A Crack in the Edge of the World

by Simon Winchester

  These days the community of Mussel Rock is the site of a ceaseless and wearying three-way battle going on between those residents who like to live there and gaze dreamily out at the foggy blue sea; those who say that the fault below them is moving so fast and so unpredictably that nature will not allow for indolent dreaming, or at least not for long, and that no one should be allowed to live there; and city officials who claim that the fault’s energies can all be sapped or countered by man’s cunning energies, and that Mussel Beach can be made safe and permanent. It can be made as enduring, some say, as that similarly situated rock in the Mediterranean, Gibraltar—and, in any case, Daly City people are perfectly at liberty to live just where they want to. They may not get good insurance, but they can live there if they feel they must. The tussle continues.

  And then there was the row that broke out over Daly City’s reputation. Just as soon as Mr. Bolt proved that the epicenter lay offshore a mile from Mussel Beach, a local historian, inspired by the news, designed a heavy brass plaque that memorialized what now appeared to be Daly City’s historic importance in the event. His plan was to erect this monument—which he had bought and paid for—on a nearby beach.

  But the mayor of Daly City, an elected official who at first blush appears to be a man with rather little sympathy for history, said no. He didn’t, he said, want people to associate his city with an earthquake. “We don’t need to put a blemish on Daly City’s shine,” he declared. “There’s no reason for it.” So the plaque could not be erected anywhere in Daly City.

  The historian tried three further times, even appearing before the city council armed with the support of a host of the region’s geologists and seismologists. But the councillors remained adamantly intransigent, too, with the result that Daly City will not now be publicly associated in this way with the events of 1906.

  Except that it will be associated with the event—thanks to a much more acceptable story that is already well known locally. Refugees from the earthquake’s firestorms fled south across the city line and camped on plots beside what were then the meadows of one John Daly’s dairy farm. Rather than return and risk another tragedy, many of them then set up home there. Daly City is, in other words, a place that owes its very existence to the 1906 earthquake, is in its very essence a memorial to the tragedy of 1906, and probably does not want to be reminded any further, thank you very much. Besides, the crumbling of the cliffs at Mussel Rock provides a daily reminder of the presence of the fault, which streaks under the entire length of the community on its way south.

  THE SAN ANDREAS FAULT can be handily divided, like Gaul, into three parts—though, unlike in Gaul, they are three unequal parts. The division is made by geologists, since in each of these three parts the fault behaves very differently.

  First, to the north, there is a 275-mile-long sector that stretches from Cape Mendocino, past Olema and San Francisco and Daly City, to a point just south of the old Spanish mission at the dust-dry ranching town of San Juan Bautista. Then there is a rather longer southern section, which runs for about 350 miles from the beginning of the Sierra Madres southwest of Bakersfield to the fault’s southern terminus at the Salton Sea, close to the Mexican border. And in between these two sectors is the shorter, 125-mile-long midriff of the fault centered around the small town of Parkfield. This middle section is very different in one hugely important way from the sections that lie to the north and the south of it.

  Some things about the fault remain essentially the same. Throughout its length the San Andreas is a right-lateral strike-slip fault: that basic characteristic, with the North American Plate on the eastern side and remaining resolutely in place, the Pacific Plate on the western side and shifting northward at an inch and a half a year, is unvarying. All the movement along the fault is sideways, and it all goes in the same way. (Having said that, there are a few places that have the fault going up or down, in the way that faults were once thought to do. But these places are few. They reflect local aberrations in rock structure, and do not affect the overall thesis, which is dominated by the now all-too-familiar scenario of two plates sliding past each other.)

  However, there is one signal difference. In the middle section of the system, in and around Parkfield, the fault is for some inexplicable reason always moving. It is shifting, all of the time, little by little, millimeter by millimeter, day by day. But in the rest of the system—in the northern section centered around San Francisco and in the southern section centered on places like Palmdale, out in the desert to the east of the eastern suburbs of Los Angeles—the fault on a day-to-day basis is not moving at all. The north and the south are locked solid. And yet the middle part of the fault is moving—a circumstance that places the locked northern and southern parts of the fault under an ever-increasing amount of stress.

  The best analogy for this is to think of a line of railway freight cars standing in a marshaling yard, each car connected to the next by a firm iron clasp. The cars in the yard are all on a slope that makes them want to move downhill headfirst, as one. But the wheels of the cars at the head of the line are rusted solid, as are those of the cars at the distant end. Only the cars in the middle have well-oiled wheels—and it is these, under the force of gravity, that move down the slope under its influence. They move, but their movement is checked all the while by the sheer inertia of the rust-blocked cars in front and behind. Enormous stresses thus build up in the iron clasps that bind the cars together, with those at the front being pressed tightly together, those clasps at the rear being stretched beyond endurance. Some of them may even bend a little, their inherent elasticity storing huge amounts of energy as they wait for the day that they either break or spring back to their original shapes.

