On the 24th of March, 1903, the Southern Pacific sent a locomotive out onto a brand-new railroad embankment in the Great Salt Lake, and the ground under it went down into the water. Partway across, the embankment gave way. The track dropped, and the engine came to rest still sitting on its rails in about two feet of salt water. Crews had to run a cable out to it and drag it back.

Nine days later, it happened again, worse. A train heading out toward the gap in the line reached a stretch of trestle near the open water, about five hundred feet of it, and the trestle sank under the load, throwing the train into the lake. Two failures in nine days, in the same part of the lake. The first sank an embankment, the second sank a trestle.

The two methods the engineers had to choose from had both gone down in the same soft ground, and the thing doing it was still invisible under a few feet of brine. That was the moment when the crossing everyone called an impossible dream looked like a disaster they could not fix. The line had been a long time coming. Two men carried it.

The first was William Hood, chief engineer of the Southern Pacific, a New Hampshire man who had started his career in 1867 as an axman with a survey crew in the Sierra Nevada and who by 1902 had his name on the plan. The second was Edward Harriman, the New York financier who had just taken control of the railroad, who did not like the idea at first, and who would end up driving a spike of solid gold into it out in the middle of the lake. The reason for the crossing sat a few dozen miles north, at Promontory Summit, where the first transcontinental railroad had been joined in 1869. To get from Ogden on the east side of the lake to Lucin on the west side, every train swung north around the top of the water and hauled itself over the Promontory mountains to a summit seven hundred feet above the lake.

On the worst grades, a train climbed ninety feet for every mile, around curves that bent ten degrees, with helper engines shoving the heavy freights over the top. A ruler laid across the map from Ogden to Lucin runs straight over the north arm of the lake. The engineers of 1869 had looked at that line and thought about a trestle. They turned it down because the lake was high in those years and nobody could drive piles long enough to reach a bottom they trusted.

Then the lake started to shrink. Lynn Farrar, a historian who spent his career with the Southern Pacific, put the high water at 4,215 feet in 1873. By 1900, the surface had fallen nearly fifteen feet. Fifteen feet less water meant shorter piles, lower trestles, and miles of lake bed sitting only a few feet under the surface.

That drop put the lake crossing back on the table—and left the trap. Along the east shore, the retreating water uncovered a flat of salt crust that looked solid enough to walk on. Under it was eight to ten feet of mud. Work began at both ends of the cutoff in March of 1902.

The land section west out of Ogden went fast. By June, the grade ran across the farm country to the east shore of the lake, carried over forty-four little drainage openings on stone culverts and steel pipe, with two steel bridges still to come. Then the grade reached the edge of the salt flat, and the ground stopped cooperating. From that shore to Promontory Point was a little over eight miles.

Close to three of those miles crossed the lake bed the water had just abandoned: eight to ten feet of soft mud under a white crust of salt. You could not set a rail on it. You could barely set a man on it. So the crews built a road to stand on before they built the railroad.

First they laid planks flat on the crust. On the planks went sandbags, about one hundred pounds each, stacked three layers deep. On the sandbags went cross timbers a foot square and ten feet long, then heavy stringers thirty feet long, then ties, then a temporary track. Think of a snowshoe.

A man in boots punches straight through deep snow, but strap a wide frame under each foot and the same weight spreads out and floats on top. The sandbags were a snowshoe for a locomotive, and it worked. Work trains rolled out over the mud loaded with shale and gravel and dumped their loads off the north side of the temporary track, and the dumped material slowly built the permanent grade beside it. For about three miles through soft crust and then through water a foot or two deep, the snowshoe held.

Then the line reached the old mouth of the Bear River, and a snowshoe was not going to be enough. The summer before, on that very stretch, a pile driver had shown what the mud could do. Its crew stood a timber twenty-six feet long under the hammer, and the hammer came down once. The pile went out of sight.

It did not stop partway down. It slid below the surface of the mud with nothing left on top for the hammer to hit. They stood a second pile, twenty-eight feet long, right on top of the first and brought the hammer down again. One blow, and both piles were out of sight.

The report, printed in Engineering News in November of 1902 by an engineer named C. W. Arthur, recorded what the crew found when they probed the spot: mud fifty feet deep, gray silt that held its shape until something heavy sat on it. Here is the part that sounds backward.

