PALEOMAGNETISM

Susceptibility

Susceptibility data from the whole cores provides important information on the content of iron-titanium oxides in the mafic and ultramafic rocks recovered from Site 1268. Magnetite has a volume susceptibility () of ~3 SI (Heider et al., 1996), which is one to three orders of magnitude higher than other ferrimagnetic minerals likely to occur in gabbroic and ultramafic rocks (e.g., pyrrhotite, hematite, and ilmenite). Iron-bearing silicate minerals will also contribute to the susceptibility signal. This paramagnetic contribution is well approximated by the total iron content (1 wt% FeO corresponds to 6 x 10^–5 SI) (Collinson, 1983). Thus, whole-core susceptibilities higher than ~50 x 10^–5 SI (corresponding to the highest measured FeO content of ~9 wt%) primarily reflect the concentration of magnetite and may be used as a semiquantitative estimate of its abundance.

Susceptibility was measured at 2.5-cm intervals with a Bartington MS2C sensor (coil diameter = 80 mm) on the MST (see "Magnetic Susceptibility" in "Physical Properties"). Two possible complications in the interpretation of the susceptibility data should be mentioned. First, the susceptibility data from the MST (x 10^–5 SI) represent the volume susceptibility for a whole core with a diameter of 6.6 cm. The diameter of cores from Site 1268 is invariably smaller (typically ~5.5–6.0 cm), and thus the true susceptibility will be underestimated by as much as 40%. In addition, readings from the Bartington sensor of >0.1 SI are clipped, so that the most significant digit of the susceptibility value is not recorded. Because this clipping effect has been noted for some core pieces, care should be taken when interpreting the width of individual high-susceptibility zones as well as in the calculation of average susceptibility or magnetite content.

Despite these complications, the susceptibility data from Site 1268 (Fig. F92) reveal large variations in magnetite content that are correlated to lithology and type/degree of alteration. Gabbroic rocks and peridotites that have experienced talc alteration typically have susceptibilities <100 x 10^–5 SI, indicating that only trace amounts of magnetite (~0.1%–0.2%) are present in addition to the contribution from paramagnetic silicates. In contrast, serpentinized peridotites from the top of lithologic Unit I and the lower part of lithologic Unit III (see "Lithology and Stratigraphy" in "Igneous and Mantle Petrology") have susceptibility values near and occasionally above 0.1 SI. These values reflect magnetite concentrations of ~3%.

Remanent Magnetization

The natural remanent magnetization (NRM) intensity of archive halves spans four orders of magnitude (Fig. F92) (0.001–10 A/m). The general pattern of NRM intensity shows a clear positive correlation with the whole-core magnetic susceptibility. Gabbroic rocks from the lower one-third of Hole 1268A (115 mbsf to the bottom of the hole) have NRM intensities in the range of 0.01–1.0 A/m. The overlying peridotite sequence (90–104 mbsf) has a distinct NRM signature, with intensities of up to 15 A/m. Some peridotites with lower susceptibility values, presumably reflecting talc alteration, have NRM intensities similar to those of the gabbros. The inclinations of the NRM also show a similar covariance with rock type. Gabbros have distinct shallow inclinations (typically 10°–20°), whereas peridotites have very steep positive inclinations prior to demagnetization. This difference is most readily attributed to a downward-directed drilling-induced magnetization that overprinted the remanence to a variable degree as a function of the magnetic stability of the different rocks. This substantial drilling overprint precludes obtaining any reliable estimate of the in situ magnetization intensity of most of the peridotites.

All archive halves were subjected to stepwise alternating-field (AF) demagnetization in an effort to isolate the characteristic remanence (ChRM) direction. Demagnetization was typically carried out up to 80 mT, the maximum field available for long-core measurements. A treatment of 20–30 mT removes much of the low-stability (presumably drilling related) overprint in the peridotites. However, this low-stability component usually represents more than half, and commonly >90%, of the remanence (Fig. F92). For example, dunites from lithologic Unit III have ~1% remanence remaining at 20 mT and ~0.1% after treatment at 60 mT. In contrast, the remanence of the gabbros was very stable, with >50% of NRM remaining after 80-mT demagnetization.

