Paleomagnetic and rock magnetic measurements at Site 1203, which is located in the summit region of Detroit Seamount, were aimed at assessing the nature of the natural remanent magnetization (NRM) of basement rock and obtaining a preliminary estimate of the paleolatitude of the site during basement formation (~75-76 Ma, as estimated by nannofossil biostratigraphy) (see "Biostratigraphy"). Basalt and sediment discrete samples (minicores) were used for measurements. In addition, the magnetization of sediment recovered above basement was measured to define geomagnetic polarity. For magnetostratigraphic analysis, half-round core sections were analyzed. We start by describing these magnetostratigraphic measurements, analyses, and results. We follow this with a presentation of rock magnetic and paleomagnetic analyses of basalt and sediment comprising the Site 1203 basement sequence.
Natural remanent magnetizations from archive half-round core sections from sedimentary units recovered above basement in Hole 1203A (Cores 197-1203A-1R to 17R) were measured using the shipboard 2-G Enterprises superconducting quantum interference device (SQUID) magnetometer. The sediments are mostly poorly consolidated oozes and chalks (see "Lithostratigraphy"). The measurement interval was 5 cm. Progressive alternating-field (AF) demagnetizations were applied to peak fields ranging from 20 to 40 mT. The effectiveness of this demagnetization varied throughout the core, as seen in orthogonal vector plots of the decay of NRM with applied field for select measurement intervals (Fig. F58). From some intervals, the NRM was found to be composed of an apparent Brunhes-age component overprinting a reversed characteristic remanent magnetization (e.g., Fig. F58).
Inclinations derived from vector end points after AF demagnetization were used to assign geomagnetic polarity (Fig. F59). Because the core sections analyzed were derived from rotary drilling, a few caveats are needed before further considering even this preliminary polarity interpretation. Rotary drilling of sediments can result in the recovery of long core sections with only minor disturbance. However, sometimes the disturbance is severe. If the sediment is not well lithified, zones (commonly called "biscuits") can rotate within the core liner. Unconsolidated sediment recovered by rotary drilling can also have a pervasive internal disruption. In general, a polarity was assigned only if several consecutive measurement intervals demonstrated a consistent polarity; single-point estimates suggesting a polarity opposite to that of adjacent intervals were ignored and assumed to reflect core disturbance (or insufficient demagnetization) in this preliminary analysis.
Correlation of the observed polarity intervals in the upper part of the sedimentary sequence (Cores 197-1203A-1R to 10R) to the geomagnetic timescale is hindered by limited recovery, core disturbance, and poor preservation of nannofossils (see "Biostratigraphy"). Using the available nannofossil data, a few preliminary correlations can be discussed for cores recovered from deeper in the section. Cores 197-1203A-11R to 13R have been assigned to the early Oligocene NP21 to NP22 nannofossil zone (see "Biostratigraphy"). These assignments suggest that the reversed polarity intervals identified in the same cores probably correspond to the polarity chron sequence between reversed polarity Chrons 12r and 13r (~32-34 Ma) (Berggren et al., 1995). Core 197-1203A-17R is assigned to the early Eocene NP12 nannofossil zone. This assignment suggests the polarity intervals identified in this core may correspond to Chron C23 (~52 Ma).
Overall, the relatively frequent magnetic reversals observed in the sedimentary cores recovered above basement are more consistent with an early Tertiary rather than a Late Cretaceous age. The latter time interval is characterized by a lower reversal frequency (and a correspondingly greater duration of individual polarity chrons) than that apparently recorded by the sediment cored in Hole 1203A above basement. This relative age assignment is consistent with biostratigraphic data indicating a significant hiatus between igneous basement and the overlying sediment at Site 1203.
The basement section at Site 1203 is composed of basalt flows and volcaniclastic sediments (see "Physical Volcanology and Igneous Petrology"). The sediments were well indurated, so samples for paleomagnetic and rock magnetic analyses were collected as drilled minicores or sawed cubes.
