On January 7, 2005, the USS San Francisco, a nuclear-powered submarine, set sail from its home port in Guam for Brisbane, Australia. Its crew of 137 had a grueling schedule of training and drills planned, along with some shore time.

The ship and its sailors were prepared for the trip. The crew had cleaned the sub and stowed anything that could possibly come loose; a simple wrench clanging to the floor could be detected by an enemy, betraying the vessel’s exact position. They secured doors and hatches against the billions of gallons of water outside the pressure-sealed hull. The San Francisco was ready to perform its duty—to patrol and conduct reconnaissance missions for the Navy while maintaining combat readiness with two dozen Mk-48 torpedoes and Tomahawk cruise missiles aboard.

The San Francisco had made a turnaround in recent years. It previously had a reputation in the Navy for not always meeting readiness standards. But its new captain, Commander Kevin Mooney, a bookish veteran with 19 years under his belt, had vastly improved operations since taking over two years prior.

Bill Cramer, the San Francisco’s master chief at the time—the most senior ranking enlisted crew member—recalls Mooney’s influence trickling down to the smallest details. He describes a small door at the very front of the sub, inside a massive sonar sphere, that had drawn Mooney’s attention. The commander had pointed out that any breach in the forward hull would send water cascading through that door. “Just a year before on any given dive, there was a 50/50 shot whether or not that door gets shut,” Cramer recalls. “In this instance, we did follow procedures. That door got shut.”

Things were improving under Mooney’s command. By all accounts, the San Francisco and its sailors were as ready as they could be as it slipped hundreds of feet beneath the Pacific Ocean and cruised comfortably toward Brisbane. On board in the cramped quarters, Mooney’s MO was well known: confirming readiness while connecting with his crew. “I knew everyone on board by their first name,” Mooney says in his first media interview about the accident in 20 years.

Submarine crews benefit from that sort of unity and camaraderie. Unlike a surface ship, a Navy sub can be completely removed from the outside world while deployed. It can remain submerged for months at a time, generating oxygen on board via electrolysis, scrubbing the air of carbon dioxide, and using its own desalination plant to supply fresh water. The nuclear reactor that powers the sub and provides electricity has no need for air from the surface. That independence is itself a devastatingly effective weapon in times of war, but it demands a lot from the sailors who sign up for it.

Furthermore, a submarine is a compact vessel, which can be further stressful for those on board. The San Francisco was 362 feet long and its primary hull only about 30 feet tall and wide. It contained a torpedo room up front—temporary crew bunks were mixed in with the weapons as well as interspersed among other spaces on board—along with a control room in the middle, a radio room, officer and crew galleys, a laundry room, a reactor compartment, an engine room, and a broad assortment of propulsion and control equipment spread throughout. It’s a fast-attack submarine—a weapon of war—not a cruise ship, and submarine work is among the most isolating experiences of the modern world. Crews operate efficiently and purposefully, and become tightly knit comrades in the process.

Between 10 and 15 crew members circulate in and out of the control room at any given time. The general route and average speed of an assignment originate from the Navy, and the sub hones the details. Usually the orders are issued to the sub three to five days before sailing, but in this case the San Francisco received them just two days prior, adding a degree of pressure to the navigation team preparing for the voyage.

For this mission, the Navy ordered the San Francisco to proceed at high speed toward Australia, maintaining an average pace of 20 knots. This meant that surges above 20 knots would be needed, since periods of lower-speed “housekeeping” duties were required, including surfacing for ventilation and carrying out training tasks. As the sub neared the sprawling volcanic archipelago of the Caroline Islands to the north of New Guinea, the chosen route should have kept the sub far from the underwater mountains. The charts they consulted showed no navigational hazards within 10 nautical miles of the sub’s intended heading. To keep the sub on pace, Commander Mooney placed no restrictions on speed. As it entered the region, the vessel dove to 500 feet and raced ahead at more than 30 knots, or 35 mph—about as fast as it could go.

But submarine navigation is “an art and a science,” says Sam Tangredi, a professor of national, naval, and maritime studies at the U.S. Naval War College and a retired naval officer. For one thing, he says, the existing bathymetric data—what’s used to measure and map the seafloor—is incomplete. (It’s a big world with deep oceans, after all.) “Despite terrain mapping using sonar, much of the ocean has not been charted,” Tangredi says. Making matters even worse, the seabed changes constantly, making the whole process of mapping the ocean floor inherently fraught. “Just as earthquakes, landslides, and tsunamis can change features on land, they can also change undersea features,” he says. “These changes may not be reflected in electronic or paper charts if the route has not been resurveyed.”

To top it off, human error plays a role. “Survey ships can always be incorrect in fixing their own position, and some charts of the same area have contradicting information,” Tangredi says. An undersea mountain may appear on one chart but not another, so it’s incumbent on the crews to cross-check every mile of a ship’s voyage to see what hazards may exist on route, even if some charts show a clear path.

One such hazard, in fact, stood undetected and directly in the path of the USS San Francisco. It wasn’t shown on the chart the navigation team was using to race past the Caroline Islands, a week into its journey.

