The Sailing Stones of Racetrack Playa

Between 1948 and 2014, the sailing stones absorbed a lot of theories. Some were reasonable. Some, per the researchers' own accounts, ranged from the complex to the supernatural. The problem was always the same: nobody had watched the rocks move, so every explanation was reverse-engineered from tracks in the mud.

Here's roughly how the science stacked up:

• Dust devils and wind alone. The 1948 McAllister and Agnew report suggested strong gusts propelled the scrapers across a muddy playa. In 1953, Shelton actually tested this by running an aircraft propeller wash over wetted playa surfaces; the experiments showed winds over 20 m/s could move natural rocks, possibly aided by algal films lowering friction. Later friction tests kept raising the bar — W. Sharp's static and dynamic towing tests put the requirement at 33 to 45 m/s, and other calculations pushed it up to 80 m/s for low-profile rocks. Death Valley winters don't reliably produce that.
• The rocks are too heavy. George M. Stanley, publishing in 1955, noted that some stones weigh as much as a human and argued the wind simply couldn't do it alone. He proposed ice sheets around the stones either caught the wind or dragged the rocks along.
• Big ice sheets. Reid et al. (1995) found highly congruent trails from rocks up to 830 m apart, implying ice sheets that could be up to half a kilometer wide, and noted that parallel-moving rocks of different sizes usually did not rotate or tumble — a pattern that points to ice, not wind.
• Ice rafts, floating the rocks off the bed. A separate line of thinking held that ice cakes formed around each rock and buoyantly lifted it off the mud, reducing friction so much that arbitrarily light winds could move it. If true, tracks would be shallower than a dragging rock would leave.
• The boundary-layer wrinkle. Bacon et al. (1996), informed by work at Owens Dry Lake, found that winds compress and intensify over a playa's smooth surface, and that the slow-wind boundary layer near the ground can be as thin as 5 cm — so even a short rock feels close to the full force of a gust, which in winter storms can reach 140 km/h.

Every hypothesis explained some of the data and stumbled on the rest. Wind alone couldn't move the heavy stones. Thick-ice flotation didn't match the shallow tracks. And Messina and Stoffer's submeter GPS mapping around 2000 found deviations in trails that suggested rocks moving independently rather than locked to one sheet. The record kept refusing to resolve.

Bob Sharp and Dwight Carey ran the definitive early monitoring program starting in May 1968. They labeled 30 stones, gave each a name, staked their positions, and tracked them for seven years. The characters that emerged are the closest thing this mystery has to a cast.

• Mary Ann (stone A) covered the longest single-winter distance in the first year: 65 m. Ten of the initial 30 stones moved that first winter.
• Nancy (stone H), the smallest monitored stone at 6.4 cm in diameter, moved the greatest cumulative distance — 260 m — including a single-winter run of 201 m. The largest stone to move over the study was 36 kg.
• Karen (stone J), a 74-by-48-by-51 cm block of dolomite weighing an estimated 320 kg, never budged during the monitoring period. Her old 170-meter track may have come from momentum off her initial fall onto wet playa. Then she vanished sometime before May 1994, possibly during the unusually wet winter of 1992–93. A truck-and-winch theft was ruled unlikely — there was no damage to the playa. San Jose geologist Paula Messina rediscovered Karen in 1996, roughly 800 m from the playa.

By the end of the seven-year study, all but two of the monitored stones had moved. No stone was ever confirmed to move in summer. Sharp and Carey also ran the famous "corral" experiment — seven rebar stakes driven in a 1.7-meter circle around a track-making stone weighing under half a kilogram. If ice collars were dragging rocks, the rebar should have deflected them. The stone escaped anyway, moving 8.5 m to the northwest, barely missing a stake. Two heavier stones dropped in the same corral behaved differently: one moved five years later in the same direction, its companion never moved at all. Sharp and Carey read this as evidence against ice being the driver. The result was ambiguous, and it stayed ambiguous for almost forty years.

The mystery ended on a sunny December morning because the right people had finally wired the playa for it. Richard Norris, James Norris, Ralph Lorenz and colleagues — publishing in PLoS ONE in August 2014 as the first direct scientific observation of the rocks in motion — had installed a dedicated weather station on the alluvial fan, several time-lapse camera systems overlooking the southeast corner, and 15 GPS-instrumented limestone blocks. Each block, cut from the Permian-aged Darwin Canyon Formation, carried a custom logger built by a firm called Interwoof, set in a bored cavity and rigged to start recording the instant the rock pulled away from a magnet buried beneath it.

