A Crack in the Edge of the World

by Simon Winchester

  Once this was done, or under way, a portion of the North American Plate started to collide directly with the Pacific Plate—most of the Farallon had by then vanished and was wallowing deep below what would in time be Southern California. As it did so, the relative motion of the plates changed—just as the directions of cannonading billiard balls do. The two plates did not hit each other square on, one of them moving west to east, the other east to west: The North American Plate ran out of steam and essentially stopped in its tracks, while the Pacific Plate began to move northward. It began to slide up and along the outer edge of the North American Plate. It began to slip along its strike, as the geologists who first discovered the phenomenon declared; it began to slip along the line that marked the edge of its outcrop. And it began to do this a little less than 30 million years ago. Where the plates scraped against each other, the land, up on the surface, became crazed and fractured with a pattern of faults.

  At first these faults, marking the strike-slip movement going on below, were some distance away from their present-day track, and ran in different directions from it as well. Near the Californian seaside town of Santa Cruz there is a fault called the Gregori-Hosgri that represents one of the early sliding-plate tracks; and another called the San Gabriel Fault a little way north of Los Angeles also shows, in what can be thought of as a fossilized way, where the plates used to slide alongside each other. And both of these go off in a very different directions from the more recent fault, which displays the more recent relative plate movements.

  About 10 million years after the process had first begun—about 20 million years ago, in other words—the relative motion of the two plates settled down, running essentially along the line of one principal fault. And though even today in most places the plate-against-plate contact line cannot quite be pinned down to a matter of inches—usually it is more a zone than a line, and is some hundreds of feet or, in places, even miles wide—today’s center point, the zone where the maximum annual slip between the two plates is noticed, is the track of that most infamous darling of seismicity, the San Andreas Fault.

  It should be noted that this plate-on-plate strike-slip zone extends between two “triple junctions”—places where the two principal plates meet up with two small relict pieces of the old Farallon Plate that did not get themselves subducted. (These parts were not subducted because, in essence, they were too far north or too far south of the main westward thrust of the North American Plate to be affected.) So the fault zone technically runs from what is called the Mendocino Triple Junction, off California’s Cape Mendocino—where the Farallon Plate’s little relict piece is called the Juan de Fuca Plate*—down to the Rivera Triple Junction, off Mazatlán and Puerto Vallarta on the Mexican west coast—where the tiny relict piece of the Farallon is called the Rivera Plate.

  These two little plates—the Juan de Fuca up north and the Rivera down south—are still subducting, as their predecessor once did. And, sure enough, where they do subduct, there are, as always in such situations, volcanoes. In the north, as a consequence of the beautifully named Cascadia Subduction Zone, there is a fleet of active volcanoes, with Mount St. Helens being the most recently notorious; and in the south there are volcanoes such as Mexico’s Colima and Paricutín, the former old and still active, the latter young and, in spite of spectacular eruptions in the 1940s and 1950s, now apparently quite defunct.

  Between these two triple junctions, then, runs the San Andreas Fault—this 750-mile-long zone marked by its near ceaseless activity and occasionally by very lethal seismic outbursts. The land on each side of the fault is moving, all the time—though not everywhere along the fault’s length at quite the same rate. Overall it is moving at a speed that in real terms seems very slow: Up at a place called Telegraph Creek, close to Cape Mendocino and the northern triple junction, the Pacific Plate is moving northward at about an inch and a half every year. In the terms that geologists understand, however, an inch and a half a year is something approaching raceway speed. According to the USGS, a velocity like this makes the San Andreas a fault like very few others, anywhere in the world. So, by most geologists’ lights, it is very fast, very interesting, and very, very dangerous.

