An initial consolidation test, run with constantly increasing isotropic stress, allowed the estimation of elastic strain anisotropies, as well as producing a yield stress on the p' axis. The yield stress, p' = 0.85
MPa, was clearly defined on the plot of 
m' vs.
v. This value strengthens the validity of the yield stress determined from Ko tests in that it is also very low. More specifically it leads to a value of (p' 2 +q2)1/2 of 0.85, less than the 0.95 MPa value for this measure of the major axis of the yield envelope along the uniaxial strain path. This comparison shows that the axis of an elliptical yield envelope would be closer to the Ko line than to the p' axis.
Both the isotropic stress test and the initial isotropic stress steps during the Ko tests showed the samples to have a very strong vertical vs. horizontal elastic strain anisotropy, with vertical strains being about 3 times the average horizontal strain. The anisotropy within the horizontal plane appeared to be 30% or less as defined by A = 100 (max - min)/ave but was not investigated further because the core section was not oriented with respect to north.
Compressional acoustic wave velocities were determined during all tests except the consolidation of the disaggregated material. Velocities for a given test have a precision better than ± 0.01 Km/s, but the accuracy is limited to about ± 0.02 km/s because of unmeasured changes in sample length between the initial physical measurement and the time of the first LVDT measurement.
The increase in velocities with increasing stress above yield represents the effect of consolidation, but the large (0.2 km/s) increase from 1 atm to
c' can be attributed to microcrack closing
(Fig. 8). Furthermore, the sum of all the tests demonstrates that there is a large scatter in 1-atm velocities, which can be attributed most reasonably to variations in microcrack development. Because Vp was measured during tests run at isotropic stress increase, under Ko conditions, and at critical state failure (not reviewed here), data were generated that show Vp to be primarily a function of stress in the direction of wave propagation. If Vp is plotted against
v', there is relatively little difference between the stress paths and no consistent relationship with respect to
m', which increases for a constant
v' from the critical state failure test to isotropic stress test. If plotted against
m', Vp shows a significant increase at a given
m' with increasing
v', most
obviously during the constant effective mean stress tests to critical state
(Fig. 8).
These results indicate the importance of measuring Vp, not only at the correct
m', but also at the correct stress ratio. They also imply that appropriate Vp-
relationships should be generated from sets of values obtained at in situ conditions rather than from the relationship generated from a single sample.
