Skip to main content

Ship of Gold

  • Author:
  • Updated:
Illustration by Andy Warner

Illustrations by Andy Warner

Ship of Gold in the Deep Blue Sea

In this excerpt from his 1998 book, author Gary Kinder takes us aboard the Pine River with her treasure-hunting crew in search of a sidewheel steamer laden with gold and lost in a storm off North Carolina in 1857.

THE STEEL BOW of the Pine River shot upward eight feet, then slammed down, shuddering all the way to the fantail. Then the bow leaped up and slammed down again. A forty-knot wind whipped the sea white beneath lightning tearing at the night sky. When the stern dropped, water exploded over the fantail and rolled a hundred feet up the flat deck to crash against the control room, where the sonar techs sat trying to figure out what was wrong with the SeaMARC. They could look out a small window cut into the steel and see the wall of water headed up the deck.

“At times,” said sonar technician John Lettow, “it seemed as if there was ten foot of ocean and you were under it.”

The Pine River was a flat-bottomed mudboat from the Louisiana oil patch built to ferry drilling mud and supplies out to the oil rigs in the Gulf. Tommy had found her in a shipyard in Orange, Texas, while scouting with a former navy commander named Don Craft. In his late fifties, Craft had retired after thirty years in the navy with an Unlimited Master’s ticket: He could skipper any vessel in any ocean. In late 1984, Tommy had called Craft because the commander now consulted for offshore operations and he knew which vessel and what equipment Tommy needed to run a SeaMARC search in deep water two hundred miles off the coast.

Editor’s Note:Ship of Gold in the Deep Blue Sea (Atlantic Monthly Press, 1998) by Gary Kinder is the tale of a problem-solving Ohioan named Tommy “Harvey” Thompson who took up the challenge of finding sunken treasure on the high seas with a logical, methodical
approach. From historical data, he and his team determined a ship named S.S. Central
sank with more than 21 tons of gold aboard. They partnered with Columbus, Ohio-based investors to finance their efforts for a share of the impending treasure, and they overlaid a swath of the big wide ocean with a probability grid showing the odds the ship might be
located in each two-mile square. We join the story as Thompson and company commission a deep-sea exploration vessel and put to sea with a team of cutting-edge sonar operators who have brought along their one-of-a-kind towable sonar sensor known as SeaMARC.

At first, Craft was leery. But Tommy sent him a check for his fee, and the check cleared, so Craft met Tommy in Houston. For four days they drove, talked, ate seafood, and stopped at every bayou shipyard from Orange to Jennings to Lafayette, from Cameron to Patterson to Houma, showcase spots along the Gulf where the offshore support industry displayed its rustbucket mudboats for charter. Tommy and Craft ventured out on scaffold piers in search of one that could be sucked from the muck, sandblasted, overhauled, and refitted for a deep-water survey of the Atlantic Ocean.

For the first day and a half, Craft wondered if Tommy would ever stop talking. “He asked me every goddamned thing you can think of, on every subject you can think of,” said Craft, “vessels, ROVs, operational techniques, equipment, shipping companies, methods used in the Gulf, how seafarers used to do things and what problems they had. We covered it all during that four days.”

Illustration by Andy Warner

In Orange, Texas, they found the Pine River, which Tommy liked because it had a helo-deck. Under the helo-deck was good control space, a small shop, and some storage, which appealed to Craft. At 165 feet, it was smaller than what Craft had envisioned, but he was satisfied.

Craft then had ripped out a lot of old equipment from a previous charter, got the vessel cleaned up for transit, measured the fuel on board, and topped it off at sixty thousand gallons. When he left Orange, he had the Pine River on charter, beginning the 14th of May and running through the end of July. Eight days later he arrived in Jacksonville, ready to place the tow point and the winch, weld a modified log boom to the deck for launching and recovering the twelve-hundred-pound SeaMARC, fire up the galley, fill the ship with groceries, and await Mike Williamson and the sonar techs for mobilization.

