In some respects the auxiliaries fared better than
the legionaries. When the two classes of troops were
acting together, they had to play the part of skir-
mishers. They provided convoys, and kept the enemy
at bay when the nightly camp was being constructed.
Consequently they were less heavily laden on the
march, and were spared some of the exacting field-work
that was imposed upon their fellow-soldiers. Again,
not being Roman citizens, they were free to marry, or
at least to contract unions, at a time when a nominal
celibacy was insisted upon within the legions. When
they received their discharge (honesta missio), they were
made Roman citizens, and their marriages were recog-
nized as legal.3 On the whole, however, their position

1 He speaks of ' Gallos et Germanos et (pudet dictu) Britannorum
-blerosque] as forming part of the Roman army (Agricola, c. 32).

2 See infra, chap. x. No. 51.

3 See Cagnat, DArmte romaine d'Afrique, pp. 454 ff.

PLACE OF THE AUXILIARIES 59

was less dignified, their privileges less secure. Their
term of service, too, was longer, extending to twenty-
five years and often more.1 Man for man, their lives
were counted less valuable than those of the legionaries.
Hence they were, as we shall see, very freely employed
for the rough and tumble work of frontier warfare.
Even in pitched battles they were put in the forefront,
if that could be done without endangering the fortunes
of the day. At Mons Graupius, for instance, the
entire Roman fighting line consisted of auxilia — 8000
foot in the centre, 3000 horse upon the wings. The
legions were drawn up in reserve before the camp —
" ingens victoriae decns citra Romanum sanguinem
bellandi, et auxilium, si pellerentur. " 2 The ordinary
Roman attitude towards the auxiliaries, an attitude
which is far older than the Empire, has never found
more pointed expression.

At the death of Augustus, the total number of
legions was twenty-five. By the time of Septimius
Severus it had grown to thirty-three. When the
Notitia Dignitatum was compiled (circa A.D. 400),
there were at least one hundred and seventy, many of
which were, however, in all probability greatly attenu-
ated. From inscriptions and military diplomas we
know definitely the names of at least one hundred and
twenty alae and four hundred and fifty cohortes, but
we cannot be certain to what extent these existed
contemporaneously. We shall not be far wrong if we
reckon that, during the period that concerns us, the

1 The regular formula in military diplomas relating to the discharge of
auxiliaries is : gut quina et vicena plurave stipendia meruerunt. See
Marquardt, Rom. Staatsverwaltung, ii.2 p. 543.

2 Tacitus, Agricola, c. 35.

60 ORGANIZATION OF ROMAN ARMY

auxiliaries were at least equal in number to the legion-
aries. If we assume for the first three centuries of our
era an average of thirty legions, each containing 5000
men, we may safely calculate that the strength of the
Roman army can hardly have fallen short of 300,000
men. When we consider the extent of the empire and
the pressure to which it was liable to be exposed at
various points, this total seems small compared with
the millions a single European state can put into the
field to-day. As a matter of fact, it ultimately proved
unequal to the burden that rested upon it. Its task
was to defend the frontier provinces against barbarian
invasion. Accordingly it was mainly concentrated in
Northern Britain, in Africa, and along the Rhine, the
Danube, and the Euphrates. As Mommsen well puts
it, the Roman army — except for the regiments stationed
in the two capitals, Rome and Alexandria, — was simply
the aggregate of the frontier garrisons. There was no
"striking force" in reserve, that could be used for
offensive purposes without disturbing the normal
equilibrium. If a crisis arose demanding a special
effort in any particular quarter, it could only be met by
a dangerous weakening of some other portion of the
line. It was this that made the final catastrophe
inevitable.1

It follows from what has been said that there was
nothing in any way corresponding to a system of relief.
Under ordinary circumstances legions, cohorts, and
alae remained permanently in the provinces to which
they had once been assigned. Thus, the Second
Le£ion Auo-usta and the Twentieth Valeria Victrix

o o

1 See Mommsen, ' Das Militarsystem Caesars ' in Hist. Zeitsckr. 1877,
pp. 1-15, and Gesammelte Schriften, iv. pp. 156 ff.

SIZE AND DISTRIBUTION 6r

had been in Britain for a hundred years before
they helped to build the wall from Forth to Clyde,
and they were there for more than two centuries after
it was completed. Nor must it be supposed that this
was regarded as a hardship. It was looked on as the
very opposite. No doubt the legionaries were apt to
murmur when they contrasted their own conditions
of service with the pleasant places in which the lines
of the favoured few who constituted the Praetorian
Guard were cast ; the strain was sometimes practi-
cally continuous, as the ringleaders in the Pannonian
mutiny pointed out to their companions.1 It was very
different when it was merely a question of transference
from one frontier station to another. In A.D. 69, the
"year of the four emperors," the soldiers of Vespasian,
then legate of Syria, were roused to fury by the bare
rumour that Vitellius contemplated moving them to
Germany. Here there was the special reason that
they would be going to a district where the climate
was more severe and the work more harassing. But
Tacitus explicitly states the general ground : long
service in one and the same spot had so familiarized
them with their station that they had come to love
it as a home. The provincials would have had a
grievance too ; for them it would have meant parting
once for all from countless friends and relatives.2 To

JTac. Ann. i. 17 : An praetor ias cohortes, quae binos denarios acceperint,
quae post sederim annos penatibus suis reddantur, plus periculorunt sus-
tipere? Non obtrectari a se urbanas excubias: sibi tamen afiud horridas
gentes e contuberniis hostem aspici.

2 Tac. Hist. ii. 80 : Quippe et provinciales sueto militum contubernio
gaudebant^plerique necessitudinibus et propinquitatibus mixti, et militibus
vetustate stipendiorum nota et familiaria castra in modum 4>enatium
diligebantur.

62 ORGANIZATION OF ROMAN ARMY

what an extent these particular legions had assumed
the colour of their surroundings is shown in striking
fashion by the picturesque incident that precipitated
the end of the battle of Bedriacum. The struggle
had raged fiercely all night, at first in total darkness,
then by the light of the moon. On the whole, the
advantage rested with the Flavian troops, but the
issue still hung in the balance. As soon as the dawn
was sufficiently advanced to let his features be recog-
nized, Antonius Primus, leader of the Flavians, hurried
from one regiment to another, urging his soldiers to
redouble their exertions. He was greeted with
answering shouts, and just at that moment the sun
rose. With one accord the men of the Third Legion
lifted up their voices and did homage to the god of
light, as they had learned to do in Syria.1 The
clamour was interpreted on both sides as a sign that
reinforcements had arrived, and in a few moments
the legions of Vitellius were in headlong flight.

Along the frontier-line the troops were not quar-
tered in or near large cities, but in separate forts or
fortresses, which were rigorously closed against all
but the soldiers themselves. Outside the gates of
these, however, there were generally settlements of
traders and camp-followers, which not infrequently
grew into towns. During the period with which we
are concerned, it was customary in time of peace to
have the castra or cantonments of the different legions
a little to the rear of the nominal boundary. The
interval was covered with a more or less intricate
network of strategical roads, guarded by the smaller
forts or castella which served to house the auxiliaries.

1 Tac. Hist. iii. 24 : Orientem solem (ita in Suria mos est) tertiani salutavere.

FORTRESSES AND FORTS 63

It had not always been so. The fortress of Novae-
sium, for instance, erected under Tiberius, was planned
for considerably more than 6000 fighting men.1 That
is, the legionaries and the auxiliaries were at this time
brigaded together. They formed a tactical unit, and
the policy of concentration was thoroughly sound from
a military point of view. Why, then, was it aban-
doned ? No doubt the separation brought with it
certain incidental advantages. It minimized the risk
of such violent outbreaks of jealousy as that which took
place at Ticinum in the brief reign of Vitellius,2 when
a quarrel over a wrestling-match led to a fierce attack
on the auxiliaries by their legionary comrades. Two-
cohorts were cut to pieces, and the risk of worse
disturbances was only averted by sending the Batavian
and Gaulish levies back across the Alps. Again, there
is other evidence to suggest that the Flavian emperors
— more particularly Domitian3 — deliberately aimed at
restricting the number of troops over which it would
be easy for a single officer to gain a personal influence.
It is, therefore, natural to suppose that the change was
not unwelcome. But the reasons that dictated it lay,
in all probability, much deeper. They cannot be dis-
sociated from the general causes by which the whole
frontier policy of the empire was determined. A brief
sketch of the process of evolution will put us in a
better position for approaching the study of the
Scottish Wall.

1 H. Nissen (Banner Jahrbiicher, cxi.-cxii. pp. 56 f.) calculates that
there was accommodation for over 9000. This is possible, but the basis-
of reckoning1 is necessarily somewhat speculative.

2 Tac. Hist. ii. 68 f. 3 See su^ra, p. 47, footnote.

CHAPTER III

THE HISTORICAL BACKGROUND: ROMAN FRONTIER
POLICY AND FRONTIER POSTS

UNDER the Republic the territory of Rome had been
protected by a fairly continuous girdle of " buffer
states." In the absence of a standing army no other
method of defence was practicable.1 During the thirty
or forty years that preceded the establishment of the
principate most of these " buffer states " had lost
their independence. Some perished by a process of
natural decay. Others were attacked and annexed
by ambitious officers in quest of plunder and prestige.
Others, again, had to suffer for the mistake of espousing
the quarrel of the losing side when war broke out
between the rival leaders. Augustus fully grasped the
significance of the altered conditions. He recognized
that the huge territory he controlled was exposed to
attack at many points, and he created an army for the
express purpose of defending it. He saw, however, what
a drain upon the imperial exchequer was involved, and
so he set his face sternly against unnecessary expan-
sion. As far as possible he sought to take advantage
of natural frontiers — the sea, or great rivers like the

1 See Mommsen, ' Der Begriff des Limes,' in Westd, Zeitschr. xiii.
pp. 134-143, and Gesammelte Schriften, v. pp. 456 ft.

THE EARLIER EMPERORS 65

Rhine ; l such vassal states as still survived within the
prescribed limits were to be peaceably absorbed as
opportunity might offer. The precepts of Augustus
were rigorously adhered to by his immediate successor,
Tiberius.2 Of Caligula, the next of the Caesars, it is
less easy to speak ; he was hardly accountable for his
actions, and we do not know how far his scheme of
invading Britain, for instance, was seriously meant.
But under Claudius Rome entered on a period of
renewed expansion. It was he who was responsible
for the conquest of our own island, that being indeed
the most important event in his reign.

It would be a mistake to assume that the policy of
Augustus was deliberately abandoned through mere
lust of power. The change was due rather to the
irresistible pressure of circumstances. We can see
this more clearly by noting what happened on the
Rhine. Even Augustus had at one time contemplated
the subjugation of Germany as far as the Elbe. That
was the end in view when the Gallic legions were
moved forward, and great fortresses established at
Xanten (Castra Vetera) and Mainz (Moguntiacum).3
The fortresses remained, after the disaster to Varus
had definitely checked the impulse that brought
them into being ; and from them there radiated
a series of roads protected by fortified posts. One
such road ran along the left bank of the Rhine

1 Tac. Ann. i. 9 : Mart Oceano aut amnibus longiquis saeptum imperium.

2Tac. Ann. i. n : Quae cuncta sua manu perscripserat Augustus
addideratque consilium coercendi intra terminos imperii; and Agricola,
c. 13 (of non-intervention in Britain) : Consilium iddivus Augustus vocabat,
Tiberius praeceptum.

3 See Ritterling, 'Zur Geschichte des romischen Heeres in Gallien unter
Augustus,' in Banner Jahrb. cxiv. esp. pp. I76ff.

£

66 ROMAN FRONTIER POLICY

between Mainz and Xanten. It was guarded by
more than fifty castella, whose garrisons would be
expected to prevent any unauthorized crossing of the
river.1 But that was insufficient. The other bank
was not in the occupation of any organized state or
states, with whose heads it would have been possible
to maintain stable political relations. It was inhabited
by a medley of tribesmen. The danger of raids was
constant, and punitive expeditions were unavoidable.
Hence the necessity for roads and outposts on the
right bank of the Rhine as well.2 These roads and
outposts were gradually and insensibly pushed forward
until a day came when it was realized that the pax
Romana could best be secured by ceasing to regard the
river as the frontier.

Thanks to the labours of the German Limes-Com-
mission,— appointed eighteen years ago by an en-
lightened government, mainly at the instance of
Mommsen, — the details of the story can now be largely
reconstructed. It is not necessary to recapitulate
them fully here. They were admirably summarized
for the benefit of British students in 1906 by the late
Professor Pelham.3 So far as the Middle Rhine is

1 Florus (iv. 12) tells us that Drusus, in the course of his offensive
operations on the German frontier, 'perRkeni quidem ripam quinquaginta
amplius castella direxit?

2 Those on the lower reaches of the river were abandoned in A.D. 47.
''Claudius adeo novam in Germanias vim prohibuit, ut referri praesidia
ris Rhenum juberet' (Tacitus, Ann. xi. 19). The historian attributes the
emperor's action to jealousy and fear of Corbulo's possible success. It is
at least as likely that the real motive was the strain that the conquest of
Britain was imposing on the military resources of the Empire.

3' A Chapter in Roman Frontier History' in Transactions of the Royal
Historical Society r, N.S. vol. xx. pp. 17 ff. (Reprinted as chap. ix. of
Pelham's collected papers, Oxford, 1911.)

DOMITIAN IN GERMANY 67

concerned, and this is the district we have been con-
sidering, the final breach with Augustan tradition was
made by the Emperor Domitian. In 83 A.D., after a
brief campaign which has already been alluded to in
another connection, he formally annexed the whole of
the Taunus district. This he did by enclosing it
within a new frontier line or limes, protected by a
series of small earthen forts with wooden watch-towers
between them. Communication with the legionary
base at Mainz was maintained by a series of strategical
roads with intervening castella. The effective frontier,
in short, was a broad belt, not a narrow line. To hold
it properly, a large number of troops, broken up into
small bodies, was required. To employ legionaries
for the purpose would have meant the dissipation of
the only homogeneous and effective field-force that
was available for the serious business of war. The
auxiliaries, on the other hand, were organized on a
basis which made their distribution a simpler matter.
Besides, their number was at this very time being
largely increased by the deportatio juvenum which
was now, as we have learned,1 a common accompani-
ment of any advance of the Roman arms. It has
been pointed out that the Cohors I Brittonum
appears for the first time in a Pannonian military
diploma of 85 A.D., the year after Mons Graupius.2
The restless spirit which we may suppose to have
animated these forced levies would furnish an addi-
tional argument for utilizing them in the manner
indicated. As frontier-police they would have little
leisure to brood over their wrongs. Lastly, if a score

1 See supra, p. 55 f.

2Fabricius, Ein Limesproblem, p. 19.

68 ROMAN FRONTIER POLICY

or even a whole regiment of them were lost, they were
more easily replaced than the legionaries.

There is no reason to believe that any visible barrier
ran continuously along the limes of Domitian. His
line seems to have been much like that which Agricola
drew from the Forth to the Clyde. In trying to
understand what a limes was, we must do our best
to rid ourselves of our associations with its English
derivative ' limit.' The Latin word does not necessarily
mean a boundary. Its original sense seems to have
been a broad straight path such as that which the sun
appears to follow in the sky.1 But it ultimately came
to have various technical significations in the ritual
and phraseology of land allotment. When we first
encounter it in the story of frontier-warfare, it is
applied to the strategic lines which were run straight
out into the enemy's territory to facilitate the rapid
movement of troops. In the centre was a made road,
the agger viae. On either side was a stretch of open
ground, left in its natural condition, except that trees
and bushes were uprooted.2 It is probably this type of
limes — an offensive weapon pure and simple — that
Tacitus has in view in his descriptions of the campaigns
of the early first century.3 But, when we remember

1 See Oxe", ' Der Limes des Tiberius,' in Bonner Jahrbucher, cxiv.-cxv.
pp. 99 ff., for a careful investigation into the various meanings of the term.

2 A proper appreciation of this is necessary to a clear understanding of
the description of the battle of Bedriacum in Tac. Hist. iii. 21 ff. See
Oxe, I.e. esp. pp. 113 ff.

3 Ann. i. 50, where Germanicus, in his expedition against the Marsi,
' agmine propero silvam Caesiam limitemque a Tiberio coeptum scindit^
castra in limite locat, etc.' Similarly in ii. 7, of his operations against the
Chatti : ' Cuncta inter castellum Alisonem et Rhenum novis limitibus
aggeribusque permunita? In the former passage scindere has much the
same force as has aperire in the phrase of Velleius (ii. 120) '•penetrat
interius, aperit limites?

THE LIMES IMPERII 69

that a line of forts was the normal complement to a
military road, we can see that an artificial boundary-
line such as Domitian created was nothing more than
a transverse limes. And it is of such boundary-lines
that the word is most frequently used from the end of
the first century A.D. onwards. Thus Tacitus, in the
brief reference which he makes in the Germania to
the annexation by Vespasian's generals of the country
beyond the upper Rhine, employs the phrase " having
drawn a limes and moved forward the garrisons M1 of a
step which was in its essence strictly analogous to that
afterwards taken by Domitian. The times, however,
it should be remembered, was not the whole strip of
more or less debatable land over which the castella of
the auxiliaries were scattered. It was merely its outer
fringe, the hiberna of the legions in the rear being the
massive supports on which the security of the frontier
really rested. The contrast is emphasized by Tacitus
in his famous indictment of Domitian's foreign policy :
"No longer was it the limes imperil and the river-
bank that were threatened ; it was the legionary for-
tresses and the supremacy of Rome."2

It will be observed that, in the sentence just quoted,
the limes imperil and the river-bank are spoken of in a
way that suggests that they meant very much the same
thing. And so they do in the present connection.
Where a river was available, the frontier-road, with
its forts, would run beside it. The drawing of an
artificial line would be unnecessary. As a matter of

1 Germania, c. 29 : Limite acto, promotisque praesidiis. An excellent
general sketch of Vespasian's annexation will be found in E. Fabricius,
Die Besitznahme Badens durch die Rotner (Heidelberg, 1905).

*Agncola, c. 41.

yo ROMAN FRONTIER POLICY

fact, however, such artificial limites tended to become
more and more numerous. The principle of expansion,
accepted — reluctantly perhaps — by Claudius and the
Flavian Emperors, was welcomed whole-heartedly by
Trajan. He was above all things a soldier, full of
tireless energy and consumed by a passion for military
renown. He pushed out the boundaries of the empire
on almost every side, with the result that, when he
died, he left behind him a heritage which no successor
could have hoped to maintain intact. Reaction followed
promptly, for Hadrian wisely declined to attempt the
impossible. The districts beyond the Euphrates were
unhesitatingly abandoned. Elsewhere the change from
the offensive to the defensive was quite as strongly
marked. Here is what his biographer Spartian says :
"In many places, where the barbarians were separated
from the Romans, not by rivers, but by limites, he
erected a barrier of great stakes driven deep into the
ground and fastened together so as to resemble a wall-
like fence."1 That is a very noteworthy statement,
and its significance has been considerably added to
since the German Limes-Commission commenced its
investigations ; we now know that it is literally true.
At various points along the German line excavation
has brought to light the trench in which the stakes
were planted, and even the actual remains of the
palisade.2

That the same device was adopted on the frontier
towards the Danube we have pictorial evidence to

1 In plurimis locis, in quibus barbari non fluminibus sed limitibus divi-
duntur, stipitibus magnis in modum muralis saepis funditus jactis atque
conexis separavit (Vit. Hadr, 12, 6).

2 Limes-Blatt, pp. 483 ff.

TRAJAN AND HADRIAN 71

prove. The reliefs on the column of Trajan represent
his Dacian campaigns. Incidentally we get a view of
the Danubian Limes (Plate III. i). What we see is a
river-bank, and on it a continuous series of square
towers, apparently of stone, each surrounded by a
stockade. These are the burgi or praesidia, mentioned
in the inscriptions,1 which formed the lesser links in the
chain of which the castella were the principal members.
It will be noted that each has a fire-signal projecting
from its uppermost story. This indicates the ordinary
method of exchanging messages. Between the towers
stand one or two soldiers, whose dress shows that they
belong to the auxilia, or regular frontier-force. That
is the pre-Hadrianic limes. It is most instructive to
compare it with the post-Hadrianic limes as depicted
on the column of Marcus Aurelius. Here too it is the
Danubian frontier that is in question (Plate III. 2).
But the blockhouses almost seem to have crept closer
to each other, and to have shrunk considerably in
height, as if it were now practicable to communicate
otherwise than by signal. These features, of course,
may be due simply to the exigencies of the space in
which the artist had to work. But, at any rate, it would
evidently be easy for messengers to pass along the
line, for between each blockhouse and its neighbours
there stretches a tall, close-set row of stakes, obviously
a reproduction of the palisade of Hadrian. It is true
that, according to Spartian, Hadrian's own palisades
were erected only where the boundary was a mere limes,
not a river.2 But the fashion, once introduced, may
have grown; 3 or the sculptor may have combined in one

1 C.I.L. iii. Nos. 3385, 10312, and 10313. 2 See supra, p. 70, footnote.
3 See Koepp, Die Romer in Deutschland, p. 73.

72 ROMAN FRONTIER POSTS

representation the characteristics of two varieties of
frontier.

As Hadrian was the immediate predecessor of
Antoninus Pius, our survey of Roman frontier-policy
has been carried far enough to prepare us for entering
on a more detailed study of the remains of the Scottish
Wall. But there is still some preliminary matter which
it would be convenient to dispose of at this point. We
shall find that the remains include traces, not only of
the Wall itself, but also of the forts once occupied by
the auxiliaries who constituted its garrison. It is,
therefore, important to get some general idea of the
original form and character of an enclosure of the kind
we may expect to meet with. What we learn from
ancient writers regarding Roman camps applies, for
the most part, to field-works such as were thrown up
nightly by large armies in the ordinary course of
campaigning. This difference is, however, less vital
than it might appear. In essential points the general
arrangement of all Roman castra and castella would
seem to have been very much alike ; within certain
limits the disposition of their parts was identical and
invariable. Of still greater value for our purpose is
the fund of knowledge that has been gathered by those
who have investigated actual Roman sites in other
localities. In France the efforts of Napoleon III. to
uncover the footprints of Caesar's armies brought
many interesting facts to light. In our own country,
and in Austria, not a little has been done in recent
years. Africa, too, has taught us much.1 But most
fruitful of all have been the operations of the German

1 See particularly Cagnat, LlArmee romaine d'Afrique (1892) and Les
deux Camps de la Ltgion III" Augusts a Lamblse (1908).

I. FROM THK COLUMN OF TRAJAN

2. FROM THE COLUMN OF MARCUS AURF.LIUS

THE DANUBE FRONTIER IX THE SECOND CENTURY A. I).

ROMAN CAMPS 73

Limes-Commission, under whose supervision some-
where about a hundred forts have been more or less
systematically examined since 1894.*

Minute descriptions of the system of castrametation
followed by the Romans at two widely separated
epochs are available for our information. Polybius,
the friend of the younger Scipio, pictures the Roman
camp as it was a century and a half before the com-
mencement of our era ; and Hyginus, or at least the
author of the treatise that bears the title Hygini
Gromatici Liber de Munitionibus Castrorum, gives an
account of the encampment of imperial times, most
probably of the period of Trajan.2 Particulars of the
defences of a camp, but not of its internal arrange-
ments, are also furnished by Vegetius, who lived in
the end of the fourth century A.D. Apart, however,
from the lateness of the epoch at which he wrote, this
last author can hardly be said to rank high as an
authority. The sketch of Polybius3 applies to a camp for
a consular army of two legions with auxiliaries, or (say)
about 20,000 men. So far as it goes, it is full, interest-
ing, and completely trustworthy. But the historian is
mainly concerned with impressing upon his readers the
uniform and symmetrical ground-plan of the interior,
the manner in which the tents are grouped, the precise
space allotted to each section, and similar details
relating to the less enduring portion of the whole. He

1 The publication of results has kept pace with the explorations. The
official reports (' Der Obergermanisch-raetische Limes') already form a
long series of invaluable monographs.

2 Von Domaszewski, its most recent editor (Leipzig, 1887), has advanced
convincing reasons in favour of this date. Formerly it was generally
assigned to the third century A.D.

3 Bk. vi. cc. 27-32.

74 ROMAN FRONTIER POSTS

has but little to say of the character of the defences.
There is more to be learned from Hyginus. His
treatise, it is true, is only a fragment. But his inten-
tion was to produce a handbook which might be
practically useful to Roman officers, and such of his
instructions as survive are correspondingly minute and
precise. At the same time we must remember that
what he has in view is a temporary camp, capable of hold-
ing three legions with their auxiliaries, on the march, —
that is, between 30,000 and 40,000 men — and we must
not expect to find all the features of his description
reproduced in the smaller and more permanent
structures which will claim our attention here.

The defences of a Roman camp of any kind con-
sisted mainly of a rampart and ditch. According to
Hyginus,1 there were two forms of ditch in use —
fossa fastigata and fossa Punica. In the former
case scarp and counterscarp alike sloped inwards
till they met, the whole assuming the shape of the
letter V. In the latter case the counterscarp sloped,
while the scarp descended perpendicularly. Exca-
vation has shown the fossa fastigata to have been much
the commonest type. Hyginus gives the minimum
breadth of a ditch for a temporary camp as 5 feet at
the surface, and its depth as 3 feet. In stationary
or permanent camps these dimensions might be
greatly exceeded, while it is not uncommon to find
castella girt with a double line of ditches. Even
a further multiplication was occasionally resorted to,
such multiplication being specially frequent in Britain ;
there were five trenches on the north and east faces of the
station at Ardoch, while Birrens had six. Turning to

1 c. 49-

THE DEFENCES 75

Hyginus again,1 we learn that the minimum breadth of
the rampart was 8 feet and its minimum height 6 feet,
exclusive of a small parapet, and also that the materials
used in its construction depended on what the locality
could most readily provide. Layers of turf or stone
were considered best. In an emergency recourse
might be had to branches of trees. If these also
failed, the place of the rampart was taken by four
rows of armed men.

Roughly speaking, one may say that all Roman
camps were either rectangular or square ; there are
exceptions, although none have been noted in Scot-
land. Apart from such modifications as the nature of the
ground might render necessary, the four corners were
usually rounded, the object being to bring within range
of missiles discharged from the ramparts the whole
sweep of ground beyond. Even so the corners were
regarded as vulnerable, and were specially strengthened
by the erection of towers. In temporary camps the
place of the towers was taken by raised platforms, on
which artillery was posted. Similar precautions were
observed at the gates, where other means of rendering
access difficult were also frequently employed. The first
of these was the titulus, a traverse or short ditch, backed
by a mound, and so placed as to cover the entrance.
Another was called the clavicula. This was formed
by giving the rampart, at the point where a gate
occurred, a semicircular bend inwards on the right
hand of those entering, the space for ingress being,
not directly in front, but on the left. A storming party
attempting to force the gates would thus be exposed to
a cross ' fire,' which would be particularly heavy on the

'Cc. 50-53.

76 ROMAN FRONTIER POSTS

side on which their shields could give them no protec-
tion. Claviculae are unknown in the forts of the
German Limes, and none have been found on the line
of the Antonine Vallum.1 Good examples were dis-
covered in Caesar's camp on the Aisne, excavated by
Napoleon III.,2 and one or two occur in Britain.3
Perhaps they were characteristic mainly of temporary
camps, being less necessary where the gates could be
made strong in themselves and where they were
flanked by towers. At all events one may hazard the
prophecy that, if claviculae ever come to light between
the Forth and Clyde, it will be in connection with
forts erected by Agricola.

The normal number of the gates themselves was
four. The main lines which formed the key to the
ground-plan of any Roman camp were limites in the
land-surveyor's sense.4 Of these the first to be marked
out was known as the decumanus maximus, the chief of
the decumani, which were so called as being drawn at
intervals of 10 actus or furrows. The gate which stood
at one end of this — originally the western end, but
latterly simply the end furthest from the enemy — was
named the porta decumana. The gate at the other
extremity was the porta praetoria, this being the gate
which led directly to the Praetorium, where the general
and his staff were accommodated. The Praetorium
lay in the centre of the camp on the line of the
decumanus maximus. Immediately in front of it the
decumanus maximus was crossed at right angles by

1 But see infra, p. 21 1, footnote. 2 Histoire de Jules Cesar, Atlas, PI. 9.

3 See Roy, Military Antiquities, PI. xi. The gates of the two earliest
Newstead periods also have arrangements that recall the clavtcula.

4 See Schulten in Pauly-Wissowa's Real-EncyclopUdie, iv. pp. 2314 ff.

THE GROUND-PLAN 77

the limes which was regarded as second in order of
importance, — that along which ran the main street (via
principalis] of the camp. This was the cardo maximus?-
At either end of the via principalis were the porta
principalis dextra and the porta principalis sinistra.
Immediately in rear of the Praetorium, and parallel to
the via principalis, was yet another limes, indicating
the course of the via quintana, so termed because the
limes along which it passed was the fifth cardo reckoned
from the cardo maximus ; it was a limes quintarius.

The Polybian camp was square, while the Hyginian
was oblong. Both authors give descriptions of the
more or less fixed arrangements of the interior, and
we can see that the differences in shape and plan
were due to the organic changes that had in the
interval taken place in the organization of the Roman
army, including the increase in the number of auxili-
aries. Thus, Polybius reckons his measurements on
a decimal basis (50, 100, 250, 500, etc.), Hyginus on
a duodecimal one (30, 60, 90, 120, etc.).2 According
to the directions given by Hyginus,3 a clear space of
60 feet had to be left on the inside of the rampart all
round the camp. This was not a street. It was a
limes, and formed an integral part of the defence,
inasmuch as its intention was to secure that every
portion of the actual encampment should be beyond
the range of weapons hurled over the rampart by a
foe outside. On the inner side of the clear space or

1 Although the decumanus was held to be the more important in the
fully developed science of land measurement, there are indications that
the cardo originally had precedence of it. The latter is a noun, not a
mere adjective, and the road that ran along it in a camp was the 'prin-
cipal' road (see Mommsen, Hermes, xxvii. p. 91).

2H. Nissen in Banner Jahrbiicher, cxi.-cxii. p. 38. 3 C. 14.

78 ROMAN FRONTIER POSTS

intervallum were pitched the leather tents of the legion-
aries, which ran in a continuous belt round the whole
camp. These were the troops upon whom most reliance
could be placed, and it was important that they should
be ready to man the rampart on the slightest alarm
and should at the same time effectively ' contain ' the
possibly disloyal auxiliaries.1 The tents of the legion-
aries were bounded on the inner side by a street, the
via sagularis. This owed its name to its position.2 A
soldier had to cross it before quitting his quarters, for
it ran parallel to the intervallum, enclosing the lines
where the whole of the force except the legionaries
lay. It therefore marked the point at which officers
and men must assume their uniform — the military
cloak or sagum — just as the citizen assumed the toga
when he left his house. Within the lines each detach-
ment and each squadron had its own duly appointed
position, and the streets, like the tents, were laid out
on a regular and perfectly familiar plan, so that there
could be no losing of the way even in the case of a
hurried night-call to arms.

So much for the testimony of the professional writers
on castrametation. It is interesting as an exposition of
the general principles which governed the design of
castella, such as those which defended the Scottish
Wall. For details of the latter we are, however, mainly
indebted to the excavation of concrete examples like
Birrens,3 Housesteads,* and Newstead.5 These make

1 Hyginus, c. 19 : ita fiet ut omni parte nationes a supra scriptis con-
tineantur.

2 H. Nissen in Banner Jahrbiicher, cxi.-cxii. p. 58.

3 Proceedings of the Soc. Antiq. of Scot. 1896, pp. 81 ff.

4 Archaeologia Aeliana, xxv. pp. 193 ff.

6 J. Curie, A Roman Frontier Post (Glasgow, 1911).

ARRANGEMENT OF INTERIOR 79

it clear that in the castellum the arrangements in vogue
in the castra were subject to modification as circum-
stances might dictate. Thus, castelta have been found
with more and with less than four gates ; that of
Agricola on the Bar Hill had only one. Nor do we
know how far the names given above continued to be
used in the smaller forts, although it seems certain
that porta praetoria at least survived.1 A few points
regarding the interior deserve to be specially men-
tioned.

The place assigned to the Praetorium in the tem-
porary camp was occupied in the castellum by a large
building, which we know to have been called the
Principia, and which served as the headquarters of
the garrison. It was oblong, and consisted, in the
main, of two open courts, one of which entered
directly from the via principalis. This outer court
often contained a well. A colonnade ran round it,
supporting a roof which sloped inwards and gave
shelter to a covered walk. Immediately opposite the
principal entrance was a door leading straight into
the inner court. Here too there were pillars ; but the
most important feature was a series of rooms — five or
three in number, according to the size of the castellum
— occupying the whole of the side that faced the door.
Four (or two) of them were used as regimental offices.2
That in the centre was the shrine of the standards.3
The image of the emperor was kept in it as well as

1 C.I.L. Hi. No. 7450.

2 As may be inferred from the analogy of the corresponding, but much
more numerous, suite of rooms in the ' Praetorium ' at Lambesis (Cagnat,
Les deux Camps, pp. 35 ff.).

3 This was first pointed out by Hettner, Westd. Zeitschrift, xvii. p. 343.

80 ROMAN FRONTIER POSTS

the colours, while it was also turned to account as the
strong-room of the regimental bank. An underground
safe or cellar is sometimes found beneath the floor, par-
ticularly from the end of the second century onwards.

Beside the Principia, and in line with it, stood the
horreum, or horrea — for in the larger castella two were
required. This was the granary or storehouse, and its
remains are always unmistakable.1 It was a long
narrow building of singularly substantial character.
The thick stone walls were strengthened with massive
buttresses, evidently to provide support for a heavy
roof such as could not easily be set on fire by any of
the enemy's missiles, or penetrated by the rainstorms
that must often have beat upon it. A set of dwarf
walls ran along the whole length of the interior to
facilitate proper ventilation (Plate IV.). The free
circulation of air below was secured by a series of
perpendicular openings in the main walls. On the
outer face these were mere slits ; but they had splayed
jambs, expanding inwards to a considerable width
(Plate XXVI. 2).

Hard by, and still in line with the Principia, there
was a large square building, usually described as
the house of the commandant. The view that it
was a residence is not, however, altogether easy to
reconcile with the general absence of hypocausts,2
and an alternative suggestion is that it may repre-
sent the fabrica or workshop of the fort.3 A feature
which seems often to occur here is a small apsed
room, identified by v. Domaszewski as the shrine

1 See Bosanquet, Arch. Ael. xxv. pp. 235 ff.

2 See Haverfield, Roman Wales, pp. 83 f.

3 Ritterling, Kastell Wiesbaden, p. 35.

THE BUILDINGS 81

of the Genius Praetorii.1 The rest of the garrison
lived in humbler fashion. The tents of the temporary
camp were represented in the castellum by struc-
tures of a somewhat more solid kind.2 Long narrow
blocks either of stone or of wood were divided by
partitions into huts, each calculated to hold a con-
tnbernium, or small group of soldiers who slept and
messed together. The hut at one end of each block
was rather larger, being intended for non-commissioned
officers. There was no provision for artificial heating,
and the accommodation must have been far from
luxurious.

One other indispensable adjunct of Roman stations
has still to be mentioned. The balneum or bath had of
necessity to be built of fire-resisting materials. As a
consequence its remains have often been found, and
they might be found still more frequently if there was
any certainty as to its probable position. That, how-
ever, cannot be determined in any given case. It is the
exception to find it within the castellum? In most
cases it lies at some distance outside the rampart, in
one of the annexes. Its discovery, like that of the
cemetery, is thus more or less a matter of chance, for
few excavations can be carried out so completely as to
exhaust all the possibilities. Its ruins are, as a rule,

1 Romisch-Germanisches Korrespondenzblatt, 1909, p. 40.

2 See Bosanquet, Arch. Ael. xxv. pp. 228 ff., and Curie, A Roman
Frontier Posf, pp. 64 ff.

3 We shall find two such exceptions on the Scottish Wall. A third was
in the East Fort at Welzheim on the German Limes (Der Obergermanisch-
Raetische Limes, Lief. 21, ' Kastell Welzheim,' PL III. fig. I. And there
are others. Of course, it must not be forgotten that, where there was a
bath within the fort, there may also have been one outside. That seems
to have been the case at Cilurnum (Chesters) on the English Limes.

F

82 ROMAN FRONTIER POSTS

not difficult to identify. The comparative abundance
of hypocaust provision is a significant clue. Then the
arrangement of the rooms is more or less conventional,
apses being usually a prominent feature. It was in
these recesses that the actual bathing-places were
situated. The rest of the space, apart from the
dressing-room, was utilized for lounging and various
forms of recreation. Altars to Fortune suggest that
the time was often whiled away in gaming.1 The
baths, in short, were really a species of club, and that is
doubtless why they were generally erected outside the
fort itself. Such a position enabled their advantages
to be shared by the civil settlement beyond the walls,
including women, a participation which would otherwise
have been impossible since, under strict military law,
only soldiers and their servants could enter the gates
of the camp. The whole constituted so extensive a
range of buildings that its remains have often been
wrongly described as a villa. An admirable example,
dating probably from the time of Agricola, is that
which was opened up nine or ten years ago at Inch-
tuthil in Perthshire.2

Regarding the annexes within which dwelt the civil
population — traders, time-expired soldiers, women, and
children — our information is still very defective. It is,
however, certain that such settlements were attached
to some, if not to all, of the forts along the Vallum.
One side was contiguous to one of the sides of the

1 Arch. Ael 3rd series, vol. v. pp. 1 58, describes a recent find of the sort
in the baths outside the fort of Aesica (cf. Year's Work in Classical
Studies, 1909, p. 88, for amended reading). Other cases in point are
Cilurnum (Chesters), Vindolana (Chesterholm), and Habitancium
(Risingham).

*Proc. Soc. Antiq. of Scot. 1902, pp. 214 ff.

ANNEXES 83

camp ; the others were protected by a ditch, and doubt-
less also by a rampart. The outline was generally
irregular, being determined, no doubt, by the conforma-
tion of the ground, and by the number of huts and
booths to be enclosed. The only large building whose
presence is at all constant is the balneum.

An interesting glimpse of a frontier fort with its
annexe is given us incidentally by Arrian, the historian,
who was governor of Cappadocia under Hadrian. In
A.D. 131 he undertook an official tour of inspection,
and sent his imperial master a formal report upon the
results of his observations. This report is unfortunately
lost, but a supplementary account, written in Greek
and containing more personal details, has reached us
under the title : Periplus of the Euxine Sea. It has
been justly characterized as "a unique specimen of a
report made by a Roman frontier officer to his master,
the emperor," and as ranking "with the letters of
Pliny to Trajan as a document of the first importance
for the history of provincial administration under the
Caesars."1 The frontier stations receive particular
notice. The short description of that on the river
Phasis is specially full : "The fort, which is garrisoned
by 400 picked troops, appeared to me to occupy a
position naturally very strong, and very conveniently
situated for the protection of vessels entering the river.
The wall is surrounded by two ditches, both of con-
siderable breadth. Formerly the wall was of earth and
the towers surrounding it were of wood ; but now it is
built of burnt bricks, with towers of the same material.
Its foundations are securely laid ; it is furnished with

1 Pelham, ' Arrian as Legate of Cappadocia,' in Eng. Hist. Review^
1896, pp. 625 ff.

84 ROMAN FRONTIER POSTS

military engines ; in short, its defences are so complete
as to prevent an enemy from even approaching, and to
relieve the garrison from all danger of a possible
blockade. As some protection was required for the
harbour, as well as for the portion of ground beyond
the ramparts occupied by non-combatants and traders,
I resolved to have another ditch dug which will connect
the two that surround the wall with the river. That
will bring within our lines both the anchorage and the
settlement which lies outside the wall of the fort."1 This
singularly vivid picture, drawn by the hand of an eye-
witness, should be kept carefully in mind.

1 Periplus, c. ix.

CHAPTER IV

THE ACTUAL REMAINS : INTRODUCTORY

So far as is known, the first person to attempt anything
approaching a systematic survey of the Scottish Wall
was the cartographer, Timothy Pont. His observa-
tions, made towards the close of the sixteenth century,
were turned to some account by Robert Gordon of
Straloch, who edited the Scottish section of Blaeu's
Atlas (Amsterdam, 1654). Pont's maps are there
reproduced, but they are not on a large enough scale
to be of any real service in connection with crucial
points of topography. His list of forts is also quoted,
the places mentioned being enumerated in so topsy-
turvy a fashion as to suggest that his rough notes have
been allowed to get into confusion.1 Gordon (who, by
the way, speaks of the Wall as Hadrian's) seems
further to have been partly responsible for the form in
which Pont's "draught" appears in Gibson's edition of
Camden's Britannia (i695).2 Absurd as that sketch

I0p. cit. v. pp. 4f. His description is reprinted in full in the Antonine
Wall Report, pp. 35 f. footnote 2. It includes also the 'forts' on the
northern side of the isthmus, which are referred to by various eighteenth-
century writers, but which are not now believed to have had any connec-
tion with the Romans.

8 Op. cit. p. 959, where the " Draught " is said to be " taken from the
Papers of Mr. Timothy Pont (who had exactly traced it) and the obser-
vations of some others who after him had taken the pains to describe it."

86 THE REMAINS: INTRODUCTORY

looks in the shape which it finally assumed,1 it may be
that its most curious features are not purely imaginary,
but have rather developed naturally out of an initial
misapprehension as to the meaning of some of the
actual phenomena. Thus, the " wall of squared and
cut stone, two foot broad ; probably higher than the
Wall to cover the Defendants, and to keep the Earth
of the wall from falling into the Ditch " may be a
deduction based upon a chance glimpse of the outer
kerb of the stone foundation on which the whole
Rampart is now known to have rested.

Timothy Pont was followed by David Buchanan
(circa 1595-1652), whose account of the remains, as he
saw them, was included among the voluminous mass
of manuscripts which he left behind him at his death.
These have long since disappeared.2 Next came Dr.
Christopher Irvine (floruit 1638-1685), physician and
philologist, a man of considerable note in his day both
scientifically and politically. He was more particularly
acquainted with the western half of the Wall, having
several times travelled along its course from the banks
of the Clyde to Falkirk. The subject is not discussed
in any of his published books. But his papers, as well
as those of Pont and Buchanan, were accessible to Sir
Robert Sibbald, the sixth chapter of whose Historical
Inquiries (1707) is entitled: "An A ccount of Antoninus
Pius his Wall, from the Vestiges which yet remain of

Sir Robert Sibbald, however, who was Gibson's informant, says in his
own Historical Inquiries (p. 27): "The learned Robert Gordon of
Stralogh hath from Mr. Ponfs Papers given us a Draught of it," etc.

1 For a reproduction, see Antonine Wall Report^ p. 36.

2 In Bishop Nicolson's Scottish Historical Library (1702), p. 1 6, they
are said to be " still in safe Custody ... in Bundles of loose Papers,
Latin and English"

SIBBALD AND OTHERS 87

it" Sibbald does not profess to have been over more
than a portion of the ground in person, and much of
what he says is admittedly based upon the notes made
by his predecessors. One cannot help regretting that
this material is no longer available.1 We should
probably have learned a good deal from it, if we may
judge from an anonymous manuscript, now in the
Welbeck Abbey Collection, from which we shall have
frequent occasion to quote. This is a letter from a
casual but observant visitor to Scotland, who rode
along the line from east to west in 1697. 2 He
describes the Wall as being "made of stone and turff,
and a ditch behind it."

Much more elaborate than any of the accounts yet
referred to is that published by Alexander Gordon in
his Itinerarium Septentrionale (1726). Born in Aber-
deen towards the close of the seventeenth century,
Gordon took the degree of A.M. at one or other of the
two colleges now united to form the northern Univer-
sity. Thereafter he travelled abroad. On his return
from the Continent, where his taste for the study of
antiquities had doubtless been broadened and strength-
ened, he spent some time, as he himself informs us, in
journeying through different parts of Great Britain
exploring, drawing, and measuring ancient remains.3

1 For the purposes of his Inquiries^ Sibbald also addressed questions to
clergymen and other prominent persons in most of the counties of
Scotland (cf. Nicolson, op. cit. pp. 19-21). These are preserved in the
Advocates' Library, where there is also a manuscript description of
Scotland from Sibbald's own pen. The latter contains little of any
importance regarding the Wall. His 'Directions' to Edward Lhuyd
are more to the point (Proc. Soc. Antiq. Scot., 1910, pp. 319 ff.).

2 Hist. MSS. Commission : Portland Papers^ ii. pp. 54%

3 In his Preface he says : '/ have not spared any Pains in tracing the
Footsteps of the Romans, and in drawing and measuring all the Figures in

88 THE REMAINS: INTRODUCTORY

In 1726 we hear of him as an enthusiastic advocate of
the project for constructing a canal between the Forth
and Clyde,1 an enterprise suggested long before his
day, but not carried out till half a century later. His
Itinerarium is chiefly devoted to the two Walls — the
English and the Scottish. The drawings it contains
are unreliable. But the full, if not always accurate,
descriptions give the book a value of its own. It has,
besides, an interest of quite another sort, in that it
undoubtedly furnished Sir Walter Scott with some of
the colouring for his immortal picture of Jonathan
Oldbuck.2 Gordon walked along the course of the
Scottish Wall from west to east, measuring at frequent

the following Sheets from the Originals; having made a pretty laborious
Progress through almost every Part of Scotland for Three Years
successively!

1 See extracts from the correspondence between Roger Gale and Sir
John Clerk, reprinted in Proc. Soc. Antiq. Scot. vol. x. pp. 380 f.

2 It was, of course, a copy of Gordon's book that Monkbarns had with
him on the journey to the Hawes Inn, and he always professes a deep
respect for the authority of ' Sandy Gordon.' But that is not all. Here
is an extract from the dialogue between Monkbarns and Lovel about
' Aiken Drum's Lang Ladle ' : " [The stone] ' bears a sacrificing vessel,
and the letters A.D.L.L., which may stand, without much violence, for
Agricola Dicavit Libens Lubens? ' Certainly, sir ; for the Dutch anti-
quaries claim Caligula as the founder of a lighthouse, on the sole authority
of the letters C.C.P.F., which they interpret Caius Caligula Pharum
Fecit'" (Antiquary, c. iv.). With this should be compared the following
passage in Gordon's discussion of Arthur's O'on : " And if ever these
Initial Letters I.A.M.P.M.P.T. mentioned by Sir Robert Sibbald were
engraven on a Stone in this Building, it may not be reckoned altogether
absurd that they should bear this Reading ; Julius Agricola Magnae
Pietatis Monumentum Posuit Templum \ but this, my Reader may either
accept or reject, as he pleases. However, I think it may as probably be
received as that Inscription on Caligula's Pharus in Holland, which
having these following letters C C P F, is read Caius Caligula Pharum
Fecit" (Itin. Sept. p. 29). Similarly, the line of Claudian in which Monk-
barns finds an allusion to the Kaim of Kinprunes (' Lie Caledoniis
posuit qui castra pruinis*} is quoted on p. 109 of the Itin. Sept.

GORDON AND HORSLEY 89

intervals and keeping a record of his observations.
He often over-estimates the dimensions of the Ditch,
giving it on one occasion a depth of as much as
35 feet.1 In his " Profile or Section of the whole," it
is made the most important element in the entire work.
Probably it was so, though certainly not to the extent
that Gordon's drawing would suggest. On the other
hand, he was the first to discover the stone foundation
of the Rampart, and to divine its true purpose correctly.2
He was, however, mistaken as to the character of the
superstructure, which he supposed to be "made of
the common Stone and Earth which was dug out of
the Ditch."3

Horsley's Britannia Romana, which saw the light
six years after Gordon's Itinerarium Septentrionale,
was undoubtedly the most judicious treatise that had
up to that date been compiled on Roman Britain as a
whole. His chapter on "The antient and present
state of the Roman wall in Scotland, and the forts
upon it " 4 is valuable chiefly for its careful descriptions
of portions that have been more or less completely
destroyed since the author examined them about 1730.
So far as the structure of the Vallum is concerned,
Horsley agreed generally with Gordon. That is, he
considered that it was composed of "earth or turf,"
resting on a stone foundation.5 His "profile" of the
Wall, however, shows a foundation of five courses of
squared stones, instead of only one, and he sets a high
parapet on the top at the northern side, — a purely

1 1 tin. Sept. p. 55. The place to which he refers is on the eastern slope
of Bar Hill.

2 Itin. Sept. p. 63. 3 Ibid.

* Brit. Rom. pp. 1 58 ff. 5 Op. cit. p. 163.

90 THE REMAINS: INTRODUCTORY

fanciful embellishment. There is more justification for
his view of the Outer Mound. In regard to this he
definitely dissociated himself from Gordon, explaining
the appearance it presented as partly natural, partly
produced by the rubbish that had been thrown out of
the ditch.1

William Maitland, whose History and Antiquities
of Scotland was published in 1757, provides his readers
with a fairly full description of the remains, evidently
based upon a personal inspection. His observations
and deductions are often shrewd enough. But the
value of his observations surfers seriously through his
scarcely concealed determination to belittle his prede-
cessors, Gordon and Horsley. Thus : " We are told by
Gordon, that this wall had a stone foundation through-
out. This is a very great mistake. ... I am of
opinion, that the very few places wherein stones were
laid in the foundation, as a security for the wall, were
placed in oozy grounds ... in other parts, in a dry
soil, it would have been both ridiculous and unneces-
sary to have given a stone foundation to an earthen
wall."" Or again: "By Horsley 's account of this
place one would imagine he had never seen it." 3 And :
" At and near this fort, Horsley seems to have been
quite bewildered."4

In 1755 General Roy brought to bear upon the
Vallum all the skill and experience of a successful
officer of engineers. The results of his survey are
'embodied in the plan which is included in his Military
Antiquities of the Romans in Britain, issued in 1793,
<L year or two after his own death. To him the Ditch

i-Brit. Rom. p. 163, * Hist, of Scotland^ i. pp. 188 f.
3 Op. tit. p. 177. 4 Op. tit. p. 181.

MAITLAND, ROY, STUART 91

was the leading feature of the whole : " the rampart,
with its parapet, as far as can be discerned from the
imperfect remains of them, seeming everywhere to
have been slight and inconsiderable."1 It will be
observed that the parapet, which Horsley invented, is
here taken for granted as a matter of course. Roy was
mistaken, too, in regard to the stone foundation. He
did not believe that it had been continuous. Oddly
enough, however, his theory was the exact opposite of
Maitland's. He held that it had been laid, not where
the ground was soft and wet, but where it was hard
and rocky, being, in fact, simply a means of disposing of
stone that had to be cleared away in making the ditch.2
There is nothing of importance to be learned re-
garding the Wall from the Caledonia of George
Chalmers (1807). The treatment of the subject in
Robert Stuart's Caledonia Romana (1845) *s f~ar more
adequate. Stuart's map, it is true, is nothing more
than a somewhat unsatisfactory copy of Roy's, and
his description of the actual wall is hopelessly mis-
leading, the dimensions being quite impossible. It is
spoken of as "a rampart of intermingled stone and
earth, strengthened by sods of turf, which measured,
it is supposed, about twenty feet in height and twenty-
four in thickness at the base. This rampart or agger
was surmounted by a parapet, behind which ran a
level platform, for the accommodation of its defenders." 3
Still, Stuart had been over the ground himself, and
his book was a really successful endeavour to bring
together all that was known at the time regarding
the Wall and the objects of antiquity that had been
found in association with it. The posthumous second

1 Military Antiquities, p. 156. 2 Ibid. 3 Caled. Rom. 2nd ed. p. 282.

92 THE REMAINS: INTRODUCTORY

edition (1852) represents the last word of the older
school of antiquaries upon the Wall and its problems.
It gained greatly from the active co-operation of Dr.
John Buchanan, whose memory deserves to be held
in high respect by all who are interested in the story
of the Romans in Scotland. For many years he kept
a specially watchful eye on the Antonine Vallum, and
it is undoubtedly to his enlightened enthusiasm that
we owe the preservation of not a few of the surviving
inscriptions.

Stuart and his predecessors had drawn their conclu-
sions almost entirely from what was to be seen above
the surface. They did not realize the latent possi-
bilities of the spade and pickaxe. The credit of being
the first to apply modern methods to the investigation
of the remains belongs to the Glasgow Archaeological
Society.1 Under their direction a series of systematic
excavations was carried out between the years 1890
and 1893, with the result that the true structural
nature of the Rampart was once for all conclusively
demonstrated. Upon that point the Report of their
Excavation Committee is final. Much light was also
thrown upon the manner in which the Road had been
built. It is only as regards the shape of the Ditch
that any real room for doubt remains. The Society's
operations proved useful in another way. Largely in
consequence of the interest they aroused, the Ordnance
Survey Department were induced to have the whole
line very carefully re-surveyed, with the view of securing
an accurate permanent record of the surface appear-
ances. Even the skilled officers of the Survey have
here and there been led astray by natural depressions,

1See The Antonine Wall Report, Glasgow, 1899.

THE GLASGOW COMMITTEE 93

which they have erroneously supposed to be the hollow
of the Ditch. But, on the whole, they have done their
work admirably. It is matter for great regret that
their instructions did not permit them to go further.
Had they had a couple of labourers at their disposal
for a month, and been free to open up the ground
occasionally, they might easily have laid down the
precise course of Rampart and Ditch, almost from sea
to sea. They could certainly have settled at very
small expense the vexed questions as to the exact
points at which the Wall began and ended.

Before attempting to describe the present condition
of the remains, it may be well to state briefly what is
known as to their original form. When intact, the
barrier was probably about 36 English miles long.
Its two most formidable elements were the Rampart
and the Ditch. To the south of these was the Road —
known to Scottish antiquaries as the Military Way.
Forts were planted at slightly irregular intervals, the
average distance between each pair being about a
couple of miles. That the whole formed a single
system, there is abundant evidence to show. The
majority of the forts abutted directly on the Vallum,
which served as their northern defence, and were
crossed from west to east by the Military Way, which
thus constituted the via principalis. Doubts have
been hinted as to whether the great Ditch was really
contemporary; it has been suggested that it was dug
first, and that it had no necessary connection with
anything else. This view, propounded before any of
the forts were excavated and while German archae-
ologists were under a misapprehension as to the
character of their own Limes, would hardly be seriously

94 THE REMAINS: INTRODUCTORY

maintained now by any one competent to form an
opinion. Both at Castlecary and at Rough Castle
the road issuing from the northern gate ran on solid
ground all the way into the open. That is, the Ditch
stopped short on either side to give it a free passage.

That the system was a defensive work is obvious.
No one who has walked along any considerable section
of its course will require to be told that the line was
selected by a soldier. Nature, indeed, had already
done so much that the task of selection must have
been comparatively easy. Ditch and Rampart run
along a series of low hills that rise directly from the
southern edge of a trough-like valley that extends
from sea to sea across the isthmus. The outlook
from Ferguston Muir, for instance, or from the
summit of the Bar Hill is most commanding ; nor
must it be forgotten that in the days of Lollius
Urbicus the difficulty of approach must have been
much greater than it is at present, owing to the
swamps and morasses that doubtless stretched in
front. Curiously enough the fortifications hardly
ever occupy the crest of the slope, but usually lie
on its northern side a little below the summit —
sometimes 30 or 40 yards, sometimes more, some-
times a good deal less. The one notable exception
is on Croy Hill, and that is readily accounted for by
the fact that the northern face is rocky and precipi-
tous. Apart from this, which was plainly deliberate,
the keenest modern strategist would have but little to
cavil at. The choice of position is admirable, much
better than in the case of the English line.1 There

1 See ' Die Abgrenzung des Romerreiches ' by General von Sarwey,
in Westd. Zeitschr. xiii. pp. I flf. The theory there put forward as to the

LINE OF THE WALL 95

is no straining after mathematical straightness. On
the contrary, the dominant motive has been a deter-
mination to secure all the advantages that the ground
could be made to yield. Almost everywhere there is
a clear view to the bottom of the valley. When it
is not so, it will be found upon examination that the
intervening ridge could not have been included except
at a serious sacrifice. No really prominent height is
left out, and, as a rule, the higher hills are crowned with
forts — not, in all probability, because these positions
were hard to assail, but rather because they were
conspicuous and therefore convenient for signalling.
Some such supposition seems necessary, to account for
so notable a departure from ordinary Roman practice.
As we shall see, but little of the Rampart is now
left standing. Except in woods or on the open moor,
it has been levelled to the ground. Where it survives,
it is generally not more than two or three feet in
height ; five feet is probably the maximum. Allusion
has already been made to its stone foundation. In
spite of Roy and Maitland, it is certain that this
substratum was continuous, not intermittent. It con-
sists of two lines of rudely squared kerb-stones
about 14 feet apart, having between them a compact
mass of rough undressed stones, varying in size,
but much smaller than the kerbs which range
from 8 to 1 1 inches in height.1 Across the founda-
tion there run, at intervals, well-built drains or

Scottish Wall is no longer tenable. But the paper is full of very acute
remarks.

1 Mr. Alexander Park, whose opportunities for observation have been
altogether exceptional, is of opinion that the mass of stones had been
thrown in promiscuously and afterwards levelled to about the same height
as the kerbs, which were presumably laid first.

96 THE REMAINS: INTRODUCTORY

culverts, with an entrance at the one kerb and an
exit at the other (Plate V., and Plate IX. 2). These
culverts, which were protected by a covering of
large flat stones, we may take to have been intended
to allow the escape of moisture from the body of the
Rampart, as well as to fulfil the more obvious purpose
of carrying off the water that would otherwise have
tended to accumulate along its southern front

The preparation of this stone base must have cost a
great deal of labour. The material, usually freestone,
did not always lie ready to hand. Much of it would
have to be brought from a distance. Yet the appear-
ance which the whole presents is wonderfully uniform
and regular. Naturally it has proved far more per-
manent than the superstructure which it supported.
There are, no doubt, a few places where it has been
systematically uprooted. Otherwise it must be still
in situ from end to end of the isthmus. Even in
fields which have been under cultivation for centuries,
a transverse trench, cut two or three feet deep upon
the proper line, will expose the stone foundation
resting upon the till, exactly as the legionaries laid
it. Few of the farmers through whose ground it
passes have any difficulty in locating its position.
The plough is apt to grate hard upon it in winter,
occasionally dislodging heavy stones, while in spring
and summer a corresponding poverty of growth can
readily be detected above it.

This, of course, refers only to the stone base. Two
features of the superstructure that had been unnoticed
by all previous observers early attracted the attention
of the Glasgow Committee, — the systematic layering in
the soils of which it is composed, and the complete

PLATE V

I. ON EAST SLOPE OF CASTLK HILL (BAR HILL

2. AT ROUGH CASTLK

CULVERTS BENEATH THE RAMPART

STRUCTURE OF THE RAMPART 97

absence of stones. Their sections, which were cut
transversely, showed traces of thin lines running
across, with breaks and irregularities, but as a rule
parallel to one another. The lines were sometimes
dense black, sometimes merely dark or purple. They
varied in thickness from a quarter of an inch to one
inch and occasionally two inches. The layers sepa-
rated by the lines were seen to be, on an average,
from three to four inches thick (see Plate VI. i and 2).
Dr. George Neilson was the first to explain these
phenomena.1 He showed that the wall had been
built of sods, course upon course, like a wall of stone.
There would be in such a work an unfailing series of
relatively thick horizontal layers of earth, separated by
thin horizontal lines of heath or grass and by a thicker
or thinner strip of dark soil formed from the decay of
previous vegetation. "Hence," Dr. Neilson concluded,
"a structure of sods must, so long as it stands, bear in
its interior the evidence of its origin, for the rows of
aligned sod surfaces could never cease to contrast with
the soil on which they rested, and would thus pencil
the face of any section with their ineffaceable parallel
dark lines."2 The descriptive phrase of Julius Capito-
linus was therefore literally accurate. The Wall of
Antoninus Pius was a murus cespiticius in a sense
that had never before been thoroughly understood.

The breadth of the stone foundation suggests that
the breadth of the lowest course of the rampart was
about 14 feet, and, as will be noted below, the
traces of the sods prove that this was actually the

1 See also Dr. G. Lowson in Trans. Stirling Nat. Hist, and Arch. Society,
1892, p. 66.

2 Antonine Wall Report, p. 123.

G

98 THE REMAINS: INTRODUCTORY

case. As the height increased, the breadth must
have been diminished in order to secure stability.
The precise extent of the " batter " cannot now be
ascertained. But the Glasgow Committee, proceed-
ing on the assumption that the top was about 6 feet
broad — an assumption that would satisfy the canon
laid down by Vitruvius1 for the wall of a fortified town-
arrived tentatively at the conclusion that the rampart
might originally have been about 10 feet high. This
calculation suits the apparent " batter " in those sections
where it is more distinctly discernible, and it is also in
reasonable accord with the proportions indicated by
Hyginus,2 when he says that the minimum thickness of
a rampart was 8 feet and its minimum height 6.
Whether the top was surmounted by a palisade or by
any form of battlement, we cannot possibly say. The
evidence has disappeared beyond hope of reconstruc-
tion. The explanation of its disappearance is not far
to seek. A mass of debris, of greater or less depth,
extends for about 12 feet beyond the kerb on either
side. It was this debris that misled Stuart as to
the original breadth ; it will be remembered that his
estimate was as much as 24 feet. As a matter of fact,
the Glasgow investigation showed that the material
lying outside the kerbs " is always of much the same
character as that above the stone foundation, with the
single fundamental difference that, while occasional
faint irregular traces of something like lamination can
be detected, these are rare. . . . Only in one or two
exceptional instances are the layers in the vallum pro-
tracted more than 6 inches beyond the line of kerbs.
This material therefore is, normally, not stratified."3

1 Bk. i. c. 5. 2 C. 50. 3 Antonine Wall Report, p. 45.

Pl.ATK VI

I. OX THE BAR HILL

2, NEAR BONNVSIUE

SECTIONS THROUGH THE RAMPART

RAMPART AND DITCH 99

It is plain that the unstratified mass represents the
upper part of the rampart which, under the influence
of various agencies, has fallen or been thrown down,
and now slopes from the sides of the lower portion,
commingled with it indistinguishably until the interior
is laid open.

The great Ditch runs on the outer or northern side
of the Rampart. Among the older writers, Gordon,
Horsley, and Stuart have definitely formulated their
ideas as to its original shape. Their diagrams show
that they conceived of it as having had steeply sloping
sides and a broad flat bottom. That opinion is cer-
tainly erroneous. It is much more nearly true to say
that the Ditch was of the V-shaped type, — a fossa
fastigata. The Glasgow Committee found that the
angles of scarp and counterscarp were very much the
same, averaging from 26° to 30°. l They were unable
to decide whether the two actually met so as to form
an acute angle, or whether they descended in such a way
as to leave a narrow strip of level ground between their
extremities ; in the latter event the plane surface that
constituted the true bottom could not have been more
than a couple of feet wide. The difficulty of making a
clear pronouncement was due to the lack of evidence.
Only three or four sections of the Ditch were taken,
and these were cut at points where it was still open to
a very considerable depth and where the apparent
bottom was therefore covered with a thick mossy
growth. Consequently, in the words of the Report,
" water followed the spade too quickly for leisurely
examination."2 It is easy to be wise after the event,
and one can now see that it would have been better to

lAntonine Wall Report, p. 136. 2 Antonine Wall Report^. 106.

ioo THE REMAINS: INTRODUCTORY

select for investigation one or two spots where the
Ditch had long been almost or altogether filled up.

The recommendation just made may seem para-
doxical. But the risk of a premature inrush of water
would certainly have been much smaller. And it
is well known that, if the surface of the earth as
consolidated by geological forces be once disturbed by
the hand of man, the traces of that disturbance can
never be wholly obliterated, though in some soils and
places they may be clearer and sharper than in others.1
Except where many feet of soil have been bodily
removed, as in cuttings for canal or railway, the outline
of the Antonine Ditch will be safely preserved under-
ground for centuries to come. It sometimes happens
that sections are accidentally exposed to view. In
October, 1909, for instance, there was a beautiful
example to be seen in the sand-pits at Kirkintilloch.2
The shape of the Ditch was silhouetted in dark soil
against the light-coloured bank of sand in which it had
originally been cut. The result was to suggest that
the bottom had had a form not heretofore suspected.
A foot or two from the lowest point, scarp and counter-
scarp suddenly became perpendicular, and so continued
to the end of the descent (Figure i). This impres-
sion possibly requires confirmation, as sand is, in the
nature of things, less compact than boulder-clay. But
it may be pointed out that the same feature was noted
in the ditches round the Antonine fort at Bar Hill. It
had been previously observed in the ditch attached to

1 For a singularly lucid explanation of the methods of scientific exca-
vation, based upon this principle, see the letter from Colonel Stoffel
printed by Rice Holmes in his Caesar's Conquest of Gaul (pp. xxviii ff.).

2 Examined by Mr. Alexander Park and myself.

SHAPE AND SIZE OF THE DITCH 101

the Turf Wall at Appletree, on the line of the English
Limes, and it has since been found in the ditches of
one of the five Newstead periods.1

If the shape of the Antonine Ditch has still to be
accurately determined, its exact dimensions are at least
equally doubtful. There was obviously considerable

FIG. i

variation, the extent of this, no doubt, depending in
large measure on the nature of the ground. Along
the greater part of its course it probably did not
average much less than 40 feet in breadth. In special
circumstances such as obtain on the northern slope
of Croy Hill and near the summit of the Bar Hill,
where it had to be hewn out of the solid rock, it is
considerably narrower — sometimes hardly more than
20 feet. As it usually lies upon an incline, the counter-
scarp is almost invariably to some extent 'made up.'2
The normal depth is difficult to estimate, in the

1 Since the above was written, Mr. Mungo Buchanan has shown me a
sketch, made in his notebook in 1894, of a section of the Ditch which was
accidentally exposed during the building of a villa, just at the point where
the road to Camelon passed the line of the Wall. This sketch exhibits
the precise peculiarity here described. Mr. John M'Intosh informs me
that it is characteristic also of that portion of the Ditch which lies
immediately to the north of Bar Hill fort, but that in a section which he
opened up a little further east, where the cutting is through solid rock,
there was no perpendicular drop at the bottom.

2 Antonine Wall Report, p. 137.

102 THE REMAINS: INTRODUCTORY

absence of satisfactory cross-sections. The Glasgow
Report reckons it as about 12 feet, and this may not be
far from the truth, albeit it is little more than half of
what Gordon was disposed to allow for as a minimum.
But there are certainly places where the Glasgow
figure would be an under-statement. A case in point
is the first part of the steep ascent from Dullatur to
Westerwood, where to this day the hollow makes a
very close approach to the dignity of a ravine. As a
rule, it is of course comparatively inconspicuous. And
yet it is astonishing how successfully the huge furrow
has resisted the efforts time has made to obliterate its
traces.1 If the sections of it that are still distinctly
visible could be placed end to end, they would stretch
for many miles.

Naturally the line is always seen to greatest
advantage where the ground has never been under
cultivation (Plate VII. 2). In its passage over arable
land it shows as a more or less well-marked depression
of the surface (Plate VII. i). A practised eye can
often detect it, even where there is nothing that would
attract the notice of a casual observer, by accepting the
guidance of a dip in the hedge or stone dyke that
separates field from field ; these boundaries are
usually of long standing, so that the contour whose
memory they perpetuate is generally older than that
of the existing levels. The task of discovery is also
rendered easier by the circumstance that experience
soon teaches one to look for the indentation in just
its right place — a little way down the northern slope of
the hill. Lastly, should other indications be wanting,

1 So, too, on the English Limes the ditch has survived far more visibly
and continuously than even the stone wall.

THE MILITARY WAY 103

there is a chance that the farmer or his men may be
able to point to a strip of land on which the grass
never fails to grow greener or the corn deeper. In
spite of all, however, there are occasions where doubt
and uncertainty will be inevitable. More than one
important question of direction waits for the solution
that only the spade can bring.

