The Sailing Stones of Death Valley: How Paper-Thin Ice Pushes Rocks Across a Dry Lake

In a high, empty valley on the northwest side of Death Valley National Park lies a dry lakebed so flat that a marble would barely know which way to roll. Scattered across it are hundreds of rocks, some no bigger than a fist and some as heavy as a piano. Behind many of them, scratched into the cracked mud, is a long, shallow groove. Some grooves run straight for hundreds of feet. Some curve, zigzag or double back. A few rocks sit at the end of a trail with a twin trail running beside it, as if two stones had walked side by side.

Nobody carried them. There are no footprints, no tire marks, no drag marks from animals. For most of the 20th century, nobody had seen a single one of these rocks move, and geologists argued about the cause for more than 60 years. The place is called the Racetrack, and its “sailing stones” became one of the most famous puzzles in American geology.

The answer, when it finally came in 2014, was stranger than the wildest guesses and gentler than all of them. The rocks are moved by ice no thicker than a windowpane, pushed by a breeze you could fly a kite in. And the people who caught them in the act were a pair of cousins who had expected to wait a decade and got lucky in two years.

A lakebed flat to an inch and a half

The Racetrack is a playa, a lake that is almost always dry. It sits about 3,700 feet above sea level, is roughly 2.8 miles long and 1.3 miles wide, and is ringed by mountains. Its surface is fine clay that dries into a mosaic of hand-sized polygons. Fine sediment washing in has made it astonishingly level: the north end is only about 1.5 inches higher than the south end, across nearly three miles. When rain or snow does fall, water drains in from the surrounding slopes and spreads into a shallow, short-lived lake, which is the first clue to what happens next.

The rocks come from the hills. Most are dark dolomite that breaks off a steep 850-foot promontory near the playa’s southeast shore, along with tan syenite from the slopes to the west. They tumble onto the playa edge and then, somehow, set off across it. Most moving stones are 6 to 18 inches across, and their trails are typically 3 to 12 inches wide and less than an inch deep. The National Park Service says some have traveled as far as 1,500 feet.

Dust devils, airplane propellers and slime

The first known account of the trails dates to 1915, when a prospector named Joseph Crook from Fallon, Nevada, came through. Science caught up in 1948, when geologists Jim McAllister and Allen Agnew, mapping the area’s bedrock, published a short note in the Geological Society of America Bulletin. They called the rocks “scrapers” and suggested they were shoved over wet mud by gusty winds, like those that spin up dust devils.

That idea ran into the obvious problem: rocks are heavy. In 1953 the geologist John Shelton tried it with brute force. As the NPS research database summarizes his experiment, he wetted a patch of playa and blasted it with the propeller wash of a light airplane. Some rocks did move in the artificial gale of 40 to 45 mph, but since nothing about the setup was natural, the test was inconclusive. He also floated the idea that slimy films of algae on the wet mud might make the ground slicker.

Later friction tests made wind look even weaker. When researchers towed rocks across wet playa mud and did the math, they concluded that most stones would need winds of 75 mph or more to budge, and some flat ones far more. Meanwhile the geologist George Stanley argued in 1955 that the rocks must be frozen into sheets of ice that catch the wind and carry them along. Wind or ice, or both: the debate would last for half a century, mainly because no one could be there at the right moment.

Mary Ann, Nancy and the rock called Karen

The most patient early study came from Caltech geologist Robert Sharp and his colleague Dwight Carey. In May 1968 they began tracking 30 rocks with fresh trails, giving each a name and marking its position with stakes. They checked back for seven years. Ten rocks moved in the very first winter, led by Mary Ann, which traveled 212 feet. By the end, all but two of the 30 had moved, and none was ever confirmed to have moved in summer. The champion was Nancy, a stone just 2.5 inches across, which covered 860 feet in total, 659 feet of it in a single winter.

Sharp and Carey also tried a test of the ice theory. They hammered a ring of rebar stakes, about 5.5 feet across, around a small rock. If the rock were frozen into a big ice sheet, they reasoned, the stakes would snag the ice and hold the rock back. Instead, the rock left the corral in the first winter, slipping past a stake on its way out. That result was read for decades as a strike against ice.

Then there was Karen, a block of dolomite about 29 by 19 by 20 inches, weighing an estimated 700 pounds. Karen sat at the end of a straight 570-foot trail and never budged during the study. Then, sometime before May 1994, Karen vanished. There was no sign of a truck or winch on the fragile mud. It turned up again in 1996, when the geologist Paula Messina rediscovered it.