TWENTY-TWO MEN LIVED on the ship. Six men from the Louisiana bayou kept it clean, running, and pointed in the right direction. A cook ran the galley. Don Craft oversaw the operation. Twenty-four hours a day, Williamson and his sonar crew of eleven manned the electronics in the control room. Bob [Evans] had remained in Columbus with the handle “Info Bob,” a source for additional information. Barry [Schatz] documented the search on film and video, helped Tommy write letters to the partners, and was responsible for ship-to-shore communications to assure a steady flow of supplies, spare parts, and information. Tommy was Williamson’s client, and as the client he was supposed to watch and listen. But he had $1.4 million in his pocket from partners who counted on him, and he would no more leave Williamson alone to run the SeaMARC survey at sea than he would have left Larry Stone alone to produce the probability map.

That spring, Tommy had called Stone frequently to question the assumptions they had relied upon to create the map. He would ask Stone, “How critical is that assumption? Can we get better data on that? How would we go about improving it? If the hurricane that hit the Central America was actually 110 knots instead of 78, how would that affect the eventual position of the sinking?”

“He just was so detailed and careful about what he did,” said Stone. “He would keep revisiting parts of the analysis, trying to poke holes in it, seeing where the soft spots were, and ‘Let’s see if we can fix ’em before we go to sea.’”

Tommy wanted to explore the chopped-down mast, the dumped anchor, the drag sail, anything that might affect the ship’s drift. He wanted to tighten their assumptions about wind and current. He wanted Stone to investigate the accuracy of nineteenth-century sextants and chronometers. He told Stone to call experts. Just get them on the phone, he said, they’ll talk to you, and Stone was surprised at how well it worked. “It only takes about three or four phone calls to find an expert about any subject in this country,” said Stone, “and that was something he was always instructing me to do.”

A former Coast Guard officer told them that celestial readings taken with sextants and chronometers of the mid-nineteenth century would be off by no more than four nautical miles. Another expert at the University of Miami advised them on ocean currents. A meteorologist at the Naval Postgraduate School created a computer model to estimate how hurricane force winds might affect those currents.

Illustration by Andy Warner

The probability map Stone presented to Tommy was neat and precise; specific little numbers in perfect little cells, each representing a two-mile square of ocean. At the end of each line forming the grid, top and side, Stone had included the latitude and longitude down to the nearest minute; within most of the cells appeared a number from 0 to 73, indicating the probability out of a thousand that the ship would be located in that two-mile square. The highest cell had a 7.3 percent chance of containing the shipwreck site; the many cells marked zero had some probability, but less than one chance in a thousand.

Mike Williamson and a colleague had then taken Stone’s probability map, factored in topography, weather, and the velocity and direction of currents, and designed a search map, the most efficient way to run track lines back and forth through the highest-probability cells on Stone’s map. They laid out each track on a grid with finely tuned coordinates and direction of sail. But, as they were experiencing in these first few days of the search, the currents never ran true and the equipment never operated without flaw and the ocean never lay still.

THEY HAD LEFT Jacksonville just after midnight on June 3, the Pine River bucking ever higher waves and rising wind. On the morning of June 4, they were in hilly water near the start of track line 1, two hundred miles at sea, ready to begin the search for the Central America. But two hours into the search, they had to abandon the track line and recover the SeaMARC: The navigation didn’t work, so they couldn’t tell where it was or how high it flew, and if they didn’t know where the SeaMARC was and how high it flew, they would have no idea where the shipwrecks were when the SeaMARC saw them, which was the reason for being out there. By that evening, they had the guts of the SeaMARC spread across tables in the control room. Outside, the weather continued to climb.

“It was a hurricane,” said Lettow, “and then it was downgraded as it moved offshore out of the Gulf Stream. But there were storm seas and lightning strikes virtually everywhere around us.” The eye of Tropical Storm Andrew blessed them late one night with fifteen minutes of calm water and clear skies before the wind and the sea built again, buffeting the Pine River, her after deck, according to the log, “frequently awash.”

The techs took advantage of the bad weather to test one system after another, but they couldn’t even agree on the problems.

One said, “There’s a missing logic gate; we got to put this gate in.”

Another thought, “The system worked before without the gate. It’s the whole logic process that’s screwed.”

A third wanted to throw the first overboard for distracting the second, who might have a solution.

“Hours are going by,” said technician Will Watson, “days are going by, and you’re just watching this.”

