4.2.2 Middens

Middens have always loomed large in the study of Mesolithic Europe (Milner et al. 2007). The historical legacy of the ‘Obanian’ (see also the ScARF Case Study: The Obanian), has exerted a powerful influence on perceptions of coastal habitation and the significance of shell midden deposits in Scotland; so much so that midden sites have been considered a defining characteristic of the Scottish Mesolithic, often viewed as part of a putatively mobile life-cycle. Yet they are rare in the Scottish Mesolithic and new Mesolithic sites are rarely middens; in 2004/5 six new Mesolithic sites were found in Scotland and none were midden sites. Existing midden sites display little consistency in size and content, some contain evidence for internal structures; some appear to result from specialised activities. Middens occur in a variety of locations from rock-shelters and caves to open-air shorelines and there is considerable chronological variation. Some go on to be used in later periods, and are often associated with Neolithic and Bronze Age activity, including burial (Pollard 1996; Saville et al. forthcoming).

Midden remains from Morton Farm, Tentsmuir, Fife ©SCRAN

The rich organic preservation, as found within a midden, is unusual by Scottish standards so that midden sites have been well studied. Nevertheless, research has tended to lump the remains together as if they represent a uniform phenomenon, despite the fact that in many cases all that they have in common is the presence of marine shells within a coastal location.

Midden sites vary greatly in size.  Scottish middens rarely occur as upstanding monuments. The large mounded middens of Oronsay are almost unique; only Risga, Loch Sunart, Highland, is also an upstanding midden and it is much smaller. The Forth Valley middens comprise visible mounds but they are composite sites stretching over a considerable period of time and have proved hard to interpret. There are Mesolithic dates (generally 4th millennium BC; Ashmore 2004a; 2004b) from the lower levels of Inveravon, Mumrills, and Nether Kinneil, and these are backed up by the discovery of a series of Mesolithic bone and antler tools from the same area, but there has been no detailed work of the scale carried out on the west coast sites, and that which has taken place shows that some of these middens continue inuse into the Iron Age (Sloan 1982; 1984; 1989; 1993).

Other Scottish middens are much smaller. At Loch a’ Sguirr, Skye, little more than a vestige of midden survives inside a rock-shelter (Hardy and Wickham-Jones 2009). Perhaps this is closer to the norm for Mesolithic sites generally. Even where conditions are favourable it is the presence of shells rather than bone that gives the Mesolithic midden its characteristic form. All sites must have contained some organic material, but most of it has decomposed over the millennia to leave only a black sticky deposit. Between Oronsay and Loch a’ Sguirr a series of sites completes the size range. The recently excavated midden at Sand, Applecross, measures roughly 8 x 8m (no more than 50 cu m), and could have built up over one or two intensive episodes of shellfish exploitation (Hardy and Wickham-Jones 2009).  That at An Corran, Skye, on the other side of the Inner Sound, is larger and has Neolithic and later activity in the upper levels (Saville and Miket 1994a; 1994b; Saville et al. in press).

The middens of the well known Oban Cave sites are difficult to estimate in terms of size due to the early date of excavation and the fact that they were often disturbed before archaeological work took place. This latter factor also applies to those Oban sites discovered in recent times such as Raschoille (Connock 1985) and Carding Mill Bay (Connock et al. 1992). In the east of Scotland there are fewer surviving middens, but the best known, at Morton, was neither upstanding nor large, though upstanding middens may have existed along the mouths of the River Tay (e.g. Broughty Ferry; Lacaille 1954), and around the Moray Firth (e.g. Milltown of Culloden; Wordsworth 1992).

Many midden sites date to the later part of the Mesolithic: Oronsay, Morton, and the Forth valley sites all have dates in the fourth and fifth millennia BC (Ashmore 2004a; 2004b). There are other midden sites with earlier dates: Sand, Druimvargie (Oban), and Ulva all have dates that relate to the sixth and seventh millennia BC (Ashmore 2004a; 2004b). Ashmore’s list, and the more comprehensive list of Mesolithic dates included here (see the ScARF downloadable date list) serve as a reminder that middens can occur at any point in the Mesolithic (see Table 2).

Published accounts of midden sites vary greatly in quality, but it is often possible to extract some interesting information, notably the great variety of material that makes up the bulk of the midden. Some middens are limpet dominated, some dominated by cockle, and some by oyster.  Recent work at Sand has highlighted the importance of crab (Milner 2008) and this is supported at both Oronsay (Mellars 2004) and Morton (Coles 1971) while some sites contain deposits of fish bones (Parks and Barrett 2009). Animal bone may be of less importance in the make-up of most sites, but some have significant assemblages of bird bone, most notably at An Corran (puffin; Bartosiewicz forthcoming). Resources such as seaweed, samphire, birds eggs, and sea urchins, all of which are less likely to survive in the archaeological record, are often overlooked as a likely resource exploited by early peoples. The value of midden sites as indicators of terrestrial fauna and exploitation is also worthy of emphasis.

The differences in midden make-up are not confined to food remains. There is evidence for internal structures at Oronsay and Morton, but not at Sand or An Corran.  There is also variation in the artefactual component, though this is more difficult to quantify due to the differing standards of excavation. Some middens contain microliths (Sand), others do not (Oronsay). The lithic assemblages often lack formal ‘retouched’ tool types, such as scrapers. Few midden excavations have considered remains lying outwith but adjacent to the midden. The recent work at Risga, however, yielded large quantities of lithic artefacts including knapping debris, microliths, and scrapers from an area to the side of the midden (Pollard et al. 1996), and these were also present at Morton (Coles 1971). Midden data are essentially incomplete until the adjacent non-midden areas that relate to each site have been excavated. At Sand, for example, large quantities of heat-fractured stone were found on the slope below the midden. In other (later) circumstances this might have been considered more akin to the remains from a burnt mound, but at Sand it was regarded integrally as evidence of the broader spectrum of activities that had taken place (Hardy and Wickham-Jones 2009).  Bone and antler artefacts such as bevel-ended tools are present on many sites, as is a range of cobble tools including hammerstones and bevelled pebbles.

In summary, midden sites are by no means common in the Scottish Mesolithic record, but because of the rich environmental and artefactual remains they produce they play a dominant role in accounts of the period. The wide range of material recovered from midden sites has made comprehensive analysis and publication difficult, and this has hindered interpretation of the way in which these sites should be fitted into the picture of Mesolithic Scotland as a whole.