Low-field volume-normalized (bulk) magnetic susceptibility was measured on every sample collected from basement (after AF demagnetization) with a Kappabridge KLY-2 magnetic susceptibility meter. For each sample, measurements were taken three times and then averaged. In general, measurements on a single sample agreed to within 1%. Magnetic susceptibility values range from 0.2 x 10-3 to 66 x 10-3 SI (mean = 9.8 x 10-3 SI). The Koenigsberger ratio (Q) was calculated for each sample according to the following formula:
where,
Koenigsberger ratios range from 0.02 to 176 (mean = 15.4). Median destructive field (MDF) was calculated based on detailed stepwise AF demagnetization data used for constraining characteristic remanent directions (see below).
Figure F60 shows a plot of (bulk) volume magnetic susceptibility (K), the Q-ratio, and MDF vs. depth. The basalt samples measured generally have Q-ratios > 1.0. Basalt recovered between 650 and 720 mbsf is characterized by much higher Q-ratios and higher MDF values relative to other parts of the basement sequence drilled. These basalt samples correspond to basement Units 18-20 (see "Physical Volcanology and Igneous Petrology").
Basement Units 1 and 3 (~460-490 mbsf) also show high Q-ratios and high MDF values. Analyses of polished thin sections of these rocks using reflected-light microscopy reveal the presence of maghemite (see "Physical Volcanology and Igneous Petrology"). Whereas the grain size of the maghemite observed is far greater than the grains that are likely responsible for carrying a remanent magnetization, its presence indicates that caution is needed in interpreting the preliminary paleomagnetic results from these basement units.
A plot of magnetic susceptibility vs. NRM intensity is shown in Figure F61. Compared to the wide range in NRM intensities displayed by the sediment samples, the basalt samples show a rather narrow variation. Two sediments samples with high intensity are from volcanic breccia and hyaloclastite tuff. A plot of the log of bulk volume magnetic susceptibility vs. MDF is shown in Figure F62. Magnetic susceptibility data from basalt samples are clearly negatively correlated with MDF values, whereas the data from sediment samples are scattered. Basalt samples with lower MDF values (<10 mT) are from the visually fine-grained and fresh parts of the recovered core material, typified by Section 197-1203A-48R-1 to Core 197-1203A-49R (basement Unit 21). The low MDF values may reflect unoxidized titanomagnetite with a relatively large grain size. All the sediment samples with relatively high MDFs (>40 mT) are from intervals of volcanic breccia.
The Lowrie-Fuller test (Lowrie and Fuller, 1971) is based on AF demagnetization of strong-field and weak-field thermoremanent magnetization (TRM) and permits a rapid estimate of magnetic domain state in igneous rocks. Magnetic domain state can be used further to help constrain magnetic grain size. The Lowrie-Fuller test is based on the experimental observation that in large multidomain (MD) grains, strong-field TRM requires larger destructive fields than weak-field TRM to reach the same normalized remanence level. In practice, the strong-field TRM is usually represented by a laboratory-induced saturation isothermal remanent magnetization (SIRM) and the weak-field TRM is represented by an anhysteretic remanent magnetization (ARM). Given an unknown magnetic mineralogy, the test can be considered only as qualitative because changes in magnetic coercivity related to different magnetic minerals can mimic changes in magnetic grain size.
The Lowrie-Fuller test was performed on one or two samples from each basaltic flow unit. First, an ARM was produced by demagnetizing the sample in a 140-mT alternating field in the presence of a 30-µT direct-current field using the D-2000 AF demagnetizer. The ARM was then progressively AF demagnetized. An SIRM was then acquired in a 0.8-T field using an IM-10 impulse magnetizer then was subsequently AF demagnetized. All AF demagnetizations were conducted using the D-2000 AF demagnetizer, and all remanences were measured with the shipboard Molspin Minispin magnetometer.
In interpreting these data in terms of magnetic domain state, we rely on the relative shapes of the ARM and SIRM demagnetization curves. For some samples (e.g., Sample 197-1203A-26R-2, 29-31 cm [basement Unit 6]) (Fig. F63), the two demagnetization curves are similar and clear constraints on magnetic domain state are not possible. In other samples, the rapid decay of ARM relative to SIRM suggests the presence of MD grains (e.g., Sample 197-1203A-37R-2, 76-78 cm [basement Unit 16]) (Fig. F64). In still other samples, the resistance of ARM to demagnetization relative to SIRM suggests the presence of single- to pseudo-single-domain grains (e.g., Samples 197-1203A-42R-5, 65-67 cm [basement Unit 19]; 52R-2, 101-103 cm [basement Unit 23]; 61R-1, 131-133 cm [basement Unit 26]; and 67R-1, 18-20 cm [basement Unit 30]) (Fig. F63).