  And then, every so often, the compounded weight of the cars in the midsection becomes simply too much for the rust holding fast the wheels of the locked cars—and, under this terrible mounting pressure, the rust gives way and the cars shoot forward, jerking anyone foolish enough to be standing on them off their feet and upsetting any bales of cargo piled on top or inside. The pressure from the midsection of the cars suddenly eases, because all of the cars have now moved as much as they are going to—for a while. But then, day by day, the middle cars begin to shift once more—and after a while, with the intolerable stress again built up both ahead of and behind them, one of these rusted sections gives way, suddenly, dramatically, and violently.

  There is a shudder as the pressure is relieved. The energy that has been building up, the potential energy that has been growing, waiting for the weakest link in this particular chain—in this case, the rust in the locked wheels—to give, is suddenly released and converts itself in an instant into kinetic energy, which tumbles cargo and throws people around. This would be very dangerous were it ever to happen in a railroad marshaling yard. It is also, in its essentials, precisely what is happening in these sections of the San Andreas Fault.

  The forces caused by the unstoppable, irresistible movement of the plates deep below are building up and up and up. Because of the movement, barely recognizable or measurable shifts in the landscape are occurring more and more. Rocks close to the fault, in what is called the near field, are all the time bending, straining, distorting. As they do so, they absorb huge quantities of energy—and this process continues until that moment when, suddenly, the weakest link in the affected rocks—the friction that prevents them from moving under all those accumulating pressures—ruptures. This happens because at last sufficient forces have accumulated to overcome its resistance.

  In a matter of microseconds two events occur. First, the rocks that have been bent and strained and distorted rebound to their original relative positions. And second, the plates are suddenly released to jerk forward to their long-awaited new positions. As these two things happen, both plates below and surface rocks above release all the energy that has been stored in them in the years during which the stress had been a-building. This long-stored potential energy transmutes itself immediately and demonstr
ably into energy—kinetic and thermal and sonic—and, as the transmutation occurs, with drama and turmoil that is directly proportional to the amount of energy that is transmuted, buildings are torn apart, heat is created, enormous sounds rumble across the land. It becomes, in short, an earthquake.

  This the famous theory of elastic rebound. It was first adduced by the celebrated geologist Harry Fielding Reid, who had been appointed by Andrew Lawson to his 1906 Earthquake Commission, and has remained the dominant theory of the cause of earthquakes along this kind of fault ever since. In places like California, where the earth is clearly under the influence of an ever-moving fault, it is in constant tension, much like a tightly wound mainspring in an old pocket watch. The tighter it is wound, the more energy it stores; should it be wound too tightly, then some weakness within it, or its anchor point, will eventually give way and the innards of the watch will collapse dramatically. This, in its own limited way, is an earthquake, too. Although Reid’s theory has been tinkered with over the century since he introduced it (as faults are now known to be much more complicated entities than is to be supposed from their cartoonlike representations in textbooks), what he declared in 1906 continues to be central to seismic study. The earth is elastic: Tighten it and it changes; overtighten it and it breaks.

  Which brings us, if circuitously, back to the current fascination with Parkfield.

  I HAVE ALREADY mentioned that the San Andreas Fault at Parkfield is constantly moving, and as a result there is a ceaseless blizzard of very tiny earthquakes, and every so often a rather larger one. This is true—except when one looks at the situation very closely indeed. For like many other kinds of movement, Parkfield isn’t so much moving as moving in a very rapid series of very small jerks. There is a similarity, for instance, in the different ways an automobile moves when in a straight line and when going around a bend. In a straight line its tires move constantly, seamlessly; when taking the car around a curve the tires move in a series of thousands of very tiny skids, each one requiring traction that can make the tire hot and wear away its rubber. An airplane, too, skids its way around the air when it is turning; when looked at closely, in a wind tunnel, there is nothing smooth at all in the way an airfoil negotiates a turn—it merely looks smooth and seamless.

  The fault behaves in Parkfield in much the same way. Deep down, the plate is moving here just as it is everywhere else; and the rocks on top of it in Parkfield appear to be moving fluidly past one another—unlocked—in a way that they do not move up north in Olema or down south in Palmdale, where they are locked. Except that, on close inspection, the supposedly fluid movement is not always quite so fluid as it looks. To be sure, part of the movement is fluid, and is known as aseismic creep—it just stretches and bends and slowly moves along under pressure. But, as far as researchers can tell, this is a very small component of what goes on.

  Most of the time the fault in these parts is actually locking and unlocking itself, but at so fast a rate that it looks as though it is in continuous motion. Recently it has been deduced that the process of locking and unlocking that goes on scores of times each year underneath the ground in and around Parkfield is mechanically and physically exactly the same as the process that a researcher might have to wait years to see farther north or farther south along the fault (and where the unlockings produce earthquakes). And because of this Parkfield is probably the best place in the world to study, on a small scale and in real time, exactly what happens to the San Andreas Fault—and, by extension, to any other fault—just before, during, and immediately after it locks, unlocks, and finally relocks itself.

