DISCUSSION

Previous magnetic fabric analyses of oceanic basalts by Ellwood (1975) and Ellwood and Watkins (1976) focused attention on the nature of their emplacement. Ellwood (1975) presented a parameter, the F-factor, the value of which could be used to distinguish between intrusive and extrusive igneous bodies. We have calculated values of this parameter for our samples which show that ~10% (i.e., 29 samples) have F-factor values indicative of intrusive rocks. This is consistent with the extrusive nature of practically all of our samples. Based upon the standard parameters used to describe their magnetic fabric, it appears that the majority of our samples exhibit weak anisotropy. Ellwood (1978) reports similar low anisotropy values for the basaltic bodies he examined. In general this is true for both the pillowed units and massive units, although the massive units show a minor tendency to greater anisotropy. Differences between pillowed and massive units, however, are mostly small, and similarities in their magnetic properties are more pronounced. The broad similarity between the pillowed and massive units may in part be attributed to difficulties in discriminating between those massive units that are part of basaltic flows from those that are simply thicker pillows. There are some differences in the magnetic fabric that may be discerned in the orientations of the anisotropy ellipsoid. The subhorizontal plane of the ellipsoid (which is assumed to represent the flow plane) is somewhat different for pillowed samples than the massive samples. For many massive units the ₃ axis is nearly horizontal with either the ₁-₃ or ₂-₃ plane, forming a subhorizontal flow plane, whereas for many of the pillowed samples it is the ₁-₂ plane that is nearly horizontal. These differences are not pervasive, however, and there are examples of both steep ₁-₂ planes for pillowed samples and shallow ₁-₂ planes for massive units.

Laboratory experiments by Canon-Tapia and Pinkerton (2000) on simulated lava flows suggest that high values of the degree of anisotropy, A, may be associated with environments where the rocks are deformed close to their eruption temperatures and then cooled rapidly, reflecting conditions near the boundary or edge of an individual flow. Conversely, low values of A appear to be associated with more gradual cooling, suggestive of interior portions of flows (Canon-Tapia and Pinkerton, 2000). In principle, therefore, values of A together with other changes may be used to identify boundaries between individual cooling units. In the experiments of Canon-Tapia and Pinkerton (2000), the distinction between high and low values of A is not sharply defined, but generally values of A < 3 are considered low and values of A > 4 are considered high. As noted above, most of the pillowed samples in our study were obtained from portions of the pillow away from the chilled, glassy margins and, consequently, their low A values are consistent with their interior locations and presumably more gradual cooling conditions.

In their work on the Birkett flow, Canon-Tapia and Coe (2002) identify individual flow units based upon changes in both magnetic fabric parameters, especially A, and dominant AMS directions. Their results suggest individual flow units have variable thicknesses but are generally between 0.2 and 1.9 m thick. Unfortunately, sampling of the massive units in our study is generally too coarse to identify and document individual flow units with such thicknesses. Several samples have high A values, but we have insufficient samples between these to characterize individual flow units. The various pillow and massive units have been subdivided into inclination groups on the basis of their ChRM directions. Boundaries between inclination groups have been chosen on the basis of the ChRM inclination values and their variability. As such, the inclination groups are in some intervals well defined and elsewhere poorly defined. An examination of A values at or near the boundaries between inclination groups at all four sites shows that in many cases the transition is associated with higher A values (i.e., A > 4). For example, at Site 1183, values of A are generally <2, but for those samples at the transitions between inclination groups 1 and 2, 2 and 3, and 3 and 4, A > 4 (Fig. F7A). However, samples near the transition from groups 4 to 5 do not have large A values. Furthermore, within inclination group 2 there is a sample with high value (A > 7). Similar correlations of high A values with boundaries between inclination groups can be made at each of the other three sites. At Site 1186, boundaries between inclination groups 1 and 2, 2 and 3, 4 and 5, and 5 and 6 are all associated with spikes in A values that are 4 or greater (Fig. F7B). But there are a number of intervals within the presently defined inclination groups at each site where A is high but no substantial change in the ChRM inclination is recorded. Individual inclination groups often span several stratigraphically successive samples even though they are clearly from several separate pillows or flows. In a succession of flows or pillows that form during a short time interval, there may be several individual bodies that have boundaries where rapid cooling takes place. Such a succession is expected to have the same magnetic field directions and therefore would be part of the same inclination group. High values of A in the middle of an inclination group are therefore not unexpected. High values of A associated with boundaries for the Birkett flow are observed over 10- to 20-cm intervals, and it is therefore entirely possible that such boundaries have been missed by our coarse sampling.

