At 5:24 on the morning of November 21, 1944, the radar screen aboard the American submarine USS Sealion showed a Japanese battleship that had stopped moving. The torpedo tubes had just been reloaded. The submarine’s commander, Eli T. Reich, had spent more than two hours regaining an attack position he had lost earlier that night.
The range was seventeen thousand yards. The tubes were ready. Then, on the radar screen, the large return began to shrink. What happened next was not decided by anything aboard Sealion.
What was happening inside that ship during the 144 minutes following its torpedo hits at three in the morning? And why had a battleship that took those hits, then continued steaming at sixteen knots holding its place in formation, ceased to exist by 5:24? Five days earlier, Sealion had been having a bad patrol. On November 16, 1944, her number eight torpedo tube, one of the stern tubes, fired accidentally while both doors were closed.
The sea was too rough for the crew to conduct a proper damage inspection, and the tube was out of service for the rest of the patrol. Two days later, on November 18, a torpedo battery in the forward number five tube caught fire. A submarine is a sealed steel container, and a burning torpedo battery inside one is not a minor incident. The weapon was ejected, and the tube itself was returned to service.
Reich did not abort the mission. Sealion remained in her assigned waters with a lost stern tube and a torpedo room that had just burned inside it. When she later contacted a major target, she would go into action with six operational tubes forward and three aft. Sealion was the second US Navy submarine to bear the name.
She was a Balao-class boat, laid down on February 25, 1943, at the Electric Boat Company in Groton, Connecticut. She was launched on October 31, 1943, and commissioned on March 8, 1944, with Eli T. Reich as her first commander. By the standards of the American submarine force, she was a new boat with an experienced skipper operating at a time when fleet boats had reached the peak of their effectiveness.
The force she was about to face was in a very different condition. The Battle of Leyte Gulf in October 1944 had broken the Imperial Japanese Navy as an offensive instrument. Ships survived, but survival had not left Japan in a strong position. The remaining heavy units of the combined fleets were anchored at Brunei in northern Borneo, close to the oil they needed and far from the shipyards that could repair them.
Fuel was in the south. Repair capacity was in the homeland. These two demands pulled in opposite directions, and by November, compromises had to be made. On November 16, 1944, forty B-24 Liberators and fourteen P-38 Lightnings of the US Army Air Forces attacked the Japanese anchorage at Brunei.
They inflicted no damage. Not one ship was hit. Yet the raid changed everything. That same day, most of the Japanese force at Brunei put to sea, bound for home waters.
That departure was the first decision in a chain that ended with a battleship at the bottom of the Formosa Strait. A failed air attack, which cost the Japanese nothing in lives or ships, had driven those ships to sea. The force sailed under Vice Admiral Takeo Kurita. It comprised the battleships Yamato, Nagato, and Kongo, the light cruiser Yahagi, four destroyers, and two destroyer escorts.
The two escorts detached en route due to an accident, leaving the destroyers Urakaze, Isokaze, Hamakaze, and Yukikaze as defensive cover. Yamato carried the heaviest guns ever mounted on a warship. Nagato had been a symbol of Japanese naval power for more than twenty years. Kongo was the oldest of the three and had been at sea for a long time before this voyage began.
Between Brunei and Japan, Kurita had two route options. He chose the Formosa Strait, the body of water now known as the Taiwan Strait, running along the western side of the island. The logic behind that choice is the essence of what happened. The strait is shallow.
It is narrow. American submarines had spent months concentrated in the deeper passages east of the Philippines and in the Luzon Strait, and Japanese commanders believed the waters north and west of Formosa were protected by geography and by defensive barriers stretching along the entrances to the main islands. A submarine in shallow water has nowhere to go once detected. It cannot dive deep to avoid depth charges.
By the logic of 1942, the Formosa Strait was the safest route. That logic had been obsolete for two years, and nothing in Japanese planning could have told Kurita so, because the thing that made it obsolete was not visible from a Japanese bridge. By late 1944, fleet submarines were not hunting submerged and blind. They were hunting on the surface at night with radar.
