The medieval building in its basic form can readily be seen as a typical later medieval house divided into three spaces: the main living area or hall with a hearth, an inner room or chamber to the west, and a service area to the east. The hall and chamber were defined by a beam-slot and traces of an earthen floor, features that were not present in the service area, whose extent was defined only by the terrace on which the building was constructed, and the length of the drainage ditch to the rear. It is assumed to have been divided from the hall by a cross-passage. There was no clear evidence that earth-fast posts supported the structure at any time, and it is likely that it was timber-framed with the roof supported by tie-beams spanning the full 4.5m width.
Construction is only broadly dated to the 13th to 14th centuries. A sherd of pottery from the beam-slot and a few sherds from the lower fill of Ditch O came from the earliest contexts, but this is poor dating evidence and in any case is not distinguishable from the general date range of the pottery. This, which mostly comprises jars in glazed and unglazed fabrics typical of the region, suggests a relatively brief duration of existence c. AD 1250-1350, and the chronology cannot be refined further. The demise of the farmhouse corresponds closely to the time of abandonment and contraction of medieval settlement generally in Devon, a trend attributed to the transition from a primarily agrarian to a more pastorally based economy and the resultant reorganisation of landholding (Fox 1991a). In upland areas, such as Dartmoor, a retreat from marginal land due to climatic deterioration has also been seen as a factor (Weddell 1997), but this seems far less significant for East Devon. The decision not to rebuild the farmhouse after the devastating fire may, however, have been because of the wetness of the location and in a sense the land might always have been marginal for the purposes for which it had been used.
The structural form of the house does not closely correspond to examples known from this period. It does, however, appear to fit with the typical later medieval technique of timber framing, as opposed to the use of earth-fast posts, a constructional change that became ubiquitous in the country from around AD 1200 and gave rise to the possibility of houses being more stable and longer-lived by being less susceptible to decay in the ground (Gardiner 2014, 16–17). The absence of ground-cut structural features from the early 13th century typically presents a challenge of discovering the remains of houses at all, except where defined by drainage ditches or surviving stone sills (Gardiner 2014, 19). The Island Farm house had a sill-beam set in the ground to the rear (i.e. to the north), or at least set more deeply there than at the front (the south), where some of the original ground has been lost, and the different depth must have been intended to create a level building. The builders mitigated the effect of damp to some degree by starting with a raised platform of gravel as a base and by cutting a drainage ditch along two sides close up against the wall. There was no evidence for the structure of the eastern room, which was not integral to the living area. Construction may, however, have employed timber frames set directly on the ground, or alternatively have been wholly or partly of clay or cob, as the surviving evidence from later buildings shows to have been typical in Devon. It appears, therefore, that the house was not constructed as an entity, although there is no evidence that the construction platform and drainage ditch were extended eastward at a later date, and so there seems to have been the intention to provide for a building here at the outset. It is possible that the different structural approach reflected differences in specification for the different parts of the house, perhaps simply related to cost or the availability of timber.
It would appear significant that the evidence for cooking, craftwork (repairing copper-alloy vessels) and storage (crops and other items), which may all have been the purpose of the eastern room in some situations, all came from the hall and chamber rooms at Island Farm. This lends support to the interpretation of the eastern room here having an agricultural function. Its size suggests that it was a byre and the building therefore a longhouse in the strict sense of housing people and animals under the same roof, with the respective accommodations separated by a cross-passage (although a rear entry here would have been impeded by the drainage ditch). There are several examples of medieval longhouses from excavations on Dartmoor to which the Island Farm building is broadly comparable (Henderson and Weddell 1994). The examples include one at Dinna Clerks, destroyed by a fire, where the collapsed roof had sealed several pots and two wooden platters in the central living room (Beresford 1979, 135). Despite a different constructional technique, the Island Farm longhouse shares with the Dartmoor examples, among others, the common characteristic of having little evidence of a partition between the house and byre; the separation of the western room or chamber from the hall is invariably more substantial. Its overall size and the proportions of the rooms were very similar to those of the excavated mid-13th-century longhouse at Meldon, Okehampton Park (Austin 1978, fig. 5; Gardiner 2000, fig. 6), Island Farm perhaps having a slightly longer byre (7m rather than 6m) (Figure 14). The proportions provided an almost symmetrical design, supposing the common entrance and cross-passage to have been between the two units, although there was no archaeological evidence for the doorway. The eroded depression to the south of the building in this area perhaps suggests where cattle gathered. The elongated pit within the byre is suggested to have been a feature produced by 'mucking out'. There was a very similar feature in the longhouse at Meldon, Building A1 (Figure 14; Austin 1978, fig. 5), described as a 'heavily robbed central drain' (Austin 1978, 205), although it is not clear that there was any stone structure here, nor that it would have been preferentially robbed when the house walls had survived to a height of a metre and more in places. Meldon Building A1 may well have had a gable-end doorway for the disposal of effluent, as found in the smaller longhouse there, Building A3 (Austin 1978, fig. 7); the Island Farm longhouse may have had a similar design. Both Meldon buildings had lines of numerous stakeholes just inside the side walls suggesting that the animals were tethered and/or fed from racks facing the walls, and the same evidence was found at Longhouse A, Sourton, in its early phase (Weddell and Reed 1997, fig. 16). In the byre at Island Farm there was just one posthole (of uncertain purpose) although a number of stakeholes lay around the entrance to the main room. There was evidence at Black Daren, Llanveynoe, Herefordshire, that cattle were fed from the cross-passage, and at Chapple, Gidleigh, on the east side of Dartmoor, tethering posts were preserved between the byre and cross-passage (Lake 1989, 76–7). At Island Farm their interpretation is uncertain, not least because of their apparent location both inside and outside the internal doorway.
The division between the hall and chamber is thought likely to have been formed by a stone wall, perhaps supporting a loft space. At Garrow Tor, Cornwall, however, Platform House 1 had well-preserved stone furnishings that included a stone bench at the end of the hall occupying about one-third of the width of the building (Dudley and Minter 1962, fig. 88), and it is possible that the stone 'wall' at Island Farm was a bench instead.
Internet Archaeology is an open access journal. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.