Tommy was always there, watching the techs, trying to work with them, seeing hours and then days spin by on the clock. “Mike was trying to let the techs work on it,” said Tommy, “and I’d talk to the techs and try to help them think through the logic and diagnostics, and so I would be included in the circle. Mike’s telling me, ‘They’ll get it, they’ll get it,’ but when I’m working with the techs they’re saying, ‘We don’t know this part of the circuit. If we go in there, we could damage it and not ever get it fixed.’”

One night as Tommy looked at the SeaMARC electronics spread across the control room and listened to the techs trying to figure out the problems and felt the pitch and roll of the ship in storm seas and looked back over an entire week of great frustration with nothing accomplished since they had set sail, he said just loud enough for Will Watson to hear, “I was afraid of this.”

Tommy was dealing with men who knew more about electronics and sonar than he did, but they had never used electronics or sonar to find a wooden-hulled shipwreck in the deep ocean. No one had. And they hadn’t spent most of their lives studying ways to solve problems. “Typically,” said Tommy, “they just took one path at solving a problem, and then they’d get to the end of that and go, ‘Okay, that didn’t work, now what do we do?’ and they’d go down another path.”

Tommy would ask the techs, “How long before you think you have it apart and figure out if that’s the problem?” Someone would give him an estimate, and Tommy would say, “What if you can’t fix it?”

They’d say, “Oh, then we’ll do this.”

And Tommy would ask, “Can’t we be doing that right now? We’ve got eight hours here and not everybody can be working on the fish. Can we be making phone calls? Can we be getting information? Can we be ordering integrated circuits? What integrated circuits do you think? Are there five chips that are suspect? Why don’t we order those now?”

If the techs working on the problem were wrong after eight hours, at least the others would be far along a parallel path. And if they were right, they could always cancel the parts they had ordered or send them back. If a supplier charged them a few bucks, so what? Tommy was spending twenty thousand dollars a day on the best sonar experts in the world, and until their sonar worked they were surveying nothing.

The afternoon of the 8th, they redeployed the SeaMARC and began towing again along track line 1. The weather that had moderated slightly since the eye of Andrew passed over two days earlier began building back up, the seas to six feet and the wind to twenty knots. Trying again to tow along those neat little track lines on the search map, the Pine River fought strong currents and heavy seas, sometimes crabbing almost sideways. Once, a blast of lightning cracked down to the water so close to the ship it blew out a piece of cable that carried the signal from the submerged SeaMARC to the ship topside.

Tommy asked Lettow, “What do you think?”

Lettow said, “I think we should have gone back to the beach and drank beer three days ago instead of mucking around in this.”

WHEN CLIENTS CALLED Mike Williamson, they wanted him to find something on the bottom of the ocean. They told him where they thought it was, then turned Williamson loose to find it. That was their role as clients. In Tommy’s mind, his role as client was “to help direct when I thought they weren’t thinking clearly about the problem,” and he pushed Williamson to rethink everything. If Tommy was up in the middle of the night and Williamson was napping in his bunk, Tommy got him up. When Williamson’s crew started scratching their heads over what was wrong with the SeaMARC, Tommy got on the satellite phone at ten dollars a minute and called the guy who created the SeaMARC.

Don Craft had skippered several operations where Williamson was the sonar expert. He knew that when Williamson and his crew came on board, the sonar responsibilities would be handled as efficiently and professionally as possible. “Mike Williamson,” said Craft, “is a world-class sonar operator.” And Williamson had handpicked his crew after years of working with the best. “We brought to the party,” he said, “what we thought was the best team in the world.”

But Tommy was thinking, I have a hundred partners who have bet a lot of money that I know how to find and recover a shipwreck at depths no one has ever worked in before. My first test is to image the shipwreck on the SeaMARC 1A. I had forty days at sea to get that image. Eleven of those days are now gone, lost to weather, towing foul-ups, navigation problems, and other malfunctions in the SeaMARC, and all we have is one track line from a lurching tow fish constantly being dragged off course.

Ted Brockett, who had designed the sled for the tow fish, understood Tommy’s concern, but he sympathized with Williamson. “Harvey was always underfoot,” said Brockett. “If something wasn’t working right, he’d get in there with a calculator and a pencil and he’d be designing, calculating the stresses. He was down to the nuts and bolts on everything, so I spent a fair amount of time soothing egos and trying to keep these two guys calmed down. Nobody was throwing punches, but the discussions got fairly heated, and on a small boat when you can’t get away from each other, those things tend to build.”

