METHODS

Shipboard rock-magnetic methods included a series of anhysteretic remanence (ARM) acquisition and demagnetization measurements on discrete plastic box samples (Shipboard Scientific Party, 1996a). Remanences were measured on a Molspin spinner magnetometer. ARM was imparted using a DTECH bias coil mounted in a Schonstedt GSD-1 alternating-field (AF) demagnetizer, with an effective AF operating limit of 90 mT. Discrete samples, which had already been AF demagnetized, were exposed to a bias field of 0.1 mT in AF-acquisition windows 10 mT wide, centered on steps of 5 mT up to 85 mT to impart partial ARMs (pARM). A final bulk ARM was imparted over a complete 90-mT demagnetization cycle, after which some samples were again demagnetized in 10-mT steps up to 90 mT (demagnetization of ARM, dARM). Shipboard time constraints limited dARM measurements to Sites 991, 992, and 994.

Magnetic susceptibility was measured on unsplit APC cores by a Bartington MS2 susceptibility meter with a MS1/CX80 whole-core sensor attached to the multisensor track (MST), and on discrete samples with a Bartington MS2 system using a MS1B well sensor set to its "0.1 SI" scale.

Shipboard rock-magnetic samples were processed as soon as possible after sampling (within a few days at most), to minimize alteration to the magnetic mineralogy through oxidation. Samples destined for shore-based rock-magnetic analysis were collected and prepared according to one of three regimes.

1. A set of samples from Site 995 that were dedicated for shore-based microbiological and rock-magnetic analysis was taken immediately following sectioning of the core on the catwalk and was stored in an anoxic atmosphere, using Anaerocult packaging. After opening of the packaging, a split of each sample was freeze dried and stored in a sterile plastic pot with an oxygen-free nitrogen headspace. Rock-magnetic specimens were prepared from these splits immediately after opening of the plastic pots, and hysteresis analysis was completed within an hour of exposure of the samples to the air. This set of samples, which should show minimal effects of post-coring oxidation on the magnetic mineralogy, is referred to below as the "A995 set."
2. A second set of samples from Site 995 (originally sampled for magnetostratigraphy) were also subjected to rock-magnetic analysis; however, these samples were not protected from exposure to the air during storage or specimen preparation. These samples are referred to as the "B995 set." Oxidation in samples stored in this way has been noted to result in changes in rock-magnetic parameters. In particular, the ferrimagnetic sulfide greigite has been reported to decompose rapidly in air in some samples, resulting in a decrease in saturation magnetization and a change in coercivity (Snowball and Thompson, 1990; Kalcheva et al., 1990).
3. Samples for shore-based rock-magnetic analysis from Sites 991, 992, 993, and 996 were all stored in Anaerocult packaging. After opening of the packaging, samples were stored at <4°C, and rock-magnetic specimens were prepared and processed as soon as possible after opening of the package (maximum delays were a few days). These samples should also show relatively little influence of post-coring oxidation.

Hysteresis parameters were determined in the La Trobe University PALM laboratory using a Molspin NUVO vibrating sample magnetometer (VSM). Saturation magnetization (J_s), saturation remanence (J_rs), and coercivity (H_c) were determined from hysteresis loops to >800 mT (sufficient to achieve saturation in all samples), after subtraction of the paramagnetic contribution determined from the M/H gradient at >500 mT. Coercivity of remanence (H_cr) was also determined on the VSM, by imparting an IRM at 800 mT, and then applying successively increasing field steps in the opposite directions; remanence was determined by turning the applied field off after each step. Because many of the samples were very weakly magnetized, with J_rs frequently less than 50 µAm² kg^-1, the signal-to-noise ratio of coercivity of remanence was improved by averaging repeated measurements.

Stages in the reduction sequence of the magnetic mineralogy were traced downhole at each of the sites through changes in coercivity and saturation magnetization indices, which indicate changes in domain state (and hence magnetic grain size) and mineralogy. Histograms of pARM and dARM directly indicate changes in the distribution of coercivity of remanence, and can reflect the presence of distinct populations of magnetic phases. Hysteresis parameters can be combined in a Day plot (Day et al., 1977), defining fields corresponding to ferrimagnet domain states. The index D_JH = (J_rs/J_s)/(H_cr/H_c) (Housen and Musgrave, 1996) indicates position on the Day plot, and so allows domain state to be plotted as a function of depth. A single domain (SD) dominated assembly is indicated by values of D_JH > 0.33; D_JH < 0.0125 indicates multidomain (MD) or superparamagnetic (SPM) populations, and intermediate values of D_JH indicate a net pseudosingle domain (PSD) state, or a mix of SD and SPM or MD populations. Domain state reflects grain size: in magnetite, SD grains range between about 50 nm and 1 µm; PSD typically ranges between 0.1 and 10 µm; MD > 10 µm; and SPM < 50 nm.