  Its effects have been noted for a very long while and noted, more-over, in places far removed from where the fault makes itself topographically obvious, such as in the San Francisco Valley where it was first named. Its real extent was noticed in 1906, for instance, just after the San Francisco Earthquake, when a man named F. E. Matthes was sent up north to Humboldt County to help the California State Earthquake Commission see if what had happened in the Bay Area had spread into the north of the state. He found that it had—and moreover he then mapped what he imagined to be the trace of this selfsame fault, finding that all along its never-before-noticed path there was a pattern of instantly recognizable breaks, shears, landslips, and a host of other peculiar and damaging phenomena.

  All of this evidence convinced Matthes that this was not an ordinary fault line, but one that had been occasioned by two sets of rocks sliding past each other. To him it was a revelation. Back in 1906 most geologists assumed that faults operated only vertically, with huge forces throwing rocks either sharply upward or downward. But here, 200 miles north of San Francisco, there was evidence of the fault that had caused all the trouble—and it was a new kind of fault, a strike-slip fault as it came to be known, and one that was all too obviously an extension of the San Andreas. It was making itself dramatically felt up among the mountains, the clouds, and the mists of far northern California, farther away from the most gravely affected areas than most scientists of the time could ever have imagined.

  This was the first time that geologists fully realized the extent of the fault’s spread. It was the first time that the whole of California was seen to be playing host to so threatening and so massive a danger.

  CLOSE TO SAN FRANCISCO the fault, and the drama it causes when it moves, is very much more obvious than it is in the tortured coastline and hills of the state’s far north. A town called Olema, forty miles north of the city (ole is “coyote” in the language of the coast Miwok Indians), is generally reckoned to be the place on the fault that moved most dramatically during the 1906 earthquake—and such is the local certainty of this that, quite wrongly, local residents like to think of Olema as having been the event’s epicenter. There is a small clothes shop that calls itself the Epicenter. In it and other local stores you can buy caps adorned with the words OLEMA and EPICENTER. I have one. I wear it. But what it implies is not true.

  For a general truth should be pointed out here. An earthquake’s epicenter is not necessarily the same as the place of maximum ground displacement. The epicenter is the point on the earth’s surface directly above that point where the seismic energy of the quake begins to radiate outward, the earthquake’s originating point. But the parts of the earth where displacement and damage occur depend on a number of details—most important of all the nature of the underpinning rock, the kind of soil, and the slope of the land. And in Olema—where the fault zone is half a mile wide, and filled with soft soils, crumbling limestones, rotting granites, and generally pulverized rocks, much of it lying on well-watered slopes—the sudden shock of a sideways-moving fault rupture did indeed shift the ground in a dramatic and dangerous way. But that does not mean the earthquake was centered there. It means merely that it had an impressive effect there—with few casualties and less publicity, however, since it was a place where few people lived.

  Grove Karl Gilbert, the distinguished geologist who, it will be recalled, was roused by the quake as he lay in his bed in Berkeley, was among the first of the state commission members to go to investigate the situation in the Olema Valley. The devastation he found was most impressive. Buildings made of redwood and oak were crumbled as though fashioned from balsa chopsticks, huge water towers had been tossed around like cruets, long fissures spread across fields. The local dairy farmers told him fanciful stories, including one that was widely rep
orted in the papers: A cow supposedly fell into one fissure, which swallowed her up as it closed shut behind her; she then suffered the postmortem indignity of having her tail eaten off by packs of marauding and hungry dogs.

  However, it was measurement—or one measurement in particular—for which Gilbert is perhaps best remembered. And he measured anything he could lay his eyes on, particularly anything that seemed to have been displaced across the trace of the fault. He measured fences, stands of eucalyptus trees, roads and farm lanes and tracks—and, from the displacements he found, he came up with a litany of earth movements that has been exceeded only by a very few other earthquakes in history. One farmer’s fence snapped, and its posts were shifted thirteen feet apart, declared Mr. Gilbert. A line of eucalyptus trees near Bolinas Lagoon was broken and moved, also by thirteen feet. A road southwest of Point Reyes Station was displaced by twenty feet. The rough gravel pavement of the Sir Francis Drake Highway was broken clean across, and one part of the road was moved—Gilbert photographed it, memorably—just over twenty feet relative to the other. Not up or down but from side to side: If you stand and look at where the centerline of the road might be, it suddenly ends, and then reappears twenty feet off to the right, with the highway then continuing and vanishing over the horizon as though nothing had happened.

