# --------------------------------------------------------------------------- # SEARCH HEADLINE (for search engines and AI answer tools; render as , # <h1>, og:title, and JSON-LD). Rules: # 1. It states the finding, short and specific, in plain words. # 2. It may only state a claim that is established or refuted at HIGH confidence. # 3. `headline_claim` names that claim, so the link can be checked mechanically. # 4. If that claim's confidence or state changes, the headline must change. # 5. Publish only after the claim's anchors have been read directly (anchor_checked: primary). # 6. Plain words only. Name the belief being refuted AND what replaces it, and # explain any term a general reader would not know (in the summary if not the title). # 7. Search wording. A headline should carry the name people actually search for (here # "Gulf of Tonkin Incident"), the date or year, and the source of the finding, so a # reader can judge it before clicking. Aim for 70-95 characters; the first 60 must make # sense on their own because search results cut off there. The summary should open with # the question people ask ("Did ... happen?"), answer it in the first clause, and stay # under 200 characters. `url_slug` is lowercase words joined by hyphens. # headline_status values: draft | review | published | parked # review = meets rules 2, 3 and 5; waiting for the owner to approve. # --------------------------------------------------------------------------- headline: "Gulf of Tonkin Incident: The August 4, 1964 Attack Never Happened, NSA's Own Study Found" headline_claim: tonkin-aug4-attack-occurred headline_status: draft # set back to draft 2026-10-02 by owner: open dig, published as such; return to review when its anchors are read assessment: question: "Did a second attack on US destroyers happen in the Gulf of Tonkin on 4 August 1964?" answer: "no" headline: "No. The second attack did not happen (high confidence). What senior officials knew about that is unsettled." text: > NSA's own historian re-read the intercepts and concluded that no attack took place that night. The intercept reports describe aircraft and salvage of boats damaged on 2 August, not an attack on the ships, and doubts reached the White House within hours. The 2 August clash with USS Maddox did happen. Whether senior officials knew the 4 August attack had not occurred when they sought the Resolution is a separate question, and the documents read support only that they knew of doubts. Two limits: the refutation rests largely on Hanyok's re-analysis, which we read but did not repeat, and the original Vietnamese text of the key report cannot be located. key_points: - "The intercept reports describe aircraft and the salvage of boats damaged on 2 August, not an attack on the ships." - "The first clash, on 2 August, did happen." - "Whether officials knew the second attack had not happened is a separate question, and it is unsettled." basis: [tonkin-aug4-attack-occurred, tonkin-aug2-maddox-engaged, tonkin-senior-doubts-documented, tonkin-senior-knowledge] would_settle: "New intercepts, ship logs or North Vietnamese records showing boats present and firing on 4 August, or the missing original text showing it concerned 4 August." search_summary: > Did a second Gulf of Tonkin attack happen on August 4, 1964? No. NSA's own historian found it never happened, though the August 2 clash was real. What officials knew is still debated. url_slug: gulf-of-tonkin-incident-no-second-attack-august-4-1964 subject: gulf-of-tonkin title: "Gulf of Tonkin Incidents (August 1964)" description: > Two separate events are routinely fused: the 2 August 1964 clash between USS Maddox and North Vietnamese torpedo boats, and the reported 4 August attack that led to the Tonkin Gulf Resolution. Stratah keeps them apart. The first happened. The second, on the declassified record, did not. Three further questions are kept separate again: how the false 4 August picture was assembled inside NSA, what senior officials were told and when, and what they knew. The first two have documented answers. The third is a genuine gap, and it is where conspiracy claims and the evidence part ways. This dig is also a calibration case: a real, documented failure of government reporting, filed strictly on its own evidence so that it neither inflates nor excuses claims in other digs. # --------------------------------------------------------------------------- # VERIFICATION LEVELS (governance, not a claim). # anchor_checked (per claim): no | secondary | primary # no not checked, or recalled # secondary agreed across secondary sources; the primary document was not opened, # or was opened only in part # primary the primary document was opened and the cited wording read directly # `# V:` lines name what was read. Quotations marked [read] were checked against the # page text; the pages are scans, so OCR noise is possible (see sources/MANIFEST.yaml). # # STATUS as of 2026-10-02 (second verification pass). Sources READ DIRECTLY this pass: # [read] Hanyok study, 56 pages (hosted by the National Security Archive; NSA's own servers # returned HTTP 403): src-hanyok-2001 # [read] White House staff meeting memorandum, 5 Aug 1964: src-bundy-memo-1964 # [read] NSA SIGINT intercept reports, 9 pages: src-intercepts-1964 # [read] LBJ-McNamara call transcriptions (transcribed by Prados): src-lbj-tapes-transcripts # [secondary] Prados essay quoting footnoted primary documents: src-nsarchive-essay-2004 # NOT yet read: the Herrick message itself; the 28 Aug 1964 chronology; FRUS 1964-68 vol. I # (only one document read); the Congressional floor debate; SNIE 50-2-64; the audio of the tapes. # # QUANTITIES USED: # Tonkin Gulf Resolution ......... passed 7 Aug 1964; House 416-0, Senate 88-2 (Morse and Gruening # against); signed 10 Aug 1964 (secondary, several sources agree) # SIGINT withheld ................ "nearly 90 percent of all available SIGINT" on 4 August was kept # out of the post-attack summary reports and the October 1964 final # report [read, Hanyok]. Study reviewed 122 relevant SIGINT # products; NSA's official version used 15 published reports; # the six products used in the chronology are "a bit more than # 10 percent of the total available" [read, Hanyok]. # Herrick message ................ report SENT at 1:27 p.m. Washington time on 4 August # (Washington, on daylight time, is 11 hours behind the Tonkin Gulf; the Prados essay says 12), per a 28 Aug 1964 # chronology quoted by Prados [secondary]. CORRECTION: the first # draft said "received" at that time. CLOCK NOTE: Washington was on # daylight time, so it was 11 hours behind the Gulf, not 12; Prados # converts at UTC-5, so a Washington clock time may be an hour out. # McNamara-LBJ call .............. 9:43 a.m. Washington time, 4 August [read, transcription] # Bundy staff meeting ............ 