In general, the Q-ratios, median destructive field values, and results of the Lowrie-Fuller tests indicate that most of the recovered basalt and volcaniclastic sediments are capable of preserving primary magnetizations suitable for paleolatitude analyses. The only potential exception is represented by basalt from basement Units 1 and 3. The potential maghemitization of remanence carriers suggested by reflected-light observations and its potential effect on the remanence record should be addressed in shore-based studies (see "Discussion").
Minicore samples of basalt and minicore or cube samples of the volcaniclastic sediments were measured with the 2-G Enterprises SQUID magnetometer and the Molspin Minispin magnetometer. After the measurement of the untreated NRM, samples were progressively demagnetized by AF demagnetization. In general, a 5-mT field increment was used and the peak field applied was 80 mT. Paleomagnetic measurements collected after each step form the basis of the data set described below.
Most basalt samples show an extremely well-defined, simple demagnetization pattern when viewed on orthogonal vector plots (Fig. F64). After the removal of a low-coercivity (<10 mT) viscous component of magnetization, a stable single component of magnetization, identified as the characteristic remanent magnetization (ChRM), was defined. A steep overprint component, similar to that reported in some paleomagnetic studies of ODP cores and thought to be imparted by the drill string, was not observed. Volcaniclastic sediment samples showed a similar behavior (Fig. F65), although sometimes with less apparent magnetic stability.
A few samples displayed a more complex behavior, signaling potential inadequacies of AF demagnetization in isolating the characteristic remanent magnetization. In a few basalt samples, the demagnetization data defined a linear trend that bypassed the origin of an orthogonal vector plot, indicating the presence of high-coercivity magnetic minerals (Fig. F66). In a few other cases the AF treatment failed to reduce the magnetization to <10% of the untreated NRM. Sediment samples sometimes exhibited more complex behavior as well. This included the failure of the demagnetization to isolate a stable component with a trend to the origin. Whereas high-coercivity phases may be partly responsible for this behavior, resolution limits of the shipboard magnetometer may have contributed to the observed scatter.
Samples were collected from the coarse volcaniclastic sediment/breccia of basement Unit 13 to further assess the nature of the NRM. These samples can be used to conduct a variant of the conglomerate test (Graham, 1949). In the classical test, the NRM of individual clasts composing a conglomerate is measured. Magnetization directions from these clasts are expected to be random if the conglomerate and adjacent rocks have not experienced a thermal or chemical event (or both) after formation sufficient to cause remagnetization.
In our case, individual clasts were not sampled but each minicore was composed of numerous relatively large (~0.1-0.25 cm) basaltic clasts separated by a calcite matrix. Because the drilled cores are azimuthally unoriented, we must rely on an analysis of the components seen in the orthogonal vector plots. These plots show that in each sample the magnetization is stable during AF demagnetization but that several magnetic components are often recorded (Fig. F67). In general, the samples show a relatively high coercivity, and it is possible that a mineral such as hematite carrying a chemical remanent magnetization contributes to the total NRM. However, the magnetic components that can be isolated by AF demagnetization are not consistent between minicores, indicating that the breccia and other basement section rocks likely did not experience a remagnetization event.
ChRM directions for basalt and sediment were fit using principal component analysis (Kirschvink, 1980). In general, the ChRM directions were defined between 15 and 70 mT for basalt samples and maximum angular deviations of the line fits were <3°. For the volcaniclastic sediment samples, the ChRM directions were generally isolated between 15 and 50 mT and the maximum angular deviations of the fits were generally <5°. Of 202 basalt samples demagnetized and measured, 199 yielded ChRM directions. The three rejected basalt samples had demagnetization data that did not trend to the origin of orthogonal vector plots. Of the 56 sediment samples measured, 34 yielded ChRM directions useful for constraining paleolatitude. Seven sediment samples from volcanic breccia used for the modified conglomerate test were not considered. Data from 15 samples failed to show stable magnetization during AF demagnetization and were also not considered.
