We have applied a waveform cross correlation technique to study the similarity and the repeatability of more than 21,000 microearthquakes (0 < M < 4.5) in the aftershock zone of the 1984 western Nagano earthquake in central Japan. We find that the seismicity in this particular intraplate fault essentially consists of no repeating earthquakes that occurred on the same patch of the fault in a quasiperiodic manner in the study period between 1995 and 2001. On the other hand, we identify a total of 278 doublets and 62 multiplets (807 events) that occurred consecutively within seconds to days. On the basis of the relative arrival times of the P and S waves, we have obtained precise relative locations of these consecutive events with an error between several meters to a few tens of meters. There is a clear lower bound on the distances measured between these consecutive events and the lower bound appears to be proportional to the size of the first events. This feature is consistent with what Rubin and Gillard [2000] have observed near the San Juan Bautista section of the San Andreas Fault. Shear stress increases at the edge of an earthquake rupture, and the rupture edge becomes the most likely place where the second events are initiated. The observed minimum distance thus reflects the rupture size of the first events. The minimum distance corresponds to the rupture size calculated from a circular fault model with a stress drop of 10 MPa. We found that using different time windows results in a slight difference in the delay time estimates and the subsequent projection locations, which may reflect the finite size nature of earthquake ruptures.