8 a.m., 5 August 1964 [read] # Giap ........................... confirmed 2 Aug to McNamara in Hanoi, Nov 1995 (press); at a June # 1997 Hanoi conference said the 2 Aug response was ordered by a # local naval command and that nothing happened on 4 Aug # (Prados, an attendee) [secondary] # --------------------------------------------------------------------------- divergence_note: > Evidence and adoption diverge on three points. On 4 August, historians agree no attack occurred, while the public memory of 1964 is of an attack. On method, the popular framing ('false flag', 'manufactured evidence') is stronger than the record: NSA's own historian found that only the SIGINT supporting the attack claim was passed up, calls the actions deliberate, but writes that this was not conspiratorial and involved no manufactured evidence or collusion at all levels. On knowledge, the documents show doubts reaching the White House within hours and a senior adviser welcoming the events as a justification, while the same documents show that adviser still believed an attack had happened. That gap is the product. claims: - id: tonkin-aug2-maddox-engaged state: established statement: > On 2 August 1964 North Vietnamese torpedo boats attacked USS Maddox in the Gulf of Tonkin; the Maddox and supporting aircraft returned fire. evidence_class: primary_text confidence: high statement_kind: judgment # assigned by rule 2026-10-02, to be reviewed evidential_weight: 5 # of 5; 5 allowed because anchor_checked is primary anchor_checked: primary # Hanyok study and intercept reports read directly adoption_weight: 5 # of 5, shown separately; never combined with evidential_weight adoption_note: > Not polled. Universal among historians; Hanoi's own account agrees (Giap). would_change_if: > records from either side showing the 2 August clash did not occur as described anchor: nodes: - {node: tonkin-1964-08-02-maddox-fires-three-warning-rounds-north-vietnamese, verb: supports} type: contemporaneous-records-plus-adversary-account description: > Hanyok writes that on 2 August "the SIGINT system performed admirably when it provided sufficient warning to the Maddox to allow it to defend itself against the attack by the three DRV torpedo boats" [read]. Prados reports photographs of the boats and a dud shell kept by Admiral Moorer (secondary). Giap confirmed the 2 August action; at a 1997 conference he said it was ordered by a local naval command and not by Hanoi (secondary). accessible: true basis: > # V: Hanyok study read; intercept reports read; Giap via Prados and press. # TO REACH A FULLER CHECK: the Navy's own action reports (Naval History and Heritage Command). Established on records from both sides. Giap's statement is testimony and is corroboration only, never the anchor. See tonkin-maddox-warning-shots for a documented omission about the start of the engagement. - id: tonkin-maddox-warning-shots state: established statement: > On 2 August the Maddox fired three warning rounds at the approaching North Vietnamese boats before they attacked. The Johnson administration did not report this and said the Vietnamese boats fired first. evidence_class: interpretive confidence: moderate statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 3 # of 5 anchor_checked: primary # Hanyok's account read; the Navy records he draws on were not adoption_weight: 2 # of 5 adoption_note: > Not measured. Little known to the public. would_change_if: > Navy action reports or the Maddox log showing the warning rounds were not fired, or were reported at the time next_step: > Read the Maddox action report and the Navy's official history of the engagement; log whether the warning rounds appear. anchor: nodes: - {node: tonkin-1964-08-02-maddox-fires-three-warning-rounds-north-vietnamese, verb: supports} type: historian-account-of-records description: > Hanyok: "At about 1505G, the Maddox fired three rounds to warn off the communist boats. This initial action was never reported by the Johnson administration, which insisted that the Vietnamese boats fired first." [read] accessible: true basis: > # V: one author's account, read directly; the underlying Navy records not read. Filed as interpretive because it is a scholar's narration of records we have not seen. It shows the 2 August engagement began less one-sidedly than the administration said, and it does NOT undo tonkin-aug2-maddox-engaged: the boats did attack. - id: tonkin-34a-raids-preceded state: established statement: > Covert OPLAN 34-A raids on North Vietnamese coastal targets, including the islands Hon Me and Hon Ngu, preceded the Maddox engagement (about half past midnight on 31 July) and were followed by another raid on the night of 3/4 August. The Maddox's mission was to collect North Vietnamese radio and radar emissions, expected to rise after a raid. Hanoi associated the destroyer with the raids. evidence_class: primary_text confidence: moderate statement_kind: judgment # assigned by rule 2026-10-02, to be reviewed evidential_weight: 3 # of 5; capped at 4 until the primary documents are read anchor_checked: secondary # McNamara's own recorded words read; operational records not adoption_weight: 4 # of 5, shown separately; never combined with evidential_weight adoption_note: > Not polled. Accepted by historians; less widely known to the public. would_change_if: > records showing the raids were not 34-A, or that Hanoi did not connect the destroyer to them next_step: > Read SNIE 50-2-64 (May 1964) and the declassified OPLAN 34-A documents; confirm the raid dates and who controlled the boats. anchor: nodes: - {node: tonkin-1964-07-30-34-a-raid-on-hon-me-and, verb: supports} - {node: tonkin-1964-08-03-second-34-a-raid-night-of-3, verb: supports} type: declassified-operational-record description: > On 3 August McNamara told LBJ the link to OPLAN 34-A should be explained to Congress: "There's no question but what that had bearing on. And on Friday night, as you probably know, we had four [PT] boats from Vietnam manned by Vietnamese or other nationals, attack two islands." [read, transcription]. Prados reports that 34-A was directed and paid for by the United States, that the targets were chosen by the CIA, and that SNIE 50-2-64 shows the US directed 34-A to pressure Hanoi (secondary); the raids are described elsewhere as carried out by South Vietnamese forces. accessible: true basis: > # V: McNamara's words read in a transcription of the tape; the operational detail and the # SNIE are from Prados (secondary). CORRECTION: the first draft said 34-A was # "carried out by South Vietnamese forces"; sources agree South Vietnamese commandos # executed it, but the National Security Archive's reading is that the United States # directed and controlled it. Both are kept in the description. Established as context. It says Hanoi had a reason to attack the Maddox on 2 August; it does NOT say the US arranged the 2 August clash. See tonkin-bait-order. - id: tonkin-aug4-attack-occurred state: refuted statement: > On 4 August 1964 North