Seismic events reported in the bulletins of the two large arrays, LASA and NORSAR, were merged with those from the NEIS bulletin for the period 1970–1977. Using a lower cutoff of mb= 5.8, 510 ‘main shocks’ within the P range of LASA or NORSAR were selected for this period; and various seismicity trends prior to them were investigated. A search for definite foreshocks, based on a significantly short time delay to the main shock, revealed that the true rate of foreshock occurrence was less than 20%. Foreshocks are almost exclusively associated with shallow (h < 100 km) main shocks. To establish common features, a method of averaging seismicity from many regions was used to suppress the randomness of the seismic behavior of each region. This averaging shows that the seismicity level around the main shock increases somewhat for 10 days before main shocks; this feature peaks in the last 3–4 hours prior to the main shocks. The averaging also reveals that the mean magnitude of events near the main shock increases prior to main shocks but only by a few hundredths of a magnitude unit. Again by averaging, the seismicity about main shocks is shown to tend with time toward the main shock as its origin time is approached, but the average effect is small (∼10% change). By expanding or contracting each region's time scale before averaging to relate to the magnitude of the main shock, these features are enhanced. Using a new variable to track the departures from both spatial and temporal randomness, the Poisson‐like behavior of deeper seismicity (>100 km) was demonstrated. For shallow events (<100 km) this variable reveals numerous instances of clustering and spatial‐temporal seismic gaps, with little tendency toward a uniformity of behavior prior to main shocks. A statistical test of the validity of seismic precursors was performed for approximately 90 main shock regions which had sufficient seismicity. Using a five‐variable vector (interevent time, interevent distance, magnitude, epicentral distance to main shock, and depth difference relative to main shock) for each event in a ‘precursory’ time window of 500 days before the main shock and for each event in a ‘normal’ time window of 500 days before that, the null hypothesis of equal vector means between the two groups was tested. At 90% confidence level, less than 30% of the main shock regions were thus found to exhibit precursory seismicity changes.Appendices are available with entire article on microfiche. Order from American Geophysical Union, 2000 Florida Avenue, N.W., Washington, D.C. 20009. Document J81‐007; $1.00. Payment must accompany order.

von Seggern, D., Alexander, S. S., & Baag, C. (1981). Seismicity parameters preceding moderate to major earthquakes. Journal of Geophysical Research: Solid Earth, 86(B10), 9325–9351. Portico.