Gulf of Tonkin, 1964: the attack that did not happen

Series: How minds are moved · The record: documented campaigns · 2 of 7

At 11:36 on the night of 4 August 1964, President Lyndon Johnson went on American television and told the nation that North Vietnamese boats had attacked US destroyers in international waters for the second time in three days, and that American planes were already on their way to strike back. Three days later Congress gave him broad authority to use force in Southeast Asia: 416 to 0 in the House, 88 to 2 in the Senate. That resolution became the legal basis for a war in which more than 58,000 Americans and, by most estimates, well over a million Vietnamese died.

The second attack did not happen. That is not my opinion or a theory from a fringe website. It is the conclusion of the US National Security Agency’s own historian, in a study the agency classified for years and released at the end of 2005. My OSINT shelf holds the State Department and CIA side of that year, and oral histories from the officials who were in the room. Here is how a non-event was turned into a cause for war.

Contents · 10 sections

TL;DR #

• On 2 August 1964 North Vietnamese torpedo boats did attack the destroyer USS Maddox. That event was real.
• On 4 August a second attack was reported, on a dark, stormy night, from radar and sonar contacts. Within hours the commander on the scene was cabling that the reports looked “doubtful”.
• Washington retaliated anyway and used the event to pass the Gulf of Tonkin Resolution.
• The NSA historian Robert Hanyok concluded that the agency’s signals reporting was skewed to support the claim of an attack. He described “an active effort to make SIGINT fit the claim”, and found no evidence that Johnson or McNamara knew the reporting was skewed.
• In 1995 North Vietnam’s General Giap told Robert McNamara directly that on 4 August “there was absolutely nothing”.

What happened, as far as the record shows #

In the summer of 1964 the US Navy was running electronic-intelligence patrols, code-named DESOTO, along the coast of North Vietnam. At the same time, South Vietnamese commandos, under a secret US-supported programme called OPLAN 34A, were raiding North Vietnamese coastal targets. Hanoi plausibly saw the destroyer and the raids as one operation.

On 2 August three North Vietnamese torpedo boats attacked the Maddox. The destroyer and US aircraft drove them off. No American was killed. Johnson chose not to retaliate and sent a second destroyer, the Turner Joy, to continue the patrol.

On the night of 4 August, in bad weather, the two ships reported that they were under torpedo attack. They fired for hours at radar targets. James Stockdale, a Navy pilot who flew over the scene that night and later became a vice admiral, said afterwards that he had had the best seat in the house to see any boats, and saw none: only dark water and the destroyers’ own gunfire.

Within hours the commander of the patrol, Captain John Herrick, sent a cable that is now well known. A review of the action, he said, made many of the reported contacts and torpedoes “appear doubtful”, and freak weather effects on radar and “overeager sonarmen” might account for much of it. He recommended a full evaluation before any further action.

The evaluation did not come first. The airstrikes did.

How it was presented at the time #

The public story had three parts, and all three were stronger than the evidence.

Certainty. Johnson’s speech and Defense Secretary Robert McNamara’s briefings presented the second attack as established fact. Doubts expressed in the message traffic from the ships were not mentioned.

Innocence. Officials described the destroyers as being on routine patrol in international waters. McNamara told senators that the Navy had played no part in, and was not associated with, any South Vietnamese actions. The link between the patrols and the 34A raids was not explained to Congress or the public. The oral histories on my shelf from officials of the time, among them Under Secretary of State George Ball and the State Department’s intelligence chief Thomas Hughes, are where I would start for the insiders’ view of that summer.

Urgency. The resolution itself had been drafted months earlier and was waiting for an occasion. Congress debated it for less than a day of floor time in the House and two days in the Senate. Only two senators, Wayne Morse and Ernest Gruening, voted against. Morse warned that the resolution was a “predated declaration of war”.

What the NSA’s own history found #

Robert Hanyok, a historian at the National Security Agency, studied the agency’s records and intercepts for an article in its classified journal Cryptologic Quarterly around 2001. It was released to the public at the end of 2005, after the New York Times reported on it. His conclusions are blunt for an official history:

• No attack took place on 4 August.
• The signals reporting was skewed: what reached the President and McNamara was a selection that supported the claim of an attack. In his words, “the overwhelming body of reports, if used, would have told the story that no attack occurred”.
• He found no evidence that Johnson or McNamara knew the reporting was skewed, and no “smoking gun” proving a deliberate cover-up, but he described “an active effort to make SIGINT fit the claim of what happened during the evening of 4 August”.