Out in the middle of the lake, where the water ran thirty feet deep and more, the bottom held. The deep water was the easy part. The trouble lived at the shallow edges, in water you could wade, where the lake looked tame and the bottom was anything but. Arthur saw something stranger still.

A pile could be driven thirty, forty, fifty feet into that bottom. The hammer could pause for a moment, and the pile would start to rise a few inches, sometimes two or three feet. The lake bed was pushing the timber back out like a hand pushing a cork up out of a bottle. Push a stick hard into wet sand at the waterline and hold it.

Let go, and the sand around it squeezes back toward the hole, and the stick creeps up. Make that stick fifty feet long, make the sand a gray slurry of silt that a river has been dropping for thousands of years, and you have the bottom of Bear River Bay in the summer of 1902. A sane railroad would have looked at that mud and gone around it. The Southern Pacific had been going around this lake for more than thirty years, and that was exactly the problem.

The engineers tried something that sounds desperate and was entirely deliberate. After those first two piles vanished, the crew changed the recipe. They drove two forty-foot piles, tied them together, capped them, and braced them, and called that one leg. Then they built the next leg fifteen feet farther out, and the next.

Arthur wrote that the engineers hoped the doubled legs would be strong enough to carry construction trains while those trains dumped their loads, and that in time the whole thing would turn solid. That sentence is the trick behind the whole crossing. The trestle they were building across the shallow water was never meant to stay a trestle. It was a scaffold for a dump train.

The cars rolled out over the open water on top of the legs, their sides tipped, and gravel and rock poured down between the timbers into the lake. Load after load piled up around the piles until the lake bed rose to meet the track, and the wooden legs disappeared inside a new embankment. It works the way a gardener builds up a raised bed against a row of stakes. You drive the stakes first so you have something to pile against.

When the bed is full, the stakes are still in there, buried, holding the shape from the inside. On the east side of the lake, that scaffold went up at four places at once, spaced about a mile apart, with eight floating pile drivers working the water. Each leg of the temporary trestle stood on four forty-foot piles with a heavy cap across the top. Stringers thirty feet long ran from one bent to the next, and the ties and rails went straight onto them.

The plan was good, and it had a weakness nobody could see from the deck of a pile driver. The scaffold could only be as strong as the mud it stood on. Out in Bear River Bay, under the shallow water, there was a soft spot no survey had measured—a long trench of silt far deeper than the mud around it. The crews were building right over it.

They would not understand what it was until the following spring. By the end of the summer of 1902, the east side was creeping toward Promontory Point, and the men doing it had run into the shortage that no amount of gravel could fix. There was nothing out there to drink. The Great Salt Lake is many times saltier than the ocean, and a man who drinks it gets sicker the more he drinks.

The shores around the north arm were salt flat and sagebrush desert, with almost no fresh water and no towns worth the name. And the crossing needed thousands of men living out on the work for months, plus a fleet of steam locomotives and steam pile drivers whose boilers could not run on brine either. Salt water in a boiler leaves a crust that chokes the tubes. At first, the railroad hauled water from Ogden in tank cars.

That got more expensive with every mile the work moved out from shore, so the company drilled. It put down an artesian well right in the lake, about four miles out from the east shore, and struck a good flow of fresh water. Crews started laying pipe from that well to the work camps. They drilled a second well about two miles inland, and that one looked even better at first.

It went down 340 feet and came in with clean, fresh water. Then it turned. By that November, about one-eighth of what came up the pipe was sand. The well was useless, and they abandoned it.

The lake was going to be crossed with water brought from somewhere else. By the time the work hit full speed, the daily haul was roughly 1,680 tons of fresh water carried in by train from points as far as 130 miles away. Every single day, a trainload of drinking water and boiler water rolled out toward men who were standing in an inland sea. Out on the open lake, the water came by boat.

The company was building its own navy. There was a stern-wheel steamer named the Promontory, 127 feet long and 22 feet wide, built to draw only 18 inches of water so it could work the shallow bays. It carried two 500-horsepower engines and two fresh water tanks of 15,000 gallons each, plus a cabin on the upper deck for officers and visiting railroad men. Around it worked seven tugboats and a handful of smaller launches.

