IODP Proceedings    Volume contents     Search



Time permitting, selected archive-half cores from Hole U1357A were measured on the 2G Enterprises cryogenic magnetometer and demagnetized at 5, 10, 15, and 20 mT, but most were demagnetized at only 20 mT after measurement of natural remanent magnetization (NRM). No oriented discrete samples were taken, but 662 unoriented samples taken for other purposes were analyzed for bulk magnetic susceptibility. The isothermal remanence (IRM) and anhysteretic remanence (ARM) were also measured on 585 of these samples.


Natural remanent magnetization

All measurements of NRM and at the 20 mT step from archive halves from Hole U1357A are shown in Figure F22A and F22B, respectively. The entire archive section was measured at 5 cm intervals, although intervals may contain disturbed sediment, may have no material at all because of extensive whole-round sampling, or may have expanded because of gas. These portions, identified in core photographs, have been deleted in Figure F22B.

The declinations for succeeding cores should be randomly oriented and the inclinations should be near –80° on average at this site. We show core top directions from the NRM and after demagnetization to 20 mT in equal area projections in Figure F23A and F23B, respectively. The NRM data have shallow negative inclinations and even a few positive ones, displaying the distinct characteristics of overprinting by the drill string. The declinations are biased toward the direction of the sample x-direction (see Fig. F9 in the “Methods” chapter for description of sample coordinate system). After demagnetization to 20 mT, the directions are much steeper, with no positive inclinations, but the bias in declinations remains. In fact, the declinations are random around an axis tilted in the direction of the sample x.

A bias in declinations has been attributed to a “radial overprint” stemming from distortion of beds during coring (e.g., Acton et al., 2002) or to measurement of split halves that are not centered with respect to the superconducting quantum interference device coordinate system (e.g., Parker, 2000; Parker and Gee, 2002). In the case of Site U1357, the expected geomagnetic-field direction is nearly vertically up. Coring disturbance generally bends layers downward along the side of the core liner, resulting in the net deflections of the magnetization shown in Figure F24. The average magnetization in such a case would show a bias away from the double lines in the archive half (the x-direction) and not toward them as observed in Figure F23. Parker (2000) states:

“Sometimes cores are split in half along their axes, resulting in hemicylindrical samples. These are normally passed through the magnetometer with the cleaved surface horizontal and centered on the axis. Thus the centre of mass of the sample lies below the magnetometer axis, which spoils the symmetry of the simple approximation: we expect effects such as z magnetization to appear as signal on the x coils.”

It seems quite likely that it is the effect of measuring archive halves off-center in the magnetometer that leads to the bias in directions toward the x-axis. The only way to compensate for this effect is to do a full deconvolution of the remanence, a procedure that is beyond the capability of the shipboard instrument as currently designed. Ideally, one would measure U-channel or discrete samples centered in the measurement region of an instrument.

Rock magnetic measurements

The susceptibility records from the whole-round and split-section logger tracks were very close to the noise level in complicating hole to hole correlations. The Kappabridge instrument is two orders of magnitude more sensitive than the logger tracks and can produce well-constrained data even on these very weak samples. Moreover, bulk remanence measurements like ARM and IRM are well above the noise level of the instruments and potentially offer another means of hole to hole correlation. We therefore measured these rock magnetic properties on selected samples that had been taken for micropaleontological studies. These augment the relatively noisy track data and also allow us to assess the sediments for the possibility of normalizing the intensity data for relative paleointensity estimates.

Unoriented samples were taken every 10 cm starting at Core 318-U1357A-2H and were air, oven, or freeze dried to prevent the foraminifers from dissolving. We measured 662 samples from Core 2H through 8H on the Kappabridge. ARM and IRM was also measured on samples from Cores 4H through 7H.

Samples were initially measured 10 times to assess reproducibility. A quick study of approximately three measurements each was performed for the middle sections of Core 2H. After analyzing the results, we adopted a protocol of five repeat measurements for each sample, starting just before Section 318-U1357A-2H-6. Samples were measured using the SUFAM program.

ARMs were imparted in the samples using a direct-current bias field of 50 µT in an alternating-current peak field of 100 mT. Step-wise IRMs were given to 14 samples in fields up to 1 T. The saturating field was ~200 mT, consistent with a magnetite mineralogy. All samples were given a saturation IRM (sIRM) in 500 mT fields. Bulk remanences were all measured in the cryogenic magnetometer.