The overall effectiveness of AF cleaning in recovering the ChRM direction for the archive halves can be evaluated in Figure F92, where remanent inclinations are plotted at the NRM and the 20-mT demagnetization steps. Gabbros (primarily in lithologic Subunits IVA and IVC; see "Lithology and Stratigraphy" in "Igneous and Mantle Petrology") show only minor directional changes as a result of their high magnetic stability and cluster at shallow positive inclinations (10°–20°). Peridotites show more scattered inclinations as a result of the varying degrees of the low-stability overprint. The ratio of the remanence remaining at 20 mT to the original NRM intensity ([J_20mT/J_NRM] x 100) provides a measure of the likely significance of the remanence direction at the highest field steps. Directions representing less than ~10% of the initial remanence should be interpreted with caution. For samples with J_20mT/J_NRM > 10% (solid circles in right panel of Fig. F92), the inclination after 20-mT treatment is highly variable (–70° to +90°) but includes some values closer to the expected time-averaged normal polarity direction (000°/28°) at the site (e.g., lithologic Unit II and base of Unit I). This suggests that a geologically significant direction may have been isolated for these samples. In other intervals, particularly in the lower portion of Unit III, inclinations remain steep throughout demagnetization and the remanence apparently consists almost entirely of a drilling-induced magnetization. AF demagnetization of discrete samples was typically more successful in isolating a ChRM direction. This difference in demagnetization behavior as well as some other complicating factors in interpreting the archive-half data are discussed below.

The characteristic inclinations in the gabbros are significantly different from the expected inclinations of the geocentric axial dipole at this latitude (28°). With the exception of three pieces (Sections 209-1268A-22R-2 [Piece 5], 27R-2 [Piece 14], and 28R-1 [Piece 10]), all gabbroic samples yield shallow positive inclinations suggestive of normal polarity. The three pieces noted above were reexamined and their remanence directions remeasured. All three pieces preserved red markings on the piece bottom and only one (Section 209-1268A-28R-1 [Piece 10]) was found to have been measured in an inverted position. The remaining two pieces apparently were archived and measured in the proper orientation.

Stepwise AF (45 samples) and thermal (5 samples) demagnetization of discrete samples (Table T9) was carried out in order to provide accurate directional information for tectonic restoration and core orientation purposes. Figure F93 illustrates the characteristic demagnetization behavior of gabbros and peridotites. The majority of peridotites exhibit a low-stability (drilling) component of variable amplitude. In contrast to observations on long-core measurements, AF demagnetization at relatively low fields (<15 mT and often lower) (e.g., Fig. F93A, F93B) was sufficient to remove this low-stability component. A small number of peridotites have no appreciable low-stability overprint (e.g., Fig. F93C). The gabbroic samples yielded stable single-component trends with no signal of drilling overprint (Fig. F93D, F93E).

The median destructive field (MDF; the AF treatment necessary to reduce the remanence to 50% of its initial value) provides an estimate of the stability of the remanence. Because many samples have multiple remanence components, the remanence decay (and MDF) is calculated from the sum of the vector differences between successive treatment steps. Gabbroic samples have uniformly high MDF values (78–97 mT) (Fig. F94A). Peridotite samples have a broad range of MDF values. Nearly half have MDF values <5 mT. Much higher values (up to 140 mT) are apparently associated with higher degrees of talc alteration. Four of the five samples with MDF > 70 mT have a talc alteration index of 1.5 or 2.0 (see "Metamorphic Petrology"), although there is a considerable range of MDF values at a given talc alteration level.

Thermal treatment in both types of rock yields directional results that agree with AF data. Maximum unblocking temperatures range between 570° and 590°C, typical of magnetite (possibly maghemite). Two gabbroic samples have discrete high unblocking temperatures (Fig. F94B). Together with the high stability to AF, these data suggest that the remanence in the gabbros is carried by very fine (submicron) magnetite. Peridotite samples exhibit either distributed unblocking temperatures, consistent with the presence of a range of magnetic grain sizes, or discrete unblocking temperature spectra similar to the gabbroic samples. This latter type of behavior appears to be related to higher degrees of talc alteration.

