The mountainside broke without warning. No crack, no sound, just a slope of snow above the road giving way at eleven in the morning on May 26th, 1953. John Sullivan had only a moment, and he used it to make a decision. He could not outrun what was coming, so instead of running out along the road, he threw himself into the cut the plow had made, the narrow trench in the snow pressed right against the inside of the road.

Then the avalanche came over all four of the men working that stretch. The road they were clearing was the going-to-the-sun road in Glacier National Park, a 51-mile ribbon of rock and asphalt hung on the face of a cliff. Every winter, as much as eighty feet of snow piles onto Logan Pass. Every spring since the road was finished in 1932, crews have had to dig it back out of the mountain before anyone can drive it.

And on that spring day in 1953, the snow took two more lives after already taking three during construction. The road had five enemies, and the crews met them in the same order the mountain presented them: a survey racing the first blizzard, a fight over whether the road should climb in fifteen hairpins or one, a cliff called the Garden Wall, a tunnel that men could only reach by climbing down a ladder, and last, the snow itself, which comes back every winter and has never once agreed to stay gone. The first of those five started with a man looking up. In the late summer of 1924, the road had stalled at the bottom of the hard part.

About twenty-nine miles of wagon grade track existed on both sides of the park, and both ends ran into the same wall of mountains and stopped. Between them stood the Continental Divide, and the only way across it that anyone had drawn was a steep climb up a creek valley. The mountain had already shown what it thought of the lower road on the west side. That summer, the contract had pushed the track past Avalanche Creek, about sixteen miles in from the park’s west entrance.

The most expensive piece of that whole contract was a stretch only about two hundred yards long at a place called Red Rock Point. It cost so much for one reason: avalanches came down there, and the road still had to climb another few thousand feet of wall through country where the avalanches were worse. Frank Kitridge came out from the Bureau of Public Roads that September, and the park superintendent, Charles Kraebel, drove him up to the site himself. Kitridge had laid out a railroad in Alaska before this.

He knew cold. What he didn’t have was time. The survey started late in September, and in the high country of northern Montana, late September means you are already borrowing days from winter. His crew started at thirty-two men.

Each morning, they climbed something like three thousand feet just to reach the place where the day’s measurements began. They walked narrow ledges carrying their instruments. Where there was no ledge, they hung over the cliffs on ropes and took the readings from the air. Men quit, so fast that over those weeks the park service counts 135 men employed on a crew built for thirty-two.

The turnover, the records say, ran past 300 percent. A crew where every single job had to be filled three or four times over by men who looked down once and went home. Then early in November, the mountain gave its answer. Four feet of snow fell on the east side of Logan Pass and enough on the west side to stop the work.

Kitridge pulled his crews off the mountain. He had twenty-one miles of preliminary line on paper, and the start of a final location line run from the pass downhill in both directions. Over that winter, he sat with his notes and drew up a report. That report was about to start a fight, because the line Kitridge liked best did not climb the creek valley at all.

It went out onto the face of the mountain. Here is the strange part, and it is the reason the road you can drive today hangs where it does. The cliff route was about three miles longer than the valley route. A longer road, more rock to blast, more wall to build, and it was still the better road.

Think about the side of a house that gets the afternoon sun. In spring, the snow on that side is gone weeks before the snow in the shade. The valley route climbed in shadow. The cliff route faced open sky.

Because of the way that wall faced the sun, the snow would melt off it faster and the road would open sooner every year. On a mountain where the season is the enemy, the longer road bought back time. It also needed only one switchback, while the valley plan drawn up by the park’s own engineer, George Goodwin, back in 1918 used fifteen of them stacked on top of each other up the slope. The one long climb also set the grade.

Kitridge’s line held the road to 6 percent the whole way up. That number sounds like a detail until you think about the cars of the 1920s. At 7 percent, a car of that era had to shift down into second gear. Six percent was the steepest a road could be and still let an ordinary family car climb it without fighting the gearbox.

The road had to rise thousands of feet up the side of a mountain without ever getting steeper than that, and the only way to do it was to run it long and straight along the face of the cliff. That difference had already caused one of the strangest scenes in the road’s history. In August of 1924, the director of the National Park Service, Stephen Mather, rode up to look at the climb with Goodwin and a young park service designer named Tom Vint. Vint hated the fifteen hairpins.

