The winter of 1907 found America doing something its own engineers had called a physical impossibility. A dam of rock was being thrown across the runaway Colorado River, dumped off a wooden railroad trestle one flat car at a time. The trestle stood on 90-foot timber piles, spanning a gap in the riverbank more than a thousand feet wide. The whole Colorado was pouring through it.
For most of two years, the river had been leaving its bed, running downhill into a desert below sea level and filling it with a new sea. Now, before February was out, the river was supposed to run south to the Gulf of California again. This is how a railroad stopped a river, and why it had to do it twice. Stopping it took a railroad willing to wreck its own schedules.
Two men ran through all of it. The first was Epes Randolph, who ran the Southern Pacific’s lines in Arizona when the railroad took charge of the river in 1905. The second was H. T.
Corey, a former engineering professor who ended up running the fight at the break itself. The way they won came down to one stubborn idea: put rock into that river faster than the river could carry it off, and turn an entire railroad into the machine that does it. For a long time, the river was winning. In the summer of 1905, Randolph went down below the Mexican border to see the trouble for himself.
It sat about four miles downstream from the canal’s main headgate, just inside Mexico. A year earlier, silt from the Colorado had choked the canal that fed California’s new farm country. To keep the water coming, the irrigation company had cut a second intake straight into the riverbank. There was no gate on it.
That was a bet on the river behaving. The river gauges at Yuma had about 27 years of records behind them, and in all that time, no winter had brought two floods. The winter of 1905 brought several, one after another. Each one chewed at the banks of the new cut.
When Randolph arrived, it had more than doubled in width, and the river was pushing through on its own. The real trouble was which way the water wanted to go. From that spot, the Colorado’s old bed fell about 100 feet on its way south to the Gulf of California. The ground to the west fell about 400 feet down to the bottom of the Salton Sink.
A federal engineer who inspected the break that summer worked out that the slope toward the sink was roughly four times as steep as the slope toward the sea. Every flood had a shorter, steeper road to take, and each one sent more of the river down it. Randolph did the arithmetic and sent it to the man who would have to pay. E.
H. Harriman ran the Southern Pacific, and the railroad had just loaned the irrigation company $200,000 to fix its canal. Randolph wired him that $200,000 wouldn’t come close. Controlling the river, he wrote, might easily run into three-quarters of a million dollars.
The answer came back as one question: was he certain he could put the river back into its old channel? Randolph answered, “I am certain that it can be done. ” His boss’s reply was five words long: “Go ahead and do it. ”
The Colorado had two kinds of flood, and only one of them kept a schedule.
Its own rise came every spring, peaked around July, and fell off by mid-August. The floods off the Gila, the big tributary that joins it at Yuma, came in any season, and the worst of them came in fall and winter with almost no warning. They tried the old way first. That fall, crews built a long barrier of piles, brush mattresses, and sandbags to push the current away from the breach.
On the 29th of November, a flood came down the Gila carrying so much driftwood that it nearly covered the water. The barrier went out before the flood even peaked. When the water fell, the breach was about 600 feet wide. As 1905 ran out, nearly all of the river was going the wrong way.
Drowning wasn’t the danger that scared the engineers most. Every farm and every town in that valley drank water that came through canals, and the canals ran on gravity alone because the river sat higher than the fields. If the runaway river cut its own bed deep enough, the canal heads would be left hanging above a gorge and the water would stop. Thirteen months later, the president’s own message to Congress spelled it out.
Water, it warned, cannot be diverted by gravity into the canals already built. About 12,000 people had settled that valley in four years. The engineers believed they were in more danger of being driven out by thirst than by the flood. How do you shut a door that a whole river is already running through?
Nobody in America had done it on a river like this. One attempt with brush and sandbags had lost. For the next attempt, Randolph wanted rock. Plain broken rock dumped straight into the river until the pile rose high enough to turn the whole current back into its old bed.
Almost every expert who heard that plan thought it would fail. Somewhere between 40 and 50 well-known engineers came down to look at the break in those two years, and hardly any two of them agreed on what to do. Their verdict on the situation was grim: it was a desperate one without engineering parallel, with only a fighting chance of controlling the river. About rock, though, most of them agreed.