Results from the discrete samples corroborate the difference in inclination between gabbroic and ultramafic rocks at Site 1268. All but two samples yielded reliable results, as shown in Table T9. Using the method of McFadden and Reid (1982) for azimuthally unoriented cores, the average inclination for the gabbroic samples is 15° (+7°/–8° asymmetric 95% confidence limits; = 39.4; N = 11). The average inclination for the serpentinized peridotite samples (40° +5°/–9°; = 12.4; N = 38) is significantly steeper.

The anisotropy of magnetic susceptibility was measured on all minicore samples with the Kappabridge KLY-2 using the standard 15-position measuring scheme (Table T10). The degree of anisotropy (P, where P = maximum/minimum eigenvalue of the susceptibility tensor) ranges 1.02–1.40. The higher degrees of anisotropy correspond to samples of peridotites with a lower talc alteration index. In these cases the magnetic fabric has a marked foliation. Samples subject to talc alteration have low degrees of anisotropy, similar to those of the gabbros. However, this may simply reflect the combined effect of the low bulk susceptibility of the talc-altered samples and the higher level of noise of these measurements compared to shore-based laboratory environments.

Although comparison of the archive-half and discrete sample data shows general agreement, some significant discrepancies were observed. We regard the discrete sample demagnetization data as more accurate for the following reasons:

1. With the exception of five samples that were measured using the Molspin spinner magnetometer, all of the discrete samples were measured in three positions such that each component of the remanence was measured by all three superconducting quantum interference device (SQUID) sensors. This procedure ameliorates any systematic bias from the discrete sample holder in the 2G magnetometer.
2. Each sample was subjected to double demagnetization (in the DTech D-2000 AF demagnetizer) at fields >30 mT to average any bias field in the demagnetizing coil.
3. The off-line demagnetizer allowed more complete demagnetization to maximum fields of 200 mT. More than 90% of the discrete samples have demagnetization trajectories that trend to the origin, suggesting that these procedures were effective.

Long-core continuous AF demagnetization revealed some features, such as demagnetization trends not directed toward the origin or steep inclinations at high demagnetization AF fields, that were not common in off-line AF treatment of discrete samples. In a few cases, extreme directional changes at a centimeter scale led to apparent polarity reversals and multicomponent remanence. For example, Section 209-1268A-21R-1 (Piece 1) showed directional changes from shallow, northerly directions at the top of the piece to nearly antipodal directions at the base of the piece (Fig. F95A). Because adjacent pieces might influence the magnetic data, this piece was remeasured alone and header and trailer data were also recorded. The same trends in direction are seen as in the routine archive-half measurements, indicating that there was little influence from the adjacent core pieces. A transition from north to south is observed at 12 cm, with an associated increase in inclination (Fig. F95A). However, a nearly identical transition from south to north is observed at 29 cm, where no core material is present. These directional changes may be attributed to the order of magnitude higher magnetization of the pyroxene-rich lower 6 cm of the piece. The response functions of the x- and y-SQUID sensors have negative sidelobes that are ~15% of the peak response. When the high-magnetization material lies within these sidelobes of the response function, spurious directional and intensity data are produced. This effect is most serious when the magnetization varies by an order of magnitude over few centimeters, but all the archive-half directional data will be affected to varying degrees by intensity changes. The effects of neighboring pieces with different magnetization directions have been linked to the presence of spurious multicomponent magnetizations in the archive-half data. Finally, measurement of half cores rather than whole cores is expected to introduce a further bias into the continuous measurements (Parker and Gee, 2002).

Given the potential complications with the archive-half data, we do not necessarily expect close agreement between the continuous and discrete data. Nonetheless, careful piece-by-piece examination of the archive-half data in many cases is very similar to the discrete sample measurements (Tables T9, T11). Because half cores were demagnetized at five to eight different peak fields, vector endpoint diagrams could be used to assess the quality of the data. Principal component directions were calculated for the centers of apparently homogeneously magnetized core pieces. Directions were calculated only for pieces where the low-stability drilling overprint was absent (talc-altered samples or gabbros) or could be mostly removed. The resulting principal components typically represent 10%–30% of the NRM for ultramafic samples and 50%–100% for talc-altered samples or gabbros (Table T11). These data may be combined with the discrete sample directions to give orientations for ~150 pieces.

