MAGNETIC CHARACTERISTICS AND POLARITY ASSIGNMENTS

Examples of typical magnetic behavior during combined AF and thermal demagnetization of the minicores are diagrammed in Figure F2. Nearly all samples displayed an NRM with positive inclination that is generally significantly steeper than the 20° inclination expected for the present latitude of the Leg 207 sites. A similar downward magnetic overprint is a common feature observed by ODP paleomagnetic studies and is considered to be induced during the downhole coring process (summarized in Acton et al., 2002).

Upon reduction of this downward overprint using 5-mT AF demagnetization, most minicores displayed a relatively shallow inclination, low-temperature magnetic component that was gradually removed by progressive thermal demagnetization to ~200°C. As this low-temperature component was removed, the remanent magnetic directions generally either shifted slightly or else displayed a pronounced "hooklike" swing. As will be elaborated below, these two contrasting behaviors were utilized to assign magnetic polarity. During higher heating steps, the magnetic directions tended to be stable with a gradual loss of intensity—a trend toward the origin on vector plots—which permitted the isolation of a single component of magnetization.

Characteristic magnetization directions and associated variances were computed for each minicore by applying the least-squares three-dimensional line-fit procedure of Kirschvink (1980). The characteristic direction was visually assigned to the set of vectors that appeared to display removal of a single component in equal-area and vector plots during progressive demagnetization. The intensity of characteristic magnetization was computed as the mean intensity of those vectors used in the least-squares fit.

In rare cases, some "fresh" minicores of relatively unoxidized gray facies that were still damp with the original pore water displayed a fascinating and temporary magnetodiagenetic artifact upon heating through an initial thermal demagnetization step of 150° or 200°C (see example in Fig. F2B). After this first heating step, the measured magnetic vector from these moisture-bearing cores (only the gray ones) was often offset from the main trend from initial NRM to the 5-mT AF step and through the higher heating steps. Even though the color of these minicores was generally lighter or bleached after the initial heating, there were no detectable changes in magnetic susceptibility. This behavior was never observed in minicores of the same lithologies that had experienced drying prior to the initial heating step; therefore we suspect that this curious artifact is a low-temperature alteration phenomenon. One possibility is that hydration of some of the reactive iron minerals (maybe iron sulfides?) occurred in these gray sediments to form a goethite-type mineral or similar low-susceptibility compound that acquired a spurious in-laboratory magnetic field upon the initial heating below 150°C but then later dehydrated or had exceeded its Curie point upon heating >200°C.

Diagenetic alternation of iron minerals upon sample storage was also observed in some Leg 207 cores sampled at the ODP repository at Bremen. In just 2 months, many intervals in these refrigerated sediments had already changed surface color from the original greenish gray to grayish tan. Neither the tannish altered minicores from the dark gray zones nor the dried (but still retaining the original gray color) minicores displayed the "anomalous 150°C directions." However, when this phenomenon was explained to some other paleomagnetists, they expressed skepticism about a goethite mineral being involved; therefore, we simply indicate this oddity of low-temperature thermal demagnetization of still-moist, fresh, gray sediments.

Progressive direct magnetic fields up to 1.0 T were applied to typical light gray, dark gray, and reddish chalk to clayey chalk specimens from Hole 1258B (11 samples) and Hole 1259B (2 samples) in order to determine their IRM acquisition patterns (Fig. F3A).

IRM values increase quickly and with relatively weak fields. Most specimens acquire more than 90% of their maximum IRM by 0.3 T and then display a slow increase to 1.0 T. The steep IRM acquisition curve below 0.3 T suggests the presence of a low-coercivity magnetic mineral in the sediments. A few specimens display a more continuous increase up to 1.0 T, which might be caused by a relatively higher concentration of a higher-coercivity mineral.

Subsequent stepwise thermal demagnetization of acquired IRM for all specimens displays a maximum unblocking temperature of ~600°C (Fig. F3B), which indicates that a dominant magnetic mineral is magnetite. A low-blocking-temperature component is also present, as suggested by the inflection point of the IRM demagnetization curves at 200°–350°C for several samples, which may indicate the presence of iron sulfides.

The rock magnetic experiments suggest that magnetite is an important carrier of magnetization in most sediments. We interpret that the characteristic magnetization (and polarity) carried by magnetite is indicative of the original primary magnetization acquired when the sediments were deposited.

Rotary drilling generally produces relative rotation between different sediment blocks within the core liner. In higher latitudes, one would rely on the stratigraphic clustering of positive or negative inclinations of the characteristic directions to assign normal polarity or reversed polarity zones, depending whether the site was known to be north or south of the paleoequator.

The Leg 207 sites are presently between 9° and 10°N latitude. Generalized global plate reconstructions generally indicate that South America has experienced a small amount of northward drift with negligible plate rotation since the Early Cretaceous (Chris Scotese, pers. comm. 2002); therefore, the pre-Oligocene paleolatitudes of the Leg 207 sites were projected to be within a few degrees of the paleoequator, and these reconstructions and independent compilations of South American paleomagnetic data were sometimes contradictory whether the Leg 207 sites would be north or south of that paleoequator. The near-equatorial paleolatitudes and uncertain northern vs. southern hemisphere positions through time implied that it would be inappropriate to assign polarity zones based solely on inclination patterns.