They carried crews, food, messages, and water out to the temporary stations on the lake, and they herded rafts of logs to the pile drivers. The salt went after the fleet too. The boats were docked over and over and washed down with steam just to strip off the crust that built up on their machinery. Every bolt, every cable, every iron fitting out there was slowly being eaten.

So were the men. The original plan had been for the crews working in the water to wear rubber boots. The brine ruined the boots so fast that the idea was dropped. Arthur wrote that the men went into the lake at all hours in their ordinary clothes.

And he added a line worth reading more than once. “Those with scratches or sores,” he said, “had found that building a railroad across the lake was not such easy work. ” A cut on your hand from a splinter of timber, a scrape from a cable, a blister from a shovel handle. On dry land, you wrap it and keep working.

On the lake, you went back into brine strong enough to cure meat all day with that cut under the water. The lake bed at least was getting easier to read. Past the soft spots near the Bear River, the bottom across most of the bay was firmer, and the east side of the crossing kept growing west toward Promontory Point. Behind it, a contractor’s crew was blasting a deep cut more than four miles long straight through the point itself.

Beyond that cut lay the real lake, twenty miles of open water between Promontory Point and the far shore at a place called Strong’s Knob, and halfway across those twenty miles, the water was more than thirty feet deep. Here, the plan changed shape. On the east side, the shallow water had let the crews build a scaffold and bury it. Out in the deep water west of Promontory Point, burying anything would take more rock than anyone could haul.

So, for about eleven miles, the railroad would stand on wooden legs for good. This was the permanent trestle, the piece of the crossing that ended up on postcards. Scientific American printed the figures in May of 1904 in an article by a writer named Herbert Bennett. Across the deepest part of the lake, the water ran from thirty to thirty-four feet.

The engineers drove piles seventy feet long. Every fifteen feet, a row of five of them went in side by side square across the line of the track. Heavy timbers were bolted across each row to tie the piles together, and on top a cap, a beam eighteen feet long and a foot square. From each row to the next ran eleven stringers laid lengthwise with the track.

Then came the part that surprises. This was not an open trestle where you look down between the ties and see water. On top of the stringers the crews laid a solid floor of planks three inches thick, coated it with asphalt, and covered it with gravel. Bennett measured fifteen inches of ballast under the ties.

The deck was sixteen feet wide carrying eighty-pound rail. The Utah Humanities history of the cutoff counts about two million board feet of redwood in that decking and quotes a historian’s estimate that the wood above the waterline alone could have built a four-foot boardwalk from Boston to Buffalo. A train rolling onto it from the embankment would feel no change at all. A traveler crossing at night would never know when his car left solid ground for the bridge.

It is a gravel railroad carried on a wooden table eleven miles long standing in thirty feet of salt water. The timber total for the whole crossing, temporary and permanent trestle together, came to 38,256 piles. Laid end to end, they would have stretched about 535 miles. Every one of them had to come by rail from distant forests.

The permanent piles were fir, and they were so long that a single pile needed three flat cars coupled and chained together to carry it. That was the fourth enemy, and it sat a long way from the water on the railroad line feeding the job. The engineers had planned to build a mile and a half of trestle every week. The Railroad Gazette reported in December of 1903 that the best week they ever managed was a little over one mile.

The hold-up was getting material in by rail. A pile driver on the lake could only drive the piles that had already arrived. Out on the water, the work had a rhythm. Tugboats towed booms of floating logs out to the end of the trestle.

A floating pile driver lifted each log upright, set it in the line, and the steam hammer drove it down through thirty feet of brine into the mud below. Behind the pile drivers came crews setting the cross timbers and caps, then stringers, then planking, then gravel, then track, so that a work train could follow the very end of the bridge out over open water and bring up the next load. The lake did not always sit still for it. A storm across that flat water could break up a log boom and scatter timber for miles, and the boats spent part of their time rounding up logs blown loose by the wind.

Each runaway log was a pile that would not get driven that day. Every one of those dump trains was being paid for by a man who had not wanted the crossing in the first place. The idea was older than him. As early as 1892, Collis Huntington, then running the Southern Pacific, told Hood to study a line across the lake.