  Although the small Northern California village of Olema saw the greatest displacement of the earth—this roadway was shifted some twenty-one feet by the right-lateral motion of the San Andreas Fault—it has lost its much-prized position as the epicenter of the 1906 event. That distinction now goes to a point under the sea off Daly City, forty miles south.

  To the right. The effect of movement on a right-lateral fault: This is how the fault is officially known, as a right-lateral strike-slip fault. Stand anywhere and look across the fault—and the land on its far side will have been moved to the right. Hills, streams, roads, lines of trees—all of them, if they are on the distant side of the fault, will be to the right of those that lie on the side near the observer.

  Today, at the Point Reyes Visitors Center, the fault is plotted through the meadows with a line of blue posts stuck into the ground. In one place the posts show the fault spearing underneath the barn that once belonged to a local farmer named W. D. Skinner. He had told Gilbert many things about the event (including the story about the cow) and showed him how his barn had been torn apart, his neat rows of raspberry bushes offset by fourteen feet, his fences ruined and scattered. And indeed there still is a fence on the property dating from long before 1906. At one point the fence posts seem suddenly to vanish—until you look and spot them once again, half hidden in a patch of woodland. The fence had been built with one continuous line of hastily carved redwood posts, with pine or redwood crosspieces between. But at the moment of the earthquake this fence line was snapped and sundered by the force of the event, with the western end of the fence moving north. And it moved a lot—to such an extent that the two ruptured ends that are visible today have been left no less than twenty-one feet apart.

  This figure is most often cited, and impressively so, as illustrating the maximum amount by which the San Andreas Fault, and by association all of California as well, had moved on that extraordinary April morning.

  But still, Olema was not the epicenter.

  EARTHQUAKES EMIT WAVES that ring through the solid earth just as sound rings through a bell made of brass. Moreover, they emit a number of different types of waves that, most fortunately and most crucially, move through the earth at very different speeds. It is the difference in speed between the two fastest-spreading kinds of waves (which are known as body waves because they travel through the entire body of the planet) that allows us to determine generally where each earthquake has its point of origin.

  All that is required is a minimum of three recording stations, equipped with seismometers, that can observe the earthquake, measure the arrival of those two waves, and record the time between them. The first of the two body waves to arrive—and so the faster of the two—is a pressure wave, a P-wave, one that presses the rock and releases it, presses and releases, as if a Slinky toy were being stretched and then given a hard shove along its main axis. The second, slower body wave is the shear wave, the S-wave, which ripples horizontally through the rock strata—just like the sideways ripple that can be made to course through a Slinky.

  (When terrified observers speak of the ground rippling toward them, or rising and falling in great fast-moving wavelike motions, they are almost certainly seeing the next family of waves—surface waves—that propagate more slowly because they only involve the outer surface of the earth. These most destructive waves—which are very easy to see on a seismograph because they are very large and have relatively large amplitudes—are further subdivided into what are known as Rayleigh Waves and Love Waves; the behavior of these, though they can be devastating to buildings and lethal to people, generally has less relevance to determining the location of earthquakes.)