Cramer was in the chief’s quarters discussing drills with another officer when the San Francisco slammed at full speed into an underwater mountain, violently halting all forward momentum. “It sounded like an explosion,” Cramer recalls. “My head flew back into the bulkhead, and he came flying forward on top of me. My head hit a television monitor, so I was bleeding from the back of my head.”

Crew members were sent flying through the air into steel doors, pipes, and fixtures on board. Everything that wasn’t secured became a projectile. No collision alarm sounded from the control room. None was needed. Ninety-eight injured sailors suddenly lay writhing in pain throughout the 6,000-ton vessel. Fears of imminent flooding—a certain death sentence—sent every crew member who was still mobile into immediate damage assessment and recovery.

They were miles away from dry land and far from any other Navy vessels. A weakened structure could give way at any second, which meant sailors scrambled to assess damage while simultaneously triaging injured crew members.

Blood was everywhere. One officer was bleeding profusely from his mouth, having taken a direct hit to his larynx. “There were several sailors laying on the deck outside of the chief’s mess,” Cramer says. “We got them up and went to the crew’s mess. We checked for flooding everywhere.” By this time, the crew heard emergency reports over the loudspeaker announcing additional injured sailors. “That’s when it really hit me, the severity of it,” he says.

Up in the control room, Mooney, who at the time of the accident was eating lunch and was scratched up but mostly uninjured, immediately began directing efforts to secure and stabilize the boat and ensure that the 98 injured sailors were being tended to. One of the injured crew members, Machinist Mate 2nd Class Joseph Allen Ashley, 24, of Akron, Ohio, had suffered a grave head injury.

Mooney also had to manage what would prove to be a perilous surfacing. “The first thing we feared was flooding,” recalls Mooney. “We had to verify not just hull integrity, but that all the ship systems were operating properly. Some were, some weren’t. It became all about communication—keeping it brief but precise and not making a bad situation worse.”

Immediately after the collision at 11:42 a.m. local time, the chief of the watch performed an emergency blow—a procedure to surface the submarine immediately by injecting compressed air into ballast tanks, increasing buoyancy. Throwing the “chicken switches” to initiate the blow requires lifting a pair of knobs located above the ballast control panel, which starts the process of injecting air into the tanks.

But what should have initiated a brisk ascent to blue sky initially had no effect whatsoever. Most of the ballast tanks at the front of the sub were damaged, causing the forcibly injected air to simply shoot out into open water. After 30 seconds, the rear tanks filled and the ship slowly rose. It broke the surface about two minutes after the collision.

It was clear that the crew was in bad shape. “There were lots of injuries, lots of trauma,” Mooney says. Once he was sure the vessel was stable, that power systems were functioning, and that an emergency report had been radioed to the Navy’s incident authorities, Mooney said he walked “from stem to stern, top to bottom” to assess damage and injuries. In the crew’s mess hall, two corpsmen delivered medical aid to MM2 Ashley, who was unconscious, unresponsive, and had severe swelling on his face.

Though the ship was now on the surface, the struggle was far from over. The San Francisco was still alone in the middle of the Pacific, with more than half of the crew injured—one gravely—and their stricken vessel now of deeply uncertain seaworthiness. Fortunately, the nuclear reactor powering the ship and the full complement of torpedoes and cruise missiles on board all appeared to be undamaged and secure. And the pressure hull containing all the crew spaces had held together. But the crushed front end and punctured forward ballast tanks caused the sub’s nose to pitch downward. In rough seas, this made the rear propeller bob out of the water as the boat tried moving forward, drastically slowing its progress.

As Mooney turned the ship toward Guam and continued reassigning crew to critical posts, word spread quickly within the Navy that there was an emergency at sea. Teams immediately began mobilizing to assist the stricken sub, including diverting ships and aircraft in the San Francisco’s direction and flying in a doctor who’d previously served as a Navy SEAL on a helicopter, with the intent of lowering him onto the submarine.

Below, efforts continued to save MM2 Ashley, who had been placed on a stretcher and connected to life support equipment. They hoped to airlift him out via the chopper that was on its way, but Ashley’s stretcher and medical equipment wouldn’t fit through the sub’s damaged hatches. Shortly after the doctor arrived, nearly 24 hours after the collision, Ashley’s heart stopped. The physician performed CPR for 10 minutes, but Ashley couldn’t be revived. The Navy later determined that his injuries were unsurvivable and that a successful medevac operation wouldn’t have saved him. Both Mooney and Cramer, as well as many other friends and shipmates, were with him when he died.

It would take another 26 hours for the San Francisco to make its way back to Guam. Only when the sub reached port and entered dry dock could the full extent of the damage be seen. The entire nose cone of the ship was caved in, and the massive 24-foot spherical sonar array the cone covered warped like an ice cream cone that had been thrown to the ground.