Studies of magma flow directions derived from oceanic pillow basalts are very limited. Ellwood (1978) examined the anisotropy ellipsoid of samples from two pillows and found preferred azimuths but was unable to relate these to geographic coordinates. Thus, although there are broad similarities between our results and those reported by Ellwood (1978), the limited number of samples from this latter study make detailed comparisons difficult. Preferred azimuths for the magnetic fabric have been identified at each site, although the standard deviations and -95 angles are large, making it possible to only define the quadrant with any reliability. Estimates of the secular variation at ~120 Ma for a paleolatitude of ~25° (corresponding to the Ontong Java Plateau) suggest changes in the magnetic pole position of ±18° (McFadden et al., 1991). Secular variation may therefore contribute in a significant way to the large standard deviations. For two sites (i.e., 1186 and 1187) there is a strong directional bias; for Site 1185 there is a modest directional bias, and for Site 1183 there are two more or less orthogonal directions for which there is a minor directional preference (Fig. F5). To relate these directions to the overall geometry of the Ontong Java Plateau it is necessary to reorient the directions so that they are referenced to the location of the paleomagnetic north pole at the time of basalt eruption at ~120 Ma. Unfortunately, the Apparent Polar Wander Path for the Pacific plate for the Early Cretaceous is not well determined, and, consequently, the direction to magnetic north is not well defined. In a recent study based upon results from Leg 192, Riisager et al. (2004) estimated a pole for the time of eruption of the Ontong Java Plateau basalts. This pole, located at 63.0°N, 10.1°E, lies ~15° counterclockwise of geographic north with respect to the sites. To account for this difference, the preferred azimuths for each site given in Table T2 and shown in Figure F5 have been rotated counterclockwise by 15° in Figure F1. Sites 1186 and 1187, which are located >200 km apart, have essentially the same strong preferred azimuth along ~N155° (or N335°). The similarity in the preferred azimuths at Sites 1186 and 1187 is somewhat surprising because the value for Site 1187 is based mostly on pillow data, whereas the Site 1186 direction comes mostly from massive samples. The N155° direction is almost perpendicular to a line from these sites toward the present-day shallowest part of the plateau (Fig. F1). It is therefore not clear whether the directional bias relates to the flow direction or a direction perpendicular to flow. Although some studies have indicated that the ₁ axis may be perpendicular to flow (e.g., Ellwood, 1978), many more studies have indicated that flow is generally parallel to this axis. An alternative explanation is that the directions represent local flow patterns. Site 1185, with a somewhat smaller azimuthal bias, has a preferred azimuth along ~N90°/270°—a direction that points approximately toward the present-day crest of the plateau. Site 1185 also lies near the edge of the plateau (Fig. F1), and the preferred azimuth is almost orthogonal to this boundary. Site 1183, which lies closer to the plateau crest than the other sites, has two rather poorly defined preferred azimuths: N120° and N210°—both of which can be interpreted as flow directions away from the crest. Because the two directions are orthogonal to one another, it is also possible that only one represents the flow direction but, if so, it is not known which direction is the flow direction.

The results for Sites 1183 and 1185 are broadly consistent with a flow direction away from the shallowest part of the plateau. The common direction at Sites 1186 and 1187 is more problematic because it is roughly parallel to the general bathymetric trend near these sites. A more detailed fabric analysis is needed to resolve whether the N155° azimuth represents the flow direction at these sites or some other attribute of emplacement.