  Providing that both the P- and the S-waves travel through the same kind of rocks on their way to the observation station—which of course they would; but it is worth pointing out that waves run at very different speeds depending on whether they pass through granite, say, or shale—then there is always the same differential in their velocities. The farther the origination point from where the earthquake is felt and measured, the greater that differential. An earthquake originating in Olema might show, on the seismographs in the USGS machines in Palo Alto, a differential between the two waves’ arrival times of just ten seconds; that same earthquake, noted by the machines in Columbia University’s famous Lamont-Doherty Earth Observatory in New York City, might show a differential of eight minutes. Each single second of separation in the waves’ arrival times equates to about five miles of distance. This is similar to the rule-of-thumb approach that is used to compute how far away a storm might be: One sees the lightning flash and then counts the number of seconds until the thunder is heard—that number is very roughly how many miles separate you from the storm. So from Palo Alto the quake would be 50 miles distant, and from New York, 2,400 miles.*

  So, taken singly, any receiving station can tell, simply by noting the difference in the arrival times of the two waves from an earthquake, just how far away the earthquake’s center is. But, armed with at least three receiving stations, one can then triangulate these data and find out not just how far away but also exactly where the earthquake is. The exercise is simplicity itself: All one does is draw a series of circles representing the distance (calculated from the time differential) that the earthquake appears to be from each one of the three stations. If the time differential indicated the earthquake was 50 miles from Palo Alto, then draw on the chart a circle with a 50-mile radius around Palo Alto. If the time differential suggested the same quake had happened 2,400 miles away from the Lamont station in New York, then draw a second 2,400-mile-radius circle around New York; and finally if a Calgary observatory measured the quake as occuring 1,200 miles from its seismometer, draw this third circle with a 1,200-mile radius. The three circles intersect at just one point—and this point where they meet is the epicenter—the place on the earth’s surface that is directly above the originating rupture. And the exact point of that rupture, and the depth of it below the surface, is something that can itself be determined by performing the same P-wave, S-wave, and time exercise all over again, but this time in three dimensions. That will give the hypocenter, the true originating point of the event. All, in other words, extraordinarily easy, for anyone with three seismographs and three accurate stopwatches.

  Which is what a Berkeley seismologist named Bruce Bolt had—although actually he had many more seismographic records than these—when, in 1968, he recomputed the data responsible for the widely believed notion that the 1906 earthquake had its epicenter in the Marin County countryside near Olema. The 1906 event was recorded not merely by three observatories but by ninety-six seismographs around the world, and just about every single one of th
em had an accurate recording clock, each tuned to what was then Greenwich Mean Time. Moreover, a prodigious number of the local clocks—ordinary timepieces on living-room mantels, long-case clocks in parlors, clocks on church towers, clocks displayed on public buildings—were stopped dead by the force of the event, further evidence for the time of arrival of at least the strongest jolt of the quake. (It does take some technical expertise, however, to work out whether a clock from a historic earthquake has been stopped by a P-wave or by the subsequent S-wave, or perhaps even later, when a brick might have fallen on it. Generally, though, the shear waves are stronger, and so calculations on stopped clocks often assume that the stoppage was due to the arrival of the S-wave.)

  However questionable some of the data might have been, Bolt had a very great deal of them, enough to calculate what is now generally accepted to be the true epicenter of the 1906 event. And so, according to the paper that he published in 1968, the recalculated epicenter turned out to be in the sea, a little more than a mile offshore, southwest of the Golden Gate Bridge and northwest of a particularly crumbled section of the coast near the sprawling suburban community of Daly City.

  It has long been recognized that the track of the San Andreas Fault passes into the ocean a few miles south of Olema, close to that expanse of sand and shingle where so many San Franciscans go to swim known as Stinson Beach. The coastline is indented eastward from there on south, forming a funnel toward the Golden Gate and the main shipping entrance to San Francisco Bay. The fault, however, remains unpersuaded by this deflection, and continues to roll on southeastward in its usual die-straight line. It scythes on through the sea to the immediate west of where the Golden Gate Bridge now stands, and reconnects with the coastline five miles farther on, in a particularly wretched part of the suburban mass of Daly City called Mussel Rock, where—under the fault’s malign influence—houses seem always to be sliding into the ocean, landslips are an all-too-frequent occurrence, and the coast is a veritable construction site of seawalls, patched roads, and fractured gas lines. The benighted residents of Mussel Rock have paid a stinging price for the luxury of being able to watch the sun set over the Pacific each evening.