CONVENTIONS, PROTOCOLS, AND EDITING OF THE DATA FILE
Magnetic susceptibility (
) is given in dimensionless SI (Système Internationale) units. Minicore measurements made during Leg 176 are listed as volume-normalized SI units x 10-6 (Shipboard Scientific Party, 1999b, table T11), whereas output from the MST is recorded in machine units (MU x 10-6), but these measurements are not volume normalized. Conversion to SI units requires multiplication by a geometric calibration constant (C) that depends on the core diameter. For a whole core of diameter 66 mm, C = 0.66, and this is conventionally applied to MST measurements of magnetic susceptibility of sediments. However, spot measurements of full-round gabbro core during Leg 176 gave diameters of 58.5 to 59.5 mm, thus application of this constant is not quite correct.
As noted in the site report (Shipboard Scientific Party, 1999b), MST data for magnetic susceptibility recorded during Leg 176 were degraded by three other factors: (1) in some cases, the spacing of rock in the liners meant that there was no rock in the core as it passed through the instrument; (2) in other cases, the rock was either rubble or had less than full core diameter; and (3) the readings from the MST saturate and wrap around at machine values over 10,000 (e.g., a true value of 10,500 would appear as 500 in the data set). To use the data set in more than a qualitative way means that it should be edited to remove measurements made on core not having a full-round core diameter and that instances of machine saturation should be identified.
I have accordingly screened the data set in the following ways. First, all measurements not made on full-round core were deleted. Second, measurements within 2-4 cm of ends of whole-round pieces separated by plastic dividers were also deleted. During Leg 176, the half-width of the peak produced by a point-element of susceptible material (a nail) was found to be 5 cm. "For a rock of even the nominal 6.6 cm diameter to yield its true volume-normalized value requires a sample at least ten centimeters long" (Shipboard Scientific Party, 1999a, p. 77). Measurements within 10 cm of the end of a piece are degraded.
Two factors mitigate this effect, however. The first is that highly susceptible zones of core are rarely point sources; they are seams of rock rich in magnetite dispersed over some centimeters and they are usually inclined. The second is that except at the ends of sections, gaps containing plastic spacers between pieces are usually only 1-3 cm. Such narrow gaps do not produce the same drop-off in measured susceptibility as the abrupt end of a section. Thus rocks of low susceptibility usually have no obvious drops in susceptibility within 2 cm of the ends of pieces, as long as those pieces are closely spaced.
However, examination of several narrow but strong peaks in the data showed full drop-offs to background levels within 4 cm of maxima produced by very narrow oxide-rich seams. Predicted measurements are higher (Fig. F2). In the case of Sample 176-735B-139R-1 (Piece 7) shown in Figure F2, the apparent drop-off was such that the measurement should have exceeded 10,000 x 10-6 MU, and the next underlying measurement was also too low because it came at the end of Section 139R-1. The predicted values are indicated by the blue dashed line. Some uncertainty exists in the vicinity of a felsic vein that obliquely crosses the core at the end of the section and beginning of the next. This is indicated by question marks. The appropriate response is exhibited by the lower drop-off to the peak centered at 872.64 mbsf at the top of Section 176-735B-139R-2. Using this type of evaluation, I therefore usually chose to delete two measurements at spacer gaps between full-round pieces, the one at or nearest the spacer and the one measurement closest to it in the rock on either side. This was always done by comparison of the listed location of the measurements to the graphic representation of the core on the site report barrel sheets and, sometimes, to core photographs. Of the 22,678 original measurements, this left 18,971 measurements at 4-cm spacing on full-round core, representing 758.84 m of rock (18.08 m of this was at 2-cm spacing). This is 75.6% of the 1003.2 m drilled in Hole 735B during Leg 176. In the annals of attempts to recover whole-round core of igneous rock by rotary coring from a wave-tossed platform in deep water, this is an extraordinarily high figure.
Finally, in the midst of 15 high-susceptibility peaks, a single value abruptly drops to levels no greater than the nearest background measurements—too much of a drop-off. I assume these to be values above the saturation limit and have added 10,000 x 10-6 MU to them.
With all of these adjustments, I then compared minicore measurements of magnetic susceptibility (Shipboard Scientific Party, 1999b, table T11) to the nearest MST measurements (Fig. F3). Linear and power-law regressions fit the data similarly, both having standard deviations r > 0.8. The slope on the linear regression gives a value for C of 0.62. However, the regression does not fit values with low magnetic susceptibility particularly well. Forcing the regression through the origin increases C to 0.71 with little change in standard deviation. The slope of the data field is better matched, but the bulk of the data are offset from the regression. The power-law regression matches the log-log linearity of the data well, is not offset, and has a higher correlation coefficient than either variant of the linear regression.
My main interest here, however, is in the relative differences of values measured on whole-round core. To the extent that core diameters are consistent to within a small error, the machine units, which are not volume normalized, are perfectly adequate to describe variations in the core and to compare with geochemical or other data obtained on the same rocks. In most of the following, I consequently use machine units (x 10-6 MU). Comparison to the susceptibility well log obtained during Leg 118, however, requires the geometrical correction. The correction factor, C = 0.66, splits the difference between the two values for C given by the different ways of calculating the linear regression, and I use this to provide consistency with the way the data were plotted in the site report (Shipboard Scientific Party, 1999b). The power-law relationship works better for high values of magnetic susceptibility, but since these represent only a small percentage of minicores, the relationship may be an artifact of sampling. The differences between this and applying the simple correction affect only a small percentage of the core and are not sufficient to affect general relationships.