He wanted a single long climb along the wall. The men argued on the mountain until Mather looked at one, looked at the other, turned around, and walked off. By the time Vint and Goodwin got the horses together, they never caught up with him. Two days later, Mather made his decision.

He called in the Bureau of Public Roads, and that is how Kitridge got the mountain. When his report came back the following year, it said Vint was right. That walk-off changed more than one road. Mather was so taken with how the bureau handled this mountain that in 1925 he signed an agreement putting the bureau in charge of road building in every national park, made formal the next year.

A great many of the park roads Americans have driven since were built under an arrangement that started with an argument on a horse trail below Logan Pass. Kitridge also did something nobody there had seen before. He took photographs of the mountain and stitched them together into a single long panorama, then drew all three possible routes across it, so anyone could see exactly what each one would do to the face of that wall. The park service would use that method on roads for decades.

Goodwin left the park service in 1925. It is worth asking why anyone wanted this road badly enough to hang it on a cliff. When Congress made Glacier a national park on May 11th, 1910, there were only a few miles of rough wagon road inside it. The interior belonged to whoever could afford the Great Northern Railway’s trains and its hotels and chalets.

The park’s first superintendent, William Logan, wanted a road across the mountains, and so did a lot of ordinary people who wanted to see the place without paying railroad prices. Some local businessmen fought it at first until they worked out what tourist money a road would bring through their towns. By 1921, when Congress gave the first money specifically for the crossing, that argument was over. The mountain argument was just starting.

The line was chosen. It would run along the Garden Wall. The next problem was that nobody had ever built anything there. The Garden Wall is a long knife-edge ridge of rock on the west side of the Continental Divide.

On the roadside, it falls away in cliffs with avalanche chutes slicing down at every few hundred yards. In May of 1925, the government advertised a contract for the hardest 12. 4 miles from Logan Creek up to the pass. Of the road’s fifty-odd miles, about twelve would have to be cut straight out of the mountainside, and most of those twelve ran along this wall.

The plan was a road about twenty-two feet wide. Along parts of the Garden Wall, it would end up narrower than that because there was simply no more mountain to cut into without bringing the cliff above down onto it. Contractors had to see it to bid on it. So in early June, the park service and the bureau set up a base camp at Trapper Creek and took thirty-five contractors up onto the wall.

For four days, they crossed steep snow fields to look at the ground. For two of those days, according to the accounts of that trip, they were roped together in the rain. Imagine bidding on a job you can only inspect while tied to the man in front of you. The low bid came from two men in Tacoma, Washington: D.

A. Williams and A. R. Douglas.

$869,145. For that, they agreed to build more than 7,000 feet of retaining wall and guard wall, almost 11,000 feet of standalone guardrail, and the road’s first tunnel. The park service history calls it the most difficult road construction ever attempted in a national park at that time. Then, before the first hole was drilled, the government handed them a set of rules that made the job harder.

The first came straight from Tom Vint. He had written down exactly what he wanted the road to look like. Bridges, tunnels, retaining walls, and parapets, he said, should not be monumental or stylistic, but native, simple in line and retiring. Everything they built on that cliff was supposed to look as if the mountain had grown it.

The contract had rules nobody in road building was used to. Contractors had to use many small blasts instead of a few big ones because big blasts tear up a mountainside. The park’s engineers had even wanted power shovels banned entirely until somebody worked out what a road built only by hand labor would cost. And there was one rule that sounds small and isn’t: no side casting.

Side casting is what every road crew on Earth does with broken rock. You blast it and you shove it over the edge. Here, the contract said the rock had to go wherever the park’s officials pointed, on a ledge with a cliff above and a cliff below, where there is nowhere to point to. That rule meant every load of rubble became a puzzle.

To feed the work, the contractor built camps in the wilderness. Between twenty and sixty men in each, 250 to 300 on the west side in all, strung up the mountain from headquarters near Logan Creek. At the peak, fifty pack horses a day carried in the food, the fuel, the powder. Each camp was a few miles from the next, tents and rough wood frames on whatever flat ground the mountain allowed, and the cook in each one counted among the men.

Nobody could do without. The pay ran from about fifty cents an hour for a laborer to $1. 15. The best paid were the cat skinners, the men who drove the tractors and graders, and the powder monkeys, the men who handled the explosives.

