New shock wave data (to 180 GPa) for pyrite (FeS2 ) shocked along (001) demonstrate that this mineral, in contrast to other sulfides and oxides, does not undergo a major pressure-induced phase change over the entire pressure range (320 GPa) now explored. (This is probably so because of the initial, low-spin 3-d, orbital configuration of Fe+2 ). The primary evidence which indicates that a large phase change does not occur is the approximate agreement of the shock velocity when extrapolated to zero particle velocity, 5.4 km/s, with the expected zero-pressure bulk sound speed of pyrite (5.36 to 5.43 km/s on the basis of previous ultrasonic data). Pyrite displays a prominent elastic shock (or Hugoniot elastic limit) of 8¿1 GPa. The velocity of the elastic shock approaches 8.72 km/s with decreasing shock pressure, the longitudinal elastic wave velocity. As shock pressure increases, the elastic shock velocity approaches 9.05 km/s and the elastic shock becomes overdriven for shock pressures greater than about 120 GPa. Analysis of release isentrope data obtained via the pressure-particle velocity buffer method indicates that buffer particle velocities in all experiments are from 1.7% to 20% greater than expected for a Gr¿neisen ratio given by 1.56 (V/V0 )10 . This discrepancy appears to result from volume increases upon pressure release of 0.04% to 4.5% which may result from shock-induced partial melting. The normalized pressure, finite-strain formalism for reducing Hugoniot data is extended to take into account initial porosity and shock-induced phase transitions.

A least squares fit to the present and previous shock data for pyrite yields an isentropic bulk modulus, Ks , of 162¿9 GPa and a value of dKs /dP=4.7¿0.3. This is close to the 145¿3 GPa bulk modulus observed ultrasonically. If the slight discrepancy in zero-pressure modulus is taken into account in the normalized pressure finite-strain formalism, a zero-pressure density of the shock-induced high-pressure phase having a density some 2% to 3% less than pyrite is inferred to occur in the high-pressure shocked state. We suggest from this result, the release isentrope results, and limited phase diagram data that the Hugoniot states probably correspond to material which is partially to completely melted. Using the above derived equation of state and previous shock wave data for iron, both the seismologically determined density and bulk modulus distribution in the outer core are fit to models with various temperature distributions and varying weight percent sulfur. Good agreement between the shock wave derived equation of state and the density/bulk modulus relations of the liquid outer core are obtained for temperatures of ~3000 K at the core/mantle boundary extending to 4400 K at the outer core-inner core boundary. For this thermal model a calculated sulfur content of 11¿2% is obtained. ¿ American Geophysical Union 1987