By Scott Shane
The New York Times
historian's conclusion is the first serious accusation that communications
intercepted by the N.S.A., the secretive eavesdropping and code-breaking
agency, were falsified so that they made it look as if North Vietnam had
attacked American destroyers on Aug. 4, 1964, two days after a previous clash.
President Lyndon B. Johnson cited the supposed attack to persuade Congress to
authorize broad military action in
The N.S.A. historian, Robert J. Hanyok, found a pattern of translation mistakes that went uncorrected, altered intercept times and selective citation of intelligence that persuaded him that midlevel agency officers had deliberately skewed the evidence.
Mr. Hanyok concluded that they had done it not out of any political motive but to cover up earlier errors, and that top N.S.A. and defense officials and Johnson neither knew about nor condoned the deception.
Hanyok's findings were published nearly five years ago in a classified in-house
journal, and starting in 2002 he and other government historians argued that it
should be made public. But their effort was rebuffed by higher-level agency
policymakers, who by the next year were fearful that it might prompt
uncomfortable comparisons with the flawed intelligence used to justify the war
M. Aid, an independent historian who has discussed Mr. Hanyok's
material is relevant to debates we as Americans are having about the war in
Mr. Aid's description of Mr. Hanyok's findings was confirmed by the intelligence official, who spoke on condition of anonymity because the research has not been made public.
Both men said Mr. Hanyok believed the initial misinterpretation of North Vietnamese intercepts was probably an honest mistake. But after months of detective work in N.S.A.'s archives, he concluded that midlevel agency officials discovered the error almost immediately but covered it up and doctored documents so that they appeared to provide evidence of an attack.
than come clean about their mistake, they helped launch the
Asked about Mr. Hanyok's research, an N.S.A. spokesman said the agency intended to release his 2001 article in late November. The spokesman, Don Weber, said the release had been "delayed in an effort to be consistent with our preferred practice of providing the public a more contextual perspective."
Mr. Weber said the agency was working to declassify not only Mr. Hanyok's article, but also the original intercepts and other raw material for his work, so the public could better assess his conclusions.
intelligence official gave a different account. He said N.S.A. historians began
pushing for public release in 2002, after Mr. Hanyok included his
Mr. Aid said he had heard from other intelligence officials the same explanation for the delay in releasing the report, though neither he nor the intelligence official knew how high up in the agency the issue was discussed. A spokesman for Gen. Michael V. Hayden, who was the agency's. director until last summer and is now the principal deputy director of national intelligence, referred questions to Mr. Weber, the N.S.A. spokesman, who said he had no further information.
historians believe that even without the
Robert S. McNamara, who as defense secretary played a central role in the
think it's wrong to believe that Johnson wanted war," Mr. McNamara said.
"But we thought we had evidence that
Mr. McNamara, 89, said he had never been told that the intelligence might have been altered to shore up the scant evidence of a North Vietnamese attack.
"That really is surprising to me," said Mr. McNamara, who Mr. Hanyok found had unknowingly used the altered intercepts in 1964 and 1968 in testimony before Congress. "I think they ought to make all the material public, period."
supposed second North Vietnamese attack, on the American destroyers Maddox and
C. Turner Joy, played an outsize role in history. Johnson responded by ordering
retaliatory air strikes on North Vietnamese targets and used the event to
persuade Congress to pass the
authorized the president "to take all necessary steps, including the use
of armed force," to defend
Not all the details of Mr. Hanyok's analysis, published in N.S.A.'s Cryptologic Quarterly in early 2001, could be learned. But they involved discrepancies between the official N.S.A. version of the events of Aug. 4, 1964, and intercepts from N.S.A. listening posts at Phu Bai in South Vietnam and San Miguel in the Philippines that are in the agency archives.
One issue, for example, was the translation of a phrase in an Aug. 4 North Vietnamese transmission. In some documents the phrase, "we sacrificed two comrades" - an apparent reference to casualties during the clash with American ships on Aug. 2 - was incorrectly translated as "we sacrificed two ships." That phrase was used to suggest that the North Vietnamese were reporting the loss of ships in a new battle Aug. 4, the intelligence official said.
The original Vietnamese version of that intercept, unlike many other intercepts from the same period, is missing from the agency's archives, the official said.
The intelligence official said the evidence for deliberate falsification is "about as certain as it can be without a smoking gun - you can come to no other conclusion."
Despite its well-deserved reputation for secrecy, the N.S.A. in recent years has made public dozens of studies by its Center for Cryptologic History. A study by Mr. Hanyok on signals intelligence and the Holocaust, titled "Eavesdropping on Hell," was published last year.
Two historians who have written extensively on the Tonkin Gulf episode, Edwin E. Moise of Clemson University and John Prados of the National Security Archive in Washington, said they were unaware of Mr. Hanyok's work but found his reported findings intriguing.
"I'm surprised at the notion of deliberate deception at N.S.A.," Dr. Moise said. "But I get surprised a lot."
Prados said, "If Mr. Hanyok's conclusion is correct, it adds to the tragic
aspect of the Vietnam War." In addition, he said, "it's
new evidence that intelligence, so often treated as the Holy Grail, turns out
to be not that at all, just as in