Thousands of northern Californians gathered on front porches, hilltops and beaches during the unseasonably warm predown hours Thursday morning, watching the sky as the full moon slid into a total eclipse, and waited to see if the astronomical event would -- as predicted -- trigger a significant quaking in the earth.

Nothing happened.

The eclipse went off as expected, but beneath the earth, all was quiet.

But Jim Berkland, chief geologist for the county of Santa Clara and the man who forecast the quake, wasn't disturbed. He's been right before and he expects to be right again.

Berkland, 49, successfully predicted the strong tremor of Aug. 6, 14 hours before it struck, and this time, Berkland is "75 percent confident" that a quake between 3.5 and 5.5 magnitude will hit the San Francisco Bay area during an eight-day "seismic window" that began Sept. 5.

So far, Berkland says his controversial "seismic window" theory has proved right 16 out of 22 times since 1974. That's an accuracy rate of about 73 percent correct, far too high, he says, to be written off to chance.

The key to the theory is "syzygy," the straight-line conjunction of the sun, moon and earth. The resulting gravitational and tidal forces, claims Berkland, trigger shifts within already stressed areas along fault lines. Such eight-day syzygy periods, occurring twice a month, are the "seismic windows."

It was just such a syzygy, with the moon at its second-closest proximity to the earth, that reached its maximum Thursday morning. Berkland called it "the last 'good' seismic window of the year."

Berkland's theory isn't new -- it was first advanced by a French scientist, Alexis Perrey, in the late 1840s -- but the publicity generated by his espousal of its highlights the deep concern California residents feel about the possibility of another great quake like the one that flattened San Francisco in 1906.

Experts agree that another deadly earthquake is due. The only questions are where and when. The federal government is spending about $15 million this year for research into ways to find the answer. Berkland's approach, however, is not one of the techniques getting government money.

"A lot of people are looking at this approach," claims Berkland, "but the so-called experts are ignorant of what's going on." He admits he is a self-taught seismologist and says he got into prediction "by default."

Dr. Robert Uhrhammer, 33, a research seismologist at the University of California's Seismographic Station at Berkeley, says he has looked at Berkland's approach and "can lend no credence to any of his predictions or any of his methods."

"Nobody has been able to show a significant correlation between earthquakes and phases of the sun and moon. And it has never been conclusively shown that earth tides -- the one foot up and down movement of the earth's crust every 12 hours -- has a significant effect on the rate of earthquakes."

Berkland's track record, however particularly his successful prediction for Aug. 6, is impressing some people. Robert Olson, executive director of the state Seismic Safety Commission, says his predictions are becoming increasingly credible and should be reviewed by the state.

Scientists say the earth's crust is composed of great plates, and these plates are moving. These motions cause stresses that cause cracks (known as faults) in the crust. From time to time the stresses build up to the point where there is sudden movement along one of these faults. Shock waves sent out by these movements are earthquakes.

Along the San Andreas fault, which runs along The San Francisco peninsula, such stresses are present.

Scientists like Jack Evernden, a research geophysicist in the earthquake prediction group of the U.S. Geological Survey in Menlo Park, are looking at "seismic gaps". These gaps, he says, are missing parts of an earthquake pattern along a fault. The theory is that quakes begin at one point on a fault but another great quake will not occur at the same point until the rest of the pattern has been filled in.

If that's correct, Los Angeles -- which has not had a truly great temblor since 1957 -- should be jolted before San Francisco is.

Evernden's group is measuring earth strain, rock resistivity, magnetism, earth "tilt," gravity and radioactivity as means of guessing earthquake behavior, but they are avoiding syzygy.

"Berkland doesn't publish his data" says Evernden, "but I wish he would, so peer discipline could be brought to it.

"Quakes are random events. Berkland's talking about one of those straws that's going to break the camel's back, but the camel's getting tired from the big load it's been carrying for 200 years, and it appears most camels fall down quite independently of the last straw he's looking at.

"After L.A.'s had its big one, that's the time to worry if you're the worrying type. My advice is, just build your house on a rock and ride it out."