Because the magnetization of a standard paleomagnetic sediment sample is thought to have been locked in over a time interval much greater than that of a single lava flow, we give equal weight to each sample in the following analysis. Applying the inclination-only averaging procedure outlined by McFadden and Reid (1982), the 34 paleomagnetic inclinations from the characteristic remanent magnetizations derived from the volcaniclastic sediments have a mean value of 54.7°, an estimated precision parameter (k) = 24.8, and a 95% confidence region = +3.1°/-6.4°. If these data are accurate recorders of the time-averaged geomagnetic field, the results suggest a paleolatitude of 35.2° (95% confidence region = +3.2°/-5.9°). These data form a distribution that is offset by ~16°N from the latitude expected if the Hawaiian hotspot had remained stationary in the mantle over the last 75-76 m.y. (Fig. F68).
ChRM inclinations derived from basalt samples were averaged using inclination-only statistics (McFadden and Reid, 1982) to obtain lava flow means. Sixteen of the eighteen lava flows recovered from Site 1203 were sampled and yielded flow mean inclinations (Table T9). Assigning equal weight to each flow, these data yield a mean inclination of 48.0° (95% confidence interval = ±8.4°), which suggests a mean paleolatitude of 29° (95% confidence interval = +6.3°/-7.8°). The latter value assumes the directions are adequately represented by AF demagnetization, secular variation is averaged by the section, and extremely short intervals between some lava flows do not cause a bias in the mean (see "Discussion"). The paleolatitude average is 10° farther north than that predicted by a fixed hotspot reference frame (Fig. F69). Statistics of inclination averages from basalt are summarized in Table T9.
We compare the distribution of inclinations derived from the volcaniclastic sediments with a synthetic Fisher distribution (Fisher, 1953) having the same mean inclination and precision parameter (k) as displayed by the experimental data in Figure F70. A similar comparison between the flow-mean basalt inclinations and a synthetic Fisher distribution is seen in Figure F71. Inclination values from the volcaniclastic sediments appear to have a distribution that suggests the mean represents an adequate sampling of a hypothetical parent distribution. However, the magnetization of coarse sedimentary rocks such as some of the volcaniclastic rocks can be affected by detrital inclination error (King, 1955). Furthermore, compaction can induce inclination shallowing in finer-grained marine sediments (e.g., Arason and Levi, 1990; Tarduno, 1990). Paleolatitudes determined from paleomagnetic analyses of the volcaniclastic sediments should therefore be considered to be minimum values. An effect of inclination shallowing may be present in the group of inclinations with values <45°.
The distribution of paleomagnetic inclinations derived from Hole 1203A basalt seems to represent the range of values typical of a model Fisher distribution. Values near the mean, however, appear to have been undersampled. There is also a hint of potential serial correlation in the data with inclinations <45°. The overall estimated dispersion of the data (S = 17.7°) (Table T9) is slightly higher than that expected from global lava flow data spanning the 45- to 80-Ma interval (McFadden et al., 1991).
Another consideration is the possibility that the characteristic remanent magnetizations isolated by AF demagnetization in basement Units 1 and 3 do not accurately represent the field direction during basalt cooling because of later alteration (signaled by the occurrence of maghemite identified by reflected-light optical microscopy). To investigate the potential effect that inaccurate field values from these units might have, we calculated a new mean, excluding basement Units 1 and 3 (Table T10). The average inclination of 50.0° (95% confidence interval = ±9.0°) is slightly steeper than that calculated with Units 1 and 3 included and may be a more conservative estimate, given the rock magnetic uncertainties.
Overall, the sediment and basalt inclination means discussed above represent a paleolatitude range between 29° and 35°N. These values are within the uncertainty region of the paleolatitude of 36°N (95% confidence interval = ±7°) derived from analyses of older basalt (81 Ma) recovered at ODP Site 884 on the eastern flank of Detroit Seamount (Tarduno and Cottrell, 1997). The Hole 1203A data, however, are preliminary and must be confirmed with detailed shore-based rock magnetic characterizations and thermal demagnetizations to isolate remanence components.