A financial panic hit the country the following year, and the railroad came close to going under. The plan went into a drawer. Huntington died in 1900. Harriman bought into the Southern Pacific and took over, and he wanted his new railroads rebuilt to a higher standard.

Between Reno and Ogden, the company started straightening and flattening its main line piece by piece. The lake crossing was the biggest piece. Harriman opposed it at first. He had the engineers look at other routes, including one around the south end of the lake.

The surveys brought him back to the straight line from Lucin to Ogden. What makes the decision interesting is the money. With ten trains a day each way, the cutoff would save about $214,000 a year in operating costs. The same money sitting in a bank at four percent would have earned more than that.

On paper, the shortcut was a worse investment than doing nothing. Harriman built it anyway, for speed. Fewer engines on every train, no climb over Promontory, and a straight flat run that would cut about two hours off the trip between Ogden and the Pacific Coast. And it meant Hood, who had been carrying this line around in his head for ten years, finally got to build it.

He was fifty-six years old when work started. He had joined the Central Pacific as a young man in 1867, swinging an axe for a survey party in the Sierra, and he had spent thirty-five years putting railroads through country that did not want them, including the mountain line over the Tehachapi Pass in California. He knew rock, and he knew grades, and he was about to spend two years learning about brine. By the end of 1902, the work was going around the clock.

There was a complete electric light plant on the job. At night, the gravel pits and the work fronts were lit, and loading crews filled hundreds of cars of gravel in the dark for the next day’s dumping. About 800 men worked the night and winter shifts in sub-zero cold under portable electric lights. The camps strung out along the trestle like stations on a line, and they were numbered.

One of them, Camp 23, sat roughly halfway across. It would outlive the job by four decades. While the east side fought the mud, a second army was coming at the lake from the west. Track gangs had graded and laid rail east from Lucin across the desert toward Strong’s Knob, with the track layers right behind the graders.

On the west shore, at a place that became known as Lakeside, the railroad opened a quarry, and its rock would feed the western fill for years. From there, a second embankment pushed east into the water, about five miles of it, toward the far end of the permanent trestle. By the start of 1903, the crossing was a race between two ends heading for each other across the lake, with a stretch of open water in the middle that shrank week by week. The finished sections on the east side were already carrying construction trains.

To anyone watching from the shore, the line looked like it was winning. Then the track started to sag. It was small at first. A letter sent to the Railroad Gazette from Salt Lake City at the end of March reported that several hundred feet of track that had already been built had sagged between six and ten inches in a short time before.

Six inches is small, and it is also the ground telling you something. The soft spot was on the east side, out in Bear River Bay, on the stretch the crews had been filling for months. The day the locomotive went in, the newspapers across the lake had a field day. James Goodwin, writing from Salt Lake City on the 28th of March, four days after the engine went into the water, reported that there was a hole in the Great Salt Lake puzzling geologists and engineers and giving the Southern Pacific more trouble than it had ever bargained for.

Piles, rails, and thousands of tons of earth and rock had been dropped into it, and nothing seemed to fill it. He used the phrase everybody in Salt Lake City was using: a bottomless pit. The theories in town went in every direction. Some people believed the lake really did have a hole with no bottom under it.

Others had a more scientific-sounding idea: that the lake had a subterranean outlet, a hidden drain somewhere under the bed, the way the Humboldt River in Nevada runs out into the desert and sinks into the ground. If the lake was draining through a crack in the earth, the railroad was pouring rock into a funnel. Two men in Utah who actually knew the lake bed answered that. The first was the state engineer, a man named Doremus, who had made a detailed examination of the bottom years before.

His opinion was that it was not bottomless at all, just a soft place that enough material would fill. The second was a geologist, Professor James Talmage, who had headed the State University in Salt Lake City. He rejected the bottomless quagmire too, and went one step further. The site, he said, had probably once been the bed of the Bear River.

Over time, that old channel had filled with loose river silt, and the silt was not solid enough to carry the weight of a trestle and a train. Engineering News, reporting from New York, came to the same conclusion in its report on the April 2 sinking. A former river channel, filled with mud to a great depth, probably ran right under the spot. The bottomless pit had a bottom.

It was the floor of a river that no longer flowed there, buried under who knows how many feet of silt it had carried down from the mountains and dropped. When the lake had stood fifteen feet higher, all of that was deep underwater. Now the railroad was standing on top of it with a locomotive. Goodwin closed his letter with an opinion of his own.

