Continuous electric field measurements from 1988--1995 with a telluric array straddling the San Andreas fault in Parkfield, California have been analyzed for changes in electrical resistivity associated with the four largest (Mb>4.0) local earthquakes and three larger but more distant California earthquakes (1989 Loma Prieta, 1992 Landers, 1994 Northridge). Treatment of the propagation of errors through the analysis process yields formal estimates of the uncertainties in the results and thus a definition of anomalous variations based on statistical confidence. Statistically significant variations of the telluric coefficients were associated with none of the four local earthquakes and none of the California earthquakes. An obvious, but mistaken, conclusion from this disappointing result is that resistivity changes do not occur prior to earthquakes; previous observations in both the field and laboratory have shown that they do occur. A more likely explanation of the current result is that the telluric array is not sufficiently sensitive to resistivity changes associated with such small earthquakes. The next Parkfield earthquake, while overdue, should be of sufficient magnitude to test this explanation. Results from this experiment have shown the need for continuous, redundant measurements in precursory monitoring studies.¿ 1997 American Geophysical Union