In 1980 Chevron recorded a three-component seismic line using vertical (V) and transverse (T) motion vibrators over the Putah sink gas field near Davis, California. The purpose was to record the total vector motion of the various reflection types excited by the two sources, with emphasis on converted P-S reflections. Analysis of the conventional reflection data agreed with results from the Conoco Shear Wave Group Shoot of 1977-1978. For example, the P-P wave section had gas-sand bright spots which were absent in the S-S wave section.Shot profiles from the V vibrators showed strong P-S converted wave events on the horizontal radial component (R) as expected. To our surprise, shot records from the T vibrators showed S-P converted wave events on the V component, with low amplitudes but high signal-to-noise (S/N) ratios. These S-P events were likely products of split S-waves generated in anisotropic subsurface media. Components of these downgoing waves in the plane of incidence were converted to P-waves on reflection and arrived at receivers in a low-noise time window ahead of the S-S waves.The two types of converted waves (P-S and S-P) were first stacked by common midpoint (CMP). The unexpected S-P section was lower in true amplitude but much higher in S/N ratio than the P-S section. The Winters gas-sand bright spot was missing on the converted wave sections, mimicking the S-S reflectivity as expected. CRP gathers were formed by rebinning data by a simple ray-tracing formula based on the asymmetry of raypaths. CRP stacking improved P-S and S-P event resolution relative to CMP stacking and laterally aligned structural features with their counterparts on P and S sections. Thus, the unexpected S-P data provided us with an extra check for our converted wave data processing.