Two techniques were used during Leg 199 to determine the bulk density of core samples (Lyle, Wilson, Janecek, et al., 2002). Sediment mass and volume measurements were used to determine the wet bulk density of discrete samples of ~10 cm3 volume. These samples were routinely collected at ~1.5-m intervals. The bulk density of whole-core sections was determined using the gamma ray attenuation (GRA) densiometer on the multisensor track (MST). Measurement of wet bulk density by this device is based on the principle that the attenuation of a collimated beam of gamma rays passing through a known volume is proportional to the material density (Evans, 1965). The measurement width of the GRA sensor is ~5 mm, and the sample spacing was generally set at 4.0 cm for Leg 199 measurements. The narrow width of material sensed and the close sample space allow high-frequency density variation to be recorded; however, it also is accompanied by increased noise in the data. Initial steps in the processing of the GRA records include removing extraneous values and smoothing.

Calibrating the GRA device assumes a two-phase system consisting of minerals and interstitial water (Blum, 1997). An aluminum cylinder represents the mineral phase in the calibration standard. The density of aluminum, 2.70 g/cm3, is a reasonable approximation for many silicate and carbonate sediment constituents, but it is significantly greater than the density of sediments with appreciable abundances of siliceous components, smectite clays, and zeolites. The average grain density of radiolarian ooze drilled during Leg 199 is ~2.2 g/cm3. The density of pelagic clays recovered during Leg 199 is more variable, but much of the clayey sediment recovered has average grain densities between 2.4 and 2.6 g/cm3. As a result of the presence of materials with grain densities lower than that assumed by the calibration method, the GRA density typically was less than the wet bulk density determined for the discrete samples. Incompletely filled core liners can also contribute to low GRA densities. Based on observations of split cores, this factor was judged to be not as significant as grain density differences, particularly for advanced piston corer (APC) cores which comprise the bulk of the Leg 199 cores. Intervals of obviously low GRA densities resulting from incompletely filled cores were removed during data editing. At each of the eight sites drilled during Leg 199, a regression analysis of the wet bulk density and GRA density was performed (Lyle, Wilson, Janecek, et al., 2002). The equations generated by these analyses were applied to the GRA data as a correction in order to make the GRA densities better approximate the wet bulk density of the discrete samples. The agreement between the wet bulk density and the corrected GRA density, which is shown in the depth profiles for the eight sites (Fig. F1), is very good. The bulk density in the clay and radiolarian ooze intervals in the profiles typically is very uniform. A higher-frequency variation in bulk density, particularly as recorded by the GRA density, characterizes the nannofossil ooze. The variation of the density in the nannofossil ooze typically follows an alternating lithology resulting from varying abundances of clay and radiolarians.

The crossplot of wet bulk density and corrected GRA density shows a strong correlation between the two density measures (Fig. F2). The R2 value for the regression of GRA density with wet bulk density is 0.91. The crossplot also shows a separation by lithology. The density of the nannofossil ooze and chalk, as a group, is higher than that of the radiolarian ooze and clay. Although the correlation of the two density measures is high, the trend of the regression line shown in Figure F2 indicates that, despite the correction, the GRA density is less than the wet bulk density of the discrete samples, particularly in the higher-density calcareous sediments.

Despite significant differences in average grain density, variation in wet bulk density primarily is a function of porosity, as displayed by the linear relationship in the crossplot of porosity and wet bulk density (Fig. F3). The regression of wet bulk density with porosity has an R2 value of 0.97. The effect of the differences in grain density is evident in the density-porosity crossplot. At a given porosity value, the density of a radiolarian ooze sample generally is less than that of samples of clay or nannofossil ooze.


Three techniques were used to determine the compressional wave velocity of core samples (Lyle, Wilson, Janecek, et al., 2002). The P-wave logger (PWL) on the MST was used to measure the horizontal velocity in whole-core samples. The PWL determines the velocity using two transducers aligned perpendicular to the core axis and by measuring the traveltime of a sonic pulse through the sediment and core liner. The velocity of APC cores was determined at a 2-cm spacing with the PWL. The use of the PWL on extended core barrel (XCB) cores was limited by poor acoustic coupling between the sediment and core liner. Velocity measurements were made on split core sections using the insertion and contact probe systems at a spacing of ~1.5 m. The interval for which the velocity was measured coincides with the interval sampled for wet bulk density. The insertion probe system, the successor to the Digital Sonic Velocimeter, was used on soft sediments and consists of two pairs of transducers that are pressed into the sediment, allowing determination of the vertical and horizontal velocity. The depth range over which the insertion probe system was used varied from 0 in radiolarian ooze to depths of 50 meters below seafloor (mbsf) in more clayey sediment. The contact probe system, the successor to the Hamilton Frame, was used on stiff sediments and rocks. The system consists of a fixed and a moveable transducer and can be used to measure the velocity of a split core in its core liner or the velocity of a specimen extracted from the core. Most measurements made with this system during Leg 199 were made with the sediment in the core liner. Velocity measurements of samples outside the core liner were only made for very stiff sediment or rocks that could be cut from the core.

The agreement between the PWL velocity and the velocities determined by the insertion and contact probe techniques, as shown in the velocity profiles for the Leg 199 sites (Fig. F4), is not as good as the agreement between the GRA density and the discrete sample wet bulk density. The best correspondence among the different velocities typically occurs in the shallower depths of the holes. At greater depths, the differences between the PWL velocity and contact probe velocity increase for clay and radiolarian ooze. The agreement between the different velocities is better for nannofossil ooze. The crossplot of the insertion and contact probe velocity and the PWL velocity (Fig. F5) shows a weak correlation between the discrete sample and MST measurements. The R2 value for the regression is 0.59. The trend displayed by the data is that PWL velocities are less than the insertion and contact probe velocities, particularly at higher values.

