INTRODUCTION
The Roman fortress of Carpow, Perthshire, on the southern shore of the Firth of Tay, was the king-pin of the attempt by the Roman army to reoccupy and hold parts of Scotland in the late second or early third century a.d.Footnote 1 It is the largest known permanent military site north of Hadrian's Wall, except for the Flavian legionary fortress at Inchtuthil.
Carpow has long been associated with the campaigns conducted in Scotland by the Roman emperor Septimius Severus in a.d. 208–11.Footnote 2 However, support has accrued for dating the construction of the fortress, and with it the dedicatory inscriptions above its East and South Gates, to the reign of the emperor Commodus who ruled from a.d. 180 to 192. The late John Casey argued for this from the early 1990s onwards, though his views did not appear in print until 2010.Footnote 3 Peter Warry suggested that the fortress was built early in Commodus’ reign and held for an indeterminate period.Footnote 4 R.S.O. Tomlin likewise advocated a Commodan date on the basis of the style of the sculpture and lettering on the two inscribed slabs.Footnote 5 Numismatic evidence has favoured Severan occupation with some possibility of earlier activity, either in the years immediately preceding the imperial expedition or under the emperor Commodus.Footnote 6 More recently support has again been expressed for the traditional Severan date.Footnote 7
The site was first recorded in the eighteenth century.Footnote 8 Aerial photography from the 1940s onwards revealed not only its outlines and some internal buildings, but also a temporary camp and a polygonal enclosure both seeming to predate it.Footnote 9 Excavations undertaken by R.E. Birley in 1961–62,Footnote 10 then by J.D. Leach and J.J. Wilkes in 1964–79,Footnote 11 established the area of the fortress at 11 ha (27.5 acres) within the inner ditch (fig. 1). The rampart was of turf set, in places, on a clay bottoming, and there were two ditches. Stone-built gates on the north, east and south sides were investigated in 1964–70; a fourth gate, in the west side, can be presumed. The fortress faced east. A solidly-built headquarters (principia), a bath-houseFootnote 12 and a granary were identified. Traces of timber buildings of both post-hole and foundation-trench construction were located, with wattle-and-daub walling.Footnote 13 In size Carpow falls into the category of ‘vexillation fortresses’.Footnote 14 Both the Second and Sixth Legions were involved in its construction and may well have contributed to its garrison, together perhaps with other troops (see also below).Footnote 15 Use of the site by the Roman army was relatively brief and terminated in an orderly withdrawal.Footnote 16
FIG. 1. Site plan of Carpow, showing results of excavations in 1961–62 and 1964–79. (Reproduced from Proc. Soc. Antiq. Scot. 129 (1999), 482, illus. 1 and 491, illus. 5. Courtesy of Professor J.J. Wilkes)
THE GATES AT CARPOW
The stone gates at Carpow were double-portal, so that we can envisage twin archways. There were many puzzling features. Indeed, so diverse were their ground-plans that it would be easy to think that they had been found by excavation at different sites (figs 2–4). The dimensions of the gates were all different, though approximating to Roman measurements.Footnote 17
FIG. 2. Ground-plan of the East Gate. (Reproduced from Proc. Soc. Antiq. Scot. 129 (1999), 504, illus. 14. Courtesy of Professor J.J. Wilkes)
FIG. 3. Ground-plan of the North Gate. (Reproduced from Proc. Soc. Antiq. Scot. 129 (1999), 506, illus. 16. Courtesy of Professor J.J. Wilkes)
FIG. 4. Ground-plan of the South Gate. (Reproduced from Proc. Soc. Antiq. Scot. 129 (1999), 508, illus. 18. Courtesy of Professor J.J. Wilkes)
The South Gate served as the principal access point for the fortress's water supply carried in wooden pipes secured by iron collars, set in channels below the level of the road.Footnote 18 At the East Gate two drains, stone-lined and unlined, passed through the south and north passageways respectively (below).Footnote 19 Both can be seen on aerial photographs of the fortress's interior approaching the East Gate from the west.Footnote 20 No outgoing drains or incoming water-pipe channels passed through or under the North Gate, which lay on the sea-facing side of the fortress.