The vestiges of the Military Way lie a little to the
south of the Vallum. The two may be roughly de-
scribed as parallel, the average breadth of the interval
between them having been probably not more than 40
or 50 yards. The road, however, wound somewhat,
in order to accommodate itself to the inequalities of the
ground. Here and there, notably at Hutcheson Hill,
it made a wide detour to avoid an unnecessary ascent.
On the other hand, as its essential purpose was to
facilitate communication between the forts, it was
bound to come close to the great Rampart wherever
these occurred. The sections cut by the Glasgow
Committee showed it to be of simple, yet substantial
and effective, construction. "Its statumen is a base
of fairly large stones, above which a stratum of smaller
stones of various sizes is laid, rising to a rounded crown
in the centre, and giving the surface of the road that
convexity which modern as well as ancient road-makers
have found expedient. It has no squared kerbs —
indeed, ordinarily, the kerbs are hardly distinguishable
from the other stones of the base."1 Only at one
point did anything like a paved surface appear.2 It is
usually from 16 to 18 feet wide. During the excava-
tions at the Bar Hill a considerable length of it was
uncovered (Plate XVII. i), when it proved to be

1 Antonine Wall Report, p. 149. 2 Antonine Wall Report, p. 114.

104 THE REMAINS: INTRODUCTORY

about 17 feet in width, and excellently constructed.
Its foundation was formed of a stratum of fairly large
stones resting on a bed of wrought clay. This was
surmounted by a convex layer of smaller stones,
providing a surface whence the water must have
drained away quickly into one or other of the two
gutters that ran along the sides.1

It is not surprising to find that the Military Way is
in a much more ruinous condition than either the Ditch
or the Rampart. Lying as it does upon the surface, it
has been exposed to various risks. In a few places —
on the western shoulder of Croy Hill, for instance, —
it is a grass-grown track, in regular use by foot pas-
sengers to-day. In others — at Bearsden, for example,
— it has developed into a macadamized highroad of
the most approved modern type. Elsewhere its traces
are just faintly discernible, it may be as a narrow and
hardly noticeable ridge, it may be only as a ' stream of
stones ' that hampers the smooth working of the plough.
But for a large part of its course it has been torn up
so completely that its disappearance must be regarded
as absolute.

Rampart, Ditch, and Military Way were the three
most prominent features of the Limes, — a term which is
certainly the most convenient, and probably the most
correct, way of describing the combination. But there
was a fourth, regarding which a word must be said.
This was the " northern agger," whose appearance so
impressed Gordon that he walked half way across the
isthmus before he realized that there was any other
" agger " m tne question at all.2 The Glasgow Com-
mittee call it the "outer mound." They admit that

1 Roman Forts on the Bar Hill, pp. 18 f. 2 Itin. Sept. p. 58.

I'l.ATK VII

I. EAST OF CASTLECAKY

2. WEST OF ROUGH CASTI.E

VIEWS OF THE DITCH

THE OUTER MOUND 105

the term is not altogether satisfactory,1 but it is difficult
to find a better. Whatever name be applied to it, its
relation to the Ditch is intimate and unmistakable. It
is simply the upcast thrown out by the diggers as their
work proceeded. The manner of its disposition, how-
ever, would appear to have been methodical. The
bulk, if not the whole, of it was accumulated on the
northern bank, which was seemingly covered with
grass and heather when the first spadefuls were flung*
down.2 Even in the marked irregularity of form which
it exhibits, the Glasgow Committee thought they could
detect some approach to a guiding principle. "It is
that where the natural levels of the two sides of the
ditch were much the same (and particularly if the north
side had slightly the advantage), it was not expedient
to heap up the earth with a narrow surface or a sharp
crown, as that would have been to rear on the north
side of the fosse a rival to the vallum on the south side.
Hence, where there was an original approximate equality
of level on both sides of the fosse, the outer mound is
spread out with a broad, flattish surface. On the other
hand, where the ground across the ditch had a large
fall northward and lay clear below the berm, the outer
mound has the heaped-up shape, a somewhat narrower
base, and a higher and narrower crown, which some-
times is only a rounded angle."3

If this "rough and very general rule" can be
accepted — and it was formulated after careful obser-
vation— it furnishes an additional argument against
the theory that there is no necessary connection

1 Antonine Wall Report, p. 138.

2 Stuart's Caled. Rom. 2nd ed. p. 321, footnote.

3 Antonine Wall Report, p. 139.

io6 THE REMAINS: INTRODUCTORY

between Ditch and Vallum : if the latter had not
been in existence, or at least in contemplation, when
the former was dug, it would obviously have been
impossible to take account of it in disposing of the
upcast. In any event it is important to note that the
Outer Mound is not a glacis in the sense in which that
term is understood by military engineers. " It raised
the level of the ground outside, thus increasing the
relative height of the position which would be taken
by an enemy ; and it was not so spread out as to
prevent its being used for hostile purposes."1 In fact,
so far from being a defence against assault, it would
have proved a source of weakness, for there are many
points at which it would have served no other purpose
than that of affording cover to an attacking party.

A similar apparent disregard of purely military con-
siderations is observable in connection with the space
between the edge of the Ditch and the foot of the turf
Rampart. Experts in fortification regard a ' berm' as,
at the best, but a necessary evil. Some interval may
be required, if the rampart is to be prevented from
slipping down into the fosse. But it is a military
axiom that any such ledge should be as narrow as is
consistent with stability. The wider it is, the greater
the risk of the enemy getting a foothold upon it. To
be safe, the maximum should not be more than 8 feet.
In this case, however, the intervening platform is
usually about 20 feet broad, seldom less. Sometimes
it extends to 27, 35, 56, and 67 feet. Occasionally
«ven these figures are exceeded. Take, for instance,
the measurements recorded from Croy Hill by the
Glasgow Committee. At one of their sections (No. 7)

^Antonine Wall Report, p. 140.

THE BERM 107

the ' berm space ' was 1 1 2 feet, rising 24 feet further
on to 1 1 6 feet, and then contracting to 100, 95, 70, and
54 feet.1 It is true that the conditions on Croy Hill
are exceptional. But, after every allowance has been
made for this, the variations are almost as remarkable
as the original width.

1 Antonine Wall Report, p. 64.

CHAPTER V
THE LINE OF THE RAMPART AND DITCH

As has already been indicated, the revised 25-inch
and 6-inch scale maps of the Ordnance Survey1
provide, on the whole, an excellent guide to the line
originally followed by the Scottish Limes. Here and
there, however, positive correction is called for, while
at some points the evidence supplied by surface
appearances can be supplemented by reference to
records that were compiled at a time when the oppor-
tunities for observation were more favourable than
they are to-day. There is room, therefore, for a
somewhat fuller and more careful verbal description of
the route than any hitherto attempted.2 It can make
no claim to perfect accuracy. But it should prove
helpful to casual enquirers. And it may facilitate

1 The results of Colonel Ruck's observations have not yet been recorded
on the one-inch scale issue. But they are reproduced in the map pub-
lished at the end of this volume. No modification has been made upon
them in the map ; it would have been idle to attempt it without a proper
survey. Attention is, however, drawn in the text to places where caution
seems to be required.

2 The whole ground has been traversed more than once on foot.
Doubtful points have been visited several times, usually in the company
of friends in whose judgment I had special confidence. Among those
who have given me assistance in this way, I would particularly mention
Mr. A. O. Curie, Secretary to the Royal Commission on Ancient Monu-
ments, Mr. Alexander Park, and Mr. John M'Intosh.

OLD KILPATRICK TO BEARSDEN 109

the task of future explorers, should it ever prove
possible to secure a thorough archaeological investiga-
tion. We shall begin at the western end. Sibbald
and Gordon proceed from west to east, and so do
Horsley, Roy, and Stuart. The perverse Maitland
prefers the opposite direction, and he has been fol-
lowed by Hiibner.1 No advantage, however, is gained
thereby ; the only result is to render comparison with
other accounts more difficult.

i. FROM OLD KILPATRICK TO BEARSDEN 2

Practically all who have written upon the matter are
agreed that the western termination of the Wall must
be looked for on the banks of the Clyde, near the
church of Old Kilpatrick. Bede, in a passage with
which we are already familiar,3 uses the vague expres-
sion " near the city of Alcluith" ; and, on the assump-
tion that Alcluith is Dumbarton, there has been a
disposition in some quarters to prolong the line of
the Limes, if not as far as that town, at all events to
Dunglass, where a ford was in existence until the
river was artificially deepened. It was surely essential,
it has been argued, that the Romans should control
the passage. Supposing it were so, however, the
building of a small castellum on the southern shore
would have been both cheaper and more effectual.
Besides, a continuance of Rampart and Ditch along
the river-bank would have been, in the nature of

1 C.I.L. vii. pp. 191 ff.

2 Each of the sections indicated in this chapter can be readily over-
taken in a day's walk, allowance being made for the nature of the ground
to be traversed and the points of interest that are included.

3 See supra^ p. 34.

no LINE OF THE RAMPART AND DITCH

things, illogical. A limes was an artificial frontier, and
there was no need to construct limites on the enemy's
side of a natural boundary. That a Roman road,
with military posts upon it, once ran to Dumbarton or
beyond it, is of course quite possible. Perhaps it is
even probable, in view of the evidence of Pont1 and
Irvine,2 and of Horsley.3 But it should be noticed
that Pont in his map makes the line of the Wall begin
at the ' Kirk of Kilpatrick,' and that Horsley, on a
balance of all the evidence, rather inclines to agree
with him.4 George Buchanan, too, the earliest reliable
writer to express an exact opinion, regarded West
Kilpatrick as the end of the Wall,5 as did John of
Fordun before him.6 Without excavation absolute
certainty is not, perhaps, to be hoped for. But in the
meantime it will be safest to take as our starting-point
the Chapel Hill, a knoll lying on the left-hand side of
the main road to Dumbarton, some 600 or 700 yards
beyond Old Kilpatrick Church.

Standing upon this knoll and looking northwards,
one cannot fail to be impressed by the manner in
which the whole landscape is dominated by the range
of the Kilpatrick Hills. Their slope here is unusually
steep, and at no point along its entire line did the

1 On the assumption, that is, that the ' Dunvass ' which concludes
Font's list is Dunglass.

2 Quoted by Sibbald, Hist. Inq. p. 28.

3 Horsley was of opinion that 'the military way has certainly been
continued as far as Dunglass, for it is still very visible at Dunnerbuck
within half a mile or little more of Dunglass* (Brit. Rom. p. 159).

4 Op. cit. p. 164.

6 Finiebatur ad occidentem ad locum, qui nunc Fanum Patricii dicitur
(Rerum Scot. Hist. Bk. v. c. 6).

6 Scotichronicon, Bk. iii. c. 4 : In ripa fluminis de did juxta Kyrk-
patryk failebatu r.

OLD KILPATRICK TO BEARSDEN in

frontier of Agricola and Lollius Urbicus approach so
close to the majestic natural barrier that defended
Caledonia. The Chapel Hill itself, undoubtedly the
site of a Roman fort, is actually the termination of one
of the spurs which the range throws off as it sweeps
westward to hem in the Clyde. From here the course
of the Roman Wall must have been set almost due
east. On the hillside, half a mile away, stands the
farmhouse of Mount Pleasant, immediately before
reaching which one can detect very faint traces of
what may be the depression of the Ditch. The most
conspicuous is a dip in the stackyard wall. Had they
stood by themselves, these indications would have been
negligible. But their testimony becomes important,
when it is confirmed by the fact that, until recent
repairs were effected, a marked subsidence was visible
in one of the walls of the farm-buildings1 — a sure sign
that the ground below has at some time been seriously
disturbed. Beyond Mount Pleasant the line appears
to have swerved slightly to the right as it descended
the slope towards the Sandyford Burn. This view, it
is fair to add, does not quite agree with that of the
Ordnance Survey officers, who thought they could
detect vestiges of the Ditch running exactly east and
west at the foot of the hill, for about 200 yards on this
side of the burn. The configuration of the ground is
not favourable to their opinion ; such a direction would
carry the line a good deal lower than one would
have expected. At the same time it would be unwise
to formulate a definite conclusion until certainty can
be attained by cutting a trench or two across the
field.

1 Information from the Rev. Robert Munro and Mr. John M'Intosh.

ii2 LINE OF THE RAMPART AND DITCH

Immediately beyond the Sandyford Burn the ground
rises fairly rapidly, and the dip of the Ditch is unmis-
takable on the side of the hill. When the ascent
becomes less steep, the track begins to be elusive
again. For the first three fields, however, the mark-
ings on the Ordnance Map are probably approximately
correct. In the second field, in December, 1909, the
plough turned up quite a number of large hammer-
dressed stones such as were usually employed for the
kerbs of the Rampart. As they lay exposed on the
surface,1 they doubtless presented an exact reflection
of the course of the stone foundation. After crossing
the farm-road which bounds the third field on the east,
the indications on the Ordnance Map are very distinct,
— much more distinct than the present surface appear-
ances would seem to justify, and at the same time fully
30 yards further south than one would have expected
to find them.2 Here again there is urgent need of
•excavation before the line can be laid down with
perfect confidence. But just as it reaches the road
that leads to the farm of Carleith, about 1400 yards
from Mount Pleasant, the Ditch can be fairly well
made out as it emerges from the hollow that flanks
the road on the left. Then everything is once more
doubtful.

The Survey officers, indeed, show the "remains of
the Wall of Antoninus Pius " inclining sharply to the

1 Seen by Mr. A. O. Curie, Mr. John M'Intosh, and myself.

2 This opinion, unanimously reached by my friends and myself at three
successive visits, was independently confirmed by Mr. John Maclaren,
the tenant of Carleith. Here, as well as further east, he believes that the
Wall runs about 40 yards north of the old road. The old road may,
therefore, quite possibly be the representative of the Military Way ;
see infra.

OLD KILPATRICK TO BEARSDEN 113

right at this point and keeping close to the line of
the road to Duntocher, the so-called ' Roman Road.'
After about 350 yards their markings cross the road,
and run on the south of it for about the same distance,
when they recross and continue visible for some 30x3
yards more. This secures the inclusion of a deep
hollow on the crest of the hill, which has been gene-
rally supposed to be part of the Ditch. The soundness
of the supposition is, however, open to question ; as it
appears at present, the hollow lacks some of the most
characteristic features of the Roman fosse. Besides,
the tenant of Carleith unhesitatingly indicates the true
line as lying some 30 or 40 yards to the north of the
road, — precisely in the position where the formation of
the ground would suggest that it should be looked for.
If he is right, and there can be little doubt that he is,
then the ' Roman Road ' follows the track of the old
Military Way, and the Limes must have bent round
more or less quickly, to reach Duntocher Burn close
to the ' Roman Bridge ' at Duntocher village.

From the bridge one can see the Ditch climbing the
shoulder of the Golden Hill ; its line crosses the burn a
few yards higher up.1 On the hill are the remains of the
fort. The Ditch passes along its north-west front, and
can be followed thereafter for some three miles without
the slightest difficulty. At first it makes direct for a
clump of beech trees about 600 or 700 yards away, at
the foot of a long descent. When the trees are reached,
it swerves a little to the left and becomes identical with
the line of the modern road that leads past the farm of

1 The stone foundations now visible beyond the wall 'that runs between
the bridge and the church have, however, nothing to do with the founda-
tion of the Rampart. They are the ruins of a cottage.

H

1 14 LINE OF THE RAMPART AND DITCH

Cleddans.1 For some little distance this road seems to
have been laid on the berm or platform that extended
between Ditch and Rampart. Traces of the former
are sometimes visible on the left hand, and traces of
the latter on the right. The Ditch, however, is ulti-
mately crossed ; it shows with particular distinctness
on the right after Cleddans farm is passed, and
continues plain all the way to the Cleddans Burn.
From the burn the line, still clearly visible, proceeds
straight over Hutcheson Hill, and down into the Peel
Glen — where Gordon thought he found traces of a
Roman bridge2 — while the Military Way seems to
have kept round the foot of the southern slope, thus
avoiding the steep ascent.3

The eastern side of the Peel Glen is formed by the
Castle Hill, on whose tree-girt summit are the ruinous

1 We shall meet with the name ' Cleddans ' or ' Cleedins ' at two other
points on the line of the Limes. See infra, p. 121 and p. 136, footnote.
Dr. W. J. Watson informs me that it may quite well be an Anglicized
form of the Gaelic cladhan, meaning either ' the place of the ditch ' or
'the place of the rampart' ; both senses of clad are possible.

2 1 tin. Sept. p. 52. There is nothing whatever of the kind to be seen
now.

3 Stuart says : " On leaving the Peel Glen it again approached the
intrenchments gradually nearing them as they ascended to the fort in
advance. It has been thought, however, that another branch of the
causeway had been conducted along the low grounds, and almost in a
straight line towards Kilpatrick, for the purpose of avoiding the adjacent
heights" (Caled. Rom. 2d ed. p. 305). Cf. Roy, Milit. Antiq. p. 158.
As a matter of fact, the Revised Ordnance Survey Map shows very
distinct traces of the Military Way at the north edge of Garscadden
Wood. I have not been able to visit the spot personally, but it seems
unlikely that the officers would have marked the road in so very
unexpected a place unless the evidence had been beyond dispute. If they
are right, the traces must belong to such a ' branch ' as Stuart describes.
Whether it was really a branch, is another question. It may represent a
different period in the history of the Limes, as a similar 'branch' at the
fort of Rough Castle undoubtedly does.

Pl.ATK VIII

I. KAST OK THORN FARM

2. WKST OF THORN FARM

VIEWS OF THE DITCH

OLD KILPATRICK TO BEARSDEN 115

remains of the third of the Roman 'stations' on the
Limes. The depression of the Ditch, as it climbs this
hill, has been selected as a dividing line between field
and field, a hedge being planted in the middle. When
the fort on the top is reached, the Limes, which has
been pursuing a north-easterly course since the
Cleddans Burn was crossed, turns several points to the
right and proceeds due east, descending slowly for
about 650 yards. At this point it becomes very well-
marked, as it swings round towards the south-east in
order to gain the face of the rising-ground of Led-
cameroch. As it passes along the front of the low hill,
the direction is once more easterly. Here the track
runs for a certain distance among trees, and the remains
of the Ditch are consequently unusually distinct
(Plate VIII. 2). Just beyond the steading of Thorn
Farm there is yet another bend to the right, the south-
easterly inclination being maintained for 500 or 600
yards more, when the easterly course is resumed.
From the farm-steading to the public road1 the depres-
sion is quite unmistakable (Plate VIII. i). Even
beyond the road it can be followed for a little distance
through a piece of waste ground that stretches between
the backs of two rows of houses. Thereafter it loses
itself in a wilderness of villas and gardens, some of
which can boast of rockeries that have been built
of material taken from the stone foundation of the
Rampart. For about 600 yards, however, as we can
see from the Ordnance Survey Map, it must have
lain a little way back from the south bank of the

particular road did not exist when the O.S.M. was revised.
Indeed, the growth of Bearsden has been such that the map as a whole is
here very much out of date.

n6 LINE OF THE RAMPART AND DITCH

tiny streamlet known to Bearsden residents as the
Manse Burn. This section included the north front of
the ' station ' of New Kilpatrick. The surface traces
of the fort have now completely disappeared. But,
when the Survey officers visited the spot sixteen
years ago, enough remained to show that the thorough-
fare which goes by the name of * the Roman Road,'
traversed it from east to west. The claim of the
latter to represent the Military Way is therefore
beyond dispute.

2. FROM BEARSDEN TO KIRKINTILLOCH

Parting from the Manse Burn just where the Roman
Road crosses the railway to Milngavie, Ditch and
Rampart mounted the northern front of the ridge that
leads to Ferguston Muir. Modern villas have en-
croached upon the line to some extent, and have helped
to obliterate the track. But near the top, shortly before
the new Cemetery comes in sight, Rampart, Ditch, and
Outer Mound are all encountered together in fairly
good preservation (Frontispiece), the trend being more
decidedly north-easterly. The Ditch here is still 4 or
5 feet deep, and between 30 and 40 wide. Within the
Cemetery a portion of the stone base of the Rampart
which was exposed during the extension operations,
some eight or ten years ago, has been allowed to
remain open. It runs very nearly due north, and is a
highly interesting section, including as it does not only a
good example of a culvert laid across it1 (Plate IX. 2),
but also a specimen of a device that does not seem to
have been noted elsewhere ; the Rampart here climbs
a steep slope and the stone foundation appears to have

1 The covering stones have, however, been removed.

PLATE IX

I. FOUNDATION OF RAMPART, SHOWING STEP

2. FOUNDATION OF RAMPART, SHOWING CULVERT

IN NEW KILPATRICK CEMETERY

BEARSDEN TO KIRKINTILLOCH 117

been stepped (Plate IX. i).1 A little way in front is
the hollow of the Ditch.

After it has passed beyond the enclosure of the
Cemetery, the line, which has turned to the east again,
gradually approaches the public road, finally crossing
it almost immediately upon quitting the Hillfoot golf-
course. Here it curves, at first slightly towards the
south and then, with an ampler sweep and for a longer
distance, towards the north, the objective being the
wooded top of a hill, usually known as Crow Hill, but
sometimes called the Temple, about 800 yards ahead
of the last-named turn. It can be made out very
easily, for the field boundaries follow it all the way.
They follow it for rather more than 600 yards further
as it descends, after a sharp turn on the summit, to
south-east by east. Keeping straight on, it traverses
an open field and then crosses the public road leading
from Summerston Railway Station to Milngavie.
About 200 yards to the east of the road it reaches a
hedge on the brow of the hill. There it bends abruptly
to the right and drops quickly down, south-east by
south, to the farm of Summerston. Passing through
the steading, it is almost completely lost in the fields to
north and south of the railway line, but it can be faintly
distinguished again as it approaches the bank of the
Kelvin at a point a little to the east of the old ford.

On the other side of the river lay the fort of
Bemulie, one front of which must have been protected
by the Rampart and its companion Ditch. Without
excavation the precise details are not now recoverable.
But it is certain that beyond the fort the Limes turned

1 Cf. the evidences of stepping in the case of the N. Wall of the fort at
Castlecary (Proc. Soc. of Antiq. of Scot. 1903, p. 291).

n8 LINE OF THE RAMPART AND DITCH

almost at a right angle. From a point a little way
below the modern farmhouse the familiar hollow can
be traced for nearly a mile in a straight line. Dr.
John Buchanan tells us that in this neighbourhood, "so
late as 1812, \\\e fosse was sufficiently deep to render a
person on horseback within it invisible from the out-
side ; and its sides were steep and difficult of descent."1
The depth now is seldom more than 2 or 3 feet. The
direction is easterly, with a perceptible inclination
towards the north. At first it runs in the field im-
mediately to the north of the public road, from which
it is barely 100 yards distant. Road and Ditch, how-
ever, rapidly approach one another. About 400 yards
beyond the farm buildings of Easter Balmuildy they
draw together and keep on side by side for nearly 600
yards more, the road being actually on the berm for
the greater part of the distance. Close to a lint-pond,
which has been formed from the bed of the Ditch, the
road makes an abrupt turn to the right, while the Ditch
maintains an undeviating easterly course. The road
is intersected by the track of the Military Way 100
yards or so from the lint-pond, and 50 yards further on
it takes a sharp turn to the left, thus resuming its
parallel relation to the Ditch, the indentation of which
will readily be found in the field, about 1 50 yards off,
with the Military Way 100 yards behind it. A short
distance west of the point where the public road bends
a little to the north, the long straight stretch of the
Limes ends ; the Ditch swerves slightly to the south
until it comes within some 50 yards of the road. The
two then run parallel for 350 yards, when they finally
part company, just beyond a second lint-pond. Those

1 Stuart's Caled, Rom. 2nd ed. p. 320, footnote.

BEARSDEN TO KIRKINTILLOCH 119

who would follow the Limes must now quit the road
entirely. At this point it enters the policies of Cadder
House, within which its precise course speedily becomes
doubtful. On the whole, it seems probable that it
passed sinuously along the face of the hill,1 instead of
pursuing the straight line which the Ordnance Map
appears to indicate. Trenching, however, is essential
to an accurate determination of the question, although
it seems likely that the remarkable mound, which is
encountered about 200 yards west of the Forth and
Clyde Canal, and which will be described in more
detail at a later stage, was on the southern side of the
Roman boundary.

As we shall see, it is tolerably certain that there
must have been a fort somewhere in the neighbour-
hood of Cadder Manse, although its exact situation
cannot be ascertained without further digging. What-
ever its position may have been, it is clear that Rampart
and Ditch crossed the line of the Canal at the bend to
the north of the Manse. The remains of the former
show at once in a strip of wood close to the field
boundaries. Those of the latter are well-marked as
they traverse the farms of Bogton and Easter Cadder.
For noo yards they lie immediately to the north of
the main road from Glasgow to Kirkintilloch. About
100 yards west of the bridge that carries this road over
the Canal, they leave the farm of Easter Cadder,
crossing first the road and then the Canal itself, the
north bank of the latter being reached close to the end

lrrhis was the course pointed out to me as probable by the Rev.
J. B. A. Watt of Cadder, with whom I entered the policies from the east.
On a subsequent occasion Mr. A. O. Curie and myself, approaching from
the west, came to a similar conclusion.

120 LINE OF THE RAMPART AND DITCH

of a row of small houses that runs south-eastwards
from the bridge. From this point to the Peel of
Kirkintilloch, generally pointed out as the site of a fort,
the distance is rather less than a mile and a quarter.
For the first part of the way the depression indicating
the Ditch continues to be perceptible, though not con-
spicuous. By and by, however, it stops short on the
edge of the Sand Pits. These have extended con-
siderably since the Ordnance Survey Map was revised,
so that a good deal of what is laid down there has now
been completely removed, the sand being quarried out
to a level below the original bottom of the Roman
Ditch. From the western edge of the Sand Pits to the
Peel there are no definite surface indications to be
seen. But the general 'lie' of the ground would
suggest that the direction had been straight.

3. FROM KIRKINTILLOCH TO DULLATUR

For 600 or 700 yards beyond the Peel the streets
and houses of the town effectually conceal every vestige
of the work of the Romans, although it is just possible
that careful enquiry might bring one or two subsidences
to light. If so, there would be a prima facie proba-
bility that these were due to the presence of the Ditch
underneath, and then the line could be laid down with
reasonable exactness. We have already noted some-
thing of the kind at Mount Pleasant, and more than
one similar instance will be recorded in the sequel. As
it is, recourse must be had to conjecture. One account
speaks of the Wall as passing behind the Parish
Church,1 and this is where one would expect it to run,.

1N.S.A. (Dumbartonshire), p. 187.

KIRKINTILLOCH TO DULLATUR 121

assuming that the Peel of Kirkintilloch stood on the
north, instead of the south, of the Vallum, as the
eighteenth-century authorities are almost unanimous in
affirming.1 In any event it begins to reappear a little
beyond the bridge that crosses the Canal some 170
yards east of the Railway Station, the indentation that
marks the Ditch being first detected in the back
gardens of the houses lying immediately to the left of
the high road. On emerging from the gardens it
sweeps towards the north, the curve being very similar
to that adopted long afterwards by the engineers of
the Canal.

The traces are not always easy to pick up, although
there is never any real doubt. Two convenient land-
marks may be noted. The Rampart passed through
what is now the steading of the farm of Cleddans ; 2 an
excellent section of the stone foundation was exposed
there in the autumn of igoq.3 And the depression of
the Ditch runs immediately in front of the farmhouse
of Whitehill. It may be added that, when the now
deserted factory between these points was erected,
considerable trouble was caused by a subsidence.4
Rather more than 200 yards beyond Whitehill, and
about 1400 yards from the bridge mentioned above,
the Limes again intersects the Canal. The ground
has been so much altered and made up on the further
bank, that its traces are there hopelessly obscured.
Further on, however, about 300 yards beyond what
must have been the point of intersection, the usual

1 Even Maitland has no doubts upon this point (Hist, of Scotland, i.
p. 178). Only Horsley hesitates a little (Brit. Rom. p. 169).

2 On the name see supra, p. 114, footnote.

3 Seen by Mr. John M'Intosh and myself.

4 Information from Mr. John M'IntOih.

•depression commences to show itself in the field to the
north of the road. It can be followed without much
hesitation all the way to the fort of Auchendavy, which
lies 500 yards ahead, the dips in the hedge-rows being
useful indicators.

At Auchendavy the Rampart formed, as it did in
most cases, the northern defence of the fort. It is now
levelled with the ground, but the depression of the
Ditch remains distinct and continues so for more than
half a mile eastwards. For the greater part of the
way it lies in the field, about 30 or 40 yards to the left
of the road. Less than 200 yards from the farm of
Wester Shirva, however, a bend in the road reduces
the gap very considerably, while a second bend, taken
immediately beyond the farm to avoid a steep gradient,
brings the road to the north side of the Ditch, which
then becomes very plain on the slope to the right ; a
good view of it can be got by looking back from the
bridge over the Shirva Burn. For the next 450 yards
there is little or nothing to reward the most diligent
scrutiny. Indeed, for nearly a mile the Ordnance
Map shows no definite traces. At the same time it is
certain that, after leaving the Shirva Burn, the modern
road very nearly represents the line of the Limes.

Though not marked on the Ordnance Map, the dip
of the Ditch is well seen as it emerges from the Canal,
on the east bank, directly in front of the little cottage
opposite Shirva farm. From this point the road along
the south side of the waterway is the best guide as to
direction. If it be followed closely, possible signs of
the Ditch can occasionally be discovered. The mineral
railway which presently appears on the right hand
side is partly laid along the track of the Rampart, the

KIRKINTILLOCH TO DULLATUR 123

stone foundation of which was more than once exposed
during its construction.1 Passing under the cottages
which stand on the right just before the village of
Twechar is reached, the line of the Ditch crosses the
public road that comes down from the south, and then
begins to climb towards the fort on the Bar Hill.
Unlike any of the other forts whose position is known,
the station on the Bar Hill did not abut directly on
the Rampart. It lay clear to the south, the Military
Way passing between it and the great barrier. The
Ditch, it may be noted, almost immediately on
crossing the road at Twechar, ceases to be a mere
depression and is quite a pronounced hollow.

The next section presents some features of special
interest. A little beyond the fort the ground becomes
rocky to an extent that must have added seriously to
the labour of the legionaries. Nevertheless the Ditch
pursues its course unswervingly. Excavation has
shown that the cutting is neat and clean, even where
the workmen have had to hew nine feet deep into the
solid rock.2 Just below the summit of the Castle Hill,
a conspicuous height some 200 yards east of the fort,
there is a curious interruption, ten or twelve feet of
rock having been allowed to remain untouched. There
is no apparent reason for this or for a similar, but
much more considerable, break which will be met with
later on — except perhaps the obvious and extreme
difficulty of dealing with an impervious surface of hard
stone. The descent of the eastern side of the Castle

1 Information from Mr. John M'Intosh, who watched the operations
closely.

2 Information from Mr. John M'Intosh, who had a portion of it cleared
out. At a good many points on the English Limes the ditch either of the
Wall or of the so-called ' Vallum ' has been cut through solid rock.

I24 LINE OF THE RAMPART AND DITCH

Hill is rapid, the line keeping close to the northern
boundary of the wide expanse known as the Castlehill
Park ; for some distance a stone dyke runs in the
centre of the Ditch.1 The remains of the Military Way
are very noticeable here. Looking back from the
lower end of the Park, one can see it distinctly marked
upon the grass. It passes through the northern por-
tion of the little wooded enclosure in the centre, and
then, after climbing some distance, divides into two
branches, one of which leads to the east gate of the
Bar Hill fort, while the other runs between the fort
and the Rampart. On the side of the hill the Rampart
is almost completely levelled ; at one point the stone
covers of a culvert peep up through the grass.2 But
just opposite the gamekeeper's cottage there is an
unusually well-preserved portion, with the Ditch in
front of it — originally 12 feet deep and 38 broad.3
Beyond the cottage the line becomes identical with
that of the farm road, and the usual indications can be
perceived at intervals, generally on the left ; the Mili-
tary Way is easily found in the field, and afterwards
in the wood, to the right. Caution is, however, now
specially necessary, for confusion has been introduced
by the construction of a railway in connection with
an ironstone mine. Mine and railway have long been
deserted, but the embankment still remains.

Less than a mile from the top of the Bar Hill the

dyke forms the march between the estates of Gartshore and
Auchenvole so far as the latter extends ; when the land was divided, the
Limes has been chosen as the boundary.

2 On Plate V. I this culvert is shown as it appeared when exposed
during the Glasgow Committee's excavations (Antonine Wall Report^
p. 90).

3 Antonine Wall Report, pp. 88 f. and Plate II.

KIRKINTILLOCH TO DULLATUR 125

public road from Croy to Kilsyth is crossed. During
all the intervening distance the hollow of the Ditch
and the tracks of the Rampart and the Military Way
have been conspicuous in front, running up the western
slope of Croy Hill. Even in the flat ground imme-
diately to the east of the road the remains of the Ditch
are quite noticeable. After the ascent has fairly
begun, they speedily become very impressive, for the
northern face of the hill, a little distance below the
highest part of which the cutting has been made, is
sometimes almost precipitous. There are places where
no artificial hollow was needed to strengthen the wide
gully that nature had already provided. Yet the rock
has none the less been hewn out with unrelenting
persistence ; the Ordnance Survey sections record
various depths, the maximum being nearly 15 feet,1
while the breadth ranges from 20 to 30 or more.
Two of the semicircular platforms projecting on the
south of the Rampart — ' expansions,' as they are
called in the Glasgow Report — occur on the western
shoulder of Croy Hill. They will be discussed in a
subsequent chapter. The more easterly of these is not
far from the top, which is an excellent point of view ;
the Ditch is very visible, alike towards the east and
towards the west, while the hills that rise on the north
of the low ground of the isthmus are more than
usually striking.

The descent eastwards soon leads to a flat shelf-like
expanse, lying close to the summit of the hill but on

1 Measured, that is, perpendicularly from the level of the scarp to the
bottom of the Ditch. As there is a considerable slope northwards, the
corresponding measurement from the level of the counterscarp would be
a good deal less.

126 LINE OF THE RAMPART AND DITCH

its southern side. It is more than probable that this
was once the site of a fort. A shepherd's cottage
stands near its edge, and further north are the ruins
of a tiny hamlet, once known as Croy Hill Houses.
Not far to the east of these is a break in the Ditch,
similar to that already noticed on the Castle Hill, but
considerably more extensive. For more than 50 feet
the hard smooth surface of the intrusive mass has been
left entirely undisturbed. Gordon's description could
hardly be improved upon : " At the Croe-hill, there is
a great Piece of a Rock rises out of the Ditch of the
Vallum, and serves, as it were, for a Bridge to pass
from the one Side to the other."1 On the eastern
slope of Croy Hill the Rampart has almost completely
disappeared, but the Ditch continues in fine condition
nearly to the foot. The Ordnance Survey sections
indicate that it is often about 8 feet deep, with a
breadth of 40 or more. From either side of it there
project here and there great blocks of stone that have
been laid bare in the process of cutting. Some huge
boulders have been left in the bottom, evidently
because it did not seem worth while undertaking the
immense toil of removing them. About 800 yards
beyond Croy Hill Houses the line quits the open
hillside, and enters cultivated ground again. From
here to Easter Dullatur, 700 or 800 yards away,
the direction is nearly straight and the depression of
the Ditch quite distinguishable. Burnside Terrace,
as the houses immediately north of Dullatur Railway
Station are called, lies just over the course of the
Rampart.

1 /tin. Sept. p. 56.

CROY HILL AND DULLATUR 127

4. FROM DULLATUR TO BONNYBRIDGE

On both sidesof Easter Dullatur,but more particularly
beyond it, the hand of man has wrought great changes
since the Romans drew their Limes. The laying out
of the grounds, the making of the public road, the
diversion of the course of the burn and, above all, the
construction of the railway have been responsible for
these. The alterations have misled the Ordnance
Survey officers. There is no such confusion at this
point as their map suggests. On the contrary, the
line would appear to have run on perfectly straight,
passing a short distance in front of the house and then
under the outbuildings into the field on the north of
the high road.1 Presently it crosses the road, only to
be buried under the railway embankment for at least
1 50 yards. Shaking itself clear of the railway just at
the bridge, it shows itself very distinctly on the other
side at the commencement of the ascent to Westerwood
fort, which lies right ahead about half a mile away.
During the earlier part of the climb the hollow presents
a very remarkable spectacle, being sometimes as much
as 10 feet deep. Even when it becomes less con-
spicuous, it is easily followed by keeping the line of
the road that leads past the front of the farm build-
ings of Westerwood.

At Westerwood the farmhouse and farm-steading
lie just within the north-east angle of the Roman,
station. For nearly two miles further the Ditch

1 Trenching is no doubt desirable here as elsewhere. At the same
time a careful examination of the ground, first with the help of Mr. J.
M'Intosh and afterwards with that of Mr. A. O. Curie, has left little-
doubt in my own mind.

128 LINE OF THE RAMPART AND DITCH

continues very plain, the line being as a rule close
to the northern boundary of the fields. The depth
is often 6 or 7 feet, the breadth sometimes nearly 50.
About 300 yards beyond the farmhouse the road bends
to the right to avoid the little homestead of Arniebog,
in front of which it passes after resuming its eastward
course. The walker is now on the track of the Military
Way, which here lies under the modern road for a
stretch of fully 500 yards. When the two part com-
pany, the slight elevation that betrays the presence
of the former can be seen (provided the conditions of
the crop be favourable) running straight on through
the fields, while the latter swings round to the left,
crosses the line of the Limes, and, after running
parallel to it for about 180 yards, disappears down
the hill to the north. For nearly three-quarters of a
mile from this point the best path to follow is the
bottom of the Ditch itself, unless recent rain has made
it too muddy. Then a road is reached which lies on
the north bank all the way to the railway line, a dis-
tance of 500 or 600 yards. The hollow is excellently
seen here, particularly near the farm of Garnhall.

The railway introduces an element of uncertainty.
The Ordnance Map shows the Ditch emerging on the
north side of the line a little to the west of where it
had sunk under it on the south, and then curving con-
siderably so as to include all the ground on which the
Castlecary brick- works now stand. This can hardly be
right. Such a deviation would be contrary to natural
expectation ; nor is there anything to suggest it in
surveys, such as Roy's, which were made before the
railway was in existence. As a matter of fact, the
Ditch seems to have gone straight on. Part of it

DULLATUR TO BONNYBRIDGE 129

has been cleared away in the formation of the railway
cutting, and another part of it probably lies buried
beneath the brickfield, for there can hardly be a doubt
as to its having crossed the north platform of the
Station about 50 yards west of the Station buildings.
The exact spot is marked by a subsidence in the
boundary wall. At some point which cannot now be
determined, it must have turned at a very sharp angle
to the right. When it was last distinctly seen, it was
running nearly due north-east ; when it is next dis-
covered, it is running south-east by east, creeping
downwards along the face of the high bank that leads
to the Red Burn. Instead, however, of pursuing its
course straight to the ruined bridge, as the Ordnance
Map would indicate, it takes another sharp turn, this
time to the left, at a point about 70 yards before the
bridge is reached, and then, descending 50 yards to
the bank of the stream, crosses and heads direct for
the north front of Castlecary fort. On both banks
of the Red Burn, but more especially on the eastern one,
the indications are exceedingly obscure. The ruined
bridge, to which reference has been made, marks the
line of the old road, and this in its turn may indicate
the course of the Military Way.

The remains of the Roman station of Castlecary lie
within the angle formed by the modern road that runs
northwards towards Denny and the branch of it that
diverges eastwards towards Bonnybridge. As soon
as the latter is clear of the fort, it bends to the
right a little, as if for the express purpose of avoiding
the Limes, and passes in front of the new school, which
is built very nearly on the line of the Rampart. For
about a mile Ditch and modern road are almost parallel,

130 LINE OF THE RAMPART AND DITCH

the interval between them being never more than
50 yards, and usually a good deal less. The Ditch,
which is very visible all the way, particularly behind
a row of workmen's houses called Allandale Cottages,
lies on the north (Plate X. i). Beyond it is the Canal,
at first about 300 yards away but gradually drawing
nearer. At the end of the mile, just opposite Under-
wood House, the three come close together, so close
that the Canal presently encroaches on the Ditch.
Between 150 and 200 yards east of the Skipperton
Burn, the line leaves the bed of the Canal and crosses
to the right-hand side of the road. The hollow is not
to be mistaken as it climbs through the field and enters
Seabegs Wood. Within the wood, for about 450 yards,
Rampart, Ditch, and Outer Mound are all admirably
preserved (Plate X. 2), nor does it require much trouble
to discover the Military Way among the trees and
bushes in the rear.1 In one of the Ordnance Survey
sections the Ditch is about 9 feet deep and 44 feet
broad. Some 250 yards after quitting Seabegs Wood,
the depression of the Ditch passes immediately to the
north of the farmhouse of Seabegs, crosses a road
coming from the south, and makes for the western
extremity of a long belt of trees. For a quarter of a
mile or so it keeps close to the trees and to the hedge
that connects them. On reaching the railway, near
Bonnybridge Canal Station, it is momentarily lost
sight of, but it can be picked up again without diffi-
culty on the other side, crossing a piece of waste
ground behind a group of cottages.

Roy and others are probably right in holding that

1 Curiously enough, Gordon failed to find it (Itin. Sept. p. 57). Probably
he did not look quite far enough to the south.

PLATE X

I. EAST OF CASTLECARY

2. NKAR SKABKGS

REMAINS OF THE LIMES

DULLATUR TO BONNYBRIDGE 131

there was originally a fort somewhere between the
western end of Seabegs Wood and the spot at which
we have now arrived. But no vestige of anything of
the kind remains, for the curious artificial mound which
is here visible to the north of the Ditch, and which we
shall have occasion to describe in a future chapter, is
obviously an independent structure. Some sixty or
seventy yards beyond it the track of the Ditch, which
has grown much fainter, takes a considerable turn to
the right, loses itself in the bed of the old mill-dam,
and then reappears, still barely discernible, making its
way up the hill towards the public road. To the east
of this road it is more distinctly observable, although
the natural irregularity of the ground over which it
here passes renders its exact course a little uncertain.
Presently the Scottish Central (Caledonian) Railway
intervenes. All the way across the field to the west
of it the dip of the Ditch can be seen outlined against
the sky on the hillside immediately beyond the embank-
ment,— an accidental section well fitted to show how
deeply the Romans have left their mark upon the
surface (Plate XI. i). The descent from the top of
the hill is fairly rapid, the direction being due east
for about 200 yards beyond the railway. Bending
towards the north, the depression of the Ditch then
slopes very gently upwards for 300 yards, passing the
foot of the Elf Hill, and finally reaching the road
which skirts the side of what used to be Bonnyside
Plantation.1 In its course through the field it has
just traversed, the Ditch seems here and there to be
accompanied by the Rampart, although many of the

1 Since the beginning of the present year (1910) the trees have been
cut down.

132 LINE OF THE RAMPART AND DITCH

mounds and hillocks that lie beside it are so irregularly
placed that they must be natural. The gaps which are
so apparent, and one or other of which local tradition
is prone to ascribe to the efforts of the valiant Graeme,
are probably the result of the action of water ; in spite
of the formation of the reservoir above, the field at the
bottom is still wet and marshy.

5. FROM BONNYBRIDGE TO POLMONT CHURCH.

For the next two miles the Limes runs through
ground that has never been under cultivation. Its
state of preservation is consequently good, the Ditch
being remarkably perfect,1 while Rampart and Outer
Mound sometimes rise to a considerable height. An
excellent example of a semi-circular ' expansion' is to
be seen close to the boundary wall directly after the
line enters the enclosure of Bonnyside. Some 340
yards further on is a second one, in much less good
condition. A little over a quarter of a mile more
brings one to the Rowan Tree Burn. Immediately
beyond the deep glen through which it flows are
the ruins of the fort of Rough Castle. This, and
the fine stretch of the Limes lying to the east of
it, within the Tentfield Plantation, can most con-
veniently be visited in spring before the brackens
begin to shoot up. About 350 yards after leaving
the precincts of Rough Castle, and close to some
disused mineral pits, Rampart and Ditch swing round
to the right and proceed in a south-easterly direction
for more than 500 yards, when the southerly trend

'The depth of the Ditch is frequently as much as 8 feet ; in one of the
Ordnance Survey sections it is nearly 10. See Plate VII. 2, and Plate
XL 2.

PLATE XI

I. CUT BY THK CAI.KDOMAN RAILWAY

2. SCAKP OF THK DITCH, NEAR HONNYSIDK

REMAINS OF THE LIMES

BONNYBRIDGE TO POLMONT CHURCH 133

is abandoned. All the way through the Tentfield
Plantation the remains are very striking, the Ditch
being often 8 or 10 feet deep. What seem to be the
remains of a semi-circular ' expansion ' may be noted
100 yards beyond the mineral railway which has to be
crossed some time before this last bend is reached,
while 600 yards further on there is a well-preserved
example. The Limes is now once more close to a public
road. Before long one comes to a house (Tayavalla)
which is actually astride of it, and from the garden of
which one can obtain a striking view up and down the
Ditch. A couple of hundred yards beyond the house
the road turns to the left, runs down to the low ground,
and then pursues its way eastward on the north side of
the Limes. The villa (Watling Lodge) and stables on
the hillside to the right, about 200 yards beyond the
point of intersection, are also built upon the Roman
line. Their position should be remarked, for a small
fort once stood here, guarding the exit of the road that
led to the fort of Camelon, three-quarters of a mile
away. For 400 yards more the Ditch continues very
plain — at one place it is 12 feet deep — and then comes
a hopeless interruption in the shape of the basins of
the Canal and their attendant railways.

On the east side of the second basin the depression
reappears almost at once. It runs through the northern
part of the grounds of Glenfuir House, ascending first
and then descending again towards a large wooden
gate in the west wall of the grounds of Bantaskine.
On passing into the policies it climbs slowly for about
140 yards, and then runs for some distance nearly on
a level, the Rampart having occupied the crest of a
long ridge which slopes steeply towards the north. For

134 LINE OF THE RAMPART AND DITCH

the first 230 or 240 yards of this level stretch the counter-
scarp of the Ditch has been cut down to form a terrace
with a gravelled walk. The end of the walk is close
to the mansion-house which lies immediately to the
south of the line of the Limes. A little to the east of
it the hollow of the Ditch has been cleared out for
some 50 yards at a comparatively recent period. As
seen now, the counterscarp here possibly retains its
original form. On the other hand, the scarp has been
deliberately terraced during the process of clearing ;
and, deep as the whole looks, the bottom has not quite
been reached. Then follows a short section where
practically nothing is distinguishable, owing to the
alterations necessitated by the formation of the avenue,
then 100 yards or so during which the conditions are
rather more favourable, — a slight depression being
visible to the south of the avenue, — and finally the
east wall of the policies. About 60 yards south of the
main gate of Bantaskine, the line crosses the road
known as ' Maggie Wood's Loan,' and enters the
precincts of the town of Falkirk.

For the next three-quarters of a mile the remains
are difficult, and ultimately impossible, to trace. On
quitting Bantaskine grounds, the vestiges of the Ditch
can be detected at intervals in the gardens of the villas
on the south side of the road that runs along the top of
Arnot Hill — notably at Mayfield Cottage. The whole
face of the district in which we now find ourselves
has been changed by building since the date of the
Ordnance Map, and this renders accurate description
peculiarly difficult. Probably the conjectural line laid
down by the Survey Officers is approximately correct.
At one or two points, however, we may venture to be

BONNYBRIDGE TO POLMONT CHURCH 135

more definite. About 400 yards from its western end
Arnothill Lane takes a sudden bend to the right, and
then resumes its eastern course almost immediately.
In the interval it has passed from the northern to the
southern side of the Roman frontier. The north gable
of the villa which presently appears on the left is
actually founded above the Ditch, as the builder dis-
covered to his cost in the course of its erection.1 We
have now fairly begun to descend into the valley
through which the waters of the West Burn flow
northwards. To-day the burn is largely covered over,
so that it is no matter for surprise that the Roman
Ditch has vanished on both sides of the point of inter-
section. Nor is there anything more to be seen upon
the surface until the town is left behind again. At the
same time it is not yet too late to secure an accurate
plan, if only it were possible to resort to occasional
trenching. The formation of the ground is such as to
leave no room for doubt regarding the general direction ;
and not a little of the space is occupied by gardens.

Besides, there are fingerposts. Maitland, writing
soon after 1750, says: "In the southern suburb of
Falkirk, called the Pleasants, a few years since was
discovered the military way ; as it likewise was in the
arable grounds bewest the same. The causeway lies
on the southern side of the Pleasants, at a considerable
distance from the ruins in the gardens on the northern
side of the suburb, which, by their appearance, seem to
have been parts of the wall and ditch ; and the inter-
jacent space, which is now occupied by a street, court,
and houses, I take to have been the site of a station
or fort on the wall."2 It is likely enough that Maitland

1 Information from Mr. M. Buchanan. 2 Hist, of Scot. i. p. 173.

136 LINE OF THE RAMPART AND DITCH

may have been right as to the fort. Leaving that,
however, for the present, we come next to the estate
of Rosehall, within which a portion of the stone base
of the Rampart was uncovered a few years ago.1
Lastly, and this indication is particularly valuable for
its precision, one of the houses on the south side of
Booth Place, immediately to east of Rosehall, actually
rests on a stone arch thrown over the Ditch to obtain
a solid foundation.1 It follows that the conjectural
marking on the Ordnance Map is here 20 or 30 yards
too far to the south — a surprisingly small error, when
account is taken of all the conditions.

From the point just mentioned obliteration is com-
plete as far as the East Burn, although in Maitland's
time traces of the Ditch began to be apparent 100
yards to the west of the streamlet.2 Even beyond the
burn there is nothing visible at first.3 Firm ground
is, however, reached almost as soon as the western
boundary of Callendar Park is crossed. The precise
spot at which this crossing takes place is not to be
determined without excavation. According to Mait-
land4 it was a little distance to the north, according to
others a little distance to the south, of the line that the
Limes pursues in its passage through the grounds of
Callendar House. Present appearances lend some
colour to the latter opinion. On the other hand,
these appearances are by no means conclusive, and
Maitland's view certainly fits in best with the hint of

1 Information from Mr. M. Buchanan.

2 Maitland, History of Scot. i. p. 173.

3 According to Waldie ( Walks along the Roman Wall, 1883, p. 25), a
stretch of ground to the east of the East Burn was called " the cleedins.'"
The name is significant, see supra, p. 114, footnote.

4 Maitland, History of Scot. i. p. 173.

BONNYBRIDGE TO POLMONT CHURCH 137

direction which was recorded at Booth Place.1 What-
ever may be the ultimate solution of the question, the
remains are unmistakable almost from end to end of
the beautiful enclosure that forms the Park, a distance
of some 1500 yards in all, the only serious interrup-
tion being just opposite the garden. For fully a third
of the way the Ditch is admirably preserved ; at one
point it is more than 12 feet deep.

Immediately beyond the east end of the Park, the
Polmont and Falkirk Railway marks the beginning of
another break, which extends for at least 600 yards.
But there is no real difficulty in ascertaining the line
that the Roman engineers must have followed. A
depression in the wall on the left-hand side of the
main road from Falkirk, about 100 yards east of
the railway bridge, probably marks the position of the
Ditch. It thus seems to have taken a northerly turn
on leaving Callendar Park, and this agrees precisely
with the description given by Maitland, whose account
of the course of the Wall in and about Falkirk is much
more detailed than that furnished by any other of the
older writers.2 The normal direction must have been

1 The conjectural line on the Ordnance Survey Map shows no deviation
whatever at the west end of Callendar Park. But it was already 20 or
30 yards too far south at Booth Place. Maitland's bend, therefore,
supplies exactly the amount of correction that is required to reconcile
the evidence from Booth Place with the line actually visible within
Callendar Park.

2 Approaching from the east, he tells us (Hist, of Scotland, \. pp. 172 f.)
that the Wall, "descending the Gallowhill, crosseth the Gallow-sike or
rill, and, mounting the opposite eminence, deflects twenty-four feet south-
wards ; then entering the eastern end of Callender-park, recovers its
western course." This incidentally gives us Maitland's estimate of the
bend at the other end of the Park, for he expressly says (Ibid. p. 173)
that that bend enabled the Wall " to recover its rectilineal state." Irvine,
quoted by Sibbald (Hist. Inq. p. 30), also speaks of the bend at " the
Gallow-Syke."

138 LINE OF THE RAMPART AND DITCH

resumed again very quickly, for there is every reason
to believe that the more northerly of the two branches
into which the main road now divides, lies on the very
line of the Roman barrier. It is known as Graham's
Dyke Street, and forms part of the village of Laurie-
ston. Soon after the houses cease, the Ordnance Map
shows traces of what may be the Outer Mound close
to the left-hand side of the road. These continue for
fully 500 yards, and they would seem to indicate that
the modern highway has been laid along the filled-up
Ditch. The change must have taken place almost
within living memory. The author of the description
of the Parish of Falkirk in the New Statistical
Account of Scotland, writing in 1845, speaks of the
remains of the Ditch as being " particularly conspicuous
... at Laurieston."1 This is quite in harmony with
the "very visible" of Gordon2 and the ''clearly
discernible " of Horsley.3 To-day there is but little
to be seen, and that little is of a rather indefinite
character.

The low ground of the isthmus has now broadened
out into a plain not less than two miles wide, and a
view from the brink of the steep slope a little to the
north of the farmhouse of Mumrills conveys a good
idea of the commanding position of the Limes. Mum-
rills is believed to have been the site of a fort, and
we shall be able to show good reason for accepting
this opinion. The fort did not, however, lie in the
immediate neighbourhood of the farm-buildings, as it
is generally supposed to have done. As a matter of

1 Stirlingshire, p. n. This is fully confirmed by what Mr. M. Buchanan
tells me he has heard from those now dead.

2 Itin. Sept. p. 60. 3 Brit. Rom. p. 172.

BONNYBRIDGE TO POLMONT CHURCH 139

fact, these are beyond the Roman frontier, the line of
which is beginning to trend towards the south of the
public road, although here, and for fully half a mile
beyond, it is meanwhile impossible to lay down its
course with accuracy. In all probability it passed
through the field directly opposite Mumrills farm-
house at such a distance as would leave sufficient
room for the fort between it and the southern edge
of the high plateau. This would carry it along the
southern face of the great hollow between the Mum-
rills Braes. Its natural direction would then be south-
east by east until the Westquarter Burn was crossed,
when it would turn slightly and proceed almost due
east across the fields of Beancross farm, the turning
point being somewhere within the newer portion of
the Grandsable Cemetery. In a word, the conjectural
markings on the Ordnance Map are in all likelihood
very near the truth. Maitland's line is the same.
Gordon and Horsley speak with a more uncertain
tone. According to the former, "the Wall has cer-
tainly passed by the Village of Bencross? He adds :
" but hereabouts are no visible Marks of its Track,
and is only known to have gone this Way from
the Foundation of the Stones which are dug up in
Ploughing."1 Horsley agrees that it "passes" Bean-
cross, and describes the Ditch as being " very grand
at about five chains west from the village."2 Maitland
is characteristically caustic about the blindness of his
predecessors in "their leading it so far out of its way,
round by the village of Bencross."3 He may, however,
be pressing the sense of the word ' pass ' too closely ;

1 Itin. Sept. p. 60. 2 Brit Rom, p. 172.

3 Hist, of Scotland, \. p. 172.

140 LINE OF THE RAMPART AND DITCH

Horsley's language about the Ditch is more appro-
priate to the natural gap between the Mumrills Braes
than to anything else of which there is the faintest
trace left to-day, so that, after all, the divergence of
opinion is perhaps less complete than at first sight
appears. It is worth adding that the early oblitera-
tion of the Ditch in this neighbourhood is doubtless
due to the sandy nature of the soil, just as its better
preservation towards the centre of the isthmus, near
Castlecary for example, is to be attributed to the fact
that it has been cut through heavy clay.

About 900 yards beyond Mumrills the Limes must
have crossed the Polmont Burn. In the field immedi-
ately to the east faint signs of it can perhaps be
detected 120 or 130 yards to the south of the road.
At all events, from this point the depression can be
seen very distinctly in front as it mounts the slope of
the Cadger Brae, at the top of which it enters the
grounds of Polmont Park. Within very recent years
stones from the base of the Rampart have been
ploughed up in the field at the foot of the ascent.1
Inside the park it shows itself here and there, half
dubiously. But it does not become at all clear until
it approaches the farm-steading to the north of the
churchyard wall. There the indications grow quite
decided. They can be followed for more than 100
yards through the fruit-gardens on the further side
of the road that leads past the church. In the open
field they vanish.

4 Information given to myself on the spot. Mr. M. Buchanan also tells-
me that the Military Way was cut through when the foot-road, which,
leads out from the house beyond the Cadger Brae, was formed.

POLMONT 141

6. FROM POLMONT CHURCH TO BRIDGENESS.

At this point the walker had better go back to
the church and return to the main road. If he keeps
to that for 300 yards or so, he will find on his right
hand a small bridge and a footpath. After running
for about 100 yards round the base of a wooded hill
beside the Millhall Burn, the path will bring him to
a wide opening on the left, up which the Limes
undoubtedly went. At no time since leaving Cal-
lendar Park have the remains been in such good
condition. As they pass through the wood, they bend
a little towards the south in ascending the shoulder of
the hill, Rampart, Ditch, and Outer Mound all being
visible (Plate XIII. i). The easterly trend is quickly
resumed, and for more than half a mile the track
continues very apparent, except for a brief space
where it is interrupted by a modern reservoir. Its
course is at first due east and subsequently north-east
by east ; and the hog-backed range of heights, along
the upper part of which it passes, offers a splendid
outlook over the Forth, now drawing very near.
During the greater portion of the way only the de-
pression of the Ditch can be seen. But the ploughmen
on Polmonthill farm are quite familiar with the stone
foundation of the Rampart.1 About 200 yards beyond
Polmonthill the depression drops suddenly to the level
of the River Avon. A view of it from the bank of
the stream enables one to appreciate the aptness of
Dr. John Buchanan's description : "it has the appear-
ance of an immense slice cut out of the breast of the
firae, with well-preserved edges."2

1 Information given to myself on the spot.

2 Stuart's Caled. Rom. 2nd ed. p. 360, footnote.

142 LINE OF THE RAMPART AND DITCH

On the further side of the river another blank is
encountered. The soil is rich here, and the field has
been long under cultivation. Only at its upper end,
immediately to the north of a small cottage, is a certain
irregularity in the surface which the Survey Officers are
probably justified in identifying as an indication of the
Ditch. There is a general consensus of opinion that
one of the Limes forts stood in or near this spot.
Nothing, however, remains to show whether it was
situated in the field of which we have been speaking,
as Roy apparently thought likely,1 or whether, as
others have supposed, it lay on the higher ground
now occupied by the farmhouse and steading of Inver-
avon. In any event a close examination of the
ground suggests the gravest doubts as to whether
the Ordnance Map is to be trusted here. The officers
appear to have been led on a false scent along the
steep wooded face of the plateau towards the left. It
is much more likely that the line passed through the
present farm-buildings2 and then traversed the fields
beyond them, making across the high ground straight
for the bridge over the North British railway. Up
till 1842 "a large portion of the Roman causeway"
through this farm was still used by horses and carts
" in its original state." It was then so greatly worn
as to resemble "a harrow," and was therefore uprooted
by the farmer as dangerous for his horses. It is said to
have been " formed with large stones in the centre,

1 He suggests that the remains of the fort have been " washed away by
the river" (Military Antiquities, p. 162). This could not possibly have
happened if it stood upon the higher ground, even although the river may
have changed its course since Roman times.

2 This agrees with what is said by Gordon (Itin. Sept. p. 60), Horsley
(Brit. Rom. p. 173), and Roy (Milit. Antiq. p. 162).

POLMONT CHURCH TO BRIDGENESS 143

and regular curb-stones along the edges"1 — a descrip-
tion which is not applicable to the Military Way at
all,2 but which exactly fits the stone foundation of the
Vallum.

The question at issue cannot be finally settled with-
out digging. In the meantime it will be sufficient to
suggest that, when the opportunity for excavation does
arise, the Ditch might be looked for, in the first
instance, 30 or 40 yards to the left of the road that
runs past Inveravon on the south. Half a mile or
more beyond the farm-steading this road is carried
across the railway by the bridge already mentioned.
The White Bridge, as it is called, is directly in the
track of the Roman Ditch. It is true that there is
nothing very visible on the steep incline behind. But
in front the hollow is plainly discernible close to the
right of the road. It can be traced all the way to the
top of the hill and then, with less confidence, down
the slope on the other side to the foot of a second field.
Thenceforward all is doubtful. Indeed, from the hilltop
we have just left, the markings on the Ordnance Map
cease entirely. Nor do they reappear at all, except for
a very brief space — three or four miles ahead — above
the town of Bo'ness. The configuration of the ground
might suggest that, from the foot of the field mentioned
above, the line bent to the right and began to ascend
again, as if making for the middle of Kinneil Wood.
Enquiry has, however, elicited some interesting informa-
tion that points to a different conclusion.

Stuart's Caled. Rom. 2nd ed. p. 361, footnote. Dr. John Buchanan,
who is responsible for the note, saw the process of demolition when it
was in progress.

2 See supra, p. 103.

i44 LINE OF THE RAMPART AND DITCH

Mr. A. S. R. Learmonth, who was for many years
tenant of the farm of Nether Kinneil, states1 that in
the year 1861, when ploughing in the Easter Wellacres
field — the fourth field on the right hand side after
crossing the railway — he " came upon a causeway of
rough stones, varying in size from about one to two
feet, the larger stones being on the north or lower side
and the smaller ones on the south side. It was covered
by from 8 to 12 inches of soil. As the stones were
liable to break the agricultural implements, they were
removed." He adds that his uncles, who preceded
him on the farm, had previously removed many other
parts of the causeway in that same field and the other
two fields to the west. At the time Mr. Learmonth
was under the impression that it was the Roman road
which he had encountered. His recollection of the
breadth — about eighteen feet — suits well for the Mili-
tary Way. On the other hand, his description is
much more applicable to the stone foundation of the
Rampart. What he says as to position has a similar
bearing. It was "about 20 or 30 yards west of that
part of the road leading to Upper Kinneil known as
'the Stey Step,' and about the same distance south of
the road to Nether Kinneil." If the 'causeway' was
the base of the Rampart, then the Limes must have
run in a straight line from the top of the hill, instead
of swerving to the south. That it may very well have
done. It would be more difficult to account for a
sudden bend towards the north, such as it would be
necessary to assume if the stones had belonged to the
Military Way.

The difference, however, is not material. In any

1 Letters to the Rev. Robt. Gardner and myself.

PLATE XIII

I. IN THE WOOD ABOVE MILI.HALL BURN, LOOKING \V.

THE ROAD TO CAMELON CROSSING THE DITCH, WITH THE MAIDEN
CASTLE, AS IN 1893, LOOKING N.W.

POLMONT CHURCH TO BRIDGENESS 145

event, the hint as to direction is valuable, particularly
as it appears to be confirmed when we pass into the
field immediately to the east of ' the Stey Step.'
Entering the field by the gate opposite the end of
the road from Nether Kinneil and turning to the left,
one speedily meets with a slight hollow. This, Mr.
Learmonth believes, marks the line of the Roman
Ditch. When he first remembers it, the depth was
as much as 6 or 8 feet. But it was filled up by his
uncles as being very inconvenient to plough and cart
across. He himself helped to level a portion of it
about the year 1860. The Limes, then, would seem
to have swung slightly northwards here. It must have
turned southwards again very shortly. Sir Robert
Sibbald speaks of it as having passed through Kinneil
Wood,1 and such a course would be a natural one for
it to take. No positive vestiges can now be detected
either in the wood itself or to the east of the slight
hollow just described. Mr. Learmonth, however,
recalls that somewhere in the ' sixties ' he observed,
when a drain was being made, "a piece where the
sub-soil was quite different from the rest, and ending
abruptly at each end." This was in the same field
as the hollow — the Walk or Summerhouse Park, —
but considerably to the east of it and just opposite
the Pond. Though puzzled, Mr. Learmonth attached
no particular significance to these appearances at the
time. But he is now convinced that the drainers had
cut across the line of the Roman Ditch, and there is
every probability that he is correct. The Limes
would then be heading for the corner of Kinneil
Wood. Meagre as these particulars are, they are

1 Hist. Ing. p. 30.
K

146 LINE OF THE RAMPART AND DITCH

worth recording, in view of the little that is now visible
above ground between Inveravon and the Forth. The
early years of the eighteenth century appear to have
been responsible for much havoc along this section of
the Limes. Gordon and Horsley evidently saw little
more than it is possible to see to-day, while the
anonymous traveller of 1697 found the vestiges dis-
tinct, though "rather faint," the whole way from
Inveravon to Kinneil.1

The field beneath the wood is known as the
Meadows. It contains several conspicuous hollows,
which are sometimes spoken of as the remains of the
Roman Ditch.2 That the identification is erroneous
is clear from an examination of the account of
1697. "The Roman Wall which is here made of
stone and turff, and a ditch behind it" was then
visible " within a bow shoott " to the south of Kinneil
House.8 This is suitable only for a line that ran
through the lower half of the wood. And even to-day
confirmation of the traveller's statement is not alto-
gether lacking. At some time subsequent to his visit
the ground to the south of the mansion-house has been
considerably altered by the formation of two terraces,
the upper one of which is now covered with green
sward, while the lower is planted with trees that are
apparently about 100 years old. These improvements
must have swept away all remnants of the Rampart.
But 50 yards to the south of the garden gate, — that is,
very much in the position which the traveller indi-

1 Hist. MSS. Commission: Portland Paper -s, ii. p. 55.
2 They are so marked on the six inch O.S.M., but not on the twenty-
five inch issue.

3 Hist. MSS. Commission: Portland Papers, ii. p. 55.

POLMONT CHURCH TO BRIDGENESS 147

cates,1 — is a slight depression which can hardly be
anything else than the Ditch. It has a northerly
trend, and at a point 25 yards south of the modern
roadway it can be seen entering the field to the east,
where it continues to be traceable for some little dis-
tance. It is generally assumed that there was a fort
at or about Kinneil. If so, no evidence of its existence
has survived. It may, however, be noted that the
bank of the Gil Burn, near the entrance to Kinneil
House, is a likely enough position for a station.

There have been some, notably Maitland, who
would have the Limes end in the neighbourhood of
Kinneil. The majority, like Gordon, Horsley, and
Roy, would prolong it to Carriden. The chief argu-
ment in favour of Maitland's view is a philological
one : Kinneil or Cenail is identical with the Peanfahel
or Penneltun of Bede, and means "the end of the
Wall." 2 Although the etymology is sound enough, the

1The distance from the S. gable of Kinneil House to the line of the
depression is about 175 or 180 yards, and Dr. Allan Jamieson, who has
given much attention to the subject of archery, informs me that nine or
ten score yards may be taken as the conventional ' bowshot ' of the
seventeenth century.

3 Hist, of Scot. i. p. 171 and pp. 185 f. The passage from Bede is
cited in full, supra, p. 34, footnote. See also the gloss on the Historia
Brittonum, quoted supra, p. 36, footnote. Dr. W. J. Watson tells me
that there is no reason to doubt the correctness of the etymology given
above. Cenail, as the Nennius gloss has it, is simply the Gaelic form of
the Cymric Penguaul, meaning "the end of the Vallum." Peanfahel
represents a stage in the transition, fahel being Gaelic and the fh subse-
quently becoming silent in quite normal fashion. Penneltun, again, is
Pen(faA)eZ+A..S. tun. Cf. Zangemeister, Neue Heidelb. Jahrb., v. (1895),
pp. 96 f. Mr. Nicholson's ingenious attempt {Keltic Researches, pp. 21 ff.)
to prove that Penfahel is really a corruption of the Latin pinnae valli rests
mainly on the argument that the Vallum does not, as a matter of fact,
end at Kinneil. But the precise spot to which the name of Kinneil was
originally attached is quite unknown to us. As we shall see presently,

i48 LINE OF THE RAMPART AND DITCH

testimony of the place-name cannot be allowed to
outweigh the more substantial indication now available.
Two miles further on, at Bridgeness, there was dis-
covered in 1868 a fine slab (Plate XIV.) bearing an
inscription which showed that it had stood on the
Limes. Besides, we are assured from various sources
that in the seventeenth and eighteenth centuries, and
even in the early part of the nineteenth, distinct traces
of the remains could be observed at several points
above the town of Bo'ness. Thus, Sibbald "saw
some of the Foundation Stones of the Wall taken up
(for some Building)" about " the middle of the Ground
betwixt Bander stoun and Borrowstounness," l and
Gordon made out, " for a mile beyond Kinniel, a
faint Track of the Rampart."2 Horsley, who was
doubtful as to whether the Limes itself had gone so
far, thought that " what has been taken for remains
of the wall between Kinniel and Caer-ridden are
rather the remains of the military way which has
gone not only to Caer-ridden^ but probably to Cramond
and Edinburgh"* His main argument is that "the
ditch which is everywhere else the most visible part
of the work, and always appears where anything is
visible, does not appear at all in this space." Roy, on
the other hand, says : " To the eastward of the
inclosures of Kinneel a slight vestige of the ditch
may be perceived, and another on the south side of

the parish of Kinneil, until the seventeenth century, included the whole
of what is now Borrowstounness, and thus came very close to Bridgeness
— the most probable termination.