Matching trails and ice rafts

In 1995 John Reid of Hampshire College took a group of students to the playa and mapped trails from rocks that had moved in the late 1980s and the wet winter of 1992 to 1993. They found trails that matched each other turn for turn, from rocks as much as 830 meters, or about half a mile, apart. Rocks of different sizes made the same moves and generally didn’t tumble. Wind alone couldn’t easily explain that kind of choreography. A single moving sheet of ice could.

Other work pointed the other way. Messina and her co-author Stoffer later used precise GPS to map the rocks and their trails, and found enough differences between neighbors to suggest the rocks often moved independently. And in 1996 the physicists Bacon, Cahill and Tombrello noted that the layer of slowed air just above a smooth, flat surface can be only about 2 inches thick, so even low rocks catch nearly the full force of gusts that can top 90 mph in winter storms.

In 2011 Ralph Lorenz, a planetary scientist at the Johns Hopkins University Applied Physics Laboratory, and his colleagues offered a twist in the American Journal of Physics under the title “Ice rafts not sails.” Perhaps ice didn’t need to act as a sail at all. A collar of ice frozen around a rock could float it partly off the mud, making it so light against the bed that almost any wind could nudge it along. They demonstrated the idea with a kitchen experiment. They froze a 64-gram pebble into a slab of ice in a plastic food container, floated the slab in a baking tray of water over a layer of sand, and found that a very gentle push, or just blowing across the surface, made the raft drift while the pebble scratched a neat trail in the sand.

“The most boring experiment ever”

The team that finally settled it formed for the fun of it. Richard Norris, a paleobiologist at the Scripps Institution of Oceanography, and his cousin Jim Norris, an engineer at a Santa Barbara firm called Interwoof, set up what they called the Slithering Stones Research Initiative, a loose crew of friends and relatives who helped haul gear into the remote valley. Their colleague Jib Ray helped design the electronics. Lorenz and Brian Jackson of Boise State University joined from a different direction. They had already been running cheap time-lapse cameras at the playa to study dust devils and desert flooding, partly because playas like the Racetrack make a useful stand-in for a hydrocarbon lake on Saturn’s moon Titan.

The National Park Service wouldn’t let them use the playa’s own rocks, so the team brought in similar limestone from outside, bored a cavity into each block and fitted 15 of them with custom GPS loggers. Each logger sat over a magnet buried in the mud and would start recording every second the moment its rock moved off it. A weather station nearby logged gusts at one-second intervals. The setup went in during the winter of 2011, and then everyone waited. Lorenz suspected it would be “the most boring experiment ever.” Rocks at the Racetrack can sit still for a decade.

December 2013: the pond freezes

They didn’t have to wait long. A regional storm on November 21 to 24, 2013, dropped about 1.4 inches of rain and roughly 8 inches of snow. Water collected on the southern third of the playa in a pond only a few inches deep, and night after night it froze. When Richard and Jim Norris arrived that December, they found the rocks standing in a frozen lake.

The GPS rocks logged the first move on December 4, when two of them started moving within six seconds of each other, even though they sat about 500 feet apart. Both traveled about 210 feet over 16 minutes, moving at up to 16 to 20 feet per minute. On December 20, the cousins watched more than 60 rocks move in one event. “On Dec. 21, 2013, ice breakup happened just around noon, with popping and cracking sounds coming from all over the frozen pond surface,” Richard Norris said later. “I said to Jim, ‘This is it!'” On January 9, 2014, Jim Norris and Lorenz caught a rock on camera sliding for about 18 seconds past rocks that stayed put.

The fresh trails were invisible at first, hidden under muddy water. Only in the afternoon, as a steady southerly breeze pushed the water toward the north shore, did they appear, dozens of them, curving in near-perfect unison.

How windowpane ice moves a rock

The team published their observations in PLOS ONE on August 27, 2014, and the mechanism they described needs a precise recipe. First, the playa has to flood with water deep enough to freeze into floating ice but shallow enough that the rocks still poke through. Second, cold nights have to freeze the pond into a thin, clear sheet the researchers called “windowpane” ice, only about an eighth to a quarter of an inch thick. Third, on a sunny day the ice has to start melting and breaking up around midday. And fourth, a light, steady wind has to be blowing.

When all four happen, the ice sheet cracks into floating panels that can be tens of meters across. The wind, which only needs to be about 10 mph, drives those panels across the pond, and water flowing underneath helps. Where a panel runs into a rock, it doesn’t lift it. It shoves it, like a slow bulldozer. Broken ice piles up on the rock’s upwind side, which gives the ice an even bigger surface to push against. The rock slides across the soft, wet mud and plows its trail.

