• Keine Ergebnisse gefunden

Antonine Wall

Im Dokument The Antonine Wall (Seite 85-110)

The Antonine Wall: Papers in honour of Professor Lawrence Keppie: 61–66

periods of circulation before deposition – and, as a general rule, the older the coin the more worn it appears (Figure 5.2).

For virtually unworn coins of earlier reigns than that of Pius in the vicinity of the Antonine Wall we can look to Camelon, on the western outskirts of Falkirk, some 1.2 km (0.75 miles) to the north of the Wall. Anne Robertson’s catalogue of the 65 coins excavated there between 1975 and 1981 (Robertson forthcoming) includes, amongst 33 Flavian issues, two slightly-worn bronze issues of AD 77-78 and five unworn or slightly-worn bronze coins of AD 86 or 85-86. A further virtually-unworn coin of AD 86 was recovered in 1988 from the northern ditch of what appears to be another fort that pre-dates the nearby Antonine one (Figure 5.3; Brickstock forthcoming, coin no. 33).

Camelon, together with Mollins and Barochan and perhaps others beside (Hanson 1980; Breeze, 2006:

63) are thought to have been Agricolan foundations forming a line of garrison posts across the

Forth-Figure 5.2. Examples of coins from the Antonine Wall: a. a very worn denarius of Vespasian from Carriden (AD 71; Brickstock forthcoming, coin no. 27); b. a worn denarius of Hadrian from Mumrills (AD 118; Brickstock forthcoming, coin no. 5);

c. a virtually unworn denarius of Antoninus Pius from Mumrills (AD 140-43; Brickstock forthcoming, coin no. 14)

Figure 5.3. A virtually unworn dupondius of Domitian as COS XII, AD 86 from Camelon (a. obverse; b. reverse)

a b c

a b

Clyde isthmus, sixty years before the building of the Antonine Wall. Robertson also records a little-worn bronze coin, probably of AD 86, from Barochan (1983: 409).

On the basis of the coin evidence Robertston concluded that ‘the Flavian occupation at Camelon lasted from a date in or after AD 77-8, until AD 86 at the earliest’ and that ‘after a probable interval of 50-55 years, there followed an Antonine occupation which lasted until at least AD 154-5’ (forthcoming).

There are, however, a couple of aspects of the coin assemblage that might cause us to modify these conclusions somewhat.

Since the publication of David Walker’s highly influential study of the coinage of Roman Bath (1988), numismatists and archaeologists generally have been aware of a number of peaks in supply of coinage (and bronze coinage in particular) to Roman Britain in the first century AD, and have used them as significant staging posts in the dating of the sites. These peaks occurred in AD 64-67, 71-73, 77-78 and 86-87. A lesser peak is apparent for AD 95-96, at the very end of Domitian’s reign, followed by a period of rather more regular supply but, as Walker pointed out ‘the fact that a coin series ends with coins of 87 merely indicates that occupation probably ceased by about 96, not that it ended in 87 or 88’ (1988: 287). I would add to this the observation that in the north, by which I mean sites north of the Humber, coins of AD 86 appear generally much more common than those of AD 87 (at Corbridge, for example, the ratio of AD 86 to 87 is 6:1): this might be argued to be an indication of an exceptional phase of activity in AD 86 in the militarized north, but it is perhaps rather more likely to be a function of coin supply to the north - in which case the presence or absence of issues of AD 87, especially from a small assemblage, should not automatically be taken as a reflection of occupation or abandonment of a particular site in AD 86 or 87 (contrary to the argument presented in Hobley 1989).

To return then to Camelon: the earliest little-worn coins are those of AD 77-78, which ought to place the foundation soon after that, c. AD 80, which would tie in very well with the policy of consolidation on the Forth-Clyde line that is outlined in Tacitus’ account of Agricola’s fourth campaign season (Tacitus, Agricola 23).