The soft places could be bridged instead of filled, he wrote, and spanning would have been cheaper in the first place. The company had to decide whether to change methods partway through the work or keep pouring. Either way, he said, it would take time and a great deal of money. On the 24th of April, 1903, exactly one month after the engine went into the lake, the Railroad Gazette printed a statement from William Hood.

It is one of the strangest documents in the whole story. The yarns and fake stories coming out of Salt Lake City about difficulties and disasters on the cutoff, Hood wrote, were without foundation. He had never heard of any scientist examining the so-called bottomless pit for Mr. Harriman.

He did not even know that any bottomless holes existed along the line. As a matter of fact, the railroad had experienced less trouble from the settling and sinking of material dumped into the lake than was usual on work of that kind. Everybody knew that material dumped into water always settled some. The only thing to do was to keep adding to it until it stopped.

Only a small percentage of what they had dumped had sunk, and the Salt Lake fill had really been very easy to handle. Then he explained where he thought the stories came from. Salt Lake City, he said, held enemies of the company. They had a grievance because the railroad had declined to build sixty-five miles of track around the south shore of the lake and into their city.

Hood said he had given strict orders that nobody was allowed on the work or on work trains without a special permit, and that not even a superintendent could grant one. He said he had done it to avoid accidents, not to hide anything, because there was nothing to hide. Read that statement next to the reports from the weeks before it, and one thing stands out. Hood never said the track had not sunk.

He said there was no bottomless hole. He said the stories were exaggerated. He said material always settles and you keep adding until it stops. Every one of those sentences can be true on the same day a locomotive is sitting in two feet of brine.

What he left out is the engine. He left out the second of April. A statement written to answer the newspapers never mentions the two events the newspapers were writing about. And the same statement made promises of its own that the record lets us check.

Hood said the trestle would be finished in about two months, a little more. It took about six. He said the company wanted trains running over the new line long before the beginning of the next year. Regular freight did not start crossing until March of 1904.

He said the cutoff would save about forty-eight miles. When Bennett measured the finished line, the saving came to just under forty-four. Buried in the same statement is the clearest explanation of the whole method. Eventually, he said, the entire crossing would be solid ground because the company intended to keep depositing material until it was.

But if they waited to do all that filling first, it would take too long. They wanted trains running as soon as possible. That was the reason for the scaffold. Build a bridge fast enough to carry trains now, and turn it into an embankment at the company’s leisure, with the trains themselves hauling in the rock.

On the one point that mattered most, the chief engineer was right, and the bottomless pit was a story. But on the smaller point, whether the lake was giving his railroad trouble, his own crews were telling a different story with every carload of rock. Through the spring and summer of 1903, gravel trains and rock trains ran out to the soft spots and dumped, and the track sagged, and the crews raised it and dumped again. The rock came from pits near the line and from the quarries at Promontory Point and Lakeside.

The method was exactly the one Hood had described in his statement, minus the word easy. You keep adding until it stops. Mud that is soft does not hold a load, but it can be pushed. Every carload of rock that went down into the old riverbed shoved some of the silt aside and some of it down, and the rock settled lower and lower until it found something firm enough to stand on, or until there was so much rock that the pile became its own foundation.

Fill a bucket with thick mud and start dropping in gravel a handful at a time. At first, the gravel disappears. It sinks, and the mud closes over it. Keep going, and the mud starts to bulge up around the edges of the bucket as the gravel pushes it out of the way.

Keep going longer, and eventually your hand lands on a pile of gravel that holds. Nobody can tell you in advance how many handfuls it will take. That bulge was real. A United States Geological Survey guidebook published in 1915 describes ridges of mud pushed up on both sides of the fill, with remnants still visible just beyond a section house called Bagley on the Bear River Bay crossing.

The lakebed squeezed out from under the rock and rose up beside it. In the middle of July, Engineering News carried a short item. The two most troublesome sinkholes, it said, had now been filled with rock and earth. Seven steam shovels and 3,000 men were at work filling in the temporary trestles.

Trains would be running across within six months. 3,000 men, seven steam shovels digging gravel day and night, trainloads rolling out across the lake and dumping it into water a few feet deep every day for months. And in the company’s published account, all of that came down to a few places where the bank settled. The camps moved with the work.

