After a hiatus of many years the 3rd Conference on Tectonic Problems of the San Andreas Fault System was held at Stanford University and the U.S. Geological Survey in Menlo Park on the 6th, 7th, and 8th of September, 2000. The conference was sponsored jointly by the Department of Geophysics, and the School of Earth Sciences of Stanford University, together with the Earthquake Hazards Program of the U.S. Geological Survey. As in the previous two conferences there was an emphasis on providing a convenient framework for informal but pointed discussions. Since the first two conferences there has been much scientific progress, especially in the last decade, and it was felt that it would be timely to present and discuss these new developments at a third conference. A broad spectrum of papers was submitted and we arranged them into four thematic sessions, which forms the structure of this proceedings volume.

I. Seismicity and Fault-Related Experiments
Among the highlights presented at the conference were new techniques for high-resolution earthquake relocation. These new techniques indicate that much of the seismicity along the San Andreas fault system is organized into events repeatedly occuring at the same location or along linear, nearly horizontal structures (Beroza et al., pages 4-6 in this volume; Waldhauser et al., 7-13, Rubin, 14-18). This spatial organization is clearly very important and it allows many insights into the fine-scale structure of the fault zones, the dynamics of earthquakes, and the type of deformation under which specific fault patches move. There are a number of open questions, however, such as how large are the associated stress drops (Nadeau and McEvilly, 19-21)? Understanding fault zone behavior also requires lab experiments (Muhuri et al., 82-84, Reches, 85-86; Katz and Reches, 87-88) and modeling (Goertz et al., 55-66). Beeler et al. (31-42) give a model for the occurrence of repeated events. Borehole observations (Ito et al., 91-92) allow studying the earthquake source with greater detail than possible from the surface (Abercrombie, 22-28; Nadeau and McEvilly, 19-21) and allow in situ studies of structure (Paulsson et al., 89-90). The study of exhumed fault zones (Evans et al., 67-81) is also important. A number of papers were presented which discussed the spatial distribution of large earthquakes in Northern California (Toppozada, 93-105), the seismicity of Southern California (Hauksson and Jones, 43-54), and the structure in the vicinity of the fault zone (Fuis et al., 106-107).

II. The Big Picture and Deformation
An enduring paradox of fault-zone mechanics is that faults should produce considerable frictional heat, which is not observed, suggesting that perhaps the major faults slip under small shear-stress. Instead of localized high heat flow anomalies at the faults there is a broad zone of high heat flow under the Coastal Ranges (Williams et al., 108-111), which may be explained by a slabless window under most of California. Furlong and Guzofski (112-127), Merritts et al. (128-143), and Wright (144-151) discuss the long-term evolution of the San Andreas Fault system and of California as a whole, e.g., will the Pacific-North America plate boundary be to the east of the Sierra Nevada in a few million years? It is of interest to understand how this evolution can be accommodated mechanically. Bokelmann and Beroza (152-164) show evidence for a weak lower crust under the San Andreas Fault system. Nur (1-3), in the opening talk of the conference, and Ron et al. (165-176) discuss the rotation of crustal blocks, especially in Southern California, and the evolution of new faults in the Mojave. The possibilities of using new satellite-based instrumentation for measuring shorter-term surface deformations associated with earthquakes and with the interseismic period was also discussed. Zebker et al. (177-178) applied satellite-geodetic techniques (inSAR) to the recent Hector Mine earthquake in Southern California, and, based on radar interferometry, Vincent (193-208) suggested that aseismic deformation is more widespread than previously thought. The proceedings also describe the results of a study of seismic versus aseismic deformation in the San Juan Bautista area (Gwyther et al., 209-213). Surface geodetic data were analyzed and used to infer the spatially-variable slip rate of the San Andreas Fault near Parkfield (Murray et al., 179). Finally, Bosl and Nur (214-225) discussed how to model deformation in a poroelastic model, and applied it to a study of aftershocks of the Landers earthquake.