The stone gates examined in 1964–70 exhibited three very different ground-plans (figs 2–4). At the East Gate the piers differed in shape from one another.Footnote 21 There was no continuous central spina but two square foundations, the larger of which was surely designed to support the extra weight of the heavy slab placed above it (fig. 2). The other was placed off-centre, which must have affected passage through the gate and influenced the design of the superstructure. A stone-lined drain, 0.45 m wide and up to 1.2 m deep, passed through the south passageway of the gate and cut the foundation trench for its southern pier, thus establishing the constructional sequence.Footnote 22 The stone-lined drain was found on excavation to have been infilled with road metalling and a large fragment of an inscribed slab (see below and Supplementary Material, Appendix 1), with a road surface laid on top. As this drain was not infilled when it was located elsewhere within the fortress, perhaps the intention at the East Gate was to improve the stability of the stone gate-structure.
At the North Gate (fig. 3) the layout of the neatly-kerbed foundation cobbling and a surviving course of masonry in situ above it appeared to preclude any passage through it.Footnote 23 Indeed the excavators did not initially consider it to be a gate.Footnote 24 Post-pits midway across both passageways could have been part of a preceding timber structure; smaller post-holes, presumably supporting timber uprights, suggest a number of lean-to structures against the gate.
The foundation cobbling at the South Gate appeared casually laid and incomplete (fig. 4); it similarly extended across the two gate-passageways. By contrast with the carefully designed North Gate, it gives the impression of ramshackle construction.Footnote 25 Yet these foundations seemingly supported a finely sculptured slab (see below). The squared-off masonry set midway over the west pier of the gate and over the spina could testify to the actual placing of the slab and suggest that not all the cobbling was eventually built on.
At the South Gate (fig. 4) the ground level, to judge from the published plan and accompanying photograph,Footnote 26 had been lowered, perhaps to facilitate the laying down of the cobble foundations, thereby shaving away two incoming water-pipe channels in the east passageway, which presumably went out of use, to be replaced by the broader channel in the west passageway, which cut through the foundation cobbling.Footnote 27
A number of timber features were found during excavation of the gates.Footnote 28 At the North Gate (fig. 3) four post-pits were located east of the stone gate; they appeared to belong to a free-standing timber-framed tower, which measured 10 by 12 Roman feet. No matching post-pits were found west of the gate in the areas examined. At the South Gate (fig. 4) post-pits were found next to the gate, suggesting a timber tower to either side.Footnote 29 Two post-pits ‘alongside the south side of the East Gate’ (fig. 2), mentioned in the final report but not shown on the published plans, were perhaps part of a similar structure.Footnote 30 There was no indication at Carpow that the timber posts had been forcibly removed at the end of their useful life. If the stone gates were inserted into existing timber structures solely to support the heavy slabs placed above them, as is suggested here, the building of flanking stone towers, whose absence has been remarked upon, may have been judged unnecessary.Footnote 31 At the North and South Gates the stone piers were wedged tight against the post-pits of the likely timber towers (see below), partly overlying them. The placing of the north pier of the East Gate above a pre-existing pit could suggest that its builders had no choice over where to place its foundations. The possibility that stone gates were flanked by timber towers seems unlikely.
At Carpow the insertion of deep foundation cobbling for the successor stone gateways could have removed traces of any timber predecessors, and account for irregularity and oddities in construction. As Carpow was much larger than a ‘standard’ fort built to house a regiment of auxiliaries, timber gate-structures of some complexity might be looked for; however it is hard to determine what their overall layout was. Flanking towers at timber gateways were generally integral with the gate-structures, not free-standing entities.
THE SLABS AT THE EAST AND SOUTH GATES
A particular highlight of the excavations was the discovery at the East Gate in 1964–65 of two sizeable fragments of a commemorative stone tablet, lying atop the road ‘in front of the south passage’ (figs 5–6).Footnote 32 The slab was presumably set into the outer facade of the gate-structure. A detailed description of all the surviving fragments can be found in the Supplementary Material, Appendix 1.
