UNCLASSIFIEDffFOR OPIlICIAL tiS! ONLY

Distribution

Cryptologic Quarterly is published four times a year by the Center forCryptologic History, NSA. The publication is designed as a working aid and is not sub­ject to receipt, control, or accountability. Distribution is made to branch level; furtherdissemination is the responsibility of each branch. Extra copies or those for whichthere is no further need Should be returned to the Editor for disposition.

Contributions

Contributions to Cryptologic Quarterly should be sent to. I I

Managing Editor I Cryptologic QuarterlyE05, SAB 2, Door 22, Suite 6886I I (b) (3)-P.L. 86-36

Electronic submissions can be made via email. Such submissions should include apdf (portable document format) version, as well as a txt (text) version of the article.Each article should include an abstract. Authors are responsible for determining theclassification of submitted articles. If computer disks are submitted, authors mustensure that they are virus-free. All material used in the publication of an article isdestroyed when no longer needed unless the author requests that it be returned.Cryptologic Quarterly will not accept articles written anonymously or with a pseudo­nym.

Deadlines for Submissions to the Quarterly

Spring issueSummer issueFall issueWinter issue

15 February15 May15 August15 November

Reproduction and Dissemination

Contents of any issue of Cryptologic Quarterly may not be reproduced or fur­ther disseminated outside the National Security Agency without the permission of theNSA/CSS Office of Policy.

All opinions expressed in Cryptologic Quarterly are those of theauthors. They do not necessarily represent the official views of theNational Security Agency.

UNCLASSIFIEDlfFeR eFFIebliL !:JSt eNLY Page iii

IUP SECRI!T/teeMINT/RE1 Cryptologic Quarterly

(U) Skunks, Bogies, Silent Hounds, and the Flying Fish:The Gulf of Tonkin Mystery, 2-4 August 1964

ROBERTJ.HANYOK

(CffSI) The Gulf of Tonkin incidents of 2 to 4August 1964 have come to loom over the subse­quent American engagement in Indochina. Theincidents, principally the second one of 4 August,led to the approval of the Gulf of TonkinResolution by the u.s. Congress, which handedPresident Johnson the carte blanche charter hehad wanted for future intervention in SoutheastAsia. From this point on, the American policy andprograms would dominate the course of theIndochina War. At the height of the Americaninvolvement, over a half million u.S. soldiers,sailors, airmen, and marines would be stationedthere. The war would spread across the borderinto Cambodia and escalate in Laos. Thailandassumed a greater importance as a base for sup­porting the military effort, especially for the airwar, but also for SIGINT purposes of interceptand direction finding.

(D) At the time, the Gulf of Tonkin incidentsof August were not quite so controversial.According to the Johnson administration, theissue of the attacks was pretty much cut anddried. As the administration explained, our shipshad been in international waters - anywherefrom fifty to eighty miles from the DRV coastlineby some calculations, during the alleged secondattack - and were attacked twice, even thoughthey were innocent of any bellicose gesturesdirected at North Vietnam. Secretary of DefenseRobert McNamara had assured the Senate thatthere had been no connection between what theU.S. Navy was doing and any aggressive opera­tions by the South Vietnamese.' Washingtonclaimed that the United States had to defend itselfand guarantee freedom of navigation on the highseas.

(U) However, within the government, theevents of 4 August were never that clear. Even asthe last flare fizzled in the dark waters of theSouth China Sea on that August night, there wereconflicting narratives and interpretations of whathad happened. James Stockdale, then a navy pilotat the scene, who had "the best seat in the housefrom which to detect boats," saw nothing. "Noboats," he would later write, "no boat wakes, noricochets off boats, no boat impacts, no torpedowakes - nothing but black sea and American fire­power." 2 The commander of the Maddox taskforce, Captain John J. Herrick, was not entirelycertain what had transpired. (Captain Herrickactually was the commander of the destroyer divi­sion to which the Maddox belonged. For this mis­sion, he was aboard as the on-site commander.)Hours after the incident, he would radio theCommander-in-Chief, Pacific (CINCPAC) tellingthem that he was doubtful of many aspects of the"attack."

