Archive halves of core sections were measured using the shipboard cryogenic pass-through magnetometer. A few cores from the base of the advanced piston cored (APC) section at Site 907 were not measured because of excessive drilling-related deformation, and, at Hole 985A, measurements were discontinued below 180 meters below seafloor (mbsf) because of excessive drilling-related deformation in the extended core barrel section. Stepwise alternating field (AF) demagnetization was not feasible during the cruise, and most core sections were measured on board ship at a single demagnetization step (generally 25 mT). The choice of peak AF field was based on stepwise demagnetization of a few core sections and a handful of discrete samples.
The shipboard pass-through magnetometer data are perturbed by (1) magnetic overprints derived from the drill string, (2) drilling- and dropstone-related core deformation, and (3) diagenetic (secondary) remanence acquisition associated with formation of iron sulfides, particularly in ash-rich layers. Drilling-related deformation and dropstones can be avoided by discrete sampling. In addition, a complete demagnetization sequence can be performed on discrete samples to "ground truth" the shipboard magnetic stratigraphy.
Discrete samples were collected during the cruise in the standard 7-cm3 plastic boxes and measured at laboratories at Gif-sur-Yvette and at the University of Florida. Natural remanent magnetization was measured before demagnetization and during stepwise AF demagnetization using a peak field increment of 5 mT in the 5-70 mT range, or until the magnetization intensity fell below magnetometer noise level. Orthogonal projections of AF demagnetization data indicated that a characteristic magnetization component could usually be resolved at peak alternating fields above ~20 mT. A lower coercivity component was observed in most samples, particularly those with a reverse polarity characteristic component. The low-coercivity component is oriented steeply downward and is probably a viscous remanent magnetization imposed by the drill-string assembly. The characteristic (higher coercivity) magnetization component was picked visually from orthogonal projections of demagnetization data, and its direction was computed using the standard least-squares technique (Kirschvink, 1980).
Orthogonal projections of incremental AF demagnetization of discrete samples indicate that a characteristic magnetization component can be resolved at peak fields above ~20 mT (Fig. 2). The drill string-related magnetic overprint is apparent as a steep downward low-coercivity component in most samples, particularly those in which the characteristic component has reverse polarity.
For the upper 110 m at Site 907, the shipboard pass-through inclination data (after AF demagnetization at peak fields of 25 mT) indicate clearly defined polarity zones (Fig. 3), although the records are perturbed by core deformation, abundant ash layers, and associated secondary magnetizations produced by authigenic iron sulfides. Core deformation at Site 907 is exacerbated by abundant dropstones. The component inclinations determined from discrete samples at Hole 907B generally ratify the polarity stratigraphy from shipboard data (Fig. 3); however, comparison of the shipboard data (including Hole 907A from Leg 151) indicate significant discrepancies among the three holes in the placement of some polarity zone boundaries. These discrepancies are not accounted for when depths in mbsf are converted to meters composite depth (mcd) using the shipboard calculations. In addition, high-resolution IRD counts indicated that some glacial periods were missing and that others were repeated when using the shipboard composite section. By creating a new composite depth scale, we removed several instances of "double-coring" (overlap between cores), which resulted from the shipboard calculations.
The shipboard composite depth scale for Site 907 was revised by aligning the complementary gamma-ray attenuation porosity evaluator (GRAPE) and magnetic susceptibility data from each hole, with emphasis given to the matches between Holes 907A and 907C. Data from Hole 907B were used to resolve uncertainties in the correlations between Holes 907A and 907C. The methods used were similar to those used to construct composite depth sections during Leg 138 (Hagelberg et al., 1992) and during subsequent ODP legs, including Leg 162 (see Jansen, Raymo, Blum, et al., 1996). Drilling-related deformation was apparent in some parts of the GRAPE and magnetic susceptibility records and confirmed by comparison with the shipboard core descriptions. No differential stretching or squeezing was introduced within individual cores. After construction of the composite depth section, a spliced record representative of the multiply cored sedimentary sequence was assembled.
Most ash layers in the upper 100 mbsf of Hole 907A are associated with reduced GRAPE bulk density and increased magnetic susceptibility values relative to the mean values for these sediments (Rack et al., 1996). In the deeper parts of Hole 907A, below ~100 mbsf, the situation is different (ashes have higher bulk density and higher magnetic susceptibility relative to mean values) because of the relative dominance of low-density biosiliceous microfossils whose presence largely controls the measured sediment physical and geotechnical properties. The bulk of the ash layers in Hole 907A have been "ground truthed" by discrete sampling and petrologic analyses reported by Lacasse et al. (1996) and Werner et al. (1996). These authors used the above-mentioned, distinctive "signature" of these tephra layers to rapidly locate thin ash layers and intervals of dispersed or disseminated ash. The peaks in magnetic susceptibility associated with both discrete and dispersed ash layers are linked to an abundance of magnetite-bearing crystals, basaltic components (sideromelane and tachylite), or shards coated by iron sulfide. High volcanic glass concentrations were found in the sand and silt component of dispersed ash zones. The ash zones were difficult to locate visually, but they could be identified using the magnetic susceptibility record (Lacasse et al., 1996) and hence correlated from Hole 907A with similar intervals in Holes 907B and 907C (Fig. 4).