Vietnamese torpedo boats attacked USS Maddox and USS Turner Joy a second time. evidence_class: primary_text confidence: high statement_kind: judgment # assigned by rule 2026-10-02, to be reviewed assumptions: - text: "The declassified intercept reports, read through Hanyok's study, are complete enough that no unreleased 4 August material shows an attack." if_wrong: "New intercepts, ship logs or North Vietnamese records showing boats present and firing on 4 August would reopen the claim." - text: "Hanyok's re-analysis of the intercepts is sound. We read it and the reports but did not repeat his analysis." if_wrong: "A reanalysis showing that Reports 7, 12 or 13 do describe an attack on the ships would weaken the refutation." alternatives: - text: "An attack took place but left no trace in the signals intelligence." why_not_preferred: "Hanyok, with 122 relevant SIGINT products, concludes 'no attack happened that night' (read), and no source read shows one." - text: "The key after-action report did concern 4 August." why_not_preferred: "Its original decrypted Vietnamese text and translation cannot be located (claim tonkin-original-text-missing), so it cannot be tested either way." confidence_reasons: start: high start_because: "The anchor was read directly (primary)." steps: - {domain: risk_of_bias, effect: none, reason: "Hanyok writes from NSA's own records against his own institution's earlier account, and the intercept reports themselves were read. The weight rests largely on his re-analysis, which we did not repeat."} - {domain: unreported_negative_results, effect: none, reason: "The original text of the key after-action report cannot be located. Considered and not applied because the other reports read do not describe an attack on the ships. Watch this one."} - {domain: independent_corroboration, effect: none, reason: "Giap's account (via Prados, secondary) agrees; counted as corroboration only, not as a basis."} reviewed: false evidential_weight: 4 # of 5; not 5: the key originals are missing and the re-analysis is Hanyok's anchor_checked: primary # intercept reports and the study read; Herrick message read only as quoted adoption_weight: 2 # of 5, shown separately; never combined with evidential_weight adoption_note: > Historians: a consensus that no second attack occurred (Hanyok's NSA study; Prados; Moise; and a Naval Historical Center historian reversed the Navy's earlier account). The public: not measured; in 1964 the attack was widely believed. would_change_if: > new intercepts, ship logs or North Vietnamese records showing boats present and firing on 4 August, or recovery of the missing original Vietnamese text showing it concerned 4 August refutes_target: "A second North Vietnamese attack took place on 4 August 1964" anchor: nodes: - {node: tonkin-1964-08-04-second-attack-reported-then-refuted, verb: disputes} - {node: tonkin-1964-08-04-herrick-s-doubt-message-sent, verb: disputes} - {node: tonkin-1964-08-04-report-12-boat-reports-an-enemy-aircraft, verb: disputes} - {node: tonkin-1964-08-04-report-13-we-shot-at-two-enemy, verb: disputes} - {node: tonkin-1964-08-04-report-7-boats-told-to-be-ready, verb: disputes} type: signals-intelligence-reexamination description: > The intercept reports cited as evidence of 4 August read as follows [read]: Report 7 (instructions to ready two Swatow boats for "military operations" that night; boat T-333 to join if oil can be found); Report 12 (04/1630Z, a Swatow boat reports an enemy aircraft "falling into the sea; enemy vessel perhaps wounded"); Report 13 (04/1644Z, "we shot at two enemy airplanes and at least one was damaged. We sacrificed two comrades"). Report 13 mentions aircraft and not ships. Hanyok, who had access to 122 relevant SIGINT products, concludes: "no attack happened that night", and that "Hanoi's navy was engaged in nothing that night but the salvage of two of the boats damaged on 2 August" [read]. He also reports: no intercept showed an attack being planned, much less carried out [read]. The Navy's on-scene commander, Captain Herrick, sent a flash message (1:27 p.m. Washington time, 4 August) that many contacts "appear doubtful", that "freak weather effects" and "overeager sonarmen" may account for many reports, and that there were "no visual sightings" by the Maddox [quoted by Prados from a 28 Aug 1964 chronology; secondary]. A pilot overhead, Stockdale, later wrote there was "absolutely no gunfire except our own" (secondary). Hanoi has consistently denied any 4 August action; Bundy's own staff meeting recorded that "Hanoi has denied making the second attack" [read]. accessible: true basis: > # V: the intercept reports and Hanyok's study were read. The Herrick message, Stockdale and # Giap are via Prados (secondary). # LIMIT: the refutation's weight rests largely on Hanyok's re-analysis of the intercepts. # Hanyok also reports that the original decrypted Vietnamese text of the key "after-action" # report, and a translation, cannot be located (see tonkin-original-text-missing). We did # not repeat his analysis. Rule 9 satisfied: names its target and rests on primary documents (the intercept reports), read through a named analysis. Consensus is not the basis. Giap and Stockdale are corroboration only. - id: tonkin-original-text-missing state: searched_gap statement: > The original decrypted Vietnamese text of the key "after-action" intercept used to support the 4 August attack, and an important translation from the San Miguel listening station, cannot be located. evidence_class: primary_text confidence: moderate statement_kind: search_result # assigned by rule 2026-10-02, to be reviewed evidential_weight: 3 # of 5; absence anchor, capped anchor_checked: primary # Hanyok's statement read adoption_weight: 1 # of 5 adoption_note: > Not measured. Little known to the public. would_change_if: > the original text or the San Miguel translation is found next_step: > Search the National Archives, NSA, CIA and LBJ Library finding aids for the original Vietnamese text and the San Miguel translation; log every hit and every miss. anchor: type: documented-absence description: > Hanyok: "the original decrypted Vietnamese text, and an important translation from San Miguel cannot be located." [read] He calls it "unusual" that the original text could not be found (and was not used as early as October 1964), and that these are the only missing papers in the San Miguel reports "allows for suspicion to shade any further discourses". A 1972 note cited by Hanyok records NSA's deputy director, Dr. Louis Tordella, conceding that the 4 August intercept could refer to the 2 August attack, and being unable to produce the original. Report 13 above is the NSA summary that was relied on; the original it summarized is the missing item. accessible: true basis: > An absence is capped as a weaker kind of anchor. This is a documented gap: the study states it. It does not show anyone removed the