Testing the Accuracy of Paleointensity Estimates Using Experimental Pottery Assemblages

Paleointensity estimates from archaeological pottery represent a key source of information on Holocene variations in geomagnetic field intensity. Yet, pottery rarely exhibits ideal single-domain behavior that fully satisfies the theoretical assumptions underlying absolute paleointensity methods. Using two different experimental procedures, we empirically quantified the reliability, accuracy and precision of pottery archaeointensity determinations on 30 ceramic vessels produced from six geological clay types and fired in two traditional kilns. The ambient geomagnetic field inside the kilns was measured before and after the experiments, and temperatures were recorded throughout the process. Hysteresis, first-order reversal curves and thermomagnetic susceptibility curves indicate that the experimental ceramics span a range of magnetic properties representative of typical archaeological pottery. A total of 317 specimens were analyzed using the Thellier-IZZI method in three different laboratory fields, all corrected for anisotropy and cooling rate effects. The results confirmed that applying a laboratory field significantly higher than the ancient field can bias archaeointensity estimates. As non-SD behavior can produce multiple near-linear segments in the Arai plot, introducing large uncertainties in the estimates, we propose a new criterion to limit acceptable uncertainty levels. Nineteen samples, comprising 66 specimens, met the acceptance criteria currently applied with the Triaxe method, yielding indistinguishable results from those obtained with the Thellier-IZZI method. Regardless of the method used, most but not all samples reproduced the expected field value within 5%. Thus, to ensure this accuracy at the 95% confidence level, we recommend averaging data from at least three pottery samples meeting strict criteria per context.