Figure 6. Effect of the number of samples discarded per paleomagnetic site on a) paleopole position in stereographic projection, shown without outliers discarded and with one outlier discarded, and the b) A95, c) K, and d) S of the acquired paleopole.
4.4 Effect of sample distribution over sites
To average between-site scatter, it is common to collect multiple sites at a locality. The minimum number of sites recommended for determining a paleopole varies between authors (e.g., Meert et al., 2020; Van der Voo, 1990), but as shown by Vaes et al. (2021), poles with larger N tend to plot closer to a best estimate of the time-averaged pole. We performed experiments varying the number of samples per site and the number of sites, with combinations chosen such that the total amount of paleomagnetic samples remained approximately 100. We studied the effect on angular distance to the reference pole, A95, K, and S (Figure 7) of the acquired paleopole. The means were calculated from 1000 runs of the specific number of samples per site and number of sites, where the number of samples and sites was randomly selected from the total population during each run. The paleopole position is strongly influenced by the distribution of samples over sites in case of Mongolia, Turkey, and Antarctica, and lies further from the reference pole if the samples are distributed over fewer sites (Figure 7a). There also is a major effect on the A95. For all datasets, the A95 is by far the smallest when it is constrained by 100 sites of n = 1, and the highest when it is constrained by 15 sites of n=7 (Figure 7b). The effect on K and S is small, except in the dataset from Norway (Figure 7c, d). In that dataset, K is highest and S is lowest for 15 sites with n=7. Nonetheless, the paleopole position is far better determined when taking many sites with few samples.