text; it shows the check cannot be completed. - id: tonkin-sigint-selectively-presented state: established statement: > The signals intelligence about 4 August reaching senior decision-makers was assembled and presented in a way that left out almost all the material showing no attack, and did not give them the complete and objective picture. evidence_class: interpretive confidence: moderate statement_kind: judgment # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5 anchor_checked: primary # the study read; the 122 products not re-audited by us adoption_weight: 3 # of 5, shown separately; never combined with evidential_weight adoption_note: > Not measured. Accepted among historians who have read the NSA study; little known to the public. would_change_if: > an independent audit of the 122 products shows the omitted material did not contradict the attack story next_step: > Obtain the NSA's declassified set of 140 documents and spot-check Hanyok's account of which products were and were not used. anchor: nodes: - {node: tonkin-1964-08-08-cia-cover-note-sends-bundy-the-selected, verb: supports} type: internal-historian-finding description: > Hanyok [read]: "SIGINT information was presented in such a manner as to preclude responsible decisionmakers in the Johnson administration from having the complete and objective narrative of events of 4 August 1964. Instead, only SIGINT that supported the claim that the communists had attacked the two destroyers was given to administration officials." The withheld information "constituted nearly 90 percent of all available SIGINT". The handful of reports supporting an attack "contained severe analytic errors, unexplained translation changes, and the conjunction of two unrelated messages". The six products used in the NSA chronology are "a bit more than 10 percent of the total available"; the other 90 percent was not offered to congressional committees. accessible: true basis: > # V: read directly. CORRECTION: the first draft said the omitted share was "almost 90 percent # of the intercepts that would have contradicted the attack story", from a press summary. # The study's wording is that the withheld material was nearly 90 percent of ALL available # SIGINT on 4 August. This is a count by one analyst, not independently audited. # Filed interpretive because it is the analyst's reading and count; independent analyses # (Prados, Jones, Cline, Moise) reach compatible conclusions about the timing and content # of the intercepts, which is why the state is established and not proposed. Established that the presentation was selective. This claim does not say why; see tonkin-midlevel-deliberate-skew and tonkin-senior-knowledge. - id: tonkin-midlevel-deliberate-skew state: proposed statement: > Individuals at NSA deliberately presented only the SIGINT that supported the attack claim, in an active effort to make it fit the Navy's account, without manufacturing evidence and without a conspiracy at all levels. Hanyok judges that they themselves believed the attack had happened and rationalized the contradictory evidence away. The exact how and why are not known. evidence_class: interpretive confidence: moderate statement_kind: judgment # assigned by rule 2026-10-02, to be reviewed evidential_weight: 3 # of 5 anchor_checked: primary # the study read adoption_weight: 3 # of 5, shown separately; never combined with evidential_weight adoption_note: > Not measured. Rests on one internal historian's inference; the popular version ('manufactured evidence') goes further than the study does. would_change_if: > a document showing the omissions were mistaken rather than deliberate, or showing direction from outside the NSA personnel, or evidence that intercepts were fabricated next_step: > Read the oral-history interviews in the NSA's 2005 release and check what participants said about how reports were chosen. anchor: type: inference-from-pattern description: > Hanyok [read]: "From this evidence, one can easily deduce the deliberate nature of these actions... This was an active effort to make SIGINT fit the claim of what happened during the evening of 4 August." And: "This mishandling of the SIGINT was not done in a manner that can be construed as conspiratorial, that is, with manufactured evidence and collusion at all levels. Rather, the objective of these individuals was to support the Navy's claim that the Desoto patrol had been deliberately attacked." On motive: "The question why the NSA personnel handled the product the way they did will probably never be answered. The notion that they were under 'pressure' to deliver the story that the administration wanted simply cannot be supported." Also: the NSA personnel "believed that [the attack] had occurred"; "The overwhelming body of reports, if used, would have told the story that no attack had happened. So a conscious effort ensued to demonstrate that the attack occurred"; "what was issued in the Gulf of Tonkin summaries beginning late on 4 August was deliberately skewed to support the notion that there had been an attack... That the NSA personnel believed that the attack happened and rationalized the contradictory evidence away is probably all that is necessary to know in order to understand what was done." He finds no "smoking gun" memoranda or notes in the files outlining any plan or stating a justification. accessible: true basis: > # V: read directly. # CORRECTIONS from the earlier draft, which relied on a press summary: (1) it gave the # motive as "to cover up earlier errors". That is not what the study says; the study # says the objective was to support the Navy's claim and that the reason "will probably # never be answered". (2) It said Hanyok concluded that top officials "neither knew nor # condoned" it; the study text read here says there was no evidence of administration # pressure on NSA, and does not make that exact statement in the passages read. Held at proposed: intent is inferred from a pattern, by a single internal historian. - id: tonkin-senior-doubts-documented state: established statement: > By 8 a.m. on 5 August 1964, the White House national security adviser told staff that the evidence for the second attack was "less than we had yesterday", because of "eliminating double counting and mistaken signals", that details such as the number of boats and torpedoes were "still somewhat uncertain", and that Hanoi had denied making the attack. He also said "This much seemed certain: There was an attack." evidence_class: primary_text confidence: high statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5; one document, a record by an attendee and not a transcript anchor_checked: primary # the memo read adoption_weight: 2 # of 5 adoption_note: > Not measured. Known to historians; little known to the public. would_change_if: > evidence the memorandum is not authentic, or a different record of the same meeting anchor: nodes: - {node: tonkin-1964-08-05-white-house-staff-meeting-memo-evidence-less, verb: supports} type: contemporaneous-memorandum description: > Memorandum for the Record, White House Staff Meeting, 5 August 1964 [read]: "On the first attack, the evidence would be pretty good. On the second one the amount of evidence we have today is less than we had yesterday. This resulted primarily from correlating bits and pieces of information eliminating double counting and mistaken signals. This much seemed certain: There was an attack. How many PT boats were involved, how many torpedoes were fired, etc. -- all this was still somewhat uncertain. This matter may be of some importance since Hanoi has denied making the second attack." accessible: true basis: > # V: the memorandum was read (OCR; wording cross-checked against Prados's quotation). # The author is given as William Y. Smith, a military aide, by Prados (secondary). This shows senior doubts about the extent of the attack at the level of the White House two days before the Resolution passed. It also shows the speaker still believed an attack had happened. Both parts must be kept. It is therefore evidence about doubt, and evidence against the stronger claim that he knew nothing had happened. - id: tonkin-bundy-welcomed-resolution state: established statement: > At the same meeting, when a new staffer questioned the need for a congressional resolution, the national security adviser replied jokingly that the matter should perhaps not be thought through too far; as for himself, he "welcomed the recent events as justification for a resolution the Administration had wanted for some time." evidence_class: primary_text confidence: high statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5; one attendee's record of a spoken remark anchor_checked: primary # the memo read adoption_weight: 2 # of 5 adoption_note: > Not measured. Quoted in historical works; little known to the public. would_change_if: > a second record of the meeting showing the remark was different or not made anchor: nodes: - {node: tonkin-1964-08-05-white-house-staff-meeting-memo-evidence-less, verb: supports} type: contemporaneous-memorandum description: > Same memorandum [read]: "Bundy, in reply, jokingly told him perhaps the matter should not be thought through too far. For his own part, he welcomed the recent events as justification for a resolution the Administration had wanted for some time." accessible: true basis: > # V: read directly. Establishes that a senior official described the events as a convenient justification for a resolution already wanted. It does not show the events were staged, that he doubted an attack occurred, or that anyone lied; the memo says he thought an attack had occurred. It records opportunism, which is a different claim from fabrication. - id: tonkin-mcnamara-aug4-call state: established statement: > At 9:43 a.m. Washington time on 4 August, before any attack on the destroyers was reported, McNamara told President Johnson the ship was "allegedly" to be attacked that night, and they discussed what to authorize if it was attacked. evidence_class: primary_text confidence: moderate statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 3 # of 5 anchor_checked: secondary # a transcription by Prados read; the audio not heard adoption_weight: 2 # of 5 adoption_note: > Not measured. Known to historians. would_change_if: > the LBJ Library audio or its own transcript differing materially from this transcription next_step: > Listen to, or read the LBJ Library's own transcript of, the 4 August 9:43 a.m. call. anchor: nodes: - {node: tonkin-1964-08-04-mcnamara-to-lbj-ship-allegedly-to-be, verb: supports} type: recorded-conversation-transcription description: > McNamara [read, Prados transcription]: "...the tracks are laid down very clearly; they go through the three command channels to get out there; this ship is allegedly uh to be attacked tonight - we don't like to see a change in operation plan of this kind at this time." He then reports a recommendation that the commander "be authorized to pursue the attacking vessels in the event he is attacked and destroy their bases." accessible: true basis: > # V: read in a published transcription (primary_text by way of a secondary transcription). It shows Washington was already acting on intercepts mistakenly read as pointing to 4 August before the alleged action. Conditional planning ("in the event he is attacked") is ordinary and is not wrongdoing; the claim is only about timing. Prados's commentary that retaliation was discussed "in spite of the fact that no attack has yet occurred" is an interpretation. - id: tonkin-mcnamara-34a-testimony state: established statement: > On 3 August 1964 McNamara privately told President Johnson that OPLAN 34-A "had bearing on" the Maddox engagement and should be explained to Congress. In formal testimony to the joint Senate committees on 6 August 1964 he said: "Our Navy played absolutely no part in, was not associated with, was not aware of, any South Vietnamese action, if there were any." evidence_class: primary_text confidence: high statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5 anchor_checked: primary # the 6 Aug wording as read into the 1968 hearing record (text layer); the 3 Aug call via Prados's transcription adoption_weight: 2 # of 5 adoption_note: > Not measured. Known to historians; central to the 1968 Senate hearings on Tonkin. would_change_if: > the 1964 transcript itself (Executive Sessions vol. XVI, or the 1966 edited hearing) showing different wording next_step: > Read the 6 August 1964 transcript itself (the edited joint hearing, and Executive Sessions vol. XVI, pp. 291-299) and listen to the 3 August LBJ call; confirm the page numbers of the 1968 quotation on the page images. anchor: nodes: - {node: tonkin-1964-08-03-mcnamara-to-lbj-explain-oplan-34-a, verb: supports} type: recorded-statement-versus-testimony description: > 6 August testimony as read by Chairman Fulbright into the hearing of 20 Feb 1968 (S. Hrg. 90-187, p. 30, "page 23 of the transcript of hearings") [read, OCR text]: "Our Navy played absolutely no part in, was not associated with, was not aware of, any South Vietnamese action, if there were any. I want to make that very clear to you. The Maddox was operating in international waters, was carrying out a routine patrol..." McNamara, asked by Senator Cooper, confirmed it: "That is correct" and that it was his belief then and now. 3 August call [Prados transcription]: "...explain this Op Plan 34-A, these covert operations. There's no question but what that had bearing on." accessible: true sources: [src-sfrc-1968-hearing, src-lbj-tapes-transcripts] basis: > # V: the 1964 wording is read in the 1968 hearing record, where both Fulbright and McNamara treat it as accurate. The two statements are about different things as McNamara later explained (see tonkin-mcnamara-1968-account); whether the 6 August