Paleomagnetic investigations have been conducted on over 300 sediment samples from sub-bottom depths of between 125 and 1740 meters, and ranging in age from late Pleistocene to Early Cretaceous. Although NRM intensities show a considerable total variation, the section can be divided into clearly defined intervals within which the range of intensities is rather narrow. The transitions from one intensity interval to another generally coincide with lithological changes, particularly color changes, and the most marked intensity change corresponds almost exactly with a hiatus between Cenomanian and younger sediments. Most samples have been subjected to progressive a.f. demagnetization and the majority reached good stable end-points at fields of between 100 and 300 Oe. There is some evidence that the stable remanence is carried by both magnetite and hematite, and that the magnetite and hematite magnetizations are in essentially the same direction. The deviation of the hole from vertical below 1000 meters introduced an uncertainty into the remanence inclinations in unbedded samples. However, by assuming the hole deviation azimuth to be constant with depth and using it as a fixed reference, it was possible using clearly bedded samples to demonstrate that their remanence declinations were very similar, and to determine an average remanence declination with respect to the hole deviation azimuth. It was then possible to apply an appropriate correction to the inclination of unbedded samples. Down to the upper Santonian, approximately equal numbers of normal and reversed stable magnetizations are present; below this, virtually all stable magnetizations are normal. The sampling density does not allow the recognition of a detailed polarity sequence, but the results do not support the existence of an appreciable reversed interval within the Aptian corresponding to marine magnetic anomaly MO, and they suggest that the long Cretaceous normal interval extends back at least as far as the middle Barremian. Stable inclinations show some scatter but, when averaged over à dozen or so samples, are consistently shallower than would be expected from continental reconstructions for the periods of time represented, and it is tentatively suggested that this may be due to sediment compaction.