What U.S. Codebreakers Heard From Tokyo About Amelia Earhart

On July 7, 1937, five days after Amelia Earhart’s Lockheed Electra went silent somewhere near Howland Island, Japan’s foreign minister sent a cable to his consulate in Honolulu. Japan wished “to express her most profound condolences for the Earhart tragedy,” Koki Hirota wrote, on behalf of both government and people. The South Seas administration had ordered every ship, wireless station and sentinel near the Marshall Islands, “the suspected scene of the accident,” to help find the plane. The Imperial Navy was ready to search that region if the United States wanted it to.

The cable was also being read in Washington. A translation of that message went to the War Department’s intelligence staff on July 17, under a cover memo from the Army’s Signal Intelligence Service. It stayed classified until the Earhart records release that began last November.

That cable, and a handful of others like it, are the part of the release that bears most directly on the most durable Earhart theory of all. We read the intercepts, the codebreakers’ oral histories and the Army’s own files. Everyone has seen the headlines saying the files settled the capture question. Almost nobody has read the pages that actually test it.

The theory the files were supposed to answer

The release contains its own summary of the belief. In May 1965, a Weather Bureau satellite manager named J. Gordon Vaeth wrote to Adm. Husband E. Kimmel, the retired Pacific Fleet commander. Much of the speculation about Earhart, Vaeth wrote, held that “she was on an intelligence mission for the United States” and “was actually picked up by the Japanese and transported to Saipan.” Did Kimmel know anything that would support or refute it? Someone wrote across the bottom of the copy in the file, apparently initialed by Kimmel, “Answered. No knowledge.”

Fred Goerner’s 1966 book The Search for Amelia Earhart made that story a bestseller. In its usual form, Earhart and Fred Noonan, who was plotting her course, came down in the Japanese-held Marshall Islands, were taken to Saipan, and died in custody. Washington knew and kept quiet.

When President Trump ordered the records out on Sept. 26, 2025, that was the question a lot of people wanted answered. The National Archives posted 10,196 pages in five batches between Nov. 14, 2025, and Jan. 9, 2026. Within days, some outlets were reporting that the files had “put an end” to the capture theory, without naming a document that did it.

No single page in the release does that. But the intelligence files do something more interesting. They show what American eavesdroppers could hear from the Japanese side in July 1937, and what they heard was not a cover-up.

What Tokyo told its own diplomats

The two Signal Intelligence Service memos, dated July 14 and July 17, 1937, each forward three translated Japanese diplomatic messages to G-2, the Army’s intelligence staff, over the signature of Maj. W. S. Rumbough of the Signal Corps. The release includes one Earhart translation from each batch. The release files them alongside a reference set of Japanese diplomatic messages from the “Red” cipher machine, 1934 to 1938.

The first, S.I.S. No. 480, is from Ambassador Hiroshi Saito in Washington to Tokyo, dated July 5. It reports the American answer to Japan’s offer of help. A State Department official, whose name the translators left in a Japanese rendering marked as a transliterated English word, had told the embassy that the U.S. government appreciated the offer “very much” and wanted Japan’s cooperation. The plane’s last message, he said, put it “somewhere about two hundred miles north of Howland Island.” After that, nothing. Since the Navy had made Honolulu the search headquarters, Japan could get the latest information there.

That matches the State Department’s own paper trail. The National Archives documented in 2017 how the Japanese embassy phoned Joseph Ballantine of the Far Eastern division on July 5 to offer “radio stations and warships in the Marshall Islands,” and how Ballantine checked with the chief of naval operations before replying. The garbled name in the intercept is almost certainly Ballantine’s. Our read, not the document’s.

The second, S.I.S. No. 485, is Hirota’s July 7 condolence cable, sent to Honolulu and repeated to Washington. Nothing in either message reads like a government managing a secret. They read like a foreign ministry offering help near its own closed territory and asking where to look.

The phrase capture theorists will point to

Hirota’s cable does contain the line that capture researchers will quote. It calls the Marshalls “the suspected scene of the accident.”

That framing isn’t new to researchers. Capture-theory writers have argued for years, citing a July 5, 1937, Japanese navy message they say was found at the National Archives, that Tokyo placed her in the Marshalls far too quickly to be guessing. Hirota’s cable repeats the same idea two days later.