What explains the scatter about any of the regressions? Besides the imprecision inherent in the repeatability of any measurement, the objects being compared are not identical. First, the Bartington sensor loop samples a volume with a 5-cm half-width, and oxide minerals usually are unevenly distributed in whole-round core. Minicores are much smaller and sample the oxides in the working half in a different way. Over some hundreds of comparisons between MST and minicore measurements, however, this effect should even out. Similarly, X-ray fluorescence (XRF) bulk compositions were determined either on slab samples or on quarter-round samples obtained from the working half of the core. These also sampled oxide minerals in a different way than did the Bartington sensor. Once again, however, over many samples, the effect should even out.
The final editing procedure was to separate peaks from background. This involved an assumption that there were sharp boundaries to intervals bearing magmatic iron-titanium oxides. This certainly is the case for the rocks themselves, for which the presence and abundance of magmatic oxides is a critical aspect of the lithologic identification (Shipboard Scientific Party, 1999a). In almost all cases, abrupt spikes in susceptibility, even small ones of a few hundred machine units above a definable background, are the norm. However, as shall be demonstrated, background values themselves shift around, and maximum values for small peaks at one place can be the same as background levels elsewhere. The range of background measurements is from just below 100 x 10-6 to 800 x 10-6 MU, with fluctuations occurring on a scale of several to several tens of meters. Within this range, there can be oscillations of ~200 x 10-6 MU from one measurement to the next; these are still in excess of the sensitivity of the Bartington sensor. Rather than choosing an arbitrary value to separate peaks from background, I therefore examined plots of the data at 50-m intervals, in each case separating individual peaks from whatever their immediate background happened to be. On this basis, 62.8% of full-round measurements are at background levels, representing oxide-poor lithologies, mainly olivine gabbros and troctolites, and 37.2% are peaks representing oxide-bearing and oxide-rich lithologies, mainly disseminated oxide and oxide gabbros, gabbronorites, and olivine gabbronorites. Some gabbronorites, however, have low magnetic susceptibility, and some olivine gabbros have relatively high magnetic susceptibility, a factor we shall have to consider more carefully later on. Measurements of susceptibility >2000 x 10-6 MU comprise 8.6% of the total and represent prominent seams of oxide gabbro.
The following is illustrated with images scanned from whole-round core, split core, slab samples, thin sections, photomicrographs, and core photos. Usually, I have somewhat sharpened the contrast of these images by computer, and in some cases I have expanded the horizontal scale, but otherwise I have left them alone. Details of image adjustments will be found in the figure captions. Of the various image types, only those obtained from whole-round core may be unfamiliar to the reader. Whole-round images were obtained on every cylindrical piece obtained during Leg 176 using a DMT Digital Color CoreScan system by rotating the core around its cylindrical axis with the camera line scan positioned so that it is parallel to the axis of rotation (see Shipboard Scientific Party, 1999a). Core pieces were placed consistently in the liner using a reference line drawn by one of the structural geologists using a wax marker. This line was always drawn to maximize the apparent dip of inclined fabric in the rock on split surfaces of the core. This is the basis for the "Core Reference Frame," against which variations in the orientation of all structural elements and magnetic declination were compared in the site report (Shipboard Scientific Party, 1999b). The result is an "unrolled" image, from 0° to 360°, oriented left to right, in which planar features such as fractures and joints have a sinusoidal pattern, the amplitude of which increases with the dip of the feature. The images were intended to augment borehole images from the Formation MicroScanner and to allow full reorientation of the core. I use them here simply to depict features of the core related to fluctuations in magnetic susceptibility.
I have also tried to relate magnetic susceptibility and the features imaged to core descriptions, whether from the barrel sheets in the site report or itemized in spread sheets included as supplemental material to the site report (Shipboard Scientific Party, 1999b). The principal descriptive unit during both Legs 118 (Dick et al., 1991) and 176 was the "lithologic interval," which is taken to be any length of core >5 cm in thickness that consists of a single rock type that is distinguishable from its surroundings in mineralogy (principally, presence or absence of olivine or oxide minerals), grain size, or texture. Igneous intervals have sharp, diffuse, or sutured contacts that variously represent tectonic or intrusive relationships. Textures can be magmatic, thus undeformed, or they may exhibit brittle or ductile deformation, including cataclastic, gneissic, porphyroclastic, mylonitic, and ultramylonitic fabrics. An exception to the 5-cm lower limit to describable lithologic intervals is leucocratic, or felsic veins, of which 203 are annotated in the igneous veins spreadsheet in the supplemental material to the site report. These were millimeters to a few centimeters wide. I have numbered these sequentially downward in annotations on the figures. The term "felsic" was chosen as the primary descriptor because starkly white sodic plagioclase is the most abundant mineral in the veins, which have low color index. However, the veins range from diorite to granodiorite in composition, most being trondhjemite, and contain quartz, sodic plagioclase, minor green amphibole, and, in some cases, traces of magnetite, apatite, and zircon. Most felsic veins have sharp contacts, thus are intrusive, but some are dispersed or diffuse. Shipboard descriptions indicate a common association between felsic veins and intervals of oxide gabbro. Data for magnetic susceptibility provide a means of investigating this relationship.