The paleomagnetic directions estimated from discrete samples indicate significant differences between the inclination of the gabbroic and peridotite samples. Whereas the latter could be consistent with the time-averaged geomagnetic field (expected inclination = 28°), the remanent inclination in the gabbros is significantly shallower (15°). One possible explanation is that the gabbroic rocks do not average secular variation and therefore do not necessarily have an inclination corresponding to that expected from a geocentric axial dipole. However, the range of inclinations (6°–36°) is larger than would be expected solely from orientation errors, suggesting that the remanence directions from the gabbros likely represent some degree of time averaging.

Evaluating whether a particular paleomagnetic data set has adequately averaged secular variation is a key element for subsequent tectonic interpretation of the mean remanence directions. This assessment involves comparing the dispersion of the paleomagnetic directions with the expected dispersion at the site from models of paleosecular variation (e.g., Butler, 1992). For this analysis to be valid, each paleomagnetic direction should represent an independent sampling of the field; various criteria have been used to identify such independent estimates (e.g., baked sediments between lava flows and grouping adjacent stratigraphic samples with statistically indistinguishable directions). For the gabbro samples from Site 1268, the number of independent samples of the field is uncertain. Comparison of the within-site dispersion at Site 1268 with that recorded by the ~1500-m gabbroic section from Hole 735B (Shipboard Scientific Party, 1999), which might reasonably be expected to have recorded a substantial duration of secular variation, provides a qualitative indication of the possible time averaging at Site 1268. For Hole 735B the observed directional dispersion parameter ( = 59; N = 339; each sample is assumed to be an independent sampling of the field) is 1.6 times the dispersion expected for the site ( = 37). For this analysis, we have used the paleosecular variation model of McFadden et al. (1991) and calculated the expected dispersion as the average of the dispersions assuming either virtual geomagnetic poles or site directions are Fisher distributed (Cox, 1970). For the gabbros from Site 1268, the observed dispersion parameter is 39.4 and the expected value is 25, yielding an identical ratio (1.6) as that for Hole 735B. This analysis does not prove that secular variation has been adequately averaged at Site 1268 (both sites have less dispersion, i.e., higher values, than predicted) but does show that the amount of dispersion is comparable to a larger gabbroic section that could be composed of many intrusions (Dick et al., 2000).

If it is assumed that the geomagnetic secular variation has been adequately averaged upon cooling of the gabbros, this suggests that the rocks were tilted after cooling. Restoration of the rocks to their pretilt position has, however, no unique solution. Unless other independent information on core orientation is available, the indeterminate azimuth of cores precludes an estimation of the strike of the rotation axis along which tilting was produced. Because block rotations might reasonably be expected to occur about approximately ridge-parallel rotation axes, the angle between the rotation axis and the paleomagnetic reference direction (360°/28°) will be small (~30°). The amount of block rotation required to untilt the observed paleomagnetic vector to the reference value, therefore, will strongly vary depending on the trend and plunge of the rotation axis (see "Orientation of Structures" in "Structural Geology").

The mean inclination observed in the peridotites (40° +5°/–9°) is statistically distinct from the expected dipole inclination at the site. However, the steeper mean inclination may in part reflect the effect of the drilling component that partially overlaps the characteristic remanence. Indeed, the most talc-altered ultramafic samples (talc index = 2) (see "Metamorphic Petrology") have smaller drilling overprints and a mean inclination (36° +12°/–14°; N = 9) that is statistically indistinguishable from the dipole value at the 95% confidence level. Although the number of samples is small, the higher magnetic stability and smaller drilling overprint in these talc-altered peridotites suggests that they may provide a more accurate estimate of the characteristic magnetization. The mean inclination of the talc-altered peridotites does not require rotation since the magnetization of the peridotites was acquired, which in these rocks is most likely related to hydrothermal alteration and serpentinization. The remanence of the gabbros might reflect tectonic rotations at temperatures below the dominant unblocking temperature of 550°–580°C. Thus, if the alteration responsible for the remanence of the ultramafics postdates this rotation event, the alteration process must have taken place later than the intrusion of the gabbros and at a lower temperature.