(U) It would be years before any evidencethat an attack had not happened finally emergedin the public domain, and even then, most reluc­tantly. Yet, remarkably, some of the majorpartic­ipants in the events still maintained that the Gulfof Tonkin incident had occurred just as it hadbeen originally reported. Secretary of DefenseRobert McNamara, in his memoirs In Retrospect,considered the overall evidence for an attack stillconvincing." The u.S. Navy's history of theVietnam conflict, written by Edward J. Maroldaand Oscar P. Fitzgerald (hereafter referred to asthe "Marolda-Fitzgerald history"), reported thatthe evidence for the second attack, especiallyfrom intelligence, including a small amount ofSIGINT, was considered conclusive.f

Derived From: NSAlCSSM 123·224 February 1998

Declassify On: X1 TOP iIiGAi;HS9MINT,.,*1 Page 1

Cryptologic Quarterly TOP SEClt!T/ICOMINY/IX I

(U) The public literature on the Gulf ofTonkin for years has been overwhehningly skep­tical about the 4 August battle. Articles thatappeared in magazines within a few years illus­trated the general inconsistency in the descrip­tions of the incident of 4 August by simply usingthe conflicting testimony from the officers andcrews of both ships. The first major critical vol­ume was Joseph Goulden's Truth Is the FirstCasualty, published in 1969. The most completework to date is Edwin Moise's Tonkin Gulf andthe Escalation ofthe Vietnam War. Moise's workhas the dual advantage of using some Vietnamesesources, as well as small portions of a few SIGINTreports released to the author under a Freedom ofInformation Act request. Yet, even what fewscraps he received from NSA were enough toraise serious questions about the validity of theSIGINT reports cited by the administrationwhich related to the 4 August incident."

"'"(Sf/sit The issue of whether the availableSIGINT "proved" that there had been a secondattack has been argued for years. In 1968, RobertMcNamara testified before Senator WilliamFulbright's Foreign Relations Committee's hear­ings on the Gulf of Tonkin that the supportingsignals intelligence was "unimpeachable." On theother hand, in 1972 the deputy director of NSA,Louis Tordella, was quoted as saying that the 4August intercepts pertained to the 2 Augustattacks. In a 1975 article in the NSA magazineCryptolog, the Gulf of Tonkin incident wasretold, but the SIGINT for the night of August 4was not mentioned, except for the "military oper­ations" intercept, and even then without com­ment," The Navy's history of the Vietnam Warwould misconstrue the SIGINT (disguised asunsourced "intelligence") associating portions oftwo critical intercepts and implying a connectionin the evidence where none could be established.'

tel/Sf) Except for the sizable collection ofSIGINT material within NSA, and a much small­er amount from the archives of the Naval SecurityGroup (which essentially duplicates portions of

the NSA holdings), almost all relevant materialrelating to the Gulf of Tonkin incidents has beenreleased. Although the questions about what hap­pened in the Gulf of Tonkin on the night of 4August have been fairly well answered by the evi­dence from allof the other sources - radar, sonar,eyewitness, and archival - the SIGINT versionneeds to be told. This is because of the critical rolethat SIGINT played in defining the second attackin the minds of Johnson administration officials.Without the signals intelligence information, theadministration had only the confused and con­flicting testimony and evidence of the men andequipment involved in the incident. It is difficultto imagine the 5 August retaliatory air strikesagainst North Vietnamese naval bases and instal­lations being ordered without the SIGINT "evi­dence." 8 Therefore, it is necessary to recount insome detail what signals intelligence reported.

(Sf/Sf) For the first time ever, what will bepresented in the following narrative is the com­plete SIGINT version of what happened in theGulf of Tonkin between 2 and 4 August 1964.Until now, the NSAhas officiallymaintained thatthe second incident of 4 August occurred. Thisposition was established in the initial SIGINTreports of 4 August and sustained through aseries of summary reports issued shortly after thecrisis. In October 1964, a classified chronology ofevents for 2 to 4 August in the Gulf of Tonkin waspublished by NSAwhich furthered the contentionthat the second attack had occurred.

(S//Sf) In maintaining the official version ofthe attack, the NSA made use of surprisingly fewpublished SIGINT reports - fifteen in all. Theresearch behind the new version which follows isbased on the discovery of an enormous amount ofnever-before-used SIGINT material. This includ­ed 122 relevant SIGINT products, along withwatch center notes, oral history interviews, andmessages among the various SIGINT and militarycommand centers involved in the Gulf of Tonkinincidents. Naturally, this flood of new informa­tion changed dramatically the story of that night

Page 2 TePSE6AElo\'68MIN'fffl(1

'F9P SEeRE'FNe9MIN'FUK1 Cryptologic Quarterly

of 4/5 August. The most important element isthat it is now known what the North VietnameseNavy was doing that night. And with this infor­mation a nearly complete story finally can be told.