However, to muddy the waters somewhat, coins of AD 71-73 outnumber those of 77-78 in both lists by a ratio of more than three to one. This comparison is one of a number of numismatic criteria that can be used to recognise early Flavian sites (even though many of the coins may be worn and relate to later phases of occupation; see Table 5.1). I am not quite sure how to interpret this circumstance, but, given that other criteria are either not satisfied or only imperfectly so (Claudian copies, for instance, being entirely absent), the greater significance should probably be given to the observation that the earliest little-worn coinage is that of AD 77-78 rather than that of 71-73.

Beyond that, the little-worn coins of AD 86 attest occupation to at least that date; but coins of AD 87 are absent, both from the 1975-81 finds and from the fuller site list recorded by Richard Abdy (2002).

According to one’s take on my comments above, that should date the end of the Flavian phase of occupation of Camelon either to late AD 86/87 specifically or (my preference) to sometime in the decade between late AD 86 and 96.

The same argument can, of course, be applied to other sites, including, for instance, the uncompleted legionary fortress at Inchtuthil, where the small assemblage ends with six little-worn coins of AD 86. Having re-examined them twice in recent years, all appear to me to be unworn or virtually so,

consistent with abandonment of the fort in late AD 86/87, but the absence of coins of AD 87 should perhaps not be taken to be of absolute significance, which leaves open the possibility (though not, I think, the probability) of a slightly longer occupational phase.

Incidentally, Robertson alleges (forthcoming) that ‘no undisputed examples of bronze coins of either of these two Flavian groups – Vespasianic of AD 71-3, or 77-8, minted in Gaul, and Domitianic of AD 86, or of AD 85-6, or later – have so far been recorded from sites on the Antonine Wall’. This is not quite true, since a dupondius of AD 86 was excavated from the southern end of the fort bathhouse at Balmuildy (Abdy 2002: 207, no. 3). That coin, however, is very worn (according to my classification) and thus almost certainly represents an Antonine deposit. Likewise, although the Antonine Wall assemblage has a ‘tail’ reaching back to the late republic (i.e. denarii of Mark Antony, which are commonly found on second- or even early-third-century sites), it does not really satisfy any of the criteria outlined above (Table 5.1): there are, unsurprisingly, no Claudian copies; later Flavian deposits outnumber earlier Flavian; higher denominations outnumber lower (e.g. some 20 denarii and sestertii from Mumrills as opposed to 14 dupondii and asses); and, as has already been noted, no little-worn first-century coinage at all.

Beyond the Flavian period, Robertson envisages abandonment of Camelon for half a century, prior to a second phase of occupation in the Antonine period. It seems to me, however, that there are strong grounds for suggesting that the first phase of occupation continued, albeit perhaps on a limited scale, until well into the second century. There are two reasons for suggesting this. Firstly, there are a small number of little-worn issues of later years, including both a denarius and a sestertius of Trajan, both dating to AD 103-11, which cannot belong to a Flavian phase and are unlikely to have survived in such unworn condition into the Antonine period. Secondly, there are a number of earlier issues which are sufficiently worn to suggest deposition in the reign of Trajan rather than that of Domitian, but hardly so worn that they are likely to be Antonine deposits. This can be illustrated using the same (admittedly subjective) technique of calibration according to circulation wear applied above to the coins from the Antonine Wall (Figure 5.4).

It will be observed that almost as many coins are assigned by this technique to the period AD 96-117 (the reigns of Nerva and Trajan) as to that of Domitian (AD 81-96), followed by a complete gap for the reign of Hadrian (AD 117-38). Without wanting to push the limits of the technique too far, it is perhaps possible to refine the level of detail a little further since, for the Trajanic period, most of the suggested deposition dates lie in the first decade of the second century rather than the second.

The presence of some pre-Flavian coin, especially bronze, but also republican denarii The presence of Claudian copies, driven from circulation early in the Flavian period Flavian coins of AD 71-3 outnumbering those of AD 77-78

A predominance of lower denominations, especially asses, in the assemblage as a whole The presence of some little-worn coin (SW/SW = slightly worn)

Table 5.1. Numismatic criteria for recognising early Flavian sites

That circumstance, taken together with the little-worn coins of the same period, might allow one to postulate continued occupation, at least on a limited scale, until c. AD 110, followed by a complete break until c. AD 140, a gap of some three decades rather than Robertson’s five. Conversely, however, in honesty it should be admitted that the limits of accuracy of the technique are such that, although there is certainly a gap in the Hadrianic period, a proportion (perhaps 20%) of the deposits assigned to the Antonine period could arguably be late Hadrianic (though, on balance, I think this unlikely).