*Hist. Inq. p. 30. * Itin. Sept. p. 60.

*Brit. Rom. p. 159. Subsequently, however (p. 173), he is inclined
to admit the greater likelihood of the Wall having been continued as far
as Carriden.

POLMONT CHURCH TO BRIDGENESS 149

those of Grange."1 Stuart also avers that in his time
the Ditch could be seen " in a field immediately above
the House of Grange." 2

There remains a fact more significant than any of
these, except the finding of the Bridgeness stone. In
1649 an Act of the Scottish Parliament, following upon
a petition presented by the inhabitants and merchants
of Borrowstounness, erected the district round the
church of Borrowstounness into an independent parish
— it had previously formed part of Kinneil — and fixed
" Grahame's Dyk " as its southern boundary.3 Had
the parochial area remained unaltered, it would thus
have given us the exact line. As it is, twenty years
later, another Act annexed to Borrowstounness that
part of Kinneil from which it had been separated,
so that on its southern side it henceforth marched
with Linlithgow. Apparently, however, the southern
boundary of the original parish was identical with
what is now the municipal boundary. Rampart and
Ditch must, therefore, have run not very far from the
modern high road,4 which is just where we should
naturally look for them. One or two other points may
be mentioned. The large modern villa called Graham's
Dyke occupies the place of a group of old cottages
that bore the same name. It stands on the right-
hand side of the road. On the left-hand side is a row
of houses known as Riverview Terrace. It is said

1 Milit. Antiq. p. 162. 2 Caled. Rom. 2nd ed. p. 361.

3 " The saide estate of Parlement erects fe said Kirk of borrowstoun-
nesse now planted w' a minister And Separats and Divyds fe samene
from )>e Kirk of Kynneil in all tyme comeing q'of it was ance ane part
And ordaines and Declaires Grahame's Dyk to bound )>e samyn on the
south, etc." (Acts of the Scottish Parliament^ 1649, c. 206).

4 Known as Graham's Dyke Road.

150

that in digging the foundations of one of these the
workmen uncovered the stone base of the Rampart.1
But it may be noted that the ridge behind Riverview
Terrace would appear to be a natural formation2 and
not, as the Ordnance Survey Officers seem to indicate,
a remnant of the Wall.

As to the exact place where the Limes reached the
shores of the Forth, there has long been a difference
of opinion. The discovery of the inscribed slab men-
tioned above, on the spot now marked by a memorial
tablet, has gone far to settle the question. It is, of
course, not impossible that the stone may have been
moved from its original position for purposes of con-
cealment, and that the real end of the Limes may have
been at Carriden, as Gordon and many others after
him have asserted. But it is not easy to believe that
either Roman or Caledonian would have been at pains
to transport so huge a block for more than half a mile
before disposing of it. It is true that to lead the line
along the brow of the crescent-shaped hill would have
rounded it off more appropriately than could be done
by deflecting it along the spur to Bridgeness. Such a
consideration would naturally appeal to General Roy's
professional eye. The Roman engineers might, how-
ever, have been less fastidious.

A greater difficulty may seem to arise from the
scattered notices as to actual remains that have at one
time or another been visible beyond Bridgeness. Thus,
in the oldest detailed account of the Limes that we

1 Information through the Rev. Robt. Gardner. Precise details were
not, however, ascertainable.

2 This was pointed out to me by Mr. H. M. Cadell of Grange, who can
speak with authority on all that relates to the geology of the district.

POLMONT CHURCH TO BRIDGENESS 151

possess — the manuscript Description of Scotland by
Sir Robert Sibbald, now in the Advocates' Library, —
we read : "At Carin it may be traced yett as Mr.
Milne the Laird told me who hath some stones with
inscriptions and figures was taken up ther." This is
too vague to be reliable ; it may refer merely to ruins
and relics connected with the fort which has by general
agreement been located at Carriden. A century and
a half later Stuart tells us that "traces of the/0^"
could be distinguished " in a field immediately above
the house of Grange, near Borrowstouness, where the
intrench ment turned off to the south-east, just before
reaching Carriden."1 At first sight the statement
seems to throw a serious obstacle in the way of
accepting the testimony of the legionary tablet, for
the only Grange House now known in the neigh-
bourhood is situated more than 100 yards east of
Bridgeness. Properly interpreted, however, it has
precisely the opposite effect. What Stuart had in
view was the older House of Grange, built by Sir
John Hamilton in 1564 and demolished six years
ago. it lay some distance to the west of where the
slab was found, just below the point where the wall
would naturally have turned downwards if it was to
end at Bridgeness. And it did, as a matter of fact,
turn downwards there. We could gather that from
Stuart. But Horsley speaks more plainly, when
arguing against a continuance to Carriden : " the
remains, near the Grange house, make a turn, and
quit the most advantagious ground for a rampart."2
Combined with the evidence of the inscription, this
may be accepted as virtually conclusive.

1 Caled. Rom. 2nd ed. p. 361. 2 Brit. Rom. p. 159.

CHAPTER VI

THE FORTS : FROM THE CLYDE TO BAR HILL

WHEN a limes was constructed through territory that
was either actively or potentially hostile, a series of
protecting castella was its natural and inevitable
accompaniment That such a series once stretched
from the Clyde to the Forth, may be regarded as
unquestionable. Gordon found the remains of ten of
the forts sufficiently complete to enable him to lay
down plans, and in seven or eight of these ten cases
the outlines are still distinctly visible. To judge from
the average distance between those that we definitely
know to have been adjacent, the original number was
probably as large as nineteen or twenty. Excavation
on three of the sites has shown that the forts conformed
to the general type which we have seen to be con-
ventional, but that there were marked individual
differences in size, in internal arrangement, and even
in methods of defence.

Unlike the stations, 'per lineam valli] in the North
of England, the Scottish forts can no longer be identi-
fied by their Roman names, although it seems possible
that a corrupt tradition of these has after all survived.
The Cosmographia of the anonymous Ravenna Geo-
grapher, compiled about the seventh century A.D.,

THE NAMES OF THE FORTS 153

enumerates ten " cities" that lie in a straight line from
sea to sea in the narrowest part of Britain and are
connected by a road.1 The following are the names
he gives : Velunia, Volitanio, Pexa, Begesse, Colanica,
Medio Nemeton, Subdobiadon, Litana, Cibra, Credi-
gone. With the doubtful exception of Colanica, which
may be the KoAav/a of Ptolemy, none of these is known
to us from any other source. Nor does their number
correspond with the probable number of the Limes
forts. Still, the circumstances fully justify Horsley's
cautiously expressed opinion. " I am inclined to
think," he writes, "that the names relate to the
stations along the line of the Roman wall in Scotland.
But particularly to adjust or apply them is what I shall
not attempt."2 There it will be well to let the matter
rest. Hiibner endeavoured to go further.3 But the
degree of success he met with is not encouraging.
His arrangement is the merest guess-work. It rests
on the assumption that the enumeration of the Ravenna
Geographer begins at Rough Castle, and moves west-
wards in regular order ; and it compels him to ignore
at least one fort (New Kilpatrick) whose existence is
attested by irrefutable evidence. The late Mr. C. J.
Bates's effort was much more ingenious, but hardly
more convincing.4 Accepting, then, the modern
nomenclature as the only one that is feasible, we
may recapitulate briefly the main facts that have
been ascertained about each of the stations.

1 Iterum sunt civitates in ipsa Britannia recto tramite una alteri
conexae, ubi et ipsa Britannia plus angustissima de Oceano in Oceano
esse dinoscitur, id est Velunia, etc. (op. cit. v. 31).

"tBrit. Rom. p. 505. 3 C.I.L. vii. pp. 194 ff.

4 Arch. Ael. xix. pp. 1 10 ff.

154 THE FORTS

The position of the Chapel Hill has already been
described.1 None of our earlier authorities were able
to detect upon its surface any indications that could
reasonably be associated with a fort or its defences.
The revised Ordnance Survey Map, on the other hand,
shows markings that represent the indistinct remains
of ramparts or ditches. A close inspection of the spot
makes one hesitate before accepting so definite an
interpretation of the undoubted irregularities of the
ground. The vestiges are too vague to serve as a
foundation for any hypothesis. Excavation alone can
decide whether they furnish a clue to the outline of the
buried fort. But that digging would lead to the un-
covering of a fort in their immediate neighbourhood is
as certain as it is possible for anything of the kind
to be.

The reported finding of inscribed stones there con-
stituted a presumption. A much more positive index
was afforded by a discovery made in 1790 when the
canal to Bowling (which passes between the Chapel
Hill and the Clyde) was in process of construction.
This is briefly referred to by Stuart in his Caledonia
Romana? A more circumstantial account of it is
given by Mr. John Bruce in his History of the Parish
of West or Old Kilpatrick (1893). Citing as his
authority a contemporary manuscript by one John
Millar Morrison, Mr. Bruce tells us that " in the year
1 790, when the canal was being dug at the south end
of the Sufield Park, between Portpatrick and the
Ferrydyke drawbridge, a building constructed of free-
stone and lime was unearthed. In the inside were a
considerable number of partitions about two feet apart,

1 See supra, p. 1 10 f. 2 2nd edit. p. 294.

CHAPEL HILL AND DUNTOCHER 155

arched over with bricks about 9 inches long, and as
many broad, while on the top of the arch were placed
flat bricks about i £ inches thick, and of the same size.
Inside the building several rows of urns were found
about 2 feet deep and a foot and a half wide at the
mouth, and made of burnt clay, and in them a number
of silver coins were found covered over with earth. . . .
Mr. Davidson, then minister, and Sir Archibald
Edmonstone, who was staying at the manse, got a
number of the coins, as did also Mr. Colquhoun, super-
intendent of the canal." Morrison himself got one,
which afterwards found its way into the possession
of "the Scots' Antiquarian Society, London."1 The
story of the "rows of urns" as given here is not very
intelligible. But the description of a Roman hypocaust
is unmistakable. In all probability what the workmen
had hit upon was the Bath attached to the station.
The circumstance that it lay clear of what is generally
assumed to have been the site of the fort itself is an
additional confirmation of the truth of the whole story.
That is exactly where one would expect the Bath to
have been found.2

The remains of the fort of Duntocher are still easily
distinguishable on the summit of the Golden Hill,3
rather more than two miles distant from the probable
site of the first of the stations. The interval is thus a
normal one, for we have already noted that, in the
case of this particular Limes, the space between fort

1 Op. cit. p. 36. 2 See supra, p. 81.

3 Their degree of visibility depends to some extent upon the condition of
the grass. A good light, too, is required. Professor Haverfield suggests
to me that the name of the hill may have been given to it through gold
coins having at some time or other been discovered there ; there is a
'Silver Hill' near Silchester.

156 THE FORTS

and fort appears to have been, roughly, two miles.
The north-west face of the hill rises abruptly from the
bank of the Dalmuir Burn. The actual top, however,
consists of a very gentle slope, presenting a pleasant
exposure to the south and commanding a wide view
in every direction. The Rampart and Ditch of the
Limes served to protect the fort on the north-east.
The remaining three sides had been defended by ram-
parts of their own. Whether they had had more than
a single ditch, it is hardly possible to say. But it is
clear that the whole station had formed an oblong,
whose major axis lay parallel to the great Rampart
and had an interior measurement of between 300 and
400 feet.1 More precise dimensions cannot be stated
in the absence of accurate knowledge as to the extent
of the defences. In the earlier half of the eighteenth
century the track of the Military Way seems to have
been plainly visible ; Gordon,2 Horsley, and Roy agree
as to its position. Even yet it is perhaps faintly dis-
cernible— not forming, as it usually does, the Via
Principalis of the fort, but passing a little distance to
the south and throwing off a short branch which enters
the Porta Decumana.3 This exception to the general
rule is not difficult to explain ; the following of the
ordinary course would have entailed an intolerably
stiff ascent.

Apart from a gold coin of Hadrian, which is figured
by Gordon,4 little or nothing that can be associated

1 Roy's plan works out to 440 feet by 300 feet, or thereby. The mark-
ings on the O.S. Map indicate a considerably smaller enclosure.

»///«. Sept. PI. 1 6.

'Horsley's plan also shows a second branch connecting the Military
Way with the Porta Principalis Dextra.

in. Sept, ' Plate of Medals and Intaglios,' Fig. 7.

DUNTOCHER 157

with the Romans appears to have been found within
the limits of the station.1 Shortly before 1785 the
surface was scratched by a treasure-hunter,2 but no
systematic excavation has ever taken place. None of
the inscribed stones discovered in the neighbourhood
are definitely associated with the fort. Here, however,
as at Chapel Hill, the Bath has come to light through
an accident. A clear and interesting account of what
then transpired is contained in a work entitled A View
of Scotland, published in 1785, and written by John
Knox, a London Scot, who was a native of West
Kilpatrick, and uncle of the John Millar Morrison
referred to in connection with Chapel Hill. Knox
happened to be on the spot at the time, and he tells
us that it was through his efforts that something like
a proper examination of the place was secured. His
description is as follows : " Upon the declivity of the
Golden Hill, in the vicinity of the bridge, in the year
1775, a countryman, in digging a trench, turned up
several tiles of uncommon form. The tiles were of
seven different sizes, the smallest being seven, and
the largest twenty-one inches square. They were from
two to three inches in thickness, of a reddish colour, and
in perfectly sound condition. The lesser ones com-

1 Other Roman coins (Domitian, Trajan, Pius, and Faustina) have been
found "in the vicinity": see Stuart's Caled. Rom. 2nd. ed. p. 304, footnote.

2 " On the summit of the hill stood the Roman fort or castella, of which
Mr. Gordon hath given a drawing. The foundation was lately erased by
a clerk or overseer of an iron factory in that neighbourhood, who was,
however, disappointed in his expectations of finding treasure. The same
Goth expressed a strong desire to erase a fine remain of the Roman Wall,
which is carried along the base of the hill, but he hath not succeeded in
his wishes, and it rests with the family of Blantyre to prevent such prac-
tices in future upon grounds of which they are the superiors." Knox's
View oj 'Scotland (1785), ii. p. 611. See also infra, pp. 159 f.

158 THE FORTS

posed several rows of pillars, which formed a labyrinth
of passages of about 18 inches high, and the same in
width. The largest tiles, being laid over these pillars,
served as a roof to support the earth on the surface,
which was two feet deep, and had been ploughed
through from time immemorial. The building was
surrounded by a subterranean wall of hewn stone.
Some professors in the University of Glasgow, and
other gentlemen, having unroofed the whole, dis-
covered the appearance of a Roman hot bath. The
passages formed by rows of pillars were strewed
with bones and teeth of animals, and a sooty kind
of earth ; in the bath was placed the figure of a
woman, cut in stone, which, with a set of tiles and
other curiosities found in this place, is deposited in
the University."1

Knox's story bears the manifest stamp of truth, and
it may be at once accepted as reliable, even although
only one of the relics still survives ; the " figure of a
woman cut in stone " is in all probability identical with
a statuette which is familiar to every one who knows
the collection of Roman stones in the Hunterian
Museum.2 Furthermore, the details recorded are
sufficiently full to place it beyond doubt that Stuart
was wrong in supposing that two distinct discoveries

1 Op. tit. ii. p. 611.

3 This identification, now for the first time suggested, is hardly open to-
doubt. The statuette, which had evidently been carved as a support for
the inflow pipe of a bath or basin, is figured and described in chap. xi.
infra. Previously it has been supposed to be "of uncertain locality."
But doubtless that was merely because Knox's book had been overlooked.
His account of the " image," and what became of it, being contemporary,
is entitled to much more weight than is the version given in Caled. Rom.
2nd ed. p. 302, footnote, which probably refers to something quite
different.

DUNTOCHER 159

of the kind had been made in 1775.* Stuart's two
discoveries are simply two different versions of one
and the same incident — the first apparently supplied
orally, after the lapse of nearly seventy years, by an
old man who had as a youth borne an active part in
the exploration, the second derived from the very
complete and careful report published in Cough's
edition of Camden's Britannia? Gough gives sketches
made on the spot. These are less illuminating than
they might have been. But, taken along with the
markings on the Ordnance Survey Map, they show
that the Bath — for such it was beyond any manner of
doubt — lay about 100 yards to the north-west of the
fort, near the foot of the hill and not far from where
the church now stands.3 That it was within an annexe
may be taken for granted ; and that the annexe
extended some distance to the south is rendered
probable by the reported finding of Roman querns
about 50 yards away, near the south-east corner of
the wall of the church.4

After describing the opening up of the Bath in 1775,
Gough adds : " On prosecuting these discoveries three

1 Caled. Rom. 2nd ed. p. 302 and p. 303.

2 Vol. iv. (ed. 1806), p. 102 and PI. V.

3 It seems certain that what purports to be a ground-plan of 'the fort'
in Cough's illustration (PI. V. A) was really intended to be a ground-plan
of the newly discovered remains. It has no resemblance to any Roman
fort that ever existed. On the other hand, it is very like a Roman bath \
there are three apartments and two apses. Besides, what seems to be
the outline of the fort is shown in its proper place upon the hill-top. It
must be remembered that Gough did not make the sketches himself.
He received them from a third party through an intermediary, so that the
possibility of error is considerable.

4 The spot is marked on the O.S. Map, and the N.S.A. (Dumbarton-
shire) mentions that one was dug up when the foundations of the church
were being excavated (op. cit. p. 22).

160 THE FORTS

years after, there were found several beautiful frag-
ments of pottery, which being laid before the Society
of Antiquaries of London, drawings were made of them
by Mr. Basire." The drawings, which are repro-
duced,1 represent characteristic second-century vessels
— ' Samian ware,' with the egg and dart border, and a
portion of a mortarium bearing the name of the potter,
BRVSC.2 Mention is also made of "a piece of lead as
if cut from a bar or pig, weighing three or four
pounds," and of two pieces of window-glass. Cough's
language naturally suggests that these further researches
were carried on within the annexe. It is, however,
just conceivable that the operations may be identical
with the futile treasure-hunt which Knox speaks of
as having been conducted on the summit of the hill by
" a clerk or overseer of an iron factory in the neigh-
bourhood."3 If so, the Mr. Charles Freebairn whose
" diligent researches" are eulogized by the later writer
must be " the same Goth " whose disappointment is
chronicled by Knox with such unconcealed satisfaction.
In any case the moral is plain. Properly organized exca-
vation at Duntocher would meet with an ample reward.
The third of the forts on the Limes stood upon the
Castle Hill, rather less than two miles beyond the
second. The situation is most picturesque, the summit
being surrounded by a belt of beech-trees enclosing
the whole area within which the Roman station once
lay (Plate XVI. i). On a clear day the outlook is

1 Op. cit. PI. VI.

2 Hiibner (C.I.L. vii., No. 1334, 18) makes the stamp BRVSC • F. This
is possible, but not certain. Cf. the example from Newstead (Curie, A
Roman Frontier Post, pp. 265 f.).

3 See supra, p. 157, footnote.

PLATK XV

I. X. VV. CORNER, LOOKING \V.

2. N.K. COKNF.R, LOOKING K.

THE OUTLOOK FROM CASTLEHILL FORT

DUNTOCHER AND CASTLE HILL 161

magnificent, extending to the Clyde and far beyond it
on the west, while on the east the Bar Hill is well
within the range of vision. The commanding nature
of the position is well brought out by the two illustra-
tions in Plate XV. Inside the enclosure the traces
of the ditches still suffice to indicate the original outline
of the defences. The castellum, which is one of the
smallest of the series, has been of the normal shape.
As usual, the major axis lay parallel to the Vallum,
which was utilized as the northern rampart. According
to Roy, the interior dimensions were about 300 feet by
200 feet. At various points within, there are marks of
foundations that call aloud for investigation. One of
these is peculiar as being circular. Possibly it is post-
Roman. The solitary surviving relic of buildings was
recovered in 1847 in the field that slopes down to the
Peel Glen. This is the base of a pillar that may once
have formed part of the colonnade round the courtyard
of the Principia. The square plinth has a chevron
ornament carved on it in bold relief (Plate XX. i).
Other stones have been ploughed up in the vicinity,
some of them fortunately bearing inscriptions. But there
seems to be no record of finds of pottery or character-
istic debris of any sort. In other words, the site has
not been disturbed within the memory of man. The
prospects of excavation are, therefore, hopeful. If it
is ever undertaken, the field to the south should be
carefully trenched. There is some reason to think
that an annexe containing buildings may be hidden
underneath its surface.1

1 Information from Mr. Houston of Over Croy, who has often driven the
plough over it, and reports that there are extensive foundations at no
great depth.

L

1 62 THE FORTS

The small size of the castellum just described is
perhaps not unconnected with the fact that its nearest
neighbour on the east was exceptionally large and was
only about a mile and a third away. Hiibner found
that the presence of a fort at New Kilpatrick was
inconsistent with his scheme of nomenclature, and
he sought to dispose of the difficulty by refusing to
believe in its existence.1 The evidence, however, is
too strong to be gainsaid. Gordon, Horsley, Maitland,
and Roy in the eighteenth century, and Stuart in the
nineteenth are unanimous as to its position and practi-
cally unanimous as to its dimensions. Roy says : "The
fort of New Kirkpatrick, stands lower than most we
meet with on the wall, having the rivulet which
afterwards falls into the Allander in front. And as
the rising grounds, on the right and left of this post,
form a sort of gorge or pass, through which it seems
to have been apprehended that the enemy might pene-
trate from the north and north-west, therefore the fort
hath not only been made of larger dimensions, but
likewise, to render it more respectable, it hath been
surrounded with a double envelope ; though it is so
much defaced by the plough, that, excepting on the
south side, it is with much difficulty it can be traced."2
And Stuart tells us : " In the spring of the year, before
the vegetation has begun to show itself above the
surface, a slight appearance of the trenches may yet
be seen, especially towards the south-west angle of
the station ; where the usual curve may be plainly
followed by the eye."3 He adds : " Of an oblong form
— rounded as usual at the corners — this station was

1 C.I.L. vii. p. 201. 2 Military Antiquities, pp. 158 f.

8 Caled. Rom. 2nd ed. p. 312.

NEW KILPATRICK 163

defended on three sides by a double course of ramparts
and ditches, and on the fourth by the line of the con-
tinuous Wall with its protecting fosst. The area,
inclosed within its interior Valla, measured about
480 by 330 feet ; from which, on the East, West,
and South, the intrenchments extended outwards to
the breadth of more than 1 30 feet. The Military Way
passed directly through it."1

Stuart's measurements have probably no independent
value, being merely taken from Roy's map, and his
statement as to "a double course of ramparts" is
absurd. But his general description is fully borne
out by the markings on the Ordnance Map of 1890.
There the south-west corner is still distinctly visible,
although everything to the north of the Military Way
— now represented by the modern thoroughfare which
goes by the name of the Roman Road — has entirely
disappeared. Since 1890 there has been a notable
change. The land both north and south of the Roman
Road has been fenced in and built upon, and the
space once thronged by the Roman soldiery is occupied
by the villas and gardens of Bearsden. The fort, it is
to be feared, must be regarded as lost to archaeology.
It did not succumb without a struggle to protest
against its burial. In one case what was evidently
the stone foundation of the southern rampart caused
great trouble to those who were laying out the garden ;
it cost much labour to uproot it.2 In another the
corner of a house in process of erection subsided so
seriously that it had to be taken down and rebuilt with
the aid of a steel girder. It turned out that the

1 Op.cit. p. 313.

2 Information from Mr. John Annan.

164 THE FORTS

foundation had been laid, not upon the solid ground,
but upon a soft mass of black material which yielded
to the superincumbent weight.1 Doubtless this was a
rubbish-pit of the type with which Bar Hill and New-
stead have made us familiar.

In yet a third case the workmen, in making up
ground for gardens, came across what are said to have
been " a number of pits from 30 to 36 inches in dia-
meter and similar in depth. In the bottom there was
usually or always some ashes or charred wood."1 These
were perhaps post-holes.2 If so, the barracks of the
fort were probably of wood, as at Bar Hill and at
Ardoch. That the rampart was of turf may possibly
be inferred from the fact that, so far as can be learned,
the stone foundation was similar in character to the
stone foundation of the Antonine Vallum.3 There is
not even a tradition of any buildings of a more
substantial character, for it is clearly to the foundation
of the rampart that Stuart refers when he says : " The
remains of the Roman works about East Kilpatrick
have proved a perfect quarry for the supply of building
materials. Many hundred cartloads of stones have
been removed at different times from the line of the
Military Way, and also from the foundations of the
Station, and many hundreds more remain to be dug
out whenever they may be required."4

Considering the size of the fort and the extent

1 Information from Mr. W. M'Clure, through Mr. John Annan.

2 The diameter here given seems rather large ; the post-holes at Bar Hill
had a diameter of about 2 feet only. But the description of the Bearsden
pits is given from memory after the lapse of a good many years. There
is also the possibility of ////«, such as those found at Rough Castle.

3 Information from Mr. John Annan.

4 Caled. Rom. 2nd ed. p. 313.

NEW KILPATRICK AND BEMULIE 165

to which the ground on which it once stood has
been trenched in recent times, it is astonishing that
so little has been heard regarding any discoveries of
Roman remains. Perhaps the trenching has not been
sufficiently deep. Stuart mentions some fragments of
pottery " which much resembled those discovered at
Duntocher ; they likewise contain the figures of cen-
taurs, and what might almost be called a copy of the
Medicean Venus — all in low relief.1 The pottery was
evidently ' Samian ware ' or terra sigillata, and the
description of the decoration suggests that these parti-
cular pieces belong to the Antonine period. Such are
the few meagre details that can be gleaned regarding
the fourth of the stations on the Limes. Twenty years
ago the site was full of possibilities. Now it is
unlikely that we shall ever learn more.

The fifth of the series of forts was at Balmuildy or
Bemulie, which lies on the south bank of the river
Kelvin, a little over 2^ miles from New Kil-
patrick. Such information as is available suggests
that it must have been one of the most important of
all the Limes stations. Gordon saw there "the
Remains of a prodigious Fort very like the Ruins of
an old City." He thus describes them : " On the
South side of Kelvin, the great ruins of Bemulie begin
to appear, and shew it originally to have been a very
magnificent Place. On the West-End of the Village
are to be seen four Rows of Ramparts, with as many
Ditches between them ; but on the other Sides they
are more obscure and flat. Within the Area of these

1 Op. cit. p. 315. The fragment with the "copy of the Medicean
Venus" may be compared with that from Camelon illustrated in Proc.
Soc. Antiq. Scot. 1901, Fig. 14 (p. 385).

1 66 THE FORTS

Ditches are great Foundations of Stone Buildings, but
so embarassed with the Cottages now built upon them,
that one cannot form a right Idea of the whole.
Under the Ground here, are several arched Vaults, as
the Country People informed me, with hollow squared
Stones, by way of Conduits for bringing in of Water
from the River Kelvin.'"1 When allowance is made
for differences of temperament, Horsley seems to have
been quite as deeply impressed as Gordon : " The
ruins of the Roman town or out-buildings are very
remarkable. Several subterraneous vaults have been
discovered, and Roman antiquities found here. The
west side of the fort is still very visible, and appears
to have had a fourfold rampart and ditch." 2 Even
Maitland hardly attempts to depreciate what his pre-
decessors had admired, although he is sceptical about
the vaults and makes merry over the notion that
water could run uphill or be pumped through stone
conduits.3 Parenthetically, it may be remarked that
his scepticism is not justified; the "vaults" were in
all probability ruined hypocausts. Roy says : " This
station, as well from its size, as from the number of its
envelopes, and the many vestiges of ruinous founda-
tions within it, hath been one of the most considerable
belonging to the wall, though the whole work is very
much defaced."4 His plan represents it as almost an
exact square, each side measuring on the interior
rather more than 400 feet. The northern defence is
formed by the great Rampart, while the Military Way
is shown passing through the fort from west to east.
The two features last mentioned are not to be

1 //»». Sept. p. 53. *Brit. Rom. p. 167.

3 Hist, of Scotland, p. 1 79. * Milit. Antiq. p. 1 59.

BEMULIE 167

accepted without verification. They are not in har-
mony with the report of Horsley,1 made at a consider-
ably earlier period. At the same time Horsley 's
sketch is prima facie improbable. It may be that
neither it nor Roy's plan is correct. Excavation alone
can resolve the doubt, for before Stuart wrote, the
process of obliteration had been practically completed :
" At the present day the vestiges of its works are all
to be numbered with the things that were." 2 While
the modern visitor will readily endorse that statement,
he may be tempted to surmise that, as the result of an
abrupt bend executed at this point, the great Rampart
may have run along the east front of the station. It
seems to have crossed the Kelvin a short distance
above a solitary willow-tree, close to which is an old
ford.

Although so little is left above the surface, the
site offers a great opportunity. The fort must have
lain between the farm of Easter Balmuildy and the
river, and the field though seemingly level is said to
be one mass of foundations. The present tenant has
only attempted to crop it once during an occupancy of
25 years. Everywhere the plough grated against
heavy stones.3 It is a natural inference that at Bal-
muildy the interior buildings may have been of solid
masonry. There are appearances to suggest that some
of the material is to-day embedded in the dykes and
outhouses about the farm, while Dr. John Buchanan

" The wall seems to have come up to the north rampart of this fort
without forming the whole of it" (Brit. Rom. p. 167). See also his plan.

2 Caled. Rom. 2nd ed. p. 317.

3 Information given to myself on the spot. The attempt, it was added,
would not be repeated.

1 68 THE FORTS

is the authority for the following : " At the beginning
of the present [nineteenth] century, a small hamlet of
a dozen cottages existed within the ramparts of
Bemulie Fort, entirely built from the Roman ruins.
Finely sculptured stones, one in particular with a
human figure in high relief and wreaths of flowers,
were visible in the walls of the cottages, but are now
all lost. Facing the Kelvin was a mass of ruins,
probably of a watch-tower, which cost much trouble to
root out."1 He adds that "in 1848 the farmer, while
trenching the sloping field between the Kelvin and the
rampart, came upon a mass of ruins, of circular shape
and resembling the cradle of a well, within which was
a quantity of blackish-coloured stuff, like charred
wheat, and a coin of middle-brass of Antoninus Pius,
in fair preservation."2 This will immediately recall
the pits at Bar Hill and Newstead ; or possibly it
was a kiln. No other minor antiquities are specifically
recorded as having been discovered in or near the
fort. But the yield of inscribed stones has been
unusually rich. It includes the fragmentary slab
with the name of Lollius Urbicus, declared by the
enthusiastic Gordon to be "the most invaluable Jewel
of Antiquity, that ever was found in the Island of
Britain, since the Time of the Romans" 3

The exact site of the sixth station is doubtful. Pont
and Sibbald, the latter perhaps following Irvine, place
" a great Fort" at " East Gaidar" or " Easter Calder." 4
But none of the eighteenth-century antiquaries saw
at what is now called Easter Gadder anything to
suggest that there had ever been a station there. And

1 Stuart's Caled. Rom. 2nd ed. p. 320. 2 Ibid,

3 Itin. Sept. p. 63. 4 Sibbald, Hist. Inq. p. 29.

BEMULIE AND CADDER 169

its relative position is not favourable to the assumption.
It is fully 3 miles from Balmuildy, and only a little
more than a mile from Kirkintilloch on the other side.
On both grounds the claims of Kirktown of Cadder
are decidedly stronger. It is 2 miles from Balmuildy
and about 2^ from Kirkintilloch, while the soil has
not proved altogether barren of Roman antiquities.
Gordon, it is true, saw nothing. But William Hamilton
of Wishaw, who died a year or two before the Itiner-
arium Septentrionale was published, says explicitly :
" Near to the church of Calder there are very lyvely
vestiges of ane Roman encampment, and its fortifica-
tions."1 It might, of course, be argued that what
Hamilton had in his mind was the Cadder tumulus,
of which we have already spoken. It is otherwise
with Horsley. He was quite familiar with the
tumulus. But he states the case for something more
considerable with studious moderation : " The dis-
tance, situation, and some faint appearance of remains,
plead for a station near Calder church. But if there
has been a Roman fort or town here, it has been
much plundered by the village, and levelled by the
plough ; for what remains there are now, are but
doubtful and faint. The station is most likely to have
been in the grounds called the crofts, and these have
been in tillage time immemorial."2 Two subsequent
discoveries enable us to speak on the point with much
more confidence, and even to fix the site of the fort
with some approach to precision.

The first of these discoveries does not seem to be

1 Description of the Sheriffdom of Lanark, etc. (compiled about 1710
and published for the Maitland Club in 1831), p. 32.
*Brit. Rom. p. 168.

170 THE FORTS

noticed in any of the printed accounts of the Wall.
Some details regarding it have, however, been pre-
served in a manuscript by Professor John Anderson,
now in the library of the Glasgow and West of Scot-
land Technical College. The finding of the Auchen-
davy altars in 1771 interested Anderson immensely.
In that year he wrote, evidently for communication
to some local society,1 a paper on the Wall which
comprises a sketch of its history and its track, a
description of the inscribed stones then in the posses-
sion of the University, and full particulars regarding
the contents of the Auchendavy pit. What he has to
-say regarding the course of the Limes has no inde-
pendent value, the facts and much of the phraseology
being borrowed direct from Gordon and Horsley. Nor
does the rest of the original communication contain
anything of importance that is not included in the
appendix which Anderson contributed to Roy's Mili-
tary Antiquities. Two years later (1773) a new
discovery suggested a supplementary paper : "In
digging the Canal near the village of Cadder close
upon the track of the Roman wall, and several feet
underground, the workmen found the top of an altar
together with an upper and the half of a nether miln-
stone. These . . . were in October last given to this
University by the Proprietors of the Canal, and are
to be deposited in the same place with the former
Presents from that publick spirited company." The

1"The Society is not named, but there can be no doubt it was the
Literary Society of Glasgow, which during the last half of the eighteenth
century met within the University weekly while the session lasted for the
discussion of literary and philosophical subjects" (J. Macdonald, The
Roman Room of the Hunterian Museum, p. 5). Cf. Coutts, Hist, of the
Univ. of Glasgow, p. 3 1 6.

CADDER 171

altar top is in all probability to be identified with
No. 35 of the Titiili Hunteriani, where it is illustrated
as Fig. 4 of Plate XVI. It is there said to be of
unknown origin.1 The quern-stones are undoubtedly
those that appear as Fig. 2 on the same plate of the
Titu/i.9

As to the second discovery Dr. John Buchanan
says: "In 1852, while trenching part of the glebe,
near the manse of Cadder, four unfinished altars, a
thin and neatly dressed tablet ready for an inscription,
and a quantity of Roman pottery, were discovered.
Three of these altars were lying together — the other
altar and slab by themselves."3 Dr. Buchanan here
speaks from personal observation. He subsequently
amplified his brief note, in a paper read before the
Society of Antiquaries of Scotland on April nth,
1 853.* After telling how he saw the Military Way

1 Op. cit. p. 78.

2 The account of them given in the text is that they are " said on the
Monumenta Plate to have been dug up along with Nos. 19 to 23 [the
Auchendavy altars] and presented to the University by the Canal Com-
missioners " (op. cit. p. 93). This evidence cannot stand against Ander-
son's contemporary M S. record. Besides, the confusion is easily accounted
for : the donors were the same, and (as Anderson tells us in the MS.) the
iron hammers from Auchendavy were handed over along with the Cadder
stones, the altars having been presented two years earlier. The sugges-
tion in the Tituli (p. 3) that Anderson may have been the editor of the
Monumenta Imperil Romani, published by the University in the latter
part of the eighteenth century, may very well be true as regards the
earlier series of plates (Nos. I. -XX.), which seems to have appeared about
1767 (Coutts, Hist, of Univ. of Glasgow, p. 332). It can hardly apply to
the later (Nos. XXI. -XXXI I.), in which the stones now under discussion
were figured. Mr. James Coutts gives "about 1789 or 1790" as the date
when this second instalment was issued (Hist, of the Univ. of Glasgow^
p. 333)- At that time Anderson had been for several years at open war
with the Faculty and was in no mood to edit anything on their behalf.

3 Stuart's Caled. Rom. 2nd ed. p. 322, footnote.

4 Proceedings^ i. pp. 170-175.

172 THE FORTS

rooted out for some 200 yards where it crossed the
manse garden from west to east, he proceeds : " Close
to this causeway, a considerable quantity of Roman
pottery was turned up. It consisted of portions of
amphorae, vases, bowls, jugs, and large circular shallow
vessels, apparently mortaria. Many of these were of
the fine red Samian ware, highly glazed ; and when
cleaned the beautiful crimson was as clear and fresh
as if of yesterday. Several of these fragments had
evidently been impressed when soft by the potter's
stamp, within the usual small oblong border; but the
letters were quite illegible."1 Nothing, of course, is
more characteristic of Roman occupation than the
brilliant lustrous ware familiarly, if not quite accurately,
known as ' Samian/ But that was not all. " Near
this group of pottery were found a number of large
iron nails, with very broad, round heads. These nails
are 6 or 7 inches long, and though much corroded, are
still thick. They closely resemble those . . . found
within the ruins of the great station of Borcovicus, in
Northumberland. Alongside of them lay several
hones for sharpening knives."2 All of these, as well
as a fragment of an inscribed stone,3 were brought to
light within the garden. The unfinished altars were
discovered in the field outside, about 2 feet below the
surface. With them were "some fragments of what
appear to have been weapons . . . but all were greatly
corroded, and crumbled to pieces on being handled."*

1 Op. cit. p. 174. 2 Op. cit. pp. 172 f.

3 From the traces of letters Dr. Buchanan was inclined to think that
the stone had formed part of a slab erected by the Second Legion
Augusta. But this was a mere suggestion. The inscription, as restored
by Dr. Buchanan, is C.LL. vii. No. 1128.

*Op. cit. p. 173-

CADDER 173

The cumulative effect of Dr. Buchanan's evidence is
very hard to resist.

And there is unanimity of testimony even as to the
position of the station. The present Manse of Cadder
stands on a gentle height, round the foot of which the
Forth and Clyde Canal passes, coming north from
Glasgow and bending abruptly eastward towards
Kirkintilloch. The new garden, in trenching which
the discoveries described by Dr. Buchanan were made,
lies on the north side of the Manse, and is, of course,
part of the glebe. Its north wall is about 100 yards
south of the line of the Antonine Rampart, which
again is close to the bank of the Canal. It is thus
almost certain that the ruins of the Principia are buried
somewhere within the field that separates the garden-
wall from the water-way. That agrees absolutely with
the evidence provided by the finds of 1773. They
must have been dug up either at the north-west
corner of the fort or, less probably, immediately
beyond its northern front, for nowhere else is "the
Canal near the village of Cadder close upon the
tract of the Roman wall."1 But this is also the
exact situation which Horsley conjectured to be
"most likely."2 The manse and glebe occupy "the
grounds called the crofts" the former having been
erected there in the closing years of the eighteenth
century. The farmhouse two fields to the east of
the manse is named Crofthead, and within living
memory one or two old houses stood within the manse
avenue gate, with the 'rigs' attached to each still
stretching behind them.3

1 Anderson MS., quoted supra, p. 170. 2See supra, p. 169.

3 Information from the Rev. J. B. A. Watt.

i?4 THE FORTS

On the supposition that there has been a fort at
Cadder, Kirkintilloch may well be the site of the
seventh of the series ; as has been already noted, it
lies about 2^ miles further east. And since Timothy
Font's time the Peel there has been recognized by
common consent as the remains of a Roman castellum.
Sibbald, in the account which he communicated to
Bishop Gibson, calls it " the greatest fort of all."1
Gordon treats it rather cavalierly, not questioning its
Roman origin for a moment, but remarking : " What
Antiquities and Inscriptions have been found at this
Fort, I never could learn."2 Horsley is much more
detailed: "At the west end of Kirkintilloch . . . stands
another Roman fort, called the Peel, small but very
strong, and the best preserved of any. It has had
a double rampart of hewn stone, strongly cemented
with lime. They were just at the time of the survey
working stones out of it, and it was surprising to see
how fresh both they and the lime seemed to be, and
some of them were chequered. The east entry only
is visible. On the north side is a considerable descent,
and the prospect from it is pretty good. According to
the common opinion and tradition the wall has passed
on the south side of this fort ; which, if true, might
account for the extraordinary strength of it. But per-
haps the military way has been mistaken for the wall,
which, notwithstanding this common opinion, may have
formed the north rampart of the station."3 The most
surprising element in the foregoing description is " the
common opinion and tradition" that the fort stood on
the north, or Caledonian, side of the Limes. Sur-

1 Gibson's Camden (1695), p. 959. zltin. Sept. p. 54.

*Brit. Rom. pp. i68f.

KIRKINTILLOCH 175

prising as it is, it has been very generally accepted as
correct, Horsley alone throwing any doubt upon its
accuracy. Roy, whose judgment in such a matter is
entitled to great respect, is quite positive that the
Rampart "passed to the southward of the fort, called
the Peel, situated just in front of it."1 Maitland's
explanation should be noted ; he concluded that the
Peel " from its situation without the Wall, seems to
have been one of Agricola's castella, built before the
erection of this fence."2 There is, however, another
solution that deserves to be considered. The Peel of
Kirkintilloch may have nothing at all to do with the
Romans.

So far as present appearances go (Plate XVI. 2), it
is certainly impossible to detect in it any trace either
of Roman design or of Roman workmanship. The
main feature is the great ditch, by which it is still
partly surrounded. Here shape and dimensions com-
bine to suggest a mediaeval moat much more than a
Roman fossa ; the depth is fully 1 8 feet on the eastern
side, with a wide, flat bottom. And indeed there is
good reason to believe that in the Middle Ages the
place was a formidable stronghold. As early as 1 1 70
Kirkintilloch was erected into a burgh of barony by
William the Lion in favour of William Cumin, Baron
of Lenzie and Lord of Cumbernauld. The Peel
undoubtedly marks the place which the castle of his
successors occupied.3 We can hardly help inferring

1 Milit. Antiq. p. 159. * Hist, of Scotland, i. p. 178.

3 During the Wars of Independence it long held out in the English
interest, the garrison being loyal to Edward II. at least as late as 1309.
According to Hill Burton (Hist, of Scot., ed. 1876, vol. iii. p. 429), the
charge upon which Robert Wishart, the " warlike " Bishop of Glasgow
was arrested by Edward I., was "that having been licensed by King

1 76 THE FORTS

that it was to this castle that the masonry which
Horsley saw in process of demolition belonged,
although it is, of course, far from impossible that
the " chequered " stones may represent the debris of
a Roman fort, turned to good account by subse-
quent builders. Since Roy's day there have been
extensive alterations ; the interior area is now only
about 200 feet square as compared with 300 feet in
the plan published in his Military Antiquities? and
recently the Town Council have put themselves to
some expense in laying it out for use as a public
park. A few years ago a portion of the ground
within the enclosure was opened up. No account
of the operations has been published. But it is
known that they revealed the existence of an old
well, which was cleared out to a considerable depth.
Nothing distinctively Roman appears to have been
found, while the construction of the well itself, which
was circular and built of neatly-dressed and closely-
jointed freestone, seemed to be mediaeval.2

The evidence so far is, therefore, almost entirely
negative ; and alongside of it we are bound to place
the general scarcity of Roman antiquities from the
site. Anderson, in the MS. already cited, speaks of
"altars without Inscriptions" as having been dis-
covered at Kirkintilloch. He gives no details. Dr.
John Buchanan had in his collection a stone with
a filleted bucranium sculptured upon it, which was,

Edward to cut timber for the completion of the woodwork of Glasgow
Cathedral, this timber he employed in the construction of instruments of
war for the siege of the Peel of Kirkintilloch."

1 The exact dimensions are : Roy's plan, 285 ft. (N. to S.) x 300 ft.
<E. to W.) ; and O.S.M., 210 ft. (N. to S.)x 1756. (E. to W.).

2 Information from Mr. Alexander Park.

PLATE XVI

I. CASTLE HILL, FROM BEHIND THOKN FARM

2. THE I'EEL OF KIKKINTILLOCH, LOOKING N.E.

KIRKINTILLOCH 177

he tells us,1 "dug out of the ruins of this Fort," and
which had at one time borne an inscription. He also
mentions the finding of coins of Domitian, Antoninus
Pius, Commodus, and Constantine, some of which had
passed into his own possession. The "several slender
pillars with ornamented capitals, seeming to have be-
longed to a building, probably a small temple," might
conceivably have come from the colonnade round the
courtyard of the Principia. These, together with "a
number of hollow pipes of burnt clay, were dug up
within the ruins and adjoining field."2

The ordinarily received record seems thus at first
sight not quite so meagre as it was in Gordon's time.
Unfortunately, however, with the possible exception
of the coins,3 the individual items do not emerge very
successfully from the ordeal of a critical scrutiny. The
notice of the "altars" is vague. Dr. Buchanan's de-
scription of " a stone, having sculptured on it, in bold
relief, the head of a bull with distended nostrils and
a fillet across the forehead," is scarcely suitable either
for an altar or for an inscribed slab of the normal type.4
The "hollow pipes of burnt clay" have long since dis-
appeared, but, so far as we can now judge, they might
have been of any age. The same holds good of the
" bar of lead " spoken of by Stuart.5 Even the "slender

1 Stuart's Caled. Rom. 2nd ed. p. 324, footnote. 2 Ibid.

3 The inclusion of a stray Constantine need not vitiate their evidence.
More serious is the absence of particulars as to the find-spot ; it might
have been Auchendavy, which was near enough to be confused.

4 Stuart's Caled. Rom. 2nd ed. p. 324, footnote. The Rev. John
Skinner, however, who saw it in 1825 {Brit. Mus. Addit. MS. 33686, fol.
54-58), notes it as " base of altar with head of some animal in alto relievo."

6 Op. cit. p. 323, where the inscription is said to be illegible. In the
Skinner MS. the reading appears once as P • CCLXX and once as
ccxx. In C.I.L. vii. (No. 1219) it is given as CCLXX, the ultimate
source being Dr. John Buchanan.

M

i y8 THE FORTS

pillars " must be regretfully relegated to the category of
objects whose Roman character is doubtful. Waldie in
his Walks along the Northern Roman Wall, published
in 1883, speaks of having seen in the narrow lane
leading to the Peel, not merely "some of the chequered
stones so distinctive of the old workers," but also " the
capital of a small pilaster built in the wall of one of the
cottages" and "another and more perfect pilaster built
in the back wall of the cottage— a flat pilaster about
three feet high, with a base, and a capital with two
Ionic volutes."1 While the "chequered stones" — other
examples of which Waldie noted " in a ruined dyke
within the Peel" — remind one of Horsley, the "pil-
asters" are doubtless identical with Dr. Buchanan's
" slender pillars." Cottages and dyke were pulled
down many years ago. The " pilasters " still survive
in the garden of Mr. Caldwell, slater, Kirkintilloch.
But the trend of expert opinion regarding them is
hardly such as to encourage the belief that they
can be Roman. They probably date from about
A.D. 1 200.

There is a popular idea that inscriptions of the
Roman period have been found in or near the Peel,
but no properly authenticated facts have been cited to
support it. The loose statement in Stuart's text to
the effect that a legionary stone " was turned up near
the Fort of Kirkintilloch, and is the only discovery of
the kind which has been made about that station"
refers to a slab which really came from Inchbelly
Bridge, a mile or so to the east.3 And the inscription
catalogued as No. 1123 in C.I.L. vii. is the flimsiest of

1 Op. tit. p. 53. 2 Caled. Rom. 2nd ed. p. 324.

3 See chap. ix. infra, No. 1 5.

KIRKINTILLOCH 179

phantoms.1 Again, local historians say that Joseph
Train, the correspondent of Sir Walter Scott, was
about 1822 "removed for temporary duty to Kirkin-
tilloch, where he got possession of several valuable
Roman relics, a sword, a tripod, and a brass plate.
These he transmitted to Abbotsford." 2 The brass
plate is in the Abbotsford Collection. It came, accord-
ing to Train's MS. account, from Castlecary, but
there is no trace of Roman work about it. The
"tripod" is doubtless what Train describes as "a
Roman flagon found on the i4th of this month
amid the ruins of Graham's Dyke about a mile from
[Kirkintilloch]." It is a jug of copper or bronze,
with a spout, a single handle, and three legs — a well-
known type, now generally recognized as mediaeval.
The Collection contains no other objects that can be
associated with Kirkintilloch, and Train does not
mention any others in his MS. notes.3 There is
one more piece of evidence that must be noticed
before the depressing task of destructive criticism
is completed. This is a " fragment of dark red

lHubner (I.e.} refers to Sinclair's Statistical Account of Scotland (vol. ii.
p. 276) as authority for the statement that the inscription he quotes was
found at " Kirkintilloch, in the Peel fort." But the writer of the Statistical
Account is very far from being so precise. After describing the forts of
Bar Hill and Auchendavy, as well as the Peel, he proceeds : " Stones,
bearing inscriptions, have been dug up among the ruins of all these
forts : But the only words of these inscriptions that could be read, were,
LEGIO SECUNDA AUGUSTA FECIT." Hiibner rightly points out that no
inscription can possibly have had this form. The words may be an
echo of the stone described as No. 19 in chap. ix. infra.

2T. Watson, Kirkintilloch Town and Parish (1894), p. 77.

3 1 have to thank Mr. James Curie for information regarding the Abbots,
ford Collection and Train's MS., which he was good enough to examine
at my request.

i8o THE FORTS

pottery, with figure of Satyr embossed in relief, found
near Kirkintilloch," and presented to the National
Museum of Antiquities in i846.x It is a tiny little
bit of unmistakable Castor ware, and it would there-
fore have been welcome, if its * find-spot ' had been
definitely ascertainable. As it is, we must set it aside.
There is no security that " near Kirkintilloch " does
not mean "at Auchendavy."

The Peel, then, is not Roman. Moreover, there is
no proof that objects of Roman origin have been
recovered from among its ruins. When such objects
are produced, it will be time to discuss its exact
relation to the Wall. In the present state of the
evidence, speculation on the point is futile. It does
not, however, follow that the claim of Kirkintilloch
to be regarded as the site of a Roman ' station ' must
also be left in abeyance. As a matter of fact, the
presumption of situation is entirely in its favour.
Hitherto the forts have followed one another at stated
intervals with unfailing regularity, and as we proceed
eastwards we shall experience no difficulty in picking
up link after link in the chain till we pass beyond
Castlecary. There is no apparent reason for a break
at the point we have now reached. On the contrary,
the position is so excellent that it is difficult to believe
that it could have been overlooked by the engineers
who designed the Limes. The broad expanse of
ground on the northern edge of which the Peel stands,
occupies the highest portion of a long ridge that rises
somewhat abruptly on the west and slopes slowly
downwards towards the east. The descent on the
north is steep, so that the general effect is that of

1 Catalogue of the National Museum of Antiquities of Scotland, p. 221.

KIRKINTILLOCH 181

a tall bluff or headland projecting boldly into the
wide flat plain of the isthmus. Between Forth and
Clyde there is no more appropriate spot for a fort.
And that there was a fort of some sort there long
before the Peel was built is rendered certain by the
place-name ; the modern form ' Kirkintilloch ' is the
quite natural evolution of the Cymric 'Cairpentaloch,'1
and ' Cairpentaloch,' which is at least as old as the
tenth century2 — how much older we cannot tell, —
means ' the fort at the end of the ridge/ We cannot,
of course, argue from this that ' the fort ' was Roman,
but in the circumstances a Roman fort is more than
a mere possibility. Judicious excavation, directed in
the first instance towards determining the precise
course of the Limes, would doubtless settle the
question finally. The extent of ground still unbuilt
upon is probably sufficient to afford ample scope for
investigation. In the meantime we may note a few
objects which can with confidence be assigned to the
neighbourhood. Some are certainly Roman.

On August 28th, 1893, a labourer digging in a
sand-pit near the Lion Foundry, at a spot between
300 and 400 yards east of the Peel, came upon a
number of denarii, with which were an iron spear-head

1 Dr. W. J. Watson tells me he had occasion to look into the point
carefully a few years ago, when he was satisfied that there was no
reason to question this. Dr. George Neilson has kindly sent me a
selection of charter spellings, going back as far as circa 1200. These
show a general tendency to approximate to the form ' Cairpentaloch,'
except that the 'p' is never found. The exception is not surprising;
the 'p '-Celts had been displaced in this district by the *q '-Celts long
before the time of the oldest charter.

2 The Nennius gloss in which it occurs (see sufira,p. 36) is believed
by Zimmer to go back to the year 910 (Nennius Vindicatus, pp. 42 ff.).
In making the Wall end at 'Cair Pentaloch' the scribe was probably
misinterpreting the place-name through lack of local knowledge.

1 82 THE FORTS

and a large nail.1 The deposit obviously belonged
to the Antonine period, for the denarii ranged from
Vespasian to Faustina Junior. About fifteen years
previously, either in 1878 or in 1880, workmen engaged
in laying a sewer in Cowgate Street, as the street
leading up towards the Peel is called, unearthed the
upper part of a large amphora, having a maker's
stamp upon the handle ; it is now in the Hunterian
Museum.2 Twenty or thirty years earlier two very
curious statuettes of coarse sandstone, sadly mutilated,
were got a little way down the slope west of the Peel,
close to the site of the Washington Hotel. They
passed into the possession of the late Dr. Stewart, and
subsequently into that of Mr. G. A. Buchanan of Cask.
They are clearly not Roman. An Oriental origin seems
probable, though a Celtic one is just possible.3

About a mile and three-quarters beyond the Peel of
Kirkintilloch there are still to be seen distinct traces of
the eighth of the Limes stations — the fort of Auchen-
davy. Horsley's brief description may be quoted : "It
has been encompassed with a triple rampart and ditch.
The ground on which it stands is marshy, no descent

1 Proc. Soc. of Antiq. of Scot. 1894, p. 276. Only 24 coins are there
described (Titus to Hadrian). For others, see infra, p. 379.

2 The label in the Museum describes the fragment as "found near
the Roman Peel or Fort, Kirkintilloch," and "presented by Dr. Donald
Patrick Stewart, 1878." In the third edition of Nimmo's Hist, of
Stirlingshire (p. 30) it is said to have been dug up "in May last
(1880)" in Cowgate Street. The details there given — "the neck and
handles are in perfect preservation but the under portion is completely
g0ne» — correspond so exactly to the Museum specimen that some con-
fusion of date is certain. In Home's In and Around Kirkintilloch (1905),
p. 16, the fragment is erroneously stated to be " in the National Museum."

3 1 have examined them personally, and Professors Haverfield and
A. J. Evans have seen photographs.

KIRKINTILLOCH AND AUCHENDAVY 183

from it but to the north, and but little there ; so that
the trenches are for the most part filled with water.
The military way is very visible, passing by the south
rampart of the fort, where there is a visible entry into
it."1 Gordon and Roy are in substantial agreement
with this, except that, while Gordon does not refer to
the Military Way at all, Roy makes it run through the
fort from west to east. If he is right — and, in view of
the configuration of the ground, it appears highly pro-
bable that the usual plan was adhered to — then the
modern highway has, as at Bearsden and elsewhere,
been laid on the line of the Roman road. For about
a hundred years after Gordon's examination of them
the aspect of the remains seems to have altered but
little. " Until within the last twenty years," says
Stuart, writing in 1846, "the outline of the ramparts,
the excavation of the moats, and the general form of
the inclosure, were tolerably distinct ; but finding those
'canals,' as the trenches were called by the people of
the neighbourhood, to be rather in his way, the pro-
prietor of the ground has had the surface levelled, and
almost every vestige of the ancient works removed."2
One is glad to be able to say that the demolition has
hardly been so complete as Stuart's words might
indicate. There is enough left, more particularly on
the eastern side, to confirm Roy's estimate of the
original size. The interior measurement has been
about 370 by 330 feet, the major axis being, as usual,
parallel to the Antonine Rampart, which was also the
northern defence of the station. Within the enclosure
there stand, to the north of the road, the dwelling-
house of the farm of Auchendavy and, to the south

1 Brit. Rom. p. 169. 2 Caled. Rom. 2nd ed. p. 327.

1 84 THE FORTS

of it, the steading. Some of the outbuildings of the
steading lie directly above the south ditches of the
fort. A conspicuous subsidence in the east wall of one
of them, close to a brick chimney, tells its own tale.
Auchendavy is distinguished for the large number
of antiquities that have been found in and about it.
Gordon1 and others speak of coins ; a gold solidus
of Trajan is definitely mentioned.2 Then stones of
obviously Roman origin, notably a small altar or two,
were observed by Horsley and his contemporaries
built into the walls of houses.3 Stuart records " among
the principal discoveries of recent years . . . various
fragments of Roman pottery — a deep hollow, contain-
ing charcoal and ashes — a deposit of stone bullets,
about the size of a 24-pound shot, lying many feet
under the surface." The ballista bullets are said to
have been upwards of fifty in number, and to have
been " lying in small pyramidal heaps, like those of our
cannon-balls."4 It is a little difficult to reconcile this
with the statement that they were found "many feet
under the surface." The expression just quoted, like
the " deep hollow containing charcoal and ashes,"
rather reminds one of the rubbish-pits that have
proved such store-houses of treasure at Newstead.
And it was clearly such a receptacle that was dis-
covered in May, 1771, immediately to the south of
the station — that is, probably, in one of the annexes
of the fort — by the workmen who were engaged in
making the Forth and Clyde Canal.

1 Itin. Sept. p. 54. Cf. Maitland, Hist, of Scotland, i. p. 178.

2 Stuart's Caled. Rom. 2nd ed. p. 328, footnote.

3 Brit. Rom. p. 169. Cf. Maitland, l.c.

• * Stuart's Caled. Rom. 2nd ed. p. 328, footnote.

AUCHENDAVY 185

A detailed account of this remarkable find, written
by Professor Anderson of Glasgow, was printed as an
addendum to Roy's description of the Limes.1 The pit
was a small one, being only 9 feet deep, and having a
diameter of 7 feet at the top, narrowing to about 3 feet
at the bottom. It contained four Roman altars and
a portion of a fifth, along with a mutilated stone bust,
which " seems to represent an auxiliary soldier," and
two huge iron mallets. The altars, which are now in
the Hunterian Museum, are exceedingly interesting.
They will be described when we come to discuss the
inscriptions. The mutilated bust was also presented
to the University of Glasgow. To judge from the
reproduction in the Military Antiquities? it is iden-
tical with a fragment still preserved beside the altars,
and figured and described in chap, xi.3 The mallets
appear to be now lost. Anderson's actual words regard-
ing them deserve quotation : " One of the iron mallets
weighs two stone fourteen and one -half pounds
Dutch, and the other, two stone two pounds ;4 and
from the first at least five or six pounds is broken off.
They have been much used, for their faces are greatly

1 Milit. Antiq. pp. 200-204. 2 PI. XXXVIII.

3P1. XLVIII.Fig.3. Stuart also gives an illustration (Plate VII. Figs. 4
and 5). In his text he says the bust was found at Auchentoshan, near
Duntocher (2nd ed. p. 295), a statement obviously inspired by the
descriptive title employed in the Monumenta Imperil Romani : see Tit.
Hunt. p. 92. There seerns never to have been more than one bust in the
Museum, so that the title in the Monumenta is doubtless a mere mistake.
It is significant that the plate on which it is figured is No. XXXI. This
belongs to the second series, with which Anderson had apparently
nothing to do ; see supra, p. 171, footnote.

4 These weights correspond to about 50^ and 37 Ibs. avoirdupois
respectively. If Anderson's estimate of loss is correct, the larger
must originally have weighed 56 or 57 Ibs.

1 86 THE FORTS

battered.1 The little impression which the rust has
made upon them is surprising, but the smallness of the
holes in which the handles have been fixed is more so.
The workmen say that when they found them there
were the marks of wood in the holes : but in this they
might be mistaken ; for when iron is rusted it becomes
foliaceous, so that it might be mistaken for corrupted
wood. If the handles were really of wood, they could
not have been used as mallets are at present, because
of the weight of the head, and the smallness of the
handles : if the handles were of iron, they would even
in that case, without a wooden covering, have been
very inconvenient. Perhaps they were wrought by a
machine, and were used for quarrying stones." All
this is fully borne out by Anderson's illustration,2 and
also by the comment of Dr. John Buchanan, who seems
to have seen them : " These hammers have their faces
greatly battered, and appear to have seen much service.
From their great weight and general appearance, it
appears probable that they were wrought by a machine
similar to that employed at the present day for driving
piles."3 However that may be, it is clear that they
would have completely dwarfed the largest of the
collection of tools from Newstead ; the heaviest of
the Newstead hammers weighs only about 10 Ibs.
Yet another object from Auchendavy calls for passing

1 In his MS. (see supra, p. 170) Anderson says : " It is plain from viewing
them that this loss has been effected by the violence of strokes, which at
first sight made me suspect that they are entirely of cast-iron, but
a thorough examination makes it certain that they are entirely of
hammered iron."

2 Mi fit. Antiq. PI. XXXVIII.

3 Stuart's Caled. Rom. 2nd ed. p. 329, footnote. Evidently they were
still in existence in 1852. Recent search for them has proved fruitless.

AUCHENDAVY AND BAR HILL 187

mention, — a small intaglio, apparently of lapis lazuli,
which had dropped from a signet ring.1 Such finds
are not so uncommon as Stuart and Buchanan seem
to have supposed.

The fort on the Bar Hill, the ninth of the Limes
stations, stood, as nearly as may be, two miles to
the east of Auchendavy. In 1902 and following years
the site was opened up and carefully examined at the
expense of the proprietor, Mr. Alexander Whitelaw of
Gartshore, who has laid all who care for our ancient
monuments under a deep obligation through his
enlightened liberality. As the particulars are easily
accessible elsewhere,2 a very short summary will be
sufficient for our purpose. And first as to the situa-
tion. The fort is the highest on the whole line, lying
as it does 495 feet above the level of the sea.
Towards the west the view is uninterrupted, and the
eye can travel over Auchendavy and Kirkintilloch as
far as the Castle Hill on the other side of East Kil-
patrick ; on a clear day the belt of trees that encircles
the top can be made out quite distinctly. Towards
the south-west the Firth of Clyde and the lofty hills
of Arran are plainly visible. Turning to the east, one
finds that other heights — the Bar Hill Wood (51 1 feet)
and the Castle Hill (507 feet) — intervene to check the
prospect somewhat. The summit of the latter can,
however, be gained in a few minutes, and offers
an extensive outlook in almost every direction. East-
wards the line of the Ditch is strongly marked as it
descends the shoulder of Croy Hill, over whose top

1 Caled. Rom. 2nd ed. p. 328. The intaglio is now in the National Museum.

2 Macdonald and Park, The Roman Forts on the Bar Hill, 1906
(Proc. Soc. of Aniiq. of Scotland, 1906, pp. 403 ff.).

1 88 THE FORTS

peep the chimneys of the cottage known as ' Croyhill
Houses.' Beyond is the site of Westerwood ; and
then the long range of undulating ground from which
the Limes and its stations once looked down on the
Carse of Falkirk and the flats of Grangemouth. In
an exceptionally brilliant day one may even catch a
glimpse of the glistening waters of the upper reaches
of the Forth.

Unlike all of the other Limes stations of which
anything definite is known, the fort on the Bar Hill lay
clear to the south of the Antonine Rampart ; there was
no direct connection between them (see Plate XXII.).1
The reason for this difference was revealed by Mr.
Whitelaw's excavations, which confirmed a conjecture
hazarded long before by Roy. The site — an admirable
one — was originally chosen by Agricola, and the troops
of Lollius Urbicus found it convenient to reoccupy it.
It may be that the existence of an excellent well was
the main factor that determined their decision. At all
events, when the ground was trenched, it turned out
that an earlier outpost, traceable only by the line of its
filled-up ditches, nestled within the area of the fort
whose remains had been apparent on the surface.2
The older fort (Figure 2) was of the conventional
oblong shape with rounded corners.3 The major axis
had a total length of 191 feet, measured over the ditch
at either side. The minor axis, similarly measured, had

1 This Plan should be referred to throughout the description that
follows.

2 See Roman Forts on the Bar Hill, pp. 1 1 ff.

3 At the points indicated on the plan (Fig. 2) by letters, the dimensions
of the ditch were as follows : at A, 9 ft. wide, 4 ft. 3 ins. deep ; at B,
8 ft. wide, 4 ft. 2 ins. deep ; at C, 8 ft. wide, 3 ft. 7 ins. deep ; and at D,
1 1 ft. wide, 4 ft. 6 ins. deep.

BAR HILL

189

a length of 160 feet. After deduction for the breadth
of the defences, this gives an interior area of scarcely
half an acre. The size seems small. Such smallness,

FIG. 2. — THE EARLY FORT AT BAR HILL

however, is not unusual in work of the Domitianic
period.

The castellum had had but a single gate, and the
rampart had apparently been of earth. Besides the
ditch by which it was surrounded, there were others
outside of it — on north, south, and east — ingeniously
devised so as to render difficult any approach to the
entrance. To the west was a small annexe, indicating

190 THE FORTS

that the fort had been constructed for permanent
occupation. But the period during which it was
held seems to have been very short. Had it been
otherwise, broken pottery and similar debris would
inevitably have gathered in the ditches. As it was,
they yielded practically nothing of so substantial a
character. Yet their silent testimony was of real
historical value. It told us that, when the builders
of the second and larger fort arrived upon the scene,
the site had long been abandoned. Lollius Urbicus
and his men found the ditches silted up to a depth of
about two feet. Above that they were still open, but
their sides were overgrown with brushwood, mainly
whin and hazel. The whole is a strange commentary
on the boast which Tacitus makes Agricola address to
his soldiers : " Finem Britanniae nonfama nee rumore,
sed castris et armis tenemus ; inventa Britannia et
sub act a." ^

The later or Antonine fort (Plate XXII.) was fully six
times as large as its predecessor. It was more nearly
square in shape, but had the usual rounded corners.
With a major axis of 375 feet and a minor axis of
369, the area of the interior was just over three acres,
probably designed for the comfortable accommodation
of a single auxiliary cohort. As already stated, it lay
some distance to the south of the Vallum ; opposite
the north gate of the fort the interval was 120 feet.
The Military Way, which was in excellent preserva-
tion, ran between.2 The rampart, which constituted
the principal defence, was built on precisely the same
plan as the great Rampart of the Limes. That is, it
consisted of a wall of turf resting upon a foundation
1 Agricola^ c. 33. 2 See supra, p. 103, and Plate XVII. I.

