The excavations at Broxmouth identified three roundhouses (4, 5 and 7) with scooped floors. All were built using stone walls with timber roof supports and in scoops up to 1m in depth, indicating that sunken floors were a deliberate feature of the building, rather than simply a by-product of the levelling of a stance for the house. All houses had been reconstructed at some point and one, House 4, has been rebuilt at least four times. It is likely that the stone walls were probably in excess of 1m. Some of the interior walls may have been lined with wattling. Often some of the construction circuit may have been made up of timber, not stone. In subsequent phases the areas of the houses were reduced by the refacing of the enclosing wall and their floors were paved. House 4 was particularly important, being the best preserved, the most rebuilt and the structure with the most finds.
Houses of similar construction to the stone-walled houses at Broxmouth have been identified widely across South East Scotland and North East England (Jobey 1974; Hill 1982; Pope 2003). More widely, Hill identified a pattern of unenclosed settlement across the Tyne–Forth area, often overlying derelict settlements and fortifications (1982).
On their initial discovery at Broxmouth they were called Votadinian houses, placing them in the native settlement context at the time of the Roman arrival. However, this cultural affiliation is not entirely helpful (Hill 1982; Armit and McKenzie 2013). That said, Hill was broadly correct in terms of period – subsequent dating of the stone houses at Broxmouth places them within Phase 6 of the site, that is 100 BC to AD 210. Several examples can be seen at nearby Chesters (Drem) and examples have been excavated at Edins Hall, Berwickshire (Dunwell 1999). Excavations at St Germains produced evidence comparable to that from Broxmouth, comprising scooped floors and paved areas set within the derelict remains of an iron age enclosure, and possibly dating to the first to the third century AD (Alexander and Watkins 1998, 247–248). To this evidence can now be added that from Knowes, Whittingehame Tower and Standingstone and from Eweford and Phantassie (Haselgrove 2009; Cowley 2009; LeLong and MacGregor 2007).
As Cowley suggests, this excavation evidence can now be marshalled to sustain a compelling case for unenclosed settlements of scooped houses and yards as a widespread component of the settlement pattern in the period between the 2nd to 1st centuries BC and the 2nd or 3rd centuries AD (2009). Further, the excavated evidence, in particular from Broxmouth and Knowes, bears directly on the interpretation of irregular features (blobs) recorded as cropmarks on aerial photographs, both within enclosures and also in apparently unenclosed contexts. The widespread ‘speculative’ distribution of unenclosed settlements of Roman Iron Age date suggests a dense pattern of occupation of the East Lothian plain.
That said, recent work at Glenrath has hinted at complexity that may not have been fully appreciated. Work in Glenrath focused on a stone roundhouse considered likely to belong to the broad middle to late Iron Age stone-built tradition described above. Radiocarbon dates demonstrated that the roundhouse was probably Middle Bronze Age in date, with no indication of later activity (Cavers and Heald forthcoming). Furthermore, the structure overlay a field bank that was considered to be Iron Age in date, raising the possibility that formalised field systems associated with stone walled roundhouses may have emerged earlier than previously considered (RCAHMS 1977).