Here a comparison with the site of Newstead, some 85 km (53 miles) to the south-east on Dere Street, is instructive, since it is another site where both Flavian and Antonine periods of occupation are attested.

At Newstead, however, it is accepted (on wider archaeological grounds, not least the rebuilding of the fort in the late Flavian period) that occupation continued into the Trajanic period (Hanson 2012) – but there is currently no other evidence that this was the case at Camelon and the suggestion therefore represents a new departure.

The coin assemblage from Newstead has recently been discussed in detail (Holmes 2012) but a few further remarks may be permissible here, based on my recent re-examination of the coins with an eye to circulation wear (Figure 5.5). Here the picture suggested is of a primary, Flavian, phase but with some level of occupation continuing into the second century (again indicated both by suggested deposition dates and also, significantly, by several little-worn denarii of Trajan). As with Camelon, there would appear to have been a gap in deposition from sometime in the mid-late Trajanic period through to late in the reign of Hadrian or, more likely, early in the reign of Antonius Pius since, given the limitations of the technique, much of the peak for AD 117-38 that distinguishes the Newstead histogram from that of the Camelon could well belong not to the later years of Hadrian but to the early years of Pius.

Where Newstead differs from both Camelon and the Antonine Wall forts, however, is in its clear continuation through the reign of Marcus Aurelius and into the early 180s, with a very few coins also suggesting some continued presence in the Severan period (the latter, however, in all probability linked to Severus’ Scottish campaigns rather than to any continuous occupation).

0

pre‐1 41‐54 54‐68 69‐81 81‐96 96‐117 117‐38 138‐61 161‐80 180‐92 193‐222

Camelon  1975‐81

From there we come back to the main point of this short paper, that is the realisation that although the coin profile of the Antonine Wall forts is superficially similar to both Camelon and Newstead from the late republic up to and including the reign of Pius himself, the pre-Antonine coinage of the Wall is all demonstrably residual in comparison to both Camelon and Newstead, for which we are able to demonstrate periods of earlier occupation or activity.

0

pre‐1 41‐54 54‐68 69‐81 81‐96 96‐117 117‐38 138‐61 161‐80 180‐92 193‐222 222‐38 238‐60

Newstead

Abdy, R. 2002. A Survey of Coin Finds from the Antonine Wall. Britannia 33: 189-217.

Breeze, D.J. 2006, The Antonine Wall, Historic Scotland, John Donald.

Brickstock, R. J. 2017. Continuing the search for an ‘Antonine Gap’ on Hadrian’s Wall, in N. Hodgson (ed.) Roman Frontier Studies 2009, Proceedings of the XXI International Congress of Roman Frontier Studies (Limes Congress): 331-35. (Archaeopress Roman Archaeology 25).

Oxford: Archaeopress.

Brickstock, R. J. forthcoming. Coins from the Antonine Wall, in G. Bailey The Antonine Wall in Falkirk District: excavation and observation 1991-2017. Falkirk.

Hanson, W.S. 1980. Agricola on the Forth-Clyde isthmus. Scottish Archaeological Forum 12: 55-Hanson, W.S. 2012. Newstead and Roman 68.

Scotland: the Flavian to Antonine periods, in

F. Hunter and L. Keppie (eds) A Roman Frontier Post and its People, Newstead 1911-2011: 63-76.

Edinburgh: National Museums Scotland.

Hobley, A.S. 1989. The Numismatic evidence for the Post-Agricolan Abandonment of the Roman Frontier in Northern Scotland.

Britannia 20: 69-74.

