Much knowledge of the conquest and garrisoning of North Britain in the Flavian period was originally derived from Tacitus' biography of his father-in-law, Agricola, governor of the province from AD 77-84 or 78-85- either from AD 77 or 83 (e.g. Birley 1999a; Ogilvie and Richmond 1967; study of this by classicists continues to provide fresh possibilities, e.g. Campbell 2010, 84-7). However, earlier writers indicate some knowledge of the geography of Scotland (Breeze 2002a). This includes the existence of Orkney, Shetland and the Hebrides, as well as that Thule lay even further north. The interrelationship between the historical and archaeological data was discussed and analysed in the 1980s, and a broad consensus achieved (Breeze 1982, 42-67; Hanson 1991a). In the last two decades there has been considerable debate about the role that Agricola played in the conquest of the north, linked primarily to the results of extensive fieldwork on the Gask Ridge (below), with various publications proposing a revised dating (notably Caruana 1997; Shotter 2000; Woolliscroft and Hoffmann 2006). Many of the arguments relate to the discovery of early finds and their role in dating the sites in which they were found. A conference and subsequent monograph has made the current lack of consensus apparent (Breeze et al. 2009). There is still considerable debate about the nature of the process of conquest (campaigning vs. fort construction), the significance of structural phasing within such forts and other structures, and the character of the archaeological dating evidence.
There are two literary references to pre-Agricolan military activity (Pliny NH 4, 102; Statius Silvae 5.2, 140-9), both unspecific and open to various interpretations, but sufficient to indicate that there was some penetration into Scotland before Agricola (listed in Breeze 2009; for a handy compilation of translated sources, see Ireland 2008). In addition, ancient geographers provide snapshots of knowledge, for example, identifying that Britain was an island (e.g. Breeze 2002a); the main source of evidence is Ptolemy's Geography, from the mid 2nd century AD, although some of his data was gathered from earlier sources (Rivet and Smith 1979; Mann and Breeze 1987).
The dendrochronological dating of the fort at Carlisle to AD 72 (Caruana 1992) shows pre-Agricolan activity on the doorstep of modern Scotland. In the absence of similar dating evidence from other sites, the claims of those asserting an 'early' conquest date are unlikely to be readily assuaged. Recent geophysical work at Dalswinton (Bankfoot) suggests that the postulated early vexillation fort there is actually a camp (Hüssen et al. 2009). Early finds have been identified from other sites, but in such small numbers that the evidence is as yet unconvincing; date of production can be divorced by some distance from date of deposition. The only other fort site with a potential claim to be an early foundation (on the basis of the coin evidence) is Newstead (Shotter 2000, 197). However, excavations at Red House, Corbridge (Hanson et al. 1979) and Elginhaugh (Hanson 2007), both on Dere Street (as is Newstead), do not suggest foundations earlier than Agricola. Indeed, a probable foundation deposit at Elginhaugh provides a terminus post quem for its construction of AD 77-8.
The ongoing debate is a valuable reminder of the need to keep an open mind over even the supposedly solid foundations of the period, although the verdict at present for the revisionists seems to be 'not proven'.
The most likely 'early' sites are going to be camps, which are notoriously difficult to date without excavation; large-scale extensive geophysical survey can highlight features in their interior, such as ovens, which could be targeted. Trial excavation at Dalswinton, Bankfoot, might serve to confirm the geophysical results, while geophysical survey and trial excavation of the enigmatic enclosure to the east of the camp at Ardoch would help to confirm its identification.
Full assessment of early work at sites such as Loudoun Hill and Milton would be beneficial in exploring potentially early activity in Scotland.
Any possibilities of obtaining dendrochronological dates from secure contexts should be seized.