On that wall, the two jobs most likely to kill you paid the most. Where the ledge was wide enough, the contractors brought in machines. The record lists three power shovels on the job, two running on steam and one on gasoline. Behind them ran little narrow-gauge railroads the men called dinkies: a small gasoline locomotive pulling about a dozen dump cars shuttling broken rock along the shelf.

A railroad on a ledge on the side of a cliff, built to move the rubble of the ledge itself. The partnership running all this was new, and it was not a calm one. The park service wanted the mountain disturbed as little as possible. The bureau’s engineers wanted a road built on time.

Williams and Douglas were caught between them, and over four seasons they brought in subcontractors thirty different times to keep the work moving. The powder was everywhere. Over the life of the job, the contractors would go through almost 500,000 pounds of explosives, 250 tons of blasting powder and dynamite, and almost all of it set off a few pounds at a time against one wall. Before one big shot about a mile east of the Loop, the laborers pulled wool socks on over their boots so nothing on their feet could strike a spark against the rock and set it off early.

Most of the drilling happened the hard way. Men stood on the ledge with pneumatic drills or hung beside the face on ropes to reach the rock above the roadway and set the holes one at a time. Some of the work happened where there was no ledge yet at all, only the line on Kitridge’s drawings and the drop beneath it. In 1926, a foreman named Charles Ruddberg was going down one of those ropes about a mile above the Loop.

Hanging on a rope beside the face was how the crews checked overhanging rock above the road to see whether it would hold or come down on the men below. He lost his grip. He fell sixty feet to the roadway. He was the first of the three men the road would take.

The park service record of his death is one line long, but the work did not stop for him. That fall, the crews ran straight into the next problem: a shoulder of mountain sticking out of the wall that the road could not go around. On a cliff road, a shoulder like that is a trap. You can’t swing the road out around it because out there is air.

You can’t cut the whole shoulder away because that means blasting down half a mountain, and the contract had just told them they couldn’t. The crews started on the one answer left. They went through it. Two subcontractors, H.

W. Bennett and Phil Scogalia, took the tunnel job. The bore they drove was 192 feet long through the solid shoulder of the mountain, and it pushed on into the cold months and was finished early in 1927. Three thousand seven hundred twenty-nine cubic yards of rock came out of that hole.

Imagine filling a schoolhouse to the rafters with broken stone and then doing it again and again and carrying all of it out through one opening on a cliff. Then the designers did something that tells you exactly how these people thought. They cut windows in it: big arched openings through the outer wall of the tunnel so that a driver passing through the dark would suddenly see framed in rock Heaven’s Peak across the McDonald Creek Valley. In the middle of the hardest contract in the park system, they took the time to aim a hole in a mountain at a view.

But the windows hid what was happening on the rest of the wall. The no side casting rule had a problem. On much of the Garden Wall, there was simply nowhere else for the rock to go. The broken stone went over the side anyway, load after load, and it slid down the slope through the trees and the meadows below.

And every place the slope got buried, the road needed something new to hold it up, so the walls multiplied. Williams and Douglas ended up building more retaining walls than the contract had called for, out of the same rock they were blasting off the mountain. They laid it up as rubble masonry, stone fitted to stone, with a low parapet along the edge where a car met the drop. Some of those walls stand on the face of the cliff with nothing under them but more cliff.

At Haystack Creek, where a stream comes down the wall, they built a stone arch for the water and carried the road over it. Farther up, at a place that would come to be called Triple Arches, the road could not be held up by a wall at all. So they built three stone arches out from the cliff and set the roadway on top of them, like a shelf bracketed onto the side of a house. The Williams and Douglas contract ran four seasons.

On October 20th, 1928, the western leg was finished up to Logan Pass. Kitridge came back and walked it, and what he wrote is not a celebration. He was dismayed at how much rock had been thrown over the side, burying the vegetation below when it could have filled the gullies along the road. Those extra walls had driven up the cost.

But the road was there. The west side was open. And the next spring, the park found out what it had actually built. In the spring of 1929, for the first time, someone had to open the new road to Logan Pass.

The park bought a single gasoline shovel with a bucket about three-quarters of a cubic yard across. A crew of six men ran it in ten-hour shifts for three weeks, digging the road back out of the drifts so cars could reach the pass by the fifteenth of June. Think about what that means. They had spent four seasons blasting a shelf into a cliff, and every winter that shelf filled back up with snow to the top.

The road they had just cut was going to have to be cut again out of snow instead of rock every year forever. On the far side of the pass, the road still didn’t exist. East of Logan Pass, the last ten miles had to come down the mountain toward St. Mary Lake, and that side was worse.