FIG. 5. Sculptured side-panel found at the East Gate in 1964 (RIB III, 3512(a)) (© Dundee City Council)
FIG. 6. The major fragments from the East Gate (RIB III 3512(a)–(b)), as drawn by R.D. Grasby. (Reproduced by permission of the Administrators of the Haverfield Bequest)
Publication was undertaken by R.P. Wright who interpreted the fragments as belonging to a commemorative slab specifically naming Caracalla who reigned as sole emperor from late a.d. 211 to 217.Footnote 33 Wright restored the first two lines of the inscription to read: [I] mp(erator) e[ t d(ominus) n(oster) M(arcus) Aur(elius) Antoninus / Piu] s F[ elix … ]; ‘Emperor and Our Lord Marcus Aurelius Antoninus, Dutiful, Fortunate …’ (see fig. 7). ‘Marcus Aurelius Antoninus Pius’ is not, as we might easily suppose, the emperor Antoninus Pius or his adopted son and successor Marcus Aurelius, but rather the elder son of the emperor Septimius Severus, whom we know as Caracalla.
FIG. 7. Text of the East Gate slab as restored by R.P. Wright. (Reproduced by permission of the Administrators of the Haverfield Bequest)
The left side-panel shows a winged Victory, a Capricorn and two Pegasi, sculptural imagery which presumably alludes to successful military campaigns. The presence of the Capricorn and Pegasi, emblems of the Second Augustan Legion, is a clear indicator of building work at the gate by its soldiers.
Presumably a matching right-hand side-panel bore more motifs, such as a kneeling captive or a military standard. Wright argued that the right-hand side-panel could have shown one or more emblems of the Sixth Legion which we know, from the survival of quantities of stamped roof-tiles at Carpow, was also active in building the fortress.Footnote 34 However, the appearance of the emblems of two legions on the same stone would be a rarity.
Wright's interpretation, that the inscription honoured Caracalla, seemed to conflict with the accepted chronology of the site at Carpow which most then believed was constructed in the aftermath of the invasion of Scotland by the emperor Septimius Severus and his forces in a.d. 208, and abandoned in early 211. Severus died at York in February 211 and, according to contemporary historians, his sons Caracalla and Geta who had been with him in Britain left immediately for Rome.Footnote 35
Wright's view was soon queried by J.C. Mann and M.G. Jarrett who argued that a dedication involving more than one emperor (i.e. Severus and Caracalla, with Geta as nobilissimus Caesar (a.d. 208–9), or Severus, Caracalla and Geta (a.d. 209–11), or Caracalla and Geta (a.d. 211)) was possible on his readings;Footnote 36 but Wright later reiterated his view in favour of a dedication recording building work by Caracalla alone.Footnote 37 More recently both Peter Warry and the late John Casey have offered restorations naming the emperor Commodus who ruled from a.d. 180 to 192.Footnote 38 See Supplementary Material, Appendix 2.
In 1969–70 three fragments of a similarly substantial slab were found during excavation of the South Gate ‘in the filling of a pipe-channel’, one of those carrying the fortress's main water supply into the fortress (fig. 8).Footnote 39 Presumably this means the curving channel which cut the South Gate's cobble foundations (see below). A full description of the fragments can be found in the Supplementary Material, Appendix 3.
FIG. 8. Sculptured side-panel found at the South Gate in 1969 (RIB III, 3514(a)) (© Dundee City Council)
SCULPTURAL DECORATION AND STYLE OF LETTERING
The left side-panel of the East Gate slab depicts a range of sculptural motifs familiar at Roman military installations in North Britain (figs 5–6); they call to mind the distance slabs from the Antonine Wall, presumed to have been carved in a.d. 142–43. The pelta-motif was in regular use by the Roman army in North Britain; bird-headed terminals were especially popular with the Second Legion.Footnote 40 The pelta-motif continued in use after the death of Antoninus, though increasingly ‘devolved’ and poorly executed, until the Severan period.Footnote 41 There are numerous continental examples of its use over a long period.Footnote 42
The slab at the South Gate was provisionally assigned by Wright and Hassall to the Second Legion also,Footnote 43 but the difference in the style of carving should indicate that it was not the work of the same craftsman as at the East Gate, and could therefore commemorate building work by a separate detachment or another legion. The pelta-motif on the South Gate slab (fig. 8) is carved with plumage which recalls a distance slab of the Sixth Legion.Footnote 44 Perhaps the vine-leaf with grape-cluster was a ‘signature’ of the stonemason.Footnote 45
The carving on the two slabs from Carpow, especially that on the slab from the South Gate, is superior in quality to much of the known relief sculpture from the military zone of Roman Britain. The carving on the distance slabs from the Antonine Wall is much admired, but in fact the quality varies.Footnote 46 On the other hand, identifying Severan-period relief sculpture in Britain is a difficult task, in the absence of datable contexts. The relief sculpture of Brigantia found at Birrens, Dumfriesshire, has been adjudged Severan,Footnote 47 but the fort there appears not to have been occupied at that time.Footnote 48 Professor Martin Henig advises me that, in his view, a number of stylistic features could favour, or at least not preclude, a Severan date for the sculptured panels at Carpow.
