Because of the long antiquarian interest in the Roman occupation, the distinctive morphological characteristics of Roman forts which are readily identifiable from the air, and the refinement of artefact dating criteria, there is a good overall knowledge of the distribution pattern of forts period by period (e.g. Breeze et al. 1997). Nonetheless there are obvious gaps in the pattern, not least in SW Scotland where further forts might be anticipated, potentially at strategic coastal locations such as Irvine or Ayr, or in the E at the crossing of the Tweed near Berwick (see section 3; Keppie 1990). The recent identification of a fortlet at Kirwaugh, on the south side of the river Bladnoch near Wigtown (Britannia 42 (2011), 336, fig 9) necessitates reconsideration of the likely pattern of occupation in Wigtownshire. Given the level of intensity of aerial survey and the favourable conditions for the production of cropmarks, the paucity of sites in the Lothians is generally considered to represent a real gap (Bishop (2004, 175-6, fig 116) has suggested a road east of Inveresk, but the evidence is not yet published).
Forts are usually located on communications routes, though the road lines are not always known and rarely independently dated. Often forts are found at river crossings. They are generally about one day's march apart (15-20 miles), though distances vary. There has, however, been no systematic study of the location of forts in terms of their relationship with roads/rivers, orientation, tactical considerations, articulation with other installations, or relations to indigenous settlement patterns or landscape features beyond small-scale case studies. There has, however been a recent systematic study of the location of fortlets, not restricted to the Scottish evidence (Symonds 2007). The increase in the numbers of fortlets located between forts in SW Scotland in the Antonine period is generally taken to indicate greater concern for local security (Maxwell 1977; Breeze 1974; see 3.4 above).
Recent work has made progress with interpreting the literary sources for the occupation, notably Ptolemy's Geography (Strang 1997, 1998), and ongoing work by classicists on the sources may provide further insights into the problematic texts which remain. These are often frustratingly difficult to correlate with the archaeological evidence.
Roman forts in Scotland show a remarkable consistency of design and layout, both generally and in relation to specific building types, indicating adherence to a set of general principles. In the past this has led to assumptions about determining the overall layout of a fort from limited sampling, as for example at Fendoch (Richmond and McIntyre 1939). More recent extensive excavation has provided a more reliable type site for timber-built forts at Elginhaugh (Hanson 2007), whilst at the same time manifesting unique features, demonstrating that all forts are different in significant detail (in contrast to all-too-frequent stereotypes). It is also important to note that every site exhibits uncertainties, as recent work has shown at Carriden (suggesting that the conventional location of the fort and annexe should be reversed) and Mumrills (where the sequence is increasingly complex) (e.g. Britannia 33 (2002), 287; 34 (2003), 303; 41 (2010), 350 and Bailey 2010).
It has also long been axiomatic that there is some direct connection between the layout of an auxiliary fort and the type of unit in garrison, and, moreover, that it was the norm for different types of unit to be housed in their own custom-built forts. Thus, one of the primary foci of excavation strategies has been to establish the character of the garrison (e.g. Frere and Wilkes 1989). It is now more widely accepted that forts constructed for single units were the exception rather than the rule (e.g. Maxfield 1986, 59), so that estimating the garrison strength from limited evidence of the interior layout is no longer feasible. Furthermore, it is only relatively recently, as a result of excavations at Wallsend on Hadrian's Wall and at Elginhaugh (Hodgson 2003; Hanson 2007), that it has become more widely accepted that horses for cavalry units were accommodated with the men in stable-barracks. Thus, in the absence of epigraphic evidence, virtually all previous assumptions about the character of units in occupation at forts in Scotland require revision.
Very few bathhouses have been identified outside Flavian auxiliary forts and it has recently been suggested that this is the period when they begin to appear, with higher-status cavalry forts being the first to build them (Bidwell 2009).
