A recently developed paleomagnetic extraction technique successfully permits sample collection from unconsolidated sediments. Traditional measurement and demagnetization procedures can subsequently be performed, and several studies have already shown that soft-sediment samples yield reliable magnetic polarity information. To test the robustness of the extraction technique, as well as the resolution limits of the resultant magnetic vector directions, a total of 238 paleomagnetic samples were taken from unconsolidated strata exposed in an excavation across the San Andreas fault zone in the Carrizo Plain, California. Variable fault offset values for the most recent earthquakes at this locality favored the likelihood of rotational, nonbrittle, deformation. Such deformation could possibly be recorded in the DRM (detrital remanent magnetization) of sediment deposited across the zone. Declination directions of sample groups taken from presumably unrotated strata show general agreement with the most recent secular variation curve constructed for southern California, supporting the assumption that the stratigraphic units sampled do record an accurate DRM. However, significant variations in declinations directions are found between magnetically stable samples within sample groups (i.e., taken as close as 1 cm apart within the same stratigraphic horizon). Probable causes of such declination variations include postdepositional, physical disturbances associated with rupture events and errors generated by the sampling technique.

Inadvertent sample rotation about the axis of the glass holders is one of the most evident sources of error found in this study, although only samples which underwent large rotations (≥60¿) could be identified. Regardless of its origin, the variability of the magnetic declination directions found for the 29 magnetically stable groups forces us to conclude that paleomagnetic results calculated in this study are not precise enough to discern reasonable (≤20¿) amounts of rotation. These results question a similar soft-sediment study upon which this study was modeled which claims to have successfully detected less than 20¿ in rotational deformation using an identical sampling method. Additional studies are needed, preferably not within active fault zones, to better define sources of error for this relatively new soft-sediment sampling technique and to determine its resolution limits for discerning small changes in paleomagnetic vector directions. ¿ American Geophysical Union 1993