“AF Is Midway”: Rochefort, Station HYPO and the Codebreakers

Joseph J. Rochefort, officer in charge of Station HYPO, the Combat Intelligence Unit at Pearl Harbor, in 1942. U.S. Navy photograph, public domain, via Wikimedia Commons.

The Battle of Midway · Chapter 1 of 9

“AF Is Midway”: Rochefort, Station HYPO and the Codebreakers

In a windowless basement at Pearl Harbor, a small staff of codebreakers, linguists and traffic analysts turned fragments of Japanese naval traffic into the estimate that put three American carriers in the right place.

March–June 1942Pearl Harbor · Washington · Melbourne · Midway~25 min read

Nobody fired a shot in the first battle of Midway. It was fought in a basement at Pearl Harbor, in a Navy communications office in Washington, in a radio unit evacuated from Corregidor to Melbourne, and on the desk of Admiral Chester W. Nimitz. Between March and the end of May 1942, the U.S. Navy’s radio intelligence organization turned fragments of intercepted Japanese naval traffic into a working estimate of where the Imperial Japanese Navy would strike next, roughly when, and with what. That estimate let Nimitz put his only three available carriers northeast of Midway Atoll before the Japanese arrived.

The estimate was never a complete reading of the Japanese plan. The codebreakers read only parts of messages, often only a few groups in each. They worked alongside traffic analysts who studied call signs, routing and volume rather than content, and their conclusions were contested by senior officers in Washington until about 22 May. The best-known episode, the bogus report that Midway was short of fresh water, did not reveal Midway as the target for the first time. It confirmed an identification the Pacific analysts had already made and gave Washington a document it could not argue with.

This chapter follows that work from the inside. It explains how the Japanese fleet’s main operational code, called JN-25 by the Americans, actually worked, and how an additive cipher can be stripped away by subtraction. It traces the identification of the designator “AF,” the dispute between Pearl Harbor and Washington, the water ruse and the Japanese report that answered it, the date-time cipher, and the decisions Nimitz made on the strength of it all. It also sets out what the codebreakers did not know, which was considerable, and how the officer at the center of the effort, Commander Joseph J. Rochefort, was removed from his post within months of the battle and decorated only in 1986, ten years after his death.

The surviving record is uneven. Some messages survive in contemporary translations; much of the inside story comes from memoirs written decades later by men who were there, notably Edwin T. Layton’s And I Was There (1985) and W. J. Holmes’s Double-Edged Secrets (1979). Where this chapter relies on later recollection rather than wartime documents, it says so.

The dungeon under the Fourteenth Naval District

The unit that did most of the work on Midway had an awkward set of names. Within the Navy’s radio intelligence network it was Station HYPO, from the phonetic alphabet’s word for H, for Hawaii. Its sister stations were NEGAT in Washington, the headquarters element of the Navy’s communications intelligence section, designated OP-20-G, and CAST on Corregidor in the Philippines, which was evacuated to Australia by submarine in early 1942 and re-formed in Melbourne. In the summer of 1941 the Hawaii unit was renamed the Combat Intelligence Unit of the Fourteenth Naval District. Its people simply called their workplace the dungeon.

That workplace was the basement of Building 1, the administrative headquarters of the Fourteenth Naval District at Pearl Harbor. Navy Region Hawaii’s own history describes it as “a large windowless space” crowded with tables, boxes of index cards and tabulator printouts, reached by a single entrance guarded at all times by an armed Marine. The unit stayed there until April 1943, when it moved to new quarters at Makalapa near Pacific Fleet headquarters. The basement then spent decades as storage and offices before its history was rediscovered; plaques now mark the stairways.

Rochefort had been sent to Hawaii in early 1941 by Captain Laurance F. Safford, the founding figure of Navy cryptanalysis, to take charge of the station. He was an unusual officer for the post because he combined three skills the Navy normally kept apart. He had trained as a cryptanalyst in 1925 under Safford and Agnes Meyer Driscoll, had served as the second chief of OP-20-G from 1926 to 1929, and had then spent 1929 to 1932 in Japan as a language officer. NSA’s biographical note calls him “an expert Japanese linguist, an experienced and very talented intelligence analyst, and a trained cryptanalyst.”