The San Francisco’s undersea grounding was a major event. It was one of the most significant accidents in the post-WWII Navy, though not as tragic as the disasters involving the USS Thresher in 1963 and the USS Scorpion in 1968, two nuclear-powered subs that were lost with all souls on board. The former sank with 129 crew during dive tests. Its loss—the first nuclear submarine to sink—spawned the Navy’s modern safety program, called SUBSAFE. The Scorpion sank with 99 crew members in the North Atlantic after what naval authorities suspect was a catastrophic implosion. Fortunately, the San Francisco fared better, returning to port under its own power with just a single casualty. The damages were extensive nonetheless. The Navy was forced to remove it from service for a year, and the repairs cost a staggering $88 million.

But while the specifics of the event place it high in the rankings of naval mishaps—and it certainly made headlines at the time—the San Francisco saga has been largely forgotten. Had the ship suffered greater losses—had a wrinkled piece of steel caved in before it could surface—the result could have been years of hand-wringing within the Department of Defense.

As it was, the investigation revealed uncomfortable truths that few disputed, including Mooney himself. At the heart of it was the fact that the navigation team on board had relied on a chart, designated E2202, to plot a course that the sub intended to travel at high speed and significant depth. The use of only one chart contradicted Navy procedures that required crews to cross-check multiple other charts that are available to them. One of those on board the San Francisco, Chart 81023, contained a “discolored water” site near the sub’s intended route. Those color distinctions tend to mean shallower depths and the possibility of a navigation hazard. The E2202 chart, however, showed a clear path.

Vice Admiral Jonathan W. Greenert, commander of the U.S. 7th Fleet at the time, wrote in support of the final report that consistent reliance on a single chart, when “other charts with critical information were readily available,” helped lead to the grounding. The report also noted that the crew didn’t exercise enough caution given the sub’s speed and depth. It was an unnecessary risk, says former submarine officer Bryan Clark, a senior fellow and director of the Center for Defense Concepts and Technology at Hudson Institute. “They were going pretty fast, which is unnecessary in that area unless you’re getting shot at,” he says.

Undersea navigation is very different from the relative ease of GPS location tracking on land; it relies on a multitude of technologies and processes for determining a sub’s position on the charts being used, any one of which can be flawed for multiple reasons, says Tangredi, the U.S. Naval War College professor. “Poor observations, incorrect calculation, interference with signals from electronic aids…relying on any one method of navigating greatly increases risk,” he says. “Navigators skilled in the ‘art’ of navigation are those who use all techniques, but are always thinking the question: What are the risks to the current operation (and the ship) if I am wrong?”

Even now, more than 20 years after the USS San Francisco disaster, undersea navigation remains a challenge. In October 2021, a Seawolf-class nuclear-powered attack submarine, the USS Connecticut, collided with an underwater seamount in the South China Sea. Eleven sailors were injured. The subsequent report faulted the commander and navigation team for the incident, most of whom were fired, including the captain. The final report filed by the 7th Fleet leadership sounds strikingly familiar: “To meet the mission, the crew plotted a course through regions that weren’t wholly charted, and the navigator and the boat’s leadership had conflicting ideas as to how well they understood the route.”

According to Tangredi, the risks inherent in undersea navigation haven’t improved as much as one might have expected. “There are places in the South China Sea marked simply ‘Dangerous Ground’ because it has areas too shallow to allow a survey ship to operate safely. Obviously, one avoids those areas, but what if one’s navigation is incorrect?”

Science and technology are improving. Ring laser gyros, which measure the distance between multiple laser beams to track a submarine’s slightest movements, have become more accurate for tracking positions, while even more precise “quantum navigation” technologies are being tested. These new systems—one of which was recently tested on an uncrewed submersible by the United Kingdom’s Royal Navy—will use motion-sensitive sensors to analyze the movement of atomic particles, allowing them to calculate changes in a vehicle’s position, speed, and orientation. Those technologies could appear broadly within about a decade, Clark says.

During the San Francisco investigation, Commander Mooney was reassigned to a role on shore and then ultimately relieved of command via a career-ending letter of reprimand. Six other crew members received reprimands for their involvement in the navigational errors that led to the grounding, while 20 officers and sailors received commendations and service medals for their actions in the wake of the collision.

Mooney accepted his responsibility. “I knew when this happened that I was going to be held accountable, and despite whatever circumstances there might have been, I knew it was going to live with me for the rest of my life,” he says. “I told myself at the time that I’m not going to try to wriggle my way out of this or try to cover it up. The investigation made me look really bad, but I just have to live with that. Investigations don’t say the good things that you did.”

Except, however, that this one actually kind of did—even if obliquely. Vice Admiral Greenert wrote: “The San Francisco immediately took actions which enhanced the ship’s ability to minimize the effects of the grounding. The ship appropriately conducted an emergency surfacing and subsequently sustained the low-pressure blow continuously until return to port, maintaining post-grounding stability.” Also: “Post-grounding damage control and medical care were proper and exceptional in all critical respects.”

There’s also, of course, that front door in the sonar sphere. “If that door wasn’t shut, we would have lost that ship,” says Cramer.