Supplies, mail, and water came out by train or by boat. A man finished a shift at the end of a wooden bridge miles out on a salt lake with the brine drying white on his clothes and slept within sight of the next morning’s piles. Men died building the Lucin Cutoff. The company did not publish how many.

A student history written at Weber State University counts 44 workers dead from train wrecks, explosions, and drownings, but the figure cannot be traced to a primary record. The Deseret News, looking back in 1993, found no official count at all and mentioned that at least one foreman died in a knife fight. Put those side by side and you get the honest answer: men died, and nobody since has pinned the number down. Meanwhile, the permanent trestle kept walking east across the deep water from the west side and west from Promontory Point toward a gap in the middle.

Out there, the deep water legs held, and that surprised the people who had predicted where the trouble would come from. Go back to Arthur’s report from the fall of 1902. Describing the twenty miles west of Promontory Point, he passed along a warning from what he called local experts. The salt crust out there was thinner, they said, but underneath it they expected a very hard layer that would give the pile drivers a great deal of trouble.

Hard bottom, in other words—piles that would refuse to go down. The experts had picked the wrong side of the lake. The western piles went down through the brine and held. The trouble came on the east, in the shallow water near the Bear River, where the bottom was soft enough to swallow piles.

The deep stretch was a different kind of hard. October of 1903. The two ends of the bridge are close enough to see each other across the water. The pile drivers were working toward each other now from both sides of the gap, each one setting its seventy-foot logs into the deepest water on the crossing, five to a row, a row every fifteen feet.

Every new row let the crews behind it lay a little more cap, a little more stringer, a little more plank, and gravel and rail so the work train could creep forward and bring up the next load of timber. Two miles of water at the start of October meant something like 700 rows still to drive, and each row meant five logs that first had to arrive by rail, ride a tugboat across the lake, and be stood upright in thirty feet of brine. Sometime that fall, a pile driver set the last log into the gap. The hammer came down on it, and for the first time there was wood from one side of the Great Salt Lake to the other.

Wood is not a railroad. The crew still had to cap the last rows, lay stringers and planks across the gap, coat the deck, spread the gravel, and lay the rail. Eighteen days after the last pile, the rails from the east and the rails from the west met in the middle of the lake, at the camp that had been Camp 23. By then it had a new name: Midlake.

The two ends of the Lucin Cutoff came together out over the deepest water on the line, on the permanent trestle, the part of the crossing that had never sunk at all. And that is where Harriman chose to finish it. On Thursday, the 26th of November, 1903, Thanksgiving Day, a special train ran out across the new line. Engineering News called it the formal opening.

Bennett described the scene the following spring. “Out in the middle of the north arm of the Great Salt Lake,” he wrote, “in front of railroad men gathered from all over the country, Edward Harriman drove the last spike. It was solid gold. ” A special train standing on a gravel deck on top of thousands of wooden legs, miles from either shore in late November light.

A crowd of railroad men on a bridge sixteen feet wide. A man who had argued against this crossing two years earlier, swinging a hammer at a gold spike over about forty feet of salt water. He drove it on the one part of the line that had behaved. The part that had nearly beaten them was well over a dozen miles to the east, buried under a summer of rock.

So here is the answer to the question this whole story started with, the one the locomotive raised when it settled into the lake in March. Why would anyone lay a railroad across an inland sea whose bottom kept swallowing it? And how did they finally make it stay? They did it because the lake had shrunk enough to let them, and because Harriman wanted a straight, flat railroad more than he wanted the interest on his money.

And they made it stay with two different answers for two different kinds of water. Where the lake was deep and the bottom firm, they built a bridge on seventy-foot fir piles and let it stand. Where the lake was shallow and the bottom soft, they built a bridge on purpose to bury it. And then they poured rock into the buried bed of the Bear River, carload after carload, until the old riverbed quit moving.

Filling the hole took most of a year. What it cost depends on who was counting. Engineering News, reporting the opening, put the bill at nearly $4,200,000. Bennett heard close to five million.