The extent to which velocity variations are controlled by differences in lithology is shown in the crossplot of wet bulk density and velocity (Fig. F6). Low wet bulk density characterizes clays (1.1–1.5 g/cm3) and radiolarian oozes (1.1–1.4 g/cm3), and a relationship between velocity and bulk density is lacking. The velocity of the radiolarian oozes (1500–1590 m/s) is slightly higher than that of the clays (1475–1575 m/s). Nannofossil ooze and chalk are characterized by a greater range in wet bulk density and velocity. Most of the nannofossil ooze samples have bulk densities between 1.15 and 1.75 g/cm3 and velocities ranging from 1480 to 1560 m/s. Six chalk samples were collected with densities between 1.8 and 2.2 g/cm3 and velocities from 1600 to 2050 m/s. The pattern of variation in velocity with bulk density in the nannofossil ooze is consistent with that demonstrated for calcareous sediments in previous studies (Mayer et al., 1985; Jarrad et al., 1993). A slight increase in velocity with density characterizes nannofossil oozes with densities less than 1.75 g/cm3 (porosities greater than 55%). The sharp increase in velocity in the limited higher-density chalk samples follows the pattern demonstrated for lower-porosity calcareous sediments.

Factors Influencing Density and Velocity

Compositional factors affecting the bulk density and velocity were examined using the mineralogical and bulk chemistry data generated during Leg 199. These data consist of (1) calcium carbonate (CaCO3 in weight percent), determined with a Coulometrics 5011 carbon dioxide coulometer; (2) calcite, opal, illite, and smectite percentages, estimated by light absorption spectroscopy (LAS) (Vanden Berg and Jarrad, 2002); and (3) bulk sediment geochemistry determined by inductively coupled plasma–atomic emission spectroscopy (ICP-AES). Sampling frequency for the composition factors was at least one sample per core for CaCO3 analyses, two samples per section for LAS analyses, and for ICP-AES analyses, one sample per section for Sites 1215 through 1217 and three samples per core for Sites 1218 through 1222. The discrete sample densities and velocities were used in the comparison with mineralogy and bulk chemistry data because the samples for these measurements are either the same (LAS) or immediately adjacent to (CaCO3 and ICP-AES) samples used in the compositional analyses.

The CaCO3 content of the sediments has a significant influence on bulk density and less influence on velocity. The regression of wet bulk density with CaCO3 has an R2 value of 0.73. The crossplot of wet bulk density and weight percent CaCO3 shows that the relationship between density and CaCO3 depends on lithology (Fig. F7). Density variation in clay is not a function of CaCO3 content. Excluding the clays and considering only the biogenic sediments, the regression of wet bulk density with CaCO3 improves slightly, with R2 = 0.79. The chalk samples with bulk densities higher than 1.8 g/cm3 fall off the trend for density and CaCO3 regression. Loss of intraparticle porosity and reduction of porosity resulting from compaction and cementation may be the cause of the higher density in these materials. Inclusion of depth in the regression of wet bulk density with CaCO3 for all of the samples has a minimal influence, increasing the R2 value to 0.74. The crossplot of velocity and weight percent CaCO3 (Fig. F8) shows that velocity does not vary as a function of CaCO3 content.

The relationship between wet bulk density and velocity and LAS mineralogy was examined by stepwise multiple regression. Stepwise regression adds one variable at a time to the regression model based on its contribution to explaining the variance in the data. Variables not included in the final model are determined by the procedure to not contribute significantly to explaining the variance. In the regression of density with LAS mineralogy, calcite is the first variable entered into the model, with R2 = 0.73 (Table T1), which matches the results of the regression with weight percent CaCO3. The variables opal, depth, and illite, in that order, were subsequently added in the model, with a final R2 value of 0.87. In the multiple regression of velocity with depth and the LAS data, depth is the first variable entered, followed by opal and illite (Table T2). The variables in the regression model do a poor job of explaining the variance in velocity. The R2 value for the model is only 0.25.

Stepwise multiple regression was also used to explore the relationship between wet bulk density and velocity and the bulk sediment geochemistry data. Elements in the bulk sediment data set, in weight percent, include Si, Al, Ti, Fe, Mn, Ca, Mg, P, Sr, and Ba. In the regression of wet bulk density with the geochemistry data, Ca is the first variable entered in the model (R2 = 0.63) (Table T3). The R2 value for the overall regression model is 0.87. The association of Ca and calcite is obvious, and its importance in the regression is consistent with the regressions with CaCO3 and LAS mineralogy. Subsequent variables entered in the model largely follow the pattern set in the regression of wet bulk density and LAS data, where the variable order is calcite, opal, depth, and illite. The second variable in the geochemistry model, Ti, occurs in greater abundance in siliceous sediments and clay. Depth, as in the LAS regression, is the third variable entered. The last variables entered in the model, Al and Mg, are those found in greater percentages in clay. The correlation of velocity and the bulk geochemistry data is weak (Table T4); however, the R2 value of the regression model, 0.47, is higher than that for the model with the LAS data. As in the velocity-LAS regression, depth was determined to be more important than any of the composition-related variables.