The noise level of the 2G cryogenic magnetometer has been investigated by many workers, and estimates of a moment sensitivity of 10-10 Am2 are frequently quoted. There have been numerous suggestions that with very low signals close to the noise level, results are biased toward the "0°" declination, which corresponds to the positive x-component in the ODP coordinate system used on the 2G cryogenic magnetometer.
To investigate this effect, we measured the noise level on each component for 20 min. The plots in Figure F1 give the results for 10 min and reveal a small drift in the x-component, whereas the other components show noise with little drift. In each component there are fluctuations with a period of ~200 s, or 2-3 min. These drifts and noise levels are consistent with the usual performance of these magnetometers and can be mitigated by standard procedures for noise control.
In a second experiment, we measured the tray over a long period of time by moving it very slowly through the sense region. This gave us an estimate of the long-term drift, but it contains the weak signal from the tray. It was evident that the correction for the smaller drift of the y- and z-components is better than the correction for the larger drift of the x-component.
The predominance of the drift in the x-axis and the partial failure of a linear drift correction on this long time scale suggests that the drift correction may indeed be one explanation of the "0°" declination results, if similar effects occur in the short time scale. However, these instrumental effects cannot be a major source of artifacts because they are so weak.
Several empty core liners were measured and produced variable results with some being magnetically quiet, whereas others had magnetic hot spots of the order 10-4 A/m, which is larger than the magnetization of much of the sediment (Fig. F2). Some of these hot spots were caused by attached particles and could be cleaned, but others seemed to be from the intrinsic magnetization of the plastic. Because the core is measured in these liners, no correction similar to the tray correction is possible. Such hot spots appear to give rise primarily to one- or two-point reading anomalies, so the problem was at least mitigated by our standard procedure of removing such anomalous high intensities of order 10-4 A/m.
Traditionally, the paleomagnetic measurement of long cores has been made on archive-half cores after the sediment core has been described. Earlier experiments during Leg 182 suggested that the process of core splitting might introduce a spurious magnetization. As noted above, there has also been the suggestion that an important aspect of the noise encountered may be a radial moment, which would, of course, cancel any measurement of the whole-round cores. Finally, unless the primary noise is in the core liner, the signal to noise ratio should be enhanced by a factor of 2. For these reasons, it was decided to test the effect of the splitting during Leg 189.
The primary interest was in the APC cores, but XCB and RCB cores were also compared. At the first site, whole-round cores from Hole 1168B were measured, demagnetized, and remeasured before they were split to investigate the magnetization acquired in the splitting process. The responses varied, but there was no doubt that an increase in the horizontal component of magnetization was brought about by the splitting of the core, and in some cases, this was not removed by demagnetization after splitting. Figure F3 gives a summary of the effects of splitting for inclination, declination, and intensity when the core was measured with the two halves kept together as a "whole core." Figure F4 shows the changes in inclination before and after the splitting and separation of the core halves. Initially, the strong drilling overprint is seen, but this is largely removed by demagnetization to 20 mT (Fig. F4). After splitting the core, the core was reassembled and measured again as a whole-round core (Fig. F4). This plot shows that the inclination has somewhat shallowed and that two strong negative, or normal, spikes have emerged. Taking the split core apart and measuring the archive half appears to have had the effect of flattening the inclination still further (Fig. F4). Demagnetization to 20 mT removes much of the magnetization that has shallowed the inclination, although differences between this record and the demagnetized result before the splitting remain, particularly at ~100 cm in the section (Fig. F4). The experiments conducted with XCB and RCB cores will be analyzed postcruise. As noted above, the response to splitting the cores varies, but there is, in general, some difference between the demagnetized record before and after splitting.
Between Sites 1170 and 1171, a bit change was made. There was a remarkable difference in the appearance of the intensity records from the holes of the two sites, which is illustrated in Figure F5. It is natural to interpret this as a consequence of the change in bits. In this case, a moment that survives AF demagnetization has been induced in the tops of cores and affects the results to a depth of ~1 m. The drill bits were not measured in detail before use, and because both of them are now ~1 km below the seafloor, we cannot now measure the fields involved. However, it was clear from an earlier cursory examination of the drill bits in the core technician's shop that the second drill bit type was much more magnetic than the first. This result reinforces the earlier suggestion that passage up to the ship may generate hard magnetic moments that cannot be readily demagnetized. The inclination of these same parts of the cores, which were remagnetized, was shallow. To interpret the records, we applied a filter to the data removing all values >10-4 A/m.
Throughout the leg the nonmagnetic APC core barrel assembly was used on alternate cores during APC coring. Preliminary comparisons have been made, and it is already apparent that there were no major differences between cores with the different assemblies. As an example of the small differences seen in many cores, Figure F6 shows the inclination and intensity for Cores 189-1170B-4H, 5H, 6H, and 7H. After AF demagnetization, both sets of core have very similar intensities and both appear to record similar inclinations. Although there were some larger differences in other cores, the overall impression is that the nonmagnetic APC core barrel does not make a substantial difference in these cores. A more detailed analysis will be made postcruise using discrete samples to constrain core orientation.
The results obtained during Leg 189 have demonstrated the following:
A frustrating aspect of this work has been the inconsistency of the results obtained from essentially similar experiments (e.g., the effect of core splitting and the effect of the nonmagnetic cutting shoe). This inconsistency is obviously because we do not yet understand the relevant phenomena well enough to design our experiments correctly. However, the preliminary recommendations made should significantly help.