(8//BI) Two startling findings emerged fromthe new research. First, it is not simply that thereis a different story as to what happened; it is thatno attack happened that night. Through a com­pound of analytic errors and an unwillingness toconsider contrary evidence, American SIGINTelements in the region and at NSA HQs reportedHanoi's plans to attack the two ships of theDesoto patrol. Further analytic errors and anobscuring of other information led to publicationof more "evidence." In truth, Hanoi's navy wasengaged in nothing that night but the salvage oftwo of the boats damaged on 2 August.

(8//81) The second finding pertains to thehandling of the SIGINT material related to theGulf of Tonkin by individuals at NSA Beginningwith the period of the crisis in early August, intothe days of the immediate aftermath, and contin­uing into October 1964, SIGINT information waspresented in such a manner as to precluderesponsible decisionmakers in the Johnsonadministration from having the complete andobjective narrative of events of 4 August 1964.Instead, only SIGINT that supported the claimthat the communists had attacked the twodestroyers was given to administration officials.

(S/fSI) This mishandling of the SIGINT wasnot done in a manner that can be construed asconspiratorial, that is, with manufactured evi­dence and collusion at all levels. Rather, theobjective of these individuals was to support theNavy's claim that the Desoto patrol had beendeliberately attacked by the North Vietnamese.Yet, in order to substantiate that claim, all of therelevant SIGINT could not be provided to theWhite House and the Defense and intelligenceofficials. The conclusion that would be drawnfrom a review of all SIGINT evidence would havebeen that the North Vietnamese not only did not

attack, but were uncertain as to the location of theships.

-f.S/fSlr Instead, three things occurred withthe SIGINT. First of all, the overwhelming por­tion of the SIGINT relevant to 4 August was keptout of the post-attack summary reports and thefinal report written in October 1964. The withheldinformation constituted nearly 90 percent of allavailable SIGINT. This information revealed theactual activities of the North Vietnamese on thenight of 4 August that included salvage opera­tions of the two torpedo boats damaged on 2

August, and coastal patrols by a small number ofDRV craft. As will be demonstrated later in thischapter, the handful of SIGINT reports whichsuggested that an attack had occurred containedsevere analytic errors, unexplained translationchanges, and the conjunction of two unrelatedmessages into one translation. This latter productwould become the Johnson administration'smain proof of the 4 August attack.

(Sf/SI) Second, there were instances in whichspecious supporting SIGINT evidence was insert­ed into NSAsummary reports issued shortly afterthe Gulf of Tonkin incidents. This SIGINT wasnot manufactured. Instead, it consisted of frag­ments of legitimate intercept lifted out of its con­text and inserted into the summary reports tosupport the contention of a premeditated NorthVietnamese attack on 4 August. The sources ofthese fragments were not even referenced in thesummaries. It took extensive research before theoriginal reports containing these items could beidentified.

(8//SI) Finally, there is the unexplained dis­appearance of vital decrypted Vietnamese text ofthe translation that was the basis of the adminis­tration's most important evidence - the so-calledVietnamese after-action report of late 4 August.The loss of the text is important because the SIG­INT record shows that there were critical differ­ences in the English translations of it issued bothby the navy intercept site in the Philippines and

TOil S!C"!TltCOMINTI1X1 Page 3

Cryptologic Quarterly l'6P SEe~E1We6MIN1'HX I

tb) (1)(0)(3)-50 USC 403(bk(3)-P.L. 86-36

(e//SI) Desoto was the covername for a U.S.Navy signals intelligence collection programbegun in 1962 in which naval SIGINT direct sup­port units (DSO) were placed on board Americandestroyer patrols along the Asiatic coastline in thewestern Pacificl I

NSA. Without the individual texts (there weretwo of them), it is difficult to determine why thereare critical differences in the translations andmore importantly, to understand why two sepa­rate North Vietnamese messages were combinedinto one translation by NSA.

(0) Before a discussion can begin, it is neces­sary to understand how the Gulf of Tonkin inci­dents came to happen, the way they did, and whattheir significance was for the Johnson adminis­tration. To do that, we need to consider theDesoto mission that the Maddox was conductingat the time, as well as the Defense Department'sOPLAN-34A missions against the DemocraticRepublic of Vietnam (DRV). It was the conver­gence of the two that embroiled that ship in thecrisis in the Tonkin Gulf.

