Agria 39. (Az Egri Múzeum Évkönyve - Annales Musei Agriensis, 2003)

Domboróczki László: Radiokarbon adatok Heves megye újkőkori régészeti lelőhelyeiről

two time intervals. The calendar date is more likely to be within the 2a domain where the measured specimens have a 94.5% chance of falling within its range than in the la domain where the probability falls to 68.3%. As the diagrams show, we are fortunate that with most specimens the range of the 2a domain is sufficiently narrow to allow us to work out dates with some accuracy. In other cases, however, we have to make conclusions based on intervals which are on the wide side. Wide intervals are due primarily to the irregularities in the calibrated graph caused by having to calibrate the whole interval of the standard deviancy. If the specimen is representative it isn't the specimen itself which is causing the problem. Using larger amounts of specimen does not usually make it possible to narrow down the wider intervals. It was for this reason that using large quantities of material was perhaps wasteful and that it would perhaps have been better to give one part of the specimen to another laboratory for a control sample to be taken. Accuracy may perhaps have been improved if we had taken the specimens from one single source. The more bones you use in one sample the more likely you are of getting a greater time spread, as calculating the average age causes a modification in the age of the sample (as with archaeological objects the older finds and more recent additions can also find their way into the sample). Whilst in future we will be using human burial remains to get the most accurate datings, for the time being we will have to be satisfied with the data we have. For the sake of accuracy we will primarily be using the 2a probability domain, although la readings will be used for instructive and orientational purposes. It would be tempting to treat the la readings as if they were indeed more accurate or to give their mean values precidence, but it would mean the probability of our aforementioned results would drop to below 68.3% leading to conclusions which are just as likely to be false as true. We would like to avoid criticism of this find wherever possible. Some of the readings within the 2a probability domain have been given letters. Whilst these letters were originally intended to show the degree to which individual dates can be of use, they have, as can be seen, proved useful in the grouping of earlier and later data. We have marked the earlier data with the letter E and the more recent data with the letter L. This data is particularly interesting where the interval is narrow enough to produce reliable dates. If marked with the letter U the data is reasonably accurate, with an interval of no more than two hundred years. If marked with R, however, the date is redundant and in all likelihood a bad piece of data. Those pieces of data which have not been marked are difficult to evaluate, in these cases one might even consider the la values. Finally, one should mention the al3C values as well, which all fall between -19.34 and-21.34.

19 19 The Ô13C value is a correction factor reflecting the degree of isotope fractionalisation taking place during the metabolic processes. Taking this into account increases the accuracy of the measurements. In the case of bone and collagen its value varies between -18 and -24 thousandths. This value tells us how much protein originating from water creatures (eg. shellfish, fish) there was in the diet as compared with protein derived from other sources (eg. seeds, bread, beef, milk). If for example the diet was completely made up of sea fish then the human bone ô 1

3C value would be about -12±1%, in the case of freshwater fish the value is generally lower, around -20% because of the carbon content in the groundwater (WHITTLE, Alasdair ET AL. 2002. 75.) 11