A boat fighting that way does not need deep water to win. It needs only darkness and speed. By the evening of November 20, the force had entered the strait and turned north. Sealion was there.
She had moved into the area days earlier to operate in the Formosa Strait, and on the night of November 20 she was making her way through it under overcast skies and poor visibility. Twenty minutes after midnight on November 21, Sealion’s surface search radar picked up a contact at roughly forty-four thousand yards, about twenty-two nautical miles. It was a report that was hard to believe. The radar was being asked to see farther than the crew was used to seeing for a ship.
Reich’s first assumption was that he was looking at land. The northern tip of Formosa lay in that general direction, and a coastline produced exactly that kind of broad, strong echo on a radar screen. But instead of committing to either interpretation, he ordered Sealion to close the contact and watched what the echo did. Land does not change its bearing.
This contact did. As the distance shrank, the single broad echo began to break apart. By 12:48, the men in Sealion’s conning tower could distinguish several large separate echoes moving together. Their initial classification was imprecise, and the ships did not yet have names.
But their size was not in doubt. These were not merchantmen. They were arranged in a disciplined column with smaller targets protecting the flanks, moving on a steady line to the northeast at sixteen knots. They were not zigzagging.
For Reich, the situation was defined by a hard physical fact. Submerged and on batteries, Sealion could not keep pace with a battle line steaming at sixteen knots. If he dove, the Japanese would simply increase the distance and vanish into the night, and the contact would become no more than an entry in his patrol report. To attack, he had to get ahead of them.
The only way to get ahead was to stay on the surface. On the surface, a submarine has no protection worth the name. Her pressure hull is built to resist water, not gunfire. She sits low in the water with very little freeboard, and in front of a battleship that can see her, one accurate burst from a destroyer’s guns or a cruiser’s secondary battery would end matters.
None of the ships ahead needed to use their fourteen- or eighteen-inch shells to kill Sealion. They only needed to find her. There were complications already working against him. Reich had accepted one the moment he turned on his radar and left it running.
Radar emissions go outward. They can be intercepted. Aboard Yamato, Japanese radar detection equipment picked up the American transmissions and reported them. Kurita’s force now possessed something not usually attributed to any Japanese formation in this story.
They had warning. They knew a transmitter was operating somewhere ahead of them in the dark, and the only ship likely to be there broadcasting signals in the middle of the Formosa Strait at midnight was an American submarine. The Japanese were not blind on the night of November 21, 1944. Their own equipment had told them something was out there.
But what they got was a warning without a location. A warning without a location is difficult to act on at sea in the dark. The equipment aboard Yamato could tell Kurita that American radar was operating. It could not give the range or bearing of the device sending the signals.
It gave him a sector of ocean, nothing more, at night under overcast skies that already made visibility poor. His options were limited, and each carried a price. He could slow the formation, reducing noise and bow wave, but slowing a battle line in waters where an enemy was known to operate meant spending more time in the danger zone. He could break formation and maneuver hard to spoil any enemy firing solution, but that would scatter his protection and slow his passage.
Or he could keep moving and run past the point of contact before it developed into an attack. Kurita kept moving. The force held roughly sixteen knots with only limited evasive movements. The decision was justifiable.
Speed was the only protection the heavy formation had, and the fastest way out of the submarine’s operating area was to go forward. From the bridge of a battleship, an unlocated transmitter was a reason for vigilance, not a reason to surrender the initiative to something out in the dark. That decision preserved precisely the conditions Reich needed. His problem was purely mathematical.
Sealion could make about twenty knots on the surface at full power. The Japanese formation was moving at sixteen knots. That left him a margin of roughly four knots, meaning that in one hour of maximum effort he could gain about four nautical miles on ships that never stopped moving. Four nautical miles an hour was not everything, because gaining distance is not the same as gaining an attack position.
