Archaeomagnetic Dating

In the conventional application of archaeomagnetic research, the data from an archaeomagnetic sample of unknown age are compared to a regional record of secular variation in order to determine the best-fit date range for the feature's last firing event. This is what Sternberg (1997:325) has referred to as a "regional pattern-matching method" of dating because the sample direction is compared to the pattern of regional secular variation, via the reference curve, in order to determine the best date range. Unlike radiocarbon or, in some cases, even tree rings, the data recovered from an archaeomagnetic sample directly refer to a specific cultural event of archaeological interest (Dean 1978). Thus, an archaeomagnetic sample, in theory, should more accurately date the target event than other dating sources (Wolfman 1990a:346).

In the statistical method of sample dating (Sternberg 1982; Sternberg and McGuire 1990), the data from an archaeomagnetic sample are compared to to the mean VGPs of a statistically-created curve. F-tests are calculated at the 5% significance level to identify the mean VGPs that differ from the location of the undated sample. The remaining mean VGPs cannot be statistically distinguished from that of the sample, and their associated date range(s) is assigned to the sample VGP. Because secular variation is a repetitive looping motion through time, it is possible to have multiple date options for a feature. All date options are reported to the archaeologist, then makes a decision as to which best matches the other data from the site (Sternberg 1982, Eighmy 1990).

In contemporaneity studies, or relative dating, sample VGPs are compared to each other to determine whether they are statistically different at the 5% significance level. The same statistical tests (McFadden and Lowes 1981: equations 23 and 25) are used for both statistical dating against a curve and contemporaneity studies. The difference is in the type of question that is addressed.