Because of the long-standing contribution of aerial reconnaissance, Scotland has long been at the forefront of the discovery of Roman temporary camps (e.g. St Joseph 1973, RCHME 1995; see 2.4 above). Leaving aside those which probably relate to construction or training activities, lines of march are apparent in the overall pattern of camp distribution, though the picture is, inevitably, partial (see Figure 2, Figure 6, Figure 10). Some distinct groupings of camps have been identified on grounds of morphology and/or size (e.g. Stracathro camps with their distinctive clavicular gateways, generally dated to the 1st century (Jones 2009c), or the '63-acre' series; Jones 2011, 100-2). Associating the various 'series' of camps with specific campaigns is problematic because of the difficulties of obtaining independent archaeological dating evidence from anything other than very large scale excavation, as emphasised in a recent major re-examination of the evidence (Jones 2006a; 2011). Where large scale excavation has occurred, as at Kintore (Cook and Dunbar 2008), the results suggest that the picture is even more complicated, with sites being potentially reused across a long period of time. Similarly, excavation at Dunning has indicated reuse which was hitherto unsuspected (Dunwell and Keppie 1995). The very size of many of the camps provides some confirmation of the evidence from pollen analysis that the contemporary landscape in the lowlands had already been extensively cleared of forest (Hanson 1997, 208-9). Even the largest of recent excavations have been confined to camp interiors, but it would be worth assessing the area outside camps for traces of activity as well; some are known to have annexes, while one might wonder about the presence of exterior rubbish pits or other features, and even the presence of camp followers.
One major battle between Roman and indigenous forces is attested in the literary evidence, as taking place at an unknown location named Mons Graupius in the Flavian period. Despite much debate over many years (e.g. Maxwell 1990; Fraser 2005; Campbell 2010), scholars are no further forward in the confident identification of even the general area of the battle, let alone its exact location. Other battles no doubt also occurred and there are two postulated sites, where Roman siege works surround an Iron Age hillfort. The supposed practice works at Woden Law (Richmond and St Joseph 1983) may be dismissed as made up of a combination of outer defences for the hillfort and land divisions of Iron Age date (Halliday 1982, 82). The Roman character of the siege works at Burnswark is, however, not in doubt, but here opinion is divided over their interpretation, with some favouring a genuine siege (e.g. Campbell 2003; Davies 2009; Hodgson 2009; Keppie 2009) and others preferring to see it as a training exercise (e.g. Breeze 1982 ; Hanson and Maxwell 1986; both following Steer 1964 and Jobey 1978).
Ongoing aerial survey to enhance the known distribution of camps, with a particular focus on 'gap' areas and making best use of dry summers should be undertaken. Vertical air photographs taken in summer months should also continue to be analysed.
Further opportunities for area excavation within temporary camps where ditch sections alone are insufficient, should be grasped. There is also a need to sample areas immediately outside the camps, for comparative purposes. Geophysics has considerable value in detecting internal features for investigation and perhaps in identifying different camp uses.
Undertake systematic metal-detecting survey of possible sites of Mons Graupius, since this is how the battle site of the Varus disaster was finally identified in Germany (Schlüter 1999; Clunn 2005; Wells 2003). Casual finds should also be monitored with this in mind, as the iron and lead finds likely to be diagnostic are often discarded by detectorists.
The interpretation of the Burnswark earthworks is highly controversial and renewed fieldwork would offer some resolution.
See also the ScARF Case Study: Burnswark