Radiocarbon Dating: Avoiding Errors by Avoiding Mixed Samples
Ashmore, P. J., Antiquity
Time-depth makes archaeology special in studies of human societies, and the chronologies provided by radiocarbon, with all their faults, are crucial to the interpretation of evidence about past ways of life (Taylor 1987; Baillie 1990; 1995) over most of the last 40,000 years. During the last two decades Historic Scotland (the government agency which acts on behalf of the Secretary of State for Scotland) and its predecessors have accepted well over 1000 applications for dating of archaeological samples, and rejected between 200 and 300, in addition to accepting or rejecting many applications for palaeo-environmental dating. Laboratory standards have improved greatly during that period (International Study Group 1982; Scott et al. 1988). But sample selection by palaeontologists and by archaeologists, except those dating artefacts, has not in general improved as much. Despite the lessons provided by sites such as Creag nan Uamh (Murray et al. 1993: 1-10) use of samples of bulked-together organic material has remained popular.
At Creag nan Uamh (TABLE 1), the samples used to produce dates SRR-1788 and SRR-1789 may have contained antler of more than one period. When first obtained, SRR-1788 was thought to reflect deposition of bones during the Loch Lomond Stadial. This was a seemingly important result; but it now seems spurious (Murray et al. 1993: 4, 9). This lesson has not been carried over consistently into dating of charcoal from archaeological sites in Britain, for a variety of reasons.
Recent Historic Scotland dating programmes for several sites have demonstrated that combining pieces of charcoal to produce carbon for dating will cause serious errors more often than previously supposed, because the survival of charcoal on archaeological sites can occur even when there is no stratigraphic evidence for multiperiodicity. The purpose of this paper is to convince those who want to use charcoal to obtain radiocarbon dates that only single entities should be dated. Probably the same is true for other materials.
The definition of a single entity with which I will defend this proposition is generalized to include other materials. It is any thing, being demonstrably a single part of an organism, in which the absolute chronological relationship between all components forming that part can be established to the nearest calendar year. It is important that the chronological coherence of the supposed single entity can be demonstrated.
A single part of a single entity is always a single entity; but two different parts of what was once a single entity need not, when mixed, be a single entity. Groups of bones from an articulated skeleton form a single entity, but what seems to be a group of bones from a non-articulated skeleton need not be (although a specialist may demonstrate that it is). A mixture of the heart-rings and year-old twigs of the same tree is not a single entity because the relationship in time between the heartwood and the twigs cannot be demonstrated with existing techniques; but either on its own is a single entity. A wattle hurdle is not a single entity, nor are groups of more than one of the individual elements of the hurdle, unless a specialist can demonstrate that they were taken living from the same plant at the same time. Twigs found on a hearth do not form a single entity even if they are of the same species, again unless a specialist can demonstrate otherwise.
Of course, choosing to date a single entity is not enough in itself to guarantee a 14C date useful for dating a human activity. The heartwood of an old oak tree is not much use for dating the creation of, or sealing of, the context in which it survives, unless its date relative to its sapwood can be determined and the interval between sapwood death and incorporation in the context can be estimated accurately.
Housley and colleagues have recently covered many of the problems associated with the dating of bone in their article on late-glacial human re-colonization of Northern Europe (Housley et al. …