The total sample mass was determined on all samples using the Mettler-Toledo balance systems in either the Chemistry or Physical Properties laboratories. Sample bag plus label masses were measured on five randomly selected sample bags and measured five times each. The average “bag mass” was then subtracted from all samples to give the sample mass used in normalizing all susceptibility and rock magnetic data.



There are large shifts in declination trends in the data shown in Figure F22B, many of which span core boundaries. Because core orientation was impossible at this site (virtually at the south magnetic pole), core to core declination is uncorrelated. To examine the long term trends, we assumed that not much secular variation occurred during the missing intervals at coring gaps and we shifted the declination of each succeeding core top to match the last declination of the preceding core. We also shifted the initial declination to 45° to avoid the awkward jump from 360° to 0°. Data after processing are shown in Figure F25A and F24.

There is a geomagnetic secular variation model spanning the last 7000 y (CALS7k.2 of Korte and Constable, 2005). This model, similar to the international geomagnetic reference field model, although lower in resolution and accuracy, allows us to predict the geomagnetic field vector anywhere on Earth at 100 y intervals. The declinations and inclinations that are predicted for 66°S/144°E by the CALS7k.2 model of Korte and Constable (2005) are shown in Figure F25B and F25D, respectively.

A radiocarbon date at 48.72 mbsf from a nearby piston core gives an age of ~2340 y BP (Costa et al., 2007). This provides a loose age constraint in Hole U1357A. At 50 mbsf, there is a long trend in the declination curve (Fig. F25A) from a (relative) declination value of 200° at ~70 mbsf decreasing to ~50° at ~37 mbsf. At about the 2000 y mark in the predicted curve (point labeled “3” in Fig. F25B), positive values of ~5° decrease to negative values of around –15° (equal to 345° in the modular 360° scale). These trends in declination are both westward, although of different amplitudes. Assuming that these two westward declination trends are the same and that there are no large changes in sediment accumulation rate or hiatuses in the record, we can tie hairpins in the observed declination curve to those in the predicted curve. We label the most easily recognized tie points from 1 to 8 in Figure F25A. These tie points are listed in Table T5, and we plot the inferred age-depth relationships in Figure F26. Note that Core 318-U1357-8H has virtually no recovery because of the sediment “blowing up,” resulting in an interval that is difficult to tie to the predicted secular variation curve with confidence.

Several points should be considered when using the age-depth information listed in Table T5. First, there is the problem of bias in the magnetic vector estimations inherent in half-core measurements discussed above. Second, the magnitude of the declination trends observed is far larger than those predicted. However, the CALS7k.2 model is heavily smoothed and frequently underestimates the extent of the secular variation at a particular site. We expect this to be especially true for southerly latitudes for which there are scant data available to constrain the details of the model. Third, the trends in inclination are unimpressive. It is likely that the drill string overprint, which compromises the inclination more than the declination, has obscured features that very well may be recoverable in discrete samples with full demagnetization. These caveats aside, the age model in Table T5 constitutes a testable prediction. If it can be verified at a few horizons, then the rest of the predictions may provide a high-resolution timescale to ~6000 ka for Site U1357.

Rock magnetism

We show a comparison of the sIRM versus susceptibility in Figure F27A and versus ARM in Figure F27B. The former is quite scattered, probably because susceptibility responds to all electronic spins and orbits in the sample (including the diamagnetic effect of silica), whereas sIRM reflects only the remanence carrying magnetic minerals, here apparently exclusively magnetite. The IRM versus ARM plot, however, is quite linear, suggesting that the magnetic mineralogy is very consistent in grain size. The dominant control of the variability is from changes in concentration of the magnetic phases on a decimeter scale. These data bode well for the potential to normalize intensity data for relative paleointensity studies on discrete samples.

In Figure F28 we show a comparison of the NGR data (normalized by density; see also “Stratigraphic correlation and composite section”), the bulk susceptibility data, and the sIRM data versus meters below seafloor. Whereas some features appear in both the susceptibility and the NGR data (e.g., a marked drop at ~65 mbsf), little correlation in detail is apparent. Moreover, because of the more scattered susceptibility data with respect to the IRM, good correlation between the susceptibility and sIRM downcore is not apparent. It is possible that large features like those in the susceptibility data at ~43 mbsf could be correlated from hole to hole, but we have no way to insure that this is the case at this point. The sediments are inhomogeneous at a decimeter scale but quite homogeneous on larger scales, making the prospects for relative paleointensity data bright but hole to hole correlation difficult.