Kerry Sieh was a 26-year-old geologist in 1976 when he struck out with his wife, his brother, a handful of graduate students and a couple of shovels for Pallett Creek, a marshy patch in the desert on the southern San Andreas in northern Los Angeles County. They started digging a trench by hand. By noting where the ground had shifted and dating the decomposed vegetation around those shifts, Sieh began to construct a "rupture chronology" of the fault. In time, he pulled together enough funds to rent a backhoe and dig down through two millennia of geological time. The "several meters of strata freshly exposed in the cut before me had seen fully 2,000 winters and summers and autumns and springs," he wrote in 1980, "the coming and going of the Little Ice Age, generations of forest and chaparral, animals and Indians, and even several great earthquakes."

Sieh is now known as the father of paleoseismology -- the fine art of determining earthquake patterns by logging data revealed in trenches. When he started out, he was able to nail down earthquake dates only to within 30 years. Today, Sieh and other geologists can be more precise. Kate Scharer, an assistant professor at Appalachian State University who works with Biasi, goes into the trenches with a microscope and tweezers to extract just one species of grass from the peat. She then dates it and compares it to another substance in the same layer -- maybe a piece of charcoal that burned 50 years before it ended up in the soil. "We've been able to get the error ranges on those sites down to plus or minus five years," Hudnut says. "Now we're chipping away at it, trying to get it down even tighter than that."

In 2005, Scharer and Biasi, along with USGS geologist Tom Fumal and Ray Weldon from the University of Oregon ("our visionary," says Biasi), took the published data on 56 prehistoric earthquakes, derived from 12 sites on the southern 400 miles of the San Andreas Fault, and plotted them in three different scenarios. Their graph, with its brightly colored lines stretching across segments of the fault over time, reveals that earthquakes may have been happening on the southern San Andreas as frequently as once every 100 years. At any rate, no lull in the past 1,600 years has gone on longer than 200 years. It defies logic to think that the current calm will last much longer.

What will happen to Southern California when a 7.8 earthquake breaks open on the San Andreas Fault, exactly where Hudnut calculates that it might? Unstable desert sand will turn into something like barely set gelatin and many of the houses built on it will collapse, as will any remaining unreinforced masonry buildings (known as "URMs" -- or, as one seismic engineer calls them, "FPRs": Future Piles of Rubble). Hillsides throughout the San Gabriel Mountains and on Cajon Pass will collapse in cascades of rock and debris; dams will splinter. Seventy seconds into the rupture, seismic waves will reach deep into downtown Los Angeles, where 55 seconds of sustained shaking, measured from the moment the waves hit, could bring down older "steel-moment" high rise buildings.  (By contrast, a 6.7 earthquake that hit Kobe, Japan, in 1995 -- one year to the day after Northridge -- shook for seven to 10 seconds. One third of the city's 600-some steel-moment frame buildings came down.)

Fire will likely follow the destruction: The Los Angeles Department of Water and Power anticipates 1,600 ignitions in the urban wildlands interface, hundreds of which will merge into larger blazes before they can be contained. Throw a dry, hot Santa Ana wind into that scenario and Los Angeles, like San Francisco in 1906, may well burn to cinders.

Closer to the fault, it will be mayhem. In 1857, Fort Tejon, the estimated epicenter of that year's quake, was an outpost inhabited by a handful of military personnel, Native Americans and 28 camels. The 7.8 temblor temporarily changed the courses of several rivers, tossed fish out of nearby lakes and left a surface rupture more than 300 miles long. But only two people were killed, one of them crushed when her adobe house collapsed. Nowadays, however, the Los Angeles exurban area stretches north beyond that fault segment, and several million people live along its edge.

Back on Google Earth, Hudnut zooms in on a construction site teetering at the edge of the fault. He worries. "They want to get them right along the fault because that's where you have the nice view. It makes me think perhaps our state law is not strict enough."

Hudnut worries, too, about the "lifeline corridors" entangled with the fault. Interstate freeways run alongside the San Andreas; aqueducts, natural gas lines and electrical transmission corridors crisscross it. Hudnut scrolls through the landscape to find them. "Here's where the state water project comes down," he says, tracing his finger along a thin dark line on his computer screen. "You have Interstate 15, two major freight rail lines and a lot of fiber-optic cable running along Old Route 66. You have gasoline pipelines and high pressure gas lines and also high-tension power lines all nested in close together." All of them thread through the Cajon Pass, a major gateway into the northern Los Angeles Basin. Hudnut fears that a rupture today would leave metropolitan Los Angeles without water or power for weeks, maybe months.

"These lifelines that cross the fault, they were not designed not to break. We don't know for sure, but we think they're going to all break. People haven't thought that through. People haven't decided yet that they really need to fix these so that they won't break. I'm not sure they'll never decide that, but we're trying to encourage people to think about what the real consequences would be."

Before he died in 1952, Andrew Lawson built himself a house in the Berkeley Hills that was designed to withstand any earthquake or fire. It has yet to be tested by geology, but the fire that swept through the area in 1923 didn't even singe a library book inside. A little engineering goes a long way: As seismologist Nick Ambraseys famously said in 1968, "Earthquakes don't kill people. Buildings do."

Hudnut might like to amplify that: Earthquakes may not obliterate civilizations, but weak infrastructure can. As many have noted, it wasn't the shaking ground that destroyed San Francisco. It was the fact that the shaking cut off the city from its water supply, so no one could fight the fires that followed.

In Southern California's infrastructure, Hudnut says, "We think we've identified something like the levees in Katrina. The scientists saw the vulnerability of the levees; they predicted what could happen. But there wasn't the political will to raise the money to fix them." He hopes Los Angeles can avoid a similar fate. But time may be running out. And geological forces don't wait for budgets -- or repair crews.