On December 20, 2013, more than 60 rocks moved in a single event — the largest ever recorded. And the mechanism turned out to be almost anticlimactically gentle. Thin "windowpane" ice sheets, just 3 to 6 mm thick, formed on a shallow winter pond (about 10 cm at its deepest) overnight as temperatures dropped below freezing. In the late-morning sun they began to melt and break up under light winds of about 4 to 5 m/s, accompanied by widespread popping sounds as the ice fragmented. Floating ice panels tens of meters wide then shoved the rocks along at 2 to 5 m/min. Trails formed under the ice-covered water and only became visible when light winds blew the muddy water away — which is exactly what happened by 3:15 pm that day, revealing more than 60 fresh trails as the southern pond drained from about 7 cm deep to under 1 cm.

The GPS data is the part that closes the case. On December 4, 2013, two instrumented rocks 153 m apart — one 16.6 kg, one 8.2 kg — began moving within 6 seconds of each other, each traveling roughly 65 m over 16 minutes, both starting at 11:05 am and both slowing from 5–6 m/min to 3–4 m/min as the event went on. That's not wind independently nudging two stones. That's a single sheet of ice, moving them together. The surprise wasn't the ice — it was how little of it you need: sheets a few millimeters thick, far too thin to float a rock, were enough to bulldoze one across the desert. The force comes from ice panels stacking on the upstream side of each rock, increasing the surface area exposed to wind and to the water flowing underneath — and the stones slide rather than roll, over a fully saturated mud surface with almost no friction.

Freezing nights and light daytime winds are common at Racetrack Playa. What's rare is the water. The whole phenomenon depends on a specific, fragile sequence.

• A flooded playa surface, deep enough to submerge the southern rocks but shallow enough to leave many partly exposed
• A thin clay layer underneath, saturated and slick
• Overnight temperatures below freezing to form the ice
• Steady light daytime winds — around 3 to 4.5 m/s — to drive the ice and push the water across the pond
• Morning sun to trigger the breakup near midday

The 2013–14 pond came from a single storm: 3.61 cm of rain plus about 20 cm of snow on November 21–24, 2013, for roughly 5.64 cm of total precipitation. The resulting pool persisted until it evaporated in the second week of February 2014, long enough to support multiple movement events — one instrumented stone logged a total trail of up to 224 m across separate moves. Time-lapse records kept since 2007 show how uncommon this is. Winters in 2012 and 2012–13 were essentially dry; a 30-day flood in late winter 2010 rarely dropped below freezing, so little ice formed. The only comparable window before 2013 was a few days in February 2009, when a single small trail was suspected to have formed.

This is one of the rare Anomaly Daily cases where we can write the word "solved" and mean it. The driving mechanism is confirmed by direct observation, GPS tracking, and a weather station: ice shove. Not wind alone, not thick-ice flotation, not the more exotic candidates. In 2020, NASA even ruled out microbial mats and wind-generated water waves as contributing causes.

What's genuinely still open is smaller and stranger. Nobody can yet predict which rocks move in a given event and which don't. Fractures in the ice can decouple stones sitting centimeters apart — one catches a panel and travels tens of meters while its neighbor stays put. Water depth matters too: ice can float or slide right over a low-profile rock, leaving it stationary while a taller neighbor stays engaged and keeps moving into deeper water. Sharp and Carey's ambiguous 1970s corral result now reads as an early glimpse of exactly this. The 2014 team pointed out a stake sitting just upstream of the unmoving stone that may have shattered the ice sheet before it reached the rock — decoupling by accident, four decades before anyone understood the mechanism.

The other footnote is climate. A statistical study by Ralph Lorenz and Brian Jackson, reviewing published reports of rock movements, suggested at roughly 4:1 odds an apparent decline in movement between the 1960s–1990s and the 21st century — consistent with drier winters and warmer winter nights shrinking the flood-then-freeze window. The mechanism took a century to catch on camera. On a geological timescale that century is nothing; on a human one, if the ponds keep failing to freeze, the show may be quietly winding down.