Given that so few letters survive on the slabs at Carpow, we can make only limited comments about their style and date. The lettering on the East Gate slab is deeply cut, equal in quality to many monuments from Rome or from the cities of Roman North Africa. Part of a single letter A is preserved on one of the fragments at the South Gate (see Supplementary Material, Appendix 3); the cutting is broader and shallower than the lettering at the East Gate. By contrast with the many surviving Antonine-period slabs, the limited number of datable Severan inscriptions erected by legionaries in North Britain can on occasion exhibit a good standard, but they still fall short of the quality which can be found in the Antonine period.Footnote 49
REPOSITIONING THE MAJOR FRAGMENTS FROM THE EAST GATE
In publishing the inscription R.P. Wright made a basic assumption, which I believe has gone unchallenged by subsequent commentators, that fragments RIB III, 3512(a) and 3512(b) were set close together, so that the first line of the text began [I]MP E[…, i.e. the first word was a grammatical form of the title imperator (see fig. 7).Footnote 50 The surviving letters in line 2 were seen by him as preserving parts of two imperial titles, [PIV]S and F[ELIX]. The nominative case of the restored word Pius in line 2 seemed to demonstrate that the nominative imp(erator) was to be read in line 1 (see above), in preference to the dative imperatori, making the emperor the subject of the sentence rather than its dedicatee. Wright expanded E in line 1 to ET (‘and’). Indeed, as he noted, no other restoration seems likely. He went on to suggest, on Sir Ian Richmond's authority, that the connective ET could have been followed by the abbreviation D N (dominus noster), ‘Our Lord’.Footnote 51
In the early Empire, building inscriptions naming an emperor regularly began with the abbreviations IMP CAES (for the titles imperator Caesar (nominative case) or imperatori Caesari (dative case)) followed by the family names and titles of the incumbent. From the middle of the second century other ways of referring to an emperor were being introduced, regularly involving the abbreviation D N (dominus noster). The earliest epigraphic use of the word-sequence dominus noster imperator can be dated to a.d. 155,Footnote 52 and of imperator dominus noster to the Severan period.Footnote 53 There is no need of a connective ‘et’.
The designation IMP ET D N (imperator et dominus noster) appeared to be unique, as Wright admitted. He could only argue that unfamiliarity with changing fashions in describing emperors’ titulature, during a transitional phase between styles, resulted in a superfluous ‘et’.Footnote 54 More significantly, the opening sequence IMP CAES remained the standard formulation on building records during and after the reign of Severus.
However, whichever dating is preferred, I can find no compelling reason to place the two major surviving fragments of the East Gate slab in close juxtaposition.Footnote 55 Fragment RIB III, 3512(a) preserves the upright stroke of an initial letter (see above). This may indeed be I (of the abbreviation IMP). However, to me it seems just as likely that Fragment 3512(b) preserves some letters from a quite different part of the text, in my view (see below) close to the ends of its first two lines.
The upper line on Fragment RIB III, 3512(b) could well read IMP ET but I consider ET as foreshadowing either more imperial titles or a reference to a separate personage named in line 2.Footnote 56 Seen in this light it is tempting to restore the second surviving letter of line 2 as E (rather than F),Footnote 57 again forming part of the word ET, with the preceding letter forming the end of an abbreviation such as [PROCO]S for proconsul (see below). The ET here, if close to the end of line 2, could be a connective preceding the titles of another person who is named in line 3. A survey of inscriptions which include the word ET indicates that it often serves as a link between named individuals. The connective ET regularly occurs at the ends of lines or at the beginnings of lines, incidentally demonstrating forethought in the laying out of texts.
On the well-known Arch of Severus in the Forum at Rome, erected by the Senate and People jointly to Severus, Caracalla and Geta in a.d. 203 (fig. 9), the first three lines of the text, as originally inscribed, all ended in ET.Footnote 58 In line 1 of that text, the word ET linked elements of Severus’ own titulature, in line 2 it served to presage the names and titles of his elder son Caracalla, and in line 3 (as originally inscribed) the names and titles of the younger son, Geta.
