Paleomagnetic results from Site 1095 provide a magnetostratigraphy that correlates well with the GPTS from the termination of Subchron C4Ar.2n (9.580 Ma) at ~515 mcd upward through the Brunhes Chron (Figs. F7, F8, F9, F10). Exceptions occur in a few intervals where core deformation, dropstones, overprints, or low recovery bias or obscure the paleomagnetic signal.
Differences between U-channel and split-core results, such as those in the upper 5 m of Hole 1095A (Fig. F11) and the interval from 18 to 25 mcd in Hole 1095D (Fig. F12), can be partly attributed to unremoved overprints in the split-core data. Other differences can be attributed to core deformation, such as that evident in core photos (see Barker, Camerlenghi, Acton, et al., 1999) in the interval from 178-1095D-3H-2, 80 cm, to 3H-4, 70 cm (18.48-21.38 mcd), which corresponds directly to where the split-core and U-channel data have their largest differences (Fig. F12). Other biases are also apparent in the split-core data. Dropstones, which are the coarse fraction of ice-rafted debris (IRD), appear to be a problem in a few intervals. Stones larger than 2 cm in diameter, which do not fit within a U-channel sample, along with other IRD may strongly bias the split-core results. In splitting whole cores, these stones may also be dragged several centimeters, resulting in additional core deformation. IRD or deformation probably explains the two shallow-inclination spikes in the split-core data at 4-7 mcd in Hole 1095D (Fig. F12). A large dropstone is evident in the core photo at Section 178-1095D-1H-4, 55 cm (5.05 mcd), and others may be present below the surface of the split cores.
Discrete samples were collected to better assess the magnetic signal and help refine the magnetostratigraphy. For most intervals, the inclinations from the discrete samples agree to within ~10° with those from U-channel samples, though some spurious discrete-sample results occur directly above 120 mcd (Fig. F7). For unknown reasons, ~20 of the discrete samples in this region of the core give shallow inclinations that differ significantly from the steep inclinations expected and observed from the U-channel samples and split-core sections. Given that the upper 120 m is soft sediment, it could be that some systematic deformation occurred as the discrete samples were collected or that the discrete samples have magnetizations that are near the resolution of the magnetometer, resulting in noise-dominated signals. Perhaps even a couple samples were misoriented, but errors during collection or operator errors during measurement are highly unlikely for so many samples in this interval. None of these are particularly appealing explanations for what are clearly anomalous results. Noise seems particularly unlikely because linear demagnetization paths are generally obtained for the samples with anomalous directions. In a few cases, the paths are planar rather than linear, which could indicate that some form of viscous magnetization has been acquired by the discrete samples, which were not measured in a magnetically shielded room, in contrast to the U-channel samples. Similar viscous components are absent from the split-core sections, so even that explanation has little appeal. Below 120 mcd, the agreement between results from discrete, U-channel, and split-core samples indicates that biases are negligible regardless of sample type. Thus, even in the absence of discrete or U-channel data, the split-core data accurately record the paleomagnetic field, with exceptions for intervals with IRD and core deformation as noted above.
In Table T1, we provide a depth range within which each reversal occurs along with the best estimate of the depth of the reversal boundary. The size of the depth range is generally dependent on core recovery and coring gaps, although the length of time for the reversal to complete and the complexity of directional changes during the reversal are also factors. For example, the field direction over the Brunhes/Matuyama reversal shows a change from steep positive inclinations of the Matuyama Chron to shallow negative inclinations and then steep positive inclinations within the transition zone, before finally reaching steep negative inclinations of the Brunhes normal polarity chron (Fig. F11). The transition spans ~24 cm (15.92-16.16 mcd), which corresponds to ~12 k.y. A similar swing in inclination has been noted for the Brunhes-Matuyama transition in sediments from Leg 172 Sites 1060 and 1063 in the northwest Atlantic and from Leg 162 Site 983 in the North Atlantic, where again the transition for all three sites spanned ~5 to 12 k.y. (pp. 318-320 of Keigwin, Rio, Acton, et al., 1998; Channell and Kleiven, 2000). Other reversals appear to be virtually instantaneous within the resolution provided by the sedimentary record, such as the reversal at 84.56 mcd, which we interpret as the termination of Subchron C2An.2n. The best estimate of the reversal depth is generally the midpoint within the depth range. Exceptions to this, as noted in Table T1 and discussed further below, are restricted to reversals that fall within wide or uncertain polarity transition zones. Placement of the reversal to one side of the transition zone may be preferred because less erratic changes in the calculated sedimentation rates may result. Also, when core recovery is poor, magnetic logging data may narrow the region in which the reversal should be placed.
