Massive hydraulic fracturing is used to enhance production from the low-permeability diatomite fields of Kern County, CA. Although critical for designing injection and recovery well patterns, the in-situ hydraulic fracture geometry is poorly understood. In 1990, Shell conducted an extensive seismic monitoring experiment on several hydrofractures prior to a steam drive pilot to characterize hydrofracture geometry. The seismic data were recorded by cemented downhole geophone arrays in three observation holes (MO-1, MO-2, and MO-3) located near the hydraulic fracture treatment wells. Using lowpass filtering and moveout analysis, events in the geophone recordings are identified as conical shear waves radiating from tube waves traveling down the treatment well. These events appear to be created by the hydraulic pumps, since their amplitudes are correlated with the injection rate and the wellhead pressure. Conical wave amplitudes are related to the tube wave attenuation in the treatment well and to wave-propagation characteristics of the shear component traveling in the earth. During the main fracturing stage, geophones above the fracture zone for wells MO-1 and MO-2 (both roughly along the inferred vertical fracture plane) exhibited conical-wave amplitude increases that are caused by shear wave reflection/scattering off the top of a fracture zone. From changes in the reflection amplitude as a function of depth, we interpret that the fracture zone initially extends along a confined vertical plane at a depth that correlates with many of the microseismic events. Toward the end of the main fracturing stage, the fracture zone extends upward and also extends in width, although we cannot determine the dimensions of the fracture from the reflection amplitudes alone. For all wells, we observe that the reflection (and what we infer to be the initial fracture) begins during a time period where no marked change in fracture pressure or injection rate or slurry concentration is observed. As the main fracturing stage progressed, we observed a significant decrease in amplitude for geophones below the top of the fracture zone. The attenuation was most pronounced for wells MO-1 and MO-2 (along the fracture plane). However, near the end of the main stage, well MO-3 also exhibited a significant amplitude decrease, suggesting the development of a fractured “process zone” around the main fracture plane. In addition, well MO-3 also exhibited an amplitude decrease in an interval well below the initial fracture zone. Both the interval and the direction (toward MO-3) correspond with temperature log increases observed during later steam injection.