OFFSHORE NO ONE functions at l00 percent. Williamson figured you could take anyone who performs well on the beach and put them in a small ship at sea and their productivity would drop by 90 percent. Ships the size of the Pine River pitched and rolled. They were powered by diesel engines, which were noisy and belched fumes that filled the head when the head already felt light from the rocking of the ship. Anyone who said they never got seasick was lying; seasickness incapacitated some people, greatly reduced the abilities of others, and dropped the productivity in all. Every crewman had a wave train out there somewhere with his name on it. A storm like Andrew produced enough wave trains with enough frequencies to nauseate the entire crew. When you haven’t slept well for days, and the engines are groaning and causing the hull itself to vibrate, and the very place you plant your feet is slick steel constantly in motion, and the entire space you have is much less than half the size of a football field, and crowded into that space are twenty other men, at least half of whom you’ve never seen before, and you’re trying to get your work done in a sailor’s three point—two feet on the deck, one hand holding on—you get to where you just can’t tolerate certain things.“Little insignificant things,” noted Williamson. “Like you can’t stand the way somebody ties their shoes.”

AFTER THE WEATHER cleared and the swells and the wind subsided, they finally had the SeaMARC calibrated and tuned and the two crews coordinated on the towing. 

Mowing the lawn, they called it: moving through the water at one to two knots, back and forth, overlapping each track so as not to miss anything, thirty miles down, five hours to turn around, crab over twenty-five hundred meters, then thirty miles back. The tow fish flew a few hundred meters above the ocean floor, its ears listening for a sonar return, a bounce of its sound waves off a solid object anywhere within its range of five thousand meters.

Enclosed at the head of the back deck, the control room was filled with racks of electronics, recorders, and computers. Williamson ran two towers a day, the day tower from noon to midnight, the night tower from midnight to noon. On each tower was a pilot who controlled the altitude of the tow fish by winding in or paying out on the winch; a navigator who monitored the ship’s navigation computers; a sonar operator who watched the charts rolling out of the recorders; a SeaMARC technician who recorded each change in the settings; and a watch leader, who had to ensure that no one got so involved in his own little project that the fish crashed into a mountain or the seafloor. It had happened before.

As the SeaMARC passed slowly over the ocean floor a mile and a half beneath the surface of the sea, five graphic recorders, looking like large IBM Selectric typewriters, ran full time, the sonar operator continuously tuning them and changing the paper and the stylus belts. 

When the SeaMARC had passed a target, they radioed the bridge for the ship’s speed, plotted its position, checked the altitude of the tow fish, checked the gain on the signal, read the cable gauge for how many feet of tow line they had out, and recorded all of it in the logs. Later they could return to that spot for a closer look.

“No target acquisition data has any validity unless it’s repeatable,” noted Craft. “You have to be able to go back to the same spot; otherwise, you haven’t learned anything when you see a target.”

WHEN WILLIAMSON CAME on the night tower, he analyzed the data from the previous twelve-hour shift. No one questioned Williamson’s eye. He could see things in a sonar return no one else could see. As he reviewed the strip charts, he would call out targets for the techs to bring up on the computer screen, and they would play with the image: blow it up, filter it, alter the colors, measure it. The strip charts still were the first stage of elimination, where they could approximate length and width; where they could draw inferences about wood and steel. But with the computer they could see things in a target no one had ever seen.

Williamson dismissed most of them because they obviously were too short. He eliminated most of even the larger targets for being too round, too hard, or apparently some form of geology. The few left became key targets; Williamson called this shortlist the “hit parade” and every target on it had a reasonable possibility of being the Central America. He next arranged the hit parade in order, according to each target’s resemblance to the models Tommy had developed of what the wreck of the Central America would look like after 130 years on the ocean floor. And from the moment he saw it, one stood out, even on the strip chart: It appeared to be a sidewheel steamer, resting upright on the bottom, a dark-humped shadow amidships, indicating paddle wheels. Williamson designated the target “Sidewheel.”

They had already imaged so many targets in the high-probability cells that Williamson now wanted to shut down the broad-swath survey, drop the SeaMARC to a lower altitude, tighten the swath from five thousand to one thousand meters, and go back for a much closer look at the good targets, especially Sidewheel. With the swath at five thousand meters, all they could tell was that something large sat on the ocean floor and stood out in contrast to its surroundings; at one thousand meters, even images produced by sound would look almost like a picture.

