Discovery and Excavation in Scotland 1997, p. 18:

Port na Cille, Isle of Gunna (Coll parish)
Survey and trial trenching
NM 099 511 Topographic survey and trial trenching were carried out near the harbour of Port na Cille, on the supposed site of a medieval chapel (NMRS NM 05 SE 2) and burial ground, in advance of a proposed house construction. The survey recorded over 30 features, including the vestiges of 20 turf-walled structures, a turf dyke, several areas of lazybeds, and a ditch and upcast-bank. Ten trial trenches, opened over areas to be disturbed by the proposed development, uncovered several archaeological features. These included a structure, probably of late medieval date, consisting of stone footings with a rubble core, and the footings of a second, probably earlier, structure beside it on a slightly different alignment; a ditch, 1.5m deep, which produced a sherd of low-fired pottery and a flint blade in its primary fills, as well as post-medieval pottery and nails in its upper fills; and a collapsed field wall, with low-fired pottery and layers of midden (including shells, animal bones and charcoal)
associated with it.
The original house plot had been excavated previously, in the absence of a professional archaeologist, by mechanical digger. Surviving in the sections of this trench were the remains of a substantial midden, 1m deep, consisting of layers of organic material rich in shell and animal bone lying above burnt horizons. The layers were cleaned, recorded and sampled, and
sherds of low-fired pottery and a few flint flakes were recovered from them. A spread of organic, rnidden-rich material on the surface of the trench proved to lie above and around a linear spread of stones, which may have formed a structure. Finds recovered from the spoil heap include a retouched flint blade, several sherds of low-fired pottery, fragments of copper alloy and many pieces of animal bone and shell. The pottery appears to be early medieval in date.
Sponsor: Marcus de Ferranti.

Discovery and Excavation in Scotland 1998, p. 17:

Isle of Gunna (Coll parish) H F James (GUARD)
Late medieval/post-medieval settlement
NM 1035 5115 A five week excavation took place on the Isle of Gunna in advance of the construction of a new house. The excavation revealed that the terrace on which the new house is to be constructed was occupied at least since the 15th century. A prehistoric presence on the island was indicated by at least one sherd of pottery, but there were no structures from this period.
The earliest structures were in a fragmentary condition but were single-celled, oval structures and three post-medieval structures were constructed over them. The site may have been occupied until the early 18th century.
Structure A was an oval structure, 7.6m long and 3.6m wide internally with walls up to 2m thick constructed of an internal and external skin of drystone. infilled with organic rich sand. The corners were rounded both externally an internally. There was an entrance on the N side and a blocked window on the S side. Stone sub-dividing walls had been constructed at the W end over an earlier hearth. An annexe attached to the W end of this structure was not excavated.
Structure B was 4.1m long and 2.4m wide internally, with similarly thick walls and a hearth at the E end. The internal corners were square and this structure also had an opposing door and window. The N and E walls had been strengthened internally with the insertion of an extra skin of stone and sand.
Structure C was very fragmentary, but appeared to be the remains of an oval, thick-walled structure, the extent of which could not be ascertained. No hearth or other internal features
were identified.
Four 1 x 1m test trenches were dug outside the visible ruins to investigate the deposits on the terrace. In Test Trench 1 post- medieval midden layers sealed windblown sand, within which a single sherd of prehistoric pottery was found. In Test Trench 2, beneath a post-medieval midden, were the remains of a collapsed wall associated with a layer of peat ash. This was later identified as. the edge of Structure D. In Test Trench 3 a stone wall was immediately visible beneath the post-medieval midden spread and this was later identified as part of Structure G. Within Test Trench 4 a post-medieval midden deposit sealed a ploughsoil with evidence of a single phase of ard marks cut into the windblown sand.
Because of the discovery of structures to the S of Structure B, within the new house plot, the whole area of the house plot was deturfed and the area excavated. Significant lengths of wall from Structure B were removed and the deposits beneath the floors were examined to windblown sand. A total of five other structures were identified.
Structure D consisted of the footings for an oval shieling with two drystone wall skins and a hearth. It measured approximately 3m long and 2m wide internally and had been truncated by Structure G.
Structure E was a drystone structure at least 3m in diameter which was dug into windblown sand. The walls of this structure survived to a height of l.l m and the primary deposits were 1.5m below the present surface. It was interpreted as a latrine. A whale vertebra had been used as a door pivot. It contained a complete hand-thrown pot.
Structure F was an oval, drystone structure measuring 3m wide and at least 5m long internally. The walls were fragmentary, but survived in places up to 0.4rn high. It contained a deep deposit of ash. Midden deposits beneath the walls and floor contained sherds of a late medieval glazed jug (c 15th century) and a circular, undecorated medieval bronze brooch. The structure had been infilled with midden deposits consisting of animal bone and limpets before Structure B was built on top.
Part of Structure G was exposed in the new house plot. This structure had two entrances within the exposed length of walling. There was a small hearth in between these entrances. Structure H consisted of very fragmentary remains of an oval structure which pre-dated Structure B. It measured at least 3m long and 2m wide, and contained a small hearth.
Beneath the floor of Structure A, at the E end, there was a stone base for a hearth surrounded by stake and post-holes. No stone walls were identified at this level within the exposed area
At the W end there were several lenses of midden, windblown sand and possible old turf lines exposed over windblown sand. This area was examined to a depth of up to 1.5m below the
present surface.
Beneath Structure C there was a midden deposit which sealed ard marks, a shallow curving ditch and other linear features which did not form any particular pattern. There are few finds
associated with these features and they are thought to be late medieval or later in date. (GUARD 502.3).
Sponsor: Marcus de Farranti