The speeds were the real surprise. The rocks moved at only about 7 to 20 feet a minute, slower than a stroll. “It’s possible that tourists have actually seen this happening without realizing it,” Jim Norris said in the Scripps announcement. “It is really tough to gauge that a rock is in motion if all the rocks around it are also moving.” One instrumented rock moved more than once that winter and ended up 224 meters, about 735 feet, from where it started. The team also saw grounded ice panels scrape trails with no rock at the end, the kind of trail the park had suspected was left by thieves.

The explanation tidied up the old contradictions. Large ice panels explain why rocks hundreds of feet apart make matching trails. Ice that cracks between neighbors explains why some rocks part ways. It may also explain the famous corral. The team noted that one stake stood just upwind of the rock that stayed inside, and could have shattered an incoming ice sheet before it reached it. The thin ice was too weak to float the rocks as rafts, though, so the ice-raft tray experiment, neat as it was, didn’t describe what happened in 2013 and 2014. Like the steady wind that brought down Galloping Gertie, it wasn’t a violent force but a modest one, applied the right way.

The same thing happens elsewhere. Rocks pushed by lake ice leave trails on the bottom of Great Slave Lake in northern Canada and on the shores of the Baltic. Death Valley just has to work much harder to make the ice.

A phenomenon that may be fading

Rocks at the Racetrack move very rarely. “The last suspected movement was in 2006, and so rocks may move only about one millionth of the time,” Lorenz said in 2014. That same year, he and Jackson published a study in Geomorphology that combed the historical record of rock movements and found odds of about 4 to 1 that movement had become less common in the 21st century than in the 1960s through 1990s. Fewer freezing nights are a likely factor, since without a hard freeze there’s no ice. Movement needs cold nights, a flood and a breeze all in the same few weeks, and in a warming desert that combination is getting harder to come by.

The cousins were careful not to overclaim. The rocks they saw move were modest stones. “We have not seen the really big boys move out there,” Richard Norris said. “Does that work the same way?” A 700-pound block like Karen may need a deeper pond, thicker ice or a stronger wind, and no one has caught one in the act.

Tire tracks and stolen stones

Fame has been hard on the Racetrack. In 2013, the Los Angeles Times reported that park officials were investigating the theft of several rocks. In August 2016, someone drove an SUV onto the playa and spun and swerved across it, cutting about 10 miles of ruts into the clay. Ruts like that can last for years on a surface that only heals when it floods. In February 2018, 19 volunteers from the Park Service and the group Friends of the Inyo spent a weekend with garden rakes and 750 gallons of hauled-in water smoothing out 512 feet of tracks, so the natural mud polygons could form again.

Seeing the sailing stones yourself

Getting there is an expedition. The Racetrack is about 200 miles northwest of Las Vegas, much farther out than the classic Vegas day trips. The usual route is a rough gravel road that runs 28 miles south from Ubehebe Crater. The park recommends high clearance and heavy-duty tires, and warns that standard rental cars often get flat tires. From Furnace Creek, the drive takes at least three and a half hours each way. There’s no cell service, and the nearest gas is more than 60 miles away. Bring extra water and be ready to spend the night if something breaks. Avoid summer, when parts of the park top 120°F.

For the rocks, the park suggests driving about 2 miles south of the Grandstand parking area, then walking at least half a mile toward the playa’s southeast corner. Stay off the playa when it’s wet, since footprints in soft mud last for years. Never drive on it or move a rock, which spoils the trails for everyone else and for the science. If the road is too much, the park points to Bonnie Claire Playa, just over the Nevada line east of the park, where trails from sliding stones can be seen near the highway on land managed by the Bureau of Land Management.

While you’re in the north of the park, the Eureka Dunes are one of the few places on Earth where sand booms like a cello when it slides. The singing sand and the sailing stones are both natural mysteries with everyday physics behind them, much like Saudi Arabia’s Al Naslaa rock, split so cleanly it looks cut by a laser.

What sticks with you about the sailing stones is how undramatic the truth turned out to be. No hurricane winds, no magnetic fields, no pranksters, no mystery force. Just a few inches of meltwater, a skin of ice and a sunny winter breeze, working so slowly you could stand among the rocks and miss it. Like the slowest experiments in science, it took patience, decent instruments and a bit of luck. As Richard Norris put it: “Science sometimes has an element of luck. We expected to wait five or ten years without anything moving, but only two years into the project, we just happened to be there at the right time to see it happen in person.”

Further reading