Pass-through magnetometer measurements were taken on all split-core archive sections at 5-cm intervals. Pass-through magnetic susceptibility measurements were taken on all unsplit core sections at 4-cm intervals. In addition, we subjected several discrete samples to detailed thermal and alternating-field (AF) demagnetization as well as isothermal remanent magnetization (IRM) experiments.
To isolate the characteristic remanent magnetization (ChRM), cores were subjected to AF cleaning. The number of AF demagnetization steps and the peak-field intensity varied depending on the lithology, the natural remanent magnetization (NRM) intensity, and the amount of time available. In general, following the measurement of their NRM, the calcareous ooze split-cores were demagnetized using three AF steps up to 15 mT. The volcaniclastic split cores were demagnetized using a minimum of six AF steps. The maximum applied field ranged between 30 and 80 mT. We analyzed the results in Zijderveld and stereoplot diagrams and, where possible, calculated the ChRM direction using principal component analysis (Kirschvink, 1980). Examples of good-quality demagnetization results are shown in Figure F71.
We found the NRM intensities of sediments at Hole 1184A to be extremely variable, ranging over five orders of magnitude from <5 × 10-5 to >5 A/m. The lower Miocene (see "Biostratigraphy") ooze (Cores 192-1184A-2R through 8R) is weakly magnetic with a mean NRM intensity of 3 × 10-4 A/m and negative susceptibility values, reflecting the dominance of diamagnetic calcite. The ooze was disturbed by drilling, rendering NRM related measurements unreliable (see also "Paleomagnetic Directions").
The middle Eocene (see "Biostratigraphy") volcaniclastic rocks (Cores 192-1184A-9R to 46R), in contrast, are strongly magnetic (Fig. F72). Susceptibility values range from 10-3 to 5 × 10-2 SI, which is higher than the bulk susceptibility of common, weakly magnetic minerals (e.g., goethite and hematite). Hence, the magnetic signal is carried by more strongly magnetic minerals (e.g., titanomagnetite, maghemite, or pyrrhotite). This conclusion agrees well with the few IRM acquisition experiments we carried out, in which we did not detect hematite, goe-thite or other high-coercivity magnetic minerals. Although hematite and goethite may be present, they do not contribute significantly to the magnetic properties of the volcaniclastic rocks. Thermal demagnetization on discrete specimens often showed a decrease in NRM intensity and a small increase in susceptibility after heating to 350°C (Fig. F71C). We interpret this behavior to be the result of the breakdown of maghemite to magnetite, a process that typically takes place at ~300°C.
In conclusion, the preliminary rock-magnetic observations point to a magnetic mineralogy dominated by magnetite (possibly titanomagnetite) that, to a variable degree, has undergone low-temperature hydrothermal oxidation to maghemite. These rock-magnetic observations agree well with the finding of maghemitized titanomagnetite in polished thin sections (see "Igneous Petrology") and the detection of magnetite in XRD studies (see "Alteration"). Simple calculations suggest that as much as 1.5% of a magnetite-type mineral is required to explain the observed susceptibility values.
NRM intensity, magnetic susceptibility, Koenigsberger ratio, and the percentage of NRM removed after AF demagnetization to 20 mT for the volcaniclastic rocks are shown in Figure F72. In a gross sense, the downhole variations in magnetic properties reflect the presence of three major layers: 201.1-304.2 mbsf, 304.2-442.5 mbsf, and 442.5 mbsf to the bottom of the hole. The boundaries between these magnetic layers coincide with lithologic boundaries (see "Lithostratigraphy"), but not all lithologic boundaries are associated with major changes in magnetic properties. For example, there is no significant difference in the NRM intensities between Subunits IIA and IIB nor between Subunits IIC and IID. Below we describe the variations in magnetic properties for each lithostratigraphic unit of the volcaniclastic rocks:
The Miocene oozes were badly disturbed by drilling. For this reason, we were unable to obtain reliable paleomagnetic directions from Cores 192-1184A-2R to 8R. In contrast, the volcaniclastic rocks were largely undisturbed by drilling, and relatively long, continuous pieces were recovered. The demagnetization data from Cores 192-1184A-20R through 22R were scattered, and interpreting them using principal component analysis (PCA) was impossible. The scatter may be caused by the presence of many large clasts in this interval. The AF demagnetization of the remaining volcaniclastic rocks produced straightforward results, and we could define the ChRM direction precisely (e.g., Fig. F71). We used PCA to calculate the ChRM direction for ~2200 points, corresponding to roughly every 10-15 cm of recovered core (i.e., every second to third measurement point). In order to better represent the large amount of data, we have used the method of Kono (1980) to calculate the mean magnetic inclination and associated alpha-95 value for 5-m intervals (Fig. F74). The inclination values used are those we determined from PCA, and each value in Figure F74 therefore represents the mean of ~25-30 estimates.