(U) The Desoto Missions

(8//81) Physically, Desoto mission destroyerswere unique in their configuration - a small vanlashed to the ship which housed intercept posi­tions for voice and manual morse communica­tions. There also was a position which intercept­ed noncommunications emissions such as radars,referred to as electronic intelligence or ELINT.Finally, a communications position, whichallowed the detachment to send and receive mes­sages from the other monitoring stations in thearea, as well as other SIGINT organizations andcommands, via the Criticomm communicationssystem, was located in the hut. The hut wasmanned in shifts from a complement of twelve toeighteen officers and men from the Navy's cryp­tologic element, known as the Naval SecurityGroup (NSG). However, contrary to some asser-

tions, the Desoto missions were not the function­al or operational equivalent of the ubiquitousSoviet electronic collection trawlers." The Desotomissions primarily served the mission needs oflocal commanders, although they received techni­cal support in the way of technical working aidsand intercept data from NSA.

(S//SI)

~The Desoto patrols had a two-part mis­sion: to collect intelligence in support of theembarked commander and higher level authori­ties and to assert freedom of navigation in inter­national waters. The earl Desoto missions in thewaters

had been tracked by the coastal radar sur­"'v-e"'l"'.a..lnce networks •. belonging to the naval forcesI IWhile an occasional commu­nist patrol ship would come out and shadow theUiS, patrol, little else happened.

(D) However, when the Desoto patrol firstwas proposed for the waters in Southeast Asia, its

(V) Desoto mission V2lD

Page 4 l'6P SEe~El'lie6MIN'fiiX I

TepSESRETuseMINTfRE1 Cryptologic Quarterly

south of the islands. In response, Admiral Sharp,CINCPAC, issued a new directive for a Desotopatrol whose purpose was "determining DRVcoastal patrol activity." 13

(V) GulfofTonkin region of interest to OPLAN-34Aand Desoto missions' alphabetic pointsdenote Desotomission start and stoppositions.

(Courtesy of Naval Htstorical Center)

(U) That the two missions might run up phys­ically against one another was a consideration atboth MACV in Saigon and CINCPAC (and CINC­PACFLT) in Honolulu. But Westmorelandassured the navy commanders that as long as theDesoto patrol stayed within its schedule and areaof operations, there would be no problem.Westmoreland added that all the Studies andObservations Group (SOG), which ran theOPLAN-34A missions, needed in the way of an

17

,...I

,Halnan

Gulf

of

Tonkin

10.- 107

'00·

L80s

(U) In mid-January 1964, COMUSMACVrequested that the Desoto patrol scheduled forFebruary (USSRadford, DD-446) be designed toprovide the forthcoming OPLAN-34A programwith critical intelligence regarding NorthVietnam's ability to resist its projected comman­do operations. However, in this case, theRadfords mission was canceled so as to not inter­fere with OPLAN-34A missions planned for thefirst two weeks of February."

(D) This is an important point, although asubtle one, for understanding the events of 2 to 4August. Inasmuch as there was an interworkingbetween the two programs, and this remained apoint of contention in later congressional hear­ings, as well as a source for speculation by thepress, the Desoto mission remained merely one ofcollection of intelligence which could be of use tothe OPLAN-34A planners and commanders backin Danang and the Pentagon. There was no directoperational connection between the two pro­grams. They were managed under separateofficesand were not known to coordinate missionplanning, except for warnings to the Desotopatrol to stay clear of 34A operational areas. Atleast that was the understanding back inWashington.12

(U) In early July, General Westmorelandrequested more intelligence on Hanoi's forceswhich were capable of defending against anexpanded OPLAN-34A program. Specifically,Westmoreland required intelligence on the DRY'sdefenses in those areas targeted for July opera­tion - Hon Me, Hon Nieu, and Hon Matt Islands,as well as the area around the port of Vinh Son,

mission was expanded. First of all, the command­er, Seventh Fleet, wanted the patrol to move incloser than the original twenty-mile limit - asclose as twelve miles. Additionally, the Desotomission was expanded to include a broader col­lection of "all-source intelligence," namely, pho­tographic, hydrographic, and meteorologicalinformation."

Tep6ESRETNSeMINTfRE1 Page 5

Cryptologic Quarterly 'Fap SEeRE'F}leaMIN'F/8E1

lOP &EGAE'Fo\'e9MIN'FMC1

alert, was thirty-six hours' notice of any change.They could then adjust any planned 34A opera­tion. The navy accepted these reassurances fromMACV.14

(S//81) The first Desoto mission in the TonkinGulf region ran from February to March 1964.The USS Craig (DD-88S) sailed near HainanIsland towards the Vietnamese coast and thenturned back north towards Macao and Taiwan.

