PALEOMAGNETISM
Introduction
The vertical structure
of the sources of lineated marine magnetic anomalies have remained poorly known ever since
the recognition, more than 30 yr ago, that the ocean crust records reversals of the
geomagnetic field. Inferences on the magnetization of lower crustal rocks from studies of
dredged rocks (Fox and Opdyke, 1973; Kent et al., 1978) are ambiguous because these
surficial samples have been subjected to varying degrees of seawater alteration that may
have significantly affected the magnetic properties. To date, Hole 735B constitutes by far
the deepest penetration into plutonic basement and thus arguably provides the best
available information on the magnetization of gabbroic rocks generated at slowly spreading
ridge crests.
During Leg 118, ~500 m
of gabbroic rock was drilled in Hole 735B, with a high recovery rate of ~87%. Both
discrete sample magnetizations and logging data from Hole 735B during Leg 118 indicate a
stable inclination with reversed polarity, consistent with the location of the site, which
was mapped within sea-surface magnetic Anomaly 5r (Dick et al., 1991b). The average
inclination of 71.3º (+0.4º/-11.0º) (using the inclination-only averaging technique of
McFadden and Reid, 1982) is approximately 20º steeper than the expected inclination (
51º) for the latitude of the rift
valley at 32ºS. The average in situ natural remanent magnetization (NRM) estimated for
Leg 118 gabbroic rocks (~2.5 A/m; Pariso and Johnson, 1993), together with the relatively
high ratio of remanent to induced magnetization, suggests that gabbros may constitute a
significant source for lineated marine magnetic anomalies (Kikawa and Pariso, 1991; Kikawa
and Ozawa, 1992; Pariso and Johnson, 1993). Indeed, they apparently constitute the sole
source of lineated sea-surface anomalies observed near Site 735 where there are no
overlying basalts (Dick et al., 1991b).
During Leg 176, Hole
735B was deepened to 1508 mbsf, again with a high core recovery of 86%. Magnetic
measurements were made on minicores as well as on half and whole cores. Magnetic
susceptibilities were measured on all whole cores before they were split. Remanence
measurements were conducted on the archive halves with stepwise alternating-field (AF)
demagnetization. We collected 346 discrete samples for magnetic studies that included
remanence measurements, AF and thermal demagnetization, and determination of the
anisotropy of magnetic susceptibility.
Whole-Core
Measurements
Susceptibility
The susceptibility (
) of Leg 176 whole cores was measured
at a 4-cm interval with a Bartington MS2C Sensor (80-mm diameter) integrated in the MST
system (see "Physical Properties"). Susceptibilities
were also measured (at a 2-cm interval) on archive-half cores from the lowermost 50 m of
core recovered during Leg 118. The dynamic range of the sensor is 1 × 10-6 to
1 × 10-1 (SI). This is multiplied by a geometric calibration constant (C) that
depends on the core diameter. For a whole core with a diameter of 66 mm, C = 0.66. We use
this correction factor even though the core diameter is variable and generally less than
60 mm. All readings in excess of 0.1 (SI) are clipped, such that a susceptibility of 0.11
results in a stored value of 0.01, for example. Because this clipping effect has been
noted for several core pieces, care should be taken when interpreting the width of
individual high-susceptibility zones as well as in the calculation of average
susceptibility. Similarly, the inherent smoothing imparted by the large sensor diameter
(the response function has a half width of ~5 cm) and the relatively coarse measurement
spacing (4 cm) limit the resolution of narrow features with magnetic susceptibility above
background levels.
For most of the gabbros
measured during Leg 176, the susceptibility is proportional to the magnetite content.
Magnetite has a susceptibility of
3 (SI units) with little
dependence on grain size (Heider et al., 1996). In contrast, the susceptibilities of other
ferrimagnetic minerals likely to occur in gabbroic rocks (e.g., pyrrhotite, hematite, and
ilmenite) are 1-3 orders of magnitude smaller than that of magnetite. The susceptibility
of paramagnetic silicates is to a good approximation proportional to their iron content
(1% FeO corresponds to 6 × 10-5 (SI); Collinson,
1983). Therefore, the contribution of silicate minerals to the susceptibility of gabbroic
rocks recovered during Leg 176 is unlikely to exceed 10-3 (SI).