Work began in 1929, and the hardest contracts went out in 1931 to two companies: the Colonial Building Company of Spokane, Washington, and A. R. Guthrie of Portland, Oregon. Stephen Mather, the director who had walked away from the argument on the mountain, never saw it.

He died of a stroke in 1930. Colonial’s contract held the worst job on the whole road. Just east of the pass, the line ran into Pagan Mountain, and it had to go through. The east side tunnel would be about 408 feet long, more than twice the length of the one on the west side, and there was no way to get a machine to it.

Colonial built a rough track out from the summit along the mountain about two hundred feet above where the road would run. Everything came along that track behind a small caterpillar tractor, thirty horsepower. Then the track ended. Men carried everything down a switchback trail so steep that in the contractor’s own words, in three hundred feet of distance it lost about one hundred feet of elevation.

At the bottom of the trail, a ladder about one hundred feet long, nearly vertical, dropped down the face to the little bench where the tunnel crew worked. Picture carrying a fifty-pound box of dynamite down that. The completion report timed it. A well-conditioned man needed about half an hour to carry one box down and climb back up empty, ready for the next.

And some men never made the first trip. The report says that on several occasions, men hired for the tunnel walked to the top of the cliff, looked down the ladder, and turned in their resignation, saying they could not stand the elevation. The men who stayed drilled and blasted into Pagan Mountain, and the tunnel crept forward about five feet four inches every twenty-four hours. Six thousand seven hundred seventy-eight cubic yards of rock had to come out of it.

No power shovel could reach the place, so that rock left the tunnel by hand at five feet and a few inches a day. That bore is something like two and a half months of boring around the clock before you even count the days the weather shut it down. Guthrie’s crews lower down had their own version of the same problem. There was no road to their work.

So Guthrie put a power shovel on a barge, floated it up St. Mary Lake, then drove it off onto the mountainside. In 1931, the mountain started collecting again. That year, a laborer named Carl Rosenquist was struck by falling rock and killed.

The same season, one of the caterpillar tractors slipped off the edge of the new road. It rolled down the mountainside, and then, in what the engineering record treats almost as a joke, the machine climbed back up two hundred feet with only minor damage. Two accidents on the same mountain in the same year. One machine came back up.

One man didn’t. By the summer of 1932, it all came down to one season. The crews from the east were closing the last gap toward the pass, working against the calendar, because everyone on that mountain knew the snow would shut the job down by October whether they were done or not. A stonemason named Gus Swanson was on the wall that summer.

He was caught in a rock slide. He was the third: a foreman, a laborer, and a stonemason. The park service count of the road’s dead stops at three. One man who fell from a rope, one struck by a rock, one buried in a slide.

Gus Swanson’s name is the last on it. The crews kept going because the alternative was another winter with a gap at the top of the mountain. By October 8th, 1932, the work was far enough along for traffic to cross. That fall, for the first time, an automobile drove the entire fifty-one miles from the west entrance up the Garden Wall, over Logan Pass at 6,646 feet, down past Pagan Mountain, and out along St.

Mary Lake. It had taken eleven years from the first construction money. The first appropriation back in 1921 had been $100,000. Goodwin’s old estimate for the road had been around $600,000.

The bill came in at more than $2. 5 million and three lives. Every history of the road pairs those two numbers together, the dollars and the dead. Then the snow came, and the first car’s tracks were buried under the winter, and the road had to wait for its own party.

On July 15th, 1933, about four thousand people climbed to Logan Pass for the dedication. Think about where that crowd was standing. Ten years earlier, there hadn’t even been a survey line up here. The park superintendent, Ivan Skovran, ran the program.

He called it the most beautiful piece of mountain road in the world. He read a message from the new director of the park service, Horace Albright, who wrote that the rest of Glacier should stay reachable only by trail and that this road should stand supreme and alone. The Civilian Conservation Corps put together a chorus. The Blackfeet tribal band played the national anthem, and the afternoon ended with a ceremony of peace among the Blackfeet, the Flathead, and the Kootenai tribes up on the divide the road had just crossed.

That day, the road got the name everyone uses now: Going-to-the-Sun, borrowed from the mountain that rises above St. Mary Lake. Frank Kitridge, the man who had mapped the line in a race against the November snow, went on to run engineering for the whole Bureau of Public Roads and later for the park service itself. In the early 1940s, he became the superintendent of Grand Canyon National Park.

