Three hundred pipe joints, and one in seven of the ones they tested was bad.

April 9, 1963.
43.0803° N, 70.7402° W

The Navy wanted a submarine that could dive deeper than anything it owned and run quiet enough to sit in the water and listen for Soviet boats. It got one at Portsmouth Naval Shipyard, on an island in the river between Kittery, Maine, and Portsmouth, New Hampshire. The keel went down in 1958. The hull went into the water in 1960, and she was commissioned in August 1961. Her name was Thresher, after the shark, and the whole class that followed was named for her. Her pressure hull was rolled from a high-strength steel that let her work at roughly twice the depth of the fleet boats then in service. Her machinery sat on rafts and rubber mounts so the sound of it would not carry out into the ocean. She was the best attack submarine anyone had built.

Everything that made her useful also made her harder to keep alive. A submarine at depth is a sealed steel cylinder with holes in it. Seawater is pumped in through piping to cool the machinery and pushed back out again, and every joint in that piping holds back the ocean at a pressure that climbs the deeper the boat goes. Thresher's seawater systems had hundreds of joints. A great many of them were silver brazed rather than welded. Brazing is faster, and it can be done in spaces too cramped for a welder to get his hands and his hood into, and on a boat packed as tightly as she was, that mattered. The strength of a brazed joint depends entirely on how far the filler metal has flowed into the gap between the pipe and the sleeve around it. From the outside, a joint with full penetration and a joint with almost none look exactly alike.

She came back to Portsmouth in July 1962 for her post shakedown overhaul, the long yard period where a new boat's problems get sorted out. It ran about nine months and covered thousands of separate work items. During it, a portion of the silver brazed joints in her seawater systems was checked with ultrasonic equipment, which can see into a joint without cutting it open. About one in seven of the joints tested came back below standard. The equipment was new. The men reading it were learning it. The boat had a date to keep.

Lieutenant Commander John Wesley Harvey took command that January. He was a nuclear officer from the first generation of them and had served aboard Nautilus. Under him were fifteen other officers and ninety-six enlisted men, most of them young, many of them married and living in the towns on both sides of the river.

On the morning of April 9, the boat took on more people than her crew. Seventeen civilians came aboard for the trip: shipyard engineers, technical men from the yard's design and testing side, and representatives from the companies that had built equipment installed in her. They were going along to watch their own work perform at depth and take notes on it. It was a day at sea and then another day, and then home. One hundred twenty-nine men went out through the river mouth that morning.

She met the submarine rescue ship Skylark offshore. The plan for the trip was straightforward: run out to deep water about two hundred and twenty miles east of Cape Cod, make a shallow dive first to check the boat, then take her down in stages to her test depth, with the Skylark holding station on the surface above and the two of them talking over the underwater telephone. Thresher made her shallow dive on the ninth, came back up, and ran on the surface through the night.

The deep dive began early on the tenth. It was done in steps. The boat descended, leveled off, held, and the crew checked the hull and the fittings for leaks and for the small distortions that come with pressure. Then down another increment. Every so often the Skylark asked for a depth and got one. The messages came through the water as voices, flattened and metallic but readable. The boat reported her course and her depth and that she was proceeding to test depth. On the surface the Skylark plotted her and waited.

A few minutes after nine, a call came up from the boat that was hard to follow. Then a clearer one. Thresher reported that she was experiencing minor difficulties, that she had a positive up angle, and that she was attempting to blow. Behind the voice, the men on the Skylark could hear the sound of high-pressure air roaring through the boat's ballast lines. Another transmission followed and could not be made out at all. Then there was a noise in the water that the sonarman later described as a ship breaking apart, and after that the water was quiet.

The Skylark called and got nothing. She dropped hand grenades over the side, the prearranged signal ordering a submarine to surface, and the grenades went off and nothing came up. She kept calling through the morning and into the afternoon. Aircraft came out. What was on the surface by then was an oil slick and pieces of cork insulation and bits of yellow plastic, the kind of debris that comes out of a submarine when a submarine comes apart, and nothing else.

At the shipyard the phones started early in the afternoon. Wives came to the gate and then into the yard, and the Navy put them in a building and told them that contact with the boat had been lost and that the search was continuing. It continued through the night. The next day the Navy announced that Thresher was presumed lost with all hands. The President ordered flags to half-staff. Portsmouth and Kittery buried nobody, because there was nobody to bury, and held services anyway.

A court of inquiry convened at the shipyard the day after the loss and sat for weeks. What it assembled was a chain, and every link in it was ordinary.

The most probable beginning was a failure in a silver-brazed joint in a seawater line in the engine room, at depth, with the sea behind it. Water at that pressure does not leak. It comes through as a jet, and it does not stay in one place. Spray from a line that size would have reached electrical panels and switchboards. Salt water in an electrical panel causes short circuits, and short circuits in the wrong panels shut down the reactor. The reactor scrammed. With the reactor down, the boat lost propulsion, and a submarine without propulsion cannot drive itself up. She had one way left to get lighter, which was to blow her main ballast tanks with high pressure air.

They blew them. The air system had strainers in the lines. Compressed air carries moisture, and moisture in air expanding rapidly through a small opening gets extremely cold. Ice formed in the strainers and choked the flow. The boat did not get the air she needed. Testing after the loss, on the same kind of system, produced the same blockage.

She was heavy, she was without power, and she was already deep. Below a certain depth, the hull stops resisting. The collapse took a fraction of a second. Hydrophone records and the sonar trace from the Skylark bracketed the sound. Everything that happened aboard her happened inside about ten minutes, from the first hard-to-read call to the noise in the water.

A congressional committee held hearings on the loss that summer. The court of inquiry's findings went out with much of the technical detail classified, and portions of the record stayed classified for decades. No charges were brought against anyone at the yard, in the design office, or in the chain of command.

What did come out of it was a program. In June 1963, the Navy established SUBSAFE, which set requirements for the material, fabrication, and testing of everything in a submarine that touches the sea or holds pressure, and required documented proof for every certified part. Silver brazing was largely driven out of critical seawater systems in favor of welds. The class stopped being called the Thresher class and became the Permit class, after the next boat in the line. In the years since SUBSAFE was put in place, no submarine certified under it has been lost.

That summer the bathyscaphe Trieste dove on the site, in roughly eighty-four hundred feet of water, and photographed what was there. The boat lay in six major sections spread across the bottom. Trieste brought up a short length of pipe with the number 593 stamped into the metal, her hull number. The Navy still sends ships back to the site to sample the sediment for radioactivity.

None of the one hundred twenty-nine were recovered.