Reich had made contact from a point abeam of the enemy fleet. To attack, he had to reach a point ahead of their projected track, then turn inward and let the formation come to him. Part of his speed advantage was consumed by that lateral movement, by the distance he had to cover sideways rather than forward. Every minute spent closing to identify the target was a minute not spent gaining forward position.
Any course change by the Japanese would have forced him to start his geometric calculations over again. Those calculations explain the shape of the next two hours. The contact was made at 12:20. The first torpedo did not leave its tube until nearly 3:00.
That entire period was the price of moving the submarine from behind the battle line to a point ahead of it, and every minute of it was spent on the surface. In retrospect, the work aboard the submarine was slower and less certain than the final accounts suggest. There was no combat display gathering all targets into a single clear picture. The radar operator called out ranges and bearings as the antenna rotated.
Other men plotted those numbers on a paper chart, comparing one position to another, deducing course and speed from the difference between them. Reich made his decisions based on that moving construction, and the construction was only as accurate as the latest sweeps. The escorting ships made it harder. On a radar screen, a destroyer steaming near a battleship did not always appear as a separate target.
The larger echo could swallow the smaller signal. A track toward a main battleship that looked open on the chart might not be, and a destroyer could cross the torpedo path at the last moment. Any misclassification of a target, or any missed change in the formation’s speed, would have moved the firing point before Sealion reached it. The chase continued past 1:00.
By 1:46, the picture was clearer. There were three heavy ships in a column, with destroyers guarding the flanks and rear. Sealion had advanced until she was nearly abeam of the formation, and from there began the slow work of moving ahead. Above the conning tower, bridge lookouts scanned a horizon that showed them almost nothing.
The clouds that hid Sealion also hid the ships she was chasing. For most of those two hours, the enemy existed aboard the submarine only as numbers coming up from below. By 2:45 on the morning of November 21, 1944, Sealion had reached a position ahead of the Japanese track. The surface race was over.
Reich had won it, and winning it brought him only to the point where exposure became deliberate. He slowed and turned inward, letting the formation close on him. Through a gap in the darkness, a lookout reported the sight of a tall pagoda mast, the characteristic stepped superstructure of a Japanese battleship. The radar had promised a battleship.
The silhouette confirmed it. The ship Reich selected, the second in the column, was Kongo. She was the oldest battleship still in Japanese service and the only one built abroad. Ordered in 1911, she was built by Vickers at Barrow-in-Furness, England, launched on May 18, 1912, and commissioned on August 16, 1913, as a battlecruiser.
Two major reconstructions in the 1920s and 1930s had turned her into a fast battleship, fast enough to operate with carriers. She carried eight fourteen-inch guns. She had covered the invasion of the Dutch East Indies, served at Midway, bombarded Henderson Field, and fought in the battle off Samar less than a month earlier, where she helped sink the escort carrier USS Gambier Bay’s escort, the destroyer escort USS Samuel B. Roberts.
Her guns could reach farther than any range at which a submarine could survive on the surface. They needed the target first, and in that overcast dark, Sealion was a low strip of steel in black water. At 2:56, Sealion fired six torpedoes from her forward tubes at Kongo. Reich did not hold his course to watch them run.
He ordered a hard turn to bring the stern tubes to bear on another heavy ship in the column, later identified as Nagato. The forward tubes were empty, and there was a second battleship using the opportunity created by two hours of pursuit. The submarine heeled over in the turn while the tracking party tried to maintain a solution on a target whose bearing was changing for two reasons at once. At 2:59, Sealion fired the three available torpedoes from the stern.
There were now nine weapons running through the Japanese battle line while the submarine that had fired them was still on the surface. At 3:00, the results began to show. The men in Sealion’s conning tower saw and heard three heavy explosions. The crew believed at the time that three torpedoes hit Kongo, and that is what Sealion reported.
Japanese records examined after the war indicate that Kongo was hit twice. That discrepancy deserves to be stated plainly, because it is the first illustration of a problem that shaped the entire night. No one involved could see the whole picture. Reich could see water columns, flashes, and radar contacts.
