In May 1942 the United States Navy told a lie about running out of water, in plain language, on a frequency the Japanese were certain to be listening to. It was probably the cheapest decisive move in the history of naval warfare.

The garrison on Midway had been instructed to radio, uncoded, that its water distillation plant was wrecked and fresh water was needed immediately. Nothing was wrong with the plant. Soon after, American listening posts caught a Japanese intelligence report relaying that “AF is short of water”, as documented in the National Security Agency’s published history of the battle by Patrick Weadon. AF was the designator Japanese signals kept attaching to the target of a very large operation, and nobody in Washington could agree on where that target sat. The argument was over.

Three weeks later Japan lost four aircraft carriers in a day.

Nimitz’s arithmetic problem

The numbers were grim and everyone involved knew it. Weadon puts Admiral Chester Nimitz’s strength in June 1942 at three aircraft carriers, 45 fighting ships and 25 submarines. That was the entire barrier between a Japanese fleet that had not yet lost a single ship and the American west coast.

Japan sent four fleet carriers with roughly 250 aircraft aboard, trailed several hundred miles back by a battleship-heavy main body whose job was to finish off whatever came out to fight. The Americans could muster about 230 aircraft across their three flight decks and another 130 on the island, aircraft counts confirmed by the American Battle Monuments Commission.

Nimitz’s advantage sat entirely outside the arithmetic.

How JN-25 came apart

JN-25 was the Japanese navy’s general operational code, and it was a monster. It held around 45,000 five-digit groups, each standing in for a word or a phrase, and those groups were then enciphered a second time by adding figures drawn from a table. Reading a message meant stripping that added layer off mathematically, then working out over months, from patterns of use, what the underlying numbers referred to.

Weadon identifies the crack that let the Americans in: those additive tables were reused, and repetition is the enemy of any cipher. Even so, the work crawled, and before Pearl Harbor the team was reading barely a tenth of the traffic.

Limits of what Hypo could read

The popular version of this needs a correction, and the sharpest one comes from Samuel Cox, director of the Naval History and Heritage Command. Even at their best, months later, Cox writes, the codebreakers were intercepting only about 60 per cent of Japanese naval communications, analysing roughly 40 per cent, and actually breaking and reading 10 to 15 per cent, no better than before Pearl Harbor and often nothing more than fragments of a message body. What came through arrived as dense technical naval Japanese, and Pacific locations routinely went by several different names.

So Nimitz never sat down with a copy of the Japanese operation order. He had scraps, traffic analysis, and the instincts of his fleet intelligence officer, Lieutenant Commander Edwin Layton, whom he had told to spend his days thinking like Yamamoto.

Cox also reverses the water stunt’s purpose. Most accounts, including the National Museum of the Pacific War, have Nimitz ordering the fake signal to satisfy himself that AF meant Midway. In Cox’s telling, Nimitz and his codebreakers were already convinced, and the performance was staged for an audience 8,000 kilometres away in Washington, where senior officers had decided the intelligence was too good to be true.

Ambush at Point Luck

General George Marshall credited codebreaking with sparing the fleet from being “3,000 miles out of place”. Nimitz stationed his carriers northeast of the atoll at a spot he named Point Luck. On 18 May he asked Layton to predict where and when the Japanese carriers would first be sighted. When a patrol flying boat found them on the morning of 4 June, the report landed almost exactly on Layton’s estimate.

Japan’s own reconnaissance collapsed in every direction at once. A planned flying-boat sweep over Pearl Harbor was scrapped on 31 May because the submarines meant to refuel the aircraft found American ships parked on the refuelling spot. The picket line strung out to catch the US carriers leaving Hawaii sailed late, and by the time those boats reached station all three had passed. One submarine did report that Midway’s defences were heavily reinforced and its patrol aircraft flying to extreme range. Senior Japanese commanders filed the warning and carried on.

The result, tallied by the National WWII Museum, was four Japanese carriers, a cruiser, hundreds of aircraft and about 3,057 men lost, against one American carrier, one destroyer, 144 aircraft and roughly 362 men.

Rochefort’s reward

Commander Joseph Rochefort, who ran the Pearl Harbor codebreaking unit, was recalled to Washington on a pretext within weeks, never returned, and was handed command of a floating dry dock. Nimitz recommended him for the Distinguished Service Medal. Admiral Ernest King refused, on the advice of a Washington office that then claimed credit for having predicted the Midway operation, a claim Cox says was false. The relevant records stayed classified until the 1970s. Rochefort finally received the Navy Distinguished Service Medal in 1985, posthumously, after a long campaign by a former junior officer from the same basement.

None of this sank a ship. Codebreaking put no bombs through any flight deck. The men who did that flew obsolete torpedo aircraft into curtains of fire and mostly died doing it. The intelligence decided which patch of empty ocean the two fleets met in, and who was looking the right way when they did.

Yamamoto had more of everything at Midway except an accurate idea of what his opponent knew, and I suspect that is the one shortfall in modern warfare that cannot be covered by building more ships. Which is why every serious navy since has spent so freely on rooms containing nothing but people reading.