Joseph T. Pardee, the USGS geologist who first recognized landforms of this kind for what they were, was sitting in the audience during Bretz's 1927 roast. He, too, had worked in the Scablands region, as well as in Montana. In 1910, he had found evidence in the Missoula hills of a large glacial lake; the high-water marks, known as strandlines, are still visible today. (At the meeting, while Bretz was hemming and hawing his way through possible, if unlikely, sources for the water, Pardee reputedly leaned over and whispered to a colleague, "I know where Bretz's flood came from." He never told Bretz this, however.)

Later, while surveying Camas Prairie, Mont., in the late 1930s, Pardee discovered some curious rolling hills in the former lake's bottom sediments. They were so big that he realized what they were only after he flew over them: From the air, the hills became giant ripples, 50 feet high and 500 feet apart, which could only have been shaped by powerful currents, which themselves were only likely to occur if, for instance, an ice dam holding back an enormous lake had burst and that lake drained very, very quickly.

Pardee presented his findings at a 1940 meeting of the American Association for the Advancement of Science in Seattle, and later published them after his retirement in 1942. Why he had not been more immediately forthcoming is not known. Vic Baker, a hydrology professor at the University of Arizona who has studied the Ice Age Floods and made a hobby of their history, thinks that Pardee's silence was partly political: His supervisor was one of Bretz's most prominent critics, and Pardee, a retiring personality, was probably discouraged by the contempt that rewarded Bretz's dogged iconoclasm. "I don't think it's a coincidence that he waited until he retired to release that paper," says Baker.

But here, at last, was a source of water large enough to support Bretz's hypothesis. With Pardee's work, the resistance to Bretz's catastrophic explanations steadily eroded, although broad acceptance would not come for some time. When, in 1979 at age 96, Bretz was awarded the American Geological Society's highest award, the Penrose Medal, it was something of an attritional victory, bringing to mind Max Planck's adage that new ideas don't necessarily triumph on their merits so much as the deaths of their opponents. Bretz himself was all too aware of this. "All my enemies are dead," he told his son after he had won, "so I have no one to gloat over."

Later in the afternoon, Kleinknecht takes me to Dry Falls, near Coulee City. A series of cliffs nearly 400 feet tall and three-and-a-half miles wide, Dry Falls is one of the floods' more spectacular leavings. It marks the middle of the largest channel in the scablands, the Grand Coulee, a canyon 50 miles long, five miles wide, and 900 feet from rim to floor in places. Here, floodwaters ran more than 250 feet deep, rushing over the basalt shelf and ripping out large chunks of it; at their peak, the falls would have been a small dip in a swift cascade. Now, they sit almost absurdly empty.

Sights like this are what tease the imagination, giving the floods their made-for-TV appeal. The Discovery Channel has run a couple of specials on them, and NOVA did an episode in 2005, Mysteries of the Megaflood, which is for sale at the Dry Falls interpretive center. The show happens to be playing on a small TV inside, and I watch, transfixed, as the announcer intones things like, "Few areas on Earth are as mysterious and controversial as the Channeled Scablands," and "No river in the world can form what you are about to see," as if I am about to be treated to something more titillating than odd landforms. The jump cuts and synthesizer music are weird, but it's also strangely riveting when, to a pounding beat, a cellophane CGI wave crashes down a valley and sweeps away all the trees.

I'm not sure how long I've been slack-jawed before I notice that Kleinknecht is behind me, watching me watch the show. He looks mildly disappointed.

"Too dramatic," he mutters, which seems to be my cue to leave.

Outside, I ask Kleinknecht what about the NOVA episode he doesn't like. "That stuff shouldn't be necessary," he says. "The floods don't need that kind of treatment to be interesting." He considers for a moment. "The thing for me about the floods is what they make you do with your mind if you want to understand what happened." How they mess with our idea of linear time. How we have, in recent years, become all too familiar with the idea of a planet that can undergo sudden, catastrophic shifts. How we have to look at the landscape in its apparent stasis and know that what we see is really layers of activity, different episodes -- not one flood but many. How some geologists think that floods may have roared millions of years ago, during even more ancient ice ages. How the Earth has its own long rhythms, warm and cool, and the planet may be warming now, but someday it will cool again, and the floods will return.