His senior people were chosen for the same mix. Commander Thomas H. Dyer ran the cryptanalytic effort and had pioneered the use of IBM punched-card tabulating machinery on Japanese naval codes. Lieutenant Commander Wesley A. “Ham” Wright was another veteran cryptanalyst. Lieutenant Commander Joseph Finnegan and Marine Major Alva B. Lasswell were the principal Japanese linguists, able to make sense of a message in which only a handful of words had been recovered. Lieutenant Commanders Thomas Huckins and John A. Williams handled traffic analysis, and Lieutenant Commander Jack S. Holtwick ran the machine room. W. J. “Jasper” Holmes, a retired submariner recalled to duty, kept the plot that turned decrypts into positions and courses. After the attack of 7 December 1941, members of the band of the battleship California, sunk at her moorings, were assigned to the unit to operate the tabulating machines.

The working conditions became part of the legend. Accounts of Rochefort describe him working in a bathrobe and slippers over his khakis, and spending twelve hours or more a day in the basement, sometimes for days at a stretch. These details come from later recollection rather than records, but they are consistent across the memoirs of men who worked there. What the wartime record does show is the volume. NSA’s history notes that in May 1942 Rochefort personally reviewed as many as 140 decrypted messages a day, and reports from the unit went directly to the senior fleet commanders.

JN-25: a codebook, an additive book and a great deal of arithmetic

The Japanese navy used many systems, but the one that mattered for Midway was its general-purpose operational code, which American analysts called JN-25. It was a two-part system. The first part was a codebook. Each word, phrase, place name, number or syllable the navy expected to use was assigned a five-digit code group; the published estimates of its size run from about thirty thousand to more than forty thousand groups. The codebook alone would have been vulnerable to anyone who captured or reconstructed it, so the Japanese added a second layer.

The second part was an additive book: tens of thousands of random five-digit numbers printed in pages and columns. A code clerk encoding a message looked up each code group, then took a run of additives from a starting point in the additive book, and added each additive to each code group. The sum, not the code group, was transmitted. An indicator buried in the message told the receiving clerk where in the additive book the sender had started, so that he could subtract the same additives and recover the code groups.

The addition was done digit by digit without carrying, what cryptanalysts call false or non-carrying addition. Seven plus eight is fifteen, but in non-carrying arithmetic the clerk writes only the five and drops the ten. Subtraction works the same way in reverse: if the top digit is smaller, borrow ten for that column only. Arithmetic modulo ten in each column is quick to do by hand, it never produces a sixth digit, and it can be undone only if you know the additive.

A worked example shows both the encipherment and the weakness the Americans exploited. Suppose the codebook gives the group 13458 for a word. The clerk’s next additive is 72916. Column by column, 1+7 is 8, 3+2 is 5, 4+9 is 13 (write 3), 5+1 is 6, and 8+6 is 14 (write 4). The transmitted group is 85364. The receiving clerk subtracts the same additive: 8−7 is 1, 5−2 is 3, 3−9 borrows to 13−9 for 4, 6−1 is 5, and 4−6 borrows to 14−6 for 8. Out comes 13458. These numbers are illustrative, not real JN-25 groups, but the arithmetic is exactly the kind the system used.

The Japanese also built in a check that turned out to help the enemy. Every valid code group in the book was divisible by three (its digits summed to a multiple of three), so that a clerk could spot a garbled group. For an American analyst trying a candidate additive against an intercepted group, the same rule worked as a test. If subtracting the candidate additive left a number not divisible by three, the candidate was wrong.

The decisive weakness was reuse. An additive book was meant to be replaced before its pages had been used too often, but wartime traffic outran the replacement schedule. When two or more messages were enciphered with the same stretch of additives, analysts said they were “in depth.” Subtracting one enciphered group from another at the same position cancels the shared additive and leaves the difference between two code groups. With enough messages in depth, the punched-card machines could sort and compare thousands of groups, the divisibility check could weed out false trials, and the analysts could recover additives, then strip them from other messages, then work on the meaning of the underlying code groups. Each step fed the next. Recovered code meanings went into an index, so that the more the unit read, the faster it could read.