The downhole variation
of susceptibility as determined with the whole core sensor is displayed in "Physical Properties." This record has been cleaned for
edge effects; however, the clipping effect for the highest values ( = > 0.1) has not been taken into
account. The main characteristics of the susceptibility data are short-wavelength (<0.1
m) spikes and an overall downhole decrease of mean susceptibilities. This is further
illustrated by the variation of mean susceptibility, calculated over 20-m intervals (Fig. F108; see also "Physical
Properties"). The average susceptibility for Leg 176 Hole 735B whole cores is ave = 5.37 × 10-3
(±7.1), after filtering out data within 5 cm of a core gap and correcting for the
geometric factor C, described above, which is related to the core diameter. This value is
similar to the average value (ave
= 7.4 × 10-3 SI) calculated for the minicores (see below) for the same depth
range 500-1500 mbsf and slightly lower than the average value (ave = 23.6 × 10-3)
for Leg 118 minicore measurements (0-500 mbsf).
Remanence
Measurements
Several problems were
encountered during remanence measurements on the archive halves with the pass-through
magnetometer. The largest intensities of magnetization that can be measured on half cores
are on the order of 5-10 A/m (depending on direction) at the slowest possible tray speed
of 1 cm/s, the limiting factor being the slew-rate of the superconducting quantum
interference device (SQUID) sensors. Therefore, some pieces with higher magnetization had
to be removed for NRM measurements. These pieces were generally replaced for the first
demagnetization step. Well into the leg we discovered that the magnetometer software had
an error that resulted in false directional readings. The Y-axis calibration constant had
the wrong sign, resulting in declinations being west instead of east. This error
apparently has been present since the installation of the present LabView software for the
2G magnetometer during Leg 169. A good deal of time was spent correcting the data already
acquired and uploading these data to the JANUS database.
Remanence measurements
were performed at 2-cm intervals for all archive halves. AF demagnetization steps were
typically 5, 10, 15, 20 mT for Cores 176-735B-89R through 168R and 10, 20, 30, and 40 mT
for Cores 176-735B-169R through 210R after we realized that 20 mT might not suffice to
remove secondary drilling-related overprints completely. The archive-half core remanence
data were also filtered to remove edge effects (similarly to that for the susceptibility
whole-core measurements) owing to the response functions of the SQUID sensors, which have
half widths of ~10 cm. Therefore, only data from core pieces longer than 15 cm were
considered, and from these pieces only measurements more than 5 cm from the end of pieces
were retained. Fisher average directions were calculated for all pieces greater than 15 cm
in length.
The resulting downhole
variation of NRM declination, inclination, and intensity are displayed in Figure F109. NRM inclinations for archive-half cores
measured during Leg 176 (below 500 mbsf) are predominantly positive and steep. In
contrast, initial inclinations measured on half cores from Leg 118 (above 500 mbsf) are
dominantly negative, reflecting the presence of a substantial upward-directed,
drilling-related overprint (Robinson, Von Herzen, et al., 1989). The small number of
negative inclinations evident in the lower portion of the hole are almost always related
to pieces archived and/or measured in an inverted position. NRM declinations from
archive-half cores tend to be clustered near 360º, an unexpected result for cores that
are azimuthally unoriented. Comparison of the NRM from a whole core before splitting and
the archive-half core after splitting suggests that the declination clustering may be an
artifact resulting from a drilling-related overprint (Fig. F110). The declination of the whole core contrasts strongly with that of
the half core because the latter clusters near 360º as a result of the radial overprint,
whereas whole-core declinations are not confined. The inclination for the whole core is
generally steeper than for the half cores because the radial components cancel out. The
prevalence of NRM declinations near 360º in the archive-half cores is also well
illustrated by the distribution of average NRM declinations from more than 1900 core
pieces (Fig. F111A).
Average declinations
calculated from archive-half core pieces generally become more scattered after the removal
of the overprint by 20 mT demagnetization (Fig. F112).
However, a significant concentration of declination values is evident near 260º (Fig. F111B). As discussed more fully below, this preferred
orientation of declination is identical to that determined on minicores. As NRM
declinations of minicore samples (from the working half) are biased toward 180º by the
radial overprint, the correspondence between the stable remanent declinations in minicores
and the archive-half core data suggests that the radial overprint has been largely removed
by AF demagnetization at 20 mT. The simple arithmetic average inclination value for the
filtered half-core data of Leg 176 is 71.8º ± 13.3º, with no apparent downhole trend
but with significant scatter.