The panorama trick he invented for this mountain, the stitched photographs with the routes drawn across them, became a standard way of judging what a road would do to the country around it for decades. The fight over the road, though, had moved to the season. The crew that made the party possible had been up there weeks earlier, and their job was not in any of the speeches. To open the road for that summer, the plow crew ran the gasoline shovel in double shifts.

They cleared about fifteen thousand cubic yards of snow and rockslide debris, and in places they dug through drifts thirty feet high. Fifteen thousand cubic yards is enough to bury a football field, end zones and all, under about seven feet of snow. They dug that out of a cliff road with one shovel so four thousand people could stand at the top and hear speeches. In a way, the road still wasn’t finished.

Everyone involved knew it. The older sections built before the bureau arrived were narrower and rougher than the new ones. Through the rest of the 1930s, crews went back to widen the road, to add more stone walls and bridges, especially along the east side at a place called the Golden Stairs. One retaining wall there alone took 788 cubic yards of masonry.

By the end of 1937, the whole road had been built or rebuilt to at least twenty-two feet of crushed rock. The asphalt took even longer. Paving started in 1938, then the war stopped it. The last of the road wasn’t paved until the end of 1952.

And while all of that was going on, the mountain kept doing something to the road that nobody could fix with more stone. Along the high part of the road, the engineers counted more than sixty avalanche chutes: sixty places where every winter snow came down the wall with enough force to batter the stone guard walls along the edge. Avalanches and falling rock damaged the masonry so often that repairing it became a job that came back every single year in the same places. Stone guard walls line about 43,000 feet of the high road, more than eight miles of them.

Rebuilding pieces of that each year by hand from rubble on a cliff was a fight they were losing. So in 1948, the engineers of the Public Roads Administration, the same agency Kitridge had worked for, suggested something that no one in the park system had seriously tried: a guardrail you could take away. The idea was simple once you say it out loud. If the avalanches are going to sweep the edge of the road every winter, don’t leave anything on the edge for them to sweep.

Put the rail up in summer, pull it off before the snow, and let the slides pour across a bare road. The park service didn’t love it. They had spent twenty years making those stone walls look like part of the mountain, and they would not replace them wholesale. They agreed to try it only where the slides kept coming back.

A later contract put in about 1,600 feet of removable rail in the avalanche paths. In 1968, a newer version arrived: steel posts and square timber rails that pinned into a concrete curb. The park’s own history describes that removable design as the system still on the road today. The people who built this road had decided that in some places the only way to keep a guardrail on a cliff was to take it down every winter and put it back every spring.

The snow had won that argument, and the road was rebuilt around losing it. There was something else the snow could take that couldn’t be pulled off and stored for the winter: the people clearing it. By the early 1950s, opening the road each spring had become its own season of work. The single gasoline shovel of 1929 was gone.

In its place came bulldozers and rotary plows, machines that chewed into a drift and threw the snow out over the edge in a long white arc. The crews worked up from both sides of the divide toward Logan Pass, the west side crew climbing the Garden Wall, the east side crew coming up from St. Mary. The worst of it waited just east of the pass: a snow field on the flank of a ridge running south from Pollock Mountain where the snow lies deeper than anywhere else on the road.

The reason is the wind. All winter it pours over the ridge and across the divide, and on the sheltered side where it slows down, it drops what it is carrying. It is the same thing that happens behind a snow fence along a highway in the plains where the drift builds up on the quiet side. Here, the fence is a mountain ridge, and the drift ends up deeper than a house is tall, sitting right on top of the road.

The crews had a name for it: the Big Drift. It is a little more than a mile long, and it can take a whole month to dig through that one mile. But in the spring of 1953, the danger did not come from the Big Drift. It came from below the pass on the west side, on a slope the crew was watching as closely as they knew how.

Gene Sullivan was fifty-nine years old that spring. He was part of a road crew working up the west side toward the pass, about five miles below Logan Pass and less than a mile above the crew’s road camp. With him were his foreman, George Beaton, forty-five years old; an equipment operator named William Whitford, also forty-five; and a younger man named Frederick Klene, thirty-one. They were working a stretch where the drift stood around ninety feet high, according to the newspaper accounts of that spring.

