Just when one of the most startling new theories in science seemed to be gaining acceptance, challenges have revived doubts.

The theory is that mass extinctions, including the one that wiped out the dinosaurs, occur at regular intervals of 26 million years or so when extraterrestrial forces fling asteroids or comets toward collisions with the Earth.

According to this scenario, the collisions throw up enough dust to shroud the Earth, blocking solar heat and bringing on a freeze that devastates a perennially tropical world. It was this scenario that led to the nuclear winter hypothesis, bombs taking the place of asteroids.

Even as prominent an evolutionary biologist as Harvard's Stephen Jay Gould had declared himself convinced at least by the asteroid explanation for the demise of the dinosaurs.

Now, however, several groups of scientists are challenging the theories on what causes these supposedly periodic collisions. These theories include an unseen companion star to the sun, called Nemesis, and a 10th planet, dubbed Planet X.

One group is suggesting there was no impact at all but a spate of volcanic eruptions that lofted enough smoke and ash to do the same thing. Yet another group is challenging the idea that mass extinctions, regardless of cause, have been periodic.

The challenges are a reminder that few ideas in science are immune to questioning. Despite science's popular image as a source of immutable truths, most of its pronouncements and publications are based on fallible interpretations of imperfect data.

A new theory -- or, as scientists prefer to call it, a new hypothesis -- gains credibility only after it has been confirmed by others working independently and has survived all challenges, something the periodic-impact hypothesis has yet to do.

The hypothesis arose in 1980 when Luis W. Alvarez, a physicist, and his son Walter Alvarez, a geologist, both at the University of California at Berkeley, speculated on the origin of an unusual layer of clay formed at the time the dinosaurs died out. The layer marks the boundary between deposits laid down during the Cretaceous period, the last great age of dinosaurs, and the ensuing Tertiary period, when mammals began to diversify.

The clay had a startlingly high concentration of iridium, which is rare in the Earth's crust. Iridium, however, is thought to be more abundant in asteroids.

The Alvarezes speculated that the extra iridium came from an asteroid of perhaps six miles diameter that struck the Earth with enough energy to vaporize the asteroid and its iridium. When the dust and the iridium fell to Earth, the Alvarezes suggested, it formed the boundary layer.

The impact theory took on a new dimension in 1983 when two University of Chicago paleontologists, David Raup and John Sepkoski, reported that the mass extinctions seem to have occurred at 26 million-year intervals.

This set off the search for long-period cycles. The search led to outer space because no one could think of any Earthly cycles that operate over such long periods. All the extraterrestrial mechanisms involve bringing a source of gravity periodically near a vast cloud of objects that orbit the sun beyond the path of Pluto. Called the Oort cloud, it contains millions of massive chunks of matter, rocky asteroids and icy comets.

Any source of gravity that could slow the speed of orbiting objects would cause them to fall out of orbit and, like a decaying artificial satellite, spiral down toward the sun. Such a cloud would intersect the path of every planet as it fell.

One gravity source -- quickly discounted -- was the vast cloud of loose molecules that occupy the plane of the galaxy. In the aggregate, their gravity is equal to that of millions of stars. Because the solar system bobs "up" and "down" through this plane every 33 million years as it orbits the center of the galaxy, two scientists at the National Aeronautics and Space Administration's Goddard Institute for Space Studies in New York had speculated a few months ago that the molecular cloud's gravity could dislodge asteroids from the Oort cloud.

Patrick Thaddeus and Gary Chanan of Columbia University now have shown that the molecular cloud is so thick that the solar system never really escapes it. Only if it were more concentrated in a small layer could it have much effect.

The Planet X hypothesis, created by Daniel Whitmire and John Matese at the University of Southwestern Louisana, has fallen aside on the grounds that it could not produce a sharply defined release of objects from the Oort cloud.

The idea was that Planet X orbited within the Oort cloud, sweeping out a gap that separated the cloud into two rings. Every 26 million years continual shifts in the planet's orbit would bring it near a ring and dislodge a swarm of asteroids.

Scott Tremaine of the Massachusetts Institute of Technology now has shown that any planet big enough to clear a gap in the Oort cloud could not be small enough to avoid pulling more distant asteroids into the gap, filling it. Unless the gap stays devoid of asteroids, there can be no sudden event of Planet X grazing the ring and triggering an asteroid shower.

The Nemesis star, proposed by Marc Davis and Richard Muller of the University of California at Berkeley, and Piet Hut of Princeton's Institute for Advanced Study, although still a possibility, looks like a long shot.

For Nemesis to be a cause, it would have to be in a highly elliptical orbit that brings it near the solar system every 26 million years.

Although binary star systems are the rule in the galaxy, Hut now calculates that any companion star of the sun in such an orbit would have been torn away by outside gravitational influences in less than a billion years. Nemesis would have to have survived the 4.5 billion years since the solar system formed.

Because every hypothetical mechanism for periodic extinctions is in trouble, some scientists are reexamining the evidence that mass extinctions are periodic at all.

Raup and Sepkoski never said the evidence was clear cut. Their interpretation was based on comparisons of times at which various species disappear from the fossil record. Extinctions occurred at all times, but at some times there were many more than usual -- from 20 percent to 70 percent of the known species dying out at what looks like, given the coarseness of the fossil record, the same time.

Unfortunately, as several paleontologists have pointed out, the reliability of dating extinctions is not always high. Moreover, the peaks in a graph of presumed extinction rates do not occur in precise 26-million-year intervals. About half do, but others are off by a few million years or at levels below the rates of extinction between other peaks.

The interval also has been questioned on the grounds that randomly fluctuating extinction rates could produce peaks that look periodic. Tremaine and a colleague, Julie Heisler, have analyzed randomly generated sets of hypothetical extinction rates and concluded that there is a 10 percent chance that truly random extinction rates could have fooled Raup and Sepkoski.

In science, if not in other fields, a 90 percent confidence level -- another way of expressing the same finding -- is considered worrisomely low. For most purposes, 95 percent confidence is thought to be the lowest that merits the term "statistically significant."

Even if the periodic extinction hypothesis were discarded, it still could be possible that one impact was responsible for ending the Cretaceous period 65 million years ago -- the biggest mass extinction on record.

Even this, however, is under challenge.

Because the unusual iridium layer has been found in several parts of the world in exactly the right stratum (a core sample from the North Pacific floor, for example, shows only one iridium-rich layer -- exactly at the 65-million-years level) there is agreement that something strange happened then.

Charles Officer and Charles Drake of Dartmouth College say the iridium layer's other components point to a volcanic origin. Iridium is higher in the Earth's molten mantle than in the crust. A massive series of volcanic eruptions, the two say, could have produced the iridium layer.

Officer and Drake also say the iridium layer was not deposited all at once but over a period of 10,000 to 100,000 years. Beyond suggesting vague climatic effects from a long reign of volcanic gases, however, they do not say how the eruptions might have led to a mass extinction.

Until further study sorts out conflicting lines of evidence, all that is certain is that there have indeed been times when something happened to wipe out half or more of the Earth's life forms