Discrete Sample
Measurements
Natural Remanent
Magnetization, Intensity, and Direction
The natural remanent
magnetization (NRM) is often considered to be a good indicator of the in situ
magnetization, which is the only relevant parameter for comparison with magnetic
anomalies. Unfortunately, the NRM of a significant number of the hundreds of minicores
measured during Leg 176 (Table T12) was affected
by a secondary magnetization with apparent radial symmetry that was probably acquired
during the drilling process. Many of Leg 118 gabbros also show evidence of a significant
secondary drilling-induced magnetization, although the orientation of the secondary
component in this case was vertically upward (Kikawa and Pariso, 1991; Pariso and Johnson,
1993). Because of the prevalence of secondary magnetizations, the NRM intensity and
direction from Hole 735B minicores should be viewed with some caution.
NRM inclinations from
discrete samples from Leg 176 are generally similar to values (near 70º) determined from
the archive halves (Fig. F109). In contrast, NRM
declinations from minicore samples tend to lie near 180º, providing nearly a mirror image
of the declination data from the archive halves (Fig. F109).
Since remanent magnetization induced by drilling is radially directed about the axis of
the core, the concentration of declination values near 180º in working half minicores
likely represents the complementary signal to the overprint observed in the archive
halves. However, declination bias for minicore sample NRMs is more difficult to
demonstrate because the stable remanence directions from both minicores and archive halves
are preferentially located near 260º as a result of systematic splitting of the cores
relative to mesoscopic structural features such as rock foliations. NRM intensities for
Leg 176 minicores vary from 0.114 to 95.9 A/m, a range of ~3 orders of magnitude, but show
no consistent trend downhole (Figs. F113, F114A). The range in NRM intensities is smaller than
that observed for Leg 118 gabbros (from 0.00026 to 131 A/m), which have more scattered NRM
intensity values. The arithmetic and geometric means calculated for Leg 176 gabbros are
2.54 (±5.77) A/m and 1.58 A/m (±0.39 log units), respectively.
Magnetic
Susceptibility
Values for magnetic
susceptibility for discrete samples are plotted vs. depth in Figure F113 and listed in Table T12. The
susceptibility values vary from 8.12 × 10-4 to 0.123, a range of about 3
orders of magnitude, and have an arithmetic mean of arth = 7.39 × 10-3 and a geometric mean of geom = 4.65 × 10-3 (Fig.
F114B). These two mean values are smaller than
those calculated for Leg 118 gabbros (arth = 2.37 × 10-2; geom = 8.75 × 10-3). Furthermore,
susceptibilities for samples from Leg 176 are less scattered than those for Leg 118
gabbros and slightly decrease with depth (Fig. F113).
Königsberger
Ratio
The Königsberger
ratio, or Q-factor, is the ratio of remanent over induced intensities of magnetization, Q
= NRM/ × H. We used the value of
the ambient geomagnetic field at Site 735 (H = 30 A/m) for our calculation. As
discussed later, a large number of the samples have secondary overprinting probably
acquired during drilling. Therefore, care should be taken when considering the calculated
Q values. However, as Figure F114C shows, most
samples have a Q-factor much larger than 1. The arithmetic mean (13.4 ± 9.45) indicates
that induced magnetizations contribute relatively little to the magnetic anomalies at the
surface and in the borehole. Including the Leg 118 data reduces the arithmetic mean to
10.9 because of the lower Q values for Leg 118 (Qmean = 7.4). For Hole 735B, Q
values increase with depth (Fig. F113), primarily
reflecting the decrease in susceptibility with depth.
Demagnetization
Data and Characteristic Magnetization
More than half of the
discrete samples analyzed during Leg 176 were collected as paired minicores in conjunction
with shipboard thin-section and XRF samples. One minicore from each pair was subjected to
thermal demagnetization, and the other to AF demagnetization. Adjacent minicores yield
essentially the same remanence directions (Fig. F115A,
F115B, F115C, F115D). Most discrete samples from Leg 176 have two magnetization
components (Fig. F115): one with low stability
that is apparently related to the drilling process, the other with higher stability and
steeper inclinations. Samples in which the near-vertical, low-stability component
predominates, presumably related to drilling, are prevalent in gabbros sampled during Leg
118, yet they are almost entirely absent in gabbros of Leg 176. Instead, although most
samples from the lower kilometer of Hole 735B have low stability of magnetization, they
have a southerly declination and moderate but variable inclination. In most cases, AF
demagnetization at 10-20 mT (or temperatures of ~500ºC) is sufficient to remove this
secondary-magnetization component (Fig. F115). In
a smaller number of samples, AF demagnetization at 40-50 mT was required to remove this
component (Fig. F115E). Identification of this
secondary component was hampered when it was subparallel to the primary remanence
direction (Fig. F115F). Most samples have both a
very "hard" component of magnetic remanence (typically with high and sharp
unblocking temperatures; Fig. F115A, F115B) and a
"softer" magnetic component with a wider spectrum of unblocking temperatures
(Fig. F115C, F115D). Nevertheless, the average
demagnetization characteristics are fairly hard, as reflected in the high median
destructive fields (MDF; Fig. F113). We note that
temperature readings of the Schönstedt thermal demagnetizer TSD-1 are probably too low by
about 20ºC because the magnetite-bearing gabbros require settings of 590º to 600ºC for
complete demagnetization, depending on location within the oven.