Original interpretation of the shipboard data was difficult for the upper 100 m of the section owing to conflicting results from Holes 1095A, 1095B, and 1095D. Part of the conflict results from inaccuracy in the meters below seafloor (mbsf) depth scale, which produces artificial vertical offsets between laterally continuous, coeval features (Fig. F13). Shipboard identification of Subchrons C1r.1n (Jaramillo) through C2An.2r was particularly difficult for this reason. Also, the split-core samples gave anomalous or somewhat biased paleomagnetic directions in a few intervals that we subsequently recognized when U-channel data were obtained (Fig. F12). Specifically, the region from ~17 to 55 mcd is more complexly magnetized (possibly caused by larger drilling overprints, core deformation, or a greater influence by IRD) than intervals above or below (Figs. F8, F12). These difficulties, in association with an interpreted seismic discontinuity and lithostratigraphic boundary, led us to propose a hiatus at ~52-55 mcd in Holes 1095A and 1095D during Leg 178. When the new paleomagnetic observations from U-channel samples along with the split-core data are plotted on the mcd scale, it appears unlikely that extended hiatuses occur here or elsewhere within the sedimentary section, at least within the resolution of the magnetostratigraphic observations. A hiatus of <200 k.y. would, however, be difficult to observe paleomagnetically within this interval given the relatively slow sedimentation rate, the resolution of the paleomagnetic data, and the age constraints provided by magnetostratigraphic reversals or biostratigraphic events. For example, placing a 200-k.y. hiatus at ~52-55 mcd does not produce abnormally fast or slow sedimentation rates in the sedimentary section above or below the hiatus. Thus, we cannot preclude the occurrence of a relatively short hiatus, such as might be caused by an erosional event that could produce a discontinuity in seismic reflection profiles and a subtle lithostratigraphic boundary.
Our magnetostratigraphic interpretation agrees well with biostratigraphic constraints in most intervals as illustrated in Figure F7, which shows the polarity zones predicted using the age constraints provided by the diatom, the radiolarian, and the calcareous nannofossil events given in Iwai et al. (Chap. 36, this volume). In constructing these predicted magnetostratigraphies, we averaged some diatom events with similar ages, some of which gave conflicting relative depths for their relative ages. For example, the top of the Thalassiosira complicata Zone has an age of 3.4 Ma and occurs at a depth of 62.48 mcd, whereas the top of the Thalassiosira inura Zone has an age of 1.75 Ma and occurs at a depth of 68.33 mcd. These two zones, along with three other zones—top of Thalassiosira torokina (1.85 Ma; 57.705 mcd), top of Thalassiosira insigna (2.57 Ma; 56.495 mcd), and top of Fragilariopsis interfrigidaria (2.67 Ma; 76.53 mcd)—with similar ages and depth ranges are combined to give a mean age of 2.448 Ma for a mean depth of 64.308 mcd. Disagreement between magnetostratigraphic and biostratigraphic ages are generally no larger than the disagreement between ages estimated from the different biostratigraphic constraints. All predict similar ages near the top (e.g., note the similar depths for the Brunhes/Matuyama boundary in Fig. F7), the middle (e.g., note the similar depths for the onset of Subchron C3An.2n), and the base of the sedimentary section (e.g., all age constraints suggest that sediments below 450 mcd must be older than 9 Ma).
Our revised magnetostratigraphy for Site 1095 is also consistent with the downhole logging results presented by Williams et al. (Chap. 31, this volume) for the interval logged from 110 to 570 mbsf in Hole 1095B. The depths to reversals are available in graphical form in Figure F8 of Williams et al. (Chap. 31, this volume). The logging depths are estimated from the wireline length during logging, whereas the mcd scale is built on between-hole correlation of recovered cores, whose depths were estimated from the length of the drill pipe. For a coeval feature, the logging depths average 4 to 6 m deeper than the mbsf depths for Hole 1095B or 5 to 11 m deeper than depths from the mcd scale. In Table T16, we give the offsets between the two depth scales. These were determined by correlating the susceptibility data from whole-core measurements to the susceptibility data from the second logging run of the Geologic High-Resolution Magnetic Tool (GHMT), which includes a susceptibility measurement sonde. The correlation was done using the program AnalySeries (Paillard et al., 1996), which outputs the user-selected tie points for the signals being correlated. Depths between tie-points are adjusted through linear interpolation between bounding tie-points. We use the susceptibility data because it was collected on the GHMT logging runs, from which the logging magnetostratigraphy is derived, and was collected along the core, from which the core magnetostratigraphy is derived. We focus on only the second run of the GHMT logging tool, although both runs gave comparable results (Williams et al., Chap. 31, this volume). Furthermore, both Barker (Chap. 6, this volume) and Acton et al. (Chap. 5, this volume) used the susceptibility data to construct mcd scales because it was the data type most easily correlated between holes at a site. We found that distinctive susceptibility anomalies could be easily correlated between the core and logging data sets (Fig. F14). Finally, use of susceptibility allows us to do core-logging correlation that is independent of the magnetostratigraphy, which thus allows us to compare the core and logging results and confirm that they give compatible magnetostratigraphies. When both the magnetic logging data and the paleomagnetic inclination data are plotted in the mcd scale, the excellent correlation between them is obvious (see Figure F9 of Iwai et al., Chap. 36, this volume).
Below, we discuss our interpretation, working downhole. We focus the discussion on changes to the magnetostratigraphy presented in the Leg 178 Initial Reports volume (Shipboard Scientific Party, 1999a) and on complicated or interesting intervals. We note in Table T1 those intervals where the interpretation is similar or identical to that of the Shipboard Scientific Party (1999a).
Using the revised magnetostratigraphy, we compute the sedimentation rates between the identified reversals. These are included in Table T1 and are shown graphically in Figure F15. As noted by Barker, Camerlenghi, Acton, et al. (1999), sedimentation rates increase downhole. Average rates were ~20 m/m.y. at the top of the hole and >100 m/m.y. near the bottom of the hole.