Pl.ATK XVII

I. THE MILITARY WAV, N. OF BAR HILL FORT, LOOKING W.

2. N.E. CORNER OF BAR HILL FORT

BAR HILL 191

of stone. The stone foundation had a uniform breadth
of 12 feet, and was formed of two parallel kerbs of
dressed stones with a mass of rubble between. Special
care had been bestowed upon the rounded corners (see
Plate XVI I. 2). At each of them the stones were larger
and the rubble better laid, as if the superstructure were
intended to be heavier, — in all likelihood an indication
that the angles of the enclosure were fortified with
towers, in accordance with the usual Roman practice.
In the present instance they were probably of wood,
like those described by Arrian on the rampart of the
fort on the Phasis.1 One of the principal objects of
such towers was to serve for the mounting of artillery.
No other trace of their existence was observed, if we
except the numerous ballista balls found scattered
throughout the camp. Here and there a considerable
portion of the original turf wall was still standing
above the stone foundation, instantly recognizable
from the familiar dark lines pencilled across its face
(see Plate XVIII.).2 The dark lines, or carbonized

1 See supra, p. 83.

2 In the 2nd ed. of Stuart's Caled. Rom. (pp. 338 f. footnote) a most
circumstantial account is given of the destruction, in 1809, of a thick
stone wall which is said to have "surrounded the Barhill fort." In the
Roman Forts on the Bar Hill (p. 6, footnote) it was suggested that the
original narrator had confused Bar Hill with Castlecary. This suggestion
can be now confirmed by a reference to the Appendix to the 2nd ed.
of Nimmo's History of Stirlingshire, published in 1817 (vol. ii. p. 636).
The editor tells how he had gone to see Castlecary for himself, his hopes
roused to the highest pitch by Gordon's description of a " magnificent
fort . . . surrounded with a wall of hewn freestone, whose vestiges still
appear." Knowing that the proprietor (Lord Dundas) had been chosen
President of the Society of Antiquaries of Scotland, he expected to find
things as Gordon and Roy had described them. "Guess, then, his surprize,
when, having, last April, by the directions of some labourers, crossed a
few ridges sown with oats, he had arrived at a newly made inclosure of
ditch and hedge, about 12 paces long and 6 broad, planted with forest

192 THE FORTS

strata, were generally about half an inch thick, and they
occurred at intervals of from four to six inches. These
dimensions suggest that the layers of turf had been
placed grass to grass, a plan not uncommon to this
day in the construction of turf fences.1

The fort had the normal four gateways. Their posi-
tion showed that the line of the Via Principalis, or
street passing in front of the Principia, was some 76 feet
nearer the Porta Praetoria than the Porta Decumana.
On three sides — north, east, and west — the gates had
been flanked by wooden towers raised on the top of
the rampart, and these wooden towers appeared to
have been connected by a wooden gangway, passing
over the top of the gate and supported on either side
by stout posts of oak. The entrance on the south—
the Porta Decumana — had apparently been more
elaborately fortified. Traces of stone foundations, just
within the fort, on each side of the gateway, suggested
that here there had been guard-chambers, built of
strong masonry. Evidently the south was the side that
was felt to be most vulnerable. The ground stretching
in front of it offered special advantages to an enemy.
Another proof that the engineer who designed the
fortifications was alive to this weakness is furnished by
the ditches. These were double on every face except
the north, which could of course rely on the extra
protection afforded by the Rampart and Ditch of the
Limes. On the south they were more formidable than
anywhere else, the outer one being as much as 17 feet
wide and 8J feet deep.

trees, and discovered that the august Roman fortress which had sur-
rounded this pitiful patch of ground had been levelled so as to be no
longer visible ! "

1 See Roman Forts on the Bar Hill, p. 21.

BAR HILL 193

The ditches were all cut upon a uniform general
plan. On leaving the surface, scarp and counterscarp
sloped inwards as if destined to meet and form a V.
The initial angle of descent ranged from 30° to 40°.
But the actual meeting never took place. About 18
inches above the lowest level the two sides suddenly
became perpendicular, the result being to provide a
flat bottom, sometimes as much as two feet broad,
sometimes no more than 8 inches. We have already
seen that there is some reason to think that the great
Ditch of the Limes presents a similar peculiarity.1
Such a device would aggravate the difficulty of getting
out of a deep trench, a circumstance that may well
account for its adoption. One other point may be
noted as evidence of the apprehensions that were
entertained as to the comparative insecurity of the
south front. On that side the entrance was completely
masked by a short ditch or titulus lying immediately
in front of it, some 30 feet long, 12 feet broad, and
7 feet deep — an effective check to the force of a direct
charge. Similar care was called for on the east, and
there too a special measure was taken to cope with
the danger. Beginning opposite the gateway, 25 feet
from the outer ditch, a covering ditch — much larger
than a mere titulus — ran parallel to the main ditches
for a distance of 93 feet towards the south. It occupied
a crest directly above an expanse of dead ground, so
that a storming party emerging from the hollow would
have found themselves confronted by an obstacle not
less than 14 feet wide and 6 feet deep.

In the light of the description just given, it will not
be hard to conjure up a picture of what the outward

1 See supra, p. lot, and Figure I.

N

I94

THE FORTS

appearance of the Antonine fort must have been while
it was entire. That picture would be typical of most
of the other stations on the Limes. Turning to the
interior, we shall find the results of the excavations

VIA PRINCIPAL-IS

FEET

10 0 10.

f«4*l 1-

30.

50FEET

FIG. 3.— THE PRINCIPIA AT BAR HILL

almost equally suggestive. The Principia (Figure 3)
occupied the usual position in the centre of the fort,
and faced north.1 It had an outside measurement of
83 feet long by 77 feet broad, and consisted of three
main divisions. The most southerly, or innermost, of

1 For full description see Roman Forts on the Bar Hill, pp. 35 ff.

BAR HILL 195

these contained three separate chambers, the central
one of which (No. 5) must have been the sacellum or
shrine, where the standards were kept — the sanctuary
consecrated to their worship and to that of the Imperial
House.1 Whether the two larger chambers at the
sides (Nos. 4 and 6) had been divided by wooden
partitions, so as to make a row of five in all, there
was no evidence to show. The middle division was
remarkable for a peculiarity that does not occur in the
corresponding building elsewhere ; the eastern end
seemed to have been a separate room (No. 3), about
22 feet square, and paved with flags. In one corner
of this room there had been a subsidence above the
line of the ditch of the Agricolan fort.2

The general character of the front or northern
division of the Principia (No. i) was not difficult to
determine. It had been an open courtyard, about
70 feet by 34 feet, apparently floored with clay and
a stratum of small stones. In its eastern half was the
well, whose discovery and clearance supplied the most
exciting episode of the excavations.3 This well was
43 feet deep and 4 feet in diameter, and was ' cradled '
all the way down with dressed stones. The lowest
course of the 'cradling' rested on five well-squared
oaken beams arranged in the form of a pentagon.
When the fort was abandoned, the water supply had
been effectually cut off by choking the well with a mass

1 See supra, pp. 79 f.

2 The track of this ditch is marked by dotted lines in Fig. 3. It may
be mentioned that the 6-foot projection shown on the plan at the N.E.
corner was not a buttress ; the slope towards the E. was steep here, and a
line of large stones had been laid down to protect the gutter and roadway
beneath them.

a See Roman Forts on the Bar Hill, pp. 8 f.

196 THE FORTS

of the most miscellaneous material,1 including, besides
many building-stones, a large altar bearing an inscrip-
tion, three considerable fragments of an inscribed slab,
and no fewer than 2 1 freestone columns, or portions of
columns, of a total length of 64 feet and an average
diameter of from 10 to 13 J inches. A view of the
columns is given in Plate XIX. There were also many
bases andr capitals. These evidences of architectural
pretension were exceedingly interesting. Considerable
variety was observable among them. One of the
capitals, for instance, was carved with upright leaves
in the bell (Plate XX. 2). Two others were of cubical
shape, and showed a neatly-cut chevron ornament
(Figures 4 and 5). The rest were plain (Figures 6
and 7). In spite of such minor differences, there was
no shadow of doubt as to where all came from. They
had been torn down from the colonnade by which the
courtyard of the Principia was surrounded.

Leaving the courtyard by the main entrance, which
was of course upon the north, and turning to the right,
one would have found oneself, after a few steps, directly
in front of the Storehouse.2 At Bar Hill the side walls
of this building were less thick than is usual, nor were
they supported by the buttresses which are generally
associated with such structures. The whole, too, was
broader than might have been expected from its length.
A stone partition which divided it longitudinally is
probably the key to the peculiarities just enumerated.
The partition could be so utilized as to relieve the side
walls of much of the pressure of the heavy roof with
which we must suppose the granary to have been

1 For a complete list see Roman Foris on Ihe Bar Hill, pp. 133 f.

2 See Roman Forts on the Bar Hill, pp. 41 flf.

Pl.ATK XIX

PILLARS FROM THE PRINCIPIA AT BAR HILL

BAR HILL

197

provided, and buttresses would thus be rendered
unnecessary. Of the other stone buildings which

FIG. 4.

FIG. 5.

FIG. 6.

FIG. 7.

would naturally be looked for, only the Workshops
and the Baths had left any signs of their presence.
The former were reduced to a few indeterminate
foundations. The Baths, to which were attached

i98 THE FORTS

the Latrines, were inside the fort, thus adding another
to the exceptional features that characterized Bar Hill.
We have seen that, as a rule, such buildings lay outside
the fort in one of the annexes.1 Here they were placed
immediately on the inner side of the northern rampart,
to the west of the north gate. Their condition was
too ruinous to admit of the plan being satisfactorily
recovered. But it was clear that two of the apart-
ments had been heated by a hypocaust arrangement.
These must have been the tepidarium and the cald-
arium. The tank where the water was stored was
placed on higher ground a little to the south, and the
arrangements for drainage were careful and complete.
The finding of coins and of fragments of vessels that
had once held unguents showed that the building was
a place where money had changed hands, and that at
least some part of it was devoted to purposes of toilet.
The outstanding features of the rest of the area of the
fort were the barracks of the soldiery.2 As has already
been explained, these were long narrow buildings,
divided by partitions into a series of small apartments,
each devoted to a contubernium or small group of
soldiers who slept and messed together.3 At Bar
Hill, in accordance with the most usual custom, they
lay parallel to the Via Principalis. As at Ardoch,4 and
perhaps at New Kilpatrick,5 they were of wood, and
had doubtless been roofed with thatch. The buildings
themselves had, of course, disappeared entirely. But

1 See supra, pp. 81 f. The possibility of there having been baths
outside as well as inside must not, of course, be lost sight of ; see footnote
on p. 81.

2 See Roman Forts on the Bar Hill, pp. 51 ff. 3 See supra, pp. 81.
4 Proc. Soc. of Antiq. of Scotland, 1898, pp. 443 ff. 5 See supra, p. 164.

PLATE XX

I. BASK OK PILLAR
KKOM CASTLE HILL

2. CAPITAL
KKOM BAR HI LI.

3. INSCRIBED SLAB FROM BAR HILL

BAR HILL 199

in one or two cases their outline could be more or less
completely traced by following the rows of holes in
which had stood the oaken posts that had supported
them. Close upon a hundred and fifty such holes were
recorded, and in nearly every one of them was found
the end of a round oaken post, which had been carefully
wedged in position with stones. The stumps had
usually a charred appearance on the top, as if the
original posts had been destroyed by fire — a significant
indication of the probable fate of the fort. In all
likelihood the dwelling-houses in the annexe or annexes
that must have lain outside the main ramparts, were
also of wood. The exploration of these annexes is,
however, a task that still remains to be accomplished.
Their whereabouts may be conjectured with some
confidence. There were many traces of inhabitation
in the ground that lies to the east, including the slopes
of the Castle Hill,1 — indications of fire-places, sherds
of pottery, and the like — while quite recently a ditch,
which was obviously part of the enclosure of an annexe,
has come to light upon the south.2 The faint indica-
tion of yet another ditch is visible, some distance
outside the fort, on the west.

The mass of relics recovered in the course of Mr.
Whitelaw's excavations was of unusual extent and
interest. Pottery of various kinds, red roofing-tiles,
window glass and fragments of glass bottles, rude
pieces of sculpture, querns, troughs and mortars of
stone all helped to add to the vividness with which
the original environment of the garrison could be
realized. Weapons were comparatively scarce. Tools

1 Roman Forts on the Bar Hill, p. 132.

2 Information from Mr. John M'Intosh, who opened it up.

200 THE FORTS

were more common, and to them one may here append
the "large Iron Shovel of a vast weight" whose
finding is chronicled by Sibbald.1 Objects of wood
were singularly well preserved.2 The upper portion
of the overhead beam of the Well, with cleft to admit
the pulley, and even two pieces of the pulley itself,
were clearly recognizable. Then there were the oaken
staves of barrels, a wooden bung, a wooden bobbin,
and some stray pieces of wooden piping. More
remarkable than any of these was a very fine wheel,
with an outside diameter of 2 feet icj inches. The
felloe, which was of ash, had been formed of a single
piece of wood, artificially softened and then bent into
a circle. The spokes, eleven in number, were beauti-
fully turned with the lathe and were made of willow,
while the nave appeared to be of elm. The whole was
firmly bound together by the iron ring that formed
the tire.

Some of the finds had about them a more intimate
personal touch. Inscribed stones registered the name
of the cohort that had occupied the station as a
permanent garrison (Plate XX. 3). Shoes that had
been worn by men, women, and children (Plate XXI.)
spoke of the individual soldiers and of the civilian
population that grew up outside the main defences.
Simple ornaments like beads showed that they had
had weaknesses much like our own. Specimens of the
money they had used were fairly frequent. Lastly, the
refuse they had left behind indicated comfortable feed-
ing. The skulls and bones of oxen, including the
Celtic shorthorn, occurred in large numbers, and shell-
fish were evidently a favourite dainty — mussels as a

1 Hist. Ing. p. 29. 2 Roman Forts on the Bar Hill, pp. 91 ff.

PLATK XXI

i. MAN'S, WOMAN'S, AND CHILD'S SHOE

2. UPPER PORTION OK LADY'S SHOE

SHOES FROM BAR HILL

BAR HILL 201

rule, oysters as an occasional luxury. The life of the
denizens of this little Roman outpost had been one of
hard work and hearty feeding. It may be regarded
as typical of many similar stations, not merely in
Caledonia, but all round the fringes of the Empire.

A large number of the relics came from the well.
The majority of the rest were taken either from the
ditches or from rubbish-pits. It is worth observ-
ing that, in the case of the ditches, by far the most
prolific spots were the outermost corners. The reason
is easily guessed ; it was only natural that it should be
the parts furthest from the gates that were chosen
as receptacles for cast-off odds and ends. The pits
deserve a word of notice. So far as Scotland is
concerned, it was at Bar Hill that the possibilities of
such hiding-places were first made manifest. The
wonderful results achieved at Newstead make it
certain that, wherever they are found in future, they
will be most carefully searched. What it is desirable
to emphasize now is that they are sure to be found,
if they are properly looked for. We have already had
evidence of their existence at New Kilpatrick, at
Balmuildy, and at Auchendavy. We shall meet with
similar evidence again at Castlecary. And it may not
be amiss to direct attention here to two other Scottish
examples that are sometimes overlooked. In 1850,
when the railway from Grahamston to Larbert was
under construction, a cutting was driven through a
series of pits that can only have belonged to the fort
at Camelon.1 Again, in the latter half of the eighteenth
century the river Almond, near the point where it joins
the Tay, encroached upon its northern bank to such

1 Stuart's Caled. Rom, 2nd ed. p. 357, footnote.

202 THE FORTS

an extent as to expose what were described as eight
semi-circular pillars of earth, filled with a dark-coloured
mould which contrasted strongly with the reddish soil
of the bank itself.1 What we are told regarding the
relics that lay at the base of these pillars makes it plain
that the pillars were really pits of the type we have
learned to know at Bar Hill and at Newstead. They
furnish incontrovertible proof of the proximity of a
Roman encampment.

1 Cant's edition of Adamson's The Muses" Threnodie (Perth, 1774),
pp. 25 f.

CHAPTER VII

THE FORTS: FROM CROY HILL TO THE FORTH

THE relative completeness of the knowledge available
regarding the fort at Bar Hill has rendered it possible
to give a somewhat detailed sketch. Matters are very
different with the tenth station, the site of which seems
to have been Croy Hill, rather less than two miles
further east. Pont places a fort there,1 and so does
Sibbald, quoting Irvine.2 Gordon speaks as if indefi-
nite traces had been visible in his day ; he mentions
the " few Houses called the Croe-hill Town, at which
Place has been another Fort, but now demolished ; so
that I could not take an exact Draught of it."3
Horsley apparently saw nothing of it, and Roy ex-
pressly says that it was "totally levelled." Both of
these acute observers, however, agree that it is " highly
probable that here a station formerly stood."4 Mait-
land, however, "after the strictest search and inquiry,
could not discover the least vestige of [it], nor learn
that there ever was a fort at or in the neighbourhood
of the said Crowyhill."6 The weight of this opinion

1 Blaeu's Atlas (Amsterdam, 1654), vol. v. p. 4.
*Hist. Ing. p. 29. * Itin. Sept. p. 56.

*Milit. Antiq. p. 1 60. Cf. Brit. Rom. p. 170.
6 Hist, of Scot. i. p. 175.

204 THE FORTS

is somewhat discounted by the passage that follows :
" I must confess, that as there is a rock in the ditch at
this place, about the length of twenty feet, whereby it is
levelled, and serves at present as it were for a bridge
to cross the said ditch by,1 I think that no place in the
course of the wall had more occasion for a fort than
this." It is tempting to quote the sequel as an illustra-
tion of the straits to which a man may be reduced if
he is to maintain his independence. Maitland argues
that " it cannot be reasonably imagined that [the
Romans] would leave a rock undemolished in this
part." And he proceeds : " Now as I am, for certain
reasons, (too long to be inserted in this place) of
opinion that rocks vegetate, the rock here, by its
form, must have sprung up since the making of the
said ditch, which is the only mean I can think of, to
secure the wall at this place without a fort."2

This attractive theory notwithstanding, the evidence
for a fort at Croy Hill is overwhelming. Its probable
position has already been indicated,8 and the walls of the
ruined hamlet hard by are full of stones, whose shape is
enough to prove that they were hewn by Roman work-
men. " Some few of them," to quote from the Antonine
Wall Report, "are tooled with various well-known
patterns of Roman broaching — the reticulated, the
feathered, and in one fine example the diamond."4
Typical specimens are grouped in Plate LI 1 1., which
also shows a ballista ball and a portion of a quern met
with in the neighbourhood when the sections were
being cut for the Glasgow Archaeological Society's
Committee. At the same time some indefinite traces

1 See supra, p. 126. 2 Hist, of Scot. i. p. 176.

3 See supra, pp. 125 f. 4 Op. cit. p 60.

CROY HILL AND WESTERWOOD 205

of foundations were uncovered.1 Again, about the
beginning of the nineteenth century a votive altar,
dedicated to the Nymphs by a vexillation of the
Sixth Legion, as well as two pieces of sculpture — one
of them evidently a portion of an inscribed slab, —
were found lying at the foot of the precipitous ridge
on the summit of which the fort is supposed to have
stood.2 To these must be added the "few small
Stones, with Inscriptions," of which Gordon speaks.
One of the two which he figures was actually found
by himself " built up in a Cottage, Part of it being
covered over with Clay."3 Finally, the Report of the
Glasgow Committee mentions that, on the occasion of
one of their visits, a small fragment of ' Samian ' ware
was picked up upon the surface.4

The eleventh station — that of Westerwood — was
rather more than 3000 yards to the east of Croy Hill.
Gordon says that here there " are to be seen the
distinct Vestiges of another great Roman square Fort
upon the Wall, where the Praetorium, in the Middle
of the Area, is very plain and distinct, and what is
very peculiar here, is the Causeway which goes round
this Fort, on the Top of the Ramparts." 6 It is difficult
to understand what the last statement can possibly
mean, and the Plan (to which Gordon refers his
readers) throws little light upon the puzzle. It may

1 Information from Mr. Alex. Park, who was in charge of the
operations and who noted "a considerable amount of rough paving
and macadamizing."

2 Stuart's Caled. Rom. 2nd ed. p. 340, footnote, where other relics are
mentioned. The sculptures will be described in chap. xi.

*Itin. Sept. p. 56. See infra, chap. ix. No. 23, for a description ; it is
now in the National Museum.

4 Antonine Wall Report, p. 62. 6 Itin. Sept. p. 56.

206 THE FORTS

perhaps be suggested that he was misled by glimpses
of the stone foundation which must have run right
round the fort, if the rampart was of turf as we should
expect it to be. At all events, Horsley and Roy are
convinced that he was mistaken.1 Maitland, strange
to say, agrees with him.2 The fort was not a large
one. Roy's plan shows an interior measurement of
about 360 feet by 295, the major axis being at right
angles to the great Rampart, instead of parallel to it
as in all the cases we have met with up till now.

Stuart writes as if the remains had almost disappeared
in his day.3 The outline of the ditches is, however,
still tolerably distinct, and the revised Ordnance Survey
Map provides a plan very nearly as complete as Roy's.
The main difference between them is that the major
axis is much shorter, the Survey officers making it only
about 320 as against Roy's 360. Very few antiquities
are known to have been found at Westerwood. Gordon
tells us that " several Inscriptions have been dug up
. . . some of which are broke, and built up within
their Houses."4 He mentions only one, which he
carried off and presented to Sir John Clerk. It seems
to be now lost. But his general assertion is corro-
borated by Maitland, who saw "stones with letters
thereon, but so defaced that I could make nothing of
them."5 On the whole, it looks as if the surface had

1 See Brit. Rom. p. 170, where the Military Way is said to pass " close
by the wall, on the north side of the fort," and the plan in the Milit.
Antiq., where it forms the Via Principals.

2 Hist, of Scot. i. p. 175. Maitland's statement, however, is at least
intelligible: "the military way runs round the [ditch] in a different
position from what it does at most of the other forts."

3 Caled. Rom. 2nd ed. p. 343. * Itin. Sept. p. 56.
6 Hist, of Scot. i. p. 175.

WESTERWOOD AND CASTLECARY 207

not been much disturbed, except indeed in the north-
east corner, which is occupied by the modern farmhouse
and its steading. When the foundations of these
buildings were being dug, a large quantity of burned
or blackened wheat was discovered.1

Castlecary, the twelfth of the Limes stations, lay
close upon two miles east of Westerwood, immediately
beyond the little glen through which the Red Burn
makes its way towards the Bonny Water. During the
seventeenth and the first half of the eighteenth century
we hear of the ruins as being very extensive. Since
then, violent hands have more than once been laid
upon them. In 1769 and 1771 the workmen engaged
on the Forth and Clyde Canal raided them freely in
order to secure an easy supply of building materials,
and their example was followed by others. Writing
in 1772, Professor Anderson speaks of the "wall of
hewn freestone . . . which the country people are
just now pulling down for building walls and houses." 2
Thirty or forty years later, in or about 1809, a large
part of the stone wall surrounding the fort was deli-
berately demolished and its foundations rooted out,
gunpowder being employed to destroy them. This
was in the name of agricultural improvement. It was
then that the little patch of trees that still stands near
the centre was planted.3 Once more, in 1841, the line
of the Edinburgh and Glasgow Railway was carried
diagonally across the fort, entering beside the west gate,
and sweeping everything remorselessly before it until

1 Information from Mr. Alexander Park.

2 MS. Account of the Wall, in the Glasgow and West of Scotland
Technical College Library : see supra, p. 170.

3 See supra, p. 191 f. footnote, and cf. Stuart's Caled. Rom. 2nd ed.
pp. 338 f. footnote.

208 THE FORTS

it passed out at the south-eastern angle. Destructive
operations of a minor character must also have been
indulged in when the present high road to Stirling was
being constructed, so that a great deal of damage had
been done before 1902, when systematic excavation
was begun under the direction of the Society of Anti-
quaries of Scotland.1 Digging had unfortunately to
be restricted within the limits of a single season, and
the results were thus less fruitful than more time and
greater thoroughness might have made them. Very
interesting information was certainly obtained. But
in regard to some highly important questions — the
internal arrangement of the Principia, for instance —
the necessary evidence is still to seek. And who can
tell what may be resting at the bottom of unopened
pits ?

The fort proved to be of the usual oblong shape, with
major axis parallel to the Limes; see Plate XXVII.,
which should be consulted throughout the following
description. The interior measurements were 455 feet
by 350 feet, giving the whole enclosure an area of not
less than 3^ acres. The principal defence consisted
of a stone wall of exceedingly substantial construction,
rising on a bottoming of boulders 9 feet wide. The
average breadth of the foundation course was about
8 feet, and the stones of which it was composed were
admirably dressed and laid. At the north-east corner,
where the breadth expanded to about 1 1 feet, they
presented a most massive appearance ; some of them
must have weighed more than half a ton (see Plate

1 An account of the results will be found in the Proceedings for 1903
(pp. 271-346). It is from this that the facts set forth in the following
pages are taken.

PLATE XXIII

DRAIN THROUGH N. WALL OF CASTLECARY FORT

CASTLECARY 209

XXIII.). While this lowest course was of solid
masonry, the rest of the wall was built upon a different
principle. It was merely faced with stone on its outer
and inner sides, " the interior being a solid block of
concrete composed of sandstone chips and rough
boulders of various sizes, all run together with lime."1
The general structure is well shown in Plate XXIV.
Scarcements at the foundation course reduced the width
of the superstructure to 6J feet. Its height can only
be guessed at, but it can hardly have been less than
jo feet. A very remarkable feature was noted on the
north side. The outer edge of the foundation was in
alignment with the outer kerb of the great Rampart
of the Limes, while within the fort was an apparently
independent stone kerb (Plate XXV. i), running from
east to west, parallel to the inner edge of the foundation
but at a distance of 6 feet from it, and therefore exactly
in alignment with the inner kerb of the Rampart. The
excavators have no satisfactory explanation to offer
of this puzzling apparition. The intervening space is
not nearly wide enough for a supporting mound of
earth such as is sometimes found behind stone walls.
Is it conceivable that the stone wall represents a
secondary stage in the evolution of the Antonine fort,
and that originally the northern defence had been, as
elsewhere, the great Rampart itself? It is true that
no other evidence to that effect was forthcoming. At
the same time it must be remembered that none was
looked for, the question never having been before the
excavators, and it is not clear that any systematic
search was made for similar kerbs on the remaining
three sides of the fort.2 The point is one to which

1 Proc. Soc. of Antiq. of Scot. 1903, p. 289. 2 Ibid. p. 291.

O

210 THE FORTS

attention might usefully be directed if the site were
ever opened up again.

The gates, of which there were four, were con-
structed on a uniform plan. On either side of each
entrance the wall returned for a distance of 14 feet,
reckoned from the outer margin of the foundation.
These returns undoubtedly implied the adoption of
special means of protection. But nothing definite was
ascertained as to guard-chambers or as to the precise
nature of the gates. The indications perhaps point to
an archway, surmounted by a tower. It should be
mentioned that the Portae Principales were exactly
twice as far from the Porta Decumana as they were
from the Porta Praetoria, 230 as against 115 feet.
There was only one of the corners of the station, that
on the south-west, which had not been completely
destroyed. It was rounded, and within it were dis-
covered the foundations of a rectangular stone tower.
We may infer that there were similar towers at all four
corners, and that the south-east angle was also rounded.

A double ditch ran outside the wall on three sides —
west, south, and east, — while a third trench was added
in the interval between the east gate and the south-
east corner. On the north side a curious variation was
revealed. One would have expected to find the Ditch
of the Limes serving as the fossa on this front. And
so it did for that part of the fort which lay to the east
of the Porta Praetoria. To the west, however, its
place was taken by a much narrower ditch, which was
not in direct alignment with it, but which nevertheless
seems to have been substituted for it all the way down
to the bank of the stream. The particulars regarding
this narrower ditch, as they appear upon the plan

CASTLECARY 211

(Plate XXVII.), are to a large extent conjectural, much
of it being buried under the modern road. But, on
the assumption that they are approximately correct, the
phenomenon of a sudden shrinkage in breadth from
40 feet to 15 feet is certainly extraordinary. Is it to be
explained by supposing that at this point the engineers
of the Limes were content to utilize a ditch that they
found in existence and that happened to coincide with
the line they had themselves marked out ? If so, then
Agricola had a fort at Castlecary as well as on the
Bar Hill, — a conclusion that the pottery supports.1

As already indicated, the evidence of the excavations
was not very satisfactory, so far as the interior buildings
were concerned. Incomplete as it is, the plan (Figure
8) suggests that the Principia had been of the
normal type, — that is, that it had consisted of an outer
and inner courtyard, both paved, with a row of three
rooms at the end furthest from the entrance.2 The
excavators thought that the most westerly of the three
rooms had been a later extension. Reconstruction,
however, is far more probable than addition. The
dimensions, it may be added, were about 96 feet by 85,
as against 83 by 77 at Bar Hill. Plate XXV. 2 shows
the character of the masonry. The Storehouse, which
lay next the Principia on the east, was 83 feet long and
15 feet wide within the walls. It was in unusually good

1 Professor Haverfield long ago suggested to me that, if one could
assume an original width of 15 feet for the ditch on the east side of
the opening, as well as on the west side, the result would be an
entrance rather resembling an early Newstead type, with the gate at
right angles to the wall, instead of in alignment with it.

2 It is possible that, as at Melandra, the two side rooms were each
divided by wooden partitions, making the row consist of five in all
(see su^ra, p. 79). The dimensions, however, do not favour this view.

212

THE FORTS

Ygjjj'f S

CASTLECARY 213

preservation, and had had twenty buttresses, ten on
each side. Rows of stones laid in regular lines showed
that the floor had been raised, while at intervals there
were splayed openings in the walls to secure proper
ventilation (see Plate XXVI. 2). Here, too, there were
clear signs of structural changes having been effected ;
a large doorway which had originally existed in the
south wall had been completely built up (Plate LV. 2).
The rows of stones on which the floor had rested bore
what looked like traces of the action of fire.

West of the Principia, at a distance of some 40 feet,
there was laid bare the line of a wall which must
represent the side of a building. At a distance of 10
feet to the east of this wall — that is, between it and the
Principia — a row of nine post-holes was discovered.
Whether these had had any connection with the wall
may be doubted. The interval appears to be too wide
for a verandah. And it deserves consideration whether
they may not date from an earlier period, when the
buildings in the fort were of wood. Coming back to
the east of the Principia and passing beyond the
Storehouse, we find the meagre remains of yet another
large building, represented only by the line of one wall
and by fragments of two others, with an apse towards
the south end. It seems certain that this building was
of far greater size than the excavators were disposed
to think. Originally it had in all probability been a
house of the corridor type, with a central courtyard.
If, as may well be, the apse marks the site of the
shrine of the genius praetorii, then the house was the
residence of the commandant.1

Further east still — so far east, indeed, as to be close

1 See supra, p. 80 f.

2I4

THE FORTS

against the inner face of the wall of the fort — the
excavators rediscovered the suite of Baths described
by Professor Anderson in Roy's Military Antiquities?-
They were much struck by the excellence of the build-
ing work displayed here. The accuracy of the plan
given by Roy (Figure 9) was confirmed, and a new

feature added in the
shape of the furnace
of the hypocaust. As
at Duntocher,2 some of
the rooms had had the
apsed recesses, so char-
acteristic of Roman
baths. It will not have
escaped notice that
Castlecary resembles
Bar Hill in having the
Baths inside the fort.
There may have been
a second suite outside.
We cannot tell. But
it may be pointed out
that the presence of a
range of buildings so
near the wall must have constituted a source of
weakness, partly as inviting attack and partly as
hampering the defenders' freedom of movement.
Possibly that is why it was deemed necessary to
cut a third ditch along this portion of the east front.
The waste water from the Baths seems to have been
drained off northwards, and to have been utilized,
as at Bar Hill, for flushing the Latrines. These

1 P. 200. 2See supra, p. 159, footnote.

20 it 10 f

20

FIG. 9.— ROY'S PLAN OF THE BATHS AT
CASTLECARY

PLATE XXV

I. KERB ON INNEK SIDE OF N. WAI.I,

2. INNEK FACE OK S. WAU, OK PKINCIFIA

CASTLECARY FORT

CASTLECARY 215

last lay very near the north-east corner. When
opened up, they proved to be in a better state of
preservation than any other part of the fort. The
most striking thing about them was the drain that
conveyed the sewage through the north wall ; it was
a splendid specimen of Roman masonry (Plate XX I II.).
Only one rubbish-pit was cleared out — a hole behind
the back wall of the Principia ; it was 24 feet deep
with a diameter of 9 feet at the top, and contained
the characteristic de"bris. There must, of course, be
many more. Not long ago a second one gave some
trouble to Post Office employes engaged in erecting
a telegraph post by the side of the road east of the
fort.1 In digging they encountered a mass of soft
black matter which had Roman remains intermingled
with it. This gives a clue to where such pits should
be looked for ; there are usually a number of them
together. And it is just what might have been antici-
pated, for the position indicated is within the limits
of the annexe whose discovery was one of the most
conspicuous results of the excavations of 1902. This
annexe was an irregular pentagon, projecting from the
east front of the station. The enclosure, which covered
an area of about 2f acres, was defended on its exposed
sides by a single ditch, and probably also by a narrow
rampart of turf; digging revealed "a stone foundation,
8 feet wide, following closely the line of the inner edge
of the trenches."2 On the north it was protected by
the Ditch and Rampart of the Limes. No buildings
of any kind seem to have been laid open within its
area.

1 Information from Mr. John M'Intosh.
2Proc. Soc. of Antiq. of Scot. 1903, p. 302.

216 THE FORTS

Although hardly any pits have been examined at
Castlecary, its yield of Roman antiquities has been
considerable. They include a certain number of in-
scribed stones. Some of these have been lost, but
the majority are fortunately in safe keeping, either in
Edinburgh or in Glasgow. We are told that, when
the railway was being made, seventy years ago, great
quantities of pottery were thrown up, including much
' Samian.' l Little or nothing of what then came to
light has survived.2 The objects recovered in 1902,
on the other hand, are in the National Museum, and
have been carefully described in the official report of
the excavations.3 They were interesting, but not in
any way remarkable. The absence of coins was
noteworthy.

Details apart, two general impressions are very dis-
tinctly suggested by a review of the foregoing evidence.
In the first place, Castlecary was a more than usually
important station ; that is indicated by its size and
perhaps also by the fact that it was surrounded by a
wall of stone. Its importance was doubtless due to its
situation. Though rather nearer to the Forth than to
the Clyde, it seems to have been regarded as the
strategical centre of the Limes ; a road issuing from the
Porta Decumana was traced by the excavators for
about 1000 feet to the south.4 In the second place,.

1 Stuart, Caled. Rom. 2nd ed. p. 348.

2 A few odd articles have, however, found their way into the National
Museum of Antiquities.

zProc. Soc. of Antiq. of Scot. 1903, pp. 330 ff.

4 This tended to strengthen the case for the opinion, so firmly held by
the older antiquaries, that there was direct communication between
Castlecary and a road that led from Carlisle, past Netherby and Birrens,.

PLATE XXVI

I. LOOKING NOKTH

2. LOOKING SOUTH KAST

STOREHOUSE AT CASTLECARY

CASTLECARY 217

the structural remains present features which hardly
admit of explanation, except on the hypothesis of
successive phases of occupation. The narrowing of
the Ditch on the north-west front and the isolated
line of post-holes may conceivably carry us back
to Agricola and the first century. The reconstruc-
tion of the Principia and the building up of the
south door of the Storehouse belong to the Antonine
period.

In themselves the latter changes need not imply a
breach of continuity in tenure. But there are clear
signs of alterations so radical in character that we
could associate them only with a reoccupation follow-
ing on a temporary abandonment. Attention has been
drawn above to the possible substitution of a wall of
stone for a rampart of turf. Other indications are
distinctly more definite. The complicated network of
drains in the interior appears to have baffled the
excavators. "No system could be followed entirely,
and there were evident alterations." l The substitution
of one drainage system for another suggests a rebuild-
ing from the foundations, or at all events something
much more extensive than mere repairs. The roads,
too, yield their quota of confirmation. As shown on
the Plan, the Military Way approaching from the east
has been constructed to suit, not the present Via
Principalis, but one running on a line parallel to it at
a distance of 35 feet further north ; instead of entering
the Porta Principalis directly, it is brought right up
against the margin of the outer ditch and has to

to the western end of the Antonine Wall. According to Stuart (Caled.
Rom. 2nd ed. p. 352) the point of junction was near Carluke.

1 Proc Soc. of Antiq. of Scot. 1903, p. 324.

218 THE FORTS

execute two rectangular turns before it can reach the
gate.1 Lastly, there is the "paved road" which is
carried over the most easterly trench of the fort.2
There is no mistaking the meaning of its testimony
(Figure 10). All of these things deserve to be
far more thoroughly probed than was possible in
1902.

For 3^ miles beyond Castlecary there are no
superficial appearances such as one would naturally
associate with a fort. But it would have been strange

FIG. io.— ROAD LAID ACROSS E. DITCH OF CASTLKCARY FORT

to leave so long a stretch of the Limes unprotected.
Gordon fails to take any notice of the gap. Horsley,
on the other hand, speaking of "a village called Dicks
House" — now represented by the group of cottages
near Bonnybridge Canal Station, — and of the "beautiful

1 Is it a mere coincidence that the Military Way, as marked, is in pre-
cise alignment with what would be the natural Via Principalis for a fort
rather more than a fourth of the size of the one we know and occupying
its north-west quarter? Such a fort would have had the long straight
strip of narrow ditch as its northern defence, while the isolated line of
post-holes would represent the side of a barrack-block in the retentura.
The alternative possibility suggested by Professor Haverfield (see supra,
p. 21 1, footnote) would, of course, mean a much larger enclosure — one not
materially different from that of the Antonine fort.

z Proc. Soc. ofAntiq. of Scot. 1903, p. 302.

SEABEGS 219

exploratory mount" beside it,1 adds : " Here are also
some ruins that possibly may be the remains of a
station.2 He admits that the traces are "doubtful
and obscure." Before Roy made his survey, they had
seemingly disappeared. At all events, he saw nothing
of them. At the same time his military instinct con-
vinced him that a fort was wanted somewhere in the
neighbourhood. " Having passed through Sebeg-
wood, [the Wall] comes to the house of that name,
situated on the south brink of the ditch ; and either
here, or at Dick's-house, a little more to the eastward
(which is likewise called Mill-quarter), there hath been
a station. From Castle Gary to Sebeg- House is three
thousand three hundred yards."3 It is not clear whether
Roy was aware that he had been anticipated in his
suggestion of Seabegs as the missing site. According
to Sibbald,4 Dr. Irvine, whose visit may be dated about
1680, observed "a great Fort at the East end of
Seabegwood" Apparently this was not mere con-
jecture ; there was something to be seen there in the
end of the seventeenth century, for Sibbald in his
manuscript, " Directions for his honoured friend Mr.
Llwyd how to trace and remarke the vestiges of the
Roman Wall betwixt Fort and Clyde, " 6 expressly

exact situation of 'Dick's House' is indicated by Roy (Mtlit.
Antiq. p. 163) as being 300 yards west of 'Chapel Hill' — the site of the
' exploratory mount ' — and 440 yards east of Seabegs House.

*Brit. Rom. p. 171. 3 Mt 'lit. Antiq. p. 161. 4 Hist. Inq. p. 30.

5 Now in the Bodleian Library (Carte M.S. 269, Fol. 129 </. etc.). It is
reprinted in the Proc. Soc. of Antiq. of Scot. 1910, pp. 319 ff. Edward
Lhuyd (1660-1709) was a man of considerable note in his day. He was
appointed Keeper of the Ashmolean Museum in 1690, and six or seven
years later he set about collecting subscriptions to enable him to
undertake a prolonged antiquarian and scientific tour. In 1699 he
went to Scotland, and next year to Ireland and Cornwall. His original

220 THE FORTS

draws attention to "a great fort at the East-end of
Seabeggwood." To-day all outward signs of anything
of the sort have vanished. Yet Irvine may neverthe-
less be right.

If there was a station at Seabegs, Rough Castle
would be the fourteenth of the series. Of all the Limes
forts it is the one with which the destroying influences
of time have dealt most gently. It lies on a moor, and
is now covered with trees and with a luxuriant under-
growth of brackens. Its remains have left a deep
impression on a long succession of visitors, from the
anonymous traveller of 1697 l and Alexander Gordon,
the latter of whom speaks of "the distinct Vestiges of
a vast Roman Fort upon the Wall, called Rough Castle,
which for Intireness and Magnificence, exceeds any
that are to be seen on the whole Track from Sea to
Sea,"2 down to the Glasgow Committee, who put "a
thorough exploration of the camp at Rough-castle " in
the forefront among their recommendations.3 In 1903
extensive excavations were carried out by the Society
of Antiquaries of Scotland.4 As at Castlecary, how-
scheme covered five years, but he encountered numerous troubles which
forced him to curtail it greatly. In 1707 he published the first in-
stalment of his results in Archaeologia Britannica, vol. i. The work
was purely philological, and so disappointed the subscribers that it was
never continued. He left a great mass of papers. Those relating to
the Roman Wall in Scotland were utilized by Stukeley when he was
writing his Account of a Roman Temple, etc. They were then (Dec.
1720) in the hands of Sir Thomas Seabright (Stukeley, op. cit. p. 9),
but they seem now to have disappeared. Possibly they formed part
of a large number that were destroyed by fire in the early years of the
nineteenth century.

1 Hist. MSS. Commission: Portland Papers, ii. p. 57.

*Itin. Sept. p. 59. *Antonine Wall Report, p. 149.

4 See Proc. Soc. of Antiq. of Scot. 1905, pp. 442-499. As in the case of

ROUGH CASTLE 221

ever, the time and the means available were hardly
adequate to ensure completeness. This is all the more
to be regretted, because the problems involved were
peculiarly intricate. Their successful solution would
have been of high value for the history of the Limes.

The fort at Rough Castle (Plate XXXIII.) was
exceptionally small, its area being only a little over an
acre. Perched on the top of a slope that rises steeply
from the east bank of the Rowan Tree Burn, it occupied
a position that was immensely strong by nature, — so
strong that its elaborate fortifications come upon one
as something of a surprise. The northern defence
consists of the Rampart and Ditch of the Limes (Plates
XII. and XXVIII.). On the other three sides the
fort is surrounded by a rampart and double ditch of
its own. It is as nearly as possible 223 feet square,
and has had to the east an annexe of slightly less
regular form, enclosing — at its smallest — an area of
about an acre and a half. The annexe, like the fort
to which it was attached, was effectually protected
on the north by the Antonine Vallum. A rampart
ran round it on the south and on the east, the
accompanying ditch being single on the south and
apparently triple on the east. The members of this
triple system of ditches are separated by what the
excavators describe as two great "platforms" as
much as 50 or 60 feet wide. The rampart of the
annexe, which rested on a stone foundation 15 feet
broad, seems to have been of earth. That of the
fort was built of turf. Outside the main defences at
sundry points, notably immediately to the right of

Castlecary, this official report of the excavations has provided the materia
for the discussion that follows.

222 THE FORTS

the entrances, the Plan shows "mounds" whose pre-
cise purpose is not very intelligible.1

The original stone foundation of the rampart of
the fort resembled that of the great Rampart of the
Limes ; but it was somewhat broader, measuring on
the average about 20 feet across. Plate XXVIII.
shows the junction between the two at the north-east
corner. The excavators found what they regarded
as undeniable proof of a subsequent reconstruction,
in the course of which the breadth of the foundation
and also of the rampart itself had been increased to
30 or even to 35 feet, dimensions which would admit
of the ' caespiticious ' superstructure being raised to
a great height. The additional stone " margins " can
be seen in Plate XXIX. Another indication of two
distinct periods in the history of the defences was
afforded by the south ditch of the annexe, which had
at one point been filled up to give a passage for a
road.

Besides these, there are other peculiarities that de-
serve to be carefully noted. To begin with, there is the
seeming absence of even a single ditch in front of that
part of the rampart of the fort that runs between the
east gate and the Vallum. Then there is, in the same
quarter, the quite irregular intrusion of a ditch that
looks as if it had belonged to a different system.
Again, the roadway over the great Ditch had been
widened at the north-east corner.2 Lastly, there is the

1 It might have been desirable to have had the opinion of a geological
expert as to whether some of these, as well as certain other appearances
recorded by the excavators, were not simply natural. The character of
the soil makes a decision difficult.

2 Proc. of Soc. of Antiq. of Scotland, 1905, p. 455.

ROUGH CASTLE 223

wholly unexpected manner in which other ditches are
prolonged beyond what one would have supposed to be
their natural termination. Thus, the inner ditch on
the west side of the fort reappears in an odd way be-
yond the rampart, being cut right across the ' berm '
to the edge of the Ditch. Exactly the same thing
happens with the innermost of the three ditches on the
east of the annexe. The ditch beyond that, again,
is carried through the Military Way, interrupting its
course so completely that a bridge would have been
required, while the south ditch of the annexe, instead
of stopping when it joins the outer ditch of the
fort, is continued straight on till it unites with the
inner one.

Leaving these anomalies for the moment, let us turn
to the buildings in the interior. Their condition was
so dilapidated that little definite information regarding
them was forthcoming. Still, a few points of import-
ance emerged. A long stone building in the annexe,
L-shaped (so far as could be judged), and furnished
with an extensive hypocaust installation (Plate XXX.
2), was probably the Baths (No. 4 on Plate XXXIII.).
According to the excavators, evidence of extensive
alterations and additions was everywhere apparent.1
Their description, indeed, suggests a re-building from
the foundations, and apparently also a second series
of changes. The distinctive differences in the two
styles exhibited in the walling are well seen in
Plate LV. i. Within the fort the Principia was re-
cognizable. As might have been anticipated, it is
small — not more than 75 feet by 44 over the walls.
Its disproportionate length, as well as the fact that

1 Proc. Soc. of Antiq. of Scot. 1905, p. 484.

224 THE FORTS

on the Plan (Figure n, No. i) it seems to have
three courts instead of the normal two, may be the
result of the reconstruction that had obviously taken
place. Or, if the foundations laid down in the plan
are all contemporary, the outer and inner courts may
have been connected by a passage, 16 or 17 feet
long, with a chamber on either side. In a small fort
this would have been a very unusual, if not a unique,
arrangement. The one thing to be said in its favour
is that it would have given the normal number of four
business-rooms within the Principia. As it is, there
seem to be only three chambers at the back, and the
central one of these is, of course, the sacellum.1

Two special features should be noticed. Beneath
the floor of the sacellum there was sunk a small stone-
lined pit or cellar, 4 feet long, 2 feet 3 inches wide, and
2 feet 6 inches deep (Plate XXX. i). The cellar must
have been a tiny strong room or safe ; for it has to be
remembered that the sacellum. besides beingf the shrine

o

of the standards, was also the regimental bank. It
may have been deeper than would appear, since it is
possible that the floor above it was raised, as at New-
stead and elsewhere. At all events, it is of some
significance as an index of date ; such strong-rooms did
not come into general use until the latter half of the
second century. The second feature calling for remark
was the discovery of three fragments of an inscribed
tablet (Plate XXXI. 2), commemorating the erection
of the Principia by the Sixth Cohort of the Nervii.
The inscription is extremely interesting as confirming

1 There is always the possibility — it is no more — that each of the two
side chambers might have been divided by a wooden partition, see supra,
p. 211, footnote.

ROUGH CASTLE

225

226 THE FORTS

the view that the proper name of the building was
' Principia,' not ' Praetorium,' as it used to be called.
And there is another point. The fragments were
found "in a hole among other debris."1 On the
assumption that there were but two courts, the hole
occupied the same position in the outer one as did
a similar hiding-place at Melandra.2 But this was
also the position occupied by the well at other forts,
such as Bar Hill,3 Birrens,4 and Gellygaer.5 That
it was the well at Rough Castle is an inference that
will immediately suggest itself as possible.6

Immediately to the west of the Principia lay the
Storehouse (Figure n, No. 2), a buttressed building
67^ long and 15^ feet wide within the walls. It had
dwarf walls running from end to end of the interior, to
support the floor, and the walls between the buttresses
were pierced with slits for ventilation. On the north
front was a loading platform, which is shown in the
foreground of Plate IV. Beyond the Storehouse were
foundations indicating a building much larger than the
Principia (Figure n, No. 3). Even its outline could
not be properly recovered, and its character and purpose
can therefore only be conjectured. The presence, near
the centre, of a small stone-lined pit, exactly like that
in the sacellum but rather larger and better con-
structed, shows that the whole, whether it had been

1 Proc. Soc. of Antiq, of Scot. 1905, p. 470.

2 Excavations at Melandra (1909), p. 26. 3 See supra, p. 195.
* Proc. Soc. of Antiq. of Scot. 1906, p. 411.

5 Roman Fort of Gellygaer, p. 53.

6 Mr. Buchanan, however, tells me that the circular markings on his
plan are conventional merely, and that he observed no indications that
would lead him to believe that the hole was a well.

I'LATK XXX

-

I. STONK-I.INKD PIT IX SAflELI.UM

2. HVPOCAUST PILLARS IX THE BATHS

ROUGH CASTLE FORT

ROUGH CASTLE 227

a residence or not,1 was a replica, on a smaller scale,
of Lyne.2 At Rough Castle, the commandant's house,
if we may venture to call it so, bore as unmistakable
traces of re-building as did the Baths in the annexe.
Not only was there " the most complete and finished
workmanship side by side with that of an inferior
character," 3 but the east wall was represented by
two distinct foundations running close together on a
different alignment. In a word, the interior buildings
emphatically confirmed what the examination of the
ramparts had led the excavators to believe. The fort
that was contemporary with the Vallum — the Antonine
fort, as it may conveniently be termed, — had been at
some time or other entirely reconstructed. Let us see
whether more light upon the two periods can be got.

If we come back to the Plan once more (Plate
XXXIII.), it will be observed that, as the Military
Way approaches the fort from the west, it seems
to divide into two branches while it is still on the
further side of the Rowan Tree Burn. The branch
on the left descends by the edge of a small tributary
until it reaches the bank of the burn at a point
where the stream can be crossed on the level. Once
over the ford, it climbs straight up the steep face of
the hill and enters the west gate. The branch on
the right takes a much easier course. It does not
go nearly so far down into the hollow. Where it
touches the burn, the banks are high, and there
must have been a bridge. The ascent eastwards is
correspondingly gentle, and the road, when it reaches
the level, appears to hug the southern margin of the

1 See supra, p. 80. 2 Proc. Soc. of Antiq. of Scot. 1901, p. 180.

s Proc. Soc. of Antiq. of Scot. 1905, p. 481.

228 THE FORTS

fortifications, establishing a connection, as it passes,
first with the Porta Decumana of the fort, and then
with the south-western corner of the annexe.

This double road \s prima facie surprising. So, too,
is the triple system of ditches at the east end of the
annexe. Had the three ditches lain close together,

o

they would have been easily enough accounted for,
except as regards the breach in the Military Way,
although even then it would have been extraordinary
to find so much pains bestowed on what was a mere
appanage of the fort, meant to be cut adrift in time
of pressure. As they stand, the ditch in the centre
would have been worse than useless in the event of
an attack, owing to the great breadth of the intervening
'platforms.' A storming party would have found it a
convenient shelter in which to gather breath for a final
rush. The danger of its being put to such a use was
all the more serious because of its exceptional width —
20 feet as against 1 5 feet for the other ditches of the
annexe. We are, in fact, beset with difficulties if we
regard all three as belonging to a homogeneous system.
But there is an alternative possibility.

The double service of roads and the triple system of
ditches may be intimately connected. The difficulties
largely vanish if the evidence yielded by the two be
combined, and the solvent of successive periods applied.
The course of events might then be reconstructed thus.
When the Antonine fort was first erected, the Military
Way followed the line represented by the left branch.
That is, it chose the more direct, but also the more
trying, route — a plan which enabled it to be utilized as
the Via Principalis of the fort. To this period would
belong the innermost and the outermost of the triple

ROUGH CASTLE 229

row of ditches, the size of the annexe having been
altered in the course of the occupation. It is well
known that changes of the kind were frequently made.1
On the other hand, when the reconstruction of the
Antonine fort took place, the course of the Military
Way was modified. By throwing a bridge across the
stream, the steeper gradients were avoided, and that
was presumably the motive for the change. An in-
direct effect was that the entrance to the annexe could
now be placed in a position that was more secure, as
being under the immediate command of the tower that
doubtless surmounted the south-east angle of the fort.
Accordingly an opening was made there, and a short
connecting road laid through the ditch, which (to
quote from the report of the excavators) is here
"observed to be filled up almost to the top, for a
width of 15 feet, with tumbled stones which make a
passage across on the level. . . . That the trench has
been unobstructed originally is evident, for the stones
and soil used in the filling up are distinctly additional,
and are only found in the hollow of the trench."2 At
the same time the making of the new entrance rendered
it possible to dispense with the former one. Con-
sequently the east ditch of the new annexe — for neither
of the earlier annexes was of suitable size — was cut
right through the old Military Way, instead of stop-
ping short on the margin as its two predecessors
had done. Hence the otherwise unintelligible breach.
An interruption of this kind was now of no moment.
Indeed, it was a positive advantage, seeing that the
road had ceased to be a thoroughfare.

1 The Newstead evidence is particularly illuminating in this respect.
8 Proc. Soc. ofAntiq. of Scot. 1905, p. 482.

23o THE FORTS

The hypothesis just put forward is, of course, ten-
tative. But it can be supported on grounds that
entitle it to the most careful consideration. It agrees
well with what we know as to the certainty of there
having been at least two periods in the life of the
Antonine fort. It also suits the facts as mapped out
upon the plan. Only on two points, neither of which
is vital, does it seem to conflict in any way with
what the excavators say in their report. Firstly, all
three ditches on the east of the annexe were dis-
tinctly visible on the surface before the digging began ;
they would seem to have been open simultaneously.
Again, the account of the ' platforms ' that separate
them reads as if these ' platforms ' were not mere
intervals but elaborate raised defences, to the holding
of which the co-existence of the three ditches would be
essential. The result would be a novelty in Roman
fortification. Nor can so remarkable a theory be ac-
cepted without a more searching investigation of the
archaeological data than has yet taken place. On the
evidence as it stands, it is really easier to explain
present appearances by supposing that, as the two
older ditches fell successively into disuse, no systematic
steps were taken towards having them filled up. And
it may be that the "added soil" on the 'platforms,'
both of which "show a decided increase in height
where facing the south-east,"1 would turn out to be
the debris of earthen ramparts, in which case it would
confirm the view of there having been three periods
in the life of the annexe. This, however, is mere
guess-work. It is sufficient to say that, even if the
opinion of the excavators as to the 'platforms' were

1 Proc- Soc. of Antiq. of Scot. 1905, pp. 468 f.

PLATE XXXI

I. FRAG. MK NT OK STATUK

2. INSCRIBED SLAB KKUM THE PKINCIPIA

ROUGH CASTLE FORT

ROUGH CASTLE 231

definitely established, it would not affect the main
part of the hypothesis we have been discussing. That
depends upon the Military Way.

One further fact may be mentioned in this connec-
tion. As it finds no place in the official report, it is
very desirable to put it on record. The hard stones
of the street that entered the Porta Decumana were
worn into ruts. Many of those who saw them, as they
lay exposed during the excavations, must have been
reminded of Pompeii. No similar phenomenon was
noted at any of the other gates. That is a circum-
stance of great significance, to which we shall have
occasion to revert. It means that the second period
of the Antonine occupation, the period during which
the carts bringing up supplies passed into the fort on
the south side, was considerably longer than the period
during which the Military Way coincided with the
Via Principalis. That nothing of the sort was
observed at the Porta Praetoria, is hardly to be won-
dered at. The road that issued thence turned almost
immediately towards the right, obviously heading
direct for Camelon fort. But it was not the main
line of communication between that outpost and the
south, and it would therefore not be in regular use for
the heavy wheeled traffic by which the commissariat
was replenished.

We come now to a piece of evidence that seems
to be older than any we have yet met with at Rough
Castle. A little to the left of the Camelon road,
30 or 40 yards after it had crossed the Limes and
immediately to the west of a small traverse ap-
parently designed to mask the passage across the
Ditch, the excavators made a most interesting dis-

232 THE FORTS

covery. Guided by indefinite indications of a previous
disturbance of the soil, they opened up the ground
and laid bare a unique series of defensive pits, cover-
ing a space 200 feet long by 48 feet broad (Plate
XXXII.). The pits were arranged in ten parallel
rows, each pit being about 7 feet long by 3 feet broad
at the top, with sides that narrowed rapidly towards
the bottom. The greatest depth was about 2^ feet.
The whole was laid out on the principle of the quin-
cunx. Every row broke joint, as it were, with the row
next to it, so that the obstruction was extremely for-
midable. No such type of defence had been met with
in Roman works before. But Professor Haverfield at
once drew attention to a passage in Caesar's narrative
of the siege of Alesia. Among the extraordinary
precautions adopted in order to maintain the blockade
of Vercingetorix and yet keep the relieving force at
bay, was a device that bears a very close resemblance
to the one we have been describing. Eight rows of
pits, each three feet deep, were dug obliquely in the
form of a quincunx. Sharpened stakes were planted
in them, the points projecting about 4 inches above the
surface. Twigs and brushwood were then strewn over
them, to produce a delusive appearance of solidity, and
the snare was ready. The soldiers of Caesar's army
called them ' lilies,' from their supposed likeness to the
flower.1

No traces of stakes could be found in the Rough
Castle pits. But there cannot be the slightest doubt
as to the obstacle having been in its essence the same
as that of which Julius Caesar made use. On general
grounds, then, it is more likely to be early than late.

1 Caesar, De Bell. Gall. vii. 73.

ROUGH CASTLE 233

That is, the chances are that it belongs to the first
century rather than to the second. The detailed
arguments in favour of the conclusion thus suggested
are convincing ; it may be taken as certain that the
lilia at Rough Castle were dug by the soldiers of
Agricola. Consider, in the first place, their position.
They were 50 yards away from the rampart of the
fort, right out beyond the great Ditch of the Limes.
Further, and surely this is proof complete and final,
there is clear evidence that they had been dug before
that Ditch was there at all ; the five rows of pits lying
to the south were buried beneath its upcast.

The excavators are careful not to commit themselves
very far : " The appearance suggested that additions
had been made at different times."1 But the particulars
they give may take us a good deal further. After stating
that "the five south rows are covered by the tailing of
the glacis " — that being the name by which they always
designate the Outer Mound, — they proceed : " The
depth of soil overlying the southmost row is nearly
3 feet, but northward it tapers down to 9 inches in a
distance of about 25 feet, and continues at this depth
over the other five rows. As regards the strata of the
covering soil, nothing particular was noted in the pits
themselves, except that there appeared a few inches of
dark soil near the bottom, such as is generally found in
trenches ; but about halfway between the top of the
pit and the surface of the glacis a distinct dividing line
was observed. This line varies from \ to i^ inches
in width, and was almost black with streaks of iron
pan through it. The soil under the mark was of a
lighter colour than that overlying it. ... In this soil

1 Proc. Soc. of Antiq. of Scot. 1905, p. 456.

234 THE FORTS

were many very small pieces of broken pottery, and
they were found even near the bottom of the pits.
Most of it was of a dark lead colour, with a black
glaze showing reticulated lines on the outside, but a
few fragments were coarser, and of a light grey
colour." 1 It is a pity that none of this pottery was pre-
served. Its testimony might have been most valuable.

Even without it, however, the central fact is clear.
The five rows of lilia to the south lay underneath the
Outer Mound. For the rest, it is worth while hazard-
ing an interpretation. The pits seem to have stood
open for some little time after they were dug ; the
" few inches of dark soil near the bottom " represent
the earth detached from their sides by frost and rain.
Then they were completely filled up, of set purpose,
with the debris of the abandoned fort, the rubbish
being actually piled rather higher than the original
surface level ; this is the soil of a lighter colour which
rose above the top of the pits and contained " very
many small pieces of broken pottery." Years passed,
and grass and heather grew again. Last of all came
the makers of the Antonine Limes, who buried the
grass and heather under the spadefuls of earth that
they threw out of the great Ditch, producing the
"distinct dividing line" which "varies from \ to i^
inches in width, and was almost black."

This detailed reconstruction of the progress of
events involves, as has been hinted, a certain element
of speculation. But those who have had experience of
excavation work will probably concede that the specu-
lation is legitimate enough. The best way of putting
it to the test would be by a careful examination of the

1Proc. Soc. ofAntiq. of Scot. 1905, pp. 456 and 459.

ROUGH CASTLE 235

black line. It is just possible that such an examina-
tion might still be secured, for it is conceivable that
the line may extend some way to the south of the
lilia — say, to the margin of the north ditch of the lost
Agricolan fort. Where that margin may be, one can
only conjecture. It is quite probable, however, that it
lies under the Outer Mound, not far from the lilia. If
its direction could be ascertained, one might hope that it
would supply the key to those mysteries of the ground-
plan that are still unsolved. The ditches, or portions
of ditches, which appear to be intrusive — those across
the 'berm,' for instance, and that in the north-east
corner of the annexe — might fall naturally enough into
their places as survivals. Just as at Bar Hill,1 and
possibly Castlecary,2 the Agricolan fortifications may
have been partially utilized by the engineers of the
subsequent occupation.

The relics found at Rough Castle were less
numerous than might have been anticipated, nor did
they include any coins. The inscribed tablet has
already been mentioned. The usual varieties of
pottery and tiles were also represented, and there
were a few fragments of bronze, as well as one or
two implements of iron and some architectural and
other fragments of stone. At the best, the list is a
meagre one. There was, however, ample compensa-
tion, as will be evident from a brief summary of what
we have learned through the excavations. The lilia,
if not certain parts of the ditches, date from the earliest
Roman invasion of Caledonia. The employment of
so exceptional a form of defence has a very definite
meaning. Comparison with what Caesar says regard-

1 See supra, Plate XXI I. 2 See supra, p. 21 1.

236 THE FORTS

ing his own situation at Alesia indicates that, on some
occasion, Agricola's garrison were hard put to it, and
that their numbers were too small to man the peri-
meter effectively. In a very few years they were
altogether withdrawn, and the lilia were presently
filled up with the debris of the abandoned outpost.
Half a century later the Romans were at Rough Castle
again. This time the Antonine Limes was constructed,
and a new fort with ramparts of turf was erected on
the hill-top, the Military Way passing through it from
west to east. Then came fresh trouble. After a
relatively short period the attempt to hold the hill
was relinquished. When the position was recovered,
the fortifications were strengthened and the stone
buildings in the interior re-erected from their founda-
tions. The Military Way was now changed to an
easier line, while the size of the annexe was altered.
In its reconstructed shape the fort continued in Roman
occupation for many years — twenty or thirty at least—
indeed, till the final debacle. Whether this continuous
occupation was ever temporarily interrupted, the evi-
dence hardly enables us to say with certainty. But
there are indications that it may have been, — particu-
larly the extensive alterations in the Baths. In this
respect the phenomena are very similar to those
observed at Castlecary.

Of the Limes stations to the East of Rough Castle
practically nothing is known. As a rule, their sites
can only be guessed at. But there is a certain con-
sensus of opinion as to the situation of the majority,
based no doubt upon the occurrence of likely positions
at appropriate intervals, while in one or two cases
a priori probabilities are supported by something

ROUGH CASTLE AND FALKIRK 237

approaching positive evidence. Some have placed the
fifteenth station at Bantaskine. Others would prefer
to locate it at Falkirk.1 Roy is the principal advocate
of the claims of the former. His reasons, however, are
not convincing — "the nature of the ground answering
for the position of a fort, and the distance corre-
sponding with the mean interval between them."2
No doubt he was right as to the distance. Bantaskine
House is about a mile and three-quarters beyond
Rough Castle, while Falkirk is rather more than three-
quarters of a mile further on. On the other hand, if
the sixteenth station was at Mumrills, as we shall see
reason to believe that it was, and not at New Mer-
chieston as Roy supposed, then the balance of
probabilities, so far as distances are concerned, is
decidedly in favour of Falkirk. This was Horsley's
view,3 and Maitland4 for once agreed with him. If
they are right, the complete disappearance of the
remains of the station is readily explained ; they are
buried beneath the streets of the town. Such dis-
appearance would be a little difficult to account for in
the case of Bantaskine ; "some rows of stones on the
inside of the ditch, probably the remains of some

1 Falkirk is sometimes derived from fahel or faat, as if it meant ' the
church on the Vallum.' As Zangemeister has given currency to this
etymology in his otherwise admirable note in the Neue Heidelb. Jahrb.,
referred to supra, p. 147, it may be well to point out here that the explana-
tion is demonstrably erroneous. The ' 1 ' in ' Falkirk' is intrusive. Even
the modern pronunciation indicates this, and the old spellings (Dr. George
Neilson tells me) 'are all fau, /owe, faw— being Old English fah, Old
Scots faw, fauch = diverse-coloured.' He adds: 'It is the best-attested
place-name story in the English language, having so many forms all
joining in one clear sense— Eiglesbrec (Gaelic), Vaire (or veyre) chapelle,

Varia capella, (The) Fau kirke= speckled church.'

2 Milit. Antiq. p. 162. 3 Brit. Rom. p. 1 72. 4 Hist, of Scot. \. p. 1 73.

23 8 THE FORTS

"^ which are all that the traveller of 1697
or any one else has noted there, were obviously too
insignificant to indicate a fort.

So much for negative evidence. But positive testi-
mony is not entirely lacking. In the last edition of
Nimmo's History of Stirlingshire, published in i88o,2
we read : " About the beginning of the present
century an immense quantity of stones was dug up
from two of these gardens [on the south side of the
town] with which several walls have been built in the
neighbourhood. Fire-places were also discovered, still
bearing blackened stains of their former servitude ;
while amid heaps of rubbish were found a number of
fire-scarred vessels of a clayey compost, and of greyish
colour, about an inch thick and upwards of a foot
broad. There was likewise a vessel of exceedingly
beautiful workmanship, about the size and shape of a
common slop-dish. The material was very hard, and
resembled red sealing-wax. It bore on the outside
the figures of four lions and other hieroglyphs, with
the word Nocturna. On one of the stones dug up the
word fecit was distinctly traced. In another garden a
coin was found, having on the obverse a bust of
Antoninus, with the legend " Antoninus Aug. Pius
P.P." We need not place too much reliance on fecit.
And the potter's stamp must not be taken seriously,5
though it possibly was NOCTVRN -F.4 But the description
of ' Samian ' ware — terra sigillata — is fairly accurate.

1 Hist. MSS. Commission : Portland Papers^ ii. p. 56.

2 Revised and re-written by R. Gillespie, vol. i. p. 38.

3 It was locally interpreted as indicating the purpose the vessel was
meant to serve ! So R. Kier in the Falkirk Magazine for 1827.

4 NOCTVRN and NOCTVRN • F occur on the exterior of bowls of Type
(Dragendorff) 37 at Westerndorf (Knorr, 'Die Westerndorf-Sigillaten

FALKIRK AND MUMRILLS 239

An earlier account of these discoveries, published in
the Falkirk Magazine (1827), suggests that they were
made about the year 1815. Fullarton's Gazetteer of
Scotland, issued at Glasgow in 1844, adds one or
two further particulars : "A number of fragments of
earthenware were dug up in a garden in the Pleasance
of Falkirk, and among them one vessel, round the
margin of which the word ' Nocturna ' was legible ;
they were generally unglazed, of a white or brown
colour, and some ornamented with raised figures on
the outer surface."1 Such explicit statements are
entitled to serious consideration, and it will not have
escaped notice that the locality indicated is precisely
that which Maitland took " to have been the site of
a station or fort on the wall."2 It should perhaps
be added that a few small objects, such as might
well be part of the debris of a Roman settlement,
are said to have been found within more recent years,
in the same neighbourhood.3

The site of the sixteenth fort is rather better authen-
ticated. It was probably at Mumrills, on the brow of

des Museums Stuttgart,' Fundbericht aus Schwaben, xiv. (1906) p. 84).
But it is not usually supposed that Westerndorf pottery ever made its
way to Britain.

1 Op. cit. i. p. 520. 2 See supra, p. 135.

8 Information from Mr. M. Buchanan. The only one of these which
I have been able to see is an interesting little head of Pan, horned and
bearded, made in a whiteish metal which has taken on a rich green
patina. Mr. Rennie, its present possessor, tells me that it was got about
1860, by his father, in a 'hole' which he encountered in trenching his
garden, now the site of a bowling-green. With it was a quantity of
pottery, but the pottery was thrown back into the ' hole,' as being of no
particular interest. The head, which is about 3^ inches high by 2$ broad,
is very well executed. The question as to whether it is of Roman or
of later workmanship is, however, hardly to be determined without a
closer scrutiny than it was possible for me to give it.