Farrar, working from the company’s later figures with the track and signals included, gives 8. 9 million. Somewhere between about four and nine million dollars, depending on what you count, to save a railroad 44 miles, more than 1,500 feet of climbing, and almost 4,000 degrees of curves. Arthur had made his own forecast back in the fall of 1902, before any of the trouble.

It was safe to say that January of 1904 would see trains running over the Great Cutoff. Regular freight started crossing two months after that date. After a buried river, a sunken engine, a sunken trestle, and all that rock, the man who wrote the first report missed by about two months. The ceremony was finished.

The gold spike was in. The question nobody at that Thanksgiving celebration asked out loud was whether the shallow fills to the east were really done sinking. They were not, and the company kept saying they were. In May of 1904, Bennett’s article in Scientific American carried the official line about as far as it could go.

No locomotives or cars had been lost during construction, he wrote. Reports of locomotives suddenly disappearing into bottomless pits out in the lake were absolutely false, because nothing of the sort had ever happened. The track had sunk slightly in a few places, he admitted, but no more than any embankment built in water. You keep dumping rock and dirt until it stops settling, and the track, he promised, would be as solid as Gibraltar.

Technically, the first part may even be true. The engine of March 1903 was not lost. It was pulled back out on a cable. But anybody who reads the March and April reports next to that article can see how much work the word lost is doing.

Freight trains started running over the cutoff in March of 1904. Passenger trains came later, and by the first day of 1905, all the through traffic between Ogden and Lucin was going across the lake. The old line over Promontory Summit, the line of the Golden Spike, stayed in place as a spare, and it was going to be needed. That 1915 geological survey guidebook, written for travelers riding the Overland Route, describes the fills across the lake sinking into the mud after the line opened.

There were successive sinks, some of them taking whole sections of track down with them and forcing the railroad to send its trains back around the lake on the old Promontory line. The guidebook’s own comment is the most honest sentence in the whole record. “It might fairly have discouraged the railroad company,” it says, “but perseverance finally conquered. ” So the old line Harriman had replaced kept rescuing the new one.

Every time a sink took out a stretch of fill, the trains went back over the summit where the two railroads had met in 1869, up the ninety-foot grades and around the ten-degree curves, until the crews had dumped enough rock to put the shortcut back. There is a strange footnote. Remember Goodwin’s letter from 1903, the one that said spanning the soft places would have been cheaper than filling them. Notes from a talk given by the railroad historian Bob Burton in 1997 record that in letters dated 1909 and 1913, Hood himself argued for replacing the fills with trestle and put a price on it: a little over $104,000 a mile.

Six years after calling the sinking stories yarns, the chief engineer was making the letter writer’s case. Even the finished track needed help. A few months after the opening, crews started placing more fill under the tracks to stop them sagging, a job that took about six more months. But it held.

Year by year, the rock won. The fills settled and were raised, settled and were raised again, and the trains kept running across the lake. By 1908, five passenger trains and seven freight trains were crossing in each direction every day. The Southern Pacific started selling the trip to tourists.

One of its brochures called it going to sea by rail. And out in the middle, the camp that had been Camp 23 turned into a town. Midlake was a train order station on the trestle, a row of houses standing on a wooden platform on top of the piles, about eighty yards long and forty feet wide, with the main line fanning out into three tracks in front of it. Farrar describes it as a town with a transcontinental railroad for a main street and the Great Salt Lake for a backyard.

At its peak, about thirty people lived there. There were no stores and no post office. Groceries, newspapers, and mail came out every day by train from Ogden, and so did the trip to a doctor or a school. The train order office was run by three telegraphers.

When the office finally closed, all three of them were women, living on a platform in the middle of an inland sea and handing orders up to passing train crews for a living. Midlake lasted 41 years. In April of 1945, the Southern Pacific replaced its old block signals on the crossing with centralized traffic control run from a distant office, and a train order station out on the lake was no longer needed. The three telegraphers were transferred.

By August, the section crews and their buildings had been moved to Promontory Point. By then, the cutoff had already seen the worst day in its history, and it had nothing to do with mud. On the last day of 1944, before dawn, in thick fog, a westbound Southern Pacific Mail Express ran into the back of a slower passenger train on the fill across Bear River Bay, near the section house at Bagley. The passenger train was full of wartime travelers, many of them in uniform.

