Approximately 1100 discrete minicores were extracted from the "A" holes at the five sites, the "C" hole extension of Site 1050, and the Cretaceous/Paleogene boundary interval in the "C" hole of Site 1049. Sampling density averaged about two minicores per core section. Soft sediment samples were taken using oriented plastic cylinders (~10 cm3) followed by extrusion, trimming, and air drying. Indurated sediment minicores were drilled into the cut face of the working half of the core with a water-cooled nonmagnetic drill bit attached to a standard drill press. Minicore orientations are perpendicular to the axis of the core, and therefore horizontal relative to bedding strata. Some of the advanced hydraulic piston cores (denoted by the suffix "H" in the core name) had attempts at downhole orientation relative to magnetic north using a tensor tool mounted on the core barrel, but these attempts were generally unreliable. Therefore, declinations of magnetization of the minicores are random, although stability or progressive rotation of the magnetic declinations during demagnetization can be used to identify characteristic magnetic polarity.
Shipboard pass-through cryogenic magnetometer measurements were made at ~5-cm intervals on all cores from all holes using alternating-field demagnetization at 10-20 mT. Even though these shipboard measurements were commonly distorted by coring recovery artifacts, such as biscuiting slurry, and were generally inadequate to remove magnetic overprints, the main polarity patterns were often remarkably indicative of the detailed demagnetization results from the suites of minicores.
Paleomagnetic analyses of the suites of minicores were performed at the University of Oxford and at the University of Michigan. Progressive thermal demagnetization was tailored to each facies based on pilot studies of each lithology from each site, but it generally consisted of 30°C increments from ~100°-360°C, with continuation to higher thermal steps for the more stable samples. Bulk susceptibility was generally monitored after every second thermal demagnetization step <300°C, and after every higher temperature step to monitor changes in magnetic mineralogy and onset of viscous magnetization associated with formation of new magnetic phases. Magnetic directions were measured on a cryogenic magnetometer isolated from the Earth's magnetic field in a shielded room (Michigan) or with Helmholtz coils (Oxford). The practical limit on resolution of the natural remanence of the minicores was ~5 × 10-6 A/m, and duplicate measurements were taken whenever the intensity was <2 × 10-5 A/m.
Even though magnetic characteristics varied greatly among the various lithologies, it is possible to summarize the typical behavior. Initial natural remanent magnetization (NRM) directions are dominated by a downward inclination, which is presumed to be the combined secondary overprints of normal polarity oriented parallel with the present dipole field and a drilling-induced remanence. The drilling-induced overprint has always plagued Deep Sea Drilling Project (DSDP) and ODP paleomagnetic studies and is characterized by a downward (high positive inclination) and radially inward orientation (e.g., the series of investigative studies made by the paleomagnetists during Leg 154, as summarized in successive site chapters in Curry, Shackleton, Richter, et al., 1995). This radial-inward overprint is exhibited as an apparent preferential declination toward 0° ("north"); therefore, relatively few samples displayed initial NRM declinations between 100° and 260°, in contrast to the expected random distribution.
Secondary overprints were largely removed upon heating above 180°C, and dual polarity was generally observed from 200° through 360°C. Upon heating >360°C, most samples displayed a rapid increase in susceptibility accompanied by viscous magnetization. Despite performing the thermal demagnetization and measurements within a shielded room, the onset of spurious viscous magnetizations generally rendered continued demagnetization to be fruitless. A minority of samples, generally from intervals with a more condensed accumulation or a reddish colored facies, could be demagnetized at temperatures >360°C without displaying viscous magnetizations (Fig. F1).
The general magnetic behavior during thermal demagnetization suggests that a common carrier of characteristic magnetization (generally in the 200°-400°C range) is magnetite. The surge in susceptibility above 360°C observed in many samples is probably due to a combination of dehydration of iron-rich clays, oxidation of iron sulfides, and minor contributions from the alteration of iron minerals in the organic-rich claystone by organic combustion.
