3.6 Forts

There are 17 forts or probable forts currently known along the line of the Antonine Wall. Most of them were first recorded in antiquarian accounts either as extant earthworks or concentrations of Roman finds. The locations of the latter were later confirmed by excavation, aerial reconnaissance or geophysical survey. Two further fort sites, at Seabegs and Kinneil, have been postulated because the gaps between the forts on either side were wider than the norm. Neither have been confirmed, though fortlets have been identified in the immediate vicinity (see Fortlets). The distribution of the forts along the line has generally been taken to indicate that they were meant to be between 2 and 3 miles apart, but there is, in fact, no set interval. Other factors, such as the local topography and the relationship with north-south routeways, should also be considered (Graafstal et al 2015, 63-4).

All of the forts but two (Bar Hill and Carriden) are attached to the barrier. However, their structural relationship with it varies (see Planning and building the Wall). Old Kilpatrick, Balmuildy and Castlecary clearly predate the construction of the rampart, and geophysical evidence suggests that the same may have been the case at Auchendavy (Jones et al 2006, 13–14; Hanson 2020b, 11). The small fort at Duntocher also predates the Wall, though it post-dates a fortlet on the same site. The same seems to apply to the fort at Castlehill (Hanson and Jones 2020). Inveravon, Westerwood, Croy Hill, Rough Castle and Cadder have all produced stratigraphic evidence to suggest that they were constructed after the Wall rampart was laid out, though the presence of a causeway across the Wall ditch at the latter could imply that this fort was constructed, or at least planned, before the Wall ditch was created. Despite the implication of the published plan (Robertson 2015, fig 40), there is no direct evidence of the relationship between the fort and the Wall at Falkirk. Most forts were, like the Wall, defined by ramparts of turf or clay on a stone base, but two (Balmuildy and Castlecary) had stone walls.

We are now in a position to provide at least reasonable estimates of the sizes of all the known forts, which vary considerably (0.12–2.6ha). In other contexts the two smallest (Inveravon and Duntocher) would be referred to as fortlets (Symonds 2017, 5–12) and several others (Rough Castle, Westerwood and Croy Hill) were not sufficiently large to house a full auxiliary unit. Indeed, it is difficult in most cases to see how the fort sizes relate to the recorded garrisons. At several forts more than one auxiliary unit is recorded epigraphically, though cavalry are poorly attested in comparison to Hadrian’s Wall and legionary detachments are thought to have been quite widely used (Breeze 1993, 288–90; Breeze 2006, 81–94 and 189–92; Robertson 2015, 31–4).

All the forts on the line of the Wall are oriented towards it, which usually means north because of the general orientation of the Wall. The one exception is Cadder, which faces east, though it may have originally been designed to face north given the apparent central location of its north gate. We are fortunate in having a wealth of data for the central range of buildings (principia, praetoria and granaries) mainly from sites excavated between 1900 and the 1930s. These were usually stone-built, but several forts (Bearsden, Old Kilpatrick, Cadder) have one or more central buildings of timber construction, and in some cases (Mumrills and Cadder) there are buildings that appear to have both timber and stone phases. Evidence for the existence of workshops in the central range is slight, while the identification of non-standard buildings is often hampered by the use of posthole construction (eg Breeze 2016a, 314–20 and 335–43).

Several of the forts (Old Kilpatrick (possibly), Bearsden 1, Balmuildy, Cadder, Castlecary, Bar Hill, Westerwood and Mumrills) have internal bathhouses, which is unusual because these were more often located outside the fort, presumably due to the potential fire risk and associated social activities. The significance of this arrangement in the Wall forts is debated (Bailey 1994; Keppie 2004, 204–09). However, external bathhouses, sometimes in an annexe (see Annexes), are equally common (Duntocher, Bearsden 2, Auchendavy, Croy Hill, Rough Castle, Falkirk and Carriden) and the small size of several of the forts makes provision of an internal bathhouse seem unlikely. Currently only Old Kilpatrick, Cadder and Balmuildy are known to have both internal and external bathhouses. The sizes of the bathhouses do not seem directly to reflect the size of the fort, while variations in the box flue tiles suggests that each unit was responsible for producing its own (Keppie 2004, 218–19).

Excavation in the front (praetentura) and rear (retentura) sections of forts has generally been more limited, but where this has taken place, long narrow, timber buildings are indicated by postholes. These are generally interpreted as barracks, though they are often smaller than the norm (eg McIvor et al 1979, 280–1; Breeze 2016a, 337). It has been suggested that some forts may never have been fully provided with internal buildings (Keppie 2009, 1138), though this is difficult to demonstrate given the lack of large-scale modern excavation in their interiors (Breeze 2020). The known plans of accommodation blocks are too fragmentary to contribute meaningfully to any calculation of fort garrisons, which have been postulated on the basis of the epigraphic record (eg Hanson and Maxwell 1986, 153–8; Breeze 2006, 91–4; Keppie 2009). Various issues need to be addressed: whether stable-barracks can be identified; the structural implications of the presence of different units attested epigraphically; the physical manifestation of any postulated legionary garrison; the identification of areas of specific activity such as food preparation and consumption; and the implications of finds that indicate the presence of women and children within forts (Allason-Jones et al 2020).

It was long considered that the Wall went through more than one phase of occupation (eg Hanson and Maxwell 1986, 137–43), but now a single phase of use over no more than 21 years is widely accepted (Hodgson 1995; see The Wall in its historical context). Nonetheless, there is structural evidence of more than one phase of building within some forts (eg Mumrills and Cadder) that seems to require explanation beyond that of repairs or refurbishment.

Research issues

• Establish if there are any other forts along the line of the Wall.
• Establish the relationship of the forts to the linear barrier where this is not known or in dispute.
• Encourage the publication of the excavations of the fort at Camelon.
• Take any opportunities to establish the nature of, and any phasing within, interior buildings, particularly in the smaller forts and those that have seen no internal investigation.
• Take any opportunities to investigate the character of the barrack/stable accommodation.
• Encourage the detailed analysis of finds and their distribution within the forts.
• Seek to establish any links (structural and artefactual) between the forts and the soldiers who built and were housed in them (eg Chorus 2009).
• Undertake systematic comparative analysis of the buildings within the forts, including the number and types of troops, their deployment and comparison with other Roman forts.