Holmes, N.M.McQ. 2012. The Roman Coins from Newstead in context, in F. Hunter and L. Keppie (eds) A Roman Frontier Post and its People, Newstead 1911-2011: 123-36. Edinburgh:

National Museums Scotland.

Robertson, A.S.  1983  Roman coins found in Scotland, 1971-1982. Proceedings of the Society of Antiquaries of Scotland 113: 405-48.

Robertson, A.S. forthcoming The Coins, in V.A.

Maxfield Excavations at Camelon 1975-81.

Walker, D.R. 1988. Roman Coins from the Sacred Spring at Bath, Part 6, in B. Cunliffe (ed.) The Temple of Sulis Minerva at Bath, Volume 2: The Finds from the Sacred Spring: 281-358. Oxford:

Oxford University Committee for Archaeology.

Nick Hannon, Lyn Wilson, Darrell J. Rohl

Introduction

In a previous paper analysing the Antonine Wall Distance Stones, the need to reassess the issue of spacing of the installations along the frontier was highlighted (Hannon et al. 2017). The primary reason necessitating this reassessment is the enhanced metric framework now available for the frontier’s linear components derived from the recent LiDAR survey of the World Heritage Site commissioned by Historic Environment Scotland (Wilson et al. 2013); data which formed the basis of the first author’s PhD thesis. These improved measurements for the first time accurately account for the changes in elevation encountered along the Wall’s course, resulting in the measured length of the rampart increasing from 60.24 km to 62.03 km; an increase of 1.79 km (Hannon et al. forthcoming). It was argued that this three dimensional measurement approach reflected the techniques used by the Roman surveyors, who originally set out the frontier’s installations, more closely than a two dimensional map-based approach could. The following paper serves two purposes: it will explicitly publish the three-dimensional distances between the known installations along the Antonine frontier and will also reconsider the issue of installation spacing.

In 1975 John Gillam hypothesised a model to describe the constructional sequence for the Antonine Wall (Gillam 1975). This model suggested that the Wall was originally planned to have six forts constructed at around eight Roman mile intervals (known as primary forts). Between the forts, fortlets were built, at intervals of around 1.1 miles, in a pattern similar to that observed with Hadrian’s Wall’s milecastles, which were usually built at one-mile intervals (Breeze 2006a: 64). Gillam further suggested that a change was enacted before this initial plan was completed, leading to the intervals between forts being reduced to around two miles (a fort spacing first proposed by Horsley 1732: 173) resulting in the addition of at least 11 forts to the frontier (known as secondary forts). At the time of Gillam’s proposal the existence of four fortlets had been established, but renewed interest in the Wall generated by his hypothesis led to the discovery of additional fortlets at Seabegs Wood in 1977, Kinneil in 1978, Cleddans in 1979 (Keppie and Walker 1981), Croy Hill in 1977 (Goodburn 1978: 413-15) and Summerston in 1980 (Grew et al 1981: 32; Maxwell and Hanson, this volume). These discoveries added weight to Gillam’s argument supporting the view that a chain of regularly spaced fortlets once existed between the Forth and Clyde, as the locations of these newly discovered fortlets fit within the model’s framework. However, the newly discovered fortlets did not complete the proposed sequence, leaving around three-quarters of the hypothesised series still undiscovered.

A number of hypothetical reconstructions of the series, building on Gillam’s work, have proposed specific locations for the frontier’s missing fortlets (notably, Keppie and Walker 1981; Hanson and Maxwell 1986; Woolliscroft 1996). These reconstructions are similar in that they all use a two dimensional map based approach to establish the distances between the known fortlets and, thus, the positions of those The Antonine Wall: Papers in honour of Professor Lawrence Keppie: 67–85

that are merely hypothesised. These models also struggle to fit the proposed fortlets neatly within a regular spatial pattern, requiring in a number of cases for the intervals to be lengthened or shortened to accommodate the proposed number of fortlets. Now that an enhanced metric framework is available for the frontier with an updated overall length, both Gillam’s hypothesis and the fortlet models can be reassessed in order to test their validity. Two conclusions pertinent to the issue of the Wall’s installation spacing were reached in the aforementioned analysis of the Distance Stones: firstly, that the standard of measurement referred to on the inscriptions was probably based on the pes Drusianus and, secondly, that the inscribed measurements anticipated or acknowledged the existence of not just the forts considered primary but also those deemed secondary (Hannon et al. 2017).