One of the primary areas of Roman military studies is concerned with the development of frontiers and the Antonine Wall has been a focus of study for over a century (e.g. GAS 1899; Macdonald 1934; Hanson and Maxwell 1986; Keppie 2001; Breeze 2006a) (see Antonine Scotland (c. AD 139-165) and The Antonine Wall above). Although its development, periodisation and character seem to be quite well understood, there are still many gaps in understanding among which are:
- an incomplete sequence of fortlets;
- an absence of anticipated watchtowers (Bailey 1995, with the possible exception of one at Garnhall (Woolliscroft 2008) whose identification is disputed);
- uncertainty about which of the two central forts (Auchendavy or Bar Hill) is primary and;
- the function and full distribution of the so-called 'minor enclosures' (cf Hanson and Maxwell 1983). Similarly, the full implications of the secondary character of annexes has still to be worked out (c.f. Bailey 1994).
Given the length and central location of the Wall, elements of it are regularly under development threat. As a result ongoing small-scale work continues to provide the opportunity for cumulative enhancement of knowledge, exemplified in the discovery of a previously-unsuspected series of obstacles on the berm (Bailey 1995). There has been a long tradition of periodic roundups of such evidence, and it is vital that such surveys are facilitated in the future.
More fundamental questions about the function of the linear barrier remain in dispute, directly paralleling the debate about Hadrian's Wall. The debate essentially concerns whether the Wall was designed to be a 'permeable' boundary designed to control movement into and out of the province, or to be a defended barrier primarily intended to deal with military threats. The debate tends to focus on whether or not there was a defended walkway along the wall top, a question not readily susceptible to archaeological proof (although the very absence of watch-towers would have made a wall-walk essential).
The narrow neck of the Forth-Clyde isthmus followed by the Antonine Wall is an obvious potential frontier location which also seems to have been utilised briefly during the Flavian conquest of Scotland (Tacitus, Agricola 23). Unfortunately, there is very little supporting structural archaeological evidence, other than the fort at Camelon and fortlet at Mollins. The suggested earlier use of Antonine Wall sites goes largely unsubstantiated, with the possible exception of Cadder, Castlecary and Mumrills which have all produced Flavian finds (Hanson 1980), the latter also revealing early phases of building in the annexe apparently yielding Flavian pottery (DES 1996, 42). At none of these sites is the evidence sufficiently strong to support Flavian occupation and a review of the pottery dating evidence would be timely.
The only other postulated frontier in Scotland, along the Gask Ridge, has been extensively studied in the last decade, with many of the sites undergoing excavation (e.g. Woolliscroft 2002; Woolliscroft and Hoffmann 2006). By contrast to the Antonine Wall, it is largely defined by its watchtowers, along with a decrease in the spacing between forts and the addition of some fortlets, all features which characterise frontiers more widely. Suggestions for its chronological context vary from after the withdrawal from the more northerly forts (Breeze 1982 , 61-5), to the temporary halt on the isthmus (Hanson 1991b, 1765-7), to an even earlier, pre-Agricolan, establishment (Woolliscroft and Hoffmann 2006, 178-90), although the latter interpretation has been strongly challenged (Hanson 2009a). Moreover, the very interpretation of the Gask system as a frontier has been reconsidered, and an alternative role as a protected supply line to the legionary fortress at Inchtuthil proposed (Dobat 2009).
Aerial survey should be used to continue to enhance the known distribution of forts, with particular focus on 'gap areas' such as the SW, and making best use of relevant dry summers, or on locations where stray finds may hint at missing sites (c.f. Keppie 1990). Vertical air photographs taken in summer months should continue to be analysed.
The systematic study of forts within their wider landscapes should be encouraged (see further below).
Any opportunities for more detailed survey and excavation along the line of the Antonine Wall should be taken, with a view to identifying further smaller structures.
The evidence for fort 'types' and the character of garrisons should be reviewed in the light of the identification of stable-barracks.
See also the ScARF Case Study: Elginhaugh Roman Fort