wording left senators with a false impression is an interpretation and is held on the positions of Morse (disputed it) and McNamara (defended it). Prados calls it "highly misleading"; that stays his judgement. - id: tonkin-mcnamara-1968-account state: established statement: > In his prepared statement of 20 February 1968 McNamara said he had told the Senate Foreign Relations Committee about the South Vietnamese raids "in an informal meeting on August 3, 1964, after the attack on the Maddox", that the subject "was again raised in lesser detail" in his testimony on 6 August, that Senator Morse asked that day about a connection between the patrol and the raids and he answered "that there was no connection", and that he "learned subsequent to my testimony of August 6, 1964, that another South Vietnamese bombardment took place on the night of August 3-4". On the second attack he said "some uncertainty existed, and to this day exists, about some of the precise details of the attack. But there should be no uncertainty about the fact that an attack took place." evidence_class: primary_text confidence: high # that he said it; what he said is his own account four years later statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5 anchor_checked: primary # S. Hrg. 90-187 read (OCR text); page numbers from the text layer adoption_weight: 2 # of 5 adoption_note: "Not measured. Known to historians of the Tonkin hearings." would_change_if: "the 3 and 6 August 1964 committee records showing that what he told the senators then differs from this account" next_step: "Read the 3 August informal meeting record, if any survives, and the 6 August transcript (Executive Sessions vol. XVI) to test his 1968 account against 1964." anchor: nodes: - {node: tonkin-1968-02-20-mcnamara-cites-unimpeachable-intercepts-to-the-senate, verb: supports} type: congressional-hearing-testimony description: "Hearing before the Senate Committee on Foreign Relations, 20 Feb 1968 (S. Hrg. 90-187), prepared statement pp. 14-16 and the questioning pp. 30-31 [read, OCR text]." accessible: true sources: [src-sfrc-1968-hearing] basis: > An account given four years after the events, by the person whose conduct is at issue. It is evidence of what he said in 1968, and of what he claims he said in 1964; it is not independent confirmation of the 1964 events. - id: tonkin-fulbright-1968-evidence-uncertain state: established statement: > At the hearing of 20 February 1968 Fulbright, as chairman, said: "I think this committee ... it was very unfair to ask us to vote upon a resolution when the state of the evidence was as uncertain as I think it now is ... none of those intercepts were mentioned to us, I don't believe, in the testimony on August 6. Your statement and General Wheeler's was without any doubt, any equivocation that there was an all-out attack." evidence_class: primary_text confidence: high # that he said it statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5 anchor_checked: primary adoption_weight: 2 # of 5 adoption_note: "Not measured." would_change_if: "the 6 August transcript showing the intercepts were mentioned to the committees" next_step: "Read the 6 August 1964 transcript to test 'none of those intercepts were mentioned'." anchor: nodes: - {node: tonkin-1968-02-20-mcnamara-cites-unimpeachable-intercepts-to-the-senate, verb: supports} type: congressional-hearing-statement description: "S. Hrg. 90-187, 20 Feb 1968, p. 80 [read, OCR text; the page number is from the text layer]. Fulbright qualified the point with 'I don't believe'." accessible: true sources: [src-sfrc-1968-hearing] basis: > This is Fulbright's 1968 view and recollection about what the committees were told in 1964. It bears on what senators knew when they voted (P8) but does not itself show what they were told. - id: tonkin-senior-knowledge state: contested statement: > Senior administration officials knew the 4 August attack had not occurred when they ordered reprisal strikes and sought the Resolution, as opposed to acting on a genuine, if flawed, belief that it happened. evidence_class: primary_text confidence: low statement_kind: judgment # assigned by rule 2026-10-02, to be reviewed disputed_by: - {who: "historians, not named in the sources read", kind: unnamed, position: "divided on what senior officials knew", source_read: false, via: "this claim's adoption note"} evidential_weight: 2 # of 5 anchor_checked: primary # the memo, the tape transcription and the study read adoption_weight: 3 # of 5, shown separately; never combined with evidential_weight adoption_note: > Not measured. 'They lied us into war' is a popular framing; historians are divided on who knew what. would_change_if: > a dated senior communication stating, or clearly showing knowledge, that no attack had occurred, or a document showing senior officials were told the SIGINT was being filtered next_step: > Read FRUS 1964-68 vol. I documents dated 4-7 August 1964 and the LBJ Library tapes for those days; log who said what about whether an attack occurred. positions: - label: "Knew it was false" holder: "The popular 'false flag' claim" ground: > The Herrick doubt cable (sent 1:27 p.m. Washington time, 4 August) reached Washington; McNamara acted on the intercepts before any attack was reported; the adviser welcomed the events as justification; McNamara's later statement to Congress about 34-A was misleading. Prados argues the record shows Washington "rushing to a judgment". - label: "Believed an attack happened, on weak and filtered evidence" holder: "Hanyok; the reading of Bundy's memo" ground: > Bundy recorded "This much seemed certain: There was an attack." Hanyok found only SIGINT supporting the attack reached administration officials; he writes that the notion NSA staff were under pressure from the administration "simply cannot be supported"; and he found no "smoking gun" memoranda or notes in the NSA files outlining a plan or stating a justification. The National Security Archive's 2005 posting summarizes the study as finding senior officials "none the wiser about the gaps in the intelligence" (secondary). Prados's own essay says "senior officials knew of the 'double counting and mistaken signals'", which is knowledge of doubt, and not that nothing had happened. anchor: type: live-dispute description: > Doubts reached the White House within hours, and a senior official welcomed the events as justification. The same record shows that official still thought an attack had happened, and the study found the SIGINT reaching senior officials had been filtered. "Knew of doubts" and "knew nothing happened" are different claims; the documents read support the first and do not settle the second. accessible: true basis: > # V: the memo, the tape transcription of the 4 August call and the study were read directly. # The earlier draft carried an exact clock time of "received 13:27" for the Herrick message; # the chronology says