Several of the forts on the Wall were provided with an annexe, that is an enclosure with a variety of functions attached to one side of the fort, though Carriden, Mumrills and probably Falkirk have two. Many annexes were encountered during the initial investigations into the layouts of the forts themselves between 1900 and the 1930s, though there was no systematic search for them. The absence of an equivalent provision on Hadrian’s Wall has led to the suggestion that they were intended as an alternative to the Vallum (Bailey 1994, 305), though annexes are commonly provided at forts, and occasionally fortlets, across northern and western Britain from the Flavian period onwards.
The annexes vary considerably in internal area (0.3–1.7ha). Two (Duntocher and Rough Castle) were larger than the forts to which they were attached, though the identifications of the two enclosures at Duntocher may need reassessment (cf Swan 1999, 431–3).
The annexes were usually defined, like the forts, by a rampart and one or more ditches, though detailed excavated evidence for their defences is limited. Two ditches are attested in the north-west corner at Mumrills, but no meaningful rampart was detected (Bailey 2021, 311–54). The provision of annexes does not seem to have been part of the original scheme for the Wall, though only at Rough Castle has it been possible to investigate the junction between the Wall and annexe rampart. The latter appears to abut the Wall, but some slight uncertainty is introduced by the positioning of a culvert (Buchanan et al 1905, 466 and fig 1). Elsewhere (eg Balmuildy, Castlecary and probably Falkirk) there is evidence, such as the infilling of ditches between fort and annexe or differences of construction material used for the ramparts, which suggests the annexes were later additions. Though superficially of one build with the fort, the annexe at Bearsden also seems to be secondary as neither the north and south gates nor the headquarters building were centrally located within the fort. Rather they were central to the whole fort/annexe enclosure. This suggests that the annexe had been created by subdividing what was originally intended to be a larger fort during its construction (Breeze 2016a, 330–4).
When the decision was made to add annexes to forts and whether it was, indeed, a single decision is a matter of much debate (see The Wall in its historical context). The evidence from both Bearsden and Duntocher clearly indicates that it took place there while the Wall was still being built. Arguing for a hiatus in the whole construction process before the decision was taken, Swan dated it to the return of a task force from Mauretania in AD 149/50 on the basis of the widespread distribution of pottery with strong North African influences (1999, 445–6). A slightly later date for the return of these troops is indicated by several diplomas (Eck and Pangerl 2016), which would imply a considerable delay before the Wall was completed if this were to provide a terminus post quem for all annexes. However, despite Steer’s assertion to the contrary, the southern ditch of the western annexe at Mumrills was clearly earlier than the outer fort ditch, whose infilling has been dated by pottery and coins to the mid-150s AD (Macdonald and Curle 1929, 402; Steer 1961; Bailey 2021). This would suggest that the infilling of the fort ditches was not a necessary precursor to the construction of the annexe at Mumrills or, by extension, at any other fort. Though Bailey now postulates, somewhat tendentiously, that the southern annexe ditch at Mumrills must therefore relate to an earlier Antonine fort on the same site (2010).
There has been very little attempt, other than occasional exploratory trenches and some geophysical survey, to shed light on the interiors of the annexes. Bathhouses occur in several, in some cases in what is clearly a secondary context such as at Balmuildy (see Forts). At Rough Castle, the annexe contains what appears to be an additional enclosure of some sort, once suggested as a fortlet though that now seems an unlikely interpretation based on its size and shape (see Fortlets; also Macdonald 1933).
Where annexes have been more extensively investigated, notably at Mumrills and Falkirk, they have produced significant quantities of Roman material culture as well as evidence of buildings of varying sizes and complexity that were predominantly of timber construction. They also contain traces of cobble surfaces, pits, ovens or furnaces and metalworking, suggesting areas of semi-industrial activity (Bailey 1994, 305–09 and 2021, 399–462; Dunwell and Ralston 1995, 572). Geophysical survey has recently indicated that a large tile kiln excavated at Mumrills over a century ago was located in the eastern annexe (Macdonald 1915, 123–8 and plates II and III; Hanson et al forthcoming). There is insufficient evidence to be certain about the density of that activity, though the fact that at some sites (such as Balmuildy and Falkirk) buildings were erected over the infilled ditches of the forts would seem to suggest that space was at a premium. However, excavation or watching briefs in the northern and southern areas of the annexe at Bearsden failed to reveal any certain buildings other than the bathhouse and latrine (Breeze 2016a, 320–3).
There is considerable debate about the function of annexes generally. Some see them as serving entirely military requirements for the production and maintenance of equipment for the Roman army and the provision of secure areas for goods and vehicles in transit, or the protection of livestock, such as cavalry horses (for example Breeze 2006, 95; Hanson 2007, 667–9; Hanson 2021). Others prefer to equate them with the provision of protection for civilian settlement (Sommer 2006, 118–23).
- Take any opportunities to investigate larger areas within the interior of annexes, which have seen very little excavation, in order to understand their function better.
- Bring to full publication the various geophysical surveys undertaken at several sites along the Wall (see Research and methodological issues).
- Take any opportunities to obtain more information on the structural relationships of annexes with forts and the Wall itself.
- Undertake further work, including geophysics, on the additional enclosure within the annexe at Rough Castle to try to ascertain its possible function.