This was slow, cumulative work, and it never produced complete messages on demand. Estimates of how much could be read vary with the date and the source. An NSA history gives “only 10% to 15%” of the code being read before Pearl Harbor; other accounts put the figure lower. By the spring of 1942, according to a history of the Australian Central Bureau cited in the literature on JN-25, the Allies could read roughly one word in five. What mattered was that the words recovered included place designators, unit names, dates and operational verbs, and that HYPO, CAST in Melbourne and NEGAT in Washington shared their recoveries.

The spring of 1942: from Coral Sea to Operation MI

The codebreakers’ first great test of 1942 came not at Midway but in the Coral Sea. From March onward, traffic and partial decrypts showed the Japanese preparing an operation against Port Moresby on New Guinea. That forecast allowed Nimitz to send Lexington and Yorktown south, and on 7–8 May the first carrier battle in history was fought. The Americans lost Lexington, and Yorktown was damaged; the Japanese lost the light carrier Shoho, Shokaku was badly damaged, and Zuikaku‘s air group was depleted. Both large Japanese carriers would miss Midway. Frederick D. Parker’s NSA history of the period treats Coral Sea and Midway as one continuous intelligence campaign, and the success at Coral Sea gave Nimitz reason to trust HYPO’s next forecast.

Even before Coral Sea, Japanese planning had turned toward Midway. Parker records that the Imperial General Headquarters agreed to Admiral Isoroku Yamamoto’s Midway operation on 16 April 1942, with early June chosen so as not to disrupt operations planned for mid-May. Two days later, on 18 April, sixteen Army B-25s launched from USS Hornet bombed Tokyo and other cities. The Doolittle Raid did no serious damage, but it ended resistance within the Naval General Staff to Yamamoto’s plan to extend the defensive perimeter and draw out the U.S. Pacific Fleet. The Combined Fleet’s war games on the plan ended on 5 May. Navy Order No. 18 directed Yamamoto to “carry out the occupation of Midway Island and key points in the western Aleutians in cooperation with the army.” The plan and its politics are the subject of Chapter 2.

From the American side, the picture assembled itself gradually. Through late April and early May, the analysts saw a large operation forming that was distinct from the Moresby operation: new task organizations, fueling arrangements, the movement of air units and base personnel, and repeated references to an objective designated by a two-letter code. Traffic analysis, the study of the outward forms of messages rather than their content, did much of this early work. Parker notes that before JN-25 could be read in volume, analysts relied on call signs, routing and message volume, and that these patterns helped separate Midway-related activity from Moresby-related activity.

The early estimates were not tidy. On 30 April, HYPO’s evaluation suggested the Aleutians as a possible objective, and Layton reported that a Japanese offensive “in the Aleutian chain seems possible in late May.” On 14 May, according to Parker, Layton confused the preparations for a second “K” operation, a planned flying-boat reconnaissance of Pearl Harbor that would refuel from submarines at French Frigate Shoals, with the Midway preparations. The Pacific analysts were not infallible; what distinguished them was that they corrected course quickly as the traffic accumulated.

What was “AF”?

Japanese naval messages referred to places by geographic designators, usually two or three kana letters that American analysts transliterated into Roman letters. Designators beginning with A seemed to belong to the Hawaiian area and the central Pacific. The NSA’s short history of the battle notes that Rochefort and Layton reasoned from the A-series designators they had already seen assigned to Hawaiian locations; “AH,” for instance, had appeared in connection with Oahu. “AF” turned up in partially readable messages as early as 4 March 1942, according to Parker, the same day Japanese flying boats staging through French Frigate Shoals made a night raid on Oahu.

The first formal correlation of AF with Midway came from the Corregidor unit, CAST, on 13 March 1942; its report of 23 March listing recoveries for eighty-nine designators included “AF for Midway.” HYPO followed, and the Melbourne unit agreed in April. By the time the Midway traffic surged in May, then, the Pacific units had held the equation AF = Midway for two months. What changed in May was the stakes. A designator that had appeared in reconnaissance traffic was now the objective of the largest operation the Japanese navy had ever mounted, and the fleet’s dispositions depended on whether the identification was right.

