Secular Variation Curves

Once the paleodirections of enough independently dated archaeological features are determined, they can be used to compile a secular variation record for a particular region. This is often depicted as either the path of movement of the geomagnetic pole with respect to the specific archaeomagnetic region within which the features were magnetized (i.e., a polar plot), or as changes in D and I through time with respect to a single location within that region (Batt 1997; Sternberg 1982; Sternberg and McGuire 1990a). These archaeomagnetic reference curves are usually presented as bands with varying widths that reflect the uncertainty associated with each segment of the curve. The uncertainty can be decreased, and the curve refined, by increasing the number of high precision, well-dated samples used to generate each segment of the curve. Thus, curve construction and refinement is an ongoing process. As the dataset of independently dated sample VGPs increases, the precision in estimating points along the VGP or D and I curve(s) will improve and documentation of the curve(s) can be extended to earlier or later periods.

In North America, archaeomagnetic reference curves are typically constructed via the moving-window approach (Sternberg 1982). This approach statistically averages directional data over incremental windows of time to produce a series of averaged VGPs that can be connected to form a curve. Each averaged, or mean, VGP is calculated from sample VGPs that have been weighted by both their respective precision parameters and by the extent to which their individual date ranges overlap the averaging window. For instance, a sample VGP with the date range AD 700-775 would fully overlap the averaging window for the AD 700-800 mean VGP, whereas a sample VGP with the date range of AD 675-750 would only partially overlap the window. In this way, the uncertainty in both the dating and magnetic direction of the data can be accounted for in the derived reference curve.