FIG. 9. The Arch of Septimius Severus in the Forum at Rome, south façade. The inscription was altered in a.d. 211 to remove the names and titles of Severus’ younger son, Geta. (Photo: Margaret J. Robb)
However, if Wright's restoration of the Carpow text is dismissed as implausible, little is gained if an equally unlikely restoration is offered here. The particular challenge is finding a suitable expansion for [I]MP in line 1, if these letters appear, as is proposed, towards the end of the line. Assuming for the moment that [I]MP is an abbreviation for imperator or imperatori, a quite narrow range of possibilities, among epithets and titles with which emperors were honoured, includes invictus imperator, ‘unconquered emperor’,Footnote 59 and optimus imperator, ‘best of emperors’.Footnote 60
On the other hand, we might wonder whether IMP is abbreviated not from imperator but from imperium (i.e. ‘power’, ‘empire’), and forms part of the phrase propagator imperii, ‘enlarger of empire’. Emperors such as Augustus, Claudius and Trajan had enlarged the Empire beyond its previous limits, and would with good reason have merited the title propagator imperii, but it is not attested for them. The first recorded epigraphic occurrence of the title comes surprisingly late, in a.d. 166 with reference to the eastern campaigns of Lucius Verus, then joint emperor with Marcus Aurelius.Footnote 61 It also occurs on a medallion issued jointly by Marcus Aurelius and Commodus in a.d. 178–80.Footnote 62 There appears to be no record of its use by Commodus during his own reign. The title can be written out in full, or is sometimes abbreviated to PROPAG IMP.
The title propagator imperii is attested epigraphically above all for Septimius Severus.Footnote 63 Caracalla too can be so described, when co-emperor with his father, but also as sole emperor from a.d. 211.Footnote 64 Thereafter the title is found only occasionally, before coming back into regular use under Constantine, his sons and successors in the early fourth century. The designation propagator imperii, when used to describe Septimius Severus, is found almost exclusively in his home province of Africa.Footnote 65 There is one example from Parentium (Poreč) in modern Croatia.Footnote 66 The title is unattested in Britain;Footnote 67 this lack may be due principally to the smallness of the surviving epigraphic corpus. A reference to Severus as propagator imperii in north Britain, where his campaigns had substantially extended the geographical extent of the province, and particularly at Carpow where Severus or Caracalla, or more probably both, had spent time, would be particularly apt.Footnote 68 Their northwards advance beyond the Cheviots and their military victories had resulted in a sizeable extension to the Empire, in recognition of which both, together with the younger son Geta, adopted the titles Britannicus Maximus early in a.d. 210, presumably reflecting victories won during the previous summer's campaign season.Footnote 69
In line 2 of the Carpow inscription Wright and subsequent commentators read [Piu]s F[elix], a combination of titles used by Commodus, Caracalla and many of their successors.Footnote 70 However, as already stated, I suggest that the first word was [PROCO]S, the standard epigraphic abbreviation for proconsul. The emperor had always been notionally proconsul of the provinces he had taken under his control, matching the title used by the governors despatched by the Senate to provinces which remained nominally within its ambit. Epigraphically the title proconsul is attested for emperors from Augustus onwards. It is not found in use by Commodus,Footnote 71 but Severus and Caracalla both employed it.
RESTORING THE EAST GATE INSCRIPTION
The suggestion offered above is that both lines 1 and 2 of the inscription on the East Gate slab ended in et, and that this word introduced either further imperial titles or the name or names of other emperors. We need to consider therefore whether it named Marcus Aurelius and Lucius Verus (joint emperors a.d. 161–69); or Marcus Aurelius and Commodus (joint emperors a.d. 176–80); or Severus and Caracalla (a.d. 198–209), with Geta as nobilissimus Caesar; or Severus, Caracalla and Geta (a.d. 209–11); or Caracalla and Geta (a.d. 211). If three individuals are named, then we are restricted to Severus, Caracalla and Geta, and a date-range of a.d. 208–11. Further, if the title propagator imperii is correctly restored in line 1 (see above), we have a firm pointer to the reigns of Septimius Severus or Caracalla rather than earlier emperors. So few letters survive that certainty is unattainable. Three possible restorations are considered in the Supplementary Material, Appendix 4. The likeliest (Appendix 4, no. 2) commemorated the emperor Severus, with his titles spread over two lines. All three restorations require an exceptionally elongated slab, but the heights of the few surviving letters serve to indicate its overall monumentality.