There was also a definitional problem. Parker writes that OP-20-G in Washington treated AF as a communications designator rather than a geographic one, a distinction that “caused trouble later.” If AF was an address or call-up in the communications system rather than a place, then a message about operations “against AF” did not necessarily say where the attack would fall. That technical disagreement fed a larger one about Japanese intentions.

Washington and Pearl Harbor disagree

The Navy’s communications intelligence organization was run from Washington. OP-20-G was part of the Office of Naval Communications, headed in 1942 by Rear Admiral Joseph R. Redman; his younger brother, Commander John R. Redman, took charge of OP-20-G early that year, with Commander Joseph N. Wenger as a senior figure in its cryptanalytic work. Rochefort’s formal chain ran through them, though in practice he worked for Nimitz and briefed Layton daily. That arrangement produced friction long before Midway. Rochefort’s biographer Elliot Carlson, interviewed by NPR in 2011, described the Redman brothers as jealous of a subordinate who went around them.

Substance mattered as much as personality. Senior officials in Washington, Parker writes, sometimes produced assessments that were “perilously inaccurate,” at one point expecting an all-out attack on the West Coast “at any time.” On 15 May, NEGAT wrongly warned the Navy’s War Plans Division that the second K operation would be a large-scale attack on Hawaii. In his later article on the AF dispute, Parker found that between 15 May and 20 or 21 May OP-20-G and others in Washington, including the War Plans staff, considered that AF might be Samoa, the U.S. West Coast, or even Hawaii. Navy Region Hawaii’s history adds that “many in Washington remained convinced” the main blow would fall in the Aleutians.

Timing was disputed too. An unsigned 1944 account of the episode, later attributed to Safford and published by the Station HYPO historical site, says the Hawaii unit stuck to an estimate of 3 June for the opening of operations while the Navy Department found “nothing to substantiate 3 June.” That memorandum was written by a partisan of Rochefort from what its author admitted were stories told to him, and it must be read as advocacy. But the general shape it describes, with Pearl Harbor estimating an early-June opening and Washington favoring a later date, is consistent with the later histories.

Nimitz sided with Pearl Harbor. Parker states that the admiral sided with the intelligence advisors at HYPO and Melbourne about the true nature of the Japanese operation, and he did so while Washington was still unconvinced. He had already flown to Midway on 2 May to inspect its defenses personally, and through May he poured reinforcements onto the atoll. But to move the fleet he needed the agreement, or at least the acquiescence, of Admiral Ernest J. King, Commander in Chief, U.S. Fleet, in Washington, and King’s staff took its intelligence from OP-20-G. The question of AF was therefore not academic. It decided whether the three carriers would be committed to the central Pacific or held for other contingencies.

The water ruse

The idea of making the Japanese name AF themselves came out of a discussion in the dungeon in mid-May 1942. According to the account of Donald “Mac” Showers, then a junior officer at HYPO, as summarized by Parker, it was Jasper Holmes who proposed it. Holmes knew Midway’s fresh water came from a distillation plant, and he suggested that the island commander be instructed to send a plain-language message reporting a major breakdown of the plant. Finnegan speculated that the Japanese radio intelligence unit on Wake Island, which listened to American traffic, would pick it up. Showers recalled Rochefort’s reaction as “Very good, Jasper, very good.” Layton carried the idea to Nimitz, who approved it.

The mechanics mattered. The instruction to Midway went by the undersea cable between Oahu and Midway, which the Japanese could not intercept. Midway then sent the false report by radio to the Commandant, Fourteenth Naval District, in Honolulu, as though it were a genuine emergency. Layton’s memoir adds that a follow-up was sent in a strip-cipher system the Americans believed the Japanese had captured at Wake, to make the story more convincing; that detail rests on his later recollection.

The Japanese took the bait, and their report of it was intercepted in turn. Parker’s reconstruction from the surviving documents is precise. A Japanese intelligence report stated that “the AF (Midway) air unit” had sent a radio message to the Commandant of the Fourteenth District, dated it to the 20th, and summarized its content, including the line that “at the present time we have only enough water for two weeks.” The Melbourne unit decrypted and published the report on 22 May 1942. Midway lies east of the international date line and Tokyo west of it, so a message sent on 19 May in Midway would be dated 20 May in Japanese records; that is a likely explanation of why the dates in accounts vary between the 19th and the 20th, though it is an inference rather than something the documents state.