24o THE FORTS

the hill overlooking the Westquarter Burn. There is
no spot on the whole isthmus where the general con-
ditions correspond more closely with those which we
know to have been demanded by Roman military
engineers. And there have been specific indications
that the plateau was actually occupied. Gordon's
evidence is important : " Some Quantities of Roman
Vessels made of red Earth are found here, like those
near Castlecary Fort. Besides there are broken Pieces
of Urns, and hollow square Conduits of the same
Earth, very thick and hard, found at this Place."1
Horsley merely re-echoes Gordon.2 Maitland goes
further. He speaks of "many Roman chequered
stones . . . (with a hand-mill) which are erected in
the walls of the houses, besides a number of urns
and broken vessels discovered at this place."3 The
"village" of Castle-towry, to the houses of which
Maitland refers, has long since disappeared, but not
many years ago its site was still marked by a heap of
stones in the field opposite Mumrills.4 The story as
to "chequered stones" receives confirmation from an
apparently independent source — Nimmo's History of
Stirlingshire (1777), which states that "many stones
of Roman workmanship have been dug up in an
adjacent field upon the south."5 Lastly, in 1834 the

1///«. Sept. p. 60. zBrit. Rom. p. 172. * Hist, of Scot. i. p. 172.

4 Information from Mr. M. Buchanan, who says that the ruins were in
the field lying to the west of the road that runs south from the farmhouse.
The fort, however, should rather be looked for in the field to the east. It
may be mentioned that in April, 1910, Mr. A. O. Curie picked up an
undoubted fragment of Roman window-glass on the surface of the eastern
field. It was then in grass. If a look-out were kept when it was being
ploughed, many relics might be found.

6 Vol. i. p. 45.

PLATE XXXIV

I. ALTAR FROM ROL'OII CASTLK

2. TOMBSTONE FROM MUMKILLS

MUMRILLS AND INVERAVON 241

tombstone of a Roman soldier (Plate XXXIV. 2)
came to light in the immediate neighbourhood. With
it was a quern of Niedermendig lava.1 This volume
of testimony is very considerable, whereas New Mer-
chieston, which Roy preferred on grounds of distance,2
has not a single discovery to its credit.

There is virtual unanimity regarding the claims of
Inveravon to be the site of the seventeenth station.
Nor need we have much hesitation about admitting
them, albeit they rest solely on conjecture. It is true
that Sibbald quotes Irvine as saying that the Wall
"runs down to the Water of Evin, and crosseth there,
and goeth up to Innerevin where there hath been a
Fort, and the ruins of Buildings remain yet."3 But
the ruins which Irvine had in view were clearly those
of the mediaeval tower, which still stands in the wood
about 100 yards west of the house. Its masonry
is not Roman, nor is there anything about it to
suggest that Roman material was utilized in erecting
it. We may safely infer that, when it was built,
there were no Roman remains left above the surface.
Gordon, Horsley, and Roy all alike failed to find any
visible signs of the station, although at least the two
former are in agreement as to its precise position.4
Maitland would have us believe that their failure was
due to lack of observation : " Gordon and Horsley both
declare they could find no clear vestige of a fort at this
place ; whereas had they carefully searched the gardens
on the northern side of the village, they would have
discovered certain remains of a fort, with the wall for

1 Nimmo's Hist, of Stirlingshire, ed. 1880, vol. i. p. 40.
2 Milit. Antiq. p. 162. It was 400 yards further west.
3 Hist. Inq. p. 30. 4See suflra, p. 142.

Q

242 THE FORTS

its northern boundary, without reckoning the stone
circular tower at the south-western corner of the place
as part thereof, which Dr. Irvine seems to imagine."1
The Revised Ordnance Survey Map shows indefinite
markings in the field in front of the house, and again
along the western face of the plateau. But these are
too doubtful and irregular to serve as the basis of any
conclusions. Until its existence has been demonstrated
as the result of a proper search, the fort at Inveravon
will be no more than a possibility.

Einneil is in even worse case. Considerations of
distance emphatically demand a station here, and there
is at least one spot that would be quite suitable.2
At the same time there is no site that is at all so
obvious as was the plateau at Inveravon. Consequently
almost no one has ventured upon an explicit assertion.
Horsley and Maitland are exceptions. The former
says : " TJjere are at Kinniel some faint vestiges of a
fort."3 But he refrains from giving any hint as to its
whereabouts. Maitland is much more confident. In
this instance, however, we cannot accept him as a
guide. He takes us quite away from the true line of
the Limes to a point " on the southern coast of the
frith of Forth, about a mile to the westward of the
town of Borrowstonness, at the brow of a steep hill
called the Cowbank, near the pavilion or summer-house
at the north-western corner of Kinniel-park, a little to
the eastward of the village of Kinniel ; as is manifest
by the eastern end of the ditch's being plainly to be
seen at the precipice of the Cowbank, adjoining to the
road leading to the town of Borrowstonness. And
though the station or fort at this end of the wall is

1 Hist, of Scot. i. p. 172. 2See supra, p. 147. 3 Brit. Rom. p. 159.

KINNEIL AND CARRIDEN 243

become a prey to the plough, yet by the unevenness of
the ground within the ditch, resembling ruins, is shewn
its site. The rampart and military way are lost in
the arable land."1 Maitland, it will be remembered,
believed that the wall did not run further east than
Kinneil. He searched the coast for an appropriate
terminus, and he evidently convinced himself that he
had found it.

The nineteenth and last station is usually assigned
to Carriden, because, as Gordon puts it, "Roman
Altars, Inscriptions, and Coins, have been dug up at
this Place."2 Sibbald saw what he considered to be
"the vestige of a Fort" at Bridgeness.3 But this passing
notice contrasts strongly with the language he applies
to Carriden : " There have been great Buildings here
of old, and the name Cair in the old Language
signifieth a Town ; in the Rubbish here was found a
Golden Medal of the Emperor Vespasian, which was
shown to me by my worthy Friend Alexander Miln
the Proprietor, and the Owner of the Mannor and
Lands : He told me while he was building there, a
Stone was digged up with an Eagles Head graven upon
it, and some pieces of Potter-work was likewise found
there."4 The "stone . . . with an Eagles Head" was
probably identical with the centurial stone which
Gordon and Horsley saw built into the wall of Carriden
House.6

Vague as they are, these indications are not to be
lightly set aside. It should, however, be borne in mind
that, even if we allow that there was once a Roman

1 Hist, of Scot, i p. 171. 2 1 tin. Sept. p. 60.

3 Hi 'st. Ing. p. 30. *Hist. Inq. p. 31.

6 Itin. Sept. PI. 10, Fig. 6, and Brit. Rom. p. 192, n. 6 (Fig. xxiv.).

244 THE FORTS

fort at Carriden, we do not thereby decide the question
of the termination of the Limes as against Bridgeness.
It was by no means necessary that there should be a
station exactly where Ditch and Rampart reached the
river. What happened was that, at the point of
junction, the Forth became the frontier-line. The
Military Way continued its course towards Cramond.
Accordingly, if there was no convenient site at the
actual end of the Limes — and Bridgeness would not
have been specially convenient— the 'terminal station'
might quite well have been placed a little further to the
east. Carriden House was about three-quarters of a
mile away. Roy gives the distance from Kinneil as
4,050 yards — by no means an abnormal interval.

CHAPTER VIII

MINOR STRUCTURES

AT this point it may be convenient to summarize
briefly what the two preceding chapters have taught
us regarding the position of the various stations.
They probably numbered nineteen in all, inclusive of
Carriden. In the list that follows, ordinary capital
letters are used for printing the names of those sites
where structural remains still exist, or are known on
good authority to have existed until a comparatively
recent period. Italicized capitals indicate sites where
there is no definite record of structural remains, but
where evidence of identity has nevertheless been
secured in the shape of inscriptions, pottery, and the
like. Ordinary type is employed where there is a
mere presumption, resting upon suitability of situa-
tion. The distances are taken from the measure-
ments made by Roy.1 It will be seen that they
give a total of 36 miles 620 yards as the length
of the Wall. If the end was at Bridgeness, this
total would be reduced to 35 J miles. The average
interval between the stations works out at rather
more than 2 miles.

1 Milit. Antiq. p. 163.

246 MINOR STRUCTURES

LIST OF FORTS

CHAPEL HILL (OLD KILPATRICK)

to
DUNTOCHER - - 3>5?o yards.

to
CASTLE HILL - - 3,450 yards.

to
NEW KILPATRICK - - 2,450 yards.

to
BEMULIE - 4,600 yards.

to
C ADDER^ - 3,900 yards.

to
KIRKINTILLOCH - - 4,150 yards.

to
AUCHENDAVY - - 2,970 yards.

to
BAR HILL - - 3,450 yards.

to
CROY HILL - - 3,200 yards.

to
WESTERWOOD - - 3,080 yards.

to
CASTLECARY - -3,320 yards.

to
Seabegs - - 3,300 yards.

to
ROUGH CASTLE - 2,860 yards.

to
FALKIRK - 4,830 yards.

to
MUM RILLS - - 3,400 yards.

to
Inveravon - - 4,000 yards.

to
Kinneil - 3,400 yards.

to
CARRIDEN - - 4,050 yards.

1 The measurement here is taken from the probable site of the fort (see
su^ra, p. 173), not, as by Roy, from the 'castellum.'

A GUARD-HOUSE 247

The greatest gap between the forts as thus arranged
was that between Rough Castle and Falkirk. We have
already seen that this consideration led Roy to place
the fifteenth station at Bantaskine. But the difficulty
of which he was conscious was hardly so great as he
believed. About 2,200 yards to the east of Rough
Castle there was an opening in the Limes, through
which passed the road that served to connect the
Military Way with Camelon and the posts that lay
beyond. And there is good reason to think that here
there was a guard-house. The anonymous traveller
of 1697, wno nas been already repeatedly quoted, adds
at the close of his description of Camelon : " Between
the squares above said is a paved way of half a mile
long which lead up to the Roman Wall, at the end of
which stood a great castle called by the country folks
the Maiden Castle, but now little is to be seen of it."1
Gordon, Horsley, Roy, and Stuart all take notice of
the intersection of the Wall by the road to Camelon.
But they appear to know nothing of any fort. Mait-
land, on the other hand, speaks of the Wall as joining
" the Madun-castle, a fort on the wall to guard the
gateway, that gave passage to the great military way
from the southern parts of Britain to the northern,
through Camelon, Stirling, Sec., the vestigia whereof still
appear on the outward side of the said gateway." 2 The
last words are important. They enable us to identify
the Maiden Castle with a high mound that stood until
1894 on the north side of the Ditch, immediately to
the west of the road.3 This furnishes a plausible

1 Hist. MSS. Comm. : Portland Papers, ii. p. 57.

''•Hist, of Scot. i. p. 173.

3 Grammatically, ' vestigia ' might have ' the great military way ' for its

248 MINOR STRUCTURES

explanation of the silence of Gordon and others.
Owing to its position outside the Ditch, it may have
seemed to them to form no part of the Roman works.

Leaving the Maiden Castle for the moment, we
may note that, so far as could be judged without
excavation, a clear outlet for the road had been left
in both Rampart and Ditch when they were originally
constructed. That statement is made on the authority
of Mr. M. Buchanan, who carefully watched the process
of destruction that was carried out upon the spot in 1 894
during the erection of Watling Lodge, taking notes
and making sketches. His notes bear that the stone
foundation of the Rampart seemed to stop short on
either side of the opening, and that similarly the ends
of the Ditch on either side of the roadway appeared
to taper up gradually and naturally to the level. What
he says on the latter point is amply borne out by a
photograph that still preserves a picture of the Maiden
Castle (Plate XIII. 2). The most interesting part of
his report is, however, that which records his wholly
unexpected discovery of the remains of the guard-
house by which the break in the Limes was protected
(Figure 12).

A little distance to the east of the outlet for the road
the workmen uncovered 100 feet of stone-work, about
15 feet broad, very like the stone foundation of the
Rampart. At first it ran due south at right angles to
the Limes.1 Subsequently it bent westwards, con-
antecedent. The evidence of the photograph, to be referred to presently,,
is, however, altogether in favour of the interpretation adopted above.

1 Mr. Buchanan has shown upon the Plan as certain (i.e. marked with
a black line) only the section which he himself saw in situ. The portion
immediately adjoining the Rampart had been torn up by the workmen
before his attention was directed to the matter.

A GUARD-HOUSE

249

o'
•«/»•

o-

uT

Ul

a

o
ffi

250 MINOR STRUCTURES

tinuing in that direction, still perfect, for 40 feet.
Thereafter it proved to be much broken up and
disturbed. In the course of the operations great
quantities of very coarse pottery, of a light grey
colour, were found among the debris. Mr. Buchanan
observed no ' Samian' ; and the absence of finer dishes
is significant. Taken in conjunction with the size of
the post, it indicates that the occupants were common
soldiers and petty officers, who lived in simple fashion
and did not enjoy the luxury of imported table ware.
Their chief duty would be to regulate the traffic and
supervise the customs. And the road must have been
a busy one ; Camelon was strongly held during the
Antonine period, and it must have been only the first
of a series of forts that stretched at least as far north
as Ardoch.

The garden of the villa now covers the site of the
guard-house. The villa itself is built exactly where
the Maiden Castle used to stand. According to Mr.
Buchanan's description, the latter was a large mound,
obviously artificial, which rose high above the ordinary
Outer Mound of the Limes, projecting a considerable
way beyond it, and presenting to the north a front
that was approximately semi-circular.1 It was con-
structed of earth, not of turf. Its exact dimensions
are no longer ascertainable. To render it suitable to
receive the foundations of the house, it was reduced in
height by 6 or 7 feet, the soil that was removed being
thrown to the outer side to form a terrace. Its char-

1 The only notice of it I have observed in any published account, since
Maitland, is a reference made to it in 1892 by Dr. G. Lowson : "Here it
seems as if there were a little guard mound or speculatorium" (Trans, of
Stirling Nat. Hist, and Arch. Society ', 1891-2, p. 51).

MOUNDS 251

acteristic features were thus entirely obliterated. In
the circumstances its original appearance and purpose
must remain doubtful. On the whole, however, it is
perhaps permissible to treat it as one of a small group,
of which the tumulus near Cadder Church is by far
the finest example (Plate LIV.).

This tumulus is in excellent preservation. Oblong
in shape, with rounded ends, it has a 'circumference*
of 395 feet at the base and 210 feet at the summit, and
stands 20 feet above the level of the ordinary surface.
On the northern face the ditch is still 7 feet deep and
30 feet wide. The top is flat, and its major axis
(59 feet) runs from E. to W., the minor axis (57 feet)
from N. to S. In the immediate neighbourhood the
Rampart and Ditch of the Limes have both dis-
appeared. The relative position of the tumulus is,
therefore, matter of conjecture, although the prob-
ability would seem to be that it stood on the southern
or Roman side of the line.1 Gordon is the first writer
who makes any mention of it.2 His allusion is of the
briefest. Horsley's description is fuller, and might
apply almost literally to-day : "A little to the east of
this rivulet is an exploratory mount, in the figure of
the lower part of a pyramid, having a rectangular
parallelogram for its base. On the north side is a
ditch yet visible, which turns round the two corners.
The top is flat, and the figure of it is a regular
oblong."3 Roy speaks of it as "a fine rectangular
tumulus, or castellum, that hath been surrounded with a
ditch,"4 while Stuart says that it is "supposed to have
been the site of a Roman Castellum or Watch Tower,"

1 See supra, p. 1 19. 2 Itin. Sept. p. 54.

3 Brit. Rom. p. 168. * Milit. Antiq. p. 159.

252 MINOR STRUCTURES

and goes on to suggest that it was " perhaps connected
as an exploratory mount " l with the station that lay
close at hand. Roy, by the way, was doubtful as to
whether there had been any such station: "as no
vestiges of it remain, it seems doubtful whether the
castellum abovementioned may not have been the
only post on this part of the wall."2

All of these writers, it will be noted, tacitly assume
that the tumulus is of Roman origin. Not so Maitland,
who, while admitting that " by some it is taken for an
exploratory mount," proceeds : " But that it was not
erected for that purpose, I think is evident, by an emin-
ence just across the burn, which is much higher and
fitter for exploration. Now as this small mount seems
to have been neither a castle, nor a place for observa-
tion, that which bids the fairest for the use it was
designed for, I think, is a torn-moid or court-hill,
whereon courts of justice were anciently held." 3 The
view that it was a mote, is not an unnatural one. It is
scarcely likely that any one would have thought of
connecting it with the Romans, had it not been for its
situation in the immediate vicinity of the Limes. And
yet, if the matter be looked at closely, other arguments
will be found to emerge, none of them indeed con-
clusive, but all tending to establish a presumption in
favour of the common opinion. To begin with, there is
the shape — straight sides and rounded corners. Then
there is the stone conduit which is reported to have
"been repeatedly met with, and cut through, in the
course of trenching, and other operations, at different
points."4 It led from a spring on the high ground

1 Caled. Rom. 2nd ed. p. 321. * M, Hit. Antiq. p. 159.

3 Hist, of Scot. i. p. 180. *Proc. Soc. of Antiq. of Scot. i. p. 174.

MOUNDS 253

near the Manse down in the direction of the tumulus.
Lastly, and chiefly, there is the fact that the Cadder
tumulus did not stand alone. The Maiden Castle has
already been discussed, and there were certainly
others.

Of these the only one that survives is, like the
Maiden Castle, on the north of the great Ditch. It
lies in a piece of waste ground, behind some cottages,
immediately to the north of the Bonny Mill Dam, now
disused and dried up. Apparently it has suffered much
damage within the last two centuries. In 1697 it still
showed as " an heap of earth on the outside of the
ditch of a square figure, about sixty feet long, forty
broad, and twenty high ; flat on the top. 'Tis very
entire, and has had a ditch round. " l Gordon says
that the ditch is " about 20 Foot broad, joining with the
great Fossa of the Wall." He adds : " Abundance of
Iron and Lead Ore is dug up near this Hill, some
of which I carried away with me, and, probably, the
Romans, at this Place, might have had a Foundary for
melting their Metal."2 What principally impressed
Horsley was its resemblance to the Cadder example;
he calls it "a beautiful exploratory mount, not unlike
that near Calder church. It is situated on the south
brink of the ditch of the wall, and has itself a ditch
round it, except on the south side. It consists only of
earth, but seems to be more regular and beautiful than
that at Calder"* Here, as at Cadder, Maitland re-
garded the "small but beautiful little mount, . . .
inclosed on three sides by its own ditch, and on the

ist. MSS. Comm.: Portland Papers, ii. p. 57. *Itin. Sept. p. 57.
* Brit. Rom. p. 171. It is obvious from what follows that "the south
brink " is a mere misprint for " the north brink."

254 MINOR STRUCTURES

southern by that of the wall," not as Roman, but as
"either a Scotish or Pictish torn-moid, or court-hill."1
Roy, who calls it the Chapel-hill, describes it merely
as " a small castellum, situated on the north side of the
ditch, and surrounded with one of its own."2 By 1777
its height had shrunk from the twenty feet of eighty
years before to " about seven or eight feet." Nimmo,
to whom we owe this information, gives it an embellish-
ment that seems to be purely imaginary. He avers that
it is " surrounded at the top with a parapet of earth,
which is still breast high."3 To-day the whole rises
hardly more than a foot or two above the normal level,
while the trench by which it was once girt is barely
discernible, having been filled up by the debris that
has slipped down from the top and sides. As is only
natural, it has partly gained in length and breadth
what it has lost in height, so that its apparent super-
ficial area is somewhat greater than the dimensions
recorded for 1697 would indicate.

It will be seen that, as regards the three examples
discussed above, our information is more or less definite.
Notices of others are vaguer. Horsley, after speaking
of Castlecary fort, adds that "there is a round tumulus
consisting of earth and stone about a furlong east from
the fort, near a house or two called Booneck. It stands
just on the wall, and therefore one would think must
rather have been an exploratory mount, than a barrow. "4
Maitland saw it too, but he dismisses it with a con-
temptuous reference to " a tumulus on the wall, thought
by Horsley to have been an exploratory mount, which,
on inquiry, I found to be the remains of a corn or malt-

iHist. of Scot. i. p. 174. 2 Milit. Antiq. p. 161.

3 Hist, of Stirlingshire, i. p. 42. * Brit. Rom. p. 171.

MOUNDS 255

kiln."1 Roy does not commit himself further than to
say that at this point "something like a small tumulus
exists in the ditch."2 Nimmo returns to Horsley's
opinion : " Upon the northern brink, stands a small
mount, which is supposed to have been an exploratory
tower."3 This evidence is far from satisfactory, but
there is no hope of testing its value now. The tumulus
is visible no longer.

As it has disappeared, we cannot, of course, say
whether it had borne any real resemblance to the other
three or not. The possibility that it may have done
so is not excluded by the fact that it differed from
them in shape. But one thing is clear. These four
mounds cannot have belonged to the Antonine Limes,
seeing that at least three of them were on the wrong
side of the Ditch. Yet, if they are the survivors of a
series, as appears to be by no means impossible, their
close association with the barrier renders it difficult to
avoid connecting them with the Romans.

If we assume for the moment that they were
Roman, there need not be much hesitation in deciding
what their purpose was. The presence of mounds in
large numbers all along the limites of Upper Germany
and Rhaetia has been conclusively demonstrated.
They are usually round, more rarely straight-sided.
Excavation has proved that they were intended for
the support of wooden towers. Post-holes are found,
showing that they had supported scaffolding, while
the ditches yield pottery and other relics which make it
certain that they were garrisoned.4 The closest
analogy in this country is provided by the mounds

lHist. of Scot. i. p. 174. *MiKt. Antiq. p. 161.

*Hist. of Stirlingshire, i. p. 41. * Arch. Anseiger, 1899, 2, pp. 81 ff. etc.

256 MINOR STRUCTURES

along the ' Roman Road ' between Ardoch and
Dupplin. These were investigated by the Society of
Antiquaries of Scotland in 1899, and turned out to be
remarkably like the German ' Begleithugel.^ No
remains of any kind were, however, discovered in
association with them, — a circumstance that suggests a
short occupation. That again would point to the first
century ; and, if the Perthshire examples are of the first
century, so too — always on the assumption that they are
Roman — are those between the Forth and Clyde. It
is unsafe to dogmatize on so slender a basis. But one
is tempted to conjecture that the Agricolan Limes in
Caledonia, like the contemporary Limes in Germany,
had consisted of a line of road, with earthen forts at
intervals and wooden towers between. If the Cadder
tumulus ever supported a wooden tower, it is still so
little damaged that the post-holes would be easy to
discover, while an examination of its ditch might yield
sufficient evidence to enable its date to be more or less
precisely determined. It is true that both it and the
Bonnybridge mound — the Chapel-hill, as Roy calls
it, — were a good deal higher than their supposed
analogues. On the other hand, in undulating country,
such as that through which the Limes ran, ordinary
precaution demanded a considerable elevation, if the
posts were to be of any use for purposes of observation.
Whatever conclusion may ultimately be come to
regarding the Cadder tumulus and its companions,
there is no doubt that the Antonine Limes had minor
structures of its own. It is, however, quite impossible
to fit these into any regular system. "'Tis true," says
Horsley, "that besides the larger forts on this wall,

1Proc. Soc. ofAntiq. of Scotland^ 1901, pp. 15 ff.

MOUNDS AND PLATFORMS 257

which seem to have stood at regular distances, there
are a smaller kind of exploratory turrets, or mounts
yet visible here and there ; but they are so few and so
situated, that one cannot conclude from what now
appears, that the series of them had been so regular
and uniform as I have shewn that was on the wall of
Severus. These in Scotland have, as the wall itself,
consisted of earth or turf, and so were more liable to
be demolished than the castella of Severus s wall,
which were built with stone strongly cemented."

Horsley would, no doubt, have included in his list the
four mounds of which we have already disposed. If
these are left out of account, there is only one spot in
the western half of the isthmus where traces of any-
thing of the sort have ever been detected. And even
there the traces are very uncertain. Gordon himself,
their first discoverer, is far from confident : " Nor am
I sure, but that where the Houses of the Peelglen are,
there might have been another Fort, seeing the
Foundations of Stone Buildings appear pretty visible
on this Ground, tho' not so distinct, as to afford me
an Opportunity of taking its true Dimensions and
Draught."2 The Peel Glen, it will be remembered,
was only 500 yards west of Castle Hill. Horsley
accordingly argues that any fort which stood here
could have been but of minor importance : " 'Tis plain
it has not been one of the principal forts in the series,
for it is too near to that of Castle-hill" * But at the
best his testimony to its existence is rather hesitating :
" There is also some appearance of the ruins of a small
fort or castellum, just at the village."4 Maitland, as

1 Brit. Rom. p. 163. 2 1 tin. Sept. p. 52.

*Brit. Rom. p. 165. 4 Ibid.

258 MINOR STRUCTURES

usual, was sceptical : " After the strictest search and
inquiry, I could not discover the least appearance or
vestige of a fort at that place. Besides, to have
erected a fortress within a quarter of a mile of that on
the Castle-hill, was needless."1 Roy seems half-
inclined to be a believer: "The deepness of the bottom
preventing its being seen from the next station on the
wall, hence, as well as from the name, it is supposed
that a small castellum had been established here, for
the security of the gorge; though in 1755, the traces
of it were scarcely to be discerned."2 Scarcely dis-
cernible in 1755, they have completely vanished now.
No materials for an independent judgment remain
above the surface.

More interest attaches to a group still visible along
the better preserved portion of the Limes. It consists
of a series of platforms, approximately semi-circular,
which project at irregular intervals from the south face
of the Rampart. These were noticed in chapter v.
as they occurred.3 There are at least two on the
western slope of Croy Hill, two in Bonnyside grounds,
and probably two in the Tentfield Plantation to the
east of Rough Castle. There may be others.4 The
most prominent, though not the largest, is due south of
the house of Bonnyside. It appears to have caught
Gordon's eye, for in all probability it is to it that he
refers when he says : " I perceived a little Mount rise

lffist. of Scot. i. p. 182. *Milit. Antiq. p. 158.

3 See supra, pp. 125, 132, and 133.

4 It is, however, doubtful whether the Glasgow Committee are right in
placing one "near Dullatur railway station" (Antonine Wall Report,
p. 147). Evidently the mound just within the entrance to Easter Dullatur
is what is meant. But this may well be a much later erection ; it may be
contemporary with the avenue rather than with the Roman Wall.

PLATFORMS 259

upon the South Bank of the great Ditch, not unlike
the small Turrets upon Severuss Wall." * And no
doubt the second of the two Bonnyside platforms is
identical with the " square Watch Tower, in Figure and
Proportion exactly like those upon Severuss Wall in
England" which Gordon met with " farther Eastward." 2
The actual positions are not easy to reconcile with his
statements of distance, but a comparison with the map
will show that his record of measurements is apt to be
confused and spasmodic. Similar discrepancies need
not, therefore, deter us from recognizing in still existing
remains two discoveries which he made in the Tentfield
Plantation3 — a second "square Exploratory Tumulus,
like that already mentioned, being only 1 5 Feet square,"
and a second "square Watch Tower, like the other
already described, each Side about 65 Feet in
Length."

Horsley does not allude to the two examples in the
Tentfield Plantation, but he notices the Bonnyside
ones in terms very like those that Gordon had used,
adding the detail that the larger was "sixty-six foot
square."4 Maitland, like Horsley, is silent as to the
two Tentfield examples. When he comes to Bonny-
side, he ignores the smaller, saying merely : " A little
farther is the remains of a castellum, of sixty-six feet
square, and of the same dimensions with those on the
stone wall in the northern parts of England."5 Roy,
on the other hand, makes no reference to the Bonny-
side " castellum," but does touch casually on the
"exploratory turret," when he speaks of " the traces of

1 Itin. Sept. p. 58. 2 Ibid.

3 Ibid. p. 59. * Brit. Rom. p. 171.

5 Hist, of Scot. i. p. 173 f.

260 MINOR STRUCTURES

an old watch tower, near Elf-hill." a He was also aware
that there were remains in the Tentfield Plantation.
No hint of this is given in his text at the place where
we should expect to find it. But one of the fixed
points in his table of measurements is " Gilmor-seat
castellum, or watch tower," which is said to lie 1000
yards east from the " west part of Rough Castle fort." 2
Further, in his ' Plan and Section of Grime's Dyke,'
he shows a small earthwork, about 18 feet square,
abutting on the south face of the Antonine Rampart
and entitled ' Gilmor Seat.' The dimensions and the
distance from Rough Castle combine to make it certain
that Roy's " castellum " is identical with Gordon's
"exploratory turret," and also with the remains dis-
cernible to-day 100 yards east of the mineral railway
line. Of Gordon's " square watch tower," which is
still visible some 600 yards further east, Roy appears
to have seen nothing, although it is much more
distinctly marked than the "turret." Gilmor Seat is
the name of a slight rising ground that comes between
the two. Just about its highest point, the Limes
takes a turn from south-east-by-east to east.

The foregoing summary includes everything that
the earlier writers have to tell regarding the semi-
circular platforms. It is curious that all of them should
have failed to observe the two largest of the series —
those upon Croy Hill. As it happens, we are now

1 Milit. Antiq. p. 161. This expression of Roy's is doubtless the
foundation for Stuart's statement to the effect that a castellum was
supposed to have stood " upon Elf-hill, a singular conical height, which
rises, crowned with trees, within about half-a-mile of Rough Castle'
(Caled. Rom. 2nd. ed. p. 353). Stuart's information regarding the minor
structures is valueless.

2 Milit. Antiq. p. 163.

PLATFORMS 261

better informed about these than about any of the
others, as sections were cut through them by the
Glasgow Committee. In one case it turned out that
the platform had had a stone base, less carefully laid
than that of the Rampart itself, but still perfectly dis-
tinguishable.1 In the other there was no base, except
such as was afforded by the natural rock.2 The details
given in the Antonine Wall Report render it all but
certain that these platforms were not constructed until
after the Rampart was finished. Both are of sods ; but
the dark lines of the former are quite distinct from
those of the latter. The line of junction between them
is marked by a break in the lamination all the way
down, such as would be produced were a solid mound
of sods built up against an already existing sod wall.3
Whether the interval that elapsed between the erection
of the two was long or short, we cannot pretend to say
with certainty. But that there was an interval seems
clear. The judgment of the Glasgow Committee is
entitled to great weight; it is "that according to the
whole indications the vallum was first erected, and that,
probably a very short time subsequently, the expansion
was added to it — built up against the vallum."4

The size of the platforms evidently varied. The
eighteenth-century estimates of 65 or 66 feet square
for the ' watch towers ' in Bonnyside and Tentfield
are not to be relied upon ; they must have been
based upon the apparent area, which would be
magnified by the debris that had slipped down the
sides as the structure became more and more ruinous.
If we take the data from Croy Hill as a guide and

1 Antonine Wall Report, p. 77. 2 Ibid. p. 85.

3 Op. cit. p. 145. * Ibid. p. 79.

262 MINOR STRUCTURES

assume that they apply generally, we are justified in
saying that, on a rough calculation, the platforms may
have measured 15 or 20 feet from north to south,
and as many as 30 feet from east to west. The
limits thus laid down were not always reached ;
the one opposite Bonnyside, for instance, was clearly
smaller. Sometimes, on the other hand, they may
have been exceeded. On the whole, it looks as if
there had been two classes, a large and a small, corre-
sponding respectively to the " exploratory turret " and
the "square watch tower" of Gordon's Itinerarium.

Whatever their size, they must have had some
definite purpose, and we can hardly avoid speculating
as to what it may have been. Dr. George Neilson
has argued with much ingenuity, though with a frank
reservation as to the provisional character of his
hypothesis, that they were probably stands for engines
of the onager type, whence were hurled the stone balls
that have been found along the line.1 The conception
of the Limes defended by artillery is magnificent.
One only wonders whether it would have been
war. Lieutenant-Colonel Ruck, again, has advanced
the view that " the expansions may have been places
darmes for men, weapons, or stores ; they may have
been flanking places for an exit, or simply places of
assembly or parade grounds, where men could quickly
assemble or disperse."2 Reckoning the size of an
expansion as 50 feet by 40 feet = 2000 square feet —
figures that are undoubtedly too large, — he reasons :
" One row of Romans in close order, four in a row,
take up 8 sq. ft, so that 250 rows of fours, or 1000 men,

^Antonine Wall Report, p. 145 ff.

2 Trans, of the Glasgow Archaeological Society (N.S.), iv. p. 466.

PLATFORMS 263

could stand on one of the expansions of 2000 sq. ft.
in a close crowd, leaving room for 914 emergency
men to be assembled on each of the expansions, if
required, after the guard had mounted."1 That would
have been war indeed. But to have garrisoned the
Wall on this scale would have called for a force far
in excess of any total ever reached by the Roman
army in Britain. Where all is so obscure, we might be
tempted to fall back on the cautious words with which
Dr. Neilson concluded his discussion of the subject.
The platforms, he says, " may have formed the bases
for wooden turrets used as watch-towers or sentry-
boxes, or they may have been ' ramps ' or steps to
mount the wall from the south side, answering to
the 'double ascents' of Hyginus, although scarcely
occurring with sufficient frequency to make that
probable as the whole explanation. The co-existence
of more than one of these properties is feasible." 2 But
there is yet another possibility that demands considera-
tion. They may represent links in a chain of signal-
stations such as we must suppose to have run from end
to end of the Limes. They certainly seem to stand in
very suitable positions.3

In this connection we may turn once more to the
sculptures on the column of Trajan. It will be re-
membered that from the upper storey of each of the
burgi m praesidia shown in Plate III. there projected a
lighted torch. These torches were undoubtedly used
for code-signalling, much as flags and heliographs
are used to-day. Three hundred years before the

* Ibid. p. 460. ^Antonine Wall Report, p. 149.

3 Mr. A. O. Curie and myself looked at them carefully from this point of

264 MINOR STRUCTURES

Antonine Vallum was constructed, Polybius described
two or three different systems — including one perfected
by himself— under which messages of the most detailed
kind could be passed from one point to another by
the aid of torches.1 Some system of the kind was
evidently employed along the Danube in the reign of
Trajan, and we may be perfectly certain that it had its
counterpart in Scotland in the reign of Pius. But that
is not all. Provision had to be made for more serious
and more sudden contingencies, for occasions when it
was vital to communicate speedily with posts so far
distant as to be beyond reach of the flare of the
brightest of torches. And we have trustworthy
evidence as to the nature of that provision.

Plate XXXV. i shows another section of the Danu-
bian Limes as portrayed on the column of Trajan. To
the left of the praesidium are two remarkable objects
that were long a puzzle to the interpreters — a huge pile
of logs, carefully laid in the form of a solid square, and
something which resembles a rick of hay or straw or
rushes. Cichorius adopted the view that they were
meant to represent the garrison's stock of fuel and of
fodder for horses, adding that the latter suggested the
presence on the frontier of squadrons of cavalry which
space considerations prevented the sculptor from at-
tempting to depict directly ; artistic limitations would
also account for their disproportionate size, and for the
fact that they are outside of the stockade.2 This explana-
tion seems prima facie rather unlikely, and its improba-
bility is made all the more manifest if we set beside it
the corresponding scene from the column of Marcus

1 Hist. Bk. x. c. 44 ff.

2 C. Cichorius, Die Reliefs der Trajanssdule, ii. p. 20.

PLATE XXXV

I. FROM THE COLUMN OF TRAJAN

2. FKOM THE COLUMN OF MAKCUS AUKELIUS

BEACONS ON THE DANUBE FRONTIER IN THE
SECOND CENTURY A.D.

PLATFORMS 265

Aurelius. There, as will be seen from Plate XXXV. 2
the same two classes of objects recur. It is true
that their shapes have somewhat altered; the square
pile of logs has become round, and the ricks have
expanded until they resemble bee-hives. But there
is not a shadow of doubt as to their essential identity.
Further, the position of prominence now accorded
them — they actually occupy the foreground — forbids
the assumption that they are mere adjuncts to the prae-
sidia. Clearly they were a recognized and an impor-
tant element in the organization for the protection of
the frontier. Their real meaning has been pointed out
by von Domaszewski, whose solution carries instant
conviction. They are beacons laid ready for kindling.1
Is it not highly probable that we have here a clue to
the purpose of the ' expansions ' or semicircular plat-
forms on the south side of the Rampart of the Scottish
Limes ? They may mark the points from which, in
the hour of danger, the " fiery heralds " would start
upon their errand. It would not be only " swift to
east, and swift to west" that their "warning radiance"
would spread. We know that during the period of
occupation a line of forts stretched northwards from
Camelon into the heart of Perthshire. In an emer-
gency a series of fire-signals, such as those which the
sculptured columns show on the frontier where Rome
faced the Dacians and the Marcomanni, would bring
the garrison of Ardoch into touch with the garrisons
on the Limes. These again could speed the message
farther south to Hadrian's Wall, either by the chain
through Castledyke and Birrens or by that through

^etersen, v. Domaszewski, and Calderini, Die Marcus-Saule auf
Piazza Colonna in Rom. Textband, p. 109, footnote.

266 MINOR STRUCTURES

Cramond, Inveresk, and Newstead. Raiders could
be intercepted, reinforcements sent when needed. But
the contingencies that might arise in frontier warfare
were many, and the troops available to maintain the
Roman peace were few. A properly organized system of
signalling would be indispensable, if the commanders
were to make the best use of the forces at their dis-
posal and to avoid all waste of effort. This would
involve a code — less elaborate, of course, than that
which Polybius perfected for torches, but still a code ;
and a code would be impossible with single beacons.
Is it not significant that the platforms on the Limes
always seem to go in pairs ? There are two on Croy
Hill, and two on either side of Rough Castle. The
latter, if we add the fort in the centre, would form a
group of five, all visible from Camelon and far beyond.
Given a clear understanding between the parties con-
cerned, much could be said with the help of five fires.
These, however, would require to be lit in fixed posi-
tions, if the understanding were to be effective, and
the positions would require to be so far apart as to be
easily distinguishable at a distance. Both conditions
are fulfilled by the ' expansions.' And there is yet
another point. It will be recollected that the platforms
fell into two classes — a large and a small. Is it a
mere coincidence that the sculptures exhibit two classes
of beacon in simultaneous use — a pile of logs which
would burn long and fiercely, and a rick of straw or
rushes whose blaze would soon be over? The column
of Marcus shows, indeed, a third variety. The whole
question is perhaps one of those that hardly admit of a
final answer. But there is far more to be said for the
suggestion now put forward than for any other yet

PLATFORMS 267

propounded. That beacon-fires were employed upon
the Limes may be regarded as absolutely certain, and
it would be idle to argue that a code presupposes a
degree of elaboration of which the Romans were not
capable. Six centuries earlier the audience in the
theatre at Athens saw nothing improbable in the story
that Agamemnon used beacons to flash the news of
the fall of Troy across the Aegean to Argos.1 In the
interval the art of war had made great progress, and
with it, no doubt, the art of signalling.

1 Aeschylus, Agamemnon, 22 ff. The tragedy won the first prize in
B.C. 458.

CHAPTER IX
THE LEGIONARY TABLETS

THE various structural remains constitute at once
the most impressive and the most important element
in the body of archaeological facts relating to the
Limes. Next to them must be ranked the lapidary
inscriptions, the number of which is already consider-
able, and might at any time be added to. They can
conveniently be divided into three main classes —
commemorative tablets, often bearing the name of the
emperor, altars dedicated to divinities, and tombstones.
The first of these classes, however, contains a group of
so special a character as to demand separate treatment.
Nothing exactly corresponding to it has been found
anywhere else in the Roman world. It comprises
seventeen slabs, each recording that a particular
contingent of legionary troops had executed, for a
certain specified distance, some piece of work — in two
instances expressly defined as 'opus valli? the work of
the Rampart. Except as regards the use of the two
words just quoted, the form of inscription is stereotyped.
First come the name and titles of Antoninus Pius,
these being in the dative rather than in the ablative,
if we may argue from the only two cases in -which a
crucial word is written at full length. Then follow the

THE DISTANCE-SLABS 269

name of the regiment or detachment, and the statement
of distance. The former is in the nominative. The
latter is in the accusative, generally governed either by
PER or by FECIT, which may be abbreviated into
FEC or F. When the verb is employed, it usually
precedes the statement of distance. Occasionally it
follows it.1

The problem which these distance-slabs suggest has
been the subject of more or less fruitful discussion at
the hands of various writers. Argument has chiefly
centred round the significance of the terms in which
the distances are expressed. PE R M I L P and M • p • P2
each occur once, while M • p and p E R • M • P are each
found three times, P five times, and p • p three times.
The reading on the remaining slab is uncertain. Does
the P in these abbreviations refer to passus or to
pedes, to paces or to feet ? The familiar use of millia
passuiim for distances of any great extent, and notably
the analogy of mile-stones, might have seemed decisive
in favour of passus. That was the view of Horsley.
To him it appeared to be beyond question that Roman
miles and Roman paces were intended, and by a
remarkable coincidence it so happened that the sum
of the numerals mentioned on all the slabs known
in his day (39,726) was only 9 paces more than his
measurement of the actual length of the Wall (39,7 1 7).3
It is true that, in order to reach this result, he had to

1 A convenient conspectus of these inscriptions, together with a
summary of the main facts concerning them, will be found in a paper
by the late Dr. James Macdonald, printed in the Trans, of the Glasgow
Archaeological Society (N.S.), iv. pp. 49 ff.

2 Probably, however, M • P • P is merely the result of a blunder, part of
the inscription being accidentally repeated ; see infra, No. 16.

3 Brit. Rom. p. 162.

270 THE LEGIONARY TABLETS

include certain stones that are not distance-slabs at
all. Besides, the value of his calculation was speedily
destroyed by the inconvenient discovery of additional
tablets. As matters now stand, the total would reach
more than 55,000 Roman paces, a surplus of fully 15
Roman miles over the entire distance between Old
Kilpatrick and Carriden, and that without taking
account of two stones where the numerals appear to be
omitted.

Three different ways of getting over the difficulty
have been suggested. Some would interpret P as
referring to feet rather than to paces.1 Ample room
would thus be left for undiscovered slabs, while the
method of reckoning would be similar to that employed
along the line of lesser limites? At the same time it
would be a rather marked departure from ordinary
practice to measure such long distances in feet, and
inconsistent with ordinary usage to expand M • p except
into millia passuum. Accordingly, others have pre-
ferred to adhere to the old view that the measurement
is always by paces, and to be satisfied with explaining
the apparent surplus by the fact that more than one
slab seems to have been set up to commemorate the
completion of a single length.3 A via media was
devised by the late Mr. C. J. Bates, who observed
that the inscriptions found in the extreme west have,
without exception, P or p p, while all the rest have
M i L P or M P. He proposed to divide the slabs into
two distinct sets, in one of which the measurement was

1 So Hiibner, C.I.L. vii. pp. 194 ft. passim.

2 E.g. at Rome, along the Tiber (C.I.L. vi. pp. 3111 ff.) and the muni-
cipal aqueducts (C.I.L. vi. pp. 267 ff. and pp. 3123 ff.).

3 Antonine Wall Report, pp. 9 f.

THE DISTANCE-SLABS 271

by passus, in the other by pedes. " When," he says,
" we meet with M • p on the distance-slabs, we are
bound to read the figures following as Roman miles
and paces ; it is only when there is P or p p before
the figures that we are entitled to treat these as feet.
. . . The suggestion is that the Wall was begun from
the west ... As they were accustomed to do in the
construction of camps and forts, the legions recorded the
work they accomplished in so many feet ; by the time
they got to Duntocher they saw that in a great work
like the limes it was more practical to talk of so many
paces."1 Whatever may be thought of this highly
ingenious way of accounting for the sudden change
of standard — and that is a point to which we shall
return later — the division into sets has the incidental
advantage of furnishing a reason for the much greater
frequency with which distance-slabs have been found at
the western end than elsewhere along the Limes ; since
there were five feet in a Roman pace, the slabs would
originally be five times as numerous, relatively to the
length over which they were spread. Not that it
would be prudent to push such an argument too far.
When we come to deal with the altars, we shall find
that the ratio of geographical distribution exhibits
exactly the opposite proportion. Yet it would obviously
be absurd to make that ratio a basis for conclusions as
to the spiritual state of the different garrisons.

It will be best to set a priori considerations of every
kind aside, and to see what help we can get from an
examination of the slabs themselves, and from a careful
scrutiny of the circumstances attending their discovery.

1 Archaeologia Aeliana, xix. pp. 107 and 109. " Duntocher" should be
" Castle Hill." The latter is the real dividing-line between the two classes.

272 THE LEGIONARY TABLETS

In describing them we shall follow, as far as may be,
the order of the localities in which they are believed to
have been found, proceeding from west to east as we
did in the case of the Limes and its forts. Fifteen
out of the seventeen distance-slabs are now in the
Hunterian Museum (University of Glasgow), and the
question of the provenance of each of these received
particular attention in the Tituli Hunteriani?- which
gives references to the older authorities. Subsequent
to the publication of the Tituli, fresh light was thrown
on the history of some of the stones in a series of
articles by the late Mr. Alexander Gibb. Mr. Gibb's
main thesis must be pronounced untenable, but the
information which he gathered to support it is curious
and valuable.2 In so far as his researches modify the
statements in the Tituli, the results are embodied
here.

No. 1 (Plate XXXVI. i ; C.I.L. vii. 1141; Tit. Hunt.
p. 20), which has a measurement of 2 ft. 4! in. by
2 ft. 2 1 in., is an elaborate piece of sculpture, represent-
ing a distyle Corinthian fa9ade. Within the framework
is a Victory, naked to waist, reclining with feet towards
left. In her left hand she holds a palm, while her
left arm leans on a globe. The remainder of the
field is occupied by a large wreath which she holds
in her right hand. The base is enclosed in a plain

1 Tituli Hunteriani: An Account of the Roman Stones in the Hunterian
Museum. By James Macdonald, LL.D. (Glasgow: Annan & Sons, 1897).

2 'New Measurement of the Wall of Antoninus Pius' in The Scottish
Antiquary, xv.-xvii. (1901-1903). By carrying the wall to Dumbarton
Rock and 'shifting' the stones, when necessary, to sections other than
those where they are reported to have been discovered, he succeeds in
setting them all up in their proper places and in filling up " the whole
distance, except 6652 passus, between Inveravon Fort and Castel-cary
for which no stones have been found" (pp. cit. xvii. p. 196).

PLATK XXXVt

i. No. i

2. No. 2

3- No. 3
THE LECJIONARY TABLETS

THE DISTANCE-SLABS 273

moulding, having at either end an ansa-\ike projection
containing a floral ornament, and in the centre a boar
running left, the last being the well-known emblem of
the Twentieth Legion. The first three lines of the
inscription are in the pediment, the next three within
the wreath, and the last on the base, in the blank space
on either side of the boar. The whole reads :

IMP c

T- AE- HADRIA
NO- ANTotflMO AVG- Plo P- P

VEX
LEG XX
V V FEc

P P MM CBXI

A literal translation of one example will facilitate
the discussion of the rest : "In honour of (or in the
reign of) the Emperor Caesar Titus Aelius Hadrianus
Antoninus Augustus Pius, Father of his Country, a
detachment of the Twentieth Legion, the Valerian,
the Victorious, completed 4411 feet (or paces)." The
stone was presented to the University of Glasgow by
the Marquis of Montrose. According to Gordon, this
was the third Marquis.1 As he died in 1684, it was
probably the nucleus of the whole collection. Prior to
its removal to Glasgow it lay for a time at Mugdock
House, a circumstance which, taken along with the
form of the inscription — P • P, instead of M • P, — makes
it certain that it originally belonged to the western

1 He calls him " the late Marquess of Montross" (Itin. Sept. p. 50), and
he dedicates the Plate on which the stone appears to the first Duke of
Montrose, who had been the fourth Marquis. See The Scottish Antiquary,
xv. (1901), pp. 20 1 ff., for a detailed account of the various localities to
which the stone has been assigned. Mr. Gibb himself set it up at
Castlecary.

274 THE LEGIONARY TABLETS

portion of the Limes. Sibbald merely remarks that it
was "taken out of Grahames-Dyke."1 Gordon classes
it vaguely among the "many Roman Stones, with
Inscriptions," that have been dug up "at" Old Kil-
patrick.2 This is so far confirmed by the fact that
it appears to be a duplicate of the next ; the two had
apparently stood at different ends of the same section.
But it may be doubted whether Maitland had any
substantial grounds for his assertion that it came from
Ferrydyke, on the Clyde, close to Old Kilpatrick
church.3

No. 2 (Plate XXXVI. 2 ; C.I.L. vii. 1 142 ; Tit. Hunt.
p. 23) is a fragment, which however includes the
greater part of the original. The left-hand side and
the bottom are awanting. The portion that survives
measures 2 ft. by i ft. 6 in. From what remains we
can see that the general design has been identical with
that of No. 8, by a comparison with which the description
can be completed. The inscription has been enclosed
in a beaded moulding with two large ansa-like projec-
tions, each of which has contained a nude figure, doubtless
a Genius, carrying a bunch of grapes. In each of the
angles formed by the ansae and the moulding there has
been a rosette. Beneath the lettering there has been
a boar, the cognizance of the Legion, probably running
left towards a tree. The coincidence as regards the last
three digits leaves no room for doubt as to the number
of units having been identical with that on No i.

lAuctarium Musaei Balfouriani^ p. 207.

z Itin. Sept. p. 50.

3 Hist, of Scot. \. pp. 182 f. He prints it " Feny-dike" by an oversight, —
not the only name that he makes a mistake of the kind with. Mumril
for instance, figures as "Numerills" (op. at. p. 172).

THE DISTANCE-SLABS 275

(4411). The reading can therefore be restored as
follows, the only uncertain element being the verb :

[I] M P • C • T • A E

[HJADRIANO

[A]NTONINO

[AV]G PIO P P

[VEX L]EG-XX V V

[P- p.TTTTC]Bxi

[FEC]

This stone, along with No. 3, was presented to the
University of Glasgow in 1695 by William Hamilton
of Orbiston. The name of the donor gives a clue to
\h^ provenance. Hamilton was proprietor of the lands
and lordship of Kilpatrick, and his principal residence
was at Erskine House on the other bank of the Clyde.
There can hardly be a doubt but that the slabs had
been discovered on his estate. In all likelihood they
were kept for some time at the mansion-house. Hence
the popular opinion, reflected in the second edition of
Gibson's Camden, that they had been " found at
Erskine, on the river Clyde."1 Stukeley, relying on
information derived from the manuscript collections
of "the Learned and Indefatigable Mr. Edward
Lluyd,"2 states explicitly that No. 3 was got "by
plowing in the parish of Kilpatric, in the shire of
Lennox, where the Wall ended in the Lands belonging
to the Donor."3 That is perhaps the authority for

1 Op. cit. (ed. of 1722), pp. 1214 f. The inscriptions are not mentioned
in the first edition of Gibson's work. In the interval there had appeared
Fabretti's Inscriptionum Antiquarum Explicatio et Additamentum (Rome,
1699), where No. 3 is published with the title ' Marmor Erskini Scotiae
urbe repertum'' (p. 756).

2 See supra, p. 220, footnote.

3 An Account of a Roman Temple, etc. p. 8.

276 THE LEGIONARY TABLETS

Stuart's definite assertion that it was " found on the
Chapel-hill."1 While there is probably no justifica-
tion for such a precise location, we may fairly enough
assign both No. 2 and No. 3 to the extreme west of
the Limes.

No. 3 (Plate XXXVI. 3; C.I.L. vii. 1 140; Tit. Hunt.
p. 22) is a plain slab, quite without ornament. It
measures 2 ft. 8 in. by i ft. 7^ in. The letters, which
are well cut, are enclosed within a beaded moulding.

They run :

IMP-C.T. AELIO-

HADRIANO- ANTO

NINO- AVG- P P-

VEX- LEG- VT- VIC-

P- F-OPVS VALLI

P-coooooooC-XLI

The detachment responsible for the completion of
these 4141 units belonged to the Sixth Legion, "the
Victorious, the Dutiful, the Loyal " — Victrix, Pia,
Fidelis being the titles that are abbreviated. The
verb is understood, and there is no preposition.
Two peculiarities should be noted. Both recur on
No. 5, and the inference is obvious ; the stones were
designed by the same hand. The first, which is more
probably the result of accident than of ignorance, is
the omission of PIO from the emperor's titles. The
second is the use of the phrase OPVS VALLI, which is
interesting as showing by what name the Rampart of
the Limes was known to the men by whose hands it
was built. The history and probable provenance of
No. 3 have been dealt with under No. 2.

No. 4 (Plate XXXVI I.i ; C.I.L. vii. 1136; Tit. Hunt.
p. 25), which is a richly decorated stone, exhibits the

1 Caled. Rom. 2nd ed. p. 288.

Pl.ATK XXXVII

i. No. 4

2. No. 5

3. No. 6
THE LEGIONARY TABLETS

THE DISTANCE-SLABS 277

name of yet another legion. In the centre of the field
is a cable border of rectangular shape, enclosing the
last four lines of the inscription. Above is a Capricorn
and beneath is a Pegasus, both to left, these being the
emblems of the Second Legion. On either side is a
pelta-shaped ornament, with horns turned outwards and
terminating in griffins' heads — a form of embellishment
which is characteristic of many of the sculptures from
the Scottish Limes. The corners are occupied by four
large rosettes, while the first three lines of the inscrip-
tion are cut in the vacant spaces of the upper portion
of the field. The design of No. 7 is practically the
same. The reading in this case is :

IMP ANTON

AVG PIO

P P

LEG

n

AVG
FPIIICCLXXi

The number of units here is undoubtedly 3271
and not 4270, as Stuart asserts.1 The abnormal
abbreviation of the imperial titles is the result of
the superabundance of decoration. The dimensions
of the stone are 2 ft. 3 in. by i ft. 8J in.

When seen by Gordon and Horsley, No. 4 was above
the gate of Cochno House, which is in the immediate
neighbourhood of Duntocher. It was afterwards pre-
sented to the University of Glasgow by the proprietor,
James Hamilton of Barns. It seems to have been
discovered some little distance to the west of Duntocher.

1 Caled. Rom. 2nd ed. p. 298. The reading in C.I.L. vii. makes it 3270.
But this has been corrected by Haverfield {Arch. Jour. 1. p. 305).

278 THE LEGIONARY TABLETS

Mr. Gibb quotes from the Scots Magazine an interest-
ing letter written from All Souls' on April 13, 1699, by
Dr. Thomas Tanner, afterwards Bishop of St. Asaph's,
to Gibson, the editor of Camden. Its purpose was to
forward copies of five Roman inscriptions " taken last
summer in Scotland by Mr. Urry, student of Christ
Church." Mr. Urry is described as "a very curious
gentleman," and therefore as one whose report was
to be unhesitatingly trusted. One of the five is our
No. 4, regarding which it is noted : " This stone was
found at Caer Lieth, and is to be seen at Cockneugh. " l
Stukeley's information is to the same effect. Accord-
ing to him, it was "to be seen at Cockneugh, found at
Caerlyth before mention'd near Kilpatric, as in Mr.
Lluyd's Collections." 2 Carleith, it will be remembered,
is a farm between the Chapel Hill and Duntocher.'
The Limes certainly passed through some of its
fields.

No. 5 (Plate XXXVII. 2; C.I.L. vii. 1135; Tit.
Hunt. p. 27) is also elaborately ornamented. Two
winged Victories, standing to front, perched each on
a globe, raise on uplifted hands an oblong tablet con-
taining the inscription. On either side of the tablet
is the characteristic pelta-shaped ornament, described
under No. 4, while on either side of the Victories there
stand, to left, Mars in full armour, leaning on spear
and supporting shield, and, to right, Valour also armed,
with parazonium in 1. and in r. a standard bearing the

letter is reproduced in extenso in The Scottish Antiquary, xv.
(1901), pp. 85 f. It was apparently first printed in Letters on Various
Subjects to and from William Nicolson, ed. John Nichols (1809), i. p. 337*,
No. *I35, and thence copied into the Scots Magazine for i8io(p. 829).
The reference to the original source is given by Mr. Gibb.
1 An Account of a Roman Temple, etc. p. 9.

THE DISTANCE-SLABS 279

legend \^Q — " the Valour of Augustus." The inscrip-
tion, except for the number of units (3240), is almost
the same as that on No. 3 :

IMP-C T-AELIO HADR

IANO-ANTONINOAVG

PPVEXLEGVl-

VICTRICS P F-

OPVS VALLI P

IXIMXICCXL- F

No. 5 was found in 1812 on the farm of Braidfield,
near Duntocher. The Ordnance Survey Map indicates
the exact spot as having been on the north side of the
Duntocher Burn, close to the ' Roman ' bridge. The
slab seems to have passed at once into the Hunterian
Museum. It measures 3 ft. 10^ in. by 2 ft. 6 in.1

No. 6 (Plate XXXVII. 3; CJ.L. vii. 1137; Tit.
Hunt. p. 28) has little or no ornament. The only
indication we possess as to its provenance is that
afforded by the name of the donor. It was presented
to the University in 1695 by Mr. Hamilton of Barns,
proprietor of Cochno, and was therefore probably found
near Duntocher. The inscription, which is enclosed
in a cable moulding, has for some inexplicable reason
been left unfinished ; the number of units has never
been cut. It reads :

IMP-C

T • AE • H ADRI ANO
ANTON I NO • AVG
PIO- P P- VEX- LEG

XX V V FEC-
P

1 A re-measurement of the stones has shown that in this and one or
two other cases the dimensions given in CJ.L. vii. and Tit. Hunt, require
to be slightly corrected.

28o THE LEGIONARY TABLETS

The blank in the last line is partly filled by the
legionary boar to right. The slab measures 2 ft. 5f in.
by i ft. 10 in.

No. 7 (Plate XXXVIII. i; C.I.L. vii. 1138; Tit.
Hunt. p. 30) appears to have been added to the Uni-
versity Collection between 1826 and 1845. Nothing is
known as to the circumstances of its discovery. Stuart
conjecturally assigns it to Duntocher, with the caveat
that "most probably it belongs to some of the other
stations."1 His provisional classification is suggested
by its resemblance to No. 4, to which it is almost
exactly similar, except that the framework which
encloses the lettering is richly ornamented, instead of
being a simple cable pattern. Other differences are
the number of units (4140), and the omission of the
usual dedication to the emperor, the latter doubtless
more or less of an accident, due to the peculiar
way in which the inscription is divided in the
complete design. The stone measures 2 ft. 3^ in.
by i ft. 74 in. The letters are :

LEG

M
AVG-F

PMIICXL

No. 8 (Plate XXXVIII. 2 ; C.LL. vii. 1133 a; Tit.
Hunt. p. 38) is the first of a pair of duplicates which
deserve to be very particularly noticed. The descrip-
tion of its general design has already been anticipated
in speaking of No. 2. Its dimensions are 2 ft. 10 in.
by 2 ft. 2 in. The inscription runs :

1 Caled. Rom. 2nd ed. p. 298.

PLATE XXXVIII

3- o. 9
THE LKGIOXARV TABLETS

THE DISTANCE-SLABS 281

IMP C-T
AEL- HADR

IANO AN

ToNI N O A/Q •

Plo-P P- VEX

LEG XX VV

FEC
P P iTT

The even number of units (3000) is a little remark-
able. Much more remarkable is the history of the
stone. We know it only from a cast. Found in 1865,
the original fell into the hands of Professor M'Chesney,
then American Consul at Newcastle-on-Tyne, who
insisted on sending it to the Museum at Chicago,
despite the indignant protests which were evoked in
the West of Scotland. It was destroyed there in the
great fire of 187.1. We are fortunate enough to possess
a fairly detailed record of the circumstances of its dis-
covery. " Hutchisonhill," writes Dr. John Buchanan,1
" is only a short distance west of the Peel Glen. In
the spring of 1865, the farmer, while trenching a field
on the southern slope, came upon a large stone at a
depth of about three feet below the surface. It was
lying flat in the stiff * till.' . . . The precise spot
where it was discovered is on the slope of the little hill
about six yards south from the line of the Antonine
Barrier, and therefore within, or on the Roman side."
He adds that " on the back of the stone are three
well-marked wedge-shaped receptacles or indentations
for fastenings, and there are marks of a fourth, proving
that this slab had been attached to a building of some

1 Trans, of the Glasgow Archaeological Society (ist series), vol. ii. p. 14
and p. 17.

282 THE LEGIONARY TABLETS

sort." It is much to be regretted that Dr. Buchanan
did not] specify exactly how far west of the Peel Burn
the find was made. Still, what he does tell us may
possibly suffice.

No. 9 (Plate XXXVIII. 3; C.I.L. vii. 1133; Tit.
Hunt. p. 37) is an almost exact duplicate of the preced-
ing, so far as the inscription goes, the only difference
consisting in the omission of the verb. The letters are
enclosed within a rectangular moulding of the cable
pattern, and there is no attempt at decoration except
for the boar, the usual emblem of the Twentieth
Legion, running to left beneath. The tablet is not
quite complete, but the portion that remains measures
2 ft. 6 in. by 2 ft. 4 in. The reading is :

IMP c-

T AELIO

HADRIANO

ANTONINO

AVG PIO P- P

V E X L E G • X X V [V]

p p m

This slab was turned up by the plough in 1847
about half way up the slope of the field that lies
between the Peel Glen and the fort of Castle Hill. Dr.
John Buchanan, in whose possession it was for a good
many years before coming to the Hunterian Museum,
has put it on record1 that it was " found in stiff clay,
more than three feet below the surface. But it was
not lying flat ; it was on its edge, showing that it must
have been thrust into a little pit opened in the clay
and afterwards covered up." He says also that " there
is a triangular indentation deeply cut in the centre of
the upper rim to receive a dook or tongue, to steady or

1 Trans, of the Glasgow Archaeological Society (ist series), vol. ii. p. 18.

THE DISTANCE-SLABS 283

fasten the stone in some kind of building."1 Else-
where he indicates his belief that it had been attached
to some of the stone structures within the fort, and
that prior to the final departure of the garrison it had
been carried down the hill and deliberately buried.2
The view as to its deliberate burial by the garrison
is not inconsistent with the facts as stated. It is
more difficult to reconcile with the circumstance that
a substantial portion of it is amissing. That its proper
place was within the station is even less certain,
although the base of a pillar, such as one would
naturally associate with the courtyard of the Principia,
was afterwards unearthed not very far away.3 The
point is important, as will presently appear.

It will be recollected that Nos. 8 and 9 are duplicates,
each recording the completion of the same number of
units (3000) by a detachment of the Twentieth Legion.
That they refer to the same section, and not to different
sections that happened to be of the same length, will
be evident from considerations to be discussed in
describing Nos. 10 to 13. Analogy suggests that the
duplication is best accounted for by the supposition that
either end of each completed section was furnished with
a commemorative tablet. If, then, we could ascertain
the original position of each member of any pair of
duplicates, we should be able to clear up once for all
the ambiguity in which the unit of measurement has
hitherto been involved. It is worth trying whether
such a result can be achieved in the case of the pair
now being dealt with. In no other instance do the
data so nearly approach completeness.

^ Ibid. 2 Stuart's Caled. Rom. 2nd ed. p. 311, footnote.

3 See supra, p. 161, Plate XX. i.

284 THE LEGIONARY TABLETS

Our information regarding No. 9 leaves little or
nothing to be desired. According to the Ordnance
Survey Map, it was found on the steep slope of the
Castle Hill, close to the southern margin of the great
Ditch, and about no yards west of the outer fringe of
the belt of trees by which the fort is enclosed. These
particulars are entirely confirmed by the independent
testimony of Mr. Alexander Houston of Over Croy,
who is still alive (1910), and who was actually behind
the plough that brought the slab to light in 1847.
The exact spot marked on the Ordnance Map is 950
English feet east of the middle of the Peel Glen Burn,
a point from which it will be convenient to start in
examining the evidence relating to its fellow. A
reference to Figure 1 3 will render the details intelligible.

Dr. Buchanan tells us that No. 8 was discovered by
a farmer on Hutcheson Hill, which " is only a short dis-
tance west of the Peel Glen." The words are vague,
but further details are given which render it possible to
be more precise. The stone was lying " about three feet
below the surface" and "about six yards south from
the line of the Antonine Barrier," somewhere "on the
southern slope" of Hutcheson Hill. As the Limes
ascends Hutcheson Hill on the south-west side and
descends into the Peel Glen down its north-east
face, "the southern slope" can only mean the part
that looks down on Cleddans. The lower of the
two fields lying on this slope is a fit subject for
drainage operations such as those which led to the
finding of the slab in 1865, and the distances suggest
that it was actually the scene of the discovery. Its
western edge is 2100 feet, and its eastern edge
1420 feet, east of the middle of the Peel Glen Burn,

286 THE LEGIONARY TABLETS

which we were to take as our starting-point. Add
the 950 feet which lay between the Peel Glen Burn
and the find-spot of No. 9, and we get 3050 and 2370
English feet as the limits for the interval between
the known find-spot of No. 9 and the probable
find-spot of No. 8. If we assume that No. 9 had
stood within the fort, we shall have to add not less
than 330 feet to each of these figures, making the
maximum and minimum limits 3380 and 2700 respec-
tively. The Roman foot was, of course, slightly less
than the English one. As the number of units re-
corded on the slabs is 3000, two conclusions seem to
follow. In the first place, the slabs were buried not
far from where they were originally set up. In the
second place, the unit of measurement was the foot, at
all events in all inscriptions where P, and not M p, is
used. Insistance upon P(assus) would create almost in-
superable difficulties. It would mean either that No. 8
had been carried from beyond Duntocher, a distance
of more than two miles ; or that the duplicates had
been erected, not at different ends, but at one and the
same end of the section, and that even so one or other
or both had been transported several hundreds of yards
in order to be put a foot or two below the ground.

Next comes an extremely interesting group of four,
which must all relate to one and the same length. In
view of the variety that generally prevails, the identity
in the number of units, absolute even to a fraction
(3666^), would be inexplicable on any other hypothesis.
The coincidence in the case of Nos. 8 and 9 might
conceivably have been accidental. This could not
possibly be so. The group falls into two sets of
duplicates, one belonging to the Second Legion and

THE DISTANCE-SLABS 287

the other to the Sixth. It looks as if the men of the
two legions had worked side by side, one perhaps
digging the Ditch, while the other built the Rampart.

No. 10 (Plate XXXIX. i; C.I.L. vii. 1132; Tit.
Hunt. p. 43) commemorates a piece of work executed
by a detachment of the Sixth Legion. The letters are
carved on a central panel, which is flanked on either
side by a variation of the characteristic pelta-shaped
ornament, the ends terminating in rosettes instead of
in griffins' heads. It will be seen that the stone-cutter
has blundered, making the F of F(idelis) in the fourth
line into P. It should also be noted that M • P is now
substituted for P in designating the unit. The tablet
measures 4 ft. 1 1 in. by 2 ft. 6 in., and the inscription
runs :

IMP CAESAR- T-AELIO
HADRIANO-ANTONINO
AVG- PIO- P P VEX ILL At O
LEG- VI • VIC'S • P- P
PER M • P- III • DCLXVIS

In the Tituli this slab is described as being "of
uncertain locality." It seems to have been first
published in 1699, by Fabretti, who states that it was
found "in colle Castelli"1 Sibbald too assigns it to
" Castlehill, near the new Kirk of Kilpadrick, Lennox-
shire."2 Gordon, on the other hand, places it at New
Kilpatrick,3 while the Monumenta, published by the
University of Glasgow in the end of the eighteenth
century, says it came from Summerston.4 Mr. Gibb's

llnscr. Antiq. Explic. et Addit, No. 620, p. 756.
*Hist. Ing. PL 3, Title. * Itin. Sept. p. 53.

4 From the Anderson MS. (see supra, p. 170) it is clear that this
attribution rested on the authority of Dr. Simson, the well-known

288 THE LEGIONARY TABLETS

researches prove clearly that its original site was
at or near Castle Hill. Mr. Urry, the " very curious
gentleman " who has been already quoted, saw it in
1698 "at Castle Hill, near East Kilpatrick, in the end
of a small thatched cottage." l This is confirmed by
Stukeley's citation from " Mr. LluycCs Collections,"
where "it is said ... to be in the Wall of Castlehill
Dike House upon Grahams Dike."2 No. 10 must,
therefore, belong to Castle Hill or its neighbour-
hood.

No.ll(PlateXXXIX. 2; C.I.L. vii. 1130; Tit. Hunt.
p. 34) measures 4 ft. 5 in. in breadth by i ft. 1 1 in. in
height. The inscription occupies the centre of the
slab, being placed within a plain rectangular moulding.
On either side are groups of figures. On the left is a
winged Victory, three-quarter face towards right, about
to crown a helmeted horseman who gallops r., thrust-
ing downwards with spear ; beneath are two captives,
seated opposite each other, with heads turned to front,
and hands tied behind their backs ; between them is
what seems to be a standard. On the right is an
eagle, with wings open and head 1., standing to front
on the back of a Capricorn r. ; beneath, beside another
standard, is a third captive, seated 1., with head turned
to front and hands tied behind his back. The capri-
corn is the emblem of the Second Legion. The rest
of the symbolism is too obvious to call for comment.