The mean characteristic
inclination for Leg 176 discrete samples is 71.4º (+0.3º/-3.1º) calculated by the
method of McFadden and Reid (1982; Fig. F114D).
This is 5º steeper than the average reported for the upper 500 m of Hole 735B drilled
during Leg 118 (Kikawa and Pariso, 1991; Pariso and Johnson, 1993). However, our
recalculation of the Leg 118 data also applying the McFadden and Reid (1982) method
results in an average inclination value of 71.3º (+0.4º/-11.0º), which is statistically
indistinguishable from that calculated from the Leg 176 samples. The previously quoted
shallower inclination from Leg 118 gabbros was evidently the arithmetic mean.
Drilling-Induced
Overprint
The radial direction of
the drilling-induced overprint is nicely illustrated in Figure F116. Four samples were taken from one section of the working half. One
minicore was drilled in the center (as usual), and then cut in two pieces, an outer and an
inner specimen. Two other minicores were laterally offset to the sides of the piece. Upon
AF demagnetization, all samples revealed the same primary component. However, the degree
and orientation of the overprint varies with location. The direction of the secondary
component changes from southeast for the sample at the left, to south for the centered
samples, to southwest for the sample to the right. This is consistent with a radial
direction of the overprinting field and is in agreement with the measurements on the
archive halves (Fig. F109).
We measured magnetic
fields in and around drill bits, bottom-hole assemblies (BHAs), and drill pipes. The
magnetic fields of half a dozen measured roller cone bits were uniformly low with fields
smaller than 0.5 mT. Joints of BHAs had fields up to 2 mT in the center and up to ~5 mT
right on the edges, although the cored material was never close enough to these edges to
have experienced the latter, higher field. One joint of a BHA was measured before and
after magnafluxing. The field after magnafluxing was actually smaller than before,
probably because the applied field was of opposite polarity to the BHA's remanence. The
largest fields were measured on joints of regular drill pipes, with 5 mT at the center and
>10 mT on the edges. Each drill pipe constitutes approximately an along-axis magnetized
dipole. If all connected pipes are of the same polarity, then the magnetic flux closes at
joints, and the field in the interior should decrease. However, when pipes of opposite
polarity are connected, the field intensity at joints will increase, and the field lines
may point radially in (and out at the outside). We think that this situation is the most
likely cause for the radial nature of the secondary overprint on drill-core
magnetizations.
Anisotropy of
Magnetic Susceptibility
The anisotropy of
magnetic susceptibility (AMS) has been measured on all minicore samples with the
Kappabridge KLY-2 by applying the measuring scheme for the ANISO20 program with
measurements in 15 different orientations (Table T13).
The program calculates the susceptibility tensor as well as several derived parameters
characterizing the susceptibility ellipsoid. The degree of anisotropy (P) is
expressed by the ratio 
1/3, where 1 and 3 are the maximum and
minimum eigenvalues, respectively, of the susceptibility tensor. The shape of the
susceptibility tensor is conveniently expressed by the shape parameter T = (2
2 - 1 - 3)/(1 - 3), where i = ln I (Tarling and Hrouda,
1993). The shape parameter varies between -1 and +1, with the former value representing a
perfectly prolate ellipsoid and the latter reflecting an oblate ellipsoid. The direction
of the principal axes is given by the eigenvectors max, int, and min.
Most samples exhibit
significant anisotropy with an average P = 1.11 ± 0.05 and values as high as 1.5
(Fig. F117; Table T13). Anisotropies decrease below 1000 mbsf and increase again below 1400
mbsf. The AMS is predominantly oblate also down to 1000 mbsf and again below 1400 mbsf,
whereas prolate anisotropies are more abundant between 1000 and 1400 mbsf. At first
inspection, the orientation of the eigenvectors appears to be randomly scattered. (Fig. F117). However, the dip of max is mostly below 30º and
rarely exceeds 60º. The min
inclinations range up to 90º with clear clustering at ~45º for the shear zone at 960
mbsf.