The fragments above are the wording of the Allied translation of the Japanese report as quoted in Frederick D. Parker’s article “Revisiting the Dispute over the Identification of ‘AF’ as Midway” (NSA, Cryptologic Quarterly). The parenthetical “(Midway)” may be the translator’s gloss rather than part of the Japanese text; Parker does not say. Ellipses mark omitted and illegible portions of the surviving copy.

Parker called the deception “completely successful” and concluded that the Melbourne message of 22 May “ultimately ended all controversy.” Later accounts drawn from Layton’s memoir add that the Japanese soon broadcast instructions to load additional water-distillation equipment for the invasion force, a detail that would have reinforced the conclusion. Surviving records show no written Washington response to the report, but after it the debate about the identity of AF disappears from the record.

The popular phrase “AF is short of water” is a fair paraphrase of the report and appears in official short histories, including NSA’s own. Navy Region Hawaii’s history states that HYPO intercepted the Japanese message; Parker’s documentary reconstruction credits Melbourne with the decryption that circulated. Both may be true in part, since the stations shared traffic and recoveries, but the published report Parker found came from Melbourne.

When: the date-time cipher and the last week of May

Knowing where was half the problem. Nimitz also needed to know when. The Japanese protected dates and times inside JN-25 messages with an additional cipher of their own, a special encipherment of the date-time information applied before it was encoded in the main system. Groups that looked like dates did not decode into dates. Accounts based on Layton’s memoir describe HYPO searching its printouts and punched cards for recurring five-digit sequences that might hide date-time groups, and credit Finnegan with discovering the method the Japanese used to lock them up.

The final solution came in the last week of May in a sustained effort by Finnegan, Wright and others. Layton’s memoir, as summarized in later histories, says that an intercept of 26 May containing orders for two destroyer groups escorting the invasion transports “really clinched” the pivotal date. The resulting estimate was that the Aleutians would be attacked on about 3 June and Midway on about 4 June, with the invasion to follow. Navy Region Hawaii’s history puts HYPO’s estimate for the invasion “around June 4.” The memoirs also describe Rochefort arriving late, exhausted, to a CINCPAC conference at which the estimate was presented; accounts differ on whether that meeting took place on 25 or 27 May.

There was a deadline the codebreakers could see coming. JN-25B was due to be replaced, and each replacement threatened to wipe out months of recovered code meanings. David Kahn’s history The Codebreakers records that the change to the new system, JN-25c, had been deferred from 1 April and then from 1 May, and came into effect on 28 May 1942; some other accounts give 1 June for the change that blinded the Allied stations. Each postponement, caused by the difficulty of distributing new books across the Pacific, bought readable traffic while Operation MI was being organized by radio; the change closed the window just as the Japanese forces sailed.

Nimitz pressed Layton for something more specific than a date. As the Naval History and Heritage Command’s director recounts in H-Gram 006, Nimitz insisted that Layton estimate where the Japanese carriers would be found. Layton predicted that they would be sighted bearing 325 degrees, 175 nautical miles from Midway, at about 0600 on 4 June. His estimate allowed for the weather, which the Japanese plan did not. On the morning of 4 June, a Midway PBY reported the carriers close to that point. Nimitz later remarked that Layton had been only “five degrees, five miles, and five minutes off”; H-Gram 006 notes that the actual error was somewhat larger than Nimitz’s generous summary, but the estimate was close enough to be operationally decisive.

1. 4 Mar 1942“AF” appears in partially readable Japanese messages (Parker); the same day, flying boats staging through French Frigate Shoals raid Oahu.
2. 13 Mar 1942The Corregidor unit (CAST) first correlates AF with Midway; its 23 March report lists eighty-nine designator recoveries.
3. 16 Apr 1942Imperial General Headquarters approves the Midway operation for early June; the Doolittle Raid follows on 18 April.
4. 2 May 1942Nimitz inspects Midway in person and orders its defenses strengthened.
5. 15 May 1942NEGAT warns Washington War Plans of a large-scale attack on Hawaii; OP-20-G considers Samoa, the West Coast or Hawaii as possible meanings of AF.
6. Mid-May 1942Holmes proposes the water ruse; Nimitz approves; Midway sends its false report by radio after instructions by cable.
7. 22 May 1942Melbourne issues its translation of the Japanese report on AF’s water shortage; the identification dispute ends.
8. 26–27 May 1942Date-time cipher solved; estimates of 3 June (Aleutians) and 4 June (Midway) presented to Nimitz.
9. 28 May 1942JN-25c takes effect (Kahn); TF16 sails from Pearl Harbor with Enterprise and Hornet.
10. 30 May 1942Yorktown sails with TF17 after emergency repairs.
11. 4 Jun 1942Japanese carriers sighted northwest of Midway near Layton’s predicted bearing, range and time.