Recalculated composite depths for Site 907 and the splice for the composite section are given in Appendix A and Appendix B, respectively. Examples of these correlations are given in Figure 4 and Table 1. In general, the new composite depth (mcd) calculations are consistent for polarity zone boundaries from the three holes. The apparent position of the Brunhes/Matuyama boundary in Hole 907A (Fig. 3, Table 2) remains inconsistent, however, and the base of the Olduvai/Reunion interval at Hole 907A is also anomalous (Fig. 3, Fig. 4B, Table 2).
The interpreted position of the Brunhes/Matuyama boundary in Hole 907A coincides with a 10-cm-thick ash layer, designated as Layer G by Lacasse et al. (1996). A large dropstone was recovered just above this ash layer, at interval 162-907A-2H-6, 124-129 cm (Shipboard Scientific Party, 1995). The large dropstone in this core very likely disrupted both the sedimentary structure and the magnetic measurements across this interval, possibly resulting in a downward shift of the apparent (interpreted) position of the Brunhes/Matuyama boundary in Hole 907A. The same ash layer appears, both without a large dropstone in Hole 907B (interval 162-907B-3H-2, 65-77 cm; 14.35-14.47 mbsf) and with a large overlying dropstone in Hole 907C (interval 162-907C-2H-6, 145 cm, through 2H-7, 6 cm; 15.05-15.16 mbsf). In both cases, the ash layer is located below the interpreted Brunhes/Matuyama boundary. Therefore, it seems likely that the true position of the Brunhes/Matuyama boundary in Hole 907A should be placed at ~15 mbsf, which would be consistent with the composite depths determined for this feature in the other two holes (Table 2).
The position of the base of the Olduvai Subchron in Hole 907A is likely located within the core break between Cores 162-907A-4H and 5H, where a significant gap in core recovery is suggested by the new composite depth scale (see Fig. 4A, Fig. 4B). The normal polarity interval at the top of Core 162-907A-5H would then be assigned to the Reunion Subchron. The base of this normal polarity interval is located about the same distance above an ash layer as the inferred position of the Reunion Subchron in Core 162-907B-6H, where the ash is located at 43.5 mbsf (Fig. 4B). The small normal polarity interval at 41.67-41.83 mbsf in Hole 907C (Core 162-907C-5H) should correlate in some way to the base of Core 162-907B-5H, which suggests that the bulk of the Reunion should be located within the lower part of Core 162-907C-5H.
The 50-60 mbsf interval at Hole 907B is a data gap (Fig. 3) resulting from the high degree of drilling-related deformation in Core 162-907B-7H. The top and base of the Gilbert Chron are consistent at the three holes, but the subchrons within the Gilbert Chron are not easily correlated among them. Similarly, below the Gilbert Chron, the polarity pattern is not easily correlated among the three holes. We attribute these inconsistencies to disruption of the polarity zone pattern by drilling-related remagnetization and core deformation.
For the 110-185 mbsf interval at Site 907, the correlation among the three holes is fairly straightforward (Fig. 5). Below 185 mbsf, the record is perturbed by core deformation at the base of the APC section.
As at Site 907, orthogonal projections of incremental AF demagnetization for Site 985 indicate that a characteristic magnetization component can be defined, after removal of a steep downward (drill string-related) magnetic overprint (Fig. 6).
Shipboard pass-through magnetometer data (after demagnetization of peak fields of 25 mT) indicate a well-defined magnetic stratigraphy to the base of the Gauss Chron (95 mbsf) (Fig. 7). The records are perturbed by core deformation exacerbated by numerous dropstones and by ash layers with localized iron sulfide diagenesis. Authigenic iron sulfide formation affects the primary magnetization because of associated detrital magnetite dissolution and the presence of secondary chemical remanences associated with some iron sulfide compositions. Deformation in Cores 162-985A-3H and 162-985B-3H resulted in poor definition of the Brunhes/Matuyama boundary at both holes. Remagnetization results in poor definition of the Gilbert Chron. Below the Gilbert Chron, the record may be affected by increased core deformation at the base of the APC section. Component inclinations determined from discrete samples from Hole 985A ratify the shipboard magnetic stratigraphy as far down as the base of the Gauss Chron (Fig. 7). The component inclinations of discrete samples do not clearly define the normal polarity intervals within the Gilbert Chron, although the boundaries of the Gilbert can be located. Below the Gilbert Chron, discrete sample inclinations appear to define several polarity zones although, in this part of the section, the characteristic magnetization component is less well defined because of incomplete removal of secondary magnetization components.