He could not see inside the target. The Japanese felt the hits and watched compartments flood. They did not know where the submarine was or whether more torpedoes were in the water. What the hits did is not in dispute.
Water entered Kongo from the port side and flooded two boiler rooms. The battleship took on a port list. Behind her, the second salvo produced a different result. Nagato’s bridge saw explosions around the ship ahead and ordered a turn away from the side the torpedoes seemed to be coming from.
That maneuver took the battleship out of the path of Sealion’s stern torpedoes. A torpedo does not pursue a ship that dodges it. Those three weapons continued on their fixed course through a formation full of other ships. One struck the destroyer Urakaze and detonated her magazines.
Urakaze was a Kagero-class destroyer, loaded with fuel, ammunition, torpedoes, and depth charges. She did not sink in the traditional sense. She exploded and vanished within seconds, taking all hands with her, including the commander of Destroyer Division 17, whose flagship she was. There was no wreckage to search for and no sinking to watch.
A later search of the area found a large oil slick and nothing else. The flash that destroyed Urakaze also ended the darkness that had protected Sealion. The Japanese now knew without doubt they were under torpedo attack, and destroyers exist precisely for that moment. Reich took Sealion west at full speed on the surface.
That decision is easy to overlook, and it is the reason that night did not end there. Doctrine and habit say a submarine dives after firing. Diving is what a submarine does when the escorts come, but diving would have cost Reich his speed and his radar picture at the same moment, and would have left him behind a formation moving at sixteen knots with no prospect of catching it again. He stayed on the surface and used the same speed that had gotten him to the firing position to increase the distance before the escorts could organize a search.
The Japanese search went east. The precise reasoning on board each Japanese ship cannot be reconstructed with certainty, but the geometry pointed the wrong way. The most violent explosion of the night, the destruction of Urakaze, had occurred on the far side of the formation after the weapons crossed the battle line. The submarine was expected to be submerged.
The destroyers searched for a submerged submarine in the waters where it made sense for one to be. Sealion was not there. She was steaming west on the surface, preserving her speed and her contact. Depth charges exploded in empty water behind her.
By 3:23, Sealion had opened enough distance to return to her course. The torpedo rooms began reloading the tubes, and on the radar screen the largest target was still in formation, maintaining sixteen knots. Reich concluded from this that his torpedoes had done little damage. His assessment, formed and recorded, was that the weapons had done no more than scratch the battleship’s armor belt.
He had spent two hours getting into position, fired nine torpedoes, sunk a destroyer by accident, and by all appearances the battleship he had targeted was steaming north with the rest of the fleet as if nothing had happened. That belief drove everything he did for the next hour and a half. He began planning a second attack. In the torpedo rooms, reloading was hard work under bad conditions.
Each torpedo weighed more than a ton. Each one had to be pulled from its rack, controlled by chains and handling gear, aligned with the tube, and pushed inside over a deck that was beginning to pitch in a rising sea. The launches had also changed the submarine’s trim, and the tubes had to be drained or flooded in the correct order before anything could be loaded into them. The conning tower and the torpedo rooms were working against the same deadline from opposite ends.
Reich was trying to regain firing position before the Japanese slipped away. The torpedomen were trying to make sure there was something to fire when he got there. Then the conditions that had made the first attack possible began to change. After 4:00, the Japanese formation began zigzagging in earnest.
They had been warned by their own radar detectors, then attacked, then lost a destroyer. There was no longer any ambiguity about a submarine being in the area. Every course change by the battleships forced Reich to discard the planned intercept point and determine a new one, and every change consumed part of a speed advantage that had been thin from the start. The weather took what was left.
Wind and sea had been building since midnight, and the rough sea that had hidden the low submarine hull during the approach now beat against her. A fleet submarine on the surface is long, narrow, and very close to the water. In heavy seas, her bow digs into the waves instead of rising over them cleanly. Water rushes aboard.
The bridge crew takes waves in the face while the engine rooms push the diesels to their maximum. Her speed advantage over a formation steaming at sixteen knots had nearly vanished entirely. The same weather was affecting Kongo, and far more severely. A battleship handles heavy seas far better than a submarine, but Kongo was no longer an intact warship.
