The device, which is often displayed today to reinforce the idea that the Chinese were the first to create all of the various wonders that mark the track of human progress, does indeed illustrate the ancient concerns about the dangerous caprices of the planet. But seemingly it didn’t do very much. If it worked at all, it might have let the emperor know that there had been a distant vibration in his realms; and since the dragons were arranged around the points of the compass, it might also have allowed him to learn more or less in which quarter of his domains this vibration had occurred. It is said that this happened once: The ball dropped and the emperor was able to say that there had been a quake in Hunan Province—and lo, a few hours later a messenger arrived with the news that there had indeed been an earthquake in Hunan.
Joseph Needham, the great Cambridge scholar of Chinese scientific history, was one of the first to present pictures of how Zhang Heng’s second-century “earthquake weathercock” might have looked. He contends that the instrument was improved and used by subsequent court astronomers in China until at least the eighth century, with the basic idea finding its way to the West by way of Persia.
The device was beautiful, clever, and prescient—but probably so inaccurate as never to have enjoyed more than the most symbolic use. And, like so much that was created in early China, it was neither advanced nor improved upon, at least not by the Chinese.
It took Western science to do that—although it happened that in this particular case it was Western science that was by chance operating in the East. Crude vibration detectors were built in the eighteenth century, but the first modern device worthy of being called a seismograph—an instrument that is able to detect the earth’s movements with precision and, unlike the frog-and-dragon-balls affair, to make a record of it at the same time—was invented in 1896 by a Yorkshire wool-dealer’s son named John Milne. He invented it while he was studying the earthquakes that occurred with troubling frequency not in Yorkshire but in Japan.
His basic idea has remained the guiding principle behind the instrument ever since, and in essence it is the one that guided Zhang Heng’s instrument of sixteen centuries before: A pendulum with a pen is allowed to swing against a paper-covered and clockwork-run recording drum fixed solidly into a case bolted securely to the ground. Any shaking generated by an earthquake will cause the drum and the pendulum to move relative to each other, and the movement will be written by the pen onto the unrolling sheet of paper. (The terminology is more or less self-explanatory; the recording instrument is, strictly speaking, a seismometer; the entire package of instruments is a seismograph; the record to be studied later is a seismogram.)
John Milne constructed his brass, steel, and wire machine while serving as professor of geology at the Imperial College of Engineering in Tokyo, after he found himself frustrated by the inaccurate and shoddily made instruments that were put at his disposal to study the Great Yokohama Earthquake of 1880. And, while his device (sometimes called a Milne Seismograph but equally often called a Ewing, after his colleague-inventor James Ewing) was very accurate and as handsome as most Victorian scientific devices (it was manufactured by the famous, and still existing, instrument-making firm of R. W. Munro), it was soon to be supplanted by scores of instruments with much more exotic European and Asian names.* These would in due course become electronic and digital and accurate to unimaginable degrees.
In 1906 there were ninety-six seismographs in existence in the world, many of them Milnes or Ewings. So great was the shock emanating from San Francisco that practically every single one recorded it. Some were nearby—those in San Jose, Berkeley, and Yountville, California, and Carson City, Nevada, were so close to the event that their needles and drums recorded the destruction of buildings just yards away. At Stanford there was a Bristol’s Recording Voltmeter: an instrument designed to measure an altogether different kind of phenomenon, but that, since it had a needle, a pen, and a sheet of clockwork-driven paper, also noted the vibration with a spectacular set of indelible black-ink flailings.
Most of the true seismographs, though, were a good deal farther away: There were seismographs of all kinds in an almost alphabetically complete list of places that includes Apia, Bombay, Cairo, Dorpat, Edinburgh, Florence, Honolulu, Irkutsk, Jena, Kew, Leipzig, Mauritius, Osaka, Perth, Quarto-Castello, Rio, Sitka, Tiflis, Uppsala, Vienna, Wellington, the aforesaid Yountville and Zagreb, and, making up for the alphabetical absences of towns beginning with g, n, and x, an additional z, the small Chinese city that was possessed of a borrowed Japanese Omori machine, a half-forgotten place then known as Zikawei.
