Earth and Planetary Science Letters. 138:121-135.
The period some 20-25 kyr just prior to the most recent generally recognized geomagnetic field polarity transition, the Matuyama-to-Brunhes reversal, appears to be marked by significant geomagnetic variability, manifested as pronounced oscillations in intensity. We compare several previously published paleomagnetic records with new high resolution paleomagnetic data obtained from five pelagic marine sites: North Atlantic DSDP Hole 609B; equatorial Atlantic ODP Hole 665A; and western equatorial Pacific ODP Holes 803B, 804C, and 805B. Using standard rock magnetic normalization for all of the samples, as well as a Thellier/Thellier method on the sediments of Hole 804C, we consistently find a decrease in paleointensity (DIP of ) approximately 15 kyr prior to the Matuyama-to-Brunhes transition in the five new records, as well as in the previously published records. Despite sedimentation accumulation rates (SAR) that range from 11 cm/kyr to 1 cm/kyr, these sequences yield paleointensity curves that are broadly similar in form, even at the lowest SARs. The intensity of the pre-reversal tow (DIP1) appears to be of the same magnitude as that of the transition itself (DIP2 of ). In some of the records, a directional excursion to nearly full normal polarity accompanies DIP1 and remains after alternating field (AF) and/or thermal demagnetization, whereas in other records the directional changes vanish with demagnetization and appear to be caused by overprinting. A viscous remanent magnetization (VRM) contribution to NRM was identified in two of the records and, until removed by AF or thermal demagnetization, was found to blur the 'double-DIP' nature of the paleointensity profiles into an apparent single-DIP, and also resulted in an apparent, but erroneous, 'sawtooth'-like post-transitional sudden increase in paleointensity. After appropriate normalization, the magnitude of the post-transitional recovery was much reduced. The magnetic directions of three of the new records after 'cleaning' and adjusting the stable declinations to either 0 (normal) or 180 (reverse), map to VGP positions lying in the Pacific; the directional variations, however, are far less consistent than the intensity variations. The confirmed global existence of this DIP so closely preceding a major reversal invites questions about its relation to the reversal itself. The apparent normal character of this interval can also present problems for magnetostratigraphical interpretations based on coarse or incomplete sampling by mimicking the Brunhes/Matuyama reversal some 15 kyr earlier in the section than its true location.