Williamson realized the irony in what he wanted to do: He was always the one to insist they complete the broad-swath search before they looked closer at any of the targets, because the target that appeared promising on sonar might not be the target they were looking for.

“We had made a big issue of this with Harvey,” said Williamson, “that the search plan would be followed. This is the bible for this cruise, and we will not deviate from it.”

Now Williamson was the one suggesting they deviate from the plan, and he had good reason. The weather had finally died down, and the one-thousand-meter search required these smoother conditions. They could run the five-thousand-meter search in much rougher weather. “It was my judgment,” said Williamson, “that we should take advantage of the good weather and do the stuff that we could do only in good weather, and then when the weather got bad again, we could go back to doing the stuff we could do in bad weather.” All they had to do was flip switches topside to convert the SeaMARC to the high-resolution mode.

Tommy saw another irony: The world’s best sonar technician; who had once accused him of being a treasure hunter, now seemed himself to be succumbing to what Tommy called “treasure hunter syndrome.” The SeaMARC had flown through every high-probability cell on the map, and right at the juncture of a quad of high-probability cells they had imaged Sidewheel, a dark, pencil-shaped target with little humps in the middle. As tempting as it was to think that the Central America was already safely imaged in their computer and now lay waiting to be explored. Tommy wanted to avoid the mind-set of the treasure hunter, that every promising clue was the thing itself.

JUST AFTER MIDNIGHT on June 30, they attempted to run a second high-res shot of a good target near Sidewheel, but weather forced them to abort the line not a half hour into the run. They swung around and realigned, and on the third run they imaged the target only seventy-five meters to port. Williamson studied the strip chart and the computer returns, and before they had even ended the line, he recorded in his log, “S/wheel still best bet.”

They returned to Sidewheel and ran another line, and Williamson continued to study that target. He recorded in his log, “Search team recommends deployment of transponders for camera runs.” He was even preparing for the camera work, when Tommy interceded, and thirty minutes later they had raised the swath width of the SeaMARC back to the full five thousand meters and were proceeding toward a track line along the eastern edge of the search map, “against recommendations of MEW,” wrote Williamson. Tommy wanted to look into every probability cell that contained a number higher than 0; to the west were several with numbers from 2 through 8, and more to the east had numbers from 1 to 7, a few chances out of a thousand that the Central America lay beneath that two-mile square of ocean.

Now Tommy had at least one and often two or three high-resolution images of every major target they had located. He calculated they had now covered enough of the map that with only three or four days remaining, the time would be more wisely spent either running closer looks at the promising targets or trying to drop a camera on Sidewheel. He directed them to quit and sail for Sidewheel to begin laying the navigation grid for the camera runs.

“We ultimately prevailed,” said Lettow, “and we put the video camera down on that shipwreck.”

AT ABOUT SIX HUNDRED FEET, the deep ocean squeezes out the last particle of natural light, and below that everything moves in blackness. The SeaMARC flew through that blackness, shooting sound waves at the deep-ocean floor, and the system recorded the behavior of those sound waves as they encountered hard objects. Then it converted that behavior to information the eye could see: squiggles on the chart recorders and pixels in a color mosaic on the computer display. Even when they dropped the SeaMARC closer to a target and tightened the swath, they still could see only converted sound waves, not the object that bounced them back. That’s what was so frustrating about sonar: You couldn’t see the thing itself.

On July 4, they retrieved the SeaMARC and left it on deck. For the next two days, they set up a subsea navigation grid, repaired the cable, and worked on the video system. They launched the SeaMARC again to run it over Sidewheel another six times, until they thought they could find it with a camera. Then they deployed the camera sled early on the morning of July 7. By 5:00 a.m., the camera trailed in the bottom darkness nearly nine thousand feet below.

They had clamped the lights, a video camera, and a still camera to a hydrodynamic sled, but the cameras were stationary, aimed straight down, with no thrusters to position the sled and no pan and tilt to control the cameras. All they could do was hope that their navigation readouts were accurate, that they could relocate the site and drag the camera sled across the center of it, and that the remains of a ship suddenly would glide into view.