Nimitz decides

The intelligence did not make the decision. It made a decision possible. Nimitz had at most three carriers: Enterprise and Hornet, back from the Doolittle Raid and a fruitless dash toward the Coral Sea, and the damaged Yorktown. Saratoga was on the West Coast after torpedo repairs and would not reach Pearl Harbor until 6 June. Against him, if HYPO was right, would come four Japanese fleet carriers with veteran air groups, supported by battleships, cruisers and an invasion force. If HYPO was wrong, and the real blow fell on Hawaii, the Aleutians or the South Pacific, sending the carriers to Midway could leave the actual target uncovered.

Nimitz committed everything. He ordered the carriers to a position northeast of Midway, code-named Point Luck, from which they could strike the Japanese carriers on their flank while the Japanese were occupied with the atoll. He reinforced Midway with aircraft and Marines until the islands were, in the phrase used in Chapter 3, an unsinkable carrier. He sent Yorktown into Dry Dock No. 1 at Pearl Harbor on 27 May for repairs that the yard workers completed in about two days, working around the clock, and she sailed on 30 May. Task Force 16, under Rear Admiral Raymond A. Spruance in place of the hospitalized Vice Admiral William F. Halsey, sailed on 28 May. Rear Admiral Frank Jack Fletcher’s Task Force 17 followed with Yorktown. The plan was issued as CINCPAC Operation Plan 29-42.

Nimitz’s instructions to Fletcher and Spruance told them to be governed by the principle of “calculated risk,” which he defined as avoiding exposure of their forces to attack by superior enemy forces without a good prospect of inflicting greater damage on the enemy. H-Gram 006 argues that this phrase is the key to the decision: Midway was a calculated risk, not a desperate gamble, precisely because the intelligence told Nimitz that he could strike first if the Japanese did not know where his carriers were. The edge depended on surprise, and surprise depended on the Japanese not discovering that their own plan had been read.

It is worth noticing what the decision did not depend on. Nimitz did not wait for certainty about the Japanese order of battle, and he did not demand a decrypted copy of the Japanese operation order, which HYPO never had. He acted on a structured estimate whose core elements, target, approximate timing, approach direction and a count of four or five carriers, had been tested against months of traffic. Samuel J. Cox, the Navy’s director of naval history, writes in H-Gram 006 that code-breaking was “only a part, albeit significant,” of the intelligence Nimitz used; traffic analysis, direction-finding and the record of the Coral Sea forecast all fed his confidence.

What the codebreakers did not know

The story of Midway intelligence is often told as if Nimitz held the Japanese plan in his hands. He did not. The intelligence was strong on the objective, good on the timing, sound on the carrier striking force, and weak or silent on much else. Understanding those limits is necessary to understand the battle itself, because many of the decisions made on 4 June were made in the dark.

The largest gap was Yamamoto’s own force. The Main Body, with the giant new battleship Yamato as fleet flagship, followed the carriers at a distance that kept it out of the battle. Its role, as Parker summarizes the Japanese strategy, was to wait west of the striking force and destroy the U.S. Pacific Fleet once it came out. Because it moved under radio silence and was not the subject of much readable traffic, its strength and position were poorly understood in Pearl Harbor. That uncertainty shaped Spruance’s caution on the night of 4–5 June, when he turned east rather than risk a night encounter with battleships he could not see, a decision examined in Chapter 7.

