In his autobiography, Conquering the Last Frontier, Olympic Peninsula pioneer Thomas Aldwell described his first encounter with the land that would be his legacy: "Below the cabin was a canyon through which the Elwha River thundered, and 75 feet or so in front of it was a spring of crystal clear water, overhung by vine maples. ... that spring embodied all of life and beauty I thought I'd ever want."

The Elwha runs fast and steep, from its headwaters in Washington's Olympic National Park to the Strait of Juan de Fuca. At the time that Aldwell stood on the banks, salmon swam its length, including 100-pound monster chinook. In the push of current, he saw not just beauty and fish, but power -- power to light a town, to attract industry, to put Port Angeles on the map. He bought the land, and in 1913, the Olympic Power Development Company, Aldwell's brainchild, completed the Elwha Dam.

Now Lake Aldwell is draining like a dirty bathtub, leaving thin silt coating tree stumps, roots, the odd tin plate. The floodgates are open. Water roars through the canyon again, stained brown by sediment already leaching from behind the dam.

On Sept. 17, with the removal of ceremonial chunks of concrete, both the Elwha Dam and the Glines Canyon Dam, built further upriver in 1927, began to come down. A century after they were built, they powered only 40 percent of a single paper mill, the last one in Port Angeles. The toll on struggling salmon runs no longer seemed worth it. Started in 1992 by the Elwha River Ecosystem and Restoration Act and finally put in motion with the help of $54 million in federal stimulus funds, the $325 million restoration will be the second-largest in the National Park Service system, after the Everglades. The removal of Glines Canyon Dam, 210 feet high, will be the biggest dam decommissioning in the United States.

In some ways, the Elwha is a perfect test case for restoration science, a chance to see and document exactly what happens when a big dam comes down. Because much of the river is on protected national park land, there's no development or pollution to complicate restoration. "We really only have, at least on the upper part of the watershed, one major problem -- the dams," says George Pess, leader of NOAA's Northwest Fisheries Science Center restoration effectiveness team.

Restoration needs good experiments. It is a young field, striving to become more of a science after years of projects done haphazardly, without much monitoring or thought for the larger picture. There is money and need and desire for restoration, but only now a developing sense of what really works.

The idea of letting the river find its natural course and studying the outcome is attractive, but not everyone has the stomach for it. Park and tribal biologists are scrambling to anticipate the effects of the increased flow and the mudslide of sediment -- 18 million cubic yards worth -- as the dams come down. The flood of dirt and rock, now piled in the lakes behind the dams, will kill most of the fish in the river below, according to NOAA's 2008 Elwha Fish Restoration plan.

It raises several questions: How much should the river define the terms of its re-emergence, in the process providing scientists with a living laboratory? How much should biologists and engineers encourage it to take the shape they want? What is a desirable shape for a wild river? And how, with so little of the project money earmarked for monitoring, would anyone know?