Men were watching the slope for trouble. Everyone on that crew knew what a cracking snowpack looked like, and what it sounded like when a weak layer gives way underneath, that low whump that warns you to run. On May 26th, 1953, at about eleven in the morning, none of that happened. The park superintendent, Jack Emmert, told reporters afterward that the mountainside broke suddenly without warning.

No crack, no sound. The slope above the road simply let go. When the avalanche stopped, the road had disappeared under the slide. Klene was found with his left leg badly broken and beaten.

Whitford and Sullivan were simply gone somewhere under tons of snow on a cliff, and the men who came running had no idea where to dig. Here is where the clock starts. A man buried in avalanche snow usually doesn’t have long. The snow sets up around him almost like concrete, and every breath he lets out freezes into the snow in front of his face.

Every minute the crew spent digging in the wrong place was a minute that the men underneath didn’t have. One of the crew, a man named Diamond Apgar, had a theory. If anyone had lived through that, he reasoned, it would be someone who had ended up along the inner edge of the plow cut, right against the uphill bank, where the slide might have bridged over and left a pocket of air. So that’s where they dug.

Noon came. Then one. Two. Three.

Under the snow, Gene Sullivan was alive, and he was working. He had managed to scrape out a small opening near his face, enough to keep breathing. He couldn’t move his body. He just kept breathing into that little space and waited.

Four. Five. Six. In the high country in late May, there is daylight left in the evening, but it starts to go cold fast.

At about seven in the evening, eight hours after the slope came down, the crew broke through to him, right where Apgar had said, along the inner edge of the plow cut. He was alive. They put oxygen on him. They wrapped hot coffee bottles and towels and packed them against his body to bring his temperature back up.

Later, he explained what he’d done in a single sentence. He said he knew the snow would bury him, and he didn’t want to be carried down the mountain. So that is how you keep a road on a cliff that winter buries every year. You dig it out by hand and machine every spring, one drift at a time, and you accept that the mountain will sometimes come down on the men doing it.

On that day, it took two of them. William Whitford was dead. George Beaton, the foreman, had been carried away. They brought in a bloodhound to find him.

His body was found nearly half a mile below the road. The road didn’t open that year until June 24th, the latest it had opened since the dedication twenty summers earlier. And the crews were back on the same wall the next spring, because there was no other way to open it. And after that, the next.

The slides didn’t stop either. On May 16th, 1975, near the same road camp, a man named Dennis Holden was walking on the road when an avalanche hit him. It buried him four times as it carried him and swept him three hundred feet down the mountain. Road crew members lowered a rope down the slide path and hauled him back up to the road.

He lived. The machines got bigger. The danger stayed the same shape. In the spring of 2011, the crews had to clear one section of the road seven separate times because avalanches kept filling it back in behind them.

That year, the road didn’t open across the pass until the thirteenth of July. Today, the crews start in early spring. In 2018, they started on the first of April in a year of near-record snow. The west side crew was led by a man named Brian Paul running a rotary plow that weighs 20,000 pounds.

Working beside them was a snow scientist from the United States Geological Survey, whose only job was to watch the slopes above them for avalanches, a job the crew of 1953 had been doing for themselves with their own eyes on the morning the slope let go. That year, the Big Drift was about one hundred feet deep: a ten-story building made of snow sitting on the road, more than a mile long. The crews from the east and the west dug toward each other across it. On June 6th, after more than two months of plowing, they met.

Brian Paul described the whole job in three words. “It takes patience,” he said. His crew that year was about twelve people on the west side, including two avalanche technicians from the Geological Survey and a heavy equipment mechanic, because a broken machine on that wall can’t be towed anywhere. Every year, Paul said, “It changes completely.

” Asked when the road would open, he gave the answer the mountain has been giving since 1929: “It’ll be ready for the public when it’s ready for the public. ”

The equipment now can move something like four thousand tons of snow an hour. Even so, clearing the whole road takes about ten weeks every year. The method would be familiar to the crew of 1953.

Where the snow hangs off the ridges in cornices, the crews bring them down first with explosives, on purpose, before the mountain chooses its own time. Then excavators and loaders cut the snowpack down to about four feet above the pavement, and the rotary plows finish it, throwing the last of the snow over the edge. Some of it goes into dump trucks and gets hauled away, because in places there is no edge to throw it over. The danger doesn’t end when the plows are done.