Only half forgotten, however. Zikawei used to be a small town southwest of Shanghai: It is now thoroughly subsumed into its neighbor, bears the name of Xujiahui, and is as filled with skyscrapers, nightclubs, and subway lines as any other part of the world’s fastest-growing city. A century ago this village was a place of colleges, orphanages, and hospitals—and, most interesting for this story, it also had a large Jesuit mission. The Omori seismograph that recorded the San Francisco Earthquake had been loaned specifically to the Jesuit Observatory of Zikawei—a happenstance that serves to underline the generally forgotten fact that the Jesuits and seismology have enjoyed a long and intimate connection, and that no small number of the machines scribbling down the record of the Californian events of that April morning were in fact maintained and operated by monks and priests ordained in the Society of Jesus.
The society, right from its sixteenth-century beginnings, was predicated on the basis of promoting scholarship and learning, particularly in the fields of mathematics, astronomy, and what were once called the natural sciences. By the seventeenth and eighteenth centuries most Jesuit colleges had departments that studied the earth and the stars; and in China, where the society had long had a foothold, the Manchu court appointed Jesuits to direct the Peking Astronomical Observatory.
The patient, monastic observers who maintained these schools soon noticed the terrible disruptions caused by earthquakes—and once the precursor to the machine that was eventually to be called the seismograph* had been invented in the 1750s, the society’s leadership decided, as a matter of canonical policy, that in those areas where earthquakes were common it would be the Jesuits’ holy duty to investigate them and, by doing so, perhaps help to mitigate the misery that such unpredictable events brought to God’s people.
So the first Jesuit seismograph—a crude two-pendulum arrangement of a type that would these days be somewhat dismissively called a seismoscope, since it never left a written record—was installed at the Manila Observatory in 1868. The next was installed the following year in a monastery in Frascati, Italy. Then one was built in Granada, Spain, another at Mungret College, County Limerick, another at Stonyhurst College in Lancashire, another on the island of Jersey, the first American instrument in 1900 in a Jesuit college in Cleveland, Ohio. And finally, in 1904, the Omori seismograph from Japan was carefully assembled in a special observatory building at the mission college in Zikawei.
This was the device that registered the San Francisco shock waves—one of four maintained by Jesuits around the world at the time. It was responsible for detecting the impressive set of shocks from which the earthquake’s approximate epicenter was determined. (The Shanghai instrument was replaced in 1932 by a newfangled contraption winningly known as a Galitzin-Wilip instrument, and it operated outside the great city until 1949, when the Chinese Communists turfed the Jesuits out.)
The installation of Jesuit seismographs accelerated rapidly after the disaster of San Francisco (with one instrument being installed in Addis Ababa as late as 1957), and for much of the twentieth century in America the Jesuits wielded a considerable influence over the entire seismological community, helping to direct the research that has shaped much modern thinking about the planet’s mysterious convulsions. And to this day four large Jesuit institutions—in St. Louis, Bogotá, La Paz, and Manila—continue to observe and conduct research into plate tectonics, keeping the scientific tradition
of this unique form of Catholic inquiry very much alive. Most of the small observatories have closed down (though the one at Stonyhurst College still functions, courtesy of the schoolchildren themselves), and the Zikawei machine is, of course, thanks to Maoist suspicions, no more. But their records—the etched traces on smoked-paper drums, the wavy lines of ink on now-yellow graph paper, and, the least scientifically satisfying but the most spectacular of all, ink traces made by a pendulum waving this way and that over a fixed circle of paper—remain.
What was recorded in all of these observatories, Jesuit owned or not, depended on the sophistication of each seismograph. The simple pendulums—like that in the Chabot Observatory in Oakland, just across San Francisco Bay—produced not much more than ragged messes of ink, most of them “too confused to give details.” Lawson’s report for Oakland presents raw data typical of these primitive instruments: The machine allowed investigators to say little more than that “the movement of the pendulum … was in nearly all directions, and more or less irregular, tho’ this irregularity was undoubtedly in part due to the pendulum’s striking against the side of the case.”
