DISCUSSION

Several assumptions were made during the averaging of the paleomagnetic data that could change our results, so it is prudent to assess their impact. First, inclination values from 10 samples were not used in the flow averages (Table T1). These values were discarded because they appeared to be outliers when compared with other data from a given flow or flow group. They typically differed from the other samples in a given flow by >2 , frequently having an inclination sign opposite that of the other samples. It was not typical for such samples to display poor demagnetization consistency, so the discordant inclination values cannot be explained simply by samples with poor paleomagnetic characteristics. Those samples with inclination absolute values similar to those of adjacent samples, but which show a difference in sign, may be explained as samples inverted during curation or handling of the cores. This explanation does not work for samples with greatly discordant inclinations, for example, Sample 191-1179D-12R-5, 14-16 cm, which gave an inclination of 52.7° despite a relatively low MAD value of 10.7° (Table T1). Such discordant results can occur when rock fragments fall from the borehole wall and are cored out of position and orientation. Discarding such errant points as outliers seems appropriate and is standard practice in paleomagnetic studies. Including the discordant measurements would not change the mean colatitude significantly but would unnecessarily enlarge the estimated error bounds.

Five flow groups that have samples with positive inclinations were assumed to be measurements of the same reversed polarity field as the other flow groups rather than polarity reversals. At higher paleolatitude sites, where there is little chance that secular variation would cause the difference in sign, flows that give an inclination with opposite sign are sometimes assumed to record a magnetic reversal. In contrast, at a low paleolatitude site changes in inclination sign caused by secular variation are expected. The latitude variance at the equator in models of secular variation based on paleomagnetic data is ~9° (Harrison, 1980; McElhinny and McFadden, 1997), far greater than the 1.9° average paleolatitude for Site 1179. In addition, the differences in sign are unlikely to result from overlapping lava flows of opposite polarities. Site 1179 is located near the middle of 37-km-wide Anomaly M8 (Shipboard Scientific Party, 2001), and because of the near 180° skewness of the magnetic lineations (Nakanishi et al., 1989; Larson and Sager, 1992) the anomaly is located directly over the reversed polarity source body. If the assumption that the entire section is reversed polarity is incorrect, the mean paleolatitude may be slightly greater. Changing the polarity of flow units with positive inclinations and reaveraging yields a mean paleocolatitude of 84.2°, a change of 4.1°. Furthermore, if the assumed polarity is incorrect, the implied normal polarity paleolatitude is 1.9°S.

Another subtle assumption made in averaging the data is that there is no overlap in cored flows, despite an apparent overlap of 2.6 m between the bottom of Core 191-1179D-12R and the top of 13R (Table T1). This discrepancy resulted from ODP core curation convention (which "expands" fractured hard rock core specimens) and the fact that Core 191-1179D-13R recovered 0.9 m more core than the estimated interval drilled. Although the cause of actual 0.9 m overlap between these two cores is unclear, it is doubtful that it could have been caused by drilling that resumed above the bottom of the hole (to cause overlap), so the assumption appears justified.

When calculating a paleocolatitude using a small number of independent measurements of the paleomagnetic field, the result may be inaccurate if secular variation is not properly averaged (Cox and Gordon, 1984). Despite the consistency of the Site 1179 colatitude with other mid-Cretaceous Pacific paleomagnetic data (Fig. F5), the statistics of our calculation suggest some uncertainty owing to insufficient secular variation averaging. We estimate that only 13 independent magnetic units were sampled, a number that would be considered inadequate to average secular variation in most paleomagnetic studies. Although the total colatitude variance is 14.6°, much of this scatter arises from measurement error and between-group colatitude variance is only 4.8°. This value is only half the expected 9.0° from secular variation models (Harrison, 1980; Cox and Gordon, 1984). This difference suggests that Site 1179 samples may not completely average secular variation. Nevertheless, because of the method used to determine the error bounds, which used an estimate of colatitude scatter from a secular variation model (Cox and Gordon, 1984), the calculated confidence limits are not biased by the lack of secular variation sampling.

Despite the cautions above, comparison of the Site 1179 paleocolatitude with other Pacific basalt core data of mid-Cretaceous age shows that our results are in good agreement (Fig. F5). The colatitude arc for the site goes through the middle of the distribution of other colatitude arcs of mid-Cretaceous age. Furthermore, the 125-Ma paleomagnetic pole calculated mainly from magnetic anomaly skewness (Petronotis and Gordon, 1999) falls within the error bounds of the colatitude.

The 1.8°N paleolatitude calculated from Site 1179 basalts implies that the site has drifted 39.3° northward since 129 Ma, a result consistent with other paleomagnetic data of similar age. Not only do other Pacific basalt cores (many with even fewer independent magnetic units) give similar results (Fig. F5), but the data are also consistent with seamount (Sager and Koppers, 2000) and skewness data (Larson et al., 1992; Petronotis and Gordon, 1999). This agreement suggests that the paleocolatitude is reliable.

In addition to the paleolatitude information, the Site 1179 basalt data also give some insight into the construction of the igneous section. Colatitude values from flows in the bottom part of the section show markedly less variation than those above, implying the lower section was emplaced rapidly, with little time between flows. The flows that exhibit little between-flow scatter, combined into unit 13 (Fig. F4), are those that belong to the distinct geochemical Group III, the olivine-rich flows (Shipboard Scientific Party, 2001). In contrast, flows in the upper part of the section exhibit large changes between successive flow groups, indicating significant time gaps. Along with the change in geochemistry (Shipboard Scientific Party, 2001), this may indicate a shift in the magma source or mode of flow emplacement.