Professor of Mathematics, who seems, however, to have been confusing
it with No. n, where the number of units is the same; see infra,
p. 292.

1 The Scottish Antiq. xv. (1901), p. 85 = Letters to and from Wm. Nicol-
son, i. p. 338, where the last line has been omitted in the transcription.

2 Account of a Roman Temple, etc. p. 9.

PLATE XXXIX

i. No. 10

2. Xo. ii

3. No. 12
THE LEGIONARY TABLETS

THE DISTANCE-SLABS 289

There is a blunder in the fourth line of the inscription,
PEP being cut instead of PER. The whole reads :

IMPCAESTITOAELIO
HADRIANOANTOIsfNo
AVG-PIO- PP- LEG- M
AVG-PEP.MP.nTDC

LXVI • S

This stone was presented to the University of
Glasgow in 1694 by Mr. John Graham of Dougalston.
Gordon assigns it to Castle Hill : " Out of the ruins of
this Fort, the noblest Roman Stone that ever was found
in Scotland was dug."1 Horsley too says it "belongs
to the fort at Castle-hill'"2- Mr. Gibb, on the other
hand, would put it at Balmuildy, " though the recorded
evidence does not, strictly speaking, bear out that
opinion."3 His conclusion is admittedly illogical, and
it may therefore be set aside at once. At the same
time the arguments he brings forward more than
justify his refusal to accept the customary attribution
to Castle Hill. A fresh examination of the whole
question thus becomes imperative. And it is well to
recall the cardinal fact by which that examination must
be governed — the absolute identity in the number of
units on each of the four slabs we are considering.
This renders it virtually certain that all the members
of the group belong to one length. It also greatly
increases the risk of confusion between their find-spots.

It is, of course, always possible that Gordon and
Horsley may be right; either No. n or No. 12 may
originally have stood close to No. 10, a Sixth Legion
slab and a Second Legion slab being erected side by

1 Itin. Sept. p. 52. *Brit. Rom. p. 195.

3 The Scottish Antiquary, xvi. (1902), p. 180.

29o THE LEGIONARY TABLETS

side at either end of the length. Still, it is equally
possible that they may be wrong, seeing that Gordon
(who is clearly the ultimate source for Horsley) gives
no authority for his statement. Indeed, a significant
mistake in his account of No. 1 1 arouses grave sus-
picion. Both in his plate and in the Latin rendering
in his text he omits the number of the legion, although
it is plain enough on the stone, while in his English
translation the Sixth Legion actually appears instead
of the Second.1 In other words, the inscription on
No. 1 1 is transformed into a duplicate of that on
No. 10, which we already know to belong to the
neighbourhood of Castle Hill. It would be difficult
to imagine conditions more symptomatic of con-
fusion. Thus much from the negative side. But
there is positive testimony in favour of an alternative
provenance.

Mr. Urry in noting the inscription on No. n, which
had been removed to Glasgow four years before his
visit to Scotland, adds simply : " This was found at
Dougalston."2 As Dougalston House is fully a mile
north of the Limes, Urry's phrase is not to be taken
literally. At the most it means no more than that the
tablet was discovered within the bounds of the estate.
Even that is something, for Mr. Gibb has shown that
the estate did not include the property of Castle Hill.3
And there are other witnesses who take us further
forward. In June, 1729, Wodrow, the well-known
ecclesiastical historian, then minister of Eastwood,

lftin. Sept. p. 52, and Plate 11, Fig. 3.

2 The Scottish Antiquary, xv. p. 86 = Letters to and from Wm. Nicolson,
i. p- 338-

3 The Scottish Antiquary, xvi. p. 179.

THE DISTANCE-SLABS 291

paid a visit to John Graham of Dougalston, grandson
of the John Graham who had presented the stone to
Glasgow in 1694. At this time the Itinerarium
Septentrionale had been published for three years,
and it was in all probability familiar to Wodrow, if
not also to his host. The former would be specially
interested in No. n, for he had had it under his
charge for four years, having been Librarian to the
College of Glasgow from 1697 to 1701. Particular
weight therefore attaches to what he tells us in con-
nection with his visit. It almost assumes the air of a
deliberate correction of Gordon : " This moneth I was
at Dougalstoun, throu whose ground the old Roman
Wall goes. I had the pleasure to see that old vestige
of the Roman greatnes. The wall is levelled with the
ground or filled up with every year's grouth and dust
many years since. Houever, the tract of it is very
plain ; from Kilpatrick to Kirkentilloch it runs all
along on an eminency. Dougalston getts all his
stones for a large park dyke from it, and the people
just digg under a foot of earth and find them in plenty
for raising. At the place where they wer digging,
the heuen stone with inscription, gifted by Dougalston,
1694, to the Colledge, was turned up."1

The extract just quoted is aptly illuminated by a
passage which Mr. Gibb cites from the Britannia
Romana. After speaking of Gordon's discovery of
the stone foundation of the Rampart, Horsley pro-
ceeds: "And since that gentleman made his survey,
this foundation of stone has been laid open and dug
up for near a mile together, from the middle of
Ferguston moor east of New Kirkpatrick almost to

1 Wodrow's Analecta, iv. p. 66.

292 THE LEGIONARY TABLETS

the village of Simmerston. The Laird of Douglaston
has used the stones for building a park wall."1 This
narrows the limits considerably. But the really
important words are "almost to the village of Sim-
merston" They immediately suggest the description
which, on the authority of Professor Simson,2 was
wrongly attached to the Monumenta Plate of No. 10:
' Inventus est lapis hie prope Villam de Summerstoun
ad ripam fluminis Kelvin. ' 3 Who can doubt but that
Simson's information, quite possibly obtained from
Wodrow, originally applied, not to No. 10, but to
the Dougalston stone ? Its transference may conceiv-
ably have been due to a not unnatural reluctance to
traverse the very explicit assertion of the Itinerarium,
almost a canonical book in the eyes of eighteenth
century antiquaries. There is a final point. To any-
one who knows the ground, the words "near the farm
of Summerston on the bank of the river Kelvin "
will appeal as a good general description of the spot
where No. 13 was discovered in 1803. As a matter
of fact, Stuart's actual expression regarding the latter
is "found at Millochan, not far from Summerston."4
We can hardly avoid concluding that No. 1 1 and
No. 13 were originally companions.

No. 12 (Plate XXXIX. 3; C.I.L. vii. 1126; Tit.
Hunt. p. 44), the measurements of which are 2 ft. 10^ in.

* Brit. Rom. p. 163.

2 See supra, p. 287, footnote. It may be added that Simson was eighty
years of age when the scheme of the Monumenta first took shape, and
that it is uncertain whether it was issued during his lifetime.

3 Monumenta Imperil Romani, etc. PI. IV. No. n is represented on
the plate immediately preceding, a juxtaposition which would increase
the chances of confusion.

*Caled. Rom. 2nd ed. p. 315, footnote.

THE DISTANCE-SLABS 293

by 2 ft., is singularly plain. It has suffered much
from exposure. The inscription, which is enclosed by
a beaded moulding, is a duplicate of that on No. 1 1 :

IMP-CAESTITOAELIO
HADRIANOANTONINO
AVGPIO-PP-LEGI7.AVG
PER M-P-III-DCLXVIS

This slab was one of the earliest of the inscriptions
from the Scottish Limes to be published. It first ap-
peared in Scaliger's Thes. Temp. Eusedii(i6o6)^ having
been communicated to Scaliger by ' Crispinus Gericius.'
A year later Camden, who had received a copy of it from
' Servatius Rihelius,' included it in the sixth edition of
his Britannia? According to Camden, it was 'at
Cadir fixed in the wall of a house ' — " ad Cadir parieti
domus infixa " — obviously Cadder House, then as
now the property of the Stirlings of Keir. Scaliger's
description is at first sight a little puzzling. He
speaks of it as being ' in a very old tower in Glas-
gow'— ''''in turri quadam pervetusta in Glasco" But
a reference to the MS. authorities quoted by Hiibner
(in his commentary on the stone in C.I.L. vii.) will
show that Scaliger was merely abbreviating, and that
there is no real discrepancy between him and Camden.
It will also show that, when Gericke and Reichel
visited Cadder, the mansion was occupied by Lady
Stirling, widow of Sir James Stirling, who died in
1588, and one of whose daughters had been married
in 1572 to John Napier of Merchiston, the inventor
of logarithms.3 Scaliger's expression further suggests
that at this time the slab was not built into the house

1 Animadversiones, p. 175. 2P. 699.

3 See Sir Wm. Fraser, The Stirlings of Keir, p. 43.

294 THE LEGIONARY TABLETS

itself, but into the wall of the old square keep, the
ruinous foundations of which were not finally rooted
out till about I8I6.1 Before Gordon's day it had
been removed to the west wall of the mansion-house,
doubtless when the walls of the keep were demolished
and the house subjected to one of the extensive
processes of reconstruction that it has undergone in
the course of its long history. The author of the
Itinerar'mm saw it still in this position in 1726. In
the interval that elapsed before the date of Horsley's
visit (1731), it had been "taken down and placed
within."2 The most probable date for its presentation
to the University is I735.3

There is practically no external evidence as to its
original position. The "digged out of the ruinous
ditch of Graemes Dyke " of the Morton MS. is
probably true, but it is not very enlightening. Nor
can we place any reliance on Sibbald, who, after
saying that No. 14 was found "at Cadir Mannor,"
adds that No. 12 was found "near to the same Place
in the Ruines of the Wall."4 Some of his attribu-
tions are notoriously wild, and here he seems to
have been merely guessing. We are accordingly
thrown back on the clue provided by the curious
number of units recorded in the inscription (3666^).

1 Information from the Rev. J. B. A. Watt. They were removed
to admit of the lawn in front of the house being levelled. Previously
they had been visible as grass-grown mounds.

zBrit. Rom. p. 198.

3 MS. note on margin of the University Library copy of Gordon's
Itinerarium : " N.B. Mr Stirling of Kier caused take this stone out of the
wall of his house, being much spoiled by being exposed to the weather,
and sent it to the University of Glasgow, and here it now stands with the
rest A.D. 1735." * Hist. Inq. p. 50.

THE DISTANCE-SLABS 295

As has been already pointed out, that connects it
unmistakably with the length to which Nos. 10, n,
and 13 all belong. This being granted, there remain
two possibilities. On the supposition that only two
sets of men worked on the section, there would be
only two stones at each end. We should then possess
all four, and No. 12 would come, like No. 10, from
Castle Hill, seeing that No. n and No. 13 came
from Millochan. Alternatively, however, it is possible
that Rampart, Ditch, and Military Way had each
been assigned to a different set of men — two of the
three sets being drawn from the Second Legion,
which alone was present in full strength. On such
an assumption the total number of distance-slabs for
the section would be six, four of which still survive,
and No. 12 might quite well have stood at Millochan
alongside of Nos. to and 13. The choice, therefore,
seems to lie between Millochan and Castle Hill.

The conclusion we have been driven to is open to
two objections, both of which are prima facie reason-
able. In the first place it runs directly counter to
the express assertion of Gordon. After giving an
account of Bemulie, he proceeds : " At this Place
likewise have been dug up several Inscriptions and
engraved Stones, shewing, that the second Legion
Augusta lay there. Most of these Stones are now
brought from thence to Colder-House^ belonging to
Mr. Stirling of Kier, on whose Ground are the ruins
of Bemulie. The Predecessors of this Gentleman
built them within the Walls of Colder-House^ for
Preservation."1 Circumstantial as this statement
looks upon the surface, there is good reason to

1 I tin. Sept. p. 54.

296 THE LEGIONARY TABLETS

believe that it is nothing but a surmise, or, at the
best, the repetition of an empty tale not unlike that
which, a hundred and thirty years later, led so excel-
lent a historian as Sir William Fraser to attribute
the solitary inscription still at Cadder House (No. 19)
to the ' castellum ' within the policies.1

After making the general assertion quoted above,
Gordon goes on to support it by reference to particular
instances, these being No. 19 and the slab we are
now discussing. But the one thing that is certain
about the provenance of No. 19 (C.I.L. vii. 1127) is
that it did not come from Bemulie. A passage in
the anonymous Cotton MS.,2 cited by Hiibner in his
commentary on the inscription, specifically mentions
that the stone was at Cadder "ex dono dni Naperi"-
that is, as a gift from Napier of Merchiston, Lady
Stirling's son-in-law. Bemulie was on the Cadder
estate, and, if the stone had been found there, it
would have been unnecessary for an outsider to pre-
sent it. It has always been known that Napier was
alive to the importance of Roman inscriptions. He
was, in fact, the first to publish the famous dedication
to Apollo Grannus, discovered at Inveresk in the
reign of Queen Mary.3 In or about 1573, shortly
after his marriage with Miss Stirling, he took up his
residence at Gartnes, in the valley of the Endrick,
where certain lands had been assigned him by royal

1 The Stirling's of Keir (1858), p. 570.

2 The MS. is a miscellaneous collection of historical and antiquarian
papers of Cotton's and Camden's. Professor Haverfield, who has
examined it, tells me that the particular phrase to be quoted presently
is, he thinks, certainly in Camden's handwriting.

3 See A Plaint Discovery of the whole Revelation of Saint John,
Edinburgh, 1593, p. 210.

THE DISTANCE-SLABS 297

charter ; and there he lived for some five and twenty
years. His property was well to the north of what
had been the Roman frontier. But the parish of
Killearn, within which it lay, is immediately adjacent
to New Kilpatrick, and in his comings and goings
to Glasgow, the town nearest to his house, Napier
must often have crossed the Wall close to New Kil-
patrick fort. It may have been in the course of one
of his journeys that he heard of No. 19, and managed
to secure it for his relatives. At all events, Gordon's
story of No. 19 is valueless, and what he says regard-
ing: No. 1 2 must also stand discredited.

o

There remains the second objection. Cadder
House, it might be urged, is fully two miles from
Millochan and more than five from Castle Hill.
Would it be in the least likely that a heavy stone
would be transported such a distance ? The best way
to supply an answer is to draw a clear distinction.
Where inscribed stones from the Wall have been used
simply as building material — in cottages, as was No. 10,
or in byres, as was No. 20 — we may be fairly confident
that they have not travelled far from their original
location. It is different when we meet with them
in country-houses. They owe their position there to
the fact that they were looked upon as curiosities.
Indeed, towards the end of the sixteenth and the
beginning of the seventeenth century, when the in-
terest in them first began to awaken, it was evidently
a fashion with country gentlemen to have one or two
of them set up in or about their mansions. Thus,
No. i was conveyed to Mugdock, Nos. 2 and 3 were
probably ferried across to Erskine, No. 4 was let
into the wall above the gate of Cochno House, and

298 THE LEGIONARY TABLETS

No. 14 was actually carried as far as Dunnottar in
Kincardineshire, where it was placed " in a Porch "
in the Castle.1 Napier's present of No. 19 is in itself
a proof that the Stirlings of Keir and Cadder did not
wish to be behind their neighbours. If No. 19 came
from New Kilpatrick or its vicinity, No. 12 may well
have been brought from Millochan or even from
Castle Hill.

No. 13 (Plate XL. i ; C.I.L. vii. 1131) is preserved at
Levenside, near Dumbarton. The dimensions of the
stone are 4 ft. 9 in. by 2 ft. 5^ in.2 In design it closely
resembles its duplicate, No. 10, the inscription being cut
on a central panel, which is flanked by two pelta-shaped
ornaments. The latter are plainer than usual, the horns
terminating in griffins' heads and the central projection
in a rosette. The work is coarse, and some of the
abbreviations are unusual. The whole surface is
now much weathered. Owing to a mistake on the
part of the draughtsman who engraved the illustration
in Stuart's Caledonia Romana? an inaccurate reading
has gained currency ; the number of units is there
represented as 3665. As a matter of fact, the corre-
spondence in this respect with No. 1 1 is exact. Both
have 36667^. The following is the correct version :

IMP CAES-T
AELIO - HADRIA
ANTONINoAVG
PIOPPVEXILLA
LEGVl VIC P.P.
PER M -P. IIIDClivis

1 Gibson's Camden (1695), p. 939.

2 1 have to thank Mr. Archd. Macdonald, Dumbarton, for measuring
the stone, and also for help in getting it photographed.
3 PL xvi. 3.

PLATE XL

i. No. 13

2. No. 14

3- No. 15
THE LEGIONARY TABLETS

THE DISTANCE-SLABS 299

The circumstances connected with the finding of
this slab are well known. In the second edition of
Stuart's Caledonia Romana we are told, presumably on
the authority of Dr. John Buchanan, " that it was dis-
covered in 1803 in the formation of a deep drain on
the farm of Low Millochan. A person who happened
to be present on the occasion has pointed out the
precise place, about 100 yards within the Antonine
Wall, where it commences to ascend the height called
the Temple, and very near the present Farm-house."1
The fact that it was lying so far south of the Rampart
may be due either to its having been carried or to its
having originally stood on the line of the Military Way.

The terms of the description are too vague to admit
of the exact spot where it lay being identified. But,
measured on the Ordnance Survey Map, it can hardly
have been less than 1000 yards along the line of the
Limes from the point where the latter quits the north
bank of the Kelvin opposite Bemulie. According to
Roy the distance from Bemulie to Castle Hill was 7050
yards. Subtracting 1000 from Roy's reckoning, we get
6050 yards east of Castle Hill as the approximate
' find-spot' of No. 13, and also (there is good reason to
believe) of No. 1 1. But 6050 English yards are roughly
equivalent to 3700 Roman paces. The interval that
actually separates the probable 'find-spot' of No. 10
from the approximate 'find-spot' of Nos. n and 13,
calculated in Roman paces, thus comes very close
to the number of units (3666^) mentioned in the
inscriptions. The approach is too striking to admit
of any other explanation than that, the ' feet ' on
Nos. 1-9 notwithstanding,2 P on Nos. 10-13 is meant

1 Op. cit. p. 315, footnote. 2 See supra, p. 286.

300 THE LEGIONARY TABLETS

to indicate 'paces.' We are thus brought back to
the point of view which Mr. Bates was prepared to
maintain on a priori grounds.1

No. 14 (Plate XL. 2 ; C.I.L. vii. 1143 ; Tit. Hunt.
p. 46) has had a very chequered history. It is
usually classed along with the stones from Bemulie,
but the evidence for such an attribution is very
slender. The only thing that seems clear is that,
as the method of reckoning shows, it must have
been found east of Castle Hill. When first pub-
lished by Camden in 1607, it was at Dunnottar
Castle in Kincardineshire. Hitherto its previous
history has been altogether unknown. A scrutiny of
Sir Robert Sibbald's manuscript ' Collection of papers
and information in order to ye description of Scotland,'
now in the Advocates' Library, has thrown a little light
upon the matter. The collection includes "A Note of
some remarkable things within the Sheriffdom of the
Merns by Mr. John Keyth, sometime Minister at
Dunotir," written in 1642. In this there is mentioned
as "most worthy of observation" a stone "that was
found in a Dike in ye Borders of England and Scot-
land, brought to Dunotir by Earle George Marischall
sometime Ambassadour to Denmark for Queene
Anne." The ' Dike ' is subsequently described as
" Vallum Hadriani, now Grame's Dike." Earl George
succeeded to the title in 1581, and we may assume
that the removal of the stone took place at a later date.
He was a very prominent man at Court, as his selec-
tion for the task of bringing home the royal bride is
enough to prove. He must have known George
Buchanan well. It is therefore hardly doubtful but

1 See supra, pp. 270 f.

THE DISTANCE-SLABS 301

that No. 14 was one of the inscriptions that Buchanan
had in view in the familiar passage in his History.1
From Dunnottar the slab passed, about 1725, to
Marischal College, Aberdeen, the authorities of which
presented it in 1761 to the University of Glasgow,
believing that its appropriate resting-place was beside
the other inscriptions from the Limes.

The size is 3 ft. 2^ in. by 2 ft. 10 in. A beaded
rectangular moulding, with a pelta-shaped ornament
on either side and a richly floreated border above and
beneath, contains the lettering, which runs :

IMP CAESARI

TAELIOHADRI

ANOANTONINO

AVG PIO P- P-

VEXILLATIO

LEGXXVALVICF

PER MIL Pill

This and No. 15 are the two stones on which the
emperor's name is unmistakably in the dative. After
what we have learned from Nos. 10-13, we need not
doubt but that the units (3000) are in this instance
paces.

No. 15 (Plate XL. 3; C.LL. vii. 1121 ; Tit. Hunt.
p. 55) is a very large slab, the largest of all except
No. 17. It measures 5 ft. 2 in. in length by 2 ft. 6 in.
in breadth. Although so much space for decoration
was available, the treatment is singularly plain, the
lettering being contained in a central tablet with a
pelta-shaped ornament on either side. It has generally
been supposed that the inscription is incomplete, as no
numerals have been cut after M . p. But it is also

1 See supra, p. 41.

302 THE LEGIONARY TABLETS

possible, though hardly likely, that the particular length
whose completion is commemorated covered only a
single mile. The text is as follows :

IMP-CAESARI -T-
AELIO- HADRIA/0
ANTONINO AVG-
PIO- P- P- VEXILLA
LEG- VI -VIC P F
PER- M P-

Maitland records that the stone was found a few
years before he wrote (1757) "a little bewest" Inch-
belly Bridge, which crosses the Kelvin between
Kirkintilloch and Auchindavy.1 It must thus have
stood quite close to the slab next to be described, the
two commemorating the completion of adjacent lengths
by different sets of soldiers. The exact date of its
discovery was somewhere about 1740. Writing from
Edinburgh on July i6th of that year, Sir John Clerk
sent a sketch of it, as a novelty, to Roger Gale.2
Four years afterwards it was added to the Glasgow
collection by purchase.3

No. 16 (Plate XLI. i; CJ.L. vii. 1122) is in the
Hunterian Museum, although not included in the
Tituli. Its omission is due to an accident ; it was
lost for a good many years and was not recovered
until after the book had been published. The shape
is peculiar ; it is not so much a tablet as a solid
block — i ft. nj- in. long, i ft. 5 in. high, and i ft.
2 J in. deep — hewn and dressed as if it were to be built

^ Hist, of Scot. i. p. 178.

2 Bibliotheca Topographica Britannica, iii. p. 343.

3 The records of the University show that in 1744 a small sum was paid
to Professor Simson "for his outlays in purchasing and bringing from
Kirkintilloch a large stone from the Roman Wall with an inscription "
(Coutts, Hist, of the Univ. of Glasgow, p. 332).

Pi, ATE XL I

i. No. 16

2. No. 18

3. No. 19
THE LEGIONARY TABLETS

THE DISTANCE-SLABS 303

into a wall. In execution it is one of the rudest of
all. An oblong frame, with an ansa-like projection at
either side, holds the first two lines of the inscription.
In the left-hand portion of the space beneath are the
last two lines of the inscription, and in the right-hand
portion is the forepart of a boar running r. towards a
tree, an emblem that has already been more than once
encountered. The letters are :

LEG XX

V- V FEC
M P III • P
ITI C C C I V

The stone-cutter has apparently blundered into a
repetition of P • 1 1 1, unless he has combined on one stone
the methods of reckoning by passus and by pedes, using
the latter for distances less than a mile. The alterna-
tive is a very unlikely one,1 and the hypothesis of a
mistake accords well with the poorness of the work-
manship.2 The number of units will then be 3304.
Regarding the finding of the stone, Dr. John Buchanan
says : " It was lying, on its inscribed face, about three
feet under the surface, in the very centre of the Roman
ditch, which had been ploughed cross-wise, at this
locality, from time immemorial, and was discovered
during a deeper course of ploughing in 1789."'
Combining this with what Stuart says in his text : "at
that part of the line where it traverses the farm of

1 It was, however, adopted by Mr. C. J. Bates, who made it one of the
chief arguments in favour of the use of a double method of reckoning
(Arch. Ael. xix. p. 107). As we have seen, his theory does not require
such a doubtful kind of support.

2 The failure to make it clear that only a Vexillation of the Legion was
concerned, is another indication of the same sort.

3 Stuart's Caled. Rom. 2nd ed. p. 325, footnote.

304 THE LEGIONARY TABLETS

Eastermains, about three-quarters of a mile to the east
of Kirkintilloch,"1 we are brought, as already indicated,
close to the 'find-spot' of No. 15.

No. 17 (Plate XIV. ; C.I.L. vii. 1088), which is now
in the National Museum of Antiquities in Edinburgh,
is the largest and finest of the distance-slabs that have
as yet come to light. It may owe this special char-
acter to the fact that it marked the eastern termination
of the Limes. It was found in 1868 on the little rocky
promontory of Bridgeness, close to the shore of the
Forth, on the lands of Mr. Henry Cadell of Grange,
who at once presented it to the nation. Like all the
others regarding which we have detailed information,
it had the appearance of having been deliberately
hidden. When discovered, it was lying " with its face
down in a sloping direction." : It is 9 ft. 2 in. long
by 3 ft. ii in. high, and is elaborately decorated.

A rectangular beaded moulding in the centre contains
the inscription. Two pelta-shaped ornaments, with
the horns and central projections all alike ending in
griffins' heads, are placed one on either side of the
central tablet, four rosettes occupying the vacant spaces
beside the corners of the tablet. The overhanging
ledge above is carved in a conventional pattern. On
the left, within a framework of pillars is a horseman,
armed and helmeted, galloping r. ; he carries a shield
on his 1. arm, and with spear held in r. thrusts down-
ward at four naked Caledonians ; one of the Caledonians
is already decapitated, but has been armed with a spear
and an oblong shield ; another who is just falling dead

1 Stuart's Caled. Rom. 2nd ed. p. 325, footnote.

2 See Mr. Cadell's letter, printed in Proc. Soc. of Antiq. of Scotland,
1871, pp. 109 f.

Pl.ATK XLII

THE SUOVETAURILIA ON THE BRIDGENESS
LEGIONARY TABLET (No. 17)

THE DISTANCE-SLABS 305

has an oblong shield in his 1. hand, while a sword lies
by his side ; the remaining two are defenceless. On
the right, likewise within a framework of pillars, is a
representation of the suovetaurilia (Plate XLII.) ; the
animals to be sacrificed are advancing 1. towards a
figure seated r., these occupying the foreground; behind
them is a youth, the tibicen, playing the double-flute ;
in the background are five male figures, the foremost of
whom is holding a patera over an altar, while the one

LEG

behind him bears aloft a standard with the letters 1 1 .

AVG

The principal inscription is as follows :

IMPCAESTITO-AELIO
HADRIANTONINO
AVG PIO P.P. LEG -ll
AVG-PERMPfTTiBCLII
FEC

The slab, therefore, records the completion by the
Second Legion of 4652 paces of the Limes. As to
the symbolism it need only be remarked that the repre-
sentation of the suovetaurilia connotes the lustratio, or
ceremony of purification, which played so prominent a
part in the public and private life of the Romans. The
suovetaurilia were peculiarly associated with the god of
war; they were B. piaculum Marti,1 and every Roman
army was solemnly lustrated before taking the field
against an enemy.2 The companion picture of the
victorious horseman trampling his vanquished foes
underfoot requires no comment. Taken together, the
two are meant to be an epitome of the Caledonian
campaign, a monument set up to celebrate its close,
strictly analogous — longo intervallo, it is true, — to the

1Livy, viii. 10, 14.

2 Tacitus, Annals ; vi. 37, is a well-known instance.
U

306 THE LEGIONARY TABLETS

column that stands in the Forum of Trajan at Rome.
There also we have the suovetaurilia * and the Romans
dealing death to the hapless barbarians. One other
point must be mentioned. Along with the tablet there
were found " a considerable quantity of squared sand-
stones, roughly dressed."2 These had apparently
formed part of a wall to which it had been attached ;
there are holes in the back, as if for fixing it. We
have already heard of such holes in connection with
Nos. 8 and 9. Doubtless the arrangement indicated
was the usual one.

The whole of the surviving distance-slabs have now
been passed in review. It is certain that there were
once many more. Some must have been destroyed.
Others may still be resting safely beneath the earth.
To complete the record, we may note one or two frag-
ments that had probably belonged to the same class.
The imperfect inscription printed as moa in C.f.L.
vii. undoubtedly did so. It must have come from
near Bar Hill. When copied, it was l in pariete horti
domini de Kilseith! It has long since disappeared.
To this we may add the double fragment discovered
in 1868 at Arniebog, now in the Hunterian Museum
(Plate XLVI 1 1. 4 ; Tit. Hunt. p. 76). No letters at all
are visible, but the height (2 ft. 10 in.) is very suitable,
while the figure of a kneeling captive is also highly
appropriate. Possible, but less likely, is the larger
of the two pieces of sculpture built into the house at
Nether Croy, and figured on Plate LI 1 1. 2. That it
once had an inscription is fairly certain. Those who
saw it first thought they could decipher the letters vi

1 See Cichorius, Die Trajanssaule, Plates ix. xxxviii. and Ixxvi.
zProc. Soc. of Antiq. of Scotland, 1871, p. no.

THE DISTANCE-SLABS 307

(C.I.L. vii. 1 105). But the decoration is hardly such
as we usually find upon the distance-slabs. Perhaps
a faint memory of yet another is preserved by Mait-
land, who, dealing with Old Kilpatrick, speaks of
"another stone I saw lying at the threshold of the
door of the most eastern house of Fenydike," — that
is, Ferrydyke, — "of twenty-eight inches square, and
six inches thick, with a border of three inches and a
half curiously wrought ; but the inner part being
greatly worn by people treading thereon, I could
only discern that there had been an inscription there,
which was then unintelligible. This stone, which was
dug up at the eastern end of the house where it lies,
I take, by its form, to have been a legionary stone
erected in the wall at or near the place where it was
found, setting forth the name of the legion or vexillation,
with the part of the wall erected by either of them."1

The inscriptions we have been discussing are records
of labour that had been apportioned by skilled officers
and executed by disciplined troops. In view of this,
it may seem at first sight surprising that there should
be such marked differences between the distances
registered. A little consideration will show that it is
natural enough. Setting aside for a moment the slabs
from the west of Castle Hill, and limiting ourselves to
those on which the measurements are in paces, we
can see, upon reflection, that strict uniformity would
really have been inconsistent with the ideal towards

1 Hist, of Scot. \. p. 183. The carved stones of which he speaks in the
next paragraph were not, however, Roman. At all events, the detailed
description of one of them communicated by Dr. Lettson to the Society
of Antiquaries in 1776 (Cough's Camden, 2nd ed., iv. p. 103, footnote)
makes it certain that it was an early Christian monument. This is sug-
gested even by Maitland's own vague language.

308 THE LEGIONARY TABLETS

which the responsible officers had to strive. We may
presume that different sets of men would be working
on adjacent sections. Sound organisation would there-
fore demand that the sections should be such as could
be finished in approximately equal times — a result
that would have been impossible of attainment, under
the conditions that prevailed, had the distances been
mathematically equal. Account would inevitably have
to be taken of the nature of the ground, and the
length of the sections graduated accordingly. The
strip which included the rocky face of Croy Hill, for
instance, would be far shorter than the 4652 paces that
stretched inland from the Firth of Forth. Further, it
seems not unlikely that the successive lengths were
originally staked off by the help of the eye alone, and
that the process of exact measurement was postponed
until the length had been completed. Only on this
supposition is it possible to understand the choice of
so extraordinary a figure as 3666^.

But what of the different methods of reckoning
employed to west and to east of Castle Hill? Here,
again, reflection will furnish a reply. Mr. Bates, it
will be remembered, suggested that the Wall was
commenced from the west, and that the legionaries,
not fully realising the magnitude of their task, light-
heartedly began to measure their progress in feet; by
the time they got to Castle Hill "they saw that in a
great work like the limes it was more practical to talk
of so many paces."1 That is surely too ingenious.
It is much more probable that the Wall was com-
menced from the east, and that the engineers, on
approaching the Clyde, found themselves with an

^ee supra, p. 271.

THE DISTANCE-SLABS 309

odd length, so to say, upon their hands. The obvious
way to deal with such a length was to subdivide it.
The whole would thus be finished off more quickly,
while the burden on all the working parties would be
kept equal. Unless such a course were adopted, a
large body of the soldiers would be left idle. And,
when subdivision became desirable, the ordinary
unit of measurement below the pace was the foot.
The explanation of the abrupt change of standard is
therefore comparatively simple. In this connection it
is relevant to point out that the distance from Castle
Hill to the Clyde is rather more than 7000 English
yards, or about 4200 Roman paces — a length that
agrees wonderfully well with the average for the rest
of the line, so far as the information at our disposal
enables us to estimate it.

Considering that they were all produced about the
same time and all destined to serve the same general
purpose, the variety in the style and execution of the
slabs is remarkable. Some of the tablets, particularly
No. 5, are quite creditable specimens of the stone-
cutter's art. Others are rude in the extreme. Two
cases of uniformity in design were noted — Nos. 2 and 8
and Nos. 4 and 7 ; and, duplication apart, there were
two instances in which a certain affinity betrayed itself
between the inscriptions. These, however, are the
exceptions. As a rule, each artificer has been left to
the freedom of his own will. If we disregard the note
of triumph that breathes through the figure-groups, the
solitary feature that can be called characteristic is the
use of the pelta-shaped ornament.

It is worth remarking that this ornament is not a
local peculiarity ; it is simply a fashion of the time. It

3io THE LEGIONARY TABLETS

occurs, though less regularly, in the sculptures from
the English Wall, occasionally in a distinctly degraded
form. On the Rhine it appears much more rarely,
although one might cite as a good German example
a tombstone in the Museum at Mainz, erected by a
Roman Knight, Adjutorius Lucilianus, to the memory
of his mother, Ulpia Lucilla.

One other point must be noted. Dull and dingy as
the inscribed slabs may seem to-day, we are justified
in believing that they once made a braver show. The
probabilities are all in favour of their having been
brightly, if crudely, coloured. No vestige of anything
of the sort is visible upon them now ; the marks of
gilding on No. 14 belong to a comparatively modern
phase. But contemporary monuments of a similar
character abroad, along the German frontier for
instance, often retain traces of pigments, proving that
they had been brilliantly tinted.

Such are the distance-slabs themselves. What of
the information that they furnish ? The character of
the figure-groups that appear among the sculptures is
significant. If we except the three legionary crests —
the boar, the Capricorn, and the Pegasus — and the
purely decorative genii on Nos. 2 and 8, all alike
speak plainly of victory won. The erection of the
Limes, then, was not an incident in a campaign
or a series of campaigns. It marked the close of a
successful war. The work was allotted among squads
of men drawn from the legions, that is, from the troops
that had constituted the regular field force. To the
auxiliaries would be assigned the task of making sure
that their comrades were unmolested at their labour.
Probably they had also to build the forts which were

THE DISTANCE-SLABS 311

to serve as their own permanent quarters. Two or
three squads from different legions seem to have
laboured side by side, one toiling at the Rampart,
the others at the Ditch or the Military Way. As each
squad finished its appointed section, it set up at either
end a memorial slab dedicated to the Emperor and
recording the exact amount of work it had accom-
plished. That Bridgeness was the starting-point is a
natural deduction from the fact that the distances were
reckoned in feet to the west of Castle Hill, and in paces
to the east of it. Apparently the tablets were not let
into the turf Rampart, but were rather placed in a
framework of stones specially built to receive them.
Doubtless they were raised some little height above
the ground, a circumstance which, combined with
their bright colouring, would render them conspicuous
a good way off. In this position we may suppose
them to have remained until the Limes was definitely
abandoned by the Romans. There are indications
that they may then have been taken down by the
retreating soldiery and deliberately concealed.

These are general inferences. More particular
conclusions are open to us too. But, before attempting
to draw them, it is desirable to bring together the
few remaining inscriptions of a commemorative kind
in which mention is made of the Legions. In no case is
there anything to show to what the tablets had referred.

No. 18 (Plate XLI. 2 ; CJ.L. vii. 1139 ; Tit. Hunt.
p. 31) is a very plain slab, regarding whose earlier
history practically nothing - is known. Stuart says
vaguely that he has "some reason to believe that it
was found in the neighbourhood of Duntocher." l The

1 Caled. Rom.2n^. ed. p. 364, footnote.

312 THE LEGIONARY TABLETS

inscription is contained within a beaded rectangular
moulding, having an ansa-like projection on either
side. The stone measures 2 ft. 6 in. by 2 ft., and the
lettering runs :

VEXILLATIorCS
LEG- M • AVG T
LEG XX V V F

It is worth observing that this is the first time we
have met with a vexillatio or detachment of the Second
Legion. Hitherto it has always been the whole Legion
that was mentioned. It may be, therefore, that the
stone belongs to a different campaign from the
distance-slabs (Nos. 1-17). The same remark applies
to Nos. 21 and 41.

No. 19 (Plate XLI. 3; C.I.L. vii. 1127), which
measures 2 ft. 8 in. by 2 ft. f in., is built into the wall
of Cadder House. Hitherto, on the authority of
Gordon, it has been attributed by common consent
to Bemulie. That attribution we now know to be
inadmissible ; the slab reached Cadder as a present
from Napier of Merchiston, the inventor of logarithms.
It probably came from the western end of the Wall,
conceivably from New Kilpatrick.

The inscription runs :

LEG

n

AVG
FEC

These letters are placed within a large wreath which
is supported between two naked genii, each of whom
stands on the head of an eagle, which rises from the
end of a cornucopiae. The whole is enclosed in a rect-

1 See supra, p. 297.

MISCELLANEOUS SLABS 313

angular beaded moulding. It seems quite likely that,
as Hiibner suggests, No. 19 is identical with C.I.L.
vii. H23.1 Dr. John Buchanan thought that a similar
inscription had once been borne by a stone, a fragment
of which was found at Cadder.2

No. 20 (Plate XLIII. i; C.I.L. vii. 1125; Tit.
Hunt. p. 49) is Gordon's "most invaluable jewel of
antiquity."3 It is a mere fragment, about half of the
letters being awanting. The missing portion can,
however, be restored with a very close approach to
certainty. Thus :

[IMPCTAELIO-HADR]

[ANTONINOAVGPIO]

[P-]P LEG II A[VG SVB]

Q LOLLIO VR[BICO]

LEG AVG PR- PR[- F]

It is therefore a tablet commemorating the execution
by the Second Legion of some piece of work com-
pleted in the reign of Antoninus Pius, when Lollius
Urbicus was governor of Britain (legatus Augusti
pro praetore] — that is, about 142 A.D., when the Limes
was being constructed. What gave it so much im-
portance in Gordon's eyes was the fact that it
confirmed the statement of Julius Capitolinus as to
Lollius Urbicus having been the builder of the Wall.
The exact 'find-spot' of No. 20 was long a matter
of doubt. In the Tituli it is assigned to Bemulie,
and the correctness of the attribution is now beyond
question. According to the letter of Dr. Tanner,
which Mr. Gibb rediscovered, Mr. Urry saw it in 1698

1 See supra, p. 179, footnote.

2 See supra, p. 172. This is C.I.L. vii. 1128.
3Itzn. Sept. p. 63.

3 14 THE LEGIONARY TABLETS

"at a village called Balmudy, in the sole of a byyer
window."1 It may have come originally from the
buildings of the fort. What is left of it measures
i ft. 7 in. by 10 in.

No. 21 (Plate XLIII. 2; C.I.L. vii. 1117; Tit,
Hunt. p. 64), which is also a fragment, measures
3 ft. 9^ in. by 2 ft. i\ in. It exhibits more effort at
decoration than No. 20. The rectangular framework,
in which the lettering is enclosed, is highly ornamented
and has been supported at either end by a pelta-shaped
ornament. The inscription runs :

VEX
LEG II
[AV]G

[F]

Here again, then, we have a mention of a vexilla-
tion of the Second Legion. Possibly, therefore, the
tablet is not strictly contemporary with the construction
of the Limes, as the main body of the Legion was in the
field when that piece of work was being carried through.
The stone was found at Shirva, a little distance to the
east of Auchendavy, seemingly along with a notable
group of sepulchral slabs, to be dealt with in chapters
x. and xi. To what it had originally referred we
cannot, of course, tell.

No. 22 (C.I.L. vii. 1107) is a small legionary stone
from Croy Hill, where it was seen by Gordon.2 It is
now in the Edinburgh Museum. It measures 10 in.
by 7 in., and reads :

LEG VI
VICTi F

1 The Scottish Antiquary ', xv. (1901), p. 85 = Letters to and from Win.
Nicolson, i. p. 338. By a curious slip, Urry places " Balmudy " in Perth.
*Itin. Sept. p. 56.

PLATK XLIIl

'Rf

I. FRAGMENT WITH NAME OF LOLLIUS UKBICUS (NO. 2O)

2. LKGIONARY TAHI,KT FROM SIIIRVA NO. 21

3. AUXILIARY TABLET FROM CASTLECAKY (N(J. 28

MISCELLANEOUS SLABS 315

No. 23 (C.I.L. vii. 1 106) is from the same locality as
No. 22, and is also in Edinburgh. It is interesting
as having been found by Gordon himself. He took
it to be "an invaluable Rarity of its Kind, being the
only Stone that ever I found in the Island of Britain,
with the Name of the fifth Legion impressed upon it." x
The Fifth Legion was never in Britain at all. Con-
sequently we can only suppose either that v stands
for Victrix, or that the I of vi is lost in the border
line. The dimensions of the stone are 10 in. by 5^ in.
The letters, which are enclosed in a rectangular border
with ansa-like projections, certainly appear to be

LEG -V

No. 24 (C.I.L. vii. 1 108), which comes from Croy too,
may suitably be mentioned here. It is in Edinburgh,
and has evidently been a centurial stone. It measures
9 in. by 5 in., and the workmanship is very coarse.
The letters, which occasionally approach the cursive
in form, are hardly legible now. Hiibner's reading
has been corrected by Prof. Haverfield2 into:

OQLICONIS

L(?)ABRVCIVS

No. 25 (C.I.L. vii. 1 100) is another centurial stone, this
time from Castlecary, where it was found within the
fort, in 1841, during the construction of the railway.3
It is in Edinburgh, and measures i ft. 3 in. by 1 1 in.
The execution is good, and there is some attempt at
decoration, a palm-branch and an ivy- leaf rising from

1 Ibid.

2 Archaeological Journal, 1. (1893), p. 304.

3 Stuart, Caled. Rom. 2nd ed. p. 348.

316 THE LEGIONARY TABLETS

each of the upper corners of the border that holds
the inscription. The letters read :

CHO VI
OANTO
ARATI

With this we may compare the stone described in
C.I.L. vii. 1089, which was seen by Gordon and
Horsley built into the wall of Carriden House. It
has long since disappeared. Like No. 25, it simply
gave the name of the centurion and the number of
the cohort.

Except for one or two stones now lost, such as
C.I.L. vii. 1109 and IH6,1 the foregoing comprise all
the purely commemorative inscriptions from the Limes,
which can be associated with the legions. It will not
have escaped notice that the soldiers responsible for
the work were drawn either from the Second Legion or
from vexillations, or special service contingents, either
of the Sixth or of the Twentieth Legion. Reckoning
the Legion as 5000 strong, we must allow for a de-
tachment left behind at the base. Assuming, then,
that the vexillations were of the normal size,2 the total
number of men employed in building the Rampart
and digging the Ditch would not be less than 6000,
and might be 10,000 or even more.3 To what extent,
if any, forced labour may have been resorted to, we

1 C.I.L. vii. No. 1109 was on a pillar from Bar Hill, and is described
and figured by Gordon (Itin. Sept. p. 55 ; Plate 9, Fig. 4). It consisted
of the name and titles of the Emperor, followed by a reference to a
vexillation or vexillations of legionaries. C.I.L. vii. No. 1116 was found
at Auchendavy (Skinner in Archaeologia, xxi. (1827), pp. 459 ff. PI. XXI. 3).
The letters, which were within a laurel wreath, are unintelligible as tran-
scribed, but the Second Legion appears to be mentioned.

a See supra, p. 46. 3 See supra, p. 46, footnote.

THE FORCE EMPLOYED 317

have no means of judging. But it is not unnatural to
think that, in some form or other, the assistance of the
natives would be enlisted to relieve the highly skilled
legionaries, as far as possible, from the more menial
duties attaching to the task.1 It would be interesting
to have a practical man's estimate of the length of
time required to complete the whole. On general
grounds one would be inclined to allow a single season
of, say, six months' duration. Colonel Ruck has given
a solution of the problem based upon a careful com-
parison with modern practice rates.2 Unfortunately
his calculation is vitiated by the assumption that the
work was executed in the face of opposition from an
active and determined enemy. He accordingly postu-
lates an army of not less than 50,000 men, — a force
at least twice as large as Lollius Urbicus can have
taken to the front. As a matter of fact, the probability
is that Rampart and Ditch were not commenced until
organized resistance had been effectually crushed.

A word or two may be added as to the indivi-
dual legions which participated. The Legio Secunda
Augusta — Augustus's Own Second Legion — perhaps
owes its honorary title to the fact that it formed part
of the consular nucleus round which Augustus built up
the standing army of the empire, after the victory of
Actium had placed the whole burden of the Roman
world upon his shoulders.3 It was then stationed in
Further Spain. It was subsequently moved to Upper

1 Cf. the complaint of Calgacus : corpora ipsa ac manus sil-vis ac
paludibus emuniendis inter verbera ac contumelias conteruntur (Tacitus,
Agricola, c. 31).

2 Trans, of the Glasgow Arch. Society (N.S.), iv. pp. 460 ff.

3 Von Domaszewski, Rangordnung, p. 176, footnote 3.

3i 8 THE LEGIONARY TABLETS

Germany, probably in course of the redistribution
necessitated by the annihilation of Varus and his
legions ; and for a time it had its headquarters at
Argentoratum (Strassburg).1 More than half a century
later, when the momentous decision to occupy Britain
was reached (43 A.D.), it was selected for service in the
army of invasion, and thenceforward it remained con-
tinuously in the island. When it crossed the Channel,
it had as its commanding officer the future emperor,
Vespasian.2 The district assigned to it in the general
scheme of advance appears to have been the south of
England.3 Starting from Kent, it gradually fought its
way westwards until it reached I sea Silurum (Caerleon-
upon-Usk). There it built the great fortress which
became its permanent home, a bulwark to defend the
Romanized portion of Britain from possible attack at
the hands of the unruly tribes of Wales. The most
probable date for the occupation of I sea is about
72 A.D.4 It must accordingly have been from there
that the Legion was summoned to participate in the
campaign or campaigns that preceded the building of
the Scottish Wall. The distance-slabs prove that its
whole available strength was at the front on this
occasion, and that it took its full share in the work of

1 This is the view of Ritterling and Zangemeister, and it is confirmed
by tombstones (C.I.L. xiii. Nos. 5975-5978 ; cf. Banner Jahrbiicher, Ixvi.
pp. 71 ff.) as well as by a stamped tile discovered in 1905 ( Westd. Zeitschr.
xxiv. p. 330).

2 Suetonius^ Divus Vespasianus, c. 4.

3 See Haverfield, Arch. Journ. xlix. 181, Viet. Count. Hist. 'Somerset,'
i. 208, Appendix to Mommsen's Roman Provinces, etc. The view has
since been put forward, apparently independently, by Teuber, Beitrage
sur Geschichte der Eroberung Britanniens durch die Romer, pp. 57 ff.

4 There is, however, much uncertainty as to this : see Haverfield, Roman
Wales, p. n, footnote.

THE INDIVIDUAL LEGIONS 319

constructing the Limes. The inscriptions that record
the presence of a vexillation only, should probably be
referred to a different period.1

Unlike the Second Legion, the Legio Sexto, Victrix
Pia Fidelis — The Sixth Legion, the Victorious, the
Dutiful, the Loyal — was not included in the original
army of Britain. Under Augustus its station was in
Hither Spain. Afterwards it was moved to Lower
Germany. We hear of it as being active there during
the crisis of A.D. 69, and it was there that it earned,
twenty years later, the last two of the honorary titles
which it bears in British inscriptions. Previously it
had been known merely as 'the Victorious.' Ritter-
ling has made it probable that the additional distinctions
were conferred upon it in A.D. 89, when the legions in
Upper Germany revolted and proclaimed Antonius
Saturninus as emperor.2 His ingenious arguments go
to show that the movement was crushed by the army
of the Lower Province, whose loyalty remained un-
shaken, and that the grateful Princeps bestowed upon
each of the legions and auxiliary regiments that had
stood firm the additional epithets of ' Pia Fidelis
Domitiana ' — ' the Dutiful, the Loyal, Domitian's
Own.' The last was dropped as a matter of course a
few years afterwards when the damnatio memoriae was
pronounced over the fallen emperor. The two others
were retained. The precise date at which the Sixth
Legion came to Britain is doubtful. That it crossed
direct from Germany some time in the reign of Hadrian

aSee supra, pp. 312 and 314, for Nos. 18 and 21, and infra, p. 342,
for No. 41.

2 ' Zur romischen Legionsgeschichte am Rhein ' in Westd. Zeitschr. xii.
pp. 203 ff.

320 THE LEGIONARY TABLETS

is rendered certain by epigraphic evidence.1 As its
headquarters were fixed at York, it is clear that it was
brought over to replace the Ninth Legion, after the
latter had been cut to pieces by the Brigantes.2 This
cannot have been long before the construction of
Hadrian's Wall, at the building of which it must have
assisted. Two decades later, a vexillation was engaged
in the making of the Scottish Limes.

The Legio Vicesima Valeria Victrix — the Twen-
tieth Legion, the Valerian, the Victorious — was not
represented in the army of Augustus as originally
constituted. It was one of five that were subsequently
raised, in all probability when the project of conquering
Germany began to take definite shape in the minds of
the emperor and his advisers. It appears to have
seen service in Illyricum during the Pannonian insur-
rection.3 But its first unmistakable entrance on the
stage of history is in 14 A.D., when it played a pro-
minent part in the mutinous disturbances that marked
the accession of Tiberius.4 Its winter quarters were
then at Colonia Agrippina (Cologne). Ten or twelve
years afterwards it seems to have been transferred to
Novaesium (Neuss).5 It was still at Novaesium in
43 A.D., when it received orders to join the expedition

1 Thus, a well-known inscription in honour of a certain Marcus Pontius,
who lived in the reigns of Hadrian and his immediate successor, describes
him inter alia as TRIBIVNO] MILITVM LEG[IONIS] vi VICTR[ICI8]

CVM QVA EX GERM[ANIA] IN BR I TT A N [N I A M] TRANSIIT (C.l.L.

vi. No. 1549).

2 See supra, p. 3.

3 Von Domaszewski, Rangordnung, p. 177. Cf. C.l.L. iii. p. 280, and
Index.

* Tacitus, Ann. i. 31-39.

6 See Nissen, Banner Jahrbiicher^ cxi.-cxiii. pp. 10 ff.

THE INDIVIDUAL LEGIONS 321

which Aulus Plautius was to lead against Britain. In
Britain it had no lack of hard fighting. Along with
the Fourteenth Legion, it appears to have acted as the
central of the three columns of invasion, moving from
London north-westwards across the midlands of Eng-
land. As early as A.D. 49 it had established itself at
Viroconium (Wroxeter). About A.D. 60 its head-
quarters were pushed forward to Deva (Chester),
which continued to be its station until the Romans
finally quitted the island. It is interesting to remember
that at one time the Twentieth Legion had Agricola
as its commanding officer.1 Regarding any part which
it and the Second Legion took in the campaigns which
he carried on in Caledonia during his governorship,
we know nothing definite. Tacitus is silent on these
details, aqd there are no inscriptions to guide us.2 But
the distance-slabs show that there was a vexillation
of the Twentieth with Lollius Urbicus,3 while (as has
already been indicated) No. 18 appears to point to a
different period when the Second and the Twentieth
were alike represented by contingents.

1 Tacitus, Agricola, 7 f.

2 An altar from Camelon, now in the National Museum, has been
interpreted as referring to the Second Legion Adjutrix, and therefore
as belonging to the period of Agricola (Class. Review, 1904, p. 399).
It reads MILITES • L • il . A DIE VIRT • L . M. Professor Haverfield, how-
ever, has condemned it as a forgery (Class. Review, 1905, p. 57), a
verdict with which I agree.

3 The fact that No. 20 has simply LEG -XX is of no significance. It
merely indicates careless workmanship. See supra, p. 303, footnote.

CHAPTER X

OTHER INSCRIPTIONS

ALL the commemorative tablets that have so far been
described had been erected by legionaries. There re-
main three which bear the names of auxiliary cohorts.
In two out of the three cases we know where the
tablet stood and what it was intended to commemo-
rate. As the third is similar in character, it seems not
improbable that it may have served a similar purpose.
No. 26 (Plate XX. 3 ; Roman Forts on the Bar Hill,
pp. 82 f.) was found in the well at Bar Hill. It is
broken in three pieces and is incomplete. But enough
of the inscription is preserved to admit of the whole
being restored. Originally it had run somewhat as
follows :

• [IMP-CAE]SARI •
TAE[L-HADAN]TONINO

AV[G PIO P P-C]OH •
T[BAETASIOR-C-]ROB
VI[RTVTEM-ET-FI]DEM

That is : " The First Cohort of Baetasii, made Roman
citizens for their valour and loyalty, [erected this] in
honour of the Emperor Caesar Titus Aelius Hadrianus
Antoninus Augustus Pius, Father of his Country."
When complete, the slab must have been about 3 ft.
long by 2 ft. high.

AUXILIARY TABLETS 323

The particular cohort here mentioned had originally
been raised among the Baetasii, a people of the Lower
Rhine, possibly after the revolt of Civilis in which we
know that they participated. The corps is not heard
of anywhere save in Britain, whither we may suppose
it to have been moved immediately after it was levied.
Such a proceeding would have been in accordance with
the ordinary Roman practice. At all events, military
diplomas of the years 103 and 124 A.D.1 prove it to
have been in the island at least as early as the be-
ginning of the second century. It did not then bear
the title of c • R, so that the ' valour and loyalty' which
earned the immediate gift of Roman citizenship for its
members, may perhaps have been displayed in the
course of the struggle that was ended by the making
of the Scottish Limes. It may, therefore, have been
after their withdrawal from Bar Hill that they were
quartered, as they evidently were for some time, at
Uxellodunum (Ellenborough, near Maryport), close to
the western end of Hadrian's wall. Their presence
there, under two different commanders, is attested by
as many as five inscriptions.2 Fully two centuries
later we get a glimpse of them confronting danger
from another quarter. The Notitia gives their station
as Regulbium (Reculver) on the Saxon Shore.

No. 27 (Plate XXXI. 2 ; Proc. Soc. Antiq. Scot.
I9°5> P- 47O and p. 472) came from what had quite
possibly been the well of the fort at Rough Castle.3
There is a portion of the stone awanting, but the
missing letters can be supplied without any difficulty.

1 C.I.L. vii. Nos. 1193 and 1195.

2C/.Z. vii. Nos. 386, 390, 391, 394 and 395.

3 See supra^ p. 226.

324 OTHER INSCRIPTIONS

As in the case of No. 26, there is little or no attempt at
ornament. No. 27, indeed, is even plainer, the letters
being simply enclosed within a beaded rectangular
moulding. They may be read as follows :

[IMP-OA1ESARI TITO

[AELIO-]HADRIANO.

[ANTO]NINO- AVG-

[PIO-]P.P-COH-Vl

[NER]VIORVM • PRI

[N C I P] I A FECIT

That is: "The Sixth Cohort of the Nervii erected the
Principia in honour of the Emperor Caesar Titus
Aelius Hadrianus Antoninus Augustus Pius, Father of
his Country." The surviving portion of the tablet is
nearly square, being i ft. n|- in. long by i ft. n in.
high.

The Nervii were neighbours of the Baetasii, but
their name is much more familiar in history owing to
the prominent part they played in Julius Caesar's
Gallic wars. The Romans were quick to take ad-
vantage of their fighting qualities by enlisting them
as auxiliaries. Tacitus tells us that, in the life and
death struggle with Civilis, certain cohortes Nerviorum
brought the army of Vocula into the gravest peril
by suddenly retreating, either through treachery or
through panic, and so leaving the Roman flanks
exposed.1 It has been conjectured that these may
have been the Fourth and Fifth Cohorts, of whom no
mention is made in any extant inscription, and who
may therefore have been among the regiments whom
Vespasian disbanded in disgrace.2 However that may
be, the First, Second, Third, and Sixth Cohorts were

1 Hist. iv. 33. 2See Pauly-Wissowa's Real-Encyclopadie^ iv. p. 319.

AUXILIARY TABLETS 325

all ultimately incorporated in the army of Britain.
The Notitia gives the station of the last-named as
Virosidum, a fort which cannot now be definitely
identified, but which was perhaps near Bainbridge in
Yorkshire. The Cohort was certainly at Bainbridge
towards the end of the reign of Septimius Severus.1 It
was also at one time at Aesica (Great Chesters) on
Hadrian's Wall.2 Whether this was before or after its
occupation of Rough Castle, it is impossible to tell.

Returning now to the inscription, we may note the
importance of the last two words — 'Principia fecit'
As has already been pointed out,3 these show that the
proper name for the central administrative buildings of
such forts as Rough Castle was ' Principia,' not ' Prae-
torium.' They also show where the tablet had stood,
and in so doing furnish a clue to the original position
of the corresponding slab from Bar Hill. The general
form of the inscription is the same in the two cases,
and both were found in the same part of the fort.
Comparison with a similar, if rather more elaborate,
tablet from the well of the fort at Birrens in Dum-
friesshire4 justifies the conclusion that a memorial of
this kind was placed in a prominent position in the
Principia of each of the second century caslella in
North Britain. It recorded the name of the corps
that had formed the original garrison.

No. 28 (Plate XLIII. 3; C.LL. vii. 1099 ; Tit. Hunt.
p. 72) has features which suggest that it belongs to
the class of which Nos. 26 and 27 are examples.
Like them, it is plain, although the rectangular frame-

1 C.LL. vii. No. 269. * Ibid. No. 726.

3 Supra, pp. 224 and 226.

*Proc. Soc. of Antiq. of Scot. 1896, pp. 128 IF.

326 OTHER INSCRIPTIONS

work is supported on either side by a pelta-shaped
ornament, such as the distance-slabs have rendered
familiar. It measures 3 ft. 6 in. by i ft. 9 in. The
inscription is complete, and reads :

IMPCXES T-XEL-AIT

AVG- PIO P- P.

COM T TVNGRO

RVM -FECIT-co

There has always been some doubt as to the precise
significance of the sign that follows the word 'fecit'
That it denotes 1000 is, of course, clear. The fact that
it is a numeral at all has led most of those who have
discussed the question to connect it with the numerals
on the distance-slabs, and to suppose that it refers to
1000 feet or paces completed by the First Cohort of
Tungrians. There are various objections to this,
notably the absence of p or M • P, and the circum-
stance that it would be anomalous to find auxiliaries
taking any part in the actual construction of the
Limes. The analogy of Nos. 26 and 27, on the other
hand, may be cited in support of the alternative view,
according to which the ambiguous sign has no refer-
ence whatever to distance, but is used, as it often is,
to indicate that the Cohort was miliaria or 1000
strong. The solitary counter-argument is that it
follows 'fecit' instead of preceding it. But that is
easily met. The milliary sign was not an essential
part of the designation of the Cohort. As a matter of
fact, it is more usual to omit it. Here it was inserted
as an after-thought when the stone-cutter found he
had more space than he absolutely required. We
may, therefore, translate : " The First Cohort of
Tungrians-j 1000 strong, erected [this] in honour of

AUXILIARY TABLETS 327

the Emperor Caesar Titus Aelius Antoninus Augustus
Pius, Father of his Country."

The First and Second Cohorts of Tungrians are
among the oldest of the auxiliary regiments whose
history is known. Each of them was 1000 strong,
but they presented a curious difference from the ordi-
nary cohortes miliariae, in that the commander was
only a praefectus, not a tribunus}- They are first
heard of in A.D. 69, when they constituted the largest
element in the force despatched by Fabius Valens
to ensure the safety of Gallia Narbonensis against a
threatened raid of Otho's fleet.2 They must have
been sent over to Britain shortly afterwards, for both
were in Agricola's fighting-line at the battle of Mons
Graupius.3 When the Notitia was compiled, the First
Cohort was in garrison at Borcovicium (Housesteads)
on the English Wall. To judge from the number of
inscriptions they have left behind, they must have lain
there for many years.4 On one occasion they were at
Cramond.5 No. 28 shows them to have been also at
Castlecary, for it was there that the slab was found.
Beyond the date (1764), we have no particulars as to
the circumstances of its discovery. If it was got
within the area of the fort, it may well be the memorial
tablet from the Principia.

It is just possible, though hardly likely, that we
ought to associate with Castlecary one of the auxiliary

1 See supra, p. 53. See also infra, No. 43.

2 Tacitus, Hist. ii. 14. So Cichorius in Pauly-Wissowa's Real-Encyclo-
pddie, iv. p. 343. It is, however, not quite certain that these Tungrian
cohorts were identical with the similar corps afterwards found in the
Army of Britain.

3 Tacitus, Agricola, 36.

4 C.I.L. vii. Nos. 633, 635, 638, etc., etc. 6 C.I.L. vii. No. 1084.

328 OTHER INSCRIPTIONS

regiments that stood shoulder to shoulder with the
Tungrians at Mons Graupius. It 1698 Mr. Urry saw
there a fragment of an inscribed stone on which the
letters H BAT were visible.1 The same reading was
given independently thirty years later by Gordon, who
speaks of the stone as "a broken Altar."2 The frag-
ment subsequently passed into the collection of Sir
John Clerk of Penicuik, and has long since disappeared.
If it was correctly deciphered, the inscription may
conceivably have referred to the First Cohort of
Batavians, which is mentioned as forming part of the
garrison of Britain in a diploma of A.D. 1 24, and which
has left inscriptions at Procolitia (Carrawburgh) and
Magnae (Carvoran) on the English Wall.3 As Hiibner
points out, however, the absence of the number of the
Cohort makes one suspect the accuracy of the reading.4
In any case the lost fragment serves to introduce a
fresh group of inscriptions, those occurring upon
altars.

No. 29 (CJ.L. vii. 1134; Tit. Hunt. p. 32, PI. XV.
Fig. 2) is a much defaced altar discovered in 1829 near
Duntocher. It is said to have been found "in the
vicinity of the Fort." 5 The only letters now legible are :

i o M

and even these can only be made out with the greatest
difficulty. They show that it was dedicated to Jupiter
Optimus Maximus, the chief god of the army, legion-

1 The Scottish Antiquary, xv. (1901), p. ^—Letters to and from Wm.
Nicolson, i. p. 338.

»//»». Sept. p. 57, PL 15, Fig. 4.

* CJ.L. vii. Nos. 617, 621, 777. * CJ.L. vii. No. uoi.

6 Stuart's Caled. Rom. 2nd ed. p. 300, footnote.

ALTARS 329

aries, and auxiliaries alike.1 The dimensions are 2 ft.
i in. by i ft.

No. 30 (Plate XLIV. i ; C.I.L. vii. 1 129 ; Tit. Hunt.
p. 35) was ploughed up in 1826 a few hundred yards
east of the fort of Castle Hill. It is said to have been
"firmly fixed on its edge in the ground, a few feet
below the surface, close to, and on the south or Roman
side of, the Wall. It had the appearance of being
purposely buried in the stiff clay soil."2 It is 3 ft.
i \ in. high by i ft. 2^ in. broad. The inscription,
which is well preserved, reads :

CAMPES
TRIBVSET

BRTANNI

Q. PSEJIVS

IVSTVSPREF

COM -NTT GAL

V S L L- M

That is " Q. Pisentius Justus, Commander of the
Fourth Cohort of Gauls, [dedicated this altar] to the
Campestres and to Britannia.3 Willingly, gladly, justly
has he performed his vow."

The deities in whose honour this altar was erected
are interesting. Britannia is the personification of the
island. On an inscription from York she is called
' Britannia Sancta.'4 A slight variant of the abstrac-
tion appears on No. 35, to be described presently.
The Campestres or Matres Campestres are closely
analogous to the Deae Matres or Matronae, whose

1 Von Domaszewski, Religion des romischen Heeres, pp. 22 ff.

2 Stuart's Caled. Rom. p. 308, footnote.

3 Von Domaszewski would, however, expand into Britanni(cis), to
agree with matribus understood (Relig. des romischen Heeres, p. 50).

4 C.I.L. vii. No. 232.

330 OTHER INSCRIPTIONS

draped and seated figures recur so often among the
monuments on the German frontier, the three being
always represented side by side. They are of Celtic
origin. Under their special aspect of Campestres they
perhaps had a shrine near the campus or drilling
ground of the fort. They seem to have been held in
particular veneration by mounted troops. It has been
pointed out that the extant inscriptions on which they
are mentioned are chiefly dedications by alae or
cohortes equitatae? or by officers of one or other type
of regiment. No. 30 is no exception. The Fourth
Cohort of Gauls was equitata, or furnished with a
contingent of cavalry. It first meets us in Britain in a
diploma of the year A.D. 146, having probably been
included in the reinforcements which the garrison
received during the reign of Hadrian.2 The Notitia
assigns it to Vindolana (Chesterholm) on the English
Wall, and it has left inscriptions there.3 Tiles bearing
its stamp have been found at Templeborough in York-
shire,4 and other inscriptions near Castlesteads 6 as well
as at Habitancium (Risingham),6 a fort on the road
between Corbridge and Newstead. The altar we have
been discussing indicates that the regiment had once
held Castle Hill on the Scottish Limes.

No. 31 (C.I.L. vii. 1124) is described by Selden on
the authority of Camden, but has long been lost.
When copied, it is stated to have been at Cadder. It
is very doubtful whether the reading can be accepted
as correct. The nomen of the dedicator seems to be

1 Von Domaszewski, Religion des romischen Heeres, p. 51.

2 Pauly-Wissowa, Real- Ency clop, iv. pp. 290 f.

3 C.I.L. vii. Nos. 703 f., 715. 4 Ephem. Epigr. iv. No. 697.
6 C.I.L. vii. Nos. 877 f. 6 C.I.L. vii. No. 1001.

ALTARS 331

wrong, and one would have expected a note of the
corps which he commanded. Camden's version of the
inscription is :

DEO

SILVANO
L TANICVS

VERVS
PRAEF VSLLM

Silvanus was popular with the soldiery. We shall
meet him again on Nos. 36 and 38, and probably also
on No. 43.

No. 32 (Plate XLI V. 2 ; C.I.L. vii. 1 1 1 1 ; Tit. Hunt,
p. 61) is the first of a very remarkable group, the
discovery of which in 1771 in a pit at Auchendavy,
has already been recorded.1 It is the largest of the
set (3 ft. f in. by i ft. 2.\ in.), and also the only one
that has remained unbroken. The first line is made
specially conspicuous by being cut in large letters
on the cornice. The whole reads :

IOM

VICTORIAE

VICTRICIPROSA.V

TEIIVPNETSVA

SVORVM
M COCCEI

FIRMVS
>LEG- M AVG

That is, " To Jupiter, Best and Greatest, M. Cocceius
Firmus, a centurion of the Second Legion, Augustus's
Own, [dedicated this altar], and also to Victory, the
Victorious, for the well-being of our Emperor and for
his own well-being and that of his household."

It will be observed that the foremost place, and that
a very special one, is accorded to Jupiter Optimus

1 See supra, p. 185.

332 OTHER INSCRIPTIONS

Maximus, who was, as we have seen, the chief of the
dii militares. As a rule, in the religious memorials of
the Roman army, Victory appears, not as an inde-
pendent divinity, but rather as the incarnation of the
conquering spirit finding its embodiment in some par-
ticular individual — Victoria Augusti and the like.1
Here, however, as on No. 33 and No. 44, she figures
independently. The exact significance of the epithet
that attaches to her name in this case, can hardly be
determined. Some have been tempted to associate it
with the gaining of a signal success. But that is quite
uncertain.