But better instruments followed—some in America, some in the Jesuit monasteries, many in Europe. Some of the best were the Bosch-Omori machines in Washington, D.C., at the New York State capital of Albany, and at the U.S. Navy base in Vieques in Puerto Rico. Each of these left a long trace of ink on a time-marked paper roll, and, by poring over the traces later, investigators were able to determine when the first tiny tremors started to be detected; when the secondary set of preliminary tremors began to rattle the needle; when the regular shear waves and pressure waves started to arrive—with the time difference between the two types, as we have seen, enabling the instrument readers to work out how far away the earthquake’s originating point was; then when the principal part of the tremor was under way; and finally when the maximum disturbance occurred.
Close modern analyses of the data from all the seismograms show that the earthquake unfolded in a number of stages. All of them are now known to have begun at a point rather less than ten miles south-southwest of where Officer Cook was standing at the eastern end of San Francisco’s Washington Street, several miles down in the earth’s crust at a site that lies well beneath the Pacific Ocean.
Modern analyses of the information have produced a different set of results from those computed at the time. The very distant seismogram information was not fully appreciated in the months and years that immediately followed the earthquake. It was seen more as a curiosity, as a series of unanticipated confirmations of the event’s sheer power. The thinking at the time held that the quake’s origins should be deduced from a small number of local recording stations. The more distant records might be impressive, but they were perhaps not truly to be trusted.
Since the recorded time of the shock waves’ arrival is crucial to finding out where the earthquake happened, the Lawson team became obsessively concerned with determining exactly when the event was felt, detected, and recorded in places very close indeed to its supposed point of origin. Buried deep within the commission’s great report are interviews galore with those who felt and recorded the two most distinct phenomena of the earthquake: the onset of the first perceptible vibrations, and the arrival of the truly great and destructive shock. Typical of the reports that Lawson collected is one from a Professor T. J. J. See,* the appropriately named director of the (equally appropriately named) Mare Island Naval Observatory on San Pablo Bay:
I had been sleeping downstairs, lying with my head to an open window, which faced the south, and as the house was not seriously endangered at any time I was favorably situated for making careful observations of the entire disturbance. I had been awake some time before the earthquake began and, as everything was very quiet, easily felt and immediately recognized the beginning of the preliminary tremors.
The exact time of the phenomenon. This was found by the stopping of two of the four astronomical clocks at the Observatory. The violent shocks were so extreme that the pendulums were thrown over the ledges which carry the register for measuring the amplitude of the swing. The standard mean time thus automatically recorded was: by the mean time transmitter 5h 12m 36s; by the sidereal clock 5h 12m 35s. The yard clock at the gate, which is simply an office clock, though electrically corrected from the Observatory daily, and therefore approximately correct, gave the time as 5h 12m 33s. The agreement of all these clocks is very good; but I think the best time is the mean of the two astronomical clocks, viz 5h 12m 36s. I estimate that the error of this time will not exceed about 1 second.
Thus were the rigorous minds of science applied to this one small problem of when the quake occurred—because it was realized that by knowing when, they could work out where. So an obsessive, super-pedantic dedication to precision, similar to that of Professor See’s, seems to have informed all of Andrew Lawson’s witnesses—such that in the end, and without fanfare, he was able to announce his own considered view of when the shocks first began, and when the biggest of all the shocks occurred, and where these two events had their origins.
He crunched all the data from four close-in observing stations—Mare Island, of course; George Davidson’s observations at his home in Lafayette Park, San Francisco (an account of which is in the prologue); the Students Astronomical Observatory at Berkeley; and the Lick Observatory at the top of Mount Hamilton near San Jose—and from one rather distant site, the International Latitude Station* in the town of Ukiah, about eighty miles north of the city.
From all of this data he was able to declare, definitively, that the shock began at 11 minutes and 58 seconds after 13 hours, Greenwich Mean Time (GMT)—in other words, two seconds before 5:12 A.M., Pacific Time, plus or minus three seconds. The place where this opening salvo was triggered was computed to be under the sea, just a short distance off the Golden Gate. And the depth of the vibrations’ beginnings—all computed from the simple geometries of the observations from these five sites—was about twelve miles, give or take about six.