Two boiler rooms were flooded, and she was listing to port. The water inside the hull moved with each roll of the ship. The sea pressed against torn openings in her side, and the bulkheads and piping shaken by the explosions had to hold while the ship drove forward at sixteen knots through a rough sea. The generally accepted accounts describe a gradual process rather than a sudden collapse.
Flooding spread through leaks, separated joints, and damaged piping into spaces that had been dry when the torpedoes hit. The pumps that had kept pace with the inflow began losing control. What the command aboard Kongo saw supported the decision to stay with the formation. The ship still had propulsion.
The list could still be managed. There was a submarine somewhere nearby. There might be another spread of torpedoes already in the water. Slowing down meant inviting a second attack.
Stopping would make her a sitting target. So Kongo kept steaming. And steaming was what killed her. On Sealion’s radar, none of the internal damage was visible.
Reich had only the target’s behavior, and over the next period that behavior changed. The damaged battleship stopped zigzagging. Then her speed began to drop. The contact that had been holding its place alongside the other large targets began to fall behind them.
At about 4:50 on the morning of November 21, 1944, the Japanese formation split. Yamato, Nagato, and the light cruiser Yahagi continued north toward Japan with the destroyer Yukikaze. Kongo fell back with the destroyers Hamakaze and Isokaze, reduced her speed to about eleven knots, and turned toward Keelung on Formosa, about sixty-five nautical miles away, where emergency repairs might be possible. Vice Admiral Yoshio Suzuki, commanding the Third Battleship Division and embarked on Kongo, agreed with the assessment, and Rear Admiral Toshio Shimazaki, Kongo’s captain, was ordered to detach and head for the Formosan coast.
The decision was sound. Kongo could no longer maintain the formation’s speed, and keeping Yamato and Nagato tied to a crippled ship in waters already attacked by a submarine would expose Japan’s most valuable remaining units to danger. Detaching her with two destroyers was the responsible response, and no better option existed. It also took Kongo out of the protection of the battle line at the exact moment she needed it most, and it changed the picture on Reich’s radar screen.
The target he wanted was no longer one large contact buried among several identical ones. It was a slow, isolated target falling behind, guarded by two destroyers. The second race had just been won for him, but it created a different problem. Hamakaze and Isokaze were no longer protecting a full formation from an unknown threat.
Their focus was on a single damaged ship, and they knew exactly why she was slowing down. Reich began another end run, and this time he had to deal with a different issue. Catching Kongo was no longer the difficulty. A battleship doing eleven knots was within reach.
The difficulty was everything that would happen after he fired. Two destroyers focused on a single ship in a patch of ocean where they already knew a submarine existed, and this time they would not have to guess about the source of the attack. Firing a second spread would confirm to Hamakaze and Isokaze that the submarine had not left. So Reich needed more than torpedo range.
He needed an angle that would keep the destroyers from crossing the torpedo track, enough distance to get away afterward, and enough remaining night to be somewhere else by the time the escorts reached his firing position. He moved ahead of the slow group and watched the weakened contact on the screen. Kongo’s speed dropped again. Inside the battleship, the list to port was still increasing.
Damage control parties could shore bulkheads, close doors, and shift portable pumps. They could not restore hull integrity at sea. With the decks canted, every task took longer and required more effort. Equipment shifted out of place.
Openings that had been above the waterline when the ship was upright were beginning to dip toward it, and every new space that flooded made survival less likely than before. A battleship can absorb a tremendous amount of damage as long as its buoyancy remains balanced and its watertight bulkheads hold. Once the increasing list starts putting new openings near the waterline and allows water to spread around those bulkheads rather than being contained by them, the process changes from a manageable problem to a self-feeding one. Crew were moved to the high side to try to counter the list.
The list kept growing, reaching about forty-five degrees. At 5:12, Reich slowed and turned inward for the second attack. The range was much greater than the first time. He was still working on getting Kongo into the upcoming firing position when the radar operator reported that the target had slowed sharply.