In 2019, thirteen cyclists riding the closed road in spring were stranded for eight hours by avalanches. On May 13th, 2021, three bikers rode into an active slide, and two of them were trapped for several hours. The date the road opens has been sliding later. Since 1952, the road has opened across the pass in May only about ten times.

The average opening date across the history of the road is around June 10th or 11th. Since 2011, it has averaged closer to June 21st. The road is only open across the top for a few months a year, from early summer to about the middle of October. Then the snow takes it back.

And then it opens again. And for a few months, you can drive the thing these people built and keep rebuilding. Drive it west to east, the way Kitridge’s crews first fought their way up. You start at West Glacier at the park’s west entrance and run along the shore of Lake McDonald, the longest lake in the park, with the mountains standing up at the far end of it.

The road here is flat and easy, and it lulls you. Watch the trees along this stretch near Lake McDonald Lodge, and again near Avalanche Creek. In places, you can still see old cuts running off through the forest: earlier lines of the road that were abandoned when the crews of the 1930s straightened it past Avalanche Creek. The valley narrows, and the road starts to climb along the side of McDonald Creek.

This is where the Garden Wall begins to rise over your left shoulder. Watch the road edge. The low stone parapets start here, rubble masonry, stone fitted to stone, the walls Williams and Douglas built out of the rock they blasted off the mountain. Most of the names you pass on this drive were construction camps first.

Avalanche Creek, Logan Creek, the Loop, Lunch Creek, Bearing Creek, Rising Sun. Each one was a cluster of tents and a cook shack where men slept a few miles from the face they were cutting. Up Logan Creek, upstream from the road, a small cabin the engineers used is still standing. Then the west side tunnel through the shoulder of the mountain that the road could not go around.

Look out the arched windows on your right. That peak in the frame is Heaven’s Peak, exactly where the designers aimed it in 1926. Then the Loop, the single hairpin turn on the west side, the one switchback Tom Vint fought for in place of George Goodwin’s fifteen. You swing around it and start up the long six percent grade along the face of the Garden Wall.

Somewhere above the Loop, about a mile up, is the stretch where Charles Ruddberg lost his grip on a rope in 1926. There’s no marker there. The road goes on past it like any other piece of the wall. You cross Haystack Creek on its stone arch.

Farther up, you drive over Triple Arches, the roadway sitting on three stone arches bolted onto the cliff with nothing under them but air in the valley. In avalanche country along here, the rail at the edge of the road may be one of the removable kind, the rail that comes off before the snow every year. Then the road tops out at Logan Pass, 6,646 feet on the Continental Divide, the ridge the Blackfeet call the backbone of the world. This is where about four thousand people stood in July of 1933.

From the visitor center, look east toward Pagan Mountain. About two hundred feet above the road, you can still see the line of the old tote track Colonial cut to reach the east side tunnel, the track that led to the top of the ladder. Just east of the pass, you drive across the place where the Big Drift lies each winter. In summer, it’s an ordinary stretch of road with a few patches of old snow beside it.

In the spring of 2018, it was one hundred feet deep. Then the east side tunnel through Pagan Mountain. Every bit of rock in it came out by hand, about five feet a day, by men who reached their work by climbing down a ladder with a box of dynamite. Below the tunnel, the road drops through the Golden Stairs, past that big masonry wall, and down past Siyeh Bend and the Bearing Creek Bridge, the plainest and maybe the most beautiful stone bridge on the road.

Then it comes down to St. Mary Lake, the same lake A. R. Guthrie floated his power shovel up on a barge.

You run along the shore past Going-to-the-Sun Point and out to the east entrance at St. Mary. Fifty-one miles. If you’re driving anything longer than twenty-one feet or wider than eight, the park won’t let you take the alpine section at all.

The ledge is still about as narrow as it was when the first car crossed it. The government recognized what it was in stages. The road went on the National Register of Historic Places in 1983. In 1990, a team from the Historic American Engineering Record spent a summer measuring and drawing it.

In 1997, it was made a national historic landmark, and thirty-three of the park’s old red touring buses, the open-topped ones that have carried visitors over it for generations, were rebuilt to keep running on it. Remember the photograph from the start? The ledge of fresh rock on the cliff, the valley far below, the men so small you have to look twice to find them. Every winter, that ledge disappears.

Eighty feet of snow on the pass, one hundred on the Big Drift, the guardrails taken down, the road gone white from end to end. And every spring, a small crew goes up from each side, looks at the slope above them, and starts to dig.