Most samples displayed univectorial decay toward the origin of an orthogonal vector plot with little or moderate scatter in direction between successive demagnetization steps (Figs. F2, F3). A total of 89 (73%) samples gave MAD angles of <10° (Table T1), implying well-determined characteristic magnetization vectors. It was typically possible to use five to six demagnetization steps (range = 3-9) to define the characteristic remanence direction (Table T1), another indication of demagnetization consistency. Only eight samples were so erratic that they gave MAD angles >30°. Most of these samples are present in flows 7 and 8, suggesting a link to flow characteristics.
A steep, downward-directed overprint, like that often attributed to an isothermal remanent magnetization (IRM) imparted by the drill string (e.g., Roberts et al., 1996), was present in many samples. This overprint was usually removed by thermal demagnetization of >100°-300°C or AF demagnetization in fields >15-25 mT. Many samples treated with AF demagnetization did not show decay toward the orthogonal vector plot origin at high field steps (Fig. F2), a behavior that probably results from an anhysteretic remanent magnetization (ARM) imparted by imperfectly balanced AF demagnetization coils. These samples were useful nevertheless because they typically displayed univectorial decay toward the plot origin at lower field values, which were used to determine the characteristic magnetization direction.
Orthogonal vector plots show that characteristic magnetization directions usually have small inclination values (Figs. F2, F3; Table T1), consistent with formation of the basalt section near the equator. Negative inclinations were measured from 95 samples (78%), implying that most of the section records an upward-directed magnetization. Positive inclination samples were sometimes grouped (as in flows 3, 8, 9, 11, and 21), and sometimes solitary. The solitary positive inclination data may result from inverted samples or erroneous inclinations. Single discordant samples were found in flows 9, 15, 20, 22, 24, and 45, whereas two discordant samples were found in flow 44 (Table T1). Because the source of the discrepancy is not known, these data were not used in averaging calculations. The groups of positive inclination values likely indicate flows with coherent downward-directed magnetizations. Although flows 3, 9, and 21 have predominantly positive inclination values, flows 8 and 11 each contain two groups of opposite sign but with equal numbers of samples (Table T1). Because the two groups in each of these flows show consistent inclinations, we think that the changes in sign are real and result from shipboard scientists failing to recognize a flow boundary. Thus flows 8 and 11 were each subdivided into two separate subunits (Table T1).
Combining data to calculate flow mean colatitudes produced 23 statistically distinct flow groups (Fig. F4). As a consequence of large interflow changes in colatitude above 400 mbsf, most flows above this depth appear to be independent units. In contrast, the lower section shows little variation and adjacent flow mean inclinations are often statistically distinct only as a result of low within-flow inclination scatter. Because flows 23-47 show little variation, we concluded they are serially correlated and their inclinations were averaged. This required us to ignore single inclination measurements from two flows (33 and 43) that appear to break the sequence. Although these flows may be independent units and represent actual breaks in the lava succession, with only one sample apiece the flow inclination values are not robust. We also ignored the single discordant inclination from flow 6 and averaged data from flows 5 and 7, which have similar means (Fig. F4; Table T1).
Averaging the 13 inclination groups yields a mean colatitude of 88.1° ± 6.8° (this and other error limits are 2 
, approximating 95% confidence), assuming that the section is entirely reversed polarity and that the changes in inclination sign result from secular variation rather than polarity reversals. This paleocolatitude implies a paleolatitude of 1.9°N ± 6.8°. The average colatitude is not highly sensitive to the choice of flow groupings. For example, an average of all 23 statistically distinct flow groups yields a paleocolatitude of 87.5°.
