The spectra of the Love waves from five earthquakes of the 1966 Parkfield sequence, recorded at Berkeley, have been studied for evidence of source propagation. Independent field evidence suggests that the source of the main shock of the sequence may have propagated on a line away from Berkeley. The modulation effect of propagation on the amplitude spectrum is discussed. This effect is dependent upon the total time delay due to rupture down the fault and seismic wave propagation back to the point of initial rupture, and is thus ambiguous with respect to fault rupture velocity and fault length. Consideration of a frequency dependent phase velocity can resolve this ambiguity. The amplitude spectrum of the largest (M = 5.5) Parkfield earthquake shows sharp minima at periods where the spectra of the other four shocks are relatively smooth. Considered individually, these minima are consistent with a fault rupture length of about 30 km and a rupture velocity of 2.2 km/sec. The pattern of the minima varies from that predicted by the assumed phase velocities, thus the rupture velocity/fault length ambiguity is not resolved. Similarity in the shape of the spectra of the four smaller (M = 3.8-5.1) events suggests similar focal dimensions for these earthquakes. The Love wave amplitude spectra of the four smaller earthquakes scale well, at intermediate periods, to the magnitudes determined on standard short-period intruments; the main shock scales to a much higher magnitude. The spectra are considered in the light of proposed seismic scaling laws.