How were Neolithic buildings roofed?
That must surely be one of the most frequently asked questions since Skara Brae re-emerged from the sand around 1850.
Because nothing survived of Skara Brae’s roofs, it has long been assumed they were made of perishable, organic materials – whalebone or timber beams supporting a roof of turf, skins, thatched seaweed or straw.
At Rinyo, in Rousay, however, the 1938 excavators encountered an irregular “pavement” of thin slabs on the floor of some of the buildings. Few of these were laid flat, with “quite a number” leaning up against wall and furniture.
Their conclusion was that this “pavement” represented a collapsed roof. 
But no-one seems to have paid much attention.
Especially when there were tantalising clues elsewhere a few years earlier.
At Skara Brae, rubble in the buildings led to the proposal that they had corbelled, beehive-like, roofs. But there was a fly in the ointment – the excavation team did not find enough to cover the entire span of the structures. 
But re-reading the early excavation accounts does hint at the possibility that some of Skara Brae’s buildings were stone-roofed.
In 1931, Childe described the conditions encountered in House Seven:
Had he unwittingly stumbled across roofing tiles? Tiles that he then had his labourers clear out with the rest of the “rubbish”?
Bearing this in mind, it is interesting to note his references, two years previously to fragile paving – hardly the best material to cover a floor.
The first evidence
It was the Ness of Brodgar complex that, in 2010, first produced Orkney’s first real evidence of a Neolithic roof.
In Structure Eight there was a layer of carefully trimmed, rectangular slabs lying on the floor deposits in the northern end. Such was their number that, like Childe at Skara Brae 81 years earlier, they were originally thought to represent paving. But it soon became clear that we were actually looking at fallen roofing material!
Site director Nick Card’s experience in traditional Orcadian roofing techniques proved invaluable in interpreting what had been uncovered.
The distribution of the slabs showed that their size decreased towards the centre of the building and were smaller at the end walls. This was not only indicative of roofing material but suggested the prehistoric roof followed the external curved ends of the structure.
With no evidence of post holes inside, it seems likely that a wooden framework was secured to the top of the building’s walls and the roof tiles arranged on top and held in place by their own weight.
Large quantities of white clay found among the tiles may have been used to seal and weatherproof the roof. Or possibly to bed the tiles down to prevent slippage and achieve a more regular profile.
The angle at which the tiles were lying, and their distribution, strongly suggested that Structure Eight’s northern roof section had collapsed in one sudden event.
The scale and grandeur of the monumental buildings at the Ness suggests they were meant to impress. It seems very likely, therefore, that their massive stone roofs played a major role in this.
There is little doubt that the stone roofs, particularly if they were atypical, would have dramatically enhanced their visual impact – not only from the outside but also from inside.
In Structure Eight, for example, a Carnegie Trust-funded study, by UHI Archaeology Institute masters student Neil Ackerman, calculated that the roof ridge was up to four metres above the building’s wallhead. 
Assuming the walls originally stood over 1.5 metres, the highest point of Structure Eight’s roof could have towered almost six metres above ground level!
The roofing material encountered at the Ness of Brodgar was fashioned from sandstone slabs that had been carefully shaped, with their edges trimmed, to form rectangular tiles of varying sizes (length 38-57cm; width 23-40cm; thickness 1.3-2.7cm). 
None of the tiles were perforated, suggesting they were propped against a wooden frame – the largest at the bottom and along the building’s long sides – and decreasing in size towards the ridge. The tiles were held in place by their own weight, perhaps aided by clay caulking.
Slabs like these were used for Orcadian stone roofs right through to the 20th century and it may be that the techniques use had not changed much over the millennia.
-  Childe, V. G., Paterson, J. and Bryce, T. (1929). Provisional Report on the Excavations at Skara Brae, and on Finds from the 1927 and 1928 Campaigns. With a Report on Bones. In Proceedings of the Society of Antiquaries of Scotland (Vol. 63, pp. 225–280).
-  Childe, V. G. (1930) Operations at Skara Brae during 1929. In Proceedings of the Society of Antiquaries of Scotland (Vol. 64, pp. 158–191).
-  Childe, V. G. (1931). Skara Brae: a Pictish village in Orkney. Kegan Paul: London.
-  Ackerman, N. (2020) A roof over their heads. In Card, N., Edmonds, M. and Mitchell, A. (eds) The Ness of Brodgar: As it Stands. The Orcadian: Kirkwall.