I think that I shall never see, a sample of a plain old tree,
a tree that's not been carved or painted, calcified, burned or otherwise tainted,
But bones and shells and peat I get; preparing them all day I sweat.
Won't someone please take pity on me, and send in just one plain old tree.
T.A Rafter in Procs 6th international Radiocarbon Conf, Pullman, 1965. P-761.
The quote given above illustrates the preference that radiocarbon daters have had with dating wood and wood charcoal materials. These materials are the most widely used material in radiocarbon dating. Since the inception of the technique in the 1950s, charcoal has been acknowledged as a most reliable material for dating (Libby, 1955). Charcoal and wood possess a high molecular weight, so rigorous pretreatments are able to be implemented without losing large amounts of sample. Its major source of error has been 'inbuilt', or 'presample' age (McFadgen, 1982). This may be the result of growth age; in which the age of dead wood in the centre of the living tree is dated, or storage age; which refers to the amount of time elapsed from the death of the tree to its use by people (McFadgen, 1982). Inbuilt age may result in errors of up to hundreds of years unless only short-lived species, or twigs, are selected for dating. Even then, there remains an inbuilt age effect, albeit somewhat reduced.
In the late 1950's, the issue of presample age was identified as a causal factor in the difficulties encountered in accurately cross-dating the Julian calendar with that of the Yucatan Maya of Lowland Mexico. The Mayan calendar; is centred on a complex Long Count cycle which repeats every 5125.6 solar years (Fedick and Taube, 1992:404) and its decipherment in the the 1950's gave archaeologists an opportunity to develop an accurate chronology for the Mayan culture (Sabloff, 1989). An attempt to correlate the Mayan Long count with secular time resulted in two different interpretations; the Goodman-Thompson-Martinez correlation and the Spinden correlation. The two differed by 260 years at the crucial Julian year of 1539, when one of the main Mayan cycles ended, and radiocarbon dating was used to try and solve which was correct. Samples of Sapodilla wooden beams found spanning inscribed Late Classic Mayan doorways at Tikal were dated by the University of Pennsylvania, and it was found that the Spinden correlation was correct. Shortly after the dates were released, Satterthwaite (1956; in Fedick and Taube, 1992) questioned their validity because the beams had been reduced in size and their exterior bore no resemblance to the original trees. According to him, the dates were erroneous because they failed to date the event of the building of the temple. Subsequent investigations into the dating of the Mayan Long Count focussed on dating exterior bark and samples from smaller vault beams, and it was found that the Thompson correlation was the correct interpretation when inbuilt age was negligible (Coe, 1966:29).
Attaching any error to charcoal dating using old wood is obviously impossible, since it is non-systematic. Any wood selected for dating incorporates an inbuilt age error but errors are significantly reduced if only short-lived species or twigs are selected for dating. In these cases, it is fair to say that identified charcoal will always date the event more closely than older wood. In New Zealand archaeology, radiocarbon daters and archaeologists alike have grappled with issues of inbuilt age and old wood. The problem in this context has been pronounced due to the brevity of prehistoric occupation. Polynesians finally reached New Zealand less than 1000 years ago. A recent re-analysis of New Zealand's corpus of radiocarbon determinations from the Rafter Laboratory at the IGNS has suggested that problems associated with inbuilt age of wood had created spuriously old dates which needed to be culled with the help of a discard protocol. This analysis suggested a prehistory of perhaps as short as 700 yr rather than the 1000+ which had been favoured according to conventional wisdom. McFadgen et al. (1995) reached similar conclusions.