When complete, the slabs above the East and South Gates at Carpow must have been among the most substantial erected at a military site in Roman Britain, second only to, if indeed not surpassing in length, the fragmentary slab from Risingham, Northumberland, which was set into the wall of that fort's principia during the reign of Caracalla.Footnote 72
A.R. Birley, who has remained consistently in favour of a Severan date for the occupation of Carpow, raised the possibility of three separate inscriptions over three of the gates, one honouring each of Severus, Caracalla and Geta, so allowing space for their full imperial titles.Footnote 73 However I prefer to suppose that the two (or more) inscriptions had the same or similar texts, potentially differing only in the name(s) of the legion(s) responsible.
DISCUSSION
Despite many seasons of excavation at Carpow, in the years between 1961 and 1979, a total of some six months’ work in all, evidence which would allow close dating of its construction and occupational history has always been in short supply. The quantity of pottery and small finds recovered was small, testifying perhaps to a short occupation, or a careful tidying up at its end, or both.
A number of different possibilities can be considered when attempting to reconcile the archaeological, numismatic, historical and epigraphic evidence. Before the site at Carpow was subject to any archaeological excavation, a Flavian (i.e. first-century) date had been advocated.Footnote 74 A few coins of this period were recovered during excavation,Footnote 75 but no pottery.
Given the ‘Antonine’ date (i.e. between a.d. 138 and 192) advocated by R.S.O. Tomlin for the sculptural fragments and letter-forms from the East and South Gates (see above), it could be tempting to assign to Carpow a date contemporary with the Antonine Wall. If put on a map of Antonine Scotland, the site seems appropriate as an advanced base which effectively anchored and protected the line of otherwise exposed forts extending northwards into Perthshire.Footnote 76 However, no archaeological evidence from the site exists in favour of such a dating. Another option was Commodan occupation (a.d. 180–92), favoured by Tomlin and others (see above), followed by abandonment, then re-use at the time of the Severan campaigns (a.d. 208–11). Commodus won victories in Britain through his legate Ulpius Marcellus, and in a.d. 184 took the title Britannicus Maximus.Footnote 77 The location of these campaigns remains unknown, but could easily have involved northern Britain beyond Hadrian's Wall.
A Severan date for the fortress at Carpow was embraced in the final reports on the excavations undertaken in 1961–62 and 1964–79, but it was accepted that the pottery could allow a wider date-range.Footnote 78 Stratified coins from the site suggest that the fortress was given up at the close of a Severan occupation, not Antonine or Commodan.Footnote 79 A recent survey of the pottery at Carpow, comparing it to ceramic material from South Shields and Corbridge, supply bases for the imperial expedition of a.d. 208–11, also favoured the Severan dating.Footnote 80
The results of excavations at the gates in 1964–70 could be understood as furnishing evidence of two periods, in timber and then in stone (see above). This was a scenario advanced by the excavators in their interim statements,Footnote 81 but discarded by the time the final report came to be written.Footnote 82 Timber-built gates were the norm at first-century forts, and are regularly found on the line of the Antonine Wall and elsewhere in Scotland. The use of timber is rare in later contexts.Footnote 83
If the stone phase at the gates and the dedicatory inscriptions are Severan, should we be thinking of an earlier foundation date for the fortress as a whole? However nothing suggests a meaningful time-lapse between the suggested timber and stone phases. Initial construction of the fortress could therefore belong in the years a.d. 208–9, and the stone gates in a.d. 210–11.
For a military site of this size, appearing to belong in the late second or early third century, a perimeter wall of stone rather than a turf rampart might have been expected, perhaps with an earthen or clay bank behind. Yet in the later second century, many towns in southern Britain were being fortified with earthen banks.Footnote 84
One option might be that the site was initially conceived of as a ‘campaign base’ where troops over-wintered, without the expectation of permanency.Footnote 85 The arrival on-site of the emperor Severus himself could have prompted a rethink, leading to the erection of stone gates, thus allowing suitable commemoration of this outpost of Roman power at the edge of the known world. In this case the inscribed slabs would testify to this second phase and not to the fortress's initial construction. The solidity of the stone principia and of the baths could indicate that they are contemporary with the stone phase at the gates; however, no evidence was found of preceding structures.