Therefore, we utilized the relative directions of low-temperature overprints (secondary magnetization) vs. the directions of higher-temperature characteristic magnetizations of the sediments to assist in interpreting the polarity of each minicore. The assumptions and procedure are detailed in Shipboard Scientific Party (2004a). This procedure is similar to that used by other paleomagnetists working with Deep Sea Drilling Project (DSDP) and ODP cores (e.g., Shibuya et al., 1991), who have demonstrated that relative orientation of overprints removed during progressive demagnetization can be used for identification of paleomagnetic polarity. The critical assumptions are that (1) an overprint vector of present-day north-directed polarity is superimposed on the original polarity vector, (2) there have not been major (>45°) tectonic rotations of the sites after sediment deposition, and (3) the relative declinations of the overprint vector and the underlying primary polarity vector can be deduced during progressive demagnetization. Ideally during progressive demagnetization, the north-directed vector of secondary magnetization (present-day magnetization) is removed, then the component of characteristic magnetization is resolved. If this characteristic magnetization also has normal polarity, then there would be only a minor change in magnetic declinations. However, if this characteristic magnetization is reversed polarity, then there would be a 180° rotation of declination. We called these two patterns "N-type" and the "hook-type," respectively.

For some intervals, the hook-type demagnetization plots enabled identification of the low-temperature secondary vector and the characteristic vector (e.g., Figs. F2E, F2F, F2G). By assigning the direction of this low-temperature component to be north (0° declination), we reoriented the characteristic vectors of the hook-type samples, which is a method also used by Shibuya et al. (1991). The clustering of these reoriented hook-type vectors toward "south" (180° declination) (Fig. F4) gives us renewed confidence in this procedure. However, isolation and resolution of secondary component vectors was generally not possible because either the last stages of their removal overlapped with onset of decreases in the intensity of the characteristic component or the few low-temperature demagnetization steps were inadequate to enable isolation or other behaviors during demagnetization (e.g., Figs. F2H, F2I, F2J). Of course, it is probable that there are a few erroneous assignments of polarity with this interpretation method. For example, a sample having only a reversed polarity characteristic direction without a significant secondary overprint might display a consistent linear decay upon demagnetization and be mistakenly assigned as N-type. Therefore, we do not trust the polarity interpretation of any single sample, but only the broader patterns. Approximately 80% of the minicore demagnetization behavior could be assigned as N-type or hook-type (sharp or curved), and these types were generally in stratigraphic clusters which we interpret as polarity zones.

A test of this method of polarity interpretation is possible in certain intervals that have biostratigraphic constraints on the possible primary polarity. For example, at all sites, the Paleocene/Eocene boundary occurred within an interval that was characterized by curved or sharp hook-type demagnetization paths. This is consistent with the placement of the Paleocene/Eocene boundary within the reversed polarity Chron C24r. Similarly, the Albian sediments displayed only N-type demagnetization paths, which is consistent with their deposition within normal polarity Superchron C34n.

About 20% of the samples yielded uncertain polarities for a variety of reasons, including very weak magnetizations near the background noise level of the cryogenic magnetometer upon early stages of demagnetization, persistent steep downward-directed overprints, unstable magnetic directions, or ambiguous intermediate behavior between N-type and hook-type demagnetization paths.

Each characteristic direction was assigned a polarity and a qualitative reliability rating based upon its demagnetization behavior: (1) well-defined N or R directions computed from least-squares fits of at least three vectors; (2) normal polarity (NP) or reversed polarity (RP) directions computed from only two vectors or a suite of vectors displaying high dispersion but displaying clear polarity; (3) NPP or RPP samples that did not achieve adequate cleaning during demagnetization but their general N-type or hook-type behavior indicated the polarity; and (4) samples with uncertain N?? or R?? or indeterminate (INT) polarity that are not used to define polarity zones. To reduce the bias of a single observer, each of us made independent assignments of polarity and the selection and rating of characteristic magnetization vectors. In cases where the two paleomagnetists had different opinions, the samples were assigned a lower reliability rating.

When both polarity and characteristic inclination are known for a discrete sample, its magnetic paleolatitude can be computed. The paleolatitude analyses and implications for South American plate reconstructions will be detailed in a separate publication, but the main results are summarized here.

The suites of discrete minicores revealed that the Leg 207 sites were at approximately 15°N latitude during the Albian (95 = 5°, k = 25), then progressively drifted southward to ~4°N during the Campanian–Maastrichtian (3.6°N, 95 = 2.6°, k = 14) and Paleocene (4.4°N, 95 = 2.5°, k = 19) and 1°S of the paleoequator during the early Eocene (1.5°S, 95 = 0.7°, k = 51) and middle Eocene (1.0°S, 95 = 0.7°, k = 82) (Suganuma and Ogg, unpubl. data). After the middle Eocene, there was a cumulative northward drift of ~10° to the present location of the sites. The general trends and projected paleolatitudes do not agree with published generalized global reconstructions (e.g., compilations by Alan Smith [in Gradstein et al., 2004] and Chris Scotese [2001, and www.scotese.com]). However, these paleolatitudes are consistent with some aspects of Apparent Polar Wander Path curves compiled for South America from local paleomagnetic studies combined with projections of selected North American and African paleomagnetic poles (e.g., Beck, 1998, 1999; Randall, 1998). A portion of this apparent disagreement between paleolatitudes computed from regional paleomagnetic data and those implied by global reconstructions may be an offset of the hotspot reference frame (used in global reconstructions) from the dipole reference frame (deduced from paleomagnetic analyses) and/or a "far-sighted" effect noticed in statistics of apparent poles from paleomagnetic data (A. Smith, pers. comm., 2003).