To see any detail, the camera had to be within twenty feet of the object—ten feet was better—and if the camera was farther away than thirty feet, they could see nothing at all. As the pilot watched the monitor, he paid out on the winch when the boat heaved up and hauled in on the winch when the boat heaved down, trying to keep the sled flying at a steady altitude. If he let the camera rise higher than thirty feet off the floor, they saw nothing on the monitor but backscatter—“Just like looking at a fuzzy white TV,” said Lettow. To get the camera back in range so they could see the bottom, he had to pay out on the winch just a tap without going too far and crashing the sled into the floor. Lettow likened it to playing an intense video game.

Within an hour, Lettow piloted the camera over anchor chain, huge iron links in the sand trailing off into the darkness. Then a ghostly image of what appeared to be timbers crossed the monitor. They realigned the ship for another drift, and two hours later they again saw a piece of something. And again an hour after that. Then they saw what they thought was debris and a few minutes later something that looked like a wood beam. But in these passes they could never tell what they were seeing or where they were on the site.

Each time they thought they saw the ship, the pilot began firing the still camera to shoot as many frames as he could, but every time he fired, the video monitor would go blank or suddenly fuzz with static. When they recovered the camera sled to load fresh film and tape and replace the batteries, they discovered that out of the four hundred still frames they had shot, only one hundred had been exposed, and most of those were either underexposed or double and triple exposed because the frames had not advanced.

The second night into the close-in camera work, Lettow was asleep in his bunk when Watson woke him up. “You got to get up here,” said Watson.“These guys just aren’t keeping it together.” The atmosphere in the control room had gotten so tense for the night tower that Watson and the other pilot had started picking at each other, arguing about why they couldn’t find the ship and how each was doing his job, and finally, the pilot had said to hell with it and walked out.

Lettow dressed and returned to the control room and started running the system again, as Tommy and the techs talked and watched sea cucumber trails scroll by on the monitor. The trails reminded Lettow of a dry desert lake where motorcycles had crisscrossed in every direction. In the middle of a long silence, he said, “What are the chances of us picking up that anchor chain again in the middle of all these sea cucumber trails and then finding a shipwreck attached to it?” About five heartbeats later, big and dark and right in the middle of the monitor appeared the anchor chain and Lettow had the camera running parallel. 

“JEEEZUS!” yelled Lettow. “Look at that!”

Watson yelled, “Check it OUT!

Everybody was watching the monitor now, including Tommy. The camera followed the chain for about a hundred feet, and the chain ran right up to the bow. Lettow already had started raising the sled so he wouldn’t crash it into the ship, and it flew up and over the side, and then he eased it back down.

Half the techs in the room started yelling at once. “Get it out of there! Get it out of there! You’re going to get it hung up!”

Lettow was already hauling in on the wire, and the sled lifted, but they still could see across part of a deck to what looked like cable shrouds from a mast and some sort of superstructure. As the camera glided forward, the decking appeared to be collapsed, and Lettow thought he saw cargo down in the hold. Then they saw the other side approaching, much of the gunwale eaten away, only jagged posts sticking up, and the techs yelled again to get the sled up or they were going to lose their camera system.

“I flew it up over that,” said Lettow, “and you could see the seafloor again. Then we drifted away.” The whole scene had lasted less than a minute.

Lettow started singing. “Chain, chain, chain ...”

And the other techs raised their hands, and started dancing and singing with Lettow, all of them rummy from six weeks at sea.

“… chain of foo-ools.”

Beat, beat.

“Chain, chain, chain …”

Nobody knew the rest of the words, but it didn’t matter. Then, as if they had just discovered how magicians pulled rabbits out of hats, one tech said, “What are the chances of us seeing,” and he looked quickly at the monitor, “some engine works?”

Then another, “Wouldn’t it be nice to see a ... paddlewheel?”

And another, “How about a … treasure chest?”

“We were getting out there on the reality mode,” said Lettow. We were all kind of delirious at that point.

But the magic never worked again. The weather had worsened since they began the camera runs two days earlier, the sea chopping into bigger and bigger hills, the wind rising. Williamson had to get back; he had commitments with oil companies. Besides, the camera was not right, the lighting was not right, the Pine River was not right. Everyone could see it. They had reached the limit of their capability for that summer.

This article originally appeared in the January 2016 issue of Power & Motoryacht magazine.