  They are doing this by drilling a hole—a modern-day journey, say some of the project’s more enthusiastic boosters, to the center of the earth.

  USING A LARGE oil rig, and employing a team of tool pushers and roughnecks hired directly from the oil industry, geologists are forcing a hole two and a half miles down into the fault itself. Scores of instruments have been placed at the bottom of the hole, and an array of branching tubes has been drilled out from the main shaft—and it is with these that the scientists are examining what happens at the depth at which all of the fault’s movements are known to originate.

  The experiment is called the San Andreas Fault Observatory at Depth (SAFOD), and it was formally begun in 2002 with a pilot hole that took the instruments down a mile and a half, stopping just shy of the active fault zone itself. Then, in the summer of 2004, a new drill team put up its $20 million derrick and its mighty stands of pipe and drill bits and the great square turning nut known as the kelly, and started to drill the main tube.

  The National Science Foundation and the U.S. Congress had earlier in the year approved the $250 million to be spent on a hugely ambitious program known as EarthScope, of which the Parkfield pipe was a major part. (Parkfield accounts for only a tenth of the budget: $100 million is going toward a great number of GPS devices that will measure movements all along the boundary between the tectonic plates in western America; and $70 million will go toward a huge network of seismometers that will be installed around the West to paint a seismic picture of the continent with an unprecedented degree of accuracy. SAFOD costs only $25 million, and yet it is the project that of the three has managed most handsomely to capture the romance—and the danger—of this kind of research.)

  EarthScope has had a profound effect on modern geology and played a major role in the making of the brand-new science into which the Old Geology has now indisputably evolved. At last the U.S. Geological Survey and Stanford University (and more than a hundred other universities and institutions in America and around the world that are committed to studying SAFOD’s eventual results) have at their disposal a hugely costly and very-large-scale experiment that, in its own way, seems likely to present as good a picture of a fundamental process of the planet’s working as the moon landers and bathyscaphes and cyclotrons and linear accelerators had already presented to the practitioners of their respective disciplines. Such a thing has never happened before—except for the 1969 moon landing and the astronauts’ collection of moon rock. The SAFOD experiment is undeniably Big Science, and the geologists involved are happily amazed at their good fortune, finally able to hold their heads high in the company of all those nuclear physicists and genome researchers and oceanographers who have taken Big Science for granted for so long.*

  The first part of the SAFOD drilling project, which was executed with the Nabors Industries No. 633 rig positioned on a cattle ranch that stands firmly on the Pacific Plate, was completed at the beginning of October 2004. The end of the hole was then left a tantalizing 1,600 feet above the football-field-size cube that seismologists have decided incorporates the most active part of the fault, where they want their various probes to be planted. In the second phase the shaft is to be extended to reach directly into this cube-shaped zone. (At first it was drilled entirely vertically but later bent eastward so that it runs through the fault and right out into the North American Plate.) Next, branching tubes will be bored with special directional drilling equipment, reaching to other parts of the cube and beyond it. Sensitive equipment in protective pods will be lowered into the end of each of the subterranean shafts, where all manner of physical data from within the deep will be measured: the stress, the pressure of fluids, the temperature, the heat flow, the chemistry of the fluids, the rocks, and any gases—and, of course, the seismicity.

  The big worry among local residents is that, somehow, the drilling program itself will trigger an earthquake. There is a widely held concern—held, at least, among those who are somewhat skeptical of science—that interfering with the delicate balance of the plates will start something that might prove unstoppable. Was it not possible, some vaguely exercised Californians asked rhetorically, that the drilling rig might set something off? The USGS said there was no possibility. Look at Texas, someone said. There can be no more comprehensively drilled place than the oilfields of Texas—and yet the state is seismically almost stone dead.

  Leavin
g aside the questions begged by such a statement—Texas is basically very much more geologically stable than California—it turns out that the skeptics did have some cause for concern. For it so happened that the SAFOD drill site was at its most active, and the drill bit was probing most closely to the fault, when, on the morning of September 28, 2004, a magnitude 6.0 earthquake hit Parkfield square on.

  It was the largest earthquake to strike the town for thirty-eight years. The USGS and the drill team insisted there was no connection. Locals were not so sure.

  THE COUNTRYSIDE SOUTH of Parkfield rapidly becomes very much drier, sandier, and unpopulated. Only eight inches of rain fall each year on the Carrizo Plain—a high, hot place of salt flats and soda lakes, with kangaroo rats and leopard lizards and other improbably named fauna—but with almost no trees and little vegetation to obscure what the fault does on its way through. The result is that the San Andreas is laid bare as it pushes down along the desert, as if pinned to a dissecting table. Almost all books related to the fault sport an aerial photograph of its signature across the Carrizo Plain on the jacket, so clear and obvious is its passage.