The decimal equivalent of a Roman mile is normally quoted as 1.48 km, however this oversimplifies the situation. While a Roman mile (mille passus) was always a thousand paces (passus) and a pace was five feet (pedes), the standard used for the Roman foot was variable. Analysis of Roman structures in Britain has shown that two standard Roman feet were used: the first was the pes Monetalis, which has a decimal conversion of 0.296 m and the second was the pes Drusianus with a decimal conversion of 0.332 m (Walthew 1978: 335). In the northwest of the empire on some occasions the contemporary use of both standards has been attested within the same fort, such as at Corbridge (Walthew 1981:

15). If a mile is calculated based upon 5000 pedes Monetales, we reach the commonly quoted 1.48 km.

However, if this calculation is repeated with the pes Drusianus, a length of 1.66 km is achieved. Millett (1982) has raised valid concerns that in reality it is difficult to distinguish between these two standards on the ground, due to the subtle difference between each Roman foot. However, due to the Antonine Wall’s considerable length, the differences between a system based on a mile derived from either the pes Monetalis or pes Drusianus would be exaggerated, and this larger scale makes identification of

Figure 6.1. Fortlet sequence showing distances between fortlet centres shown as miles derived from both the pes Drusianus and pes Monetalis measurement standards

the standard used easier to ascertain. Both units of measurement have also been identified for the setting out of town plans, with the pes Monetalis being suggested at Silchester and Baginton and the pes Drusianus at Verulamium and Longthorpe (Duncan-Jones 1980: 132). Reference to the Antonine Itinerary shows miles equating to metric distances of around 1.665 km, implying the pes Drusianus may have been adopted as the standard by the authors of this document (Bishop 2014: 26). These examples support the view that different standards of measurement coexisted in Roman Britain.

Fortlet spacing

Previous reconstructions of the Antonine Wall’s fortlet sequence have been based upon the assumptions that each fort was spaced at intervals of one Roman mile and that this mile is based upon the pes Monetalis standard (Keppie and Walker 1981: 161; Hanson and Maxwell 1986: 122; Woolliscroft 1996:

160). While this reassessment will respect the first assumption, as it has been identified that the pes Drusianus standard was used for the Distance Stones (Hannon et al. 2017), both measurement standards will be tested for the fortlet intervals. For the purpose of this paper, intervals have been calculated based upon the centres of the northern gate of each fortlet. Measurements making adjustments for each fortlet’s east-west dimensions were also considered, but due to the small dimensions of the fortlets the widths had little effect on the results, so for brevity have not been included here. If we accept that fortlets were spaced at regular intervals, it is clear that a significant number of the fortlets remain undiscovered, thus we are dealing with a partial dataset. In an attempt to account for the gaps in the data, the variance quoted in Figure 6.1 is calculated from the closest whole mile, e.g. the variance for an interval of 2.42 miles is 0.42 miles as the measurement is closest to a two whole mile interval

160). While this reassessment will respect the first assumption, as it has been identified that the pes Drusianus standard was used for the Distance Stones (Hannon et al. 2017), both measurement standards will be tested for the fortlet intervals. For the purpose of this paper, intervals have been calculated based upon the centres of the northern gate of each fortlet. Measurements making adjustments for each fortlet’s east-west dimensions were also considered, but due to the small dimensions of the fortlets the widths had little effect on the results, so for brevity have not been included here. If we accept that fortlets were spaced at regular intervals, it is clear that a significant number of the fortlets remain undiscovered, thus we are dealing with a partial dataset. In an attempt to account for the gaps in the data, the variance quoted in Figure 6.1 is calculated from the closest whole mile, e.g. the variance for an interval of 2.42 miles is 0.42 miles as the measurement is closest to a two whole mile interval

Im Dokument The Antonine Wall (Seite 85-110)