The second gap was tactical. Decrypts could not say where the Japanese carriers would be at 0700 on 4 June, which way they would turn, or when they would recover or rearm their aircraft. That had to come from search planes, and the American searches were thin. H-Gram 006 notes that on the morning of 4 June none of the PBY scouts had seen more than two carriers at once, while the intelligence estimate was of four or five. The commanders had to reconcile the estimate with incomplete sighting reports in real time, and the consequences of that reconciliation, from Spruance’s early launch to the scattered torpedo attacks, run through Chapter 4 and Chapter 5.

The third gap was created by the Japanese code change. Once JN-25c came into force at the end of May, the steady flow of fresh decrypts dried up. For the battle itself, the Americans relied far more on traffic analysis, direction-finding and plain-language or tactical intercepts. H-Gram 006 notes, for instance, that Nagumo’s order to the Tone No. 4 search plane to check for American carriers was sent in the clear and intercepted by U.S. radio intelligence. Such intercepts were useful, but they were not the strategic reading of the Japanese operation that had made the ambush possible.

The other side of the intelligence war

The American advantage was matched by Japanese failures, and the two cannot be separated. The Japanese plan rested, as H-Gram 006 puts it, on “inadequate intelligence” and an inaccurate picture of American intentions. Yamamoto assumed that the U.S. carriers would be at Pearl Harbor or in the South Pacific when Midway was struck and would come out only after the attack, into a trap. His surveillance and reconnaissance, Cox writes, were “poorly planned and executed.”

Several of the Japanese safeguards failed in ways that look worse in hindsight. The second K operation, the flying-boat reconnaissance of Pearl Harbor that might have revealed the carriers’ absence, was cancelled when the Japanese found American ships at French Frigate Shoals, where the flying boats were to refuel from submarines. The submarine cordon meant to detect the U.S. fleet leaving Hawaii arrived on station, H-Gram 006 notes, too late to see the American carriers pass. The submarine I-168 provided what Cox calls “accurate, and ignored, intelligence” on Midway’s readiness. In the war games before the operation, a Japanese officer playing the American side moved his carriers much as Nimitz actually did; the umpire ruled those actions “impossible.”

Nor did the Japanese suspect their main code. The water report was produced by their own radio intelligence, which was doing its job by listening to American traffic. The repeated postponement of the JN-25 change came from the practical difficulty of distributing new code and additive books across a vast area, not from any belief that the old books had been compromised. When the change came, it was routine. The irony of Midway is that the Japanese fleet was well defended against American ships and aircraft and almost undefended against the possibility that its own radio traffic was being read.

HYPO’s estimate of four or five carriers left room for Zuikaku, whose air group had not been rebuilt after Coral Sea; she stayed home. The four that sailed were the core of the force that had struck Pearl Harbor, and their record is collected on the Kido Butai at Midway page.

Secrecy, and the newspaper story that nearly gave it away

The value of the codebreaking depended on the Japanese not learning of it, and almost immediately after the battle it was put at risk at home. On 7 June 1942, the Chicago Tribune and other papers carried a story by the correspondent Stanley Johnston, who had been aboard Lexington at Coral Sea, reporting that the Navy had known the strength of the Japanese forces in advance. Most accounts trace it to a CINCPAC dispatch Johnston had seen on his way home. It did not mention codebreaking directly, but anyone who read it carefully could infer the source.

The government considered prosecution, and a federal grand jury was convened in August 1942; it declined to indict, in part because a trial would have required the Navy to confirm what it most wanted to conceal. The 1944 memorandum published by the Station HYPO site notes that the grand jury publicity broke on 8 August 1942 and that “in 6 days” the Japanese operations code was superseded. Whether that change was a reaction to the publicity or a routine replacement has never been established; there is no evidence the Japanese read the American press reports and concluded their code had been broken. The episode nonetheless shaped the Navy’s handling of communications intelligence for the rest of the war, and it partly explains why the men of HYPO could say little about their work for decades.

Credit, blame and the removal of Joe Rochefort

Nimitz did not hide where he thought the credit belonged. In the foreword to Parker’s history, he is quoted as giving communications intelligence a “major share of the credit for the victory at Midway.” H-Gram 006 records his summary of the battle as “essentially a victory of intelligence.” He recommended Rochefort for the Navy Distinguished Service Medal. The recommendation was turned down in Washington. According to Carlson’s account, King denied the award on the advice of staff members who disliked Rochefort, among them the Redman brothers and King’s chief of staff, Russell Willson, who had clashed with Rochefort years earlier. Nimitz gave Rochefort, Dyer, Wright and others letters of commendation instead.