Read against the rest of the file, though, the phrase has a simpler source. Japan’s first offer on July 5 was built around its Marshall Islands radio stations and warships. That was the help it had to give. The Marshalls sit several hundred miles northwest of Howland, and the only position the Americans shared put the plane north of the island. The cable proves Tokyo was thinking about the Marshalls. It doesn’t show Tokyo knew anything.

These are two messages out of the Red traffic for that summer. The Foreign Ministry also wasn’t the Imperial Navy, which ran the Mandates and didn’t always tell the diplomats what it was doing. Two intercepts can’t clear the Navy.

“The name Earhart never came up”

The Navy’s codebreakers are the next witnesses in the file. The last batch, released Jan. 9, includes an excerpt from the Naval War College’s tape-recorded sessions with Capt. Joseph Finnegan, a Japanese linguist and cryptanalyst who served at the Navy’s intercept station at Cavite in the Philippines and later in the Pearl Harbor unit that broke Japanese naval codes.

Asked whether the Cavite unit was reading traffic between Tokyo and the Japanese Mandates in 1937, Finnegan answered, “A little.” Then: “We were reading some traffic between Tokyo and the Mandates. But the name Earhart never came up.” He added that Jack Holtwick, who was at Cavite when she disappeared, agreed there had been no communications intelligence about her. As for rumors that the Japanese had “taken care of” her, “at Saipan probably,” Finnegan called them “wild opinions,” and when asked whether they persisted at Pearl Harbor during the war, he said, “They’ve never died.”

That is the strongest line in the release against the capture story, and it has limits. Finnegan arrived at Cavite shortly after the disappearance, by his own account, and “a little” traffic is not all of it.

He also flagged the biggest gap himself. In a short “Amelia Earhart Note” attached to the sessions, he wrote that 22 tons of Japanese records were removed from Saipan after the war, most “never microfilmed” and possibly never read by anyone competent to judge them. By one estimate he cited, translating them would take about 150 years. He had been assured there was no Earhart material the Naval History Division couldn’t reach. But if an answer exists anywhere, those crates are where the record says it would be, and the release doesn’t touch them.

An Army colonel who was sure, and partly wrong

The Army’s Military Intelligence Division file holds a sharper denial. On Nov. 1, 1938, Col. H. H. C. Richards of the Air Corps forwarded to G-2 a letter from a man in Australia claiming the Japanese had shot Earhart down. “It is definitely known that such is not the case,” Richards wrote. Army personnel on Howland had heard her radio, and from the signal strength she had passed “some 50 miles or less” from the island. She then said she was turning north, grew fainter, “until finally she stated she was out of gas.”

Richards’s certainty outruns the record. The Coast Guard cutter Itasca‘s radio log, whose operator’s papers are also in the release, has her reporting that gas was running low and, in her last logged message, running “north and south” on a line of position. It has no message saying she was out of gas. A colonel’s paraphrase a year later is not evidence of where she went down. It does show that in 1938 the Army’s working assumption was an ocean ditching near Howland, and that it treated the Japanese angle as a crank letter.

The same file opens with one of those. On July 6, 1937, a radio listener in Dennison, Ohio, wrote to report a faint voice claiming to be 150 miles southeast of Howland, asking for food and water. The War Department thanked him. A 1983 NSA oral history adds a similar story at secondhand, a Navy radio operator near Hawaii who was told “Amelia Earhart’s plane is calling” and never got an answer. The retired NSA mathematician telling it said he had never heard it from anyone else.

What the record settles

The files don’t end the capture theory, and they don’t prove it. What they add is the view from the listening posts. In July 1937 the United States could read at least some of Japan’s diplomatic and Mandate traffic. In the messages now public, Tokyo offers help, names the Marshalls because that is where its ships were, and asks nothing about a prisoner. The Navy linguist who read Mandate traffic that year said her name never appeared. The Army’s position a year later was that she ran out of fuel near Howland.

What they leave open is the 22 tons of Saipan paper and the Japanese navy’s own files, which no American release can declassify. If Earhart was ever in Japanese hands, that is where it would be written down.

Primary source: Signal Intelligence Service, translation of Japanese message S.I.S. No. 485 (Hirota to Honolulu, July 7, 1937), forwarded July 17, 1937. Full context: National Archives, Records Related to Amelia Earhart. The intercepts are two pages each. Read them yourself.