it was SENT at 1:27 p.m. Washington time. Correct the earlier text. Genuinely contested. Held at contested, low confidence. This remains the dig's central conspiracy-adjacent gap, but it is now much narrower than before: the documents show doubt and opportunism and do not show knowledge that no attack occurred. - id: tonkin-resolution-rested-on-aug4 state: established statement: > Congress passed the Tonkin Gulf Resolution (7 August 1964) in response to the reported attacks, with the 4 August report central to the administration's case. evidence_class: primary_text confidence: high statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5 (the cap); the Record's roll call, the printed address and the floor manager's opening were read; the full debate was searched by keyword, not read line by line anchor_checked: primary # Congressional Record, Senate, 6 Aug 1964 pp. 18399-18400 and 7 Aug 1964 pp. 18459-18471 adoption_weight: 5 # of 5, shown separately; never combined with evidential_weight adoption_note: > Not polled. The vote counts and the text are not disputed. would_change_if: > the full floor debate showing the attacks were not central to the case made to Congress next_step: > Read the whole Senate debate of 6-7 August 1964 and the House debate of 7 August; log which speakers cite the 4 August attack and which do not. anchor: nodes: - {node: tonkin-1964-08-07-tonkin-gulf-resolution-passes, verb: supports} - {node: crec-1964-p18470, verb: supports} type: legislative-record description: > Congressional Record, 7 Aug 1964 [read]: p. 18470 gives the Senate roll call "yeas 88, nays 2" (nays Gruening and Morse; 10 not voting) and says "So the joint resolution (H.J. Res. 1145) was passed". The debate prints, as "Exhibit 1", the President's address of 4 August: "The initial attack on the destroyer Maddox on August 2 was repeated today by a number of hostile vessels attacking two U.S. destroyers with torpedoes." The enacted preamble says Communist naval units "have deliberately and repeatedly attacked United States naval vessels". House: the Record prints 416-0 (Roll No. 210, p. 18555; see tonkin-house-debate-aug7). Signed 10 August per Senate.gov (secondary). accessible: true sources: [src-congressional-record-1964] basis: > # V: the Senate vote and dissenters read in the Record (primary). The presence of the 4 August address in the debate # and the preamble's "repeatedly" read in the Record. "Central to the case" is still an interpretation: the whole debate # (about 11 columns-pages) was searched for key terms, not read line by line. - id: tonkin-fulbright-opening-aug4-as-fact state: established statement: > Opening the Senate debate on 6 August 1964, Fulbright told the Senate: "On August 4 the Maddox and another destroyer, the C. Turner Joy, were again attacked by North Vietnamese torpedo boats in international waters. The attack, which lasted for over 2 hours, was without any doubt a calculated act of military aggression." He also said the committees had approved the resolution with only one dissenting vote. evidence_class: primary_text confidence: high statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 5 # of 5; the Record's page read on the image anchor_checked: primary adoption_weight: 3 # of 5 adoption_note: "Not measured." would_change_if: "a different printing of the speech in the permanent Record" next_step: "Read what the two committees were told before reporting (Foreign Relations and Armed Services joint session, 6 Aug; Senate Historical Series vol. XVI)." anchor: nodes: - {node: tonkin-1964-08-06-fulbright-opens-senate-debate-aug4-as-fact, verb: supports} - {node: crec-1964-p18399, verb: supports} type: floor-statement description: "Congressional Record, Senate, 6 Aug 1964, pp. 18399-18400 [read on the page image, p. 18399]." accessible: true sources: [src-congressional-record-1964-aug6] - id: tonkin-aug1964-debate-no-doubt-voiced state: searched_gap statement: > No passage has been found in the Senate debates of 6 and 7 August 1964 in which a senator questioned whether the 4 August attack on the destroyers occurred. Senators whose words were found citing it treated it as fact: Fulbright ("without any doubt"), Smathers ("The facts are indisputable"), Keating ("no doubt whatsoever that the second attack"), Scott and Russell (who mention the "second attack" and where the ship was). Others in the same debate (Morse, McGovern) raised the earlier raids and how close the destroyers patrolled to the coast. evidence_class: primary_text confidence: low statement_kind: search_result # assigned by rule 2026-10-02, to be reviewed evidential_weight: 2 # of 5; absence anchor, capped at provisional anchor_checked: secondary # keyword search of OCR text for 'August 4', 'second attack', 'night of the 4th' and nearby terms; quotes other than Fulbright's not checked on the page images adoption_weight: 2 # of 5 adoption_note: "Not measured." would_change_if: "a senator's words in the debate casting doubt on whether the 4 August attack occurred" next_step: "Read both days in full, page by page (6 Aug pp. 18399-18436; 7 Aug pp. 18441-18471), verify each quote on its image; also the House debate of 7 Aug (pp. 18538 ff.)." anchor: type: absence-in-searched-record description: > Machine search of the Congressional Record text (OCR, column order partly scrambled). The search terms were tied to 4 August; a doubt expressed in other words would be missed. Church (6 Aug) said the destroyers, "from all that we have been told, were not engaged in any aggressive action directed against the shores of North Vietnam" [OCR wording, not verified on the image]: senators were acting on what they were told, which is the P8 question the dig has not yet answered. accessible: true sources: [src-congressional-record-1964-aug6, src-congressional-record-1964] - id: tonkin-house-debate-aug7 state: established statement: > The House passed the Tonkin Gulf Resolution on 7 August 1964 under suspension of the rules: the Record prints "yeas 416, nays 0, answered 'present' 1, not voting 14" (Roll No. 210, p. 18555; the one "present" is Powell), on Morgan's motion to pass H.J. Res. 1145 "with an amendment". In the debate (pp. 18538-18555, read in full by a delegated reader) only Zablocki cited the 4 August attack by date ("on August 4--U.S. destroyers, the Maddox and the C. Turner Joy, were again attacked... The attack lasted 2 hours"); Albert, Broomfield and Roush referred to two attacks without dates; no Member doubted that the attack occurred or asked about the evidence for it; no Member named OPLAN 34-A or South Vietnamese raids. Several (Derwinski, Chamberlain, Alger) complained about the information given to Congress, and Halleck, Morgan and Barry referred to what they were told at White House or committee briefings. evidence_class: primary_text confidence: moderate statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5 (the cap); Record pages 