The riverbed there was fine, soft silt, and they said a stone dropped on it would just keep sinking. They had a second objection: once water started pouring over the top of a rock pile, it would pluck the top stones off one at a time. Every stone it took would make the next one easier to take, until the river had cut a gap nobody could close. Randolph had heard arguments like that before.
Years earlier, building railroads in the South, he had used brush mats and rock fill on the Tombigbee River in Alabama and seen them hold. He listened to the experts, weighed what they said, and kept his plan. The idea in plain terms: a river carries off whatever you drop into it at a certain rate. If rock goes in slowly, the current just rolls it away downstream, stone by stone.
The trick is to drop it in faster than the current can move it. Then the pile starts to stay. It is the same thing you see with a sand wall where the surf comes in. Build it slowly, and every wave knocks it flat.
The only walls that stand long enough to hold the water back are the ones heaped up faster than the waves can take them. That turned the whole problem into a hauling job. How do you get that much rock to one spot on a river in the Mexican desert and get it into the water fast? Randolph’s answer was to bring the railroad right down to the river.
Crews would lay a spur off the Southern Pacific’s main line straight to the break. The railroad borrowed 300 giant side-dump cars from the Union Pacific, steel cars that had helped build a railroad across the Great Salt Lake a few years before. Railroad men called them battleships. Each one carried 50 or 60 tons, and a crew could tip the whole load over the side at once.
In 1906, there were no bulldozers and no dump trucks. Every ton of that rock would be dug out by steam shovel, hauled in by steam locomotive, and dropped by crews standing on a wooden bridge over moving water. The rock would come from a new quarry blasted into a granite ledge at Andrade on the border, beside the canal’s new concrete headgate, with a working face 600 feet long. From Los Angeles came 1,100 timber piles, each 190 feet long, and 40 miles of steel cable.
The cable was for weaving brush into mats wide enough to cover the river bottom from bank to bank, so the rock would have something to sit on. The piles were for something harder: you can’t dump rock off a train into the middle of a river unless the train can get out over the water. That meant a railroad trestle built straight across the current, standing on piles driven into the same soft silt the experts were worried about. Randolph had been sick for twelve years.
He was born in Lunenburg County, Virginia in 1856, and he had been building railroads since he was twenty. By his early thirties, he was running the construction of a railroad bridge across the Ohio River, joining Covington, Kentucky to Cincinnati. Then in 1894, he came down with tuberculosis. He quit his job and came west for the dry air.
He first lived in Indio, California, a railroad town at the north end of the same basin the river would start filling. Then he moved to Tucson, and the Southern Pacific put him in charge of its lines between El Paso and Los Angeles. The man the railroad sent to save that basin had first gone there to get his own health back. Around Los Angeles, he had managed an electric railway network that grew to hundreds of miles of line until his lungs sent him back to the Arizona air in 1904.
When the river broke loose, he was already president of two railroads in Arizona and Mexico, and among the advisers the Southern Pacific’s president trusted most. After the November flood, Randolph ran two plans at once. One was a steel and concrete headgate set on solid rock near Pilot Knob, so the valley could take its water from a safe intake while the breach was being closed. The other was a wooden gate right beside the breach to carry the river’s low water around the work.
Men on alternate shifts built the wooden gate night and day. It wasn’t ready until the 18th of April, 1906. That same day, San Francisco shook apart. The earthquake and the fires after it wrecked much of the city.
They also held up the machinery for the canal’s big new dredge, which was on order there and wouldn’t be ready until November. The Southern Pacific’s president hurried to the city to help, and Randolph followed him there. In temporary offices, with the ruins still smoking, he told his boss the irrigation company’s money was spent, the river was still loose, and the summer flood could take the whole valley. The railroad’s president had a burning city on his hands, his trains busy carrying refugees out of it, and his own lines crippled.
He said yes anyway and put up another $250,000 for the river. Corey later wrote that of all the extraordinary things that happened in that fight, this one struck him as the most remarkable. The next day, the irrigation company’s chief engineer resigned, and Corey took his place. From then on, the railroad ran the fight, and every plan was drawn up by its own engineers.
Nothing could be closed yet. The summer flood was coming down the Colorado the way it did every year, and it came bigger than the new wooden gate could carry. By spring, the breach was a quarter of a mile wide. By late June, it was more than half a mile, and the whole river was pouring through it, spreading eight or ten miles wide before it found the old dry channels and ran north.
The first place to find out what that meant was a border town called Calexico. Calexico and Mexicali, across the line from it, were partly washed away. The railroad’s line into Mexico was underwater or gone in long stretches, and so was the little interurban out of Holtville. The wooden flumes that carried irrigation water across the New River’s channel were swept off toward the sea, and about 30,000 acres on the west side of the valley went dry in the middle of a flood.
The people who farmed that ground had bought it under a new name. The promoters had started calling the Salton Sink the Imperial Valley, because nobody wanted to buy land in a place called a desert or a sink. The worst thing in the valley that year wasn’t the water standing still. It was the water moving.
The fine silt of that valley floor washed out almost like powdered sugar. Wherever the flood gathered into a strong current, it started a small rapid. The rapid turned into a cascade, the cascade into a waterfall, and the waterfall began eating its way upstream. A waterfall that walks upstream sounds like a tall tale.
The engineers measured it. Those falls could move upstream 4,000 feet in a single day — three-quarters of a mile of ground gone between breakfast and supper. Thousands of acres behind those falls were so torn up they could never be farmed again. Behind them they left gorges 50 to 80 feet deep, with walls that stood nearly straight up.
Some of those gorges were more than a thousand feet across. When the season was over, the channels the runaway river had cut added up to more than 40 miles. Corey later worked out that the river had carried almost four times as much earth into the sea as was dug out for the entire Panama Canal, and it had done that in nine months. Each of those falls was working its way back toward the canals.
If one of them cut far enough to break into the irrigation system, the water in every canal and ditch in the valley would pour down into the gorge below it. Crews used dynamite to steer one cutback on the new river so it wouldn’t slice a new channel across the middle of the valley and leave half the farms without water. The dynamite worked on that one. The falls kept coming anyway, and the only way to stop all of them was to stop the river at the break.
Meanwhile, down on the basin floor, the railroad was fighting something that rose a little higher every day. The Southern Pacific’s main line to the east ran straight across the floor of that basin. For 61 miles, the tracks sat below sea level, and at the lowest point, the rails lay 266 feet below the ocean. When those tracks went down in the 1870s, the basin was bone-dry desert.
Jokes went around about Indio becoming a seaport. The railroad had no reason to laugh. A sea was coming up under the tracks. After the November flood, the new sea had covered about 150 square miles.
The Bureau of Reclamation has a photograph from those months of a Southern Pacific passenger train crossing the new sea on a trestle five miles east of the station at Salton. The railroad’s main line was running across open water on stilts. In June of 1906, the sea rose seven inches a day across 400 square miles. That’s more than four feet a week.
The station and siding at Salton went under, and so did the works of the new Liverpool Salt Company with the little railroad it used to haul salt. By September, the water over the salt works was more than 60 feet deep. The railroad did the only thing it could. It moved the track.
The Southern Pacific’s own historians count eleven temporary lines, called shoofly lines, laid between July of 1905 and February of 1906 just to keep trains running around the water. That summer, the main line had to be pushed back to higher ground five more times. In the end, the railroad built 39 miles of brand new mainline up out of the sea’s reach, running toward Mecca. Moving was harder for the people who lived there.
The Torres Martinez reservation of the Desert Cahuilla sat in that basin. Their families knew this lake. It was in their oldest stories — a sea that came into the basin and left again long before there were rails across it. When the water came back, most of them packed what they had and moved up into the mountains.
Nearly half of the reservation would end up under the new sea. Nobody could move the canal intakes, though, or the towns. The only real fix was back at the break, and in early August, the river finally dropped low enough to fight. First, the railroad finished its road to the water.
Work on the spur from the main line at Hanlon, near Pilot Knob, started on the 1st of July, and the first trainload of material rolled over it on the 15th of August. Before that, every pile and sandbag for every attempt had been floated down the river on barges. Randolph and Corey leaned on the railroad for everything: pile drivers, steam shovels, bridge crews, and repair parts. Corey later explained why a railroad could put together quickly a thoroughly organized and efficient force of men, and could grow or shrink that force overnight without confusion.
No irrigation company could do that. Then it needed hands. In that heat, unskilled labor was the hardest thing on the job to keep. Men came for a few days and left.
So Corey recruited from the nations of the region, across two countries, because every bit of this work sat on Mexican soil a few miles below the border. Pima, Maricopa, Tohono O’odham, and Kechin men came from Arizona, and Cocopah and Kumiai men came from Mexico. The company’s papers use older names for some of them — Papago, Yuma, and Diegueño. With their families, they made a camp of about 2,000 people beside the work.
The rest of the crews were Mexican workers from nearby and drifters from all over the country who had read about the fight in the papers. Most of what we know about who did this work comes from the engineers’ reports, and those reports mostly list the men as numbers. Mexico put the whole area under martial law, and a force of rurales, the Mexican mounted police, came up to keep order in the camps. First came the mattress, made on a barge more than 100 feet long, decked over and rigged like a giant loom.
Two shifts wove brush and steel cable into a mat about 100 feet wide. They kept at it twenty days and nights. They used about 2,000 cords of brush, stacked the way firewood is stacked — a woodpile four feet high and three miles long — and it went to the bottom of the Colorado as a double layer from bank to bank. Thomas Hind, the construction superintendent the dam would later be named for, started by pushing two jetties out from the banks to narrow the gap.
Then a trestle went across the rest of it on top of the mat, and in the middle of September, the first battleship trains rolled out and started tipping rock. In less than two weeks, the pile was high enough to start turning the river toward the gate. It worked. By the 10th of October, the rock had backed the river up so far that about 13,000 cubic feet per second were running through the new wooden gate — an Olympic swimming pool filled in about seven seconds.
Only a tenth of the river was spilling over the dam. The wooden gate was carrying almost all of it, and it was taking a beating. Driftwood piled against it, and water was starting to find its way underneath. Crews rushed a second trestle into place in front of the gate so they could dump rock there and take some of the pressure off.
When the first rock cars rolled out onto that new trestle, three of its bents settled, the train went over, and several battleships tumbled off the rails. Nobody was seriously hurt. The next afternoon, the 11th of October, the gate buckled in the middle and went out with a roar. Two-thirds of it floated off downstream.
A hundred yards downstream, a quick engineer ran his train off a trestle with seconds to spare. Four months of work and about $122,000 were gone. The bypass channel that was supposed to carry the river around the work was now the river, and the Colorado went through it back toward the sink. But when the engineers walked out on the rock dam the river had just abandoned, they found something that mattered more than the gate.
The rocks hadn’t sunk. The brush mat hadn’t broken under the weight, and the dam had only leaked a little. On the one question that decided everything, the experts had been wrong. The record doesn’t say what Randolph said when he heard that.
His rock worked. Now he had to close two channels with it before the Gila sent down another flood. This time, the railroad threw everything it had at the break. Ten work trains hauled rock from three or four quarries.
The labor force climbed to about a thousand men with 700 horses and mules. They built high rock dams across both the old intake and the bypass, working through the nights, and tied the two dams together with heavy levees into a single barrier about half a mile long. Rock by itself leaks, so they faced the dams with gravel and clay and drove that fill into the gaps between the stones with jets of water from big pumps. Down the old riverbed, which had been dry for more than a year and had grown up in willows, crews of native workers cleared a strip of brush more than 100 feet wide for miles.
Mule teams dragging Fresno scrapers cut a shallow channel along it to give the river a clean start back toward the Gulf. On the 4th of November 1906, the whole river turned. The Colorado went back into its old bed and ran south, and none of it was heading for the sink — which hadn’t happened in more than a year. That should have been the end of it.
Corey left 75 men to patrol the new dams and the levees that ran south of them, and took the train to Calexico to catch up on the paperwork of his other job, general manager of the irrigation company. About a month later, in the first week of December, another flood came down the Gila and caught the patrol by surprise. It pushed the Colorado from about 9,000 cubic feet per second to about 45,000 — five times as much water leaning on brand new work. The river climbed against the new rock dam, and the dam held.
What didn’t hold was an older earthen levee roughly a quarter-mile south of it, on a bank that two dry summers had baked and cracked. Water worked its way underneath, and somewhere around midnight the levee began to leak, then to give. The patrol wired Corey. He and Hind made a hurried trip down through the night, and at first light they could see three separate breaks.
There were too many leaks to fight. The banks caved and washed away, and within a day or two, the whole Colorado was running back toward the Salton Sea. It happened so fast that crews working south of the breaks were stranded. The little river steamer Searchlight was sent down to get them, and it ran aground several miles downstream on a riverbed that had just gone dry.
In a day or two, the share of the river headed for the sink went from none of it to all of it. Then the break grew. By the 20th of December, it was the gap from the start of the story: 1,100 feet wide and 40 feet deep, with the entire Colorado roaring through it. The Southern Pacific had now spent more than a million dollars on a river it didn’t own, in a country that wasn’t its own.
Randolph reported that holding the river for good would take about 20 miles of new levee along its west bank, and could cost another million and a half. The railroad itself had surprisingly little riding on it. The valley’s freight earned the Southern Pacific only $20,000 or $30,000 a year. For about $50,000, it could have moved its tracks up out of the sea’s way and walked off.
The people with the most riding on it were in Washington. The federal government owned almost all the irrigable land along the lower Colorado, and it was building a dam of its own at Laguna above Yuma, with about a million dollars already spent. If the falls kept cutting upstream, they could reach the main river, then Yuma, where the great railroad bridge stood, and then Laguna itself. Randolph told his boss it was time for the government to help.
So, in the week before Christmas, the question was simple: would anybody pay to fight this river again? The railroad’s president didn’t walk away, but he wasn’t going to carry it alone either. On the 13th of December, he put the whole thing in a telegram to the White House. President Theodore Roosevelt answered two days later.
He assumed, he wrote, that the railroad meant to keep working on the break, and he wanted to know exactly how far it intended to go. The railroad wanted help and a promise it would be repaid. The president’s first answer was that the government couldn’t act on a break in Mexico, and Congress wasn’t in session anyway. The telegrams went back and forth for a week.
Back in the valley, nobody sat still. The governor of California and towns all across the south of the state were sending the president urgent appeals to save the settlements in the sink. On the 20th of December, the order went out to close the break, and Randolph passed it to Corey by telegram: “Control the river at all cost. ” From that day, every other job on the Southern Pacific came second, except the relief work still going on in San Francisco.
When the president heard it, he wired back: “Am delighted to receive your telegram,” and promised to ask Congress for a law so it wouldn’t happen again. And the whole time those telegrams were going back and forth, the entire river was running into the sink. The man who had to close it was Corey. He was 36.
He’d been born in Montmorency, Indiana, studied mechanical engineering at Purdue, and taken advanced degrees at Cornell. He taught civil and sanitary engineering at the University of Missouri, and by 30 he was dean of engineering at the University of Cincinnati. Then he left the professor’s chair for the railroad, and in May of 1905 he became Randolph’s assistant in Tucson. The man running the hardest river job in the country had been teaching engineering in a university classroom a few years before.
He later wrote the whole fight up for the American Society of Civil Engineers and then turned it into a book. Much of what we know about the next seven weeks comes from him. What he faced was worse than August. Back then, at low water, the channel they dammed had been six or seven hundred feet wide.
Now the gap was nearly twice that. George Kennan, who wrote the whole fight up ten years later, put the flow through it at about 160 million cubic feet of water an hour — enough to flood more than five square miles of ground a foot deep every hour. There was no low water to wait for this time, and no time to weave a mattress. The spring flood was coming.
The president’s message to Congress would soon warn that if the break wasn’t closed before it, all the property built up in that valley would probably be wiped out, farms and towns alike. So they changed the method. Instead of one trestle on a mat, they built two trestles across the whole break about 50 feet apart, and dumped rock from both at once. Two ridges of stone would rise under them.
Then the space between the ridges would be packed with gravel and clay until the two ridges and the fill between them made one wide dam. Two instead of one, because the whole plan came down to speed. Two trestles meant twice as many cars could stand over the break and dump at the same moment. Two ridges also gave the gravel and clay a wall on each side to be packed against, instead of being washed away as fast as it went in.
Each trestle was a line of bents. A bent was a row of those long piles driven side by side, capped with a timber a foot square, with heavy stringers running from cap to cap to carry the rails. In the deepest part of the break, every pile had to go down through as much as 40 feet of moving water before it even reached the silt. A pile driver of that day was a tall wooden frame with a heavy hammer that a steam engine hoisted up and dropped on the head of the pile again and again until the pile was deep enough to hold.
Out over a river running at flood speed, it was dangerous work. The piles went in from both banks. A third pile driver worked from a barge out in the middle of the channel, held against the current by long steel cables. Every time that crew swung a new pile upright, the barge risked rolling over.
Boats sat manned downstream the whole time, ready to pull out anyone who went into the water. The river fought back. Three times within a month, floods loaded with driftwood tore out the piles and wrecked the trestles — sometimes partly, sometimes completely — and more than once they hit a bridge that was nearly finished. Three times the pile drivers went back to work.
The records don’t say how Corey kept the crews going out there after the third time, only that he did. A history of the fight written for the San Diego Historical Society describes Corey in those weeks as seeming to be everywhere, walking and patrolling the job at all hours and never appearing to sleep. Hind ran the northern part of the dam across the old intake and the bypass. A second superintendent, C.
K. Clark, ran the southern part across the new break. The credit for getting those trestles across belongs to men whose names mostly never made it into the story: the Southern Pacific’s bridge foremen and their crews. They had built hundreds of trestles like these all over the railroad system, and they knew how to set a pile bent in bad ground.
There was good news on the other front — the danger of the valley going dry. The big dredge the earthquake had held up was finally at work, clearing four miles of silted canal below the new concrete headgate so the valley could draw its water from a safe intake while the break was fought out. While the telegrams were still flying, Randolph had kept the whole organization together at the river instead of sending it home. Car after car of long piling, heavy cap timbers, and stringers rolled in and waited on the sidings.
The railroad got its rock ready. The quarry at the border was nearly worked out, so every quarry the Southern Pacific and its connecting lines could reach within about 500 miles was put to work loading. Just south of the border, the sidings were stretched to 7,000 feet to hold the waiting cars. On the 12th of January, while those trestles were still going up, the president’s special message went to Congress.
It said the railroad was the only agency equal to the task of controlling the river. Even with unlimited money, he wrote, the government’s own people couldn’t carry on the work with the speed it needed. The same message laid out what was at stake if nobody did: about 700,000 acres along the river that would be left desert, land it called as fertile as the Nile Valley, ground that could someday add at least 350,000 people to Arizona and California. In January of 1907, all of those future farms depended on a wooden trestle that kept getting knocked down out in the sink.
The new sea kept rising the whole time the trestles went up and came down again. Every week of pile driving was another week of the whole Colorado running into it. By the 27th of January, the first trestle was across the break for the fourth time. At 5:00 that evening, the bridge crews spiked the last rail into place.
The workmen stood back and cheered as the first train of rock eased out onto the trestle. By dawn the next morning, 145 carloads of rock had gone into the break. From then on, the dumping didn’t stop. Crew relieved crew at every shift change without a pause in the work, and at night, great searchlights lit the trestles and the water below them.
The rock came from everywhere the railroad could reach. It came from Patagonia, Arizona, more than 480 miles away, hauled over two mountain passes to get there. It came from Tecate, 60 miles to the east. And it came from quarries near Colton, 200 miles to the west, dragged over San Gorgonio Pass.
Those quarries sat on three different railroads — the Santa Fe, the Salt Lake Road, and the Southern Pacific itself. Trains of rock ran with the right of way even over passenger trains. Corey described it to a committee of Congress the next year. For three weeks, he said, two divisions of the Southern Pacific system were practically tied up by the railroad’s demands for equipment and facilities at the break.
A thousand flat cars were working for the river and nothing else. Shipping out of San Pedro, the port of Los Angeles, was practically abandoned for two weeks until some of those cars could be sent back. Frederick Newell, who ran the United States Reclamation Service, wrote down what the scene at the break looked like, and his description ended up in the Smithsonian Institution’s annual report for 1907. The stones were as big as a gang of men could push off a flat car, or as many men as could get around one stone, and hundreds of workmen were pushing them off as fast as the cars could be placed.
Some were too big even for that, so crews broke them right on the cars with small charges of dynamite they called pop shots. Newell wrote that the pop shots sounded like artillery in action. On top of that came the roar of the water, the whistle signals, the orders shouted to the men, and the noise of what he called an army working day and night to stay ahead of the river. At first, trainload after trainload seemed to vanish into the hole.
The heap rose anyway, and as it rose, it held the river back. The water climbed higher against it and started pouring over the top in a cascade. That cascade undercut the downstream face, so the pile would settle and stones would roll away down the river. Every bit the river took had to be made up with more rock, and faster.
To send the Colorado back into its old bed, the engineers had to lift the entire river about 11 feet — higher than the ceiling in most houses. Then the share headed west started to fall. As the heap climbed, less and less of the river went over it, and more and more of it turned south, back down its old channel toward the Gulf. Now the finer stuff went in.
Train after train of small stone and gravel from the nearby hills filled the spaces between the big rocks. Gravel came from the railroad’s big pit at Mammoth, 40 or 50 miles up the main line, and clay from pits near Pilot Knob. Last of all, crews washed silt into the cracks with hoses until the water stopped seeping through. On the 10th of February 1907, the break was closed, and the Colorado was back in its own bed, headed for the Gulf.
It was fifteen days after the first carload of rock went in, and 52 days after the railroad was told to go ahead. Two years for the river to build that sea, and fifteen days for a railroad to cut off its supply. Corey told Congress it had been a case of putting rock into that break faster than the river could take it away. “I suppose,” he said, “we handled rock faster than it was ever handled before.
” By the railroad’s count, more than 2,000 carloads of rock went into the break in that stretch. Corey put it at about 80,000 cubic yards — enough to bury a football field, end zones and all, in rock nearly 40 feet deep. None of it came free, and none of it was safe. The history of the closure written for the San Diego Historical Society says there were numerous accidents and luckily few deaths.
The records don’t give a count of the men who died, and no honest guess is worth making. The whole fight, from the first loan to the last carload, cost the railroad about $3,100,000. The railroad asked the government to pay it back, and in the 1920s, Congress was still sorting out how much. When the government finally paid in 1930, the check covered about a third of the bill.
Out in the basin, for the first time in almost two years, nothing new was coming in. The Salton Sea stopped rising. The share of the river running into the sink was back to none, and this time it would stay there. The finished barrier went into the records as the Hind-Clark Dam, after the two superintendents who built its halves.
Kennan tallied that dam alone at 1,200 piles and about 16,000 feet of heavy pine stringers — three miles of squared timber. He counted 5,765 carloads of rock, gravel, and clay in it, all hauled by train. If you coupled those cars into one line, it would run for something like 40 miles. And the railroad didn’t trust it for a minute.
Crews built new levees down the west bank with another 5,000 carloads of gravel from the Mammoth pit to face them, and a railroad track running along the top so men and rock could be rushed to any spot the river started to cut. A telephone line ran along the levee with a call station every mile. Trains loaded with rock stood waiting, and thousands of sandbags sat ready for the first sign of a leak. It held.
Kennan counted two floods in the years since that ran above 140,000 cubic feet per second — bigger than the November flood that had torn the break wide open in the first place. The river never got through the railroad’s defenses again. Randolph spent the rest of his life in Tucson running railroads until he died there in 1921. Four years later, the city opened a big new park and named it for him.
It carries another man’s name now, but a bust of Randolph still stands near the middle of it, with plaques telling what he built. You can still follow the ground that fight was about, from the border down to the sea. Start at the border west of Yuma at Pilot Knob, the granite hill on the California side of the river. The steel and concrete headgate Randolph pushed through that first winter is still standing at the Hanlon heading, with the Mexican town of Los Algodones just across the border fence.
The chain that ran up to the gate’s counterweight is still hanging on the structure. The railroad tracks that crossed that heading were the only solid way over the canal, and the rock for the break rolled across them. From 1907 to 1918, that gate controlled the water going into the valley’s main canal, until a newer gate was built upstream. The old canal doesn’t even run into Mexico anymore.
It stops right there at the fence. The break itself was four miles south, inside Mexico. Two years after the closure, the Colorado changed its course again several miles below the new levees and swung off toward a place called Volcano Lake on its own. Head west along the border to Calexico and Mexicali, the twin border towns.
The new river still flows north out of Mexicali, across the border through Calexico, and on toward the Salton Sea. In the channel the runaway Colorado tore deep in 1906, the New River got its name in 1849, when water turned up flowing down that channel into the desert without warning. The basin had caught the Colorado’s overflow many times. A lake about half the size of today’s sea formed here as recently as 1891.
The same year, the river went back into its bed. California carved a new county out of San Diego County for this valley and named it Imperial. Drive north from there through El Centro, the county seat, and out past Holtville and Brawley. The Alamo River runs through that farm country — the eastern of the two channels that carried the Colorado into the sink.
About 180,000 people live in Imperial County now, farming ground the railroad’s rock kept from being cut into gorges. Four years after the closure, the valley’s farmers voted to form the Imperial Irrigation District and take over the old canal company’s works. That district still runs the valley’s water today. Runoff from those fields, and from part of the Mexicali Valley across the border, is most of what still flows into the sea.
Both rivers end at the south shore of the Salton Sea. That’s where the state has been building shallow ponds for fish and birds, and in the fall of 2024, thousands of gulls, terns, and ducks were photographed out there on the new water. Swing up the east side of the sea to Niland, a mile and a half from the shore, and then to Bombay Beach, a small town sitting about 223 feet below sea level. The sea came for these shore towns one more time in the late 1970s, when heavy tropical storms pushed the water over its banks and damaged the towns and businesses along the shore, many of them beyond repair.
A main line still runs along this side of the sea toward Mecca, carrying Union Pacific freight. On the mountains along the west side of the sea, near Travertine Rock, the shoreline of the ancient lake that filled this whole basin before any of this is still visible, far higher than the water sits now. At the north end of the basin, the tribe whose land the sea took still has its headquarters at Thermal, not far from Mecca. The Torres Martinez Desert Cahuia also own a casino at Salton City over on the west shore.
And the sea itself is still there, about 120 years after it was made. It covered about 305 square miles in March of 2025, and it’s getting smaller. Since 2003, its surface has dropped about 13 feet, and around 16,000 acres of old lake bed have come up dry in the last two decades. The water is now more than twice as salty as the ocean.
Of everything in this story, that’s the turn nobody saw coming. Today, the state is building ponds to hold water along the shore of the same sea a railroad once fought to stop. And the sink that swallowed the Colorado is slowly giving its ground back. Down at the break, when the rock and gravel finally reached the tops of those two trestles, the crews pulled the rails, ties, and stringers off one of them to use again somewhere else.
The other one they left where it stood. So one of the wooden trestles from the start of this story never left the river at all. It was buried inside the rock it had dumped. A railroad bridge turned into part of a dam.
And the Colorado, after all that time spilling downhill into a desert, went back to the sea. Twenty-five years later, America had to move this same river on purpose.