At about 5:25, Kongo stopped. Then the radar contact began to shrink. Radar returns change with angle, range, and sea clutter, and a stationary ship will not look the same as a moving one. But a large contact gradually losing strength while holding its position tells the men watching it that there is less of the ship above water than there was minutes before.
At 5:22, Rear Admiral Shimazaki gave the order to abandon ship. Sealion held the contact at seventeen thousand yards with her torpedo tubes ready. The two destroyers were still close. They were minutes away from a second attack that the whole night had been spent preparing.
At 5:24, Kongo’s forward fourteen-inch magazine exploded. The blast lit the underside of the clouds and spread across the water. Four heavy explosions followed, the last of which tore the ship apart. The glow shrank, the darkness returned, and on the radar screen the large contact simply disappeared.
Only the two destroyer contacts remained. Kongo sank in less than a minute in 348 feet of water at latitude 26 degrees 9 minutes north, longitude 121 degrees 23 minutes east, about sixty nautical miles north of Keelung. Sealion did not fire the second spread. The ship had been destroyed by damage that had been inside her since three o’clock that morning.
Two torpedoes had opened the way. Speed, pounding seas, flooding, and the collapse of internal bulkheads finished the job, and the most accepted explanation is that the increasing list caused the ship to capsize, with the forward magazine exploding as the hull rolled over. That answers the question that had been building all night. The delay between the hits and the explosion was not evidence of the first attack failing.
It was the time required for localized damage to become a total loss of stability, and the fact that Kongo was still able to steam at sixteen knots had hidden that process from everyone on both sides until it was too late. An engine room can maintain steam pressure while the ship it drives is losing its ability to keep water out of its own hull. That is also why the decisions made that night looked reasonable at the time and fatal afterward. A failed air raid on Brunei that caused no damage put the fleet to sea.
A route through shallow water, chosen on the belief that submarines were a deep-water problem, placed them in Sealion’s patrol area. The decision to hold sixteen knots after the radar warning, because slowing a battle line near a submarine was worse, preserved the geometry Reich needed. The decision to steam west on the surface instead of diving kept the submarine in contact while the escorts searched east. And the decision to detach a crippled battleship rather than tie Yamato and Nagato to her, which was the only responsible response available, left Kongo alone at the moment she could least afford it.
Each decision was justifiable. Together they formed a chain no one in the Taiwan Strait could see whole that night. Hamakaze and Isokaze moved toward the wreckage and worked through heavy seas and total darkness to rescue survivors. They pulled 237 men from the water and carried them back to Kure.
Some 1,200 others went down with the ship, including Vice Admiral Suzuki and Rear Admiral Shimazaki. Urakaze had already gone down with her entire crew. Reich tried to regain contact with the rest of the force. Yamato and Nagato were still steaming north, and the conditions that had made the first attack possible were gone.
The Japanese now knew precisely what kind of threat had found them. The formation was alert, zigzagging, and increasing the distance. Sealion sank nothing else that night. That fact is part of the record, not a diminishment of it.
What was achieved in the Taiwan Strait was a single firing position, with radar, surface speed, darkness, and a margin of about four knots, used against one ship before the geometry closed again. No submarine controlled a battle fleet. One submarine out-thought one warship twice. The strategic consequences were larger than the tonnage of the sunken ship.
The Japanese had taken the Taiwan Strait route with the core of their remaining fleet because it was supposed to be the safe path home, close to friendly shores and protected by shallow water. By November 1944, there was no safe route left. A navy that had already lost the ability to mass at sea had now lost the ability to move its heavy units between its own ports without expecting attack. Eight days later, on November 29, 1944, the submarine USS Archerfish sank the carrier Shinano in home waters on her first voyage.
Kongo was the only battleship sunk by a submarine in the Pacific War. She was hit by a submarine that never dove to attack, never fired the shot that destroyed her, and never saw her sink except as a light on the horizon and a contact fading from a radar screen.