'-- ..... The North Vietnamesetracked the Craig as it swung south of HainanIsland, but had made no reaction even thoughthey knew that it was a U.S. warship. It wasuncertain to the Americans what the Vietnameseprecisely knew of the Craig or its mission,!

I

(SI/SI) During this mISSIon, there was aNaval Security Group DSU aboard whose taskwas to provide tactical intelligence to the Craig'scommander, as well as intercept unique commu­nications and electronic intelligence in reaction tothe vessel's presence. The Craig also receivedsupport from the SIGINI' facilities in the region:the navy and air force COMINT sites in thePhilip ines

No Vietnam-based sites were'-----_.....involved since the area of Craig's mission barelytouched on the DRVs territorial waters, and thenonly briefly, although it was suspected that the

Page 6

North Vietnamese navy at least once did reportthe Craig's position."

(U) However, there were two critical differ­ences between the Craig's Desoto mission andthat of the Maddox which followed it in late Julyand August: The Maddox would sail along theentire DRV coastline, while, at the same time,OPLAN-34A maritime missions against NorthVietnamese coastal installations were being car­ried out. By July, the North Vietnamese werereacting aggressively to these raids, pursuing andattacking the seaborne commando units .

.(T~/ISItIn mid-July 1964, the JCS approvedanother Desoto mission, which would concen­trate on collecting intelligence on NorthVietnam's coastal defense posture. The USSMaddox, under the command of Captain JohnHerrick, loaded up its intercept van in theI frhe sixteen membersof the DSUboarded, and the ship departed for theGulf of Tonkin. The Maddox had received noadditional instructions to its standard collectionmission and apparently was not aware of specificOPLAN34A missions in the area." However, theMaddox was not on a purely passive mission. U.S.intercept sites in the area were alerted to the realreason for the Desoto missions, which was tostimulate and record (my italics) NorthVietnameseI ~eactions insupport of the U.S. SIGINT effort.19

(TSf/SI) CINCPAC's orders to Herrick wereequally explicit and ambitious: locate and identi­fy all'coastal radar transmitters, note all naviga­tional aids along the DRVs coastline, and moni­tor the Vietnamese junk fleet for a possible con­nection to DRV/Viet Cong.maritime supply andinfiltration 'routes,zo Whether these missionscould be completed' was/questionable: the DSUwas limited byits few positions and equipment incollecting such a large amount of communica­tions. The Maddox had been ordered by CINC­PAC to stayeightnautical miles from the NorthVietnamese coaStline, but only four miles from

(b) (1)(b)(3)-50 USC 403(b) (3)-18 USC 798(b) (3)-P.L. 86-36

19P SE6RH....S9MINli'I)(1 Cryptologic Quarterly

(V) V55,A1qddox

COD 731),

in

August 1964

(V) C'lptain John J, Herrick (left), the on-site task

Force comrnander, 'Inc! Commander Herbert Ogier,

commandinq officer ofthe Mq440X

any of its islands." It would be attacks on theseislands, especially Hon Me, by South Vietnamesecommandos, along with the proximity of theMaddox, that would set off the confrontation.

(U) Operations Plan 34A

TOP SESRElN69MIN'fH'l:t

b) (1)(b) (3)

per CIA

Page 7

Cryptologic Quarterly lep SEeRElNeeMINffRE1

;:;;:/Ib) II)Ib) (3)

(TS//SI) At the beginning of 1964, theDepartment of Defense, which had started itsown program, assumed control of all ofthesecovert missions. It merged its own projectC]I lalldorganized

Id milie n_;O:A::::~: Ito last twelve months and was to be a program ofselective intrusions and attacks of graduatedintensity. The purpose of these actions was to"convince the Democratic Republic of Vietnamleadership that their continued direction andsupport of insurgent activities in the RVN andLaos should cease." 24

fTSi/SI) The OPLAN reflected the currentAmerican strategy of escalation of the warthrough graduated response. The u.s. estab­lished four levels of actions; each proceeding onewas a qualitative and quantitative increase in thesensitivity of target selection and the intensity ofthe application of force. It began with harass­ment attacks and operations, whose cumulativeeffect, though labeled "unspectacular," was tomake Hanoi aware of them to the extent it wouldallocate forces to counter them. 25 If this approachfailed, then the next level - tagged as attritional- was to attack important military and civilinstallations whose loss could cause "temporaryimmobilization of important resources" which, inturn, might create or increase oppositionamongst the North Vietnamese population to thegovernment in Hanoi. The third level, termedpunitive by the 34A planners, was meant to causedamage, displacement, or destruction of thosefacilities or installations considered critical to theDRV economy, industry, or security. To protectitself from further attacks would mean that theDRVwould have to redeploy resources originallymeant to support the war in the south to the

needs of interna1s~curitY. The planners admittedthat the ope~a~()ns at this level would involvelarge enougliforces that they would be necessari­lyoyert~But the planners felt that these attackscould be attributable to the South Vietnamese."