And then there was the biggest shock—the terrifying “second shock” that so many of the observers could not fail to notice, whether they were rocked softly in their beds far away up in Oregon, or down in Anaheim, or over in Winnemucca, Nevada, or whether they were knocked flat by it in San Francisco, Santa Rosa, or San Jose. According to Lawson, it came exactly thirty seconds later, at 13h 12m 28s GMT, or at twenty-eight seconds after 5:12 A.M., Pacific Time. This shock, he went on to say, took place somewhere between the surface and twelve miles down, and at a spot close to the village of Olema, which has long considered itself—wrongly—to have been the earthquake’s epicenter (and where the locals, somewhat shiftily, still peddle hats and T-shirts to try to underline the point).
Olema was the place where, from the evidence of broken fences and torn-up roads and ruined barns—it appears that the surface of the earth was ruptured to a greater extent than anywhere else along the entire fault trace—as much as twenty-one feet. This was a pleasing happenstance for the scientists, because they were able to see that their calculations proved—or so they thought—that the most violent shock took place just below the point where the earth itself was most dramatically torn apart.
As it happens, Lawson and his fellow researchers were quite wrong—understandably so, given the paucity of the information they chose to examine. Had they looked closely at the entire seismic record they had available to them, and been inclined to focus on the data they had collected from all the other ninety-six seismographs, then they might have been able to make a more accurate fix. But there was a problem, in that they were not necessarily fully aware of what it was exactly that they were seeing in all of these records. In those seismically unsophisticated years some of the records must have seemed little more than runes or hieroglyphs or impenetrable codes. Only recently has the science advanced to the point where the records, combined with the anecdotes and the records of earth movement, can be annealed into a comprehensible whole. And fr
om all this has now emerged a rather better understanding of just what took place beneath the ground on that fateful morning.
Specifically, and as already noted, the place where it all began is now known to have been just a few cables northwest of the tiny guanocrusted island of Mussel Rock, off what were in the early years of the century thousands of acres of empty and open grasslands, with clifftop meadows dotted with dairy cows. The channel between the island and the mainland marks where the San Andreas Fault passes by: The island stands on the Pacific Tectonic Plate, and the mainland on its North American opposite number. The fault that divides them is the line along which the Pacific Plate moves, northward, by as much as thirty-five millimeters, one and a half inches, every year.
And at 5:12 A.M. that day, along that very line, just a few hundred yards off the island and some miles down deep within the zone of crushed rock where the rocks had been grinding together for decades, something suddenly gave.
With a lurch, the pent-up energy of all the years when the Pacific Plate (on which Mussel Rock stood) had been moving northward past the North American Plate (on which the grasslands, meadows, and cows all stood), was instantaneously released, and a foreshock, like a sudden ignition of a pile of explosives, vibrated out and through the rocks, causing the entire area’s underneath to pulsate in a huge, sonorous, basso profundo roar. That was what the policeman heard: The announcement of the foreshock, the enormous snapping sound of the rocks at the hypocenter breaking, lurching forward, and suddenly setting themselves free.
Once free, they moved. The pent-up forces that had held the Pacific Plate static for the better part of the previous two centuries were unleashed. The elastic rebound was now free to occur, and it did so in a jolt that some calculate as moving northward and southward at two miles a second. The rocks lying above the Pacific Plate then shifted, some of them by scores of feet. The vibration caused by this motion spread outward like a ship’s bow wave, and it set the earth above the north-shifting plate rearing up and outward, rippling and bucking just as Officer Cook had reported. It spread outward in different ways, depending on the nature of the rocks through which it spread: Sometimes it moved slowly and trembled fairly minimally; at other places it shot out with terrifying speed and caused immense rocking and crashing, which triggered immense destruction. Sites like Olema, where the quake seemed to go especially wild, are known as asperities—patches of geological roughness that corrupt the otherwise smooth progress of the outbound shock waves.*