The NSA also published a dissenting essay by another of its staff, defending the original reading. The National Security Archive, which pressed for the release, noted that key pieces of the original evidence are missing from the files. The official record therefore contains its own disagreement, and readers can see both sides.

And in November 1995 McNamara, by then old and openly regretful, travelled to Hanoi and asked General Vo Nguyen Giap what had happened on 4 August. According to the reports of that meeting, Giap answered: “Absolutely nothing.” McNamara said that settled the question for him. Giap added his own interpretation, that the patrols were meant to provoke; McNamara rejected that part. Both men had reasons to tell the story their way; on the narrow factual point, they agreed.

Mistake or manipulation? #

Here the record allows several readings, and I will not choose for you.

• One reading: a genuine confusion at sea at night, followed by a political leadership that wanted to look strong in an election year and did not wait for the facts.
• Another: an intelligence system that, under pressure, selected the evidence that fitted what the leaders already believed, the pattern Hanyok describes.
• A third, held by some critics: a deliberate provocation and deception. The documented evidence supports the first two more strongly than the third; the pre-drafted resolution shows the political intent existed before the incident.

What is not in doubt is the outcome: a reported attack that did not occur was presented as certain, and Congress gave the war powers on that basis.

What it felt like from the inside #

For most Americans on 5 August 1964, there was nothing to question. The President said American sailors had been attacked. The newspapers repeated it with maps and arrows. Almost every member of Congress voted the same way. If you were a sailor on the Maddox, you were sure something had happened that night, and so were the radarmen calling contacts. The feeling of certainty was real. The thing it was attached to was not.

What it could and could not do #

The incident did not by itself cause the Vietnam War; US involvement had been growing for years. What it provided was a legal instrument and a moment of national unity. When the doubts surfaced, years later in Senate hearings in 1968 and in the Pentagon Papers in 1971, the war was already enormous. Congress repealed the resolution; the repeal took effect in January 1971. A correction that arrives after the decision does not undo the decision.

How to notice it #

• “Under attack” plus “already retaliating” in the same announcement. When the response is announced before an investigation could have been finished, ask what the evidence was.
• Pre-written authorisations. A law or resolution that appears suspiciously fast was probably waiting for its occasion.
• “Routine” and “international waters”. Ask what else was happening in the area at the time.
• Lopsided votes on fresh facts. Near-unanimity in the first days is a sign of emotion, not of verification.
• Field doubts that never reach the public. Look for what the people on the scene said in the first hours; it often differs from the evening statement.

Frequently asked questions #

Did anything happen in the Gulf of Tonkin at all? #

Yes. On 2 August 1964 North Vietnamese boats attacked the USS Maddox. The disputed event is the second attack, reported on 4 August, which the NSA historian concluded did not happen.

Did Johnson know the second attack was doubtful? #

Herrick’s doubts reached the Pentagon within hours, and McNamara asked the Pacific command to confirm that an attack had really taken place before the strikes went ahead; the confirmation he received was based on the signals reporting Hanyok later called skewed. Johnson reportedly joked to an aide days later that the sailors might have been shooting at flying fish or whales; that remark comes from later recollections, not a recording. What the President was shown from signals intelligence was, according to Hanyok, a skewed selection; Hanyok found no evidence that Johnson or McNamara knew it was skewed.

Is the NSA study reliable, given that it came from the government itself? #

It is an internal official history that criticises its own agency, which is unusual. The agency also released a dissenting essay. Combined with Herrick’s cable, Stockdale’s account and Giap’s statement, the evidence points in one direction, which is why most historians accept the conclusion.

Why does this matter today? #

Because it shows that you do not need to invent an event from nothing. You only need a confused event, a selection of the evidence, and an announcement that is more certain than the facts.

The lesson #

The Gulf of Tonkin case is the clearest documented example of a war authorised on an event that did not happen, and the most instructive part is how ordinary the mechanism was. Nobody needed to forge a battle. A stormy night produced ghosts on a radar screen, the doubts stayed inside the message traffic, the confirming fragments went up the chain, and a resolution that had been waiting in a drawer found its moment. When a government is certain within hours, the most useful question is the one Captain Herrick asked that night: has anyone actually checked?