No. 33 (Plate X LI V. 3; C.I.L. vii. 1114; Tit. Hunt.
p. 59) is the second of the Auchendavy group. It
reads :

MARTI

Mlf€RVAE

CAMPESTRI
BVS HER©2
EPONAE
VICTORIAE
M COCCEI
FIRMVS
>LEGMAVG

That is, " To Mars, M. Cocceius Firmus, a centurion
of the Second Legion, Augustus's Own, [dedicated
this altar], and also to Minerva, to the Campestres,
to Hercules, to Epona, to Victory." The altar is
2 ft. 9^ in. high by i ft. ^ in. broad.

Mars Ultor ranked next to Jupiter Optimus Maximus
among the old Roman gods of war, and it will be

1 Von Domaszewski, Religion des romischen Heeres, pp. 37 ff.

2 That is, HEROI, Hiibner reads HERCL. In any case Hercules
is intended.

ALTARS 333

observed that he occupies the same privileged position
on this altar that Jupiter did on No. 32. His name
is cut in large letters on the cornice. The other deities
honoured here all occur frequently in military dedica-
tions. Minerva, like the Greek Pallas, was an armed
goddess. The Campestres have been discussed under
No. 30. Hercules was, of course, a Roman god, but
his popularity with the frontier-armies of the north
and west was doubtless due to his identification with
a Teutonic divinity, probably Donar. Tacitus says
that the Germans " tell that Hercules too once visited
their country, and he is first among the heroes whose
praises they chant when they are about to go into
battle."1 Epona was a characteristically Celtic god-
dess, whose worship was very wide-spread among the
soldiery in imperial times. She was the protectress
of horses. Indeed, the root element in her name is
cognate with equus. The fact that both she and the
Campestres are singled out for veneration may perhaps
be interpreted as an indication that the garrison of
Auchendavy had included a mounted contingent. If
so, it was in all probability a cohors equitata. Of
Victory we have already spoken.

No. 34 (Plate XLIV. 4; C.I.L. vii. 1112; Tit. Hunt.
p. 58) is a smaller altar than either of the preceding,
but has a rather more elaborate cornice. Its dimen-
sions are 2 ft. 4^ in. high by i ft. \ in. broad. The
inscription runs :

D I A N A E
APOLLINI
M COCCEI
F I R M VS
>LEGilAVG
1 Gertnania, 3.

334 OTHER INSCRIPTIONS

That is, " M. Cocceius Firmus, a centurion of the
Second Legion, Augustus's Own, [dedicated this altar]
to Diana, to Apollo."

The main point to be noted here is that Diana
takes precedence of Apollo, a somewhat unexpected
arrangement, although one that has a parallel in
another quarter of the empire.1 In Britain Apollo
is more frequently honoured with dedications than
his sister, but an altar which begins with the name
of Diana Regina — the others being illegible — has
recently come to light at Newstead to help to restore
the balance.2

No. 35 (Plate XLIV. 5; CJ.L.vil 1113; Tit. Hunt.
p. 60) is a very plain altar, measuring 2 ft. 6f in.
high by 1 1 in. broad. The letters on the die read :

G E N i o

TERRAE
B R I T A
NNICAE
MCOCCEI
F I R M VS
>LEGflAVG

That is, " M. Cocceius Firmus, a centurion of the
Second Legion, Augustus's Own, [dedicated this altar]
to the Spirit that watches over the land of Britain."

The ' Genius Terrae Britannicae ' is another form
of the ' Britannia ' to which reverence was done on
No. 30. But the amplified expression enables us to see
at once how Britannia secured her niche in the military
pantheon. The conception of the genius, and above

1 Illyria : see Von Domaszewski, Religion des romischen Heeres, p. 53,
where, however, the analogy with Britain is overlooked.

2 J. Curie, A Roman Frontier Post, p. 143.

ALTARS 335

all of the genius loci, was a very characteristic feature
of Roman religion. Thus, when Aeneas landed in
Latium, his first act was to pray to the divinities of
the place :

" Sic deinde effatus frondenti tempora ramo
hiplicat, et Geniumque loci primamque deorum
Tellurem Nymphasquc et adhuc ignota precatur
Flumina" 1

No. 36 (Plate XLV. i ; C.I.L. vii. 1115; Tit. Hunt.
p. 62) was found in the same pit as Nos. 32-35, and
can most conveniently be noticed here, although its
fragmentary condition precludes its being absolutely
attributed to Cocceius Firmus. It measures 10^ in.
by 7! in. Only the first two lines of the inscription
remain. They read :

SILVA
N O

That is, "To Silvanus."

If this was followed by the name of Cocceius Firmus,
then Silvanus can have been the only deity mentioned ;
considerations of space would not admit of more.
No. 36 would in that event fall into line with Nos. 31,
38, and 43, on all of which Silvanus is alone, as he
generally is. Two well-known British inscriptions show
him associated closely with the chase.2

Apart from the points of individual interest which
they present, Nos. 32-35 are noteworthy as constituting
a homogeneous group erected by one and the same
person. Analogies could be adduced from other forts.
The most obvious is that of C. Arrius Domitianus, a
centurion of the Twentieth Legion, whose name occurs
on three out of the five inscribed altars from Newstead.

1 Virgil, Aeneid, vii. 135 ff. z C.I.L. vii. Nos. 451 and 830.

336 OTHER INSCRIPTIONS

At Auchendavy all five inscribed stones were found
outside the fort. But the pit in which they were lying
was just beyond the rampart, so that there is no such
certainty as to their original position as there is in the
case of No. 30 and No. 38, both of which must have
stood in outside shrines. At the same time the prob-
abilities are in favour of some, if not all, of them having
belonged to the annexe. The motley list of divinities
is remarkable. It illustrates admirably the syncretism,
or mixture of religious ideas, that permeated all strata
of society in the Empire before the final victory of
Christianity. And it seems possible that it has a
further significance. The cult of such an array of
gods and goddesses passes the limits of what we should
look for, even from the most catholic-minded of private
individuals. The altars are more likely to be official —
intended, that is, to voice the sentiments of the garrison
as a whole. This suggestion is confirmed ,by the
prominence given to Jupiter Optimus Maximus and
Mars Ultor, who were emphatically dii militares,
regimental deities. The Campestres, again, were sup-
posed to exercise a special care over soldiers and their
work, particularly perhaps over cavalry,1 whereas a
centurion was an infantry officer. Lastly, Epona was
very distinctively the goddess of horses, and was not
likely to be invoked except in connection with mounted
men. We know from Juvenal that to swear by Epona
was an affectation that smelt strongly of the stable.2

The conclusion to which all this points is not very
far to seek. If it was in a representative capacity that
Cocceius Firmus set up his four (or five) altars, it is
natural to assume that he was in command of the

1 See supra, p. 330. 2 Sat. viii. 1 57.

I'LATK XLIV

i. Xo. 30

2. Xo. 32

3. Xo. 33 4. Xo. 34 5- N'°- 35

ALTARS ?-ROM CASTLE HILL AXD AUCHENDAVY

ALTARS 337

garrison. On general grounds we should expect the
fort to be manned by auxiliaries. We have noted that
there is some indication that it had been held by a
cohors equitata. But, as has been already stated,1
from the reign of Hadrian onwards it was a common
practice to put a centurion in charge of an auxiliary
regiment, he retaining the while his legionary rank and
title. We shall find an authenticated instance of this
presently at Rough Castle. There would be all the
more reason for taking such a course, if the regiment
were quartered by itself in a frontier outpost. Admini-
strative capacity and experience would be almost as
essential there as courage and military skill Of course,
the custom was far from being universal ; the Gauls
at Castle Hill and the Hamii at Bar Hill were each
under the command of a praefectus. Still, it was
adopted at Rough Castle, and may therefore have been
adopted at Auchendavy too. Till evidence to the
contrary is forthcoming, it seems best to see in it
the most probable explanation of the facts regarding
Cocceius Firmus and his altars.

No. 37 (Plate XLV. 2 ; Roman Forts on the Bar
Hill, pp. 80 f.) was taken from the well of the Principia
at Bar Hill, and is 3 ft. high by i ft. 5 in. broad. The
clearly cut inscription is very brief:

COM -T-

BAETASIOR

OR

That is, "The First Cohort of Baetasii, Roman
citizens, [dedicated this altar]."

But to whom did they dedicate it ? The absence of
the name of any divinity is at first sight puzzling. It

1 See supra, p. 49. Cf. also Haverfield, Roman Wales, p. 119.

Y

338 OTHER INSCRIPTIONS

ceases to be so, if we remember that the ' find-spot '
was within the precincts of the Principia. Doubtless
the altar had stood in the shrine of the standards.
Such a setting would of itself suffice to show its signifi-
cance. Its very presence in the sacellum would imply
a direct connection with the god who was held in
highest honour there. In a legionary fortress this
would have been Jupiter Optimus Maximus.1 It was
so occasionally in the smaller castella also, as for
example at Newstead, where an altar dedicated to
Jupiter Optimus Maximus was taken from the well
of the Principia.2 But at Birrens the altar in the
sacellum had reference to the imperial cult in its mili-
tary aspect; the legend is " Discip(linae) Aug(usti)"*
At Bremenium (High Rochester), again, an altar that
had fallen into the strong room beneath the sacellum
reads : "(Genio) d(omini] n(pstrt) et signorum,"^ while
another from within the area of the same fort, and
doubtless ultimately from the sacellum, has : "Genio
et signis coh(ortis)" 5 We cannot, therefore, decide to
whom No. 37 was dedicated. But the soldiers them-
selves would understand. The cohort to which they
belonged has already been discussed.6

No. 38 (Plate XLV. 4 ; Antonine Wall Report, p. 153)
was ploughed up in 1895, along with a separate base
on which it had rested, 240 yards north-east of the
eastern gateway of Bar Hill fort. It had been rather

1 Von Domaszewski, Religion des romischen Heeres, pp. 22 ff.

2 J. Curie, A Roman Frontier Post, pp. 141 f. PI. xvi.
zProc. Soc. of Antiq. of Scot, 1896, p. 131.

*C./.Z. vii. No. 1030. Cf. Bruce, Roman Wall(\^-]\ pp. 318 f.
6C.I.L. vii. No. 1031. *Supra, p. 223.

ALTARS 339

seriously damaged. But Professor Haverfield had no
difficulty in restoring the inscription as follows :

[D]EO-SILV[ANO]

[C]ARISTAN[IVS]

[I]VSTIANV[S]

PRAEF

[C]OHTHA[MIOR]
V-S- L L M

That is, " Caristanius Justianus, Prefect of the First
Cohort of Hamii, [dedicated this altar] to the god
Silvanus. Willingly, gladly, justly has he performed
his vow." It is 3 ft. high by i ft. 5 in. broad.

This is the third altar dedicated to Silvanus which
we have met with, and a fourth has still to be described.
In the present case the shrine of the god certainly
stood outside the area of the fort. We may infer that
it did so also in the other cases (Nos. 31, 36, and 43).
The Hamii, who probably came from Syria, were
soldiers of a special class. They were bowmen, as we
learn from the descriptive epithet sagittarii, applied to
them in one of several inscriptions that prove them to
have been stationed for a time at Magnae (Carvoran)
on the English Wall.1 That their presence at Bar
Hill was more than a passing incident is indicated by
a lost inscription from the neighbourhood of the fort
which gives the name of a different prefect (No. 50).
It looks as if they, as well as the Baetasii, had been in
garrison there. This would fit in aptly with the view
that the forts on the Scottish Limes had witnessed two
successive occupations during the Antonine period.

No. 39 (Plate XLV. 3; C.I.L. vii. 1103; Ephem.
Epigraph, vii. 1093 (p- 334) > Tit. Hunt. p. 69)

1 C.I.L. vii. No. 748.

340 OTHER INSCRIPTIONS

ought also to be associated with Bar Hill, as is
proved by the extract from the Daily Gazetteer quoted
in the Tituli. It was first noticed about 1733. ^ts
dimensions are i ft. 8 in. by 12 in. But unfortunately
it is so much defaced that only the first two lines
can be read with any confidence. They run :

DEO MAR
[Tl ]CAMVLO

Some of the remaining letters are entirely obliterated.
Others are barely decipherable. One would not have
been surprised to encounter the Baetasii or the Hamii
again. But Professor Haverfield's version of the next
three lines is :

.. GTlAVG- I
. MARIO. . .
SO. .

If this be correct, the altar was a legionary dedica-
tion. Camulus was a Celtic divinity. He is known
only from inscriptions, and on these he is usually
identified with Mars. His name is obviously the main
element in ' Camulodunum/ the modern Colchester.

Before passing beyond Bar Hill, we may mention
an altar described in the Itinerarium Septentrionale^
and undoubtedly belonging to the fort or its precincts.
When Gordon saw it, it was at Auchenvole House.
It is now in the National Museum. It measures 3 ft.
3 in. by i ft. 5 in., and is notable for the richness of
its decoration. Gordon speaks of it as "a very curious
Altar, with several remarkable Figures engraved upon
it, having a Corona Triumphalis, with an Inscription
in the Middle, which is now defaced. Upon one Side
is engraved, in Relievo, a Quiver full of Arrows ;
1 Op. cit. p. 55, Plate 13, Fig. i.

PLATE XLV

i. No. 36

3- No- 39

2. No. 37

4. No. 38 5. No. 40

ALTARS FROM AUCIIENDAVY, BAR HILL, AND CROY HILL

ALTARS 341

upon the other Side, an Arcus or Bow. It is of the
best Proportion I ever met with, having several very
curious Borderings and Wavings above the Cornice."
The bow and quiver remind one of the Hamii, but it
is perhaps more likely that they indicate a dedication
to Apollo. No traces of letters are now visible within
the wreath. On the other hand, one can detect on
the fourth side the faint markings that betray a
vanished inscription.1

No. 40 (Plate XLV. 5; C.I.L. vii. 1104; Stuart,
Caled. Rom. 2nd. ed. PI. XIII. 7) is, like No. 39, a
legionary altar. It was found about the beginning of
the nineteenth century at the foot of Croy Hill, and
was seen by Stuart at the house of Nether Croy.
Soon after 1890 it was removed to Carron House,
near Falkirk, where it remained till 1910. It is now
in the National Museum of Antiquities in Edinburgh,
the gift of Carron Company. The inscription appears
to run as follows, the first word being on the upper
portion of the cornice and the remainder on the die.
The end is seemingly incomplete :

NYMPHIS

VEXILLATIO

LEG VI VIC

PF SVB FA
[B]IOL[I]BERA

That is, "A detachment of the Sixth Legion, the
Victorious, the Dutiful, the Loyal, serving under
Fabius Liberalis, . . . [dedicated this altar] to the
Nymphs." The dimensions are 3 ft. \ in. by I ft.
5i in.

1See Haverfield in Arch. Jour. \. p. 305 (No. 164).

342 OTHER INSCRIPTIONS

It is clear that No. 40 must have been erected during
or after some period of active campaigning, when
legionary troops were in the field, — possibly, though
not necessarily, when the Limes was originally con-
structed. The worship of the Nymphs was apparently
borrowed by the Romans from the Greeks, but under
the empire it spread all over Europe and even into
Africa. They were essentially local divinities. Virgil
couples them with the genius loci in a passage that
has already been quoted.1 Familiar British inscriptions
that mention them are a dedication at Chester by the
Twentieth Legion " Nymphis et fontibus"* and a
metrical one at Risingham on an altar erected as the
result of a warning received by a soldier in a dream.3

No. 41 (Plate XLVI. i ; C.I.L. vii. 1093 ; Tit.
Hunt. p. 73) was discovered about 1770 in the Baths
of the fort at Castlecary.4 It measures 2 ft. 5 in. by
i ft. i in., and reads :

FORTVNAE

VEXILLA

TIONES

LEG- 0 AVG

LEG VI VIC

P S P L L

It is not quite certain how the contractions in the last
line ought to be expanded. One would have expected
p- F for the titles of the Legion, but the s is quite firmly
and distinctly cut. Whatever the exact interpretation
of that particular line may be, the general sense of the
whole is perfectly plain : the altar has been dedicated
to Fortune by detachments of the Second and Sixth
Legions.

1 See supra, p. 335. 2 C.I.L. vii. No 171.

3 C.I.L. vii. No. 998. 4 See supra, p. 214.

ALTARS 343

Apparently, legionary troops were at one time quar-
tered temporarily in the fort at Castlecary. The fact
that only a detachment of the Second Legion was
present points to some period other than the season
in which the Limes was constructed.1 The 'find-spot'
is characteristic. There was a peculiarly close associa-
tion between Baths and the goddess Fortune, not
improbably because the building was much used for
games of chance. Thus, the tablet bearing the name
of Virius Lupus, which was found at Bowes in
Yorkshire, and which records the restoration of a
" balineum " after a conflagration, is dedicated to
Fortune.2 And the present is only one of a number of
instances in which an altar originally set up in her
honour has been discovered among the ruins of bath
buildings.3 At Castlecary a small figure of Fortune
set in a niche (Plate XLVIII. 2) was lying not far
from the altar.

No. 42 (Plate XLVI. 2 ; C.I.L. vii. 1095) is an
interesting little altar — an ar2ila rather than an ara —
found between the fort of Castlecary and the Red
Burn. It is now in the National Museum, Edinburgh,
and is only i ft. 8J in. high by 10 in. broad. The
inscription runs :

DEO
MERCVRIO

MILI-ES LEG VI

VIC-RICIS PIE F

ED ET SIGILLWX

GIVES ITALICI

ET- NORICI-

V S L- L M

1 See supra, p. 312. ZC.I.L. vii. No. 273.

3 See supra, p. 82, footnote.

344 OTHER INSCRIPTIONS

That is, " Soldiers of the Sixth Legion, the Victorious,
the Dutiful, the Loyal, citizens from Italy and from
Noricum, [dedicated] a shrine and a statuette (aedi-
culam et sigillum) to the god Mercury. Willingly,
gladly, justly have they performed their vow."

Evidently, as was not uncommon, a little shrine of
Mercury had stood outside Castlecary fort. The
interest of the inscription, however, centres mainly
round the words " cives Italici" These are important
as showing that, even in the second century, recruits
for the legions were sometimes drawn from Italy,
although it was certainly contrary to established
custom to make any levy there.1 If we could trust
Sibbald's reading of a lost inscription2 — and it is
accepted by Hiibner3 — the "cives Italici et Norici"
had Britons for their comrades in the ranks at Castle-
cary. This is, of course, possible. But the facts are
too uncertain for it to be worth while entering into
details. There are two fragments of stone concerned,
and Horsley, who compared them carefully, was con-
vinced that there was no connection between them.4
His own solution, however, can hardly be correct. He
suspects the presence at Castlecary of a cohors Brit-
tonum, whereas there is no authentic instance of these
cohortes having been employed within the limits of the
island.5 The one point that is clear is that an altar
dedicated to the Matres was concerned.

No. 43 (C.I.L. vii. 1096) was found not far from
No. 42,6and, like it, is now in the Edinburgh Museum.
It is 3 ft. 3 in. in height and i ft. 2 in. in breadth.

1 See supra, p. 50. 2 Hist. Inq. p. 48.

3 C.I.L. vii. No. 1094. *Brit. Rom. p. 201.

6 See supra, p. 56. 6 Stuart, Caled. Rom. 2nd ed. p. 350.

PLATE XLVT

i. No. 41

^^^

^^^^^^m|(J«yu "tW

M*

-i~.'+ifJ

2. No. 42

4. No. 46
ALTARS FROM CASTLECARY AND POLMONT

ALTARS 345

It had stood in a shrine, perhaps a shrine of Silvanus
outside the fort, as No. 38 had done at Bar Hill. The
inscription is exceedingly hard to decipher. But, with
the possible exception of the first two lines,1 Hiibner's
version, which is as follows, would appear to be

correct :

DEO

SILVANO

CONORS I

FID VARDVL

C R EQ oo

CVI • PRAEST

TREBIVS

VERVS PR

AEF

The name of the cohort is more important here than
the name of the divinity. It gives us two regiments
of auxiliaries, each 1000 strong, associated with Castle-
cary, just as we got two ordinary cohorts associated
with Bar Hill. The First Cohort of Tungrians was
mentioned in No. 28, and now we have the " First
Loyal Cohort of Vardulli, Roman citizens, 1000 strong,
with a contingent of cavalry." The latter body was
in Britain as early as the year 98 A.D. They appear
to have lain first at Lanchester,2 and subsequently to
have been moved up to Bremenium (High Rochester),
where they have left a large number of inscriptions.3

1 Prof. Haverfield, who examined the stone many years ago with Sir
W. M. Ramsay, wrote : " The upper part of this altar is now beyond
certain decipherment, but we could detect nothing at all like Dr. Hiibner's
Deo Silvano" (Archaeol. Journ. 1. 1893, p. 304). Nevertheless, repeated
examination in various lights has convinced me that the first line is
correct. The last three letters of the second are also discernible ; the
rest of it is very obscure and doubtful. The whole inscription is so sadly
defaced that an illustration would have been useless.

2 C.I.L. vii. Nos. 435, 440. 3 C.I.L. vii. Nos. 1030, 1031, etc., etc.

346 OTHER INSCRIPTIONS

The fact that both of the cohorts that seem to have
been at Castlecary were 1000 strong — and the milliary
sign is the most certain element in the whole of this
inscription — confirms the idea that the fort was an
important one. It has been suggested that the prefect
Trebius Verus, whose name occurs on No. 43, is
identical with the " L. Tanicus Verus Praef." on the
lost No. 3 1.1 However that may be, it is remarkable
that the commanding officer of a milliary cohort should
be a praefectus merely, and not a tribunus. The only
regiments in the case of which this normally occurs
are the two Tungrian cohorts that are known to
have served in Scotland.2 But this is not the solitary
instance in which the Cohors Prima Fida Vardullorum
appears as having a praefectus. A parallel from Africa
can be cited.3

No. 44 (Plate XLVI. 3; CJ.L. vii. 1097 I Tit. Hunt.
p. 75) is a mere fragment, bearing only one word :

DEAE

It was found at Castlecary, while the canal was being
made. The dimensions are i ft. 3^ in. by n^ in.
Another fragment from the same site, now in the
Edinburgh Museum, measures 9 in. by 9^ in., and
represents the lower part of a very much smaller altar
(CJ.L. vii. 1098). The greater portion of the last
line is visible, as well as parts of some of the letters
in the line immediately above. Few, however, will
venture further than Gordon :4 "I can make nothing
of any of them, excepting the last Four " — v • s • L • M.

1 Cichorius in Pauly-Wissowa's Real-Encydopadie, iv. p. 348.

2 See suflra, p. 53, footnote. 3 C.I.L. viii. No. 5532.

4 Itin. Sept. p. 57, Plate 10, Fig. 5. In the Catalogue of the National
Museum (1892) the fragment (p. 225, FV 12) is said to come from

ALTARS 347

No. 45 (Plate XXXIV. i ; C./.Z. vii. 1092) was dis-
covered in 1843, two or three hundred yards south of
the fort of Rough Castle. It is now in the National
Museum of Antiquities in Edinburgh. As it stands,
it measures 2 ft. 3^ in. by i ft. 5^ in., but the top is
broken away. The letters of the inscription are some-
what rudely cut. The whole appears to read :

VICTORIA[E]

COM V] NER

VI OR VM CO

FL BETTO> LEG

XX V V
V S L-L-M

The precise interpretation of the letters c c is
doubtful, although their general sense is clear. The
expansion ' cujus cur(am)-agit ' has been suggested by
Professor Haverfield,1 the last two words being treated
as one, after a manner that is not unknown. In any
case the meaning is that the centurion Betto — the
name, by the way, is a peculiar one — had been placed
in command of the auxiliary regiment that garrisoned
Rough Castle, just as Cocceius Firmus appears to
have been in command at Auchendavy. The regi-
ment concerned has been discussed under No. 27,
and the divinity under No. 32. The translation
runs : " The Sixth Cohort of Nervii, commanded by
Flavius Betto, a centurion of the Twentieth Legion,
the Valerian, the Victorious, [erected this altar] to
Victory. Willingly, gladly, justly has he performed
his vow."

Craigend, Croy. The confusion is due to its having been presented to
the Museum along with Nos. 22, 23, and 24.

1 Arch. Aeliana, xvi. p. 80.

348 OTHER INSCRIPTIONS

No. 46 (Plate XLVI. 4 ; C.I.L. vii. 1090) was dis-
covered in 1841 "near the Bridge at Brightens, to
the South-east of [Falkirk] in levelling the slope to the
south of the road " l during the formation of the Edin-
burgh and Glasgow Railway. The spot indicated is
a mile to the south of the Limes opposite Mumrills,
but the stone is usually described under the latter
heading. As we have seen, there is good reason to
believe that a fort once stood there. The inscription

reads :

HERCVLI

MAGVSAN
SACRVM
VAL NIGRI
NVS DVPLI
ALAETVN
GRORVM

That is, " Sacred to Hercules Magusanus. Valerius
Nigrinus, Duplicarius of the Tungrian Cavalry Regi-
ment [was the dedicator]."

The altar, which is now in the Edinburgh Museum,
measures 2 ft. 9 in. in height by i ft. ^ in. in breadth,
and has an unusually large base. It presents one or
two points of interest. The ala Tungrorum, to which
this non-commissioned officer belonged, is presumably
identical with the Ala I Tungrorum, mentioned as
forming part of the garrison of Britain in diplomas of
the years 98 and 105 A. D. It is also named in a lost
inscription from the western end of the English Wall.2
In any event, it is the first purely cavalry regiment we

1 Letter addressed to the Secretary of the Soc. of Antiq. of Scot, by
J. W. Reddoch, dated 5th June, 1841 ; the altar had been discovered the
week before. Cf. Gent. Mag. 1841, ii. 78, 303.

2 C.I.L. vii. No. 941.

ALTARS AND TOMBSTONES 349

have met with along the line of the Limes. We do
not know definitely that it was stationed at Mumrills.
But the position would be suitable ; the Carse of Fal-
kirk would give ample scope for the movements of
horsemen. The appearance of Hercules as a Teutonic
divinity has already been remarked upon in speaking
of No. 33. Here we have a special manifestation of
the phenomenon. Originally Hercules Magusanus
was a local god, indigenous in all probability to the
Lower Rhine.1 Later he acquired a much wider
celebrity, notably through being used as a type on
the coins of the Emperor Postumus. In this case he
probably bears a character almost as strictly local as
that of Viradecthis or Ricagambeda, honoured by
another body of Tungrian auxiliaries at Birrens.2

No. 46 completes the list of altars. The tombstones,
which alone remain to be dealt with, are very few in
number. This is matter for regret, since the informa-
tion conveyed by such inscriptions is often extremely
valuable. Three out of the five that are known were
found in a ' tumulus ' near Shirva, about a mile east
of Auchendavy. Along with them were two large
uninscribed slabs bearing a sculptured representation
of the Sepulchral Banquet (Plate L.). According to
a contemporary account, written by the then minister
of Kilsyth and reprinted in the Tituli Hunteriani?
the 'tumulus' was "in the Fossa, close by the wall."

'See R. Peter in Roscher's Lexicon, \. pp. 3018 ff. Some prefer to
regard him as Celtic.

2 Proc. of the Soc. of Antiq. of Scotland, 1896, pp. 143 and 157.

3 Pp. 87 ff. It was originally published in Gordon's Additions, a rare
tract published a year or two later than the Itinerarium itself. Another
letter from Mr. Robe, the minister in question, is quoted by Horsley,
Brit. Rom. p. 339.

350 OTHER INSCRIPTIONS

The details that are given leave no room for doubt
as to the place having been a Roman cemetery.
Nos. 47-49 all seem to have come from the im-
mediate neighbourhood, although No. 47 is the only
one which we know to have been certainly got within
the ' tumulus.' With it there were "also other stones,
whereof only parts are found, having D.M. for Diis
Manibus."1

No. 47 (Plate XLVII. i ; C.I.L. vii. 1 1 18; Tit. Hunt.
p. 65) is only the upper half of the original stone.
The top is rudely decorated with a triangle enclosing
a rosette and having a smaller rosette at either side
of it. What is left of the slab measures 2 ft. 2 in. by
i ft. 5f in. The inscription reads :

D M

FLA LVCIA
NVS MILES
L E Q 1 1 A V G

That is, " To the Divine Manes. Flavius Lucianus, a
soldier of the Second Legion, Augustus's Own." The
missing portion no doubt gave the dead man's age
and the number of his years of service, probably
also the name of the relative who had set up the
stone.

No. 48 (Plate XLVII. 2 ; C.I.L. vii. 1119; Tit. Hunt.
p. 66) was found not far from No. 46, but some four
or five years earlier. It shows a distinct striving after
elegance of ornament. In the centre of the upper part
is a large wreath, flanked by two palm-branches, each
of which has a rosette above it. The framework
enclosing the inscription is of the cable pattern at
the sides, but plain at top and bottom. The

1 Mr. Robe's letter to Professor Maclaurin. See Tit. Hunt. p. 88.

TOMBSTONES 351

dimensions are 3 ft. 8 in. by i ft. 5 in. The letters

read :

D.M Q

SALMAI E
VIX AN XV
SALMANES
POSVIT

That is, " To the Divine Manes. Salmanes, aged
fifteen. Erected by Salmanes." It is natural to
suppose that this is a stone set up by a father over a
son. The name Salmanes is, of course, not Roman.
Professor Haverfield suggests that it is Semitic, " the
same probably as Solomon and the first half of Shal-
manesar." l This seems very likely to be right. If it
is, we have here a proof that Eastern traders had found
their way as far north as the Caledonian frontier in the
wake of the Roman army. They are met with in many
different parts of the Empire.

No. 49 (Plate XLVI I. 3 ; C.LL. vii. 1 1 20 ; Tit. Hunt.
p. 67) is also from the neighbourhood of Shirva. It
is more elaborate than either of the other two, but
is unfortunately incomplete. It is i ft. 6J in. broad,
and was probably originally fully 3 ft. high. It gives
practically no information save the name of a dead
lady :

D- M •

VEREC
VNDAE

No. 50 (C.LL. vii. 1 1 10) was copied in the neighbour-
hood of Kilsyth by the German scholars who visited
Scotland in search of inscriptions in the opening years
of the seventeenth century.2 It has long been lost. In

1 Tit. Hunt. p. 67. 2 See supra, p. 41.

352 OTHER INSCRIPTIONS

general design it seems to have resembled the pre-
ceding tombstones. The inscription was as follows :

DM
c IVLI

MARCELLINI

PRAEF
COM 1 HAMIOR

That is : " To the Divine Manes of Caius Julius
Marcellinus, Prefect of the First Cohort of Hamii."
The mention of the Hamii confirms the inference one
would naturally draw from the fact that the inscription
was seen at Kilsyth. It must originally have stood in
the cemetery attached to the Bar Hill fort. We learned
from No. 38 that the Hamii were at one time in
garrison there. We now know that during that period
they lost a commanding officer. Whether he fell in
action or died of disease, it is impossible to say.

No. 51 (Plate XXXIV. 2; C.S.L. vii. 1091; Haver-
field, Archaeological Journal, 1. (1893), p. 304) is the
most interesting of all the tombstones. It measures
i ft. 6 in. by i ft. 3 in. ; and the inscription reads :

DIS • M NECTOVELIVS F

VINDICIS AN XXX

STIP • Vllll NAT

IONIS • BRIGANS

MILITAVIT IN

COM - il T4R

That is : " To the Divine Manes. Nectovelius, son of
Vindex. Aged thirty. A Brigantian by birth, he served
for nine years in the Second Cohort of Thracians."

This monument, now in the National Museum in
Edinburgh, was found in the neighbourhood of Mum-
rills. Standing alone, it hardly justifies the positive

PLATK XLVII

ti --i

$/f r,i*.t*

3 4

TOMBSTONES FROM SHIRVA

TOMBSTONES 353

assertion that the station there was ever garrisoned by
the Second Cohort of Thracians ; Nectovelius may
have been killed in a campaign. But it distinctly
suggests the possibility, just as No. 46 suggested a
similar possibility for the Cavalry Regiment of Tun-
grians. Mumrills, then, may be analogous to Bar Hill
and to Castlecary in that its garrison has been changed
in the course of the occupation. Two quite different
regiments seem to have borne the title of ' Cohors II
Thracum Equitata! One belonged to the army of
the East.1 The other was permanently attached to
the exercitus Britannicus, and it is of course this latter
that concerns us here. It was in the island at least as
early as the beginning of the second century A.D., for it
is mentioned in a diploma of the year 103. Inscriptions
show that it was posted for some time at Moresby on
the coast of Cumberland.2 When the Notitia Digni-
tatum was compiled, it was stationed at Gabrosentum,
probably somewhere in the western part of Cumber-
land. The stone has already been referred to as
furnishing an interesting example of local recruiting
for a corps that was nominally of foreign origin.3

A glance through the inscriptions collected in this
chapter shows that, while the legionaries constructed
the Limes, it was the auxiliaries who acted as its
garrison. This is in entire accord with the recognized
principles of Roman Imperial policy.4 The auxiliaries
were the pawns in the stern game of frontier war. The
legions were only set in motion when serious trouble

1 Cichorius in Pauly-Wissowa's Real-Encyclopadie, iv. p. 339.

2 C.I.L. vii. No. 363 f. ; Ephem. Epigr. vii. No. 967.

3 See supra, p. 58. * See supra, p. 67.

354 OTHER INSCRIPTIONS

was afoot. The names of eight of the particular
regiments concerned are known, and the probable
association of these with forts is as follows :

CASTLE HILL: Cohors 7777 Gallorum

_ . _ ( Colors I Hamiorum

D A K n 1 1_ L : •< _ .

\CohorslBaetasiorum C.R.

CASTLECARY: (Cohors { Tungrorum 66

\ Cohors I Fida Vardullorum C.R. °o

ROUGH CASTLE: Cohors VI Nerviorum

MUMRILLS: (Ala J Tungrorum
\Lohors H Thracum

Meagre as this information is, its value is consider-
able. To begin with, it is remarkable that in the
case of three out of the five forts the names of two
different regiments are mentioned. The probabilities
are all in favour of successive rather than of simulta-
neous occupation being indicated. Such succession
might, of course, have taken place without any breach
of continuity. But, in view of ordinary Roman practice,
the proportion of changes seems abnormally high when
account is taken of the comparatively short time that
the line remained in Roman hands. And the number
of inscriptions recovered from the two remaining forts —
one from Castle Hill and two from Rough Castle — is far
too small to preclude the possibility of a second regi-
ment having been associated with them also. The
hypothesis of a temporary abandonment may not be
necessary, but it certainly suggests itself as a very
natural explanation of the patent facts.

Again, the material is sufficient to enable us to form
an estimate of the size of the force employed to hold
the Limes. The maximum number of castella was 19.
Of the five here included, there were four which

THE GARRISON 355

seem to have accommodated 500 men apiece. The
fifth, Castlecary, evidently had quarters for 1000.
Assuming the same proportion to have been main-
tained throughout — and, if the assumption errs, it is on
the side of generosity — we should get four forts holding
1000 men each and fifteen holding 500. Even if we
reckon every regiment at its full strength, this gives
us a total of less than 12,000 men. When allowance
is made for contingencies, 10,000 will probably seem
nearer the mark. There is one further point to notice.
At Bar Hill and at Rough Castle there do not appear
to have been any horsemen. At each of the other
three forts, and probably also at Auchendavy,1 there
was a cohors equitata. On one occasion, indeed, the
garrison of Mumrills may have been entirely mounted.
Still, on the whole, the Limes was manned by infantry,
as we should naturally expect it to have been. The
only spot where cavalry figure prominently is, as has
already been remarked,2 close to the Carse of Falkirk.

1 See supra, p. 333. 2 See supra, p. 349.

CHAPTER XI
MISCELLANEOUS EVIDENCE

OVER and above the inscriptions, a multifarious mass
of objects, undoubtedly Roman in their origin, have
from time to time been recovered along the line of the
Limes. It is matter for regret, but hardly for surprise,
that the number of those now available for study is
comparatively insignificant. Only such as have come
to light within the last decade have been systematically
preserved. Except for the sculptured stones, which
were saved by their very massiveness, the finds of the
seventeenth and eighteenth centuries are known to us
merely through vague descriptions or hardly more
accurate drawings. Similarly, the rich harvest that
might have been gathered at Castlecary in 1841, when
the railway was driven right through the heart of the
fortified enclosure,1 is represented to-day by a few waifs
and strays, principally sherds of pottery, that have
drifted into the National Museum. Against these
losses may be set the fact that much more than has

1 " The soil, if it can be so called, was in many places almost one entire
mass of broken stones mingled with fragments of pottery, among which
last were many pieces of jars, vases, and basins — some of a cream
colour, and others of a lively red, elegantly ornamented with flowers and
figures." So wrote an eye-witness a few years later (Stuart, Caled. Rom.
2nd ed. p. 348).

DEBRIS OF THE FORTS 357

ever been discovered must still be in safe keeping
underneath the ground. The sites of many of the
forts are virgin soil for the explorer. Any attempt at
a detailed description of the surviving relics would be
out of place in a general sketch. But it will be
useful, even though it involve a certain amount of
repetition, to pass the various classes in brief review,
dwelling mainly on those which may be expected to
throw light on questions of chronology and so help
towards a solution of the historical problem with which
we have to deal. Apart from this, there are one or
two individual articles, not definitely datable, which call
for particular notice.

Nothing could well be more miscellaneous in
character than the debris of a Roman frontier fort.
Taken together with its annexes, it was a microcosm
not only of military but of civil life. The structural
remains have already taught us that the interior
buildings were fairly substantial. The casual finds
supply further evidence of the pains taken to make
them wind- and water-tight. Many of the corroded
iron objects scattered here and there a little way below
the surface — nails of great size, staples, holdfasts, rings,
hooks, and the like — must originally have been attached
to doors or walls. Red roofing-tiles are common, and
so is window-glass. The circumstance that the latter
was apparently manufactured on the spot1 is significant
as illustrating the extent to which the castellum was an
independent industrial unit. Of luxury there is little
trace. Yet one or two buildings seem to have had
superior appointments ; tiny fragments from Bar Hill
and from Castlecary bear testimony to occasional

1 Roman Forts on the Bar Hill, p. 79.

358 MISCELLANEOUS EVIDENCE

mosaic floors.1 The columns and capitals taken out of
the well of the former fort may be regarded as typical
of the architectural pretensions of the Principia. Some
of the capitals deserve particular study, as being calcu-
lated to throw an interesting light on the evolution
of this form of architectural decoration, the cubical
shape being specially noteworthy.2 The Baths, and
probably also the house of the Commandant, were
heated by an arrangement of hypocaust chambers ;
portions of the characteristic pillars often occur as
isolated finds, to say nothing of those that have been
discovered in situ.

The decorative impulse found expression in sculpture
of a more or less crude kind. A good example is the
"figure of a woman cut in stone" (Plate XLVIII. i)
from the baths at Duntocher.3 But for her armlets
she is naked to her waist, and she holds in front of
her a large shell pierced with an orifice for water.
Apparently it is a nymph that is represented. The
statuette is, in fact, a characteristically degenerate
reproduction of a well-known Greek type.4 The
figure is now in the Hunterian Museum. With it
is the somewhat clumsily carved torso of a man in
armour (Plate XLVIII. 3), found in a pit at Auchen-
davy with the altars of Cocceius Firmus.5 Much
more skill is displayed in a fragment discovered
at Rough Castle during the excavations of 1903.

1 Roman Forts on the Bar Hill, p. 78; Proc. Soc. of Antiq. of Scot.
1903, P- 338.

2 See Plate XX. 2, and page 197, Figs. 4-7.

3 See supra, p. 158, footnote. It is i ft. u in. high.

4 See, for example, Reinach, Repertoire de la statuaire grecque et romaine,
ii. p. 405.

6 See supra, p. 185, footnote. It measures 11 in. by 8£ in.

PLATE XLVIII

I. FROM DUNTOCHKK

2. FROM CASTLECARY

3. FROM AUCHENDAVY

4. FROM ARNIEBOG. 5. LOCALITY UNKNOWN

MISCELLANEOUS SCULPTURES

SCULPTURES 359

This is part of the shoulder of what may have been
a life-size statue in sandstone (Plate XXXI. i).
Possibly it was the Emperor with the paludamentum ;
his was certainly the statue that was most likely to have
a place in the fort, and the superiority of the workman-
ship would in this way be readily accounted for. It
presents a striking contrast to the curious group of
busts from Bar Hill (Plate XLIX.). Two of these are
fragmentary. The other two are fairly complete.
Each of them has been carefully squared upon the
bottom, as if to stand upon a pillar or pedestal. The
pillars or pedestals may have flanked the entrances to
some of the public buildings, for the busts fall into pairs
according to their subject. The two that are broken
(Nos. i and 3) have been effigies of the bald-headed
Silenus raising a wine-cup to his lips. The others
(Nos. 2 and 4) have bearded faces, and arms folded
across the chest, while each of the three hands — the
left hand of No. 3 has not been carved — shows the
middle finger thrust boldly out from a closed fist. The
gesture is that of the in/amis digitus, but the purpose
here was probably to provide a charm against the evil
eye. More frankly religious is the little statuette of
the goddess Fortune, standing within a niche and
holding a rudder and a cornucopiae (Plate XLVIII. 2),
which came from the baths at Castlecary. The stone
is i foot 3 inches high.

The sculptures that have so far been described are
all readily accessible in collections. Another, dis-
covered rather more than a century ago at the foot of
the northern face of Croy Hill, is built into the wall at
Nether Croy House along with a fragment of what
seems to have been an inscribed slab, found at the

360 MISCELLANEOUS EVIDENCE

same place and time. These relics were discussed and
figured long ago in Archaeologia by the Rev. John
Skinner, whose illustrations need correction.1 The
larger (Plate LI 1 1. 2), which is i foot 8J inches high
by i foot 4f inches broad, is the left-hand portion of
a decorated slab that appears to have been originally
inscribed. A figure of Venus quitting the bath,
glancing furtively behind her as she steps to the left,
is seen between two Corinthian pillars, from the inner-
most of which an arch has sprung. A portion of a
large wreath, which has occupied the centre of the stone
and which must have contained the inscription,2 is
visible to the right, while a naked figure is huddled
into the corner between the wreath and the base of
the pillar beside it. The one which is complete (Plate
LI 1 1. i) is the more interesting. It measures i foot
i^ inches by n^ inches, and represents three soldiers
standing to front, side by side. Each of the two
towards the left has a pilum in his right, and supports
upon the ground, with his left, an oblong shield of the
type usually borne by the Roman legionary, curved
inwards so as to encircle the body. The one on the
right carries a similar shield, raised so as to protect
himself as when in action, and holds a drawn sword 3
in his right hand. The stone must have been intended
to adorn some building.

Alongside of these two stones from Croy may be

1 Op. cit. XXI. (1827), PL XXI.

* In 1826 Dr. John Buchanan read the letters vi near the broken edge.
When he next saw the slab, a quarter of a century later, it had been
further damaged and the letters had disappeared (Stuart's Caled. Rom.
2nd ed. p. 341, footnote). The inscription is C.I.L. vii. 1105.

3 In regard to this detail, Stuart's text (Caled. Rom. 2nd ed. p. 341) is
more correct than his illustration, PL XIII. 4.

SCULPTURES 361

placed "a small thin slab, about 18 inches long, by
1 2 inches broad, much worn as if from exposure to the
weather . . . having sculptured rudely upon it the
figure of a Roman soldier, and diamonded on the back
by the chisel. Its appearance shows that it was set
into some building."1 This was turned up about 1840
or 1850 within the area of the fort of Castle Hill. It
had disappeared before 1852. Somewhat of the same
nature is "a long-shaped block of freestone" from
Castlecary, measuring 27 inches by 9 inches, and having
carved upon it a very rude representation of a stag or
goat hunt.2 It may still be in existence, for it was in
Dr. John Buchanan's collection,3 where were also the
upper part of a female figure, carrying something on
her right shoulder, likewise from Castlecary,4 and a
stone with a filleted bucranium upon it, said to have
been dug up at the Peel of Kirkintilloch.5

The most remarkable of the sculptures have still to
be mentioned — the three that were taken with the
tombstones from the Roman cemetery at Shirva. All
are in the Hunterian Museum. There is no doubt as
to the proper interpretation of the two largest (Plate
L.), the dimensions of which are 2 ft. uj in. by
2 ft., and 3 ft. 2\ in. by 2 ft. if in. Each shows
the figure of a man reclining on a couch with a
dog beside him. In all likelihood a three-legged
table was originally included in the scene. That is
what analogy would suggest, but both reliefs are so

1 Stuart's Caled. Rom. 2nd ed. p. 310, footnote.

2 Stuart's Caled. Rom. 2nd ed. PI. XIV. Fig. 11.

3 Ibid. p. 351, footnote.

4 Ibid. p. 348, footnote, and PI. XIV. Fig. 5.

5 Ibid. p. 324, footnote. See, however, supra, p. 177

362 MISCELLANEOUS EVIDENCE

seriously damaged at the foot that certainty is out
of the question. In any event the sculptures belong
to a class of funeral monument rather rare throughout
the empire generally, but very common on the German
and British frontiers in the first, second, and third
centuries A.D., a peculiarity of distribution to which
Professor Haverfield was the first to draw attention.1
They represent a banquet. The method of treating
the subject, as well as its special association with the
dead, was borrowed by the Romans from the Greeks,
either directly or through the Etruscans, and the
Greeks in their turn had adopted the form from the
Assyrians. How far the underlying idea was the same
in all these cases, must always remain a matter of doubt.
It is curious that, as Professor Haverfield points out,
this type of relief seems to be, in imperial times, often
associated with the graves of women and girls.

The last of the sculptures from Shirva (Plate
XLVII. 4) may be the upper portion of a tombstone,
and may be intended for a portrait of the dead. On
it a figure of a man, apparently, a soldier, is seen
standing to front, holding in his right hand a spear
and in his left a basket-like object with a handle.2
Horsley well compared this fragment with the monu-
ment of L. Duccius Rufinus, standard-bearer of the
Ninth Legion, now in the Museum at York.3 A com-
parison even more apt might be made with a monu-
ment at Athens to the memory of Q. Statius

1 ' The Sepulchral Banquet on Roman Tombstones ' in Archaeological
Journal, Ivi. pp. 326 ff.

2 Hiibner (ArchdoL Zeitung^ 1868, p. 41) regarded it as a set of tablets,
possibly the man's will.

zBrit. Rom. p. 308. The stone is C.I.L. vii. No. 243.

PLATE L

THE SEPULCHRAL BANQUET ON SCULPTURED RELIEFS
FROM SHIRVA

TOOLS AND WEAPONS 363

Rufinus, a marine of the Praetorian fleet. It is
much superior in execution to the Shirva stone, but
the essential elements are identical.1 Lastly, we may
note here a broken tablet, 17 inches square, which
somehow suggests a grave relief, although the locality
where it was found is quite uncertain (Plate XLVIII. 5).
It is now in the Hunterian Museum (Tit. Hunt. p. 93).
A bent figure, presumably that of an aged man, leans
forward heavily on a staff, while behind him is what
appears to be a youth seated.

Among the corroded fragments of iron, portions of
tools and implements can be recognized with more or
less confidence. These show that the Roman masons,
smiths, and carpenters used implements that do not
substantially differ from those that are employed in
similar circumstances to-day. As a rule, they are
rusted and broken. One or two in good condition
came, however, from the pits at Bar Hill, and it is
clearly only in the pits that we can hope to find
objects of metal in reasonable preservation. The huge
mallets from Auchendavy have already been described.2
The hoe from Rough Castle suggests gardening or field
work (Plate LI I. 3), and we may be sure that some
of the many sharpening stones found at Bar Hill and
Rough Castle had belonged to mowers or reapers.
Weapons have so far been very rare. Occasional
spearheads, one or two stray pieces of shield mount-
ings, and a few arrowheads from the well at Bar Hill
very nearly exhaust the list. Stone bullets for the
ballistae and other engines are of common occurrence ;
the find at Auchendavy was mentioned above,3 and

1 It is described and figured by Hiibner, I.e. PL 5, i.

2 Supra.) pp. 185 f. 3Su#ra, p. 184.

364 MISCELLANEOUS EVIDENCE

more than a hundred were recovered at Bar Hill. The
latter vary in diameter from 8 inches to i inch, thus
pointing to engines of different degrees of power.
Horses were, of course, present in the forts, partly as
mounts for the cavalry and partly for purposes of
transport. Bridle-bits and pieces of harness are there-
fore not seldom in evidence, while two different sorts
of wheel were represented at Bar Hill.1 Many of the
numerous pieces of deer-horn have obviously been sawn,
doubtless because the part removed was to be manu-
factured into something. Horn could be utilized in
many ways, and it has been employed in producing
what is certainly the most puzzling set of objects that
the excavations on the Limes have brought to light —
six pieces found in different quarters of the Antonine
fort at Bar Hill. Two of these are shown in the
accompanying illustration (Figure 14). The only
parallels hitherto recorded are from the armoury of
the legionary fortress of Carnuntum, a provenance
which suggests that it is in military equipment that
a clue to the nature and purpose of the objects should
be sought. No really plausible hypothesis has yet
been advanced to explain the peculiar features by
which they are distinguished.2

Leather is another material that has stood the test
of time fairly well. Portions of the leather clothing of
the soldiery have survived, and great numbers of cast-
off articles of foot-gear have been gathered from pits
and ditches. Some examples from Castlecary are shown

1 Roman Forts on the Bar Hill, pp. 92 flf. The same two types occurred
at Newstead : see Curie, A Roman Frontier Post, pp. 292 ff.

2 See Roman Forts on the Bar Hill, pp. 122 ff. for full description and
discussion.

HORN AND LEATHER

365

on Plate LI. (3-5). The Bar Hill collection is specially
rich in this respect. The
variety of size and of pat-
tern displayed is worthy of
remark. Sandals, properly
so called, seem to be rare ;
such a protection would be
ill-adapted for out-of-door
use in Scotland, unless per-
haps in the height of summer.
Shoes, again, are of two
distinct types. In one case,
sole and upper are cut from
a single piece, and the
former is always smooth
and without nails. In the
other, sole and upper are
quite distinct, much as in a
modern shoe, while the sole
is formed of several layers,
generally four or five, and
is studded with heavy nails
even in the smallest sizes.
The nails are sometimes
arranged in curious patterns.
But it was mainly on the
uppers that decorative effort
was lavished. One is almost
tempted to think that there
hardly any two pairs were
made alike. And this ap-
plies to all sizes, as can be
seen from Plate XXI. i, FlG" '^iS'SL0* <?)°RN FR°M

366 MISCELLANEOUS EVIDENCE

which reproduces a man's, a woman's, and a child's
shoe, mounted on modern 'trees.' Naturally, women's
shoes are the most highly ornamented. A fine example
is shown upon Plate XXI. 2. The fact that these and
the shoes of children occurred so constantly in the pits
and ditches of the fort itself, and not of any annexe,
at Bar Hill tends to raise doubts as to whether the
exclusion of all save soldiers from the castella was as
rigidly insisted upon as is usually believed.

A few trinkets tell the same story as the more
elaborately decorated shoes. The enamelled fibula
described and figured in Stuart's Caledonia Romano^
may fairly be associated with the Romans, though there
is no evidence that it was found within the limits of a
Roman fort. Its shape plainly proves it to be a second-
century type, just as are the two bronze fibulae from
Castlecary (Plate LI. if.); the one on the right is an
example of the ' knee ' fibula, a comparatively late
development, while the other belongs to a variety that
probably continued in use over a long period. Ribbed
melon-shaped beads of blue porcelain paste are reported
from more than one site, and there are traces of
bracelets and of necklaces of glass. If we set the shoes
aside, there is little to indicate any fondness for per-
sonal adornment on the part of the men, although the
rarity of such articles is perhaps most simply explained
by the circumstance that only one fort, and that not a
large one, has yet been exhaustively explored. At
Bar Hill some of the fragments of pottery from the
neighbourhood of the baths appeared to have been
intended for unguents. Seals that have dropped from
rings have been picked up both at Auchendavy and
1 Op. cit. 2nd ed. p. 295 and PI. VII. Fig. 6.

Pl.ATK LI

FIBULAE AND SHOES FROM CASTLECARY

MINOR OBJECTS 367

at Castlecary.1 But the list of objects suggesting refine-
ment is meagre at the best. Potsherds, rounded into
men such as might be used in playing draughts, are
perhaps reminiscent of the long, bleak winter evenings
that the watchers of the Caledonian frontier had to
pass. Lamps are scarcer than might have been
expected ; possibly oil was difficult to procure, and

FIG. 15.— LAMP OF STONE FROM ROUGH CASTLE

rush-lights therefore generally in vogue. But Gordon
says that at Castlecary "was dug up a most curious
Roman Lamp of Brass, adorn'd with Variety of
Engravings, which Mr. Alexander Drummond, a
learned Friend of mine, saw hereabouts, some Years
ago; but into whose Hands it has now fallen, I cannot
tell."2 One or two examples of the ordinary clay

1 See supra, p. 187, and Proc. Soc. of Antiq. of Scot. 1903, p. 338.

2 Itin. Sept. p. 57. Possibly this is the lamp seen in 1697 by the writer
of the letter preserved at Welbeck Abbey (Hist. MSS. Commission:
Portland Papers, ii. p. 57).

368 MISCELLANEOUS EVIDENCE

lamp are known, and a rudely cut piece of stone, found
at Rough Castle in 1903, seems to have served a
similar purpose (Figure 15).

In describing the Bar Hill fort, allusion was made
to the testimony which the refuse-pits afforded regard-
ing the food of the garrison.1 The shell-fish must have
been brought from the coast. The cattle we may
suppose to have been kept close at hand. Wheat was,
of course, extensively used. A large quantity must
have been left behind at Castlecary when the Limes
was abandoned. Nearly a hundred quarters are said
to have been discovered just outside the fort in i77i,2
while many scattered grains were dug up inside the
walls during the excavations of 1902. Whether it was
grown on the spot, we do not know. Not improbably
it was. It was certainly meant to be ground there,
for querns have been found at Duntocher, Cadder,
Bar Hill, Croy Hill, Rough Castle, and Mumrills,
and Gordon saw a portion of one at Auchendavy.3
Usually these are of Niedermendig or other lava, and
undoubtedly imported. For oil and wine, which with
bread and meat furnished the staple of sustenance,
the garrisons would depend upon supplies from the
south. We are accordingly justified in assuming the
existence of a well-organized system of land-transport.
But the task of the commissariat department was in
all likelihood greatly simplified by the Roman com-
mand of the sea. We may be sure that there was a
harbour somewhere on the Forth where vessels from
Gaul would discharge their cargoes.

1 See supra, pp. 200 f. 2 Stuart, Caled. Rom. 2nd ed. p. 346.

3 Itin. Sept. p. 54 : "a Stone with a good many hollow Circles upon its
Surface, which Kind of Stones the Romans used for grinding of their Corn."

PLATE LH

MISCELLANEOUS OBJECTS FROM CASTLECARY
AND ROUGH CASTLE

THE COMMISSARIAT 369

Wooden barrels, such as those whose staves were
found at Bar Hill, were perhaps used in the fort chiefly
for storage. But originally they were certainly em-
ployed for carrying purposes ; a relief on the column
of Trajan shows a number of large barrels being
landed from a boat beside one of the praesidia on the
Danube.1 The smaller ones may have held butter.
Cheese-making, by the way, is possibly suggested by
a portion of a shallow vessel of dark ware, pierced
with holes at regular intervals, recovered at Castlecary
in 1902 (Plate LI I. 2). A somewhat similar object
was got at Bar Hill. Bottles of square-moulded shape
with reeded handles, usually of bluish-green glass and
sometimes tolerably large, must have come from a
considerable distance, and it may be assumed that
they did not come empty. But the bulk of the oil and
the wine must have travelled northwards in the great
clay amphorae whose fragments occur with such fre-
quency wherever the soil in any of the forts is
seriously disturbed. A characteristic specimen was
taken, almost complete, from the bottom of the well
of the fort at Bar Hill. Apparently it had been used
for drawing water in an emergency. An example of a
bung that might have fitted such an amphora was got
at Castlecary (Plate LI I. i). It is well cut out of some
coniferous wood and has been carefully made, being
provided with an ornamental bronze cap to which is
attached a bronze ring. Altogether it is a superior
article ; possibly the liquor whose safety it had
secured was of more than ordinary quality.

Many of the amphora-handles bear a potter's stamp
or a mark, recording either the capacity of the vessel

1 Cichorius, Die Reliefs der Trajanssaule, Taf. v. 7.

370 MISCELLANEOUS EVIDENCE

or the name of the manufacturer. Sometimes it may
be the nature of the contents or the place of origin that
is indicated. These inscriptions are frequently hard to
decipher. Not seldom they are wholly illegible. So
conjectural indeed are most of the readings that it does
not seem worth while attempting to bring them together
here. Careful comparison with better preserved speci-
mens from elsewhere would be essential to the com-
pilation of a reliable list. The same is true of the
similar stamps impressed upon the lips of the mortaria
or pelves, fragments of which are probably as common
as are the portions of amphorae. Both classes of
vessel are of the same coarse ware, and both vary in
colour. The pelvis was a deep basin, not unlike a
modern milk pan. A special feature was the very
large everted lip, pierced at one point by a grooved
spout. The surface was generally harder and finer
than that employed for many of the amphorae, and
the inside was roughened by an admixture of small
pebbles or pounded quartzite. These dishes would
seem to have been occasionally used for heating food.
The roughened interior, however, supplies an un-
mistakable clue to their most ordinary purpose — the
preparation of corn, fruit, or vegetables for the actual
process of cooking.1 The broad rim was intended to
provide a firm hold, while the spout was for draining
off the water employed in cleansing or in softening
during trituration.2

Besides the large amphorae, many varieties of
vessels were employed for storage. They differ a good
deal in colour and in quality, the finest being probably

1 See Roman Forts on the Bar Hill, p. 70.

2 See Jacobi, Das Rbmerkastell Saalburg, pp. 424 f.

POTTERY 371

a variety of black ware, one or two characteristic
examples of which were recovered from a pit at Bar
Hill. Black ware seems also to have been occasionally
used for serving food. But it was chiefly utilized for
cooking-pots. This is apparent from the fact that so
large a proportion of the fragments are covered with a
thick coating of hard soot. Usually they seem to have
been set into an iron framework that stood upon the
hearth. Sometimes, however, they were suspended,
much as was the case with the ordinary camp-kettles
of bronze. These are of rarer occurrence. But Bar
Hill provided a good example, exactly such a vessel as
Trajan's legionaries carry as camp-furniture in the
reliefs that show them on the march towards Dacia.

Of recent years much attention has been paid to the
classification of Roman pottery, and the careful study
of types from sites occupied or abandoned at a known
date is slowly providing the basis of a sound chrono-
logical arrangement. This is particularly true of
the red lustrous ware, usually called terra sigillata
by German archaeologists, though popularly known as
' Samian ' ; but it applies also to the coarser kinds,
where minute differences of form and texture mark the
development that took place between the end of the
first and the middle of the second century A.D. It is
significant for the history of the Limes that, of all the
fragments whose whereabouts is now ascertainable,
there is only one that can with any probability be
set down as 'early/1 The pottery of the forts is
emphatically second century.

1 1 am glad to have this opinion confirmed by Mr. J. Curie, whose
experience at Newstead lends the greatest weight to his opinion. He has
carefully searched all the collections, including that from Bar Hill, and

372 MISCELLANEOUS EVIDENCE

That ' Samian ' ware was abundant is clear.
Authentic instances of its occurrence are recorded
from the forts at Duntocher,1 New Kilpatrick,2 Cadder,8
Bar Hill, Castlecary, Rough Castle, Falkirk,4 and
Mumrills.5 Like the ' Samian ' from the English
Limes, as well as that from Scotland generally, it
seems to be predominantly Lezoux — with, however,
an unmistakable infusion of East Gaulish or German

FIG. 16. — FRAGMENT OF DECORATED 'SAMIAN' WARE FROM CASTLECARY

fabrics. Alike in their shape and in the style of
their decoration, nearly all of the pieces that survive,
or that are known from illustrations, appear to be such

has failed to find anything that he felt justified in assigning to the first or
early second century, with the single exception that will be noticed below.

1 See supra, p. 160. 2See supra, p. 165. 3See supra, p. 172.

4 See supra, pp. 238 f. There are also fragments in the National
Museum, Edinburgh. It is not, however, certain that the latter are from
the station that is supposed to have been at Falkirk. The donor (Mr.
Drumrnond Hay) may have included Camelon under the general
description of ' Falkirk.'

5 See supra, p. 240.

'SAMIAN' WARE 373

as one would naturally associate with the Antonine
period. Even at Bar Hill and at Rough Castle, where
traces of Agricolan forts were uncovered, no pottery
of the Agricolan age was observed. A typical example
of second-century work from Castlecary is shown in
Figure 16 ; it is in the style of the potter Divixtus.1
It may be worth while inserting here a revised list
of such makers' stamps as are comparatively legible.
The originals are all either in the National Museum
or in the Bar Hill collection.2 Missing letters are
supplied, where this can be done with certainty. All
the names, except that of Cinnamus, appear to be
on undecorated ware.

A E S T I V I M (Castlecary)

A L B I N I • M (Castlecary)

M • I N N A (' Falkirk ' 3)

A VI TVS- F (Bar Hill)

B E L I N I C I M (Bar Hill)

M I M A N N I O4 (Bar Hill ; Rough Castle)

CINTVSMVSF (Castlecary)

CRACVNA-F (Castlecary)

C V D C V I M 5 (Castlecary)

DIVICATVS (Bar Hill)

D O V 1 1 C C V S (Rough Castle)

G E N I A L I S • F (Bar Hill)

L I B E R T I M (Castlecary)

MALLVRO F (Bar Hill)

1 Abroad, vessels with the name of this potter are usually assigned to
the end of the first century. In Scotland and the north of England
associations suggest that they are later : see Curie, A Roman Frontier
Post, p. 194 and p. 229, and Haverfield, Roman Wales, p. 129.

2 1 have examined them all personally.

3 See above, p. 238, for another possible Falkirk stamp. The one here
catalogued perhaps belongs to Camelon.

4 As usual, the name of Cinnamus, where it occurs, is on the outside.

5 The third letter is almost certainly D, though it might possibly be R.

374 MISCELLANEOUS EVIDENCE

AA R I • M i (Castlecary)

CECVLIARISF (Bar Hill ; Rough Castle2)

P R I S C V S F (Castlecary)

TASCILLI M (Rough Castle)

L 3? • SECVS3 (Castlecary)

Other names, more or less fragmentary, will be found
in the various Excavation Reports. Sir Daniel Wilson,
too, mentions several on pieces now lost, including
PAT i RAT I • OF and SACIRAPO, from Castlecary, which
are probably right.4 One well-known stamp — that of
L FABRIC MAS — is here omitted, as the lamp that
bears it seems to be of doubtful genuineness. Setting
aside CVDCVIII, which may be blundered or misread,
one may say that the great majority of names in the
list are recorded from second-century sites elsewhere,
while there is but one that is definitely early. The
evidence of shape and decoration is thus fully con-
firmed by the stamps.

The single stamp that can be classed as early is the
last upon the foregoing list. It was found at Castle-
cary in 1841, and its evidence must be regarded as
specially important, seeing that it confirms a surmise
suggested by certain features of the structural remains.
The fragment on which it occurs is the bottom of
what appears to have been a vessel of Dragendorff s
Form 27. This was the common cup of the first or
Agricolan period at Newstead ; its place was sub-

JThe AHIM of Wilson's Prehistoric Annals (1851), p. 402. Even the
reading given above is not quite certain.

2 The OFCVNI of Proc. Soc. of Antiq. of Scot. 1905, p. 492.

3 The IRS EC A of Wilson's Prehistoric Annals (1851), p. 402.

4 Prehistoric Annals (1851), p. 402. For PATIRATI OF cf. C.I.L.
iii. No. 6010, and also PATERATI • OF, found at Corbridge (where it is
probably second century), London, Colchester, Etaples, etc.

POTTERY AND GLASS 375

sequently taken by a shape with straighter sides.1
The texture of the fragment, too, is peculiarly hard
and close, decidedly more so than is usual in ' Samian '
of the age of the Antonines. Lastly, the name of
L. Terentius (or Tertius) Secundus, although compara-
tively rare in Britain,2 is fairly common in Germany,
where it has been found under circumstances that
justify its being assigned to the last quarter of the
first century A.D.3 The various indications of date
are thus entirely consistent one with another, and all
point to the tiny potsherd being a relic of the earliest
Roman invasion of Caledonia. They make the exist-
ence of an Agricolan fort at Castlecary something
more than a mere conjecture.

' Samian ' ware was not the only fine variety of
pottery among the ' plenishings ' of the forts. Frag-
ments of Castor ware occurred at Bar Hill.4 They are
of the usual dull slate colour, with a coppery tint,
and are ornamented with conventional foliation or
the figures of animals, laid on in ' barbotine ' with self-
coloured slip. The vessels they represent are common
on Roman sites in Britain, and also on the German
frontier. Other articles of table or toilet furniture
were even finer. Some pieces of glass from Castle-
cary, for instance, strike a distinct note of elegance.
These, however, are exceptional, and are much more
likely to have come from the Commandant's house

1 Curie, A Roman Frontier Post, p. 197.

2 An example from Oare in Kent is recorded in C.I.L. vii. No. 1336, 1113.
Cf. Walters, Brit. Mus. Cat. M. 1813.

3 So, more particularly, at Wiesbaden ; see Ritterling and Pallat in
Annalen des Vereins fur Nassauische Altertumskunde, 1898, p. 155, —
a reference I owe to Mr. James Curie.

4 See also supra, p. 180, for a piece from "near Kirkintilloch."

376 MISCELLANEOUS EVIDENCE

than from the quarters of the soldiery. Minor objects,
like spindle-whorls, can hardly be noticed separately.
But mention may be made of two fragmentary figures
of red clay, which have long been in the National
Museum, and which should not improbably be con-
nected with the Limes. There appears to be no
record of the date of their presentation or of the cir-
cumstances under which they were found. But their
provenance is stated to be West Kilpatrick,1 which pro-
bably means the Chapel Hill. Both are torsos and
both have belonged to statuettes, 8 or 10 inches high,
hollow behind, and therefore clearly intended to be
placed against a wall. They have been made in
moulds, rather larger than that discovered in 1909 at
Corbridge,2 but much the same in character, although
a good deal superior in execution and finish. One
of the figures has been a fully-draped female, seated
and grasping with her right hand a patera which rests
upon her knees ; in her left she holds an object which
is apparently a flat dish containing fruit. The other
has been a half-draped male figure, with well-developed
chest, standing, with fruit in his left hand ; Hercules
with the apples of the Hesperides is conceivable but
perhaps hardly likely. In any event one is reminded
of the "female figure, about 12 inches in height,
formed of reddish clay, said to have been discovered
in the baths at Duntocher in 1775 ; it is described as
having the appearance of a dancing figure."3 It had
been lost long before the middle of last century.

One class of finds still remains to be discussed — the

1 Catalogue of the National Museum, p. 221, FR 180-181.

2 Report on the 1909 Excavations, Fig. 6.

3 Stuart's Caled. Rom. 2nd ed. p. 303, and p. 302, footnote.

PLATK LTII

2. FRAGMENT OF SLAB, ONCE INSCRIBED

3. BUILDING STONES, FRAGMENT OF QUERN, AND BALLISTA BALL

STONES FROM CROY HILL

COINS 377

coins. The evidence they supply is, in a sense, the
most valuable of all. Many of them can be dated to
an exact year, and, as it is a reasonable supposition
that Roman money found in the forts was dropped
there during the Roman occupation, the expectation
that they can be utilized as a chronological index is
not unnatural. But there are qualifying considera-
tions. Ancient coins often remained in circulation for
a very long time after their original issue. This is
true of silver much more than of copper, and more
even than of gold. The reason is an economic one.
From the reign of Nero onwards the Roman emperors
plunged deeper and deeper into the mire of a debased
currency. Whenever it was possible, the good silver
was melted down to be reissued as part of an alloy.
But on the frontiers there were persistent survivals ;
coins of Vespasian, for instance, occur in hoards that
must have been deposited under Severus. The
superiority of the purer metal was recognized, and
it appears to have circulated at a higher value.1 A
different explanation is required to account for the
strange case of the legionary denarii of Mark Antony,
which continued to be used for 250 years after they
were first struck. Antony was far in advance of his
times, and he alloyed his silver so heavily that not till
the reign of Hadrian did the imperial denarius sink to
the same level. In the interval, therefore, it would
not have been worth the while of any government to
attempt to call in his pieces for purposes of conversion.2
Coin-finds are consequently an unreliable guide to the

1 In some cases, of course, the presence of early coins may be due to
another cause : the hoard may represent the accumulation of many years.
" See Haverfield in Archaeologia, liv. pp. 489 ff.