III. Stress, Stress Triggering, and Fault-Zone Strength
Another emphasis in the papers presented at the conference were efforts to understand the mechanical stress field in the vicinity of the fault zone. Does the direction of maximum compressive stress vary systematically with distance from the fault zones? How wide is this zone of rotation? Are the faults mechanically weak or strong (Zoback, 226-227)? Is the heat flow paradox still a paradox? Earthquake focal mechanisms were used to address the above questions for Northern California (Provost and Houston, 228-236) and Southern California (Hardebeck and Hauksson, 255-267; Townend and Zoback, 268-276). The Carrizo Plain Segment was studied by Castillo and Hickman (252-254). The stress state after the Loma Prieta event, i.e., whether it represented heterogeneous stress or a nearly total stress drop (Michael and Eberhard-Phillips, 237-251) was also discussed. The second part of the session was devoted to understanding the consequences of stress changes due to earthquakes. Hori and Kaneda 277-287) studied the stress shadow effect. Toda and Stein (288-304) showed that predictions of Coulomb stress changes match seismicity rate changes well, and allow understanding of many aspects of stress transfer, stress relaxation, and differences between the locked and the creeping sections of the San Andreas fault near Parkfield. They also explain why the expected Parkfield quake did not occur in the 1980's. Harris (318-323) studied the question of what stops earthquakes; the dynamics of earthquakes is important in explaining why quakes sometimes jump boundaries between different fault segments. Rundle et al (305-317) modeled the space-time pattern of earthquake occurrence in Southern California. Homberg (324-333) suggested that static stress field changes from past earthquakes may be detected and used to extend the seismological record to earlier times. Karakelian et al. (334-346) presented results from electromagnetic monitoring of earthquakes. Hanks and Scholz (did not submit a paper for publication) discussed the heat flow paradox, which led to a most lively debate (347-348).

IV. Paleoseismology
A number of papers were presented that discuss the extension of the seismological record to earlier times using the techniques of paleoseismology. Goldfinger et al. (352-354) studied turbidite sequences and suggested that turbidites are primarily generated by large earthquakes and that they represent a powerful tool for paleoseismology. Baldwin et al (355-369) and Prentice et al. (349-351) presented data for the same segment of the northern San Andreas fault and suggested that the penultimate earthquake ocurred in the mid 1600's, giving a recurrence time of about 250 years for this segment of the fault. The evolution of the San Francisco Bay block was studied by McLaughlin et al. (370). In this evolution, oil-bearing rocks near Davenport and Point reyes that originated together were displaced by about 115 km of slip on the San Gregorio and San Andreas fault (Stanley and Lillis, 371-384).

We are indebted to Amos Nur, Stanford University, and Mary Lou Zoback, U.S. Geological Survey, for their support and backing of this conference. Special thanks are due to Agnes Kehoe, who expertly handled all problems of communications, registration, and the preparation of manuscripts for electronic publication. In fact, this proceedings volume is also available on the Internet (http://pangea.stanford.edu/GP/sanandreasconf.html). Individual sessions were chaired by Greg Beroza, Götz Bokelmann, Bill Ellsworth, David Jackson, Manika Prasad, Paul Segall, Ross Stein, George Thompson, and Dan Ponti. A group of Stanford graduate students, Jonathan Franklin, Paul Hagin, Seth Haines, Darcy Karakelian, Martin Mai, Jessica Murray, Xyoli Perez-Campos, John Townend, Kris Walker, and Michael Zimmer helped in transcribing the discussions from the floor. The papers in this proceedings volume are arranged in the order of the presentation at the conference. Since a main emphasis of the conference was on the discussion, this proceedings volume also attempts to give an account of the pointed questions and spirited discussion presented at the conference . The discussion of these topics is certainly not over with this conference, and we hope that this proceedings volume, including the recorded discussion, will be helpful in the continuing debate.

Götz Bokelmann
Robert L. Kovach