Polarity interpretation of the 180°-350°C magnetization component was generally obvious from the attained characteristic inclinations and relative rotation of declinations. Most samples either attained a direction that remained stable through multiple heating steps as the intensity of magnetization decreased (i.e., univectoral decay to the origin on the vector diagrams) or displayed development of a quasi-stable direction and intensity prior to the onset of erratic magnetization at higher temperatures. However, for a few samples, a stable end-point characteristic direction was not achieved prior to the onset of viscous magnetization and it was necessary to deduce polarity from their progressive demagnetization trends. Characteristic magnetization directions and associated variances were computed for each sample by applying a least-squares three-dimensional line fit (procedure of Kirschvink, 1980) to sets of vectors displaying removal of a single component in equal-area and vector plots of the progressive demagnetization.
Each characteristic direction was assigned a polarity rating based upon the individual demagnetization behavior: (1) well-defined N or R directions computed from at least five vectors having a high degree of linearity and a univectorial trend toward the origin (e.g., Figs. F1B, F1C), (2) less precise NP or RP directions computed from three to four vectors having a high degree of linearity and a univectorial trend toward the origin (e.g., Fig. F1A) or from at least five vectors displaying a "noisy" linear trend, (3) NPP or RPP samples that did not achieve adequate cleaning during demagnetization and were omitted from paleolatitude computations, and (4) samples with uncertain N?? or R?? or indeterminate INT polarity that were generally not used to define polarity zones. To reduce the bias of a single observer, selection and rating of characteristic magnetization vectors and associated polarity interpretations were done independently by two people and the more conservative assignment was generally used; however, the minor differences in assigning ratings did not affect a common identification of polarity zones. Of the 1075 samples, 291 were rated N or R and 224 were NP or RP.
Assignment of polarity chrons of the standard magnetic polarity time scale to these stratigraphic polarity zones required biostratigraphic control, as will be summarized below for each site. In many cases, delimitation of polarity zone boundaries was further constrained by the high-resolution shipboard measurements with the long-core cryogenic magnetometer; however, the quality of this shipboard data was commonly compromised by a difficulty in removing secondary overprints through alternating-field demagnetization and a high degree of noise from drilling artifacts, as summarized in the Leg 171B Initial Reports site chapters.
Assignment of polarity chrons utilizes the pattern of normal- and reversed-polarity zones within each interval and the reference scales for biomagnetochronology of nannofossil and planktonic foraminifer datums compiled for the Paleogene (Berggren et al., 1995) and the Late Cretaceous (Erba et al., 1995) as compiled in the "Biostratigraphy" section of the "Explanatory Notes" chapter of the Leg 171B Initial Reports volume (Shipboard Scientific Party, 1998a) (Fig. F2). The biostratigraphy for each hole has undergone progressive refinement, and the columns for each site in this study are based upon the compilations in the Leg 171B Initial Reports volume (Norris, Kroon, Klaus, et al., 1998) with modifications to the Campanian-Maastrichtian interval presented by Brian Huber (foraminifers) and Jean Self-Trail (calcareous nannofossils) at the 1998 postcruise conference. In turn, the magnetostratigraphy has indicated inconsistencies both with some portions of the reference compilations of biomagnetochronology and with the assigned biostratigraphy at different sites. This iterative procedure of improving biostratigraphic calibrations and possible reinterpretation of the magnetostratigraphy and reassignment of polarity chrons is a common process before developing a robust biomagnetic chronology.
Paleolatitudes for successive time intervals at each site were derived from mean inclinations computed using the procedure of Kono (1980a, 1980b) for calculating statistics of inclination-only data from unoriented vertical drill cores. This method uses the mean and standard deviations of the sines of the inclinations to compensate for the circular Gaussian (Fisherian) distribution of paleomagnetic vectors. A simple mean of the inclinations gives unrealistic importance to the lower values. Kono's nonlinear simultaneous equations relating the true mean inclination and the circular dispersion parameter, K, to the statistics of the sines of the inclination data were solved using Newton's method to converge on the solutions. Samples having characteristic directions rated NP or RP were given one-half weight. To test the validity of Kono's procedure, we submitted previously analyzed sets of data from outcrops; the resulting mean inclination is within 0.1° of the inclination given by statistics on such directional data (procedure modified from Fisher, 1953). McFadden and Reid (1982) independently developed a different computational procedure to calculate mean inclinations from azimuthally unoriented paleomagnetic data, but the similarity of results suggests that it has the same mathematical basis as Kono's procedure.
The magnetostratigraphy, paleolatitudes, and any significant departures from the general magnetic behavior at each site are summarized in the following sections from the most seaward (Site 1049) to the most landward (Site 1053).