The diversity of their ground-plans suggests that the three stone gates at Carpow were the work of three separate detachments, each tasked with creating a solid structure to support the heavy dedication slabs, but achieving that end in different ways. As we know that at least three gates were (re)constructed in stone, there should have been at least three inscribed slabs, more probably four.
There were indications of secondary phases at two of the stone gates. At the East Gate (fig. 2), the stone-lined drain was a later feature (see above), into which an inscribed stone block was then tipped and a road surface laid over it (see Supplementary Material, Appendix 1). At the South Gate one fragment of an inscribed slab ended up in a debris-filled pipe-channel;Footnote 86 the final gate structure, to judge from its foundations, was poorly finished off (fig. 4). If more than a single phase is thus indicated at the gates, we might want to consider whether occupation at Carpow extended over a longer time-period, but the archaeological assemblage does not support this.Footnote 87
By its very geographical position, the fortress at Carpow in the Severan period looks to be an isolated outpost, supported potentially only by the fort and supply base at Cramond on the river Forth.Footnote 88 No evidence of a natural or man-made harbour at Carpow has ever been found in the vicinity of the fortress, to support the prevalent view that it was supplied by sea.Footnote 89 Perhaps the intention, never realised, was to build, or rebuild, many other forts as part of a full-scale reoccupation of central and southern Scotland.
Maintaining a site so far north in Britain was unrealistic once the Severan expeditionary force withdrew. The garrison of Roman Britain was surely overstretched, as it had been in the Antonine period.Footnote 90 What is clear is that there was lavish commemoration on a monumental scale over at least two gates by highly competent legionary stonemasons perhaps travelling in the emperors’ entourage, or identified in the accompanying military force, craftsmen well versed in the traditional styles.Footnote 91
SUPPLEMENTARY MATERIAL
For Supplementary Material for this article please visit https://doi.org/10.1017/S0068113X19000138.
Appendix 1: Description of the fragments from the East Gate D1
Appendix 2: Restorations of the East Gate inscription by Peter Warry and John Casey D3
Appendix 3: Description of the fragments from the South Gate D4
Appendix 4: Three possible restorations of the East Gate inscription D4
ACKNOWLEDGEMENTS
The fragmentary slabs from the East and South Gates are housed at the McManus Galleries, Dundee. I am grateful to Mr A. Zealand and Christina Donald for access to those in storage and for giving generously of their time to facilitate my research. I am glad also to acknowledge help received from E.W. Black and from Professors W.S. Hanson, M. Henig and J.S. Richardson. Professor A.R. Birley, Professor D.J. Breeze and Dr R.S.O. Tomlin read drafts of this paper, to my great advantage. Staff at the Search Room of Historic Environment Scotland gave access to the Carpow excavation archive held there. Dr Laura Morley expedited the granting of permission by the Administrators of the Haverfield Bequest for the use of figs 6 and 7. Professor J.J. Wilkes willingly gave permission for the reproduction of figs 1–4, which were originally published in Proceedings of the Society of Antiquaries of Scotland. The Society's Publications Officer, Charlotte Whiting, was supportive throughout. Carol Devlin provided I.T. advice. The sources of the photographs and of the line-drawings are acknowledged in the captions.
EPIGRAPHIC CORPORA
- AE
L'Année Epigraphique, Paris, 1888 onwards
- CIL
Corpus Inscriptionum Latinarum, Berlin, 1873 onwards
- ILS
Inscriptiones Latinae Selectae, Berlin, 1892–1906
- Inscr.Ital.
Inscriptiones Italiae, Rome, 1931 onwards
- IRT
The Inscriptions of Roman Tripolitania, Rome/London, 1952 (electronic reprint (2009) with many added photographs)
- RIB
The Roman Inscriptions of Britain, Volume 1. Inscriptions on Stone, Oxford, 1965
- RIB I Addenda
The Roman Inscriptions of Britain, Volume 1. Inscriptions on Stone. Addenda and Corrigenda by R.S.O. Tomlin, Stroud, 1995
- RIB II
The Roman Inscriptions of Britain, Volume 2. Instrumentum Domesticum, Stroud, 1990–95
- RIB III
The Roman Inscriptions of Britain, Volume 3. Inscriptions on Stone found or notified between 1 January 1955 and 31 December 2006, Oxford, 2009