(TB/fSI) The final step of the plan was the ini­tiationof an aerial bombing campaign designedto damage the DRVs capacity to support thesouthern insurrection or cripple its economy tosuch an extent that it would realize the extent ofits losses was not worth the support of the war inthe South. At this point, the planners inWashington believed that Hanoi's reaction to theattacks would be based on two factors: its will­ingness to accept critical damage to its own econ­omy by continuing supporting the war in theSouth, and the possible support of the People'sRepublic of China. The plan did suggest that thecommunists would choose to continue to supportthe southern front, and it left open the possibilityof further operations to offset the anticipatedChinese aid.27

(TS//SI) The major operational componentsof OPLAN 34A were airborne operations thatinserted intelligence and commando teams intoNorth Vietnam, and maritime operations(MAROPS) which consisted of hit-and-run raidson coastal installations and facilities. These lattermissions were known under the operational titleTimberwork. The teams were made up of mostlySouth Vietnamese Special Forces, known as LueLuong Due Biet or Biet Kieh, with some foreignmercenaries (mostly Chinese and Koreans) tocrew the attack craft. The American involvement,though extensive in the planning, training, andlogistics portions, was minimized to achieve theusual "nonattribution" status in case the raidswere publicizedby the North. NoAmericans wereallowed to participate in the actual raids.

(U) Despite all of the planning, there was lit­tle confidence in the effectiveness of the OPLAN34A operations. CIAchief John McConesuggest­ed that they "will not seriously affect the DRVor

per CIA

PageS leP 6E6RE'fNeeMIN'fffl( I

Tap SESRET....SaMINT....)(1 Cryptologic Quarterly

tv)Norweqlan­

built "Nasty"~5t p'ltrol

boat.the prirnqry

pL'ltForm for

maritime

operations

underOPlAN

34/\

cause them to change their policies." 28 DefenseSecretary McNamara, when he returned from aninspection trip to South Vietnam in March 1964,described OPLAN 34A as "a program so limitedthat it is unlikely to have any significant effect."The operations were described by other officialsas "pinpricks" and "pretty small potatoes." 29

(D) The Johnson administration was dissatis­fied with the initial results of OPLAN 34A andsought a stronger approach. ByJune 1964, a newOPLAN, designated 37-64, had been developedjointly by the National Security Council, the JCS,and MACV. This new OPLAN called for a three­pronged approach to "eliminate to negligible pro­portions DRV support of VC insurgency in theRepublic of Vietnam." Three military optionswere put forward: ground action in Cambodiaand Laos to eliminate VC sanctuaries and supplypoints, increased levels of 34A attacks on Hanoi'scoastal installations, and South Vietnamese andUnited States bombing of ninety-eight "preselect­ed" targets in North Vietnam.i'"

(U) If the commando raids had been such fail­ures, why did they continue to be staged? Thetruth is, Washington was anxious to support theshaky regime of General Khanh, who had suc­ceeded to the presidency of South Vietnam afterDiem's assassination. Until a better plan, such as37-64, could be implemented, then doing "some-

thing," even as ineffective as the raids, was thecourse Washington chose to follow. In spite ofHanoi's gains for the first six months of 1964, ifAmerica's determination to succeed could becommunicated to Khanh, then the SouthVietnamese might be reassured of the prospectsfor victory." This was Washington's policy: toprop up Saigon. Yet, this was a structure built onunsupported assertions.

(TS//SI) The reality for Washington was thatthe increased tempo of maritime commandoraids had only raised Hanoi's determination tomeet them head on. Through June and July 1964,NSA and the navy monitoring site in thePhilippines reported that the conflict along thecoast of North Vietnam was heating up.Communications about small boat actions, com­mando landings, and high-speed chases out at seawere intercepted and reported back toWashington. What the reports showed was aNorth Vietnamese navy emboldened to moreaggressive reactions to incursions by the com­mandos from the south. For example, on 28 July,after an attack on the island of Hon Gio, DRVSