The Billion Dollar Boat

A mile beneath the churning Atlantic lay the Central America and in its rotting hold a cache of wealth unimaginable: thousands of priceless gold coins, bags of gold dust, bars of solid gold. A fortune for the taking, as Tommy Thompson saw it. His taking.

Heading out the door? Read this article on the new Outside+ app available now on iOS devices for members! Download the app.

On the evening of August 28, 1988, a tugboat nudged a research trawler named the Arctic Discoverer from her dock in Jacksonville, Florida, guided her into the St. John's River, and set her free to find her own way the short distance to the river's mouth. With the sea buoy abeam at 8 p.m., Captain Bill Burlingham took the 180-foot vessel into the Atlantic and set his course for a spot in the ocean about 160 miles east of Charleston, South Carolina.

The Discoverer traveled through the night at just under ten knots, the breeze riffling the white canvas tarp stretched across the foredeck, the air ducts on the forecastle spinning into a blur. Her bow cleaved the incoming seas, rising slightly and falling to rise again in fine fashion. The ship had just been overhauled and newly outfitted from stem to stern, and the crew liked her seaworthiness. Inside a windowless, black-walled control room, technicians watched their shelves of computers and display screens and saw nothing quaking.

All day and throughout the night on the 29th they steamed north. Midmorning on the 30th, Burlingham halted to test the ship's dynamic-positioning system, designed to keep it in one place even in rough seas. The long shafts of the thrusters oscillated in sync, and the wide props spun at varying speeds, and the Discoverer held station. Delays in the manufacture of those new thrusters had cost them a two-month delay in setting out. They now had perhaps three weeks before the autumn storms arrived in the Atlantic.

By the following morning, they were drifting near their destination in calm seas. For the next three days, they prepared and tested deck equipment and checked the systems on their remote-operated vehicle (ROV), a 5,000-pound submersible bristling with sonar equipment, video and still cameras, banks of lights, mechanical arms for grabbing and manipulating, and storage compartments for recovered artifacts.

Tommy Thompson, the 36-year-old marine engineer and entrepreneur who had conceived of the operation, had had half a dozen new blocks specially designed and built for that season, and he wanted Burlingham to test them. The heavy new blocks were the pulleys over which the lines would run to launch and recover the ROV with the ship's crane. Rarely did one fail, but if a block was not going to hold under extreme force, Thompson wanted to know it before he had the block, the vehicle, and half the crew in a tense situation.

With each new block secured near the stern and the men safely out of the way, Burlingham ran the line through the testing block and torqued it tighter and tighter until the line was as taut as it would have to be during launch and recovery; then he tightened it more for a safe margin. In one test a block exploded, and the energy in the line that was suddenly released launched the block like a ball-bearing in a high-powered slingshot. A yellow blur shot past the foredeck and over the forecastle, so far out across the ocean that not even the watch on the bridge saw it splash down. If the block had hit one of the deckhands, the impact would have turned him to jelly.

At noon on the third day of floating above their destination, the winds rose from light to 15 knots and the seas from calm to three feet. Then the gale hit. The winds shot to 35 knots and the seas to ten feet, and all the next day the winds continued at gale force, with furious squalls soaring to 50 knots and the seas nearly doubling in size. Burlingham stood in the wheelhouse, 25 feet off the waterline, and saw walls of water cresting just below his ankles. The waves crashed down and flooded across the deck in gray-green sheets.

They had lashed the ROV to the deck under a heavy tarp, but the pounding of the foredeck twice ripped it loose, and they had to venture onto the pitching deck to resecure it with a dozen heavy canvas straps.

With jolts and spasms, fall in the Atlantic was arriving, and each new preview of the season to come would hit a little harder and stay a little longer. After two days, this one began to subside. By noon on September 7, the wind had dropped to 25 knots and the seas to 12 feet, and over the next three days the seas gradually diminished.

By midnight on September 10, all of the equipment had been checked out and every system was functional. Thompson and his crew were ready to resume their search for the most fantastic sunken treasure in history — a vast, glittering trove beyond the dreams of Midas, tons and tons of gold bars and gold coins and gold dust and gold nuggets that had been resting a mile and a half below, on the bottom of the Atlantic, for 131 years.

Over the past three years, Thompson, his technical crew, and his consortium of investors — the Columbus-America Discovery Group — had spent more than $4.5 million on their quest, along the way outmaneuvering and fighting a series of bruising battles with interlopers in federal court and on the high seas. But even now, after two previous seasons and nearly six months at sea, Thompson was far from certain that he was any closer to finding the gold.

He was searching for the SS Central America, a sidewheel steamer that had gone down in a storm during a voyage from Panama to New York in September 1857. The ship had sprung a leak, the source of which could not be found, and a two-day battle waged by 500 men pumping and bailing out the rising water had ended in defeat and calamity. Passing ships had rescued 149 passengers, including every woman and all except one of the children, but the Central America's legendary captain, William Lewis Herndon, and nearly 450 other men (many of them miners returning east with the fortunes they had dug out of the California gold fields) perished. The disaster's toll included the irrevocable loss of an estimated 42,000 pounds of gold that went to the bottom with the Central America. Fifteen tons of that gold had been part of a secret army shipment intended to shore up the uncertain northern industrial economy, a shipment whose existence was not declassified by the U.S. military until the early 1980s. So much gold went down with the Central America that economists say the disaster was partly responsible for the fiscal crisis known as the Panic of 1857. The sinking of the Central America remains the worst peacetime disaster at sea in American history; only after the Titanic was lost, in 1912, was the 1857 tragedy supplanted in the public imagination.

But as the decades passed, the dream of finding the treasure of the Central America remained alive. Rumors abounded that the ship lay off Cape Hatteras only a hundred feet deep, and every few years some group of treasure hunters would claim to have found the Central America in shallow waters off the East Coast. Anyone who bothered to carefully research the history, as Thompson had, knew that the ship had sunk far out at sea and rested more than a mile down. The technology simply did not exist to find and recover the gold.

By the 1980s, the situation had changed dramatically. In 1985, Robert Ballard stunned the world by locating the Titanic on a depth sounder, and the following summer he and his crew returned with a manned submersible they used to dive 12,500 feet and photograph the great ruined ship. Meanwhile, researchers studying ways to survey the mineral resources of the deep ocean had devised a sonar imaging system that could fly above the ocean floor at the end of a tow cable and render a precise picture of the contours and anomalies of the undersea topography. A number of people in the deep-ocean community began to wonder: Was it now possible to find the Central America and recover her vast cargo of gold?

For 15 years, ever since he first studied marine engineering at the landlocked campus of Ohio State University, Tommy Thompson had relentlessly researched and attacked the technical, scientific, and entrepreneurial obstacles to doing work and recovering artifacts in the deep-sea environment. Out of college, he had gone to Key West and worked for a time with Mel Fisher, a nearsighted chicken farmer from Indiana who migrated to Florida and ran a ramshackle outfit devoted to finding the gold, silver, and jewels waiting in a sunken Spanish galleon called the Nuestra Se±ora de Atocha, which had been lost off the Florida Keys in 1622. (After 16 years of searching, Fisher finally found her in 1985.)

As Thompson observed Mel Fisher's operation and others like it in Florida and the Caribbean, he had discerned a pattern: The treasure hunters operated from day to day, with no long-term plan; they all were underfunded; no one kept accurate records; they raised money primarily through the media; and investors were unhappy and filing lawsuits. And if you went looking for shipwrecks in shallow seas, there was a good chance that storms had scattered the ship's remains, and you could never be sure that no one else had already salvaged the treasure you were after. Thompson came away convinced that deep water was the only place to find undisturbed ships with their treasure intact, and he began to nurture an unshakable conviction that he was the man who could solve the technological and logistical problems that had previously made such recovery operations impossible.

Thompson began cultivating a wide range of contacts with experts and specialists in history, engineering, admiralty law, and deep-ocean exploration. In the early 1980s he worked for four years at the Battelle Memorial Institute, a research facility in Columbus, Ohio, that took on many top-secret government projects. Among other projects, Thompson did research on the feasibility of mining the deep ocean, and it was at Battelle that he learned the value of confidentiality in complex and highly classified operations. (No competitors are more ruthless than treasure hunters.) Perhaps the most remarkable thing he had accomplished was to convince 160 individual investors, most of them based in Columbus, to bankroll his search for the Central America, even though he had never run a business or managed employees in his life.

With $1.4 million in venture capital for the first stage of operations, Thompson went to sea in the summer of 1986 in a flat-bottomed Louisiana mud boat that had previously been used to ferry supplies to offshore oil rigs. Utilizing meticulously collated historical and navigational data and advanced search theory, his plan was to survey a 1,400-square-mile probability map off the Carolina coast using a prototype sonar, the SeaMARC IA. Thompson and his crew spent eight weeks flying the sonar imaging system back and forth — they called it “mowing the lawn” — on 30-mile runs a few hundred meters above the ocean floor. Watching, recording, and reviewing the data on their sonar monitors, they concentrated on distinguishing cultural artifacts from geological features. One site stood out: It appeared to be a sidewheel steamer resting upright on the bottom, with a dark humped shadow amidships indicating paddle wheels. Using an undersea camera sled, they captured only a fleeting glimpse of the ship, which they code-named Sidewheel, before they ran out of time and had to return to port.

Thompson's schedule and painstaking plans were disrupted the following spring, when credible rumors about competitors beginning to search for the Central America forced him to rush back to sea early with a stripped-down version of the remote-operated vehicle he had envisioned. The crew couldn't even test it before they set out. After a series of technical problems at sea, they finally got a long, hard look at the Sidewheel site, only to realize it could not be the Central America. Even worse, just as they began to focus on a promising second sunken ship, the rumors of rivals proved true, and Thompson had to play a series of dangerous games of chicken on the high seas with two aggressive salvage vessels circling the Sidewheel site. Despite the interference, he was able to retrieve a sample artifact from the new site (dubbed Galaxy) and thus secure a valid claim in federal court.

After all that, Thompson was assailed by new doubts when, over the winter of 1987-1988, further analysis of the sonar data indicated that a third site, about 40 miles from Galaxy — the sonar technicians had originally designated it as a rock outcropping or drifts of sand — had an even higher chance of being the Central America debris field. Telling only a few of his colleagues about this new information, Thompson decided to begin the 1988 season by heading secretly to the third site, Galaxy II, while maintaining a cover story that he was just there to conduct tests of their new ROV.

Shortly before noon on September 11, the tech crew on the Arctic Discoverer launched the ROV, which was attached to a control cable, without incident. It took an hour and a half for the vehicle to reach the bottom, 8,000 feet below. After lunch, the crew began to regroup in the control room: John Moore, a tall, lanky, longhaired wrangler of deep-sea submersibles, sat in the pilot's seat; Alan Scott, a software wizard whom everyone called Scotty, took his post at the navigation computers; John Doering, a veteran of many past treasure hunts and the possessor of an expert eye for underwater artifacts, sat to Moore's left in the copilot's seat; Milt Butterworth, a photographer and archivist, stood nearby and began scanning the audiovisual monitors; and Bob Evans, a geologist and data analyst, perched behind them in a chair against the wall, where he could see all of the monitors at once. Tommy Thompson sat in a chair in the middle of the room, his toes touching the carpet and his knees pumping with nervous energy.

The black-walled control room was small, with a low ceiling of acoustic foam. To protect the computers, the temperature was kept chilly — 65 degrees or below. On three sides, stacked floor to ceiling, computers, monitors, and digital displays filled the room. On one TV monitor, they could see the cable still feeding off the big drum out on deck. The only illumination came from the glow of monitors and the readouts on the digital displays. At 1:33 P.M., they had visual contact with the bottom.

The Mesotech sonar swept into the darkness a hundred meters out, far beyond the range of the cameras. A target would show first on the Mesotech display, and then, about three minutes later, the ROV would pass over the target and it would appear on a video monitor in a flood of light.

Half an hour after the ROV began its run, the sonar registered some small targets off to port. Moore called out the heading of the vehicle and the bearing of the targets, and Scotty turned to record those in the navigation log. Suddenly, Moore was calling out more and more targets. “I'm getting sonar action,” said Moore. “We have got some really major action coming here!”

Into the glare of light now glided the first of the smaller targets Moore had seen on the Mesotech a few minutes earlier: three white artifacts in the sediment.

“Cultural deposits here,” said Moore. “Yep, we got a plate, it looks like.”

The big object he had seen on the Mesotech would be coming under the camera on the port side, but just as the ROV reached the target, it started to twist to starboard. Moore talked to the vehicle: “I don't want you to rotate that way. Over the other way, over the other way.”

Suddenly a huge, bulbous shadow began to grow at the lower left corner of the monitor. Someone yelled, “Look at this! Look at this!”

“Whoa!” Doering hollered. “Whoa! Whoa!”

Thompson muttered, “Oh, my God!”

A gigantic paddle wheel lying in silt was nudging into view.

“You know what that is!” Doering shouted.

“No shit!” Moore replied.

They had hardly had time to realize what they were seeing when Bob suddenly called out to Moore, “You better get up right now!”

“I'm pulling up right now,” Moore said.

Evans had studied the sonar image of that site for months. He knew it better than he knew his closet. If his calculations were correct, the ROV was only seconds from colliding with something that cast a long sonar shadow on the sonagram.

Just as Moore raised the vehicle, a forged-iron crank gear the size of a file cabinet suddenly turned white in the glare of the lights not two feet below the ROV. The gear perched at the end of an iron shaft that ran 30 feet up from the center of the collapsed wheel.

“Oh, shit!” Moore cried out. “No, shhhh — “

On the ocean floor, looking like the cage off an antique electric fan, lay the starboard paddle wheel of an old steamer, the iron spokes still radiating from the center and only the wood of its paddles missing. As the vehicle drifted slowly by, the wheel cast a weblike shadow onto the ocean floor that danced in the lights; rusticles dripped from its undersides and sea stars lay draped across its spokes.

Every man in the room knew that no matter how precisely Scotty could tune the navigation system, or how meticulously Evans could render geometry and trig, or how skillfully Moore could fly the vehicle, or how clearly Butterworth could focus and light the cameras, no one ever, ever hit a deep-ocean target on the first track line with a camera. Ever. They had a greater chance of winning the Ohio State Lottery than they did making a shot like that in the deep ocean.

Galaxy II was no longer a test site.

Although the discovery of the sidewheel steamer had drastically increased the odds in Thompson's favor, it still did not prove he had found the Central America. He knew that another side-wheeler had sunk in this region of the Atlantic and that their data showed it might be within the probability area.

With the tech crew staring at the monitors, Moore waltzed the vehicle above the ship. Amidships looked like a country junkyard in winter, quiet, serene, buried in white; like piles of old tractors and old cars surrounded by grayed and weathered fence posts, all under an inch or two of snow. But in these piles were the engines and boilers and water tanks and gears of a once-proud steamer. Her sleekness, her blackness, the yellowed patina of her decks, the broad red stripe running stem to stern along her lower wale had all crumbled and turned to blue ash. Her spiderwebs of shrouds and the majesty of her sail and her real muscle — the enormous steam engines with piston strokes ten feet long — had disappeared or lay in disarray.

By the end of the four-hour dive, calm had settled over the control room, and Thompson was trying to process the implications of what they had seen. Preparing for this voyage, he had examined every possible scenario, but he hadn't spent a great deal of time wondering what he would do if on the first dive they landed on top of cast-iron side-wheels the size of a farmhouse. If this was the Central America, then somewhere on that site lay hundreds of millions of dollars in gold, and that much money made people think crazy thoughts and do crazy things. Only the six men in the room would be permitted to know what they had discovered. And from now on, only the essential technical crew would be allowed inside the control room during a dive.

Thompson, who was born in 1952 and raised in Defiance, Ohio, had always possessed a mechanical bent and a dogged interest in thoroughly figuring things out. Growing up in a family full of teachers and academics, he tinkered endlessly, building telephones, taking apart machines, constructing a working scuba apparatus with a gas-furnace regulator and four propane cylinders. In college, he cruised around in an amphibious car, and he later converted an old diesel Mercedes to run on cooking oil; after graduation he planned to haul two 55-gallon drums full of french-fry oil behind the car and drive coast-to-coast, stopping at McDonald's to refuel along the way.

Thompson had an almost manic intensity. He would discourse to friends on the effects of calcium overdosing, or laser technology, or the latest in group psychology, or the holistic approach to something or other; one acquaintance said it was like listening to someone tripping on LSD, except that Tommy kept track of his pursuits — he always had between seven and 14 personal research projects on all manner of subjects underway at any given time — and he always seemed to know exactly what he was talking about. He would read scientific papers and then call up the author to discuss arcane twists in his work; he perfected a knack for zeroing in on the key expert in whatever field he was currently interested in.

But his central and abiding interest was the challenge of doing work in the deep ocean. He dreamed of building an ROV that he could operate from thousands of feet above: the ultimate Swiss Army knife of underwater technology, with grabbers, saws, backhoes, drills, blowers, cameras, and light booms. Explorers and scientists had been using submersible vehicles for centuries, but in the early 1980s, when Thompson began to mobilize his effort to search for the Central America, the real problem with every system, manned or unmanned, tethered or untethered, was that it couldn't perform significant tasks on the bottom. Thompson wanted to be the first to solve this problem.

All of the practice from his years of ambitious brainstorming and intelligence-gathering was now being harnessed. A potential investor would start to listen to Thompson's proposals and explanations, sure that he was some kind of charming but impractical oddball, and then slowly become convinced that his ideas were sound, his plans rock-solid, and his enterprise an irresistibly exciting one. Confronted with challenging questions, Thompson proved to have anticipated every contingency, and he never oversold his chances for success. He was cautious and discreet — even secretive — in ways that made good sense. As it turned out, he also possessed the bold spontaneity of a warrior and gambler when it was necessary. He had proved that in the summer of 1987, when he faced down several treasure-hunting intruders, modern-day pirates who had tried to grab his site out in the Atlantic. It was never quite clear whether the crews of those other ships had independently focused on the same target area or had simply tried to jump Thompson's claim, but Thompson, utilizing the maritime rules of the road and some daring brinksmanship, had decisively bested them.

Most of all, Thompson had resisted the lure of what he called the treasure-hunter syndrome, the overwhelming temptation to believe the evidence of your hopes rather than the bedrock evidence of scientific research and well-founded probabilities. When everyone on his ship was celebrating the Sidewheel discovery on sonar and declaring victory, Thompson insisted on continuing their sonar survey of his probability map. Some of them thought his tenacity verged on the fanatical, but it was during those final runs that the Galaxy II site was recorded. Thompson's insistence that all the data be exhaustively evaluated had led them past several alluring chimeras and straight to the present possibility that the Central America and her gold were now waiting directly beneath the keel of the Arctic Discoverer.

September 12. They launched again at 8:30 that evening, and the ROV arrived at the bottom a little before ten o'clock. Everyone in the control room was aware of the significance of the date. In the storm-racked darkness of the same night exactly 131 years earlier, the Central America had plunged beneath the waves, plummeted 8,000 feet, and crashed into the seafloor.

At midnight they were hovering above an area they guessed had once been a little fore of the pilothouse. Moore had eased the vehicle to within three meters when Butterworth said, “What's that?”

Everyone searched the gray-and-white landscape of the foredeck but saw nothing except more silt and odd gray lines angling through it.

“It looks like a bell,” said Butterworth.

He walked over to Doering's monitor and pointed to it on the screen.

A gray hump nestled in the silt, something that appeared to be rounded, with perhaps a slight flange. Moore eased the vehicle closer, and they studied the artifact for 20 minutes. Moore tapped lightly at the forward thruster for a gentle wash of water to tumble the silt from the flange. When the scene cleared, Evans and Doering agreed with Butterworth: Not only was it a bell, but it was also inscribed, although it was so mottled they could not read the inscription. If they could bring it to the surface, the bell might tell them if this ship was the Central America.

They had passed into the latter half of September, but the late-season weather held steady at just shy of marginal, with winds bucking up to 20 knots and seas cresting at six feet. Regardless of threatening weather, their continuing exploration of the site had to be methodical: filming and photographing until they had documented the site and knew where to begin their search for the gold.

Then they lost five days to a series of equipment problems and aborted dives. Not until the afternoon of the 22nd did they get the vehicle to roam the site again.

Toward the end of the dive, Moore flew the ROV over the debris field about 50 feet out. As they passed over the seabed, someone noticed a solid geometric shape lying alone. It appeared to be another box, only much smaller than the ones they had photographed earlier. Moore guided the vehicle over and hovered above the box. As he dropped lower, they realized it was a leather suitcase, and sitting about six inches away was a white teacup.

“It looked like a train platform in foggy London,” Doering recalled. “Like somebody was sitting on his suitcase drinking his tea, and his train came in, and he set the cup down to catch the train and forgot his suitcase.”

Moore brought the ROV closer. A white feathery coral rose from the brown leather, and large pink anemones clung to the top and sides. Beneath where the handle had been was a nameplate, but the letters were covered with sediment. Moore trained the forward thruster on the suitcase and spun it a few revolutions for a light wash. When they still couldn't read the letters, he ran the thruster a touch faster, and suddenly the suitcase opened like a clamshell, hung open momentarily — it was packed with neatly folded shirts — and then slowly closed again.

After each dive, either Doering or Butterworth developed the film from the day's shoot, cut the frames into strips, sponged them off, hung them up to dry, and laid the strips on a light table to study with a loupe. The pictures from these still cameras were closer, sharper, more detailed than what appeared on the monitors during the dive. Searching for gold, the crew could understand the site only by analyzing these close-up stills.

Because water bends light, the colors of the rainbow leached out before they reached the cameras. The closer the ROV, the more vivid the color. At varying distances, colors and shapes and textures could all combine to turn lumps of coal and bits of wood into what appeared to be a bottle of wine — or a pile of coins, or a mottled bar of gold.

On one dive, Evans and Butterworth saw a bricklike object lying in the timbers. Moore reached out with the manipulator and caught the brick from the side with the tip of the jaws, nudging it upward until it flipped over, raising a tiny cloud of sediment. A gold brick weighs seven times more than a clay brick and thus would not have flipped over so easily. Hundreds or even thousands of these clay bricks might litter the site, masquerading as gold bars.

The most confusing of all the misleading signs of gold at the site were the casings left behind by tubeworms. As the worms bored into a beam, they created curlicue shavings of calcium carbonate, like shells, six to ten inches long. When the timbers finally broke apart, the white casings dropped to litter other timbers in the wreckage. Then the iron oxide bleeding from hundreds of tons of corroding iron drifted through the water, staining the shavings a yellowish orange. At a distance of only several feet, the bright lights shining on those stained shavings made them glint like collapsed piles of gold coins. The site was covered with them.

On the night of September 24, THEY brought the bell up and examined it in the glare of a floodlight underneath an awning on deck. Patches of pale orange and verdigris mottled the dark gray-green of the bell, but in the band around the top of the bell they could make out the inscription: MORGAN IRON WORKS NEW YORK. Morgan had been the foundry that cast the fittings and the huge steam engines of the Central America. They could also make out the last two numbers of a date: 53. The Central America had first set to sea in 1853. It wasn't gold, but the bell was the next best thing.

On the morning of September 27, despite the seas hitting five to eight feet, they launched the ROV and continued their photographic survey. Butterworth and Doering were uncertain how many shots would turn out, however, because the camera shutters seemed to be sticking and the strobe was often not in sync.

They concentrated on an area where the timbers were sharp and jagged and riddled with holes. As the ROV inched closer, shadows moved in the background, giving Moore a better sense of the third dimension. Doering pointed out a beam that looked almost like the head of a gargoyle. With the vehicle five feet up, Butterworth fired the still cameras, the whole scene brightening in the flash. White flecks of particulate matter floated by. He fired the cameras again, and a couple of shiny spots reflected the vehicle's lights. The cameras were pointed toward a depression where they saw many short, straight, gray lines, like small bricks covered with dust.

The next morning, Doering went to the darkroom to develop the film of the gargoyle shoot. He started with the cassettes from the port-side camera, but when he pulled the film from the chemical bath, cut it into strips, and hung them up to dry, he discovered that the first strip was black. So was the next, and so was the next. The entire cassette had somehow been overexposed. Disgusted with himself and the cameras, Doering went to the galley for lunch.

After he returned to the photo lab and processed the film from the starboard camera, he was relieved to see that there were images on these strips. After the film had dried, he laid the strips on the light table to check the exposure.

For over 15 years, Doering had been looking at things underwater. He had seen artifacts lying on the bottom of the ocean that dated back three or four centuries, back to when explorers and conquerors still plied the Caribbean in ships stuffed with booty from the New World. From societies long past, he had seen tools and jugs and dishes and bowls and canisters and vials and cut glass. He had seen armor and swords and harquebuses and cannons. He had seen collections of jewels and bars of silver. He had seen gold. But nothing had prepared him for what now awaited him on the light table.

With the first strip laid out, he placed the loupe over one of the pictures and peered through the lens at the most incredible sight he had ever seen: stacks of coins and gold bars of every size and shape, just sitting there. He looked at other strips, and there was more, piles and piles.

Doering picked up the first strip, grabbed the loupe, and took the steps two at a time up to the next deck, where he found Thompson and a crew member in the communications room. The door was open, so he walked in and ceremoniously closed the door behind him. He remembered trying to get the words out, but only being able to say, “Wah, wah, wah.”

He handed the strip to Thompson, who held it up to the light. Doering heard Thompson mutter something like, “Boy, I never thought…” Thompson permitted himself a joyous “Yahoo!” and then the three men began to quietly celebrate their triumph.

The gold might be there for the taking, but that afternoon the weather lowered, and for the next three days they couldn't dive as the wind howled and the seas rolled in at eight to ten feet.

On October 1, the ROV returned to the bottom and hovered just above the area they now called the Bank of California. After they had surveyed the scene for a few minutes, Moore eased the vehicle lower. The piles of coins and bars that had shocked Doering lay mostly covered with sediment. Doering had seen gold peeking from beneath the soft whiteness, hard angles of yellow bars askew and large spills of small, round, flat objects with hints of orange and brown and yellow showing through the silt. Along the gargoyle were a hundred or more coins, the ones whose glint had caught Butterworth's eye because they rested in a place where the slow current continually washed them clean.

Moore aimed the forward thruster downward and dispatched a gentle wash of water to blow away the cover material. The sediment was thin, but when the wake of the thruster hit, it exploded upward, swirling into clouds, blotting out the rotted timbers, turning the monitors white. For several minutes the techs could see nothing but the roiling sediment. Then the clouds began to drift with the current and the view on the monitors began to clear, revealing an astonishing vision.

“The bottom was carpeted with gold,” Thompson later recalled. “Gold everywhere, like a garden. The more you looked, the more you saw gold growing out of everything, embedded in all the wood and beams. It was amazing — bars stacked on the bottom like brownies, bars stacked like loaves of bread, bars that appear to have slid into the corner of a room. Some of the bars formed a bridge, all gold bars spanning one area of treasure over here and another area over here, water underneath, and the decks collapsed through on both sides. Then there was a beam with coins stacked on it, just covering it. You couldn't see the top of the beam, it had so many coins on it.”

In one 30-foot-wide pile, gold bars lay tumbled upon one another like the rubble of an old building just demolished. And coins single, coins stacked, coins once in stacks now collapsed into spreading piles, some coins tinted with drifted rust, others with their original mint luster. Except for a tiny squat lobster carefully picking its way across piles of coins, the scene lay perfectly still. Sticking up out of another area was a coin tower, eight stacks of gold coins, 25 coins to a stack, all of the stacks abutting one another like poker chips still in the rack. Nearby was a mound of gold dust frozen ten inches high, dotted with nuggets and capped by two small gold bars

“Isn't that amazing?” Evans said.

“That is amazing,” Doering agreed.

With the cameras sweeping the piles, Thompson watched the monitors as gold eased into view, passed by, and was replaced by even more gold, the scenes brightening again and again with the pop of the strobes. For two hours they took dozens of pictures.

That afternoon, Thompson finally directed Moore to recover the first artifacts. Using a fingertip touch, like he was playing the flute, Moore moved the master arm on his console, and the manipulator on the ROV reached out over a pile of coins, bent downward at the wrist, and paused, almost like it was contemplating the pile; then it dropped carefully, spread its Teflon jaws, closed them, paused again, and lifted a single coin. He set the coin in its own numbered compartment in a plastic tray. Soon they had six artifacts, each a small bar or coin collected at the edge of the scene.

Next, they focused on a pyramid of at least a hundred coins, slowly moving the cameras in closer and closer. On the face of one $20 coin they could see the words “United States of America,” an eagle surrounded by the rays of the sun, 13 stars in a tight oval above the eagle, and a tiny “s,” the mark of the San Francisco Mint.

“Wow,” said Thompson, pointing at a different coin. “Read that date, John.”

Moore moved the camera closer to a coin as pure and lustrous as the day it left the San Francisco Mint. It was emblazoned with the bust of Lady Liberty, her hair crowned with a tiara and cascading in ringlets down her neck, and her ringlets stopping just short of the date “1857.”

The seas continued rough, and they lost another four days to technical glitches and equipment failure. They completed the repairs late in the day on October 5, and overnight the weather improved. At daybreak the following morning, the Arctic Discoverer rocked gently in three-foot blue swells, as close to calm as they had seen in almost two weeks. They had been at sea 39 days.

In the early afternoon, not long after they launched the ROV, the blue swells began rising higher and darkening. Thompson was determined to proceed: This would probably be the last dive of the year, and he wanted to make it count, but as they resumed the harvest of gold in the calm depths below, the swells topside gradually sharpened and turned gray. At the center of the ship, the heavy insulation that kept the control room cool and dry also deadened the sound and feel of the sea. The tech crew could hear only the dull, intermittent hum of the thrusters and feel only the slightest rise as the ship lifted over the incoming waves.

At sunset, Burlingham looked out beyond the far edge of the Gulf Stream and saw dark clouds erupting between two fronts, then arching upward and stretching along the horizon. He estimated that the squall line was as close as 25 miles, and it was sweeping toward the ship. As the last light faded, the Arctic Discoverer began to rock over choppy, darkened waves, her bow lifting higher and higher. Then the temperature dropped, and the canvas awning on the foredeck began to undulate and snap in the wind.

Burlingham radioed the control room. “You'd better bring it up now,” he told them.

Thompson heard the warning but continued to work. He was determined to recover a significant amount of gold, yet do it carefully. By 8 P.M., they had picked up 32 more gold artifacts.

The wind had risen sharply to 25 knots. According to the latest weather fax, the disturbance Burlingham had seen along the horizon at sunset was no squall. It was a huge frontal system with enough energy to double the size of the waves and the intensity of the wind.

He radioed Thompson in the control room again. “This front's going to be on us in two hours,” he said. “We should start recovery operations immediately.”

“We're onto something important,” said Thompson.

“Fine,” said Burlingham. “We're going to be running 10- to 15-foot seas.” Burlingham didn't need to tell Thompson what that meant. Ten-foot seas made recovery almost impossible.

An hour later, the seas had reached eight feet, the wind blew at near gale force, and foam began to streak the cresting waves. At 9:13 P.M., with 40 more gold coins and bars tucked away in the artifact trays, Thompson ordered the team to begin recovery.

At 10:30 the ROV neared the surface, its lights intermittently visible beneath the waves. The wind had intensified to full gale. Crouching on his hands and knees, a deckhand lifted himself over the bulwark onto the hero boards just as the ship dropped away and a 12-foot sea hit broadside. The wave exploded off the ship, but he held tight. Then Burlingham and another crewman dropped onto the boards next to the deckhand as the wave fell away. One after another the big waves rolled along the side of the ship, taking the three men under.

With the ROV just below the surface, Burlingham signaled to Doering, who was operating the crane, to begin swinging the vehicle toward them. As the crane drew near, the three men on the hero boards disconnected the cable and hooked the vehicle to a wire coming off the crane. Then Burlingham gave the signal to start winching in the wire.

The crane was swinging toward the ship, which had begun to roll to starboard. Suddenly there was a loud pop, and the ROV came hurtling at the three men on the hero boards. They threw themselves over the rail and scrambled aft as the vehicle skimmed the water and crashed into the deployment arm. The ROV's electronics sphere exploded and gold sparks showered the deck. The gear at the base of the crane had blown apart, and the crane was at the will of the sea. The vehicle crashed into the deployment arm a second time, and several battery packs burst, spraying oil across the deck.

To try to control the swing, Bob Evans reached up with a boat hook and snagged a tag line coming off the side of the crane. Thompson and the other men grabbed hold and heaved together, stumbling and sliding on the oil-slick deck, trying to haul the line toward the starboard rail. But before they could pull it that far, the ROV began its outward swing, dragging the six of them back across the deck.

Doering was trying to raise the crane high enough so that on its next swing the ROV would clear the rail and he could drop it on deck. It was the worst possible moment for another malfunction, but just then the crane's motor died. The crane was locked in its bent position with the swing gear gone, and the 5,000-pound vehicle was slammed over and over against the side of the ship. In desperation, Doering jumped from the seat and grabbed onto the tag line with the others. Moore saw what had happened and disappeared into the forecastle, where he found the circuit breaker for the crane and succeeded in restarting the motor. When he rushed back into the storm, he saw Doering and the others playing a losing game of tug-of-war with the tag line across the unnatural brightness of the night deck.

Moore jumped onto the crane seat, but he had never operated the crane, and when he looked at the panel of levers with black knobs he saw that the labels had rubbed off. It was impossible to tell which lever would lift the ROV up over the side of the deck. Taking a wild guess, Moore jammed the lever he thought might control the fulcrum. He had guessed right: The crane began to straighten, raising the ROV as it swung back toward the ship.

Thompson and Evans stood five feet from the starboard rail, still tugging at the tag line, still trying to coax it around a cleat on the starboard side. They looked up and saw that the ROV, twisting wildly on the boom, had skipped over the port gunwale and was flying across the deck. The two men scrambled out of the way just before the vehicle swept the space where they had stood. Then it twisted at the end of the boom and began its swing back across the deck, over the gunwale, and back out across the water. Finally, as the vehicle came in once again, Moore slammed a lever, dropping the whole boom down on top of the vehicle. The ROV slammed into the deck, skidded back and forth, and crashed into the base of the deployment arm, where at last it came to rest.

They were exhausted and shaking and laughing with relief. Much of the ROV had been destroyed, and the artifact drawer was smashed closed, but they had the vehicle safely aboard. It took them three more hours to drag it to its parking place and anchor it with half a dozen lines.

At 3 A.M., Burlingham took the bridge and set a course in seas running 20 feet. With blue water coming over the forecastle and the vehicle strapped tight to the deck, the Arctic Discoverer traveled through the storm at eight knots, headed due west for Wilmington.

One of the more remarkable accomplishments of the Columbus-America Discovery Group is that from October 1988, when Thompson informed his 160 partners in a letter that he had found gold 8,000 feet beneath the sea, until August 1989, when the group finally announced to the public that it had recovered the treasure of the Central America, the news never traveled outside the partnership.

The partners' commitment to secrecy allowed Thompson and his tech crew to concentrate on refining their technology and returning to the site for further exploration and recovery. After much thought and conversation with his lawyers, Thompson decided not even to file a claim on the Galaxy II site, but to let the matter rest until he was ready to return the following summer. Everyone already assumed that the Columbus-America team had commenced recovery operations on the Central America at the old site. Why announce to the world that the treasure really lay somewhere else?

Before they returned to Galaxy II, Thompson had one more problem to solve. As rare antiquities, the Central America coins were worth far more than merely their weight in gold. Even the smallest nick or scratch on a coin can reduce its value by two-thirds, and a second or third blemish can drive the price even lower. Thompson's problem was that so many coins lay piled at the site that meticulously plucking them up one by one would be too expensive and time-consuming, yet to recover two or more at a time might mar their finish.

That winter Mike Milosh, an engineer from the Battelle Memorial Institute who was working for Thompson as a consultant, came up with a seemingly far-fetched but ingenious solution: Place a mold over a pile of coins, inject silicone into the mold, let it harden, and raise the whole mass in a rubber block. You could pick up a hundred coins at once, and the silicone would envelop the coins and protect them.

On July 20, 1989, the Arctic Discoverer returned to the Central America. The crew placed a mold over one pile of gold after another, cleared artifacts around the edges, and injected the silicone. Later, they returned to recover each of the gray blobs. In one block of silicone, they recovered 15 stacks of $20 coins some 30 coins high, or about 450 perfect double eagles. Night after night the vehicle came up filled with gold bars and gold coins, nuggets, gold-veined quartz, even piles of sparkling dust.

The crew continued to dive at the site into mid-September, when Hurricane Hugo chased them into port at Wilmington. After Hugo they returned to sea, but the weather remained the worst they had encountered. People calling from shore on the SatCom phone could hear the crash of falling dishes and books. Storm after local storm spun across the water and hit the ship, until Thompson finally decided it wasn't going to get any better that year. It was time, he thought, to take the treasure home.

Back on shore, Thompson had retained the services of James Lamb, head of the coin department at Christie's. When Thompson first showed him one of the 1857 $20 gold pieces, Lamb was dumbfounded. “A coin expert can go through an entire career and see about two of these coins in this condition,” he declared. One 1857 $10 eagle gold piece recovered from the Central America was, he said, “by a considerable margin the finest example known to exist.” Other coins had been struck by one of the small, private mints that existed before the San Francisco Mint opened in 1854, and these, too, were extremely rare.

Lamb was equally amazed by the bars. Most California Gold Rush bars were melted down for coins in the 1850s and 1860s, making them exceedingly rare today. The bars recovered from the Central America ranged from five ounces to over 900 ounces, and there were hundreds of them. Moreover, each came adorned with a unique set of symbols and numbers: In one corner appeared a shiny cut, where the assayer had taken a small sample to determine the gold's purity and kept the sample as a fee. Then the assayer had stamped the bar with his seal, recorded the “fineness” or purity (e.g., “891 fine,” meaning 89.1 percent pure gold), assigned the bar a serial or “identification” number, and given its weight in ounces. Based on the fineness and the weight and a value of pure gold at $20.67 an ounce, the last number on the face was the value of the bar in 1857. Besides the unique markings on each bar, California gold contained silver rather than copper, which gave the bars unusual brilliance and luster.

One of the largest bars was no. 4051, Justh & Hunter, 754.95 oz., 900 fine, $14,045 value in 1857. Today, in bullion value alone, the bar is worth almost $250,000.

The Columbus-America Discovery Group has returned twice to the Central America and has now recovered much of the gold. In 1989, 39 insurance companies filed a lawsuit in federal court, claiming that they had covered the loss of about 15 percent of the gold when the Central America sank in 1857 and therefore that a significant portion of the treasure was theirs. In 1996, after extensive litigation and a series of appeals, Columbus-America was awarded ownership of more than 92 percent of the disputed gold, and 100 percent of the other 18 tons thought to have gone down with the Central America. It was by far the largest salvage award in the history of admiralty. Until the gold is sold, no one knows how much it will be worth, but the total will certainly be hundreds of millions of dollars. Some have surmised it could eventually be a billion or more.

Meanwhile, Thompson has put his technology at the service of more than 150 scientists, historians, and other scholars around the world. Working with Columbus-America and its ROVs, biologists have identified 13 new species. On one of the return visits to the Central America, a 22-foot Greenland shark swam past the cameras; it was many times larger than any previously sighted in the deep ocean.

Tommy Thompson, who personally owns about a third of the gold, continues to consider how best to market and sell it. He spends most of his time and energy refining and developing new deep-water technologies, but he is not yet prepared to reveal what these new technologies are. His admirers believe that he is destined to become celebrated for many other accomplishments and that his discovery of the greatest treasure in history was only the beginning.

Gary Kinder lives in Seattle and is the author of three books. This article has been adapted from Ship of Gold in the Deep Blue Sea, to be published this month by Atlantic Monthly Press.

Copyright 1998, Outside magazine