378 MISCELLANEOUS EVIDENCE

superior limit of occupation ; denarii of Domitian need
have nothing to do with the invasion of Agricola.
And the same holds good with regard to the inferior
limit, for newly minted money evidently took some
time to make its way to the outskirts of the empire,
silver moving more quickly than copper.1

But, even when the fullest possible allowance has
been made for the qualifications just stated, the
evidence from coins deserves to be most carefully
weighed, and to that end the first requisite is an
accurate record. The following list, though meagre,
comprises all the finds known to have been made in
or near any of the forts :

Chapel Hill — ' A number of silver coins ' found in the
baths in 1790; no details available (see supra,
p. 155). ' Several denarii of Trajan ' circa 1850
(Stuart's Caled. Rom. 2nd ed. p. 294, footnote).
Duntocher — Gold of Hadrian (Gordon, I tin. Sept.
p. 52, and ' Plate of Medals, etc.' Fig. 7).
Silver of Domitian, Trajan, and Faustina ;
and copper of Trajan and Pius (Stuart, op. cit.
p. 304 and footnote).

Bemulie — Copper of Pius (Stuart, op. cit. p. 320, foot-
note).

Gadder — Gold of Pius2 (New Stat. Account of
Scot. ' Lanarkshire,' p. 407). Copper of Pius3
(Stuart, op. cit. p. 321).

1 See Dragendorff in Banner Jahrb. cxiii. p. 240.

2 This is doubtful. It was only "supposed to be of gold," and the date
and circumstances of the find agree with the particulars given by Stuart
as to the copper of Pius.

3 This find was not, however, made near the site of the fort, but in
dredging the pond at Cadder House, some distance away. Mr. J. Bar-
tholomew of Glen Orchard has shown me a copper coin of Pius found
on his property about a mile north of Cadder.

COINS 379

Kirkintilloch — ' Coins of Domitian, Antoninus Pius,
Commodus, and Constantine,' metal not stated
(Stuart, op. cit. p. 324, footnote). Silver of
Vespasian (2), Titus (i), Domitian (4), Nerva
(i), Trajan (15), Hadrian (18), and Faustina
Junior (i), apparently a hoard, found in 1893
with an iron spear-head and a large nail (Proc.
Soc. of Antiq. of Scot. 1894, p. 276).*

Auchendavy — Gold of Trajan, found in 1771 in pit
with altars (Cough's Camden, 1806, iv. p. 98).

Bar Hill — Silver of Trajan, Hadrian, and Pius
(Stuart, op. cit. p. 338). Silver2 of Mark
Antony, Vespasian, Domitian, Nerva, Trajan,
Hadrian, Pius (?), and Marcus Aurelius ; and
copper of Trajan and Hadrian (Roman Forts
on the Bar Hill, pp. 107 ff.).

Castlecary — Silver of Hadrian and of 'Caesar
Augustus,' found in August, 1771 (Nimmo's
Hist, of Stirlingshire, ed. 1880, p. 5). Silver
of Trajan, found in 1907 in cutting a road to
Castlecary Castle, exactly opposite the gate of
the fort.3

Falkirk — Coin of Pius ; metal not stated (Nimmo's
Hist, of Stirlingshire, ed. 1880, p. 39).*

Carriden — Gold of Vespasian (Sibbald, Hist. Inq.

P- 30-

aThe list there given contains only 24 coins and does not include
either Vespasian or Faustina Junior. I have seen photographs of 18
others belonging to the same find, and these are reckoned here.

2 See, however, infra as to the metal of some of these denarii.

3 Seen at the time by myself.

4 The coin-finds from ' near Falkirk,' recorded in Stuart's Caled. Rom.
and ed. p. 267, footnote, and Proc. Soc. of Antiq. of Scotland, i. p. 59,
belong to Camelon.

380 MISCELLANEOUS EVIDENCE

The list is scanty, but its meaning is not to be
mistaken. Trajan, Hadrian, and possibly Pius, are
represented in all three metals, the copper of Pius
predominating ; Vespasian in gold and silver ; Mark
Antony, Titus, Domitian, Nerva, Faustina, and Marcus
in silver only. Commodus is included too, but in
regard to him details are awanting. 'Caesar Augustus'
is not identifiable, and Constantine, whose occurrence
is obviously accidental, may be neglected. The absence
of copper earlier than Trajan agrees with the evidence
of the pottery, as showing that the effective occupation
of the Limes did not begin until the second century.
Even the finding of stray copper pieces of the Flavian
period could easily be accounted for by what we know
as to the actions of Agricola. But no such finds
have been made ; the money lost in and about the
forts is the currency of the age of the Antonines.
That is true, not only of the terminus a quo, but also
of the terminus ad quern. In the words of Professor
Haverfield, who was the first to approach this aspect of
the question from the proper side and whose inferences
apply to the whole country north of the Tweed, " the
conclusion is irresistible that the vallum, and indeed all
Scotland, must have been lost by the Romans not
very long after it was established. ": If we set aside
Cramond, which seems to have played a part of some
importance during the expedition of Severus, Com-
modus is the latest emperor whose coins are normally
found in Scotland on sites that have been occupied by
the Romans ; and his coins occur but rarely. Probably,
therefore, it was in his reign that the ground won by
Lollius Urbicus was definitely given up. At what

1 Antonine Wall Report, p. 161.

COINS 381

period in his reign, the evidence is not sufficiently
precise to enable us to say ; the notices are too vague
to admit of the various pieces being dated. But, if it
be borne in mind that, while Commodus did not become
emperor till A.D. 180, coins bearing his 'image and
superscription ' began to be struck some years earlier,
it will be seen that there is nothing in the numismatic
data inconsistent with the supposition that the Limes
was abandoned even earlier than A.D. 185.

Apart from their importance as confirming the
general conclusion, the coins found on the Bar Hill
in 1902-3 present a peculiarity that deserves particular
notice. Thirteen of the denarii were taken from the
bottom of the well. Ten of these thirteen are made of
pure tin, and have been run in moulds, not struck.
The tin coins are quite unlike the work of ordinary
forgers, since they can never have been intended to pass
current as silver. Their light weight and the softness
of the metal — they can readily be bent with the fingers
— would have led to instant detection. Furthermore,
the fact that in one case five, and in another case three,
of the ten have been cast in the same moulds, shows
that they cannot have found their way from a distance
to North Britain in the ordinary process of trade. On
the other hand, it is in the last degree unlikely that a
forger would have selected as a convenient centre for
the exercise of his activity a small military outpost on
the very fringe of civilization. The clue seems to lie
in the character of the ' find-spot' The throwing of
money into wells or rivers from superstitious motives
is a very familiar phenomenon. The tin denarii may
have been shams expressly manufactured for devotional
purposes. This would give a fresh significance to the

382

MISCELLANEOUS EVIDENCE

prohibition in the Digest (xlviii. 10) — " Ne quis nummos
stanneos, plumbeos emere, vendere dolo malo veht."
What is there forbidden is not the manufacture of
tin coins, but their being fraudulently passed into
circulation. x

1 See Numismatic Chronicle, 1905, pp. 10 ff. for a fuller discussion of these
Bar Hill pieces. Cf. also Riv. Ital. di Numismatica, 1905, pp. 166 ff.

CHAPTER XII
CONCLUSIONS

IT remains to glance back over the mass of evidence
that has been surveyed, and to consider how far the
outlines of a coherent story are to be discerned.
Inferences have been suggested from time to time as
various striking facts emerged. If these can be
brought to a focus now, the result may be a clearer
picture of the fortunes of the Limes than it has hitherto
been possible to obtain. And, first, as to Agricola.
We shall learn presently that his troops were beyond
question the original occupants of Roman Camelon.
The last decade has also produced indisputable con-
firmation of the statement that he built a line of forts
between the Firths of Forth and Clyde. Whether
the station at Camelon had formed a link in this
chain is meantime a matter of doubt. If we set it
aside, Bar Hill, Castlecary and Rough Castle may
safely be regarded as certain. As these are the
only three sites on which anything approaching
systematic excavation has taken place, the testimony
is singularly emphatic. It is also valuable for the
light it throws on the slenderness of the garrisons.
Bar Hill can have held but a handful of men. The
size of Rough Castle is more doubtful, but the

384 CONCLUSIONS

difference between the two cannot have been great ;
the total length of the north ditch at Bar Hill was
191 feet, and the lilia at Rough Castle covered a front
of not more than 200. At Castlecary, on the assump-
tion that the indications there are accepted, the
corresponding dimension would be about 250, showing
that the strategical importance of the position was
recognized in the first century, just as it subsequently
was in the second.

If we may judge from Bar Hill, the ramparts sur-
rounding the Agricolan forts were of earth, not of turf
or stone. That is, they were constructed in accordance
with the usual practice of the period to which they
belong ; the small first century forts on the German
frontier have all got ramparts of earth, stone being
reserved for the series of comparatively large base forts
on which the advanced line rested — Hochst, Heddern-
heim, Okarben, Friedberg, Kesselstadt. There is no
doubt that Agricola's earthen forts were connected by
a road ; facility of intercommunication would be vital.
Of this road there is as yet no certain trace, although
it is by no means impossible that the curious deviation
of the Military Way immediately to the east of Castle-
cary may preserve the tradition of its original line ; it
is difficult to imagine any other reason that would
account for the failure to approach the gate directly.
But would forts and road alone suffice ? An appeal to
the analogy of Domitian's limes in Germany suggests
an emphatic negative, and the appeal would seem to
be justified, seeing that the two limites are practically
contemporaneous. Domitian's limes is only two years
later than Agricola's. The German forts are a little
larger than the Scottish ones, their average size being

PLATE LIV

I. LOOKING N.W.

2. LOOKING E.

THE TUMULUS AT GADDER

THE AGRICOLAN LIMES 385

about 250 feet square. Yet they were not regarded as
adequate in themselves. Between each fort and its
neighbours there ran a line of wooden watch towers —
burgi or praesidia — such as we saw depicted on the
columns of Trajan and Marcus (Plate III.). The need
for such precautions would not be less on the Caledonian
frontier than beyond the Rhine. It is, therefore, prima
facie unlikely that they were omitted. We cannot now
determine definitely whether the Maiden Castle and
the mound at Bonnybridge ever supported super-
structures of the kind. But the tumulus at Cadder
(Plate LIV.) would still be able to answer for itself.

Next to their small size, the most noteworthy point
about the Agricolan forts is the comparative brevity of
the period during which they were occupied. Careful
search in the ditches at Bar Hill failed to reveal the
presence of relics of any kind. Whether the potsherds
in the lilia at Rough Castle would have been datable,
we cannot tell. But neither at these forts nor — apart
from a small fragment of the ware of L. Ter. Secundus
at Castlecary — anywhere else along the line has
anything been noted which one would be disposed to
connect with a first-century occupation, the structural
remains alone excepted. This is all the more remark-
able in view of the extraordinary wealth of objects that
other first-century sites have yielded. We do not
require to go so far afield as Hofheim or even as
Newstead for an instance. Had it been possible to
have had Camelon explored with the same thorough-
ness as the fort beside the Eildons, the indications are
that the harvest might have been even richer. And
much of what was actually found there can only be
attributed to the Agricolan epoch. The shapes, the

1 B

386 CONCLUSIONS

decoration, and the potters' stamps of the ' Samian '
ware,1 for instance, show a very large admixture of
early vessels, and they may fairly be taken as an index
to the rest. The abundance of first-century relics in a
station so close at hand increases our surprise at their
absence along the line of the Limes.

No doubt the phenomenon is to some extent to be
explained by the fact that Camelon was an important
centre. The fort was very considerably larger. The
garrison would be proportionately more numerous, and
the rank of the commanding officer would be higher.
There would be a greater throng of camp-followers,
and camp-furniture would be finer and perhaps less
carefully husbanded. With it all, however, in default
of further investigation, it must remain doubtful
whether this is sufficient to account for the contrast.
It is not intrinsically probable — particularly in view of
what happened during the Antonine period, when both
Camelon and the forts on the Vallum appear to have
been strongly held — that Agricola withdrew his
garrisons from the isthmus as early as A.D. 83, when
he definitely essayed to subjugate the tribes beyond
the Forth. No doubt, like every capable general, he
must have been alive to the supreme value of con-

1 The frequent occurrence of stamps beginning with OF ( = officina) is
characteristic. These appear to be chiefly first century (Curie, A Roman
Frontier Post, p. 228). I may take this opportunity of noting a few stamps
from Camelon that do not find a place in the published list. One or two
of them are in the National Museum. The rest I saw several years ago
in the hands of a workman in the foundry that now stands on the site of
the fort. He made a hobby of excavating for himself during the meal
hours, and had accumulated a considerable number of small objects. He
has since gone abroad, and no one knows what has become of his collec-
tion. The names are— BANEOLVJCCI • M, CARATILLI, DOVIICCVS,

MEM[ORI8-M], MirERTI-OFF, OF-PAUM], RIO[--], OF-SECVN,
SEVERVS. TERTVLLI, [VJERV8-F.

THE AGRICOLAN LIMES 387

centration. But even so, it is difficult to believe that
he would deliberately abandon a base which, owing to
its being in touch with the sea, was calculated to be so
useful in any advance he might undertake. It is
incredible that he should have done it, if in his time
there was a Roman road into Caledonia on the west
side as well as on the east.

An alternative, and much narrower, hypothesis
might claim some support on military grounds.
Unlike Rough Castle, Camelon lies out on the plain,
well in front of the ridge of heights along which the
Limes of Lollius Urbicus subsequently ran. Its
strategic significance is thus offensive rather than
defensive. Quite possibly, therefore, it may not have
belonged to the original series of posts erected in
A.D. 81, but may have been built two years later,
when the new movement towards the north was
commenced. On that assumption it would be open
to us to suppose that, once a Roman force had been
planted in Camelon less than a mile and a half away,
the continued holding of Rough Castle was felt
to be unnecessary. The one station might have
succeeded the other without the rest of the Agricolan
Limes being in any way affected. As against this,
however, it must be remembered that the proximity
of the two sites is not, after all, sufficient to preclude
the idea of simultaneous occupation. As a matter of
fact, remains of the Antonine period are abundant
in both, indicating that both had harboured Roman
garrisons then, while the distance that separates them
is actually somewhat greater than that which separated
the Antonine forts at Castle Hill and New Kilpatrick.
Besides, although the theory of succession might help

388 CONCLUSIONS

with Rough Castle, it would do nothing to throw light
upon Bar Hill, where the lack of first-century objects
was even more striking, in view of the greater thorough-
ness with which the ground was turned over. On the
whole, it will be wise to wait for a solution until fresh
material for forming a judgment is made available.

Meanwhile the end of the story of the Agricolan
Limes can be read as in an open book. It was
speedily forsaken by those who had brought it into
being, and its strongholds became a desolation. The
lilia at Rough Castle appear to have been deliberately
filled. The ditches at Bar Hill were left to silt up
gradually, while whin and hazel took root and
flourished on their sides. Two generations passed,
and then the Roman eagles once more flashed in the
sunlight on the heights above the Forth and Clyde.
We may be sure that a stout resistance was offered
to Lollius Urbicus by the children's children of the
men who had fought against Agricola. But the dis-
ciplined legionaries would accept of no denial. When
their work in the field was done, they were set the
task of preparing a new frontier-line for the auxiliaries
to guard. They seem to have followed closely in the
footsteps of their predecessors of sixty years before.
Their method, however, was different. Their forts
were very much larger, besides being constructed in
a more substantial fashion ; and burgi or praesidia
were apparently dispensed with, probably because they
were rendered unnecessary by the greater size of the
forts and their comparative nearness to each other.
Lastly, a wall of sods, with a ditch in front of it,
was drawn from end to end of the isthmus. These
changes are characteristic of the period to which the

THE ANTONINE LIMES 389

new frontier-line belonged. Under the influence of
Hadrian, limites had changed their character, and had
assumed the appearance of barriers, a development for
which peculiarly favourable opportunities were pre-
sented in Britain, where the distance from sea to sea
was short.

Not unnaturally, it is the Rampart and its Ditch
that have left the deepest impression on the popular
imagination. Doubtless they were deliberately cal-
culated to produce such an effect on the minds of
those who watched them in progress or who looked
on them when complete. But one result has been to
obscure their real meaning for subsequent generations.
It has usually been supposed that the Antonine Vallum
belonged to the class of fortifications known as ' con-
tinuous lines,' and that the Romans held it in the same
fashion as Wellington held the lines of Torres Vedras.
Colonel Ruck, for instance, in a very detailed study of
the subject, says : " We look upon the Antonine lines
in the same light as those of the lines constructed in
1877 against the Russians by the Turks to cover
Constantinople at Gallipoli as a model of economical
construction for the purpose required." * His reckoning
is based on the formula of Sir John Jones, Wellington's
engineer, and he rightly assumes that Sir John Jones's
principles are, in their essence, as applicable to ancient
as to modern warfare. But the figures only require to
be mentioned to betray that there is an underlying
fallacy.

Under ordinary conditions Colonel Ruck calculates
that 105,600 men would be the minimum for "the
whole field army to garrison and construct the

1 Trans, of the Glasgow Archaeol. Society (N.S.), iv. p. 472.

39o CONCLUSIONS

Antonine." After making allowance, however, for
the Roman superiority in weapons and postulating the
existence of a mobile force of considerable size in the
shape of a field army, he is able to reduce his estimate
and to conclude "that against an enemy such as the
Romans would have had to keep at bay a body of
50,000 would have had to be provided for the Antonine,
or at the rate of 5 men for every seven yards of lineal
parapet for the whole lines."1 The very cogency of
Colonel Ruck's military reasoning demonstrates the
unsoundness of the historical foundation on which it
rests. At no period in the history of Britain did the
strength of the Roman garrison remotely approach a
total which would have enabled it to comply with this
demand. The view that the fortifications between the
Forth and Clyde were of the nature of 'continuous lines'
is inconsistent with the teaching of military science. It
must be frankly abandoned. The underlying fallacy is
a confusion between the strategy of a campaign and
the requirements of a permanent occupation.2

That has been made clear from another side by
General v. Sarwey, Military Director of the German
Limes-Commission. He has pointed out that, if war
on a grand scale had been contemplated, no proceeding
could have been more senseless than to waste so large
a proportion of the available troops by spreading them
out over so thin a line.3 The notion that a limes was
intended to bid defiance to an invading army must, he

1 Ibid. pp. 458 f.

2 The confusion appears in many forms in discussions on Roman
Britain. See the repeated protests of Haverfield in Viet. Count. Hist.
'Derbyshire,' p. 211, footnote; Roman Wales, p. 51, footnote; etc.

3 ' Die Abgrenzung des Romerreiches ' in Westd. Zeitschr. xiii. especially
pp. 19 ff.

THE ANTONINE LIMES 391

declares, be rejected uncompromisingly ; a limes with all
its attendant organization was emphatically an arrange-
ment that contemplated peace, not war. Von Sarwey's
position, as here stated, is quite in accord with what
we gathered in chapter iii. as to the principles by
which the Roman system of frontier defence was
governed. The limes imperil was not a mere military
engine. Military characteristics it certainly had ; it
was manned from end to end by soldiers. But in the
nature of things its real function was, to a large extent,
political and administrative. It is true that the peace
which it contemplated was a restless peace at best.
Still, it was peace ; and peace means permanence.

Even a layman may find ample confirmation of this
view in the ruinous remains of the Scottish Wall.
What organized army, ancient or modern, has taken
its women and children to the front in a campaign ?
What series of ' continuous lines ' — in the sense now
given to that term by military engineers — has ever
had a feature corresponding to the set of outposts
that ran from Camelon to the gates of the Scottish
Highlands? No doubt the district was what might
be technically called disturbed. The forts of the
Limes and their garrisons were designed to overawe
the tribesmen to the south as well as to the north ;
and in this connection it is worth noting that the
southern defences of the forts are sometimes more
formidable than the northern.1 But no Roman gover-
nor could have supposed that they would be able
to stem the tide of invasion if the Caledonians
rose en masse. A recognition of this removes many
difficulties of a subsidiary kind.

1 Bar Hill is a case in point ; see supra, p. 192.

392 CONCLUSIONS

Now that we know what the Limes was not, we
may profitably go a step further and ask what it was.
It will be recalled that, in describing it, attention was
drawn to the singular skill with which the line it
follows has been selected.1 Almost everywhere it com-
pletely commands the country lying directly in front of
it, while — contrary to the usual Roman practice — the
forts are often planted on conspicuous hill-tops. This
points to one of its principal purposes ; it constituted a
chain of fortified posts of observation, in communication
by fire-signal with one another as well as with the chain
that ran towards the north and more than one similar
chain that stretched southwards. Prompt warning
would often enable trouble to be nipped in the bud or
raiders to be intercepted. And it had a very special
value in relation to the forts that lay behind it and
ahead. Its extremities rested on the sea, where the
Roman fleet was supreme. Provisions and reinforce-
ments could be comfortably landed on the shores of
the Forth at times when it might be dangerous to
bring them overland. Whether the Clyde was also
utilized, is more open to question ; the voyage round
the Mull of Galloway may have acted as a deterrent.

Further, so far as the natives of Caledonia were
concerned, the Ditch was an unmistakable warning
of ' thus far and no further ' to all unauthorized persons;
it indicated the line where Roman territory began. And
it was at the same time a practical obstacle of the most
effectual kind ; it secured that wheeled traffic should
pass the frontier only by recognized roads. The
Romans could thus control the customs, and regulate
the comings and goings of all strangers. The Rampart,

1 See supra, pp. 94 f.

THE ANTONINE LIMES 393

again, must have simplified the task of the frontier-
police. It protected the Military Way, and the supplies
that were constantly passing along it, from the risk of
attack at the hands of isolated marauders. Organized
bands of dacoits would be hunted down by the garrisons.
If any of the latter were hard pressed, they could take
shelter behind their ramparts and summon help from
their neighbours by preconcerted signals.

The question may be asked, ' Would the Romans
really have thought it worth while to erect so elaborate
a structure as the Antonine Vallum merely in order
to check possible aggression by irresponsible frontier
raiders ? ' The best reply is to be found in the inscrip-
tions which record that in A.D. 185 the emperor
Commodus constructed a great series of burgi and
praesidia along the bank of the Middle Danube at
all the points where there was a risk of "petty
raiders (latrunculi) " attempting to effect a crossing.1
A similar inscription has recently been published
relating to measures of the same kind taken in Arabia
by Valerian and Gallienus.2 It must be remembered
that it was not in the raids themselves that the
greatest danger lay. The country on both sides of
the Limes was like a powder-magazine. If sparks
were not instantly extinguished, a great explosion
might follow.

1IMP • CAES • M • [AVR • COMMODVS • AN]TONINVS • AVG- PIV8 .
8AR[MAT • ]GERM • BRIT- PONT- MAX- TRIB • POT • X • [IMP. VII .]
COS • iTTl • P • P • RIPAM • OMNEM • BV[RGI8] • A • SOLO • EXTRVC-
TI8 • ITEM • PRAES[I]DIS • PER • LOCA . OPPORTVNA • AD • CLAN-
DESTINO3 • LATRVNCVLORVM -TRANSITV8 • OPPOSITIS • MVNI-

VIT • PER (C.I.L. iii. No. 3385). Cf. C.I.L. iii. Nos. 10312 f., which,

however, are more fragmentary.

2E. Littmann and W. K. Prentice, Greek and Latin Inscriptions in
Syria (Leyden, 1908).

394 CONCLUSIONS

The frontier forts are thus readily accounted for.
The policy inaugurated by Hadrian made the Wall
and Ditch their natural accompaniment. The sug-
gestion that these were over-elaborate can most fitly
be met by the analogy of the ' Customs Hedge ' in
India, aptly cited in this connection by the late Pro-
fessor Pelham : " For the purpose of realizing the duty
on salt produced in Native States and in British
districts subject to a lower rate of duty, when imported
into Upper India, a customs line was commenced in
1843. In 1870 it stretched across the whole of India,
from a point north of Attock, on the Indus, to the
Mahanadi, on the border of Madras, a distance of
2,500 miles. It consisted of an impenetrable hedge
of thorny bushes and trees, supplemented in places
by a stone wall or a ditch and earth mound. It was
guarded and patrolled, night and day, by a force of
14,000 officers and men."1

That the prevention of smuggling was among the
objects served by the Scottish Limes has already been
suggested. Where the road passed out to Camelon
and the North, there was a guardhouse. The only
other exits were through the forts. That it would be
effective for the purpose is tolerably clear. It would
also be to the barbarians an ever-present symbol of
the power and majesty of Rome. When it was still
entire and occupied by its garrisons, it must have had
an imposing appearance. Its position would make it
readily visible for a long distance to all approaching
from the side of Caledonia. The Ditch, cut as it is
a little way beneath the crest, would be specially con-
spicuous. But the eye would soon catch the Rampart

1 Trans, of the Royal Hist. Soc. (N.S.), xx. p. 38.

THE ANTONINE LIMES 395

rising up behind, and would travel along the line,
picking up fort after fort, each girt by ramparts and
towers of its own, and each having in the centre a
knot of red-roofed buildings. By day the figures of
sentries would here and there be discernible against
the sky. In the long dark nights the blaze of waving
torches and of beacons would tell of rapid messages
running in this direction or in that. Inside the forts
themselves, life would be too busy to leave room for
weariness. The watchers had more to do than merely
to gaze across the swamps of the isthmus towards the
hills of "ancient Caledon." Drill and routine of all
kinds would have to be attended to regularly. Work
would be necessary to keep the fort and its buildings
in good repair, while even in times of peace exciting
incidents would be plentiful enough. Lastly, each
station was a society in itself. It had its trading
settlement just under its walls. Probably it had also
its colony of time-expired men, who were reluctant to
lose touch with the regiment and preferred to end
their days beneath the shadow of the ramparts.
There were certainly children to brighten the scanty
hours of leisure with their laughter.

The date of the construction of the Antonine Limes
is known to within a year or two ; Lollius Urbicus
was governor about A.D. 140-143. The date of its
abandonment can also be approximately determined
from the coins ; it is not likely to have been occupied
long after the beginning of the reign of Commodus.
In the forty odd years that intervened, it had obviously
suffered many vicissitudes. The evidence for this is
worth recalling now. The Military Way had been
radically altered at Rough Castle, and perhaps to the

396 CONCLUSIONS

west of New Kilpatrick too. The Ditch had been
partly filled up where the roadway issuing from the
north gate of the former fort runs across it. The
semi-circular platforms on Croy Hill had been added
to the Rampart at some time after it was finished.
These, however, seem comparatively trifling when set
beside the changes in the forts. The administrative
buildings at Rough Castle had been razed to the
ground, and reared again almost, but not quite, from
the same foundations. Similar destruction and restora-
tion had taken place at the Baths, but there even the
restored portions showed two distinct styles of building
which might fairly be denominated 'early' and 'late,'
seeing that one was very notably inferior to the other
(Plate LV. i). In like manner the ditches of the
annexe told plainly of two, if not of three, distinct
periods. And at the beginning of one of these the
ramparts of the fort had been greatly strengthened.
The same phenomena were noted at Castlecary.
There was, to begin with, the possibility that the wall
of stone did not represent the original Antonine
defence. The rebuilding, if rebuilding there was,
would be contemporaneous with the widening of the
Rough Castle ramparts. Then the complications of
the drainage system, the crossing and recrossing of
sewers at different levels, pointed to a wholesale recon-
struction, which had its counterpart outside in the
laying of a roadway across what had been an open
ditch. The testimony of the walls of the interior
buildings seemed to indicate change of a less drastic
sort — hasty repair and alteration by hands less skilled
or less careful of good workmanship (Plate LV. 2). To
sum up, Castlecary and Rough Castle yielded certain

PLATK LV

I. IN THI-: HATHS AT K()U(;H CASTLE

2. IN TIIK STOREHOUSE AT CASTLKCARY

EVIDENCES OF RECONSTRUCTION

EVIDENCES OF RECONSTRUCTION 397

evidence of complete reconstruction at some time in the
course of the Antonine period, and probable evidence
of subsequent and more hurried renovation before the
final abandonment. The complete reconstruction has
its parallel in other Scottish forts. It has been con-
clusively demonstrated both for Newstead and for
Birrens. In regard to the latter, for instance, Mr.
Barbour, whose admirable plan of the station marked a
new epoch in the study of Roman remains in Scotland,
says : " The walls . . . belong to two distinct periods.
Evidently the original buildings had been destroyed
and razed. ' There shall not be one stone left upon
another that shall not be thrown down ' represents
something like what appears to have happened over at
least a great part of the area ; and the place continued
waste for a lengthened interval, until the earth accumu-
lated and covered out of sight the underground
footings, which escaped. When occupation again took
place, the buildings were reared of new."1

Rough Castle made another important contribution
to the scattered fragments of evidence. The deep- worn
ruts at the south entrance proved that the duration of
the first period in the life of the Antonine fort had
been relatively short. Even the inscriptions contained
some hint of change. The legionary tablets and

1 Proc. Soc. Antiq. of Scot. 1896, pp. 113 f. For Newstead, see Curie,
A Roman Frontier Post, pp. 77 ff. Mr. Barbour writes to me (1910)
that the above passage accurately reproduces the impression left upon
his mind as to the time that must have elapsed between the two
occupations of Birrens ; he would, however, be prepared to modify his
view, if other evidence suggesting a briefer interval were forthcoming.
His opinion was mainly based upon the fact that the footings of
primary and secondary walls often ran close together without actually
coinciding. But this is precisely what happened at Rough Castle, where
a lengthened interval is quite out of the question.

398 CONCLUSIONS

dedications told of a campaign other than that in
which the 'opus valli* was carried through, one in
which the Second Legion had not participated at its
full strength, but had been represented by a vexillation.
Again, in the case of three out of the five forts whose
garrisons are known, there are fairly clear signs of two
distinct regiments being concerned — the Hamii and the
Baetasii at Bar Hill, the Tungrians and the Vardulli
at Castlecary, the Thracians and the Tungrian cavalry
at Mumrills. Such wholesale movement of troops
accords well with the theory of a temporary loss of
the Limes and its subsequent recovery. These being
the inferences suggested by the archaeological data, is
it possible to effect any adjustment with the scanty
historical facts that have come down to us ? The
archaeological investigation being incomplete, the
adjustment can only be tentative. But the attempt
is at least worth making.

Glancing back to chapter i. we may recall two
occasions between A.D. 142 and A.D. 180, when there
was apparently grave trouble on the Caledonian
frontier — once in the reign of Pius, when Julius Verus
was governor of Britain, and again in the reign of
Marcus, when " Calpurnius Agricola was sent against
the Britons." On the first of these occasions (circa
A.D. 155-160) the disturbance was clearly most
serious. The Brigantes were in revolt, and the
legions had to be reinforced by drafts of men both
from Upper and from Lower Germany. That the
struggle was prolonged is suggested by the circum-
stance that, while coins celebrating the fresh sub-
jugation of Britain were minted in A.D. 155, it was
not until three years later that the Second Cohort of

HISTORICAL 399

Tungrians built or rebuilt Birrens. The dedicatory
slab discovered there is dated A.D. I58.1

That this slab had reference, not to the building,
but to the rebuilding, seems tolerably plain. It is very
unlikely that the original Principia should have almost
or altogether escaped the destruction that overtook
the rest of the fort. Yet this is the conclusion that
inevitably follows if we accept the other alternative.
The circumstances under which the fragments were
found show that the slab was still in position when
the time came for the last retreat of the garrison ;
although some of them were in the well, others were
scattered about the courtyard.2 And is it not con-
ceivable that we may have in the name of the
builders a clue to the "inferior workmanship" which
Mr. Barbour notes as characteristic of the second
occupation ? The original fort may have been erected
by the Sixth Legion, whose name was found on a
block of stone recovered during the excavations : *
the legionaries were notoriously more highly trained
craftsmen than the auxiliaries. As to the date of
its erection, history gives no light whatever ; but it
is extremely significant that the whole of the large
collection of pottery from the site appears to be of
the Antonine age and that the same is true of the
inscribed stones.

Whether or not the original fort of Birrens was con-
temporary with the Limes of Lollius Urbicus and its
stations, then, we cannot definitely say; geographically,
it belongs as much to Hadrian's Wall as to Scotland, so

1 In Proc. Soc. Antiq. of Scot. 1896, p. 128, it is given as 153 ; but this
was subsequently corrected (Macdonald and Barbour, Birrens, p. 65).
a Proc. Soc. Antiq. of Scot. 1896, p. 129. 3 Ibid. p. 127.

400 CONCLUSIONS

that there are several possibilities. But there is little
doubt as to the great reconstruction at Rough Castle,
and presumably therefore at Castlecary, having
coincided with its rebuilding in A.D. 158. The strong
room in the floor of the sacellum at Rough Castle will
not have been forgotten. As was remarked in
describing it, that is a feature which emerged in the
second half of the second century. It occurs at
Birrens, where the strong room was " a pit 5 feet
deep, approached by descending steps. The walls are
formed of large flag stones set on end, and remains
seem to indicate that a parapet, finished with a
moulded cope clamped with a continuous bar of iron,
rose above the floor."1 It occurs again at Newstead
under circumstances of the greatest significance. Mr.
Curie has been able to show that no such provision
existed in the sacellum of the fort built there by
Lollius Urbicus, but that, when this fort was recon-
structed after a temporary abandonment, and the
Principia put into repair again, a strong-room was
inserted in the floor of the central chamber.2

Everything thus points to the Limes and the country
to the south of it having been evacuated by the
Romans during the great rising of the Brigantes about
A.D. 155, and to their having been again occupied in
or after A.D. 158. In the interval some at least of
the buildings had been razed to their foundations—
partly, it may well be, by the natives in the first
flush of their triumph, but partly also, we may
suppose, as a preliminary to the restoration. When
they were re-erected, the fashion of the strong-room

lProc. Soc. Antiq, of Scot. 1896, p. in.
2 A Roman Frontier Post, pp. 50 and 56.

- HISTORICAL 401

was adopted. The fortifications of Rough Castle were
strengthened. Perhaps it was now that Castlecary
was walled with stone. So far, the way is fairly
clear. The light grows dimmer as we approach the
next set of changes. They are noted only in the
Limes forts, and what they suggest is hurried repair
rather than entire rebuilding. The destruction that
preceded had been nothing like so complete. Indeed,
but for the fact that the phenomena were identical
at Rough Castle and at Castlecary, one would hardly
have felt justified in suggesting that they were
associated with any break in the occupation. At Bar
Hill the walls were unfortunately too much destroyed
for the phenomena to have been detected, assuming
that they were present. In any of the forts that may
be excavated in future, this type of evidence should be
carefully looked for. If its trustworthiness is estab-
lished, it will probably be safe to connect it with the
operations of Calpurnius Agricola or with a later rising
in the reign of Marcus.

As for the final abandonment of the Limes, and of
the whole country north of the Cheviots and the Solway,
the obvious conclusion is, as already stated, that this
took place early in the reign of Commodus. Perhaps
the veteran warrior Ulpius Marcellus, of whose char-
acter Dio draws a vivid picture, was the general to
whose lot it fell to bring the troops away. If so, it
was not his own fighting qualities that were at fault.
The pressure on the frontier of the Danube was grow-
ing so strong that the most rigid economy of military
power in Britain was imperative. If the current view
as to the burial of the distance-slabs by the garrisons
were certain, we might conclude that the retreat had

2 C

402 CONCLUSIONS

been orderly and deliberate, and that it had been
carried out under a deep sense of its finality. But,
in the present state of the evidence, such an inference
would be fanciful. Bar Hill and Castlecary, however,
give us a lurid glimpse of the last scene of all. It
would seem that there was a great conflagration on
the retirement of the garrisons, and that extensive
stores were burnt to prevent them falling into native
hands. The buildings were wrecked, possibly by
the Romans themselves, while the altars and slabs
were thrust away into pits and wells. When all
was over, the Caledonians appear to have shown little
disposition to try and obliterate the traces that were
left. They preferred to leave these to crumble into
ruin. Had they been determined on destruction, they
could easily have made it so complete that, long before
seventeen centuries had elapsed, not a vestige would
have been left above the ground. As a matter of fact,
the ravages of the last two hundred years have pro-
bably been more severe and more systematic than any
that took place on the departure of the Romans.

To-day two things are needed — an awakening of the
public conscience to the fact that an interesting national
monument is in danger, at some points in grave danger,
of being entirely swept away, and the organization, ere
it is too late, of a proper examination of the sites that
still admit of search. A thorough investigation of the
Limes would go far towards enabling the complete
story of the Roman occupation of the country to be
fully recovered. The task is a much simpler one than
that with which German scholars are grappling
successfully on the frontier of the Rhine. The area
involved is very much smaller, while decades take the

HISTORICAL 403

place of centuries in the term of occupation. Nor are
the difficulties anything like so serious as those that
English antiquaries have to face in Northumberland
and Cumberland. The remains are free from the
puzzling complexity of structure that is so marked a
feature of the Tyne and Solway barrier. Again, the
circumstance that the line of the Forth and Clyde was
held for comparatively brief periods, the limits of which
we now know to within a very few years at either end,
would markedly enhance the general value of the
evidence that might be expected to emerge. Lastly,
enough has been ascertained by excavation to provide
ample guidance for future explorers. There is thus a
great opportunity. It will not be creditable to
Scotland if the work remains undone.

INDEX

Abbotsford Collection, 1 79.

Abercorn (Aebbercurnig), 34.

Aeschylus cited, 267.

Agitius, 29 f.

Agricola, I f., 321 ; campaigns of,
if., 386 ff. ; Limes of, if., 26,
256, 383 ff. ; at Bar Hill, 188 f.,
383 ff. ; at Castlecary, 21 1, 2 I 3,
218, 374 f., 383 ff. ; at Rough
Castle, 233 ff, 383 ff.

Ala I Tungrorum, 55, 348 f, 354f.

Alae, organization and names of,
54 ff.

Alcluith. See Dumbarton.

Allectus, 21.

Altars, 328 ff.

Ammianus Marcellinus cited, 22 f.

Amphorae, 182, 369 f.

Anderson, Prof. John, 170, 173,
176, 185 f., 207, 287 ; his MS.,
170.

Annexes, 82 f., 366 ; at Bar Hill,
1 99 ; at Castlecary, 215; at
Rough Castle, 221, 23of.

Antiquaries of Scotland, Society of,
208, 220, 256.

Antonine Itinerary, 2 5 f.

Antonine Wall Report. See Glasgow
Archaeological Society.

Antoninus Pius, 7 f.

Antonius Saturninus, 319.

Antony, Mark, denarii of, 377.

Apollo, 33 3 f., 341.

Arabia, Limes in, 393.

Army, Roman, 43 ff, 60.

Arniebog, 128, 306.

Arnot Hill, I34f.

Arrian cited, 83 f.

Arrius Domitianus, C., 335.

Arthur's O'on, 36 f., 88.

Attacotti, 22.

Auchendavy, 122, l82ff. ; coin
from, 379 ; inscriptions from,
316, 33 iff; sculpture from,

358.

Auchenvole, 124, 340.

Augustus, army organization of,
43 ff. ; frontier policy of, 64 f.

Aurelius, Marcus. See Marcus.

Aurelius Victor cited, 20.

Auxiliary forces, organization of,
52 ff. ; composition of, 58 ; ser-
vice in, 58 f. ; Roman attitude
towards, 59 ; number of, 59 f. ;
separated from legionaries, 63 ;
as frontier police, 67 f., 71,
353ff-> 393; tablets erected by,
322 ff.

Avon, River, 141 f., 241.

Baetasii, 54, 323, 337 f.

ISallistae, bullets for, 184, 191,

363 f-

Balmuildy. See Bemulie.

Balneum. See Bath.

Bantaskine, 133 f., 237.

Bar Hill, 101, 123 f., 187 ff,
354 f. ; Agricolan fort at, i88ff,
383 ff. ; Antonine fort at, 190 ff,
abandonment of, 402 ; coins

INDEX

405

from, 379, 38 if.; inscriptions
from, 306, 316, 322 £, 337ff,
351 f. ; sculptures from, 359.

Barbour, Mr. James, 397, 399.

Barracks, 81 ; at New Kilpatrick,
164 ; at Bar Hill, 198 f.

Barrels, 369.

Batavian auxiliaries, 53, 328.

Bates, Mr. C. J., 153, 270 f., 300,

303, 3°8.

Baths, 8 1 ff., 343 ; at Chapel Hill,
154 f.; at Duntocher, 157 ff. ;
at Bar Hill, I97f. ; at Castle-
cary, 214; at Rough Castle,
223, 227.

Beacons, 264 ff.

Beads, 200, 366.

Beancross, 139.

Bearsden, 1 1 5 f., 163.

Bede, 3 if.; cited, 3 iff., 109,
147.

Bedriacum, battle of, 62, 68.

Begleithtigel, 25 5 f.

Bemulie, H7f. ; fort at, 165 ft".,
295, 299 f. ; inscriptions from,
312 ff. ; coin from, 378.

Berm, io6f.

Bertram's De Situ, 24.

Birrens, 226, 397, 399 ; inscrip-
tions from, gf., 338, 349, 399.

Blaeu's Atlas, 85, 203.

Bodotria, i f.

Boece, Hector, 38 f.

Bogton, 119.

Bo'ness, I48ff.

Bonnybridge Canal Station, 130,
218 ; mound near, 253ff.

Bonnyside, 131 f., 258 ff.

Booneck, mound at, 254 f.

Borghesi, 3.

Borrowstounness, I48ff.

Bottles, 369.

Bowshot, length of a, 147.

Braidfield, 279.

Bridgeness, isof., 243, 311 ; in-
scription from, 148, 3048".

Brigantes, 3, 6f., 9 f., 352 f., 398,
400.

Britannia, goddess, 329, 334.
Britons serving abroad, 56 f, 344.
Broaching on stones, 174, 178,

204, 240.

Bruce, Mr. John, I54f.
Buchanan, David, 86.
Buchanan, George, cited, 33, 39 ff.,

no, 300 f.

Buchanan, Mr. G. A., 182.
Buchanan, Dr. John, 92, 1 18, 141,

143, 167, 171, 173, 176 ff., 186,

281 £,284, 299, 303, 3 1 3,360 f.
Buchanan, Mr. M., 248 ff. See

also Plates XXVII and XXXIII.
Bung, 369.

Vurgi, 71, 263 ff., 385, 389, 393.
Burial of slabs, supposed, 282 f,

304, 401 f.

Cadder House, 119; inscriptions

at, 293 ff., 3l2f., 330 f.; mound

near, 25 iff., 256, 385; coin

from, 379.

Cadder, Easter, 119, 168.
Cadder, Kirktown of, 1 19 ; fort at,

i68ff.

Cadell, Mr. Henry, 304.
Cadger Brae, 140.
Cairpentaloch, 36, 181.
Calder. See Cadder.
Caledonians, 14 ff.
Calgacus, 56, 58.
Caligula, 65.
Callendar Park, I36f.
Calpurnius Agricola, 1 1, 13, 398,

401.
Camden, 296, 330; his Britannia

cited, 293, 300. See also Gibson,

Gough.
Camelon, 33, 133, 201, 247, 250,

3«, 379» 383 ff.

Campestres, 329 f., 332 f., 336.
Camps, Roman, 72 ff.
Camulus, 340.
Capitals, from Bar Hill, 196^,

358.
Capitolinus, Julius, 3 ; cited, 7,

ii, 57» 313.

406

INDEX

Caracalla, 1 8.

Carausius, 21, 36.

Car do maximus, 77.

Caristanius Justianus, 339.

Carleith, 112 f., 278.

Carriden, 148, 150 f., 243 ff. ; in-
scription from, 243, 316 ; coin
from, 379.

Carron Company, 341.

Carutius, 36.

Cassiodorus, 20.

Cassius, Dio. See Dio Cassius.

Castella, 62, 78 ff.

Castle Hill, 114; fort at, i6of.,
257 f., 387 ; inscriptions from,
282 ff., 288 ff., 295, 329 f., 354 ;
sculpture from, 361.

Castle Hill (Bar Hill), 1 2 3 f., 1 87 f.,
199.

Castlecary, brickworks at, 128 ;
railway station at, 1 29 ; Agri-
colan fort at, 211, 218, 374 f.,
383 ff. ; Antonine fort at, 117,
129, 191, 207 ff. ; strategic im-
portance of, 2 1 6 ; successive occu-
pations of, 216 ff., 396 ff. ; aban-
donmentof,4O2 ; debris of, 356 f.;
inscriptions from, 315 f., 325 ff.,
342 ff, 354 f. ; sculptures from,
359, 361 ; coins from, 379.

Castlehill Park, 124.

Castle-towry, 240.

Castor ware, 375.

Castra, 62, 72 ff.

Castrametation, systems of, 73 ff.

Cavalry, legionary, 46 ; auxiliary,
54 f. ; in garrison of Wall, 333,

348 f-, 355- .
Cenail. See Kinneil,
Centurial stones, 3156
Centuries, legionary, 45.
Centurions, 48 f., 57 ; seconding

of, 49» 337,347-

Chalmers, G., 91.

Chapel Hill (Old Kilpatrick),
1 1 o f. ; fort at, 1 54 f. ; in-
scriptions from, 276 ; coins
from, 378.

Chapel Hill (Bonnybridge), 219,

253 f-» 256.
Cheese-making, 369.
Cichorius, C., 264.
Citizenship, granted to auxiliaries,

54.571

Claudian cited, 23, 80.

Claudius, invasion of Britain by,
65,318.

Clavlcula, 75 f.

Cleddans, as place-name, 114.

Cleddans (near Hutcheson Hill),
Il4f., 284 f.

Cleddans (near Kirkintilloch), 121.

Cleedins (near Falkirk), 136.

Clerk, Sir John, 206, 302, 328.

Clodius Albinus, 14.

Clota, i f.

Cocceius Firmus, M., 331 ff.

Cochno House, 277 ff., 297.

Cohorts, legionary, 45.

Cohorts, auxiliary, 53 ff. ; Cohort
I Baetasiorum C.R., 53, 322 f.,
337 £> 354 ? Cohort I Brittonum,
67 ; Cohort II 1 1 Gallorum eyuit.,
53, 329 f., 354 ; Cohort I Ham-
iorum, 53, 339, 352, 354; Cohort
VI Nfrviorum, 53, 224, 324 f.,
347, 354 ; Cohort II Thracum
eqmt.t 5 3, 5 8, 3 5 2 f., 3 54 ; Cohors
I Tungrorum mi/., 53, 326 f.,
354 ; Cohort I Fida Vardullorum,
C.R., 53 f., 345 f., 351 ; Cohortet
Britannicae, 56; Cohortes Brit-
lonum, 56, 344.

Coins, list of 378 ff. ; deductions
from, 380 f. ; relating to Britain,
5ff., 10, 13, 19.

Coin-finds, interpretation of, 377 f.

Commandant, house of, 80 f. ; at
Castlecary, 2 1 3 ; at Rough Castle,
227.^

Commissariat, 200 f., 368 f.

Commodus, 12 f., 380 f., 393, 401.

Constantine, 22.

Constantine the Tyrant, 24 f.

Constantius Chlorus, 21.

Contubernium, 81.

INDEX

407

Cooking, vessels for, 371.

Cornelius Fronto cited, 6.

Cotton MS., 296.

Cramond, 148, 244, 327, 380.

Cramond Island, 33.

Crow Hill, 117.

Croy Hill, 94, 101, 107, 125 f. ;
platforms on, 125, 258, 260 ff.
266 ; fort at, 203 ff. ; inscriptions
from, 306, 314 f, 341 f. ; sculp-
tures from, 359 f.

Croy Hill Houses, 126.

Croy House, Nether, sculptures at,
306, 359 f.

Culverts, 95 f., 116, 124.

Curie, Mr. J., 371 ff., 400.

Dalmuir Burn. See Duntocher

Burn.
Danube, frontier on the, 70 f.,

264 f., 393.

Decumanus maximus, 76 f.
Decurio, 54, 57.
Deportatiojuvenum, 56, 67 f.
Diana, 33 3 f.
Dicalidonae, 22.
Dick's House, 218 f.
Digest cited, 382.
Dii mllitares, 328 f., 331 ff, 336,

338.
Dio Cassius, 3 ff, 14 f.; cited, i,

1 1 ff, 1 7 f.

Distances between forts, 246.
Distance-slabs, 268 ff See also P

and MP.
Ditch of the Limes, shape of, 99 ff ;

dimensions of, 101 f. ; remains

of, lozf. ; breaks in, 123, 126,

203.

Ditches round forts, 74.
Divixtus, 373.

Domaszewski, A. von, 80, 265.
Domitian, 2, 47, 56, 63, 67 f., 319.
Dougalston, 289 ff
Dullatur, East, i26f., 258.
Dumbarton, 33 f., 109 f.
Dunglass, 109.
Dunnottar, 298, 300 f.

Duntocher, 113, fort at, 113,
155 ff ; inscriptions from, 277?.,
279 f., 31 1, 328; sculpture from,
358 ; clay statuette from, 376 ;
coins from, 378.

Duntocher Burn, 113, 156, 279.

Dupficarius, 55.

East Burn (Falkirk), 136.
Eastermains, 304.
Elf Hill, 131, 260.
Enlistment of legionaries, 49 f.,
344 ; of auxiliaries, forcible,

55 ff

Epona, 332 f., 336.
Erskine House, 275, 297.
Eusebius, 28. See also Scaliger.
Eutropius cited, 20.
Expansions. See Platforms.

Fabius Liberalis, 341.
Fabretti, R., 275, 287.
Fabrica, 80, 197, 226.
Falkirk, 134 ff. ; fort at, 237 ff ;
place-name, 237 ; coin from,

379-
Ferguston Muir, 94, 1 1 6 f.

Ferrydike, I 54, 274, 307.

Fibulae, 366.

Flavius Aetius, 29 f.

Flavius Betto, 347.

Flavius Lucianus, 350.

Florus cited, 66.

Fordun, John of, 37 f., no.

Forts, list of, 246; names of, 153.

Fortune, altars to, 82, 342 f. ;

statuette of, 359.
Fossa, two kinds of, 74.
Franks, 23.

Fraser, Sir Wm., 293, 296.
Freebairn, Mr. Chas., 160.
Frontier, distribution of troops on,

62 ; defence of, 64 ff
Fronto. See Cornelius.
Fulgentius, 37.

Gale, Roger, 24, 302.
Gallowhill, 137.

40 8

INDEX

Garnhall, 128.

Garrison, size of, 354^ ; change

of, 354, 398.
Garscadden Wood, 114.
Gates, of camps, 76, 79.
Gellygaer, 226.
Genius loci, 335, 342.
Genius praetorii, 80 f., 213.
Genius Terrae Britannicae, 334 f.
Genunia, district of, 7.
Gericke (Crispinus Gericius), 41,

293-
Gibb, Mr. Alex., 272 f, 278, 287,

289, 291.

Gibbon cited, 22 f.
Gibson's Camden cited, 85 £, 174,

275,278, 298, 313.
Gil Burn, 147.
Gildas, 26ff., 33 ff.
Gilmor-Seat, 260.
Giudi, 33.
Glasgow Archaeological Society,

92 ; their Antonine Wall Report

cited, passim.
Glass, window, 240, 357 ; vessels

°f» 3.75 f-

Glenfuir House, 133.
Golden Hill, 113, 155 ff.
Gordon, Alexander, 87 ff. ; his

Itinerarium cited, passim.
Gordon of Straloch, Robt., 85.
Gough's Camden cited, 159 f., 307.
Graeme, Graham, or Gryme, 38,

40, 132.
Graham, John, of Dougalston, 289,

291.

Graham's Dyke, 38, 149.
Graham's Dyke Road, 149.
Graham's Dyke Street, 138.
Grandsable Cemetery, 139.
Grange, House of, 149, 151.
Guardhouse, 247 ft"., 394.
Guaul, 35.

Hadrian, 3, 5 ff. ; introduces local
recruiting, 50 ; frontier policy
of, 70 ff., 389. See also Wall.

Hamii, 54, 339, -352, 354.

Hamilton of Barns, Jas., 277, 279.

Hamilton of Orbiston, Wm., 275.

Hamilton of Wishaw, Wm., 1 69.

Haverfield, Prof., id, 16, 21, 30,
41, 211, 232, 277, 315, 318,
321, 339 ff., 345, 347, 351,
362, 377, 380.

Hercules, 332 f. ; Magusanus, 348 f.

Herodian, 14 f. ; cited, 17, 19, 23,

.52'
Hillfoot Golf Course, 117.

Historia Augusta, 3 f. See also Capi-

tolinus, Lampridius, Spartian.
Historia Brittonum. See Nennius.
Honorius, 24 f.
Horn, objects of, 364 f.
Horreum. See Storehouse.
Horse furniture, 364.
Horsley, John, 89 ; his Britannia

Romana cited, passim.
Houston, Mr. Alex., 284.
Hubner, E., 109, 153, 270, 362 f. ;

his C.I.L. vii. cited, passim.
Hutcheson Hill, 103, 114, 281,

284f.
Hyginus, 73 f. ; cited, 74 f., 77 f.,

Imperator, 43.

Imperium, 43.

Implements, 363.

Inchbelly Bridge, 302.

Inchtuthil, 82.

Indian Customs-hedge, 394.

Infamis digitus, 359.

Innes, Thos., 24.

Intervallum, 77 f.

Inveravon, 142 f, 241 f.

Inveresk, 296.

Irvine, Dr. Christopher, 86, no,.

2i9f., 241 f.
Itinerarium Antonini Augusti, 25 f.

Jerome, 20, 28.

Jewish wars, 6 f.

Julian, 22.

Julius Caesar, 31, 37, 39, 46, 232,

235 f-

INDEX

409

Julius Marcellinus, C., 352.

Julius Verus, 9 f., 398.

Jupiter Optimus Maximus, 328,

331^336,338-
Juvenal cited, 6, 336.

Keith, George, Earl Marischal, |
300.

Kelvin, 117, 165 fF, 292, 299.

Keyth, Mr. John, 300.

Kilpatrick, New (East), 1 1 6 ; fort
at, 162 fF., 297 f., 312, 387;
cemetery at, 1 1 6 f.

Kilpatrick, Old (West), 109 f., 376 ;
inscriptions from, 274 f., 278,
307. See also Chapel Hill.

Kilpatrick Hills, 1 1 o f.

Kinneil, 34, 36, 143*1".; place-
name, 147 f. ; possible fort at,
242 f.

Kirkintilloch, 36, I2of.; place-
name, 181 ; possible fort at,
174 ff. ; sculptures from, I76f.,
182, 361 ; coins from, 379.

Kirkintilloch, Peel of, 120, I74ff.

Knox, John, cited, 157, 160.

Lampridius cited, 12.

Lamps, 367 f.

Latrines, at Bar Hill, 198 ; at
Castlecary, 214^.

Laurieston, 138.

Learmonth, Mr. A. S. R., 144^

Leather, articles of, 364 fF.

Ledcameroch, 115.

Legatus legionis, 47 f.

Legions, 44 fF. ; officers of, 46 fF. ;
mutinies among, 48, 52, 61,
319 f. ; recruiting of, 49 f., 344 ;
conditions of service in, 50 fF. ;
number of, 59 ; how distri-
buted, 6 1 f. ; separated from
auxiliaries, 63; / Adjutrix, 2 ;
// Adjutrix, 2, 45, 321 ;
IX Hispana, 3, 6, 45, 320 ;
XXII Deiotariana, 6; II Augusta
Pia Fidelis, 45, 277, 279, 289,

293» 3°5» 312 fF., 317 fF.,
33 iff, 340, 342 f., 350; XIV
Gemina Martia Victrix, 45 ; XX
Valeria Victrix, 45, 273, 275,
279 fF., 301, 303, 312, 320 f. ;
VI Victrix, 45, 276, 279, 287,

298, 302, 3Hf-> 3i9f-» 34i ff-
Levenside, 298.
Levy. See Enlistment.
Lhuyd, Edward, 219^, 275, 278,

288.

Lilia, 164, 231 fF.
Limes, meaning of, 68 ; in castra-

metation, 76 f.
Limes imperil, 69 fF., 391.
Limes, English. See Wall of

Hadrian, Wall of Severus.
Limes, German, of Domitian, 66 fF.,

3846 ; of Pius, 57.
Limes, Scottish, of Agricola, i f.,

26, 256, 383 fF.; of Pius, 7f.,

26, 93 fF., 256 fF., 380 f., 388 fF.,

. 394 «

Limes-Commission, 66, 72 f.
Livy cited, 305.
Lollius Urbicus, 7 fF., 26, 57, 313,

388, 395 ; slab with name of,

168, 313 f.

London, 23 ; Tower of, 31.
Lupicinus, 22 f.
Lust 'ratio, 305.
Lyne, 227.

MP, 269^, 299 f., 308 f.
M'Chesney, Prof., 281.
Macdonald, James, 269 ; Tituli

Hunteriani, 272 ; cited, passim.
Maeatae, i4fF., 18.
Maiden Castle, 247 fF., 385.
Maitland, Wm., 90 ; his Hist, of

Scotland cited, passim.
Mallets, iron, 185 f.
Manse Burn (Bearsden), 116.
Marcellus. See Ulpius.
Marcus Aurelius, lofF. ; column

of, 71, 264 f.
Marischal College, 301;
Marius, 43, 45.

INDEX

Marriage of legionaries, 52 ; of
auxiliaries, 58.

Mars (Ultor), 332 f., 336 ; Camu-
lus, 340.

Marshal, Earl. See Keith.

Matres, 329 f., 344.

Maximus, 28, 30.

Melandra, 211, 226.

Mercury, 343 f.

Military Way, 103 ff. ; at Gars-
cadden Wood, 114; at Bears-
den, 104, 114, 165 ; on Bar
Hill, 124, 190 ; on Croy Hill,
104, 125; at Arniebog, 128;
at Castlecary, 129, 2 1 7 f. ; at
Rough Castle, 223, 2276

Millhall Burn, 141.

Millochan, Low, 292, 295, 299.

Minerva, 332 f.

Mommsen, 4, 21, 49 f., 56 f., 60,
66, 77.

Monumenta Imperil Romanl, 171 ;
cited, 185, 287, 292.

Montrose, Marquis of, 273.

Morrison, John Millar, 154 f.,

157.

Mortaria, 160, 370.

Morton MS., 294.

Mound, the Outer, 104 ff.

Mounds. See Booneck, Cadder
House, Chapel Hill (Bonny-
bridge), Maiden Castle.

Mount Pleasant, inf., 1 20.

Mugdock House, 273, 297.

Mumrills, 138 ff., 239 ff., 354 f. ;
inscriptions from, 348 f., 352 f.

Murus, distinguished from Vallum,

32.

Murus eespitlclus, 97.
Mutinies, 48, 52, 61, 319 f.

Napier of Merchiston, 293, 296 f.,

312.

Nectovelius, 352.
Neilson, Dr. G., 32, 97, 262 f.
Nennius, 35 f., 147, 181.
Nervii, 54, 324^, 347, 354.
New Merchieston, 237.

Newstead, 78, 335, 338, 385, 397,

400.
Notitia Dignitatem cited, 59, 323,

325>.327, 33°. 353-
Novaesium, 63.
Numeri, 57.
Numeti 'Srit 'annum, 57.
Nymph, figure of, 358.
Nymphs, altar to, 341 f.

Oman, Prof., cited, 22, 25.
Opus Valll, 268, 276, 279.
Ordnance Survey Dept., 92 f., 108;

maps of, cited, passim.
Orosius, Paulus, 20, 28, 32.

P on distance-slabs, 269 ff, 286,

308 f.

Pausanias cited, 7.
Peanfahel, 34, 147.
Peel Glen, 114, 161, 257 f., 281 f.,

Pelham, Prof., 66, 394.

Pelta-shaped ornament, 309 f.

Pelvis, 370.

Penguaul, 36.

Penneltun, 34, 36, 147.

Phasis, fort on the, 83 f.

Picts, 2 if., 25, 28 f., 33,40.

Pisentius Justus, Q., 329.

Pius, Antoninus, 7 ff, 57.

Pit, stone-lined, 80, 224, 226, 400.

See also Rubbish-pit.
Platforms, semi-circular, 258ff.
Pleasance (Falkirk), 135 f., 239.
Polmont, 140.
Polmonthill, 141.
Polybius, 73 ; cited, 77, 264, 266.
Pont, Timothy, 85, no.
Porta decumana, 76.
Porta praetoria, 76, 79
Portae principals, 77.
Portland Papers, 87 ; cited, passim.
Post-holes, 164, 199, 213.
Potters' marks, 160, 238, 373 f.,

386.
Pottery, 369 ff.

INDEX

411

Praefectui a/ae, 54 ; castrorum, 47 f. ;
cohortis, 53, 346 ; legionis, 48.

Praesidia. See ' fittrgi.'

Praetorium, 76. S^ also ' Principia.'

Primi ordines, 48.

Primipilares, 49.

Primus fi/us, 49.

Prindpla, 79 f. ; at Bar Hill, 1 94 ff. ;
at Castlecary, 2 1 1 ; at Rough
Castle, 223 f. ; name of, 325.

Ptolemy, 153.

Querns, 159, 171, 204, 241, 368.

Rampart of camps, 75.

Rampart of Limes, structure of,

95 ff. ; dimensions of, 97 f.
Ravenna Geographer, cited, I52f.
Red Burn, 129, 207.
Reichel (Servatius Rihelius), 41,

293-

Rhine, frontier policy on the, 65 ff.
Ritterling, Prof. E., 2, 55, 65, 80,

3i9-
Robe, Rev. Mr., 349 f.

* Roman Road ' near Duntocher,
113; near Bearsden, 116; to
Camelon, 1 33, 247 ; in Perth-
shire, 256. See also Military
Way.

Rosehall, 136.

Rough Castle, 132, 22off. ; evi-
dences of reconstruction at, 2 2 2 f.,
227 ff., 396 ff. ; early fort at,
233 ff-i 3^3 ff- ; inscriptions
from, 323 ff., 347; garrison of,
354 f. ; sculpture from, 358.

Rowan Tree Burn, 132.

Roy, General, 90 f. ; his Military
Antiquities cited, passim.

Rubbish-pits, 1631"., 1 68, 184^,
201 f., 215,363, 368.

Ruck, Colonel, 108, 262 f., 317,
389 f-

Sacellum, 79 f. ; at Bar Hill, 195;
at Rough Castle, 224.

Salmanes, 351.

* Samian ' ware, 165, 172, 205,

216,240, 371 ff, 386.
Sandyford Burn, in f.
Sarwey, General von, 94, 390 f.
Saturninus, Antonius, 319.
Saxons, 23, 30.
Scaliger, 41, 293.
Scots, 22, 25, 28 f., 33, 40.
Scott, Sir W., 88, 179.
Sculptures, 358 ff.
Seabegs, I3of. ; possible fort at,

219.

Seals, 366.
Selden, 330.
Seneca cited, 51.
Sepulchral Banquet, 349, 361 f.
Sesquiplicarius, 55.
Severus Alexander, 4 f.
Severus, Septimius, 14, 17?., 32,

35» 37» 38°- See ali° Wal1 of

Severus.
Shirva, 122; inscriptions from,

314, 349 ff. ; sculptures from,

361 ff.

Shirva, Wester, 122.
Shoes, 200, 365.

Sibbald, Sir Robt., 86; his His-
torical Inquiries cited, passim.
Signal-stations, 265 ff., 392.
Signalling, systems of, 263 f.
Silenus, busts of, 359.
Silvanus, 331, 335, 339, 345.
Simson, Prof., 287, 292, 302.
Skene, W. F., 21.
Skinner, Rev. J., 177, 360.
Skipperton Burn, 130.
Spartian, 3 ; cited, 5 f., 19 f., 70.
Standards, shrine of. See ' Sacellum.1
Statuettes, 182, 358 f., 376.
Stewart, Dr. D. P., 182.
Stilicho, 24.
Stirlings of Keir, the, 293, 295 f.,

298.
Storehouse, 80 ; at Bar Hill, 196 f.;

at Castlecary, 211 f.; at Rough

Castle, 226.
Strong-room. See Pit, stone-lined.

412

INDEX

Stuart, Robt., 91 f. ; his Caledonia

Romana cited, passim.
Stukeley, 220, 275, 278, 288.
Suetonius cited, 37, 44, 47,
„ 318.

Summerston, 117, 287, 292.
Suovetaurilia, 305 f.

Tacitus, Agncola of, cited, i, 56 ff.,
65, 69, 190, 317, 321, 327 ;
Annals of, 46, 48, 50 f., 61,
65 £, 68, 305, 320 ; Germania
°f> 69, 333 ; Histories of, 2, 46,
57, 61 ff., 68, 324, 327.

Tanicus Verus, L., 331.

Tanner, Thos., Dr., 278, 313.

Tayavalla, 133.

Temple (hill), 117, 299.

Tentfield Plantation, 132 f.,
258 ff.

Ter. Secundus, L., 374 f., 385.

Terra Sigillata. See ' Samian '
ware.

Tertullian cited, 5 1 .

Theodosius, 23.

Thorn Farm, 115.

Tiberius, foreign policy of, 65.

Tin, coins of, 381 f.

Titulus, 75 ; at Bar Hill, 193.

Tombstones, 349 ff., 362 L

Tools, 363.

Towers on ramparts, 75, 191.

Train, Joseph, 179.

Trajan, foreign policy of, 70 ;
column of, 71, 263 f., 306,

369-
Trebius Verus, 345 f.

Tribuni militum, 47 f.

Tribunus cohortis, 53.

Tungri, 53, 326 f.

Turma, 54f.

Twechar, 123.

Tyne, inscription from the, 9.

Ulpius Marcellus, 13, 401.
Underwood House, 130.
Urry, Mr., cited, 278, 288, 290,
313, 328.

Usipi, cohort of, 56.

Valentia, Province of, 23 f.
Valentinian, 23.
Valerius, Nigrinus, 348.
Vallum, distinguished from murus,

32-
'Vallum' on Hadrian's Wall, 6,

32»37-

Vardulh, 54, 345 f., 354.
Varro cited, 55.
Vecturiones, 22.
Vegetius cited, 32, 47 f., 51 f.
Velleius cited, 68.
Verecunda, 351.
Verus, Julius. See Julius Verus.
Verus, Lucius, 6, 10 f.
Vespasian, 50, 318.
Veteranorum vexilla, 50 f.
Vexillatio, 46.
Via principalis, 77.
Via quintana, 77.
Via sagularis, 78.
Victoria, 332 f., 347.
Virgil cited, 335.
Virius Lupus, 14, 343.
Vitruvius cited, 98.

Waldie, G., cited, 136, 178.
Wall of Antoninus. See Limes,

Scottish.
Wall of Hadrian, 5 f., 8, 34, 39 f.,

102, 320.
Wall of Severus, igff., 32, 35rF.,

40.
Walls described by Gildas, 28 f. ;

by Bede, 33 f. ; by Buchanan,

39 f-
Watling Lodge, 133, 248.

Weapons, 363.

Welbeck Abbey MS. See Portland

Papers.
Well, at Bar Hill, 195 ; at Rough

Castle, 226.

West Burn (Falkirk), 135.
Westerwood, 127 ; fort at,

205 ff.

INDEX

Westquarter Burn, 139, 240.
Wheat, 207, 368, 402.
Wheels, 200, 364.
White Bridge, the, 143.
Whitehill, 121.
Whitelaw, Mr. Alex., 187.
Wilson, Sir Daniel, 374.

Wodrow, Rev. Robt., 290 ff.
Workshops. See ' Fabrica?

Xiphilinus, 5. See Dio Cassius.

Zimmer, H., 35, 181.
Zosimus cited, 25.

GLASGOW: PRINTED AT THE UNIVERSITY PRESS BY ROBERT MACLEHOSE AND co. LTD.

DA Macdonald, (Sir) George

777 The Roman wall in Scotland

.5
M32

PLEASE DO NOT REMOVE
CARDS OR SLIPS FROM THIS POCKET

UNIVERSITY OF TORONTO LIBRARY