The mean inclination for all intervals, -54.4°, is significantly steeper than the ~-30° value predicted for Eocene age rocks at Site 1184 (see, for example, Sager, 1987; Petronotis and Gordon, 1999). We were therefore concerned that it might be an artificial remanence acquired in the recovery process. The inclination is shallower than that expected for a drilling overprint, which is commonly close to 90° (see "Paleomagnetism" in the "Site 717" chapter of the Leg 116 Initial Reports volume [Shipboard Scientific Party, 1988]). In addition, drilling overprints are usually "soft" and can be reduced significantly or eliminated entirely by low-field AF demagnetization. However, we were able to document the -54° direction in core sections that had been routinely demagnetized at fields of 40 mT and, in some cases, in fields as high as 80 mT. Another explanation for the -54° inclination is that it is an artifact of the core-splitting process. To eliminate this as a possible source of the magnetization, we performed pass-through magnetometer measurements on a whole core section (Section 192-1184A-34R-4) and compared the results with those obtained from the corresponding split-core section. We observed the same inclination in both the whole and split-core sections. To further investigate the nature of this magnetization, we selected several discrete samples for demagnetization. Five discrete samples were AF demagnetized in small steps up to 80mT, and six samples were subjected to thermal demagnetization up to 600°C. In neither the AF nor the thermal demagnetization experiments did we find evidence for magnetic directions other than the dominant -54° inclination direction. Because many core sections contained long, continuous pieces, we were able to successfully compare the discrete sample AF demagnetization results with those of the pass-through measurements. We found good correlations between these two data sets, demonstrating that the pass-through measurements reliably isolated the dominant remanence. A few pass-through measurements suggested the presence of an additional, lower-coercivity component, but such intervals were rare, and we were unable to firmly document this component. Detailed shore-based measurements on discrete samples will allow us to investigate such components further.
On the basis of the tests described above, the results of the discrete sample demagnetization and the relationship of the magnetic directions to sedimentary features (see immediately below), we tentatively conclude that the -54° direction is a natural rather than an artificial magnetization. Consequently, we interpret the negative inclination to be that produced by magnetization at a southerly latitude during a period of normal field polarity.
We are currently unable to provide a satisfactory explanation for the steepness of this component. Without any tilting of the sediments, the -54° inclination corresponds to a latitude of ~35°S, which is considerably farther south than predicted for the site in the Eocene by other authors (e.g., Sager, 1987; Yan and Kroenke, 1993) and also by our own results from Eocene cores from Site 1183 (see "Paleomagnetism" in the "Site 1183" chapter), which suggest that Site 1184 was located near 15°-20°S. Because regional tilting of the beds is generally to the north, correction for this tilting produces an even more steeply inclined magnetization vector. To obtain a magnetic inclination that corresponds to the predicted Eocene latitude, the sedimentary rocks would have to acquire their magnetization on a more steeply inclined surface whose angle was later reduced. We have considered a number of other possible factors to help explain the steep inclination, including local tilting of the beds not apparent in the available seismic reflection data, deviation of the drill hole from vertical, abnormal secular variation, a geomagnetic excursion, and a significantly more southerly Eocene position for the eastern salient than for other parts of the Ontong Java Plateau. None of these appears to be a likely candidate to explain the steep inclination that corresponds more closely to the inclinations observed at Site 1183 (see "Paleomagnetism" in the Site 1183" chapter) in 100- to 120-m.y.-old rocks. We conclude that the magnetic inclinations at Site 1184 are inconsistent with the Eocene age assigned, based upon the biostratigraphic analysis (see "Biostratigraphy").
The dominant -54° inclination component appears to vary somewhat systematically with depth in the section. This variation has an amplitude of ~5°, varying between -50° and -60°, and is observed over depth intervals of ~150-200 m. The variation is greater than the uncertainty in the individual mean values; we infer that the variation corresponds to changes in the geomagnetic field as the sediments were magnetized. Because the variations at Site 1184 are similar to secular variations of the field observed in modern lake sediments, we speculate that they may reflect processes occurring over intervals of hundreds to thousands of years. Whether such changes took place during the accumulation of the sediments or were postdepositional is unknown.
We measured the dip angle and downdip direction of all clearly distinguishable inclined layers found in the volcaniclastic rocks. Omitting features we suspected to be related to erosion rather than deposition, we measured 82 planar features. Their dip angles and downdip directions, measured in core coordinates, are shown in Figure F75A (core coordinates are defined in "Sampling Coordinates" in "Paleomagnetism" in the "Explanatory Notes" chapter). The apparent grouping of down-dip directions parallel to the splitting-surface (y-z plane) is an artifact, caused by the fact that it is much easier to estimate the dip in this plane than in the x-z plane. In cases where we were unsure of the x-z plane dip, we tended to underestimate this dip.
Using the declination of the ChRM for the continuous core pieces that contain the inclined layers, we reoriented each individual feature in paleomagnetic coordinates (Fig. F75B). Defining the ChRM direction for two data points was impossible, so these were discarded. The dip directions are more tightly grouped after reorientation, providing independent support to our conclusion that the ChRM was acquired by the sediments while they were in place and does not relate to drilling or laboratory artifacts. Furthermore, the dip angles and dip directions of the inclined layers, in paleomagnetic coordinates, agree fairly well with the estimated dip of intravolcaniclastic seismic reflections (i.e., dip angle ~9°, downdip direction ~0°, see "Background and Objectives"). We interpreted nine steeply (>20°) inclined layers of Subunit IID, dipping toward the southwest (Fig. F75B), to reflect cross-bedding, possibly indicating the sediment transport direction (for further discussion, see "Lithostratigraphy").