Discussion
The magnetic properties
of Leg 176 Hole 735B drill cores are well characterized by a large number of measurements
on discrete samples and the entire archive half. The average NRM intensity of Leg 176
minicores is 2.5 A/m, which is less than the average of 6.4 A/m for the upper 500 m cored
during Leg 118. However, the Leg 118 samples carried a significant drilling-induced
overprint, and the mean effective remanent magnetization was estimated to be 2.5 A/m
(Pariso and Johnson, 1993). A drilling-induced component is also apparent for Leg 176
samples as a radial overprint; however, the magnitude appears to be smaller than for Leg
118 samples. Despite the fact that oxide gabbros occur predominantly in the upper part of
the hole, there is no general decrease of NRM downhole. On the other hand, susceptibility
values decrease slightly downhole (Fig. F113).
Consequently, the ratio of remanent and induced magnetization, the Q-factor, increases
with depth (Fig. F113). This indicates a decrease in size of
the magnetite grains because their magnetic "hardness" increases with decreasing
size. The MDF increases accordingly with depth. Q-factor, MDF, and the demagnetization
characteristics in general describe very stable remanent magnetization. Moreover, the high
and often very sharp blocking temperatures suggest relatively rapid acquisition of
thermoremanence during cooling of the gabbros.
The gabbros thus
constitute an ideal source for marine magnetic anomalies. An average gabbroic layer
magnetization of 2.5 A/m together with a layer thickness of 3-5 km would in many places
suffice to cause the surface anomalies or at least dominate over the contributions from
the extrusive basalts and the sheeted dikes. However, a variety of effects including mixed
polarity, elevated temperature, and greater source depth might reduce the contribution of
these lower crustal lithologies to sea-surface anomalies observed elsewhere.
The primary remanence
isolated through AF and thermal demagnetization has reversed polarity throughout the hole
with an average inclination of ~71º and no observable downhole trend. This average value
of 71º (not adjusted for any deviation of the drill hole from vertical) determined for
archive halves and minicores of Hole 735B is about 20º steeper than the expected
inclination (~52º) from an axial geocentric dipole. A tectonic tilt for the gabbroic
section of 19º ± 5º (depending on the deviation of the hole from vertical), since the
time of remanence acquisition near the rift axis, can be deduced.
The rather large
scatter of inclinations (and declinations) within a piece would conventionally be
explained by secular variation of the geomagnetic field. However, for very slowly cooled
rocks such as gabbros, secular variation with periods up to a few thousand years should
average out, and more uniform directions have to be expected. Additionally, the often
pronounced magnetic anisotropy as determined by AMS measurements must have deflected the
NRM from the field direction during thermoremanence acquisition. Alternatively, if
significant quantities of magnetite grains crystallized as a result of metamorphism below
the Curie temperature (Tc = 575ºC), then these grains acquired a chemical
remanence that may not average the secular variation of the geomagnetic field.
Furthermore, stress and deformation that occurred below the blocking temperatures may have
caused the directional quality of the paleomagnetic signal to deteriorate.
The magnetic anisotropy
as determined by AMS measurements on discrete samples is fairly pronounced for most of
them. The underlying physical mechanism for magnetic anisotropy is not yet understood. It
may be related to a preferred orientation of elongate magnetite grains or to an alignment
of magnetite grains in chains or planar structures along grain boundaries between or along
crystallographic planes within silicate crystals. In any case, the AMS signal results
predominantly from magnetite grains because they dominate the susceptibility signal even
for the most weakly magnetic gabbros. At present, the relationship between AMS orientation
and other structural features such as magmatic fabric and oriented faults and veins is not
obvious on the large scale.
The consistent
declination data suggest that gross reorientation of structural features in the core may
be possible (Cannat and Pariso, 1991). This assumes that the characteristic remanence was
originally parallel to the reversed polarity geocentric axial dipole (GAD) direction
(180º, +51º). Such reorientation is subject to a number of uncertainties. Because the
measured inclinations are significantly steeper than the expected GAD inclination (~51º),
complex rotations about plunging rotation axes are not only possible, but likely.
Nonetheless, the prevalence of stable declinations near 260º from both archive-half cores
and minicores suggests a possible first-order reorientation of the core coordinate system.
Assuming a south-pointing characteristic remanence declination, the mean declination of
260º would be restored to 180º by a counterclockwise rotation of ~80º. Structural
planar features (magmatic foliation and crystal-plastic fabric) that preferentially dip
toward 90º in the core reference frame would thus dip preferentially toward the axial
rift in the north after this same counterclockwise rotation.