The first reports counted 48 dead. The federal investigation that followed put the total at 50. It is still remembered in Utah as the worst train wreck in the state’s history, and it happened on the same stretch of shallow water where the engine had gone into the lake 41 years earlier. The investigators blamed a train running too fast for its signals in the fog, but it tells you how busy this crossing had become.

In wartime, the line Hood built was carrying troop trains, mail, and freight across the lake around the clock. That traffic is what finally retired the wooden trestle. Heavier trains and decades of salt were wearing it out. Speeds on the old timber had been cut to twenty miles per hour, and by the early 1950s, the railroad’s engineers were looking at a bill to rebuild it in place.

Instead, the Southern Pacific built a solid rock and gravel causeway alongside it, about 1,500 feet away, so the old bridge could keep carrying trains while the new one went in. The Morrison-Knudsen Company of Boise did the work in the second half of the 1950s, and by 1959, the trains had moved onto the causeway. Its cost was $53 million, more than five times the highest figure anyone has given for the whole original cutoff. The old bridge did not go quietly while the new one was going in.

A fire in May of 1956, started by an overheated wheel housing, burned 645 feet of trestle and cut the line for days. In November of 1957, one of the barges working on the new causeway cut a hole through the trestle and stopped trains for two days. The trestle stayed beside the causeway for decades as a backup line, then fell out of use. In 1993, the salvage rights were sold, and by 2000, the crews had taken it apart.

Much of that old timber, after ninety years in the brine, went on to a second life as reclaimed lumber. Follow the Lucin Cutoff today from east to west, and here is what you are looking at. Start in Ogden, where the line leaves the city and runs west across flat farm country toward the lake. Somewhere along that first stretch, under the grade, are the stone culverts and steel pipes the crews finished in the spring of 1902.

At the shore, the track heads out onto the fill across Bear River Bay. This is the Bagley Fill, the stretch built on sandbags and doubled piles and months of dumped rock, the place where the engine went in and the place where the trains collided in the fog in 1944. Somewhere inside that embankment, if the plan of 1902 held, the legs of the temporary trestle are still standing where the crews buried them. Arthur’s report said exactly that: the trestle would be filled with rock, leaving the bents in place.

Then comes Promontory Point and the long rock cut through it that a contractor’s crew blasted in the fall of 1902. West of the point, where the wooden trestle once ran for nearly twelve miles, the trains now ride the causeway of the 1950s. Somewhere out there, beside the causeway, is the spot where a row of houses called Midlake stood on stilts for 41 years. The causeway did one more thing nobody fully intended.

By cutting the lake in two, it left the north arm with very little fresh river water, and the north arm turned far saltier than the south. In satellite pictures, the two halves of the lake are different colors, split along a straight line that is the railroad. On the far side, the line climbs off the lake at Lakeside, where the quarry that fed the western fill in 1902 was opened again for the causeway in the 1950s and again in the 1980s, when the lake rose and the railroad had to raise its fills against the waves. Then it runs west across the salt desert, up the gentle grade to Lucin, near the Nevada line.

There is one stop on this trip you can actually stand on. Go north instead to Promontory Summit, where the old line ran. In 1942, during the Second World War, the rails over the summit were pulled up and given to the war effort as scrap. Today, the place where the two railroads met is Golden Spike National Historical Park.

You can stand on that old grade on the climb Huntington and Harriman spent a fortune to get around. The picture on the cover of the original account—a thin line of track running out across a mirror-flat lake until it disappears into the horizon, a small train on it, no shore anywhere—only ever showed half the crossing. What you see is the permanent trestle, the fir legs standing in deep water, the bridge Harriman chose for his gold spike. That half is gone, taken apart by salvage crews in the 1990s.

The other half was never photographed for a postcard because it was built to disappear. It is the scaffold on the east side, the doubled legs across Bear River Bay that existed only so dump trains could bury them, and the rock poured on top of them in the summer when the company said nothing was sinking. That half is still out there. If Arthur’s report is right about how it was built, those legs are still standing in the mud of the old riverbed inside the embankment, holding the shape from the inside the way the stakes hold a garden bed.

Trains still roll across that fill over a wooden bridge nobody has seen for more than a century. The bottomless pit had a bottom after all. They just had to bury a bridge in a river to reach it.