Essentially a victory of intelligence.

— Adm. Chester W. Nimitz on the Battle of Midway, as quoted by Samuel J. Cox, Director of Naval History (NHHC H-Gram 006, Attachment 1)

Worse followed. In the autumn of 1942, OP-20-G reorganized the Pacific communications intelligence effort, and Rochefort was removed from HYPO. The details come largely from partisan sources and should be read with care. The 1944 memorandum states that John Redman sent Wenger messages describing Rochefort as uncooperative, one of them reading “Get rid of Rochefort at all costs,” and that the decision was made in Washington without consulting Nimitz. A captain named Goggins, described in that account as a friend of John Redman, was sent to Pearl Harbor and took over the unit, and Rochefort was detached shortly after Goggins arrived. Rochefort was ordered to Washington on what was presented as temporary duty and did not return.

He was then assigned to command a floating dry dock under construction on the West Coast, an advance base sectional dock designated ABSD-2, according to histories by Michael Smith and Holmes. He never served at sea again. Later in the war his skills were put to use once more as head of the Pacific Strategic Intelligence Group in Washington, a post the NHHC profile dates from 1944. The dispute between Rochefort and OP-20-G has been debated ever since. Defenders of Washington point out that NEGAT and the Melbourne unit contributed a large share of the recoveries and that centralization of a growing intelligence effort was reasonable in itself. Rochefort’s defenders answer that the manner of his removal, coming within months of the most important intelligence success of the war, looked very much like punishment for being right.

A medal forty-four years late

Rochefort retired in 1953 as a captain. Through the 1950s and 1960s, the secrecy surrounding communications intelligence meant the full story could not be told, and the published histories of Midway emphasized the aviators and the admirals. The codebreakers’ role became widely known only in the 1970s and 1980s, with the declassification of wartime records and the publication of memoirs by Holmes (1979) and Layton (1985). Rochefort died on 20 July 1976, before most of that material appeared.

The campaign for his medal began soon after the war, with one of his own former intelligence officers, and was revived in the 1980s by Rear Admiral Donald “Mac” Showers, the junior HYPO officer whose recollection of the water-ruse meeting Parker later used. Showers, then working for Director of Central Intelligence William Casey, had connections to Secretary of the Navy John Lehman. Carlson told NPR that Showers “brought the muscle.” The effort succeeded. On 30 May 1986, in the Roosevelt Room of the White House, President Ronald Reagan presented the Distinguished Service Medal to Rochefort’s family for his role in breaking the Japanese code. The Reagan Library’s photo record lists those present, among them Vice President George Bush, Secretary of Defense Caspar Weinberger, Admiral William Crowe, Casey, Lehman, Showers and Pineau.

NSA inducted Rochefort into its Hall of Honor in 2000. The Naval History and Heritage Command’s profile of Rochefort also states that he posthumously received the Presidential Medal of Freedom. This chapter could not confirm the date or citation of that award in a primary record; the NHHC profile places it alongside the 1986 Distinguished Service Medal, while some other accounts associate it with 2000, the year of the Hall of Honor induction. NSA’s own biographical note, which also misdescribes the 1986 medal, mentions no Medal of Freedom. Until a citation or presidential record is found, the Distinguished Service Medal of 1986 and the Hall of Honor of 2000 are the honors that can be firmly documented.

The basement where the work was done survives. Navy Region Hawaii notes that the dungeon, vacant or used for storage for decades after the unit left in 1943, has been rediscovered and marked, with plaques outside each stairway recording what happened there. Rochefort’s name has been given to awards for Navy information warfare officers. The fuller story of the unit, its people and its later history as Fleet Radio Unit Pacific is told on the Station HYPO special-unit page.

Midway did not show that codebreaking wins battles by itself. It showed that intelligence becomes decisive when a commander will act on an estimate before it is certain. Rochefort and Layton told Nimitz where, when and how many; Nimitz turned that into a position on the ocean. What happened when the two fleets met there, beginning with the attack on the atoll in Chapter 3, was decided by men who had never heard of Station HYPO.