18538-18555 read in full by a delegated reader with page-image checks of quoted passages; later pages keyword-searched only anchor_checked: primary adoption_weight: 3 # of 5 adoption_note: "Not measured." would_change_if: "a re-reading of the pages showing a Member who questioned the attack or the evidence, or the later pages (18556-18625) containing relevant debate" next_step: "Re-check the Zablocki, Halleck, Barry and Chamberlain quotes on the images myself; read the committee amendment text (the 'Constitution of the United States' wording) and reconcile 'with an amendment' in the House with the Senate message that it passed H.J. Res. 1145 'without amendment' (p. 18612); read what the Foreign Affairs Committee was told (its report)." anchor: nodes: - {node: vote-RH0880197, verb: supports} type: legislative-record description: "Congressional Record, House, 7 Aug 1964, pp. 18538-18555 (roll call p. 18555), with a Senate message at p. 18612. Delegated reading; quoted passages image-checked by that reader." accessible: true sources: [src-congressional-record-1964-house] - id: tonkin-resolution-repealed-1970 state: established statement: > On 24 June 1970 the Senate voted 81 to 10 to agree to the Dole amendment repealing the Gulf of Tonkin Resolution effective at the end of the 91st Congress. The ten nays were Allen, Bellmon, Eastland, Ervin, Fulbright, Hollings, Long, McClellan, McGee and Stennis. evidence_class: primary_text confidence: high statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 4 # of 5; roll call read in the Record; the amendment's text not read anchor_checked: primary adoption_weight: 3 # of 5 adoption_note: "Not measured. Known to historians of the war powers debate." would_change_if: "the Record's roll call being corrected" next_step: "Read the amendment text (Amendment No. 715) and the later vote on S. Con. Res. 64 (10 July 1970, 57-5 per Voteview) in the Record." anchor: nodes: - {node: vote-RS0910398, verb: supports} - {node: crec-1970-p21132, verb: supports} type: legislative-record description: > Congressional Record, Senate, 24 June 1970, p. 21132 [read]: "The result was announced-yeas 81, nays 10" [No. 167 Leg.], "So Mr. DOLE's amendment (No. 715) was agreed to." Voteview numbers the same roll call 398 in its index. accessible: true sources: [src-congressional-record-1970] - id: tonkin-1970-vote-not-clean-measure state: established statement: > In the 24 June 1970 Senate debate on the Dole amendment, the ten senators who voted Nay gave different reasons, or none in the pages read. Fulbright said he had intended to vote for repeal and changed his vote after "the assurance of the majority leader" that the Mathias concurrent resolution would be brought up. Allen and Hollings spoke against repeal itself. Stennis mixed opposition to repeal with a procedural complaint. McGee called the repeal a "political charade". No stated reason was found for Bellmon, Eastland, Ervin, Long and McClellan. Several senators who voted Yea (Javits, Moss) said a concurrent resolution would be better but still backed the amendment. A 1970 vote on this amendment is therefore not a clean measure of a senator's view of the 1964 Resolution. evidence_class: primary_text confidence: moderate statement_kind: reported # assigned by rule 2026-10-02, to be reviewed evidential_weight: 3 # of 5; Fulbright's words verified on the page image by me; the other senators' reasons by a delegated reading (images checked by that reader, not re-checked by me) anchor_checked: primary adoption_weight: 1 # of 5 adoption_note: "New today; not a position anyone is recorded as holding." would_change_if: "a reading of the pages showing a different reason for any of the ten, or of the pages 1-14 and 34-45 not yet read in full" next_step: "Re-check each non-Fulbright quote on its page image; read the 22 June debate (the Record file read did not include it) and the text of Amendment No. 715." anchor: nodes: - {node: vote-RS0910398, verb: supports} - {node: crec-1970-p21119, verb: supports} - {node: crec-1970-p21131, verb: supports} - {node: crec-1970-p21132, verb: supports} type: floor-statement description: > Congressional Record, Senate, 24 June 1970. Fulbright, p. 21131 [read on the page image]: "I said earlier that I would vote for the repeal because, of course, I am committed to repeal and have been for a long time. However, upon further consideration, in view of what has been said on the floor of the Senate, and also in reliance upon the assurance of the majority leader that he would bring up the Mathias resolution--which is a concurrent resolution..."; p. 21132: "I shall vote against the repeal today and preserve the integrity of the procedure which I believe is very important to this body." Earlier that day, p. 21119: "I am going to vote for its repeal. I am committed to it, of course." Nay list p. 21132. Reasons of Allen (p. 21123), Hollings (p. 21122), Stennis (p. 21121), McGee (pp. 21130-31) per the delegated reading. accessible: true sources: [src-congressional-record-1970] - id: tonkin-bait-order state: searched_gap statement: > No document read or reported so far shows an order to provoke the 2 August clash by sending the Maddox in as bait. evidence_class: primary_text confidence: low statement_kind: search_result # assigned by rule 2026-10-02, to be reviewed evidential_weight: 2 # of 5; absence anchor, capped anchor_checked: secondary # searched in the study text, the essay and the tape page; not the full releases adoption_weight: 2 # of 5, shown separately; never combined with evidential_weight adoption_note: > Not measured. Circulates in popular accounts; no document found. would_change_if: > a document ordering or planning the Maddox patrol as a provocation is found next_step: > Read SNIE 50-2-64 and the NSA's remaining declassified Tonkin documents (about 140 released in 2005); search each for provocation or bait language; log every hit and every miss. anchor: type: absence-in-searched-record description: > Text searches of the Hanyok study found no mention of bait; its only uses of "provocation" concern the Aug 4 incident as the pretext for the Resolution and September 1964 patrols. Prados's essay and the tape commentary say the Maddox's mission was to record North Vietnamese radio and radar emissions expected to rise after a 34-A raid (so the patrol was coordinated with the raids), and that the mission commander "had been briefed in Taiwan previously that there would be" no attack and was unaware of the raids. None describes an order to provoke. accessible: true basis: > # V: the search of the study was by machine text search of OCR output (noisy); the essay and # tape page were read. The full release set was not. A searched gap, not a refutation: absence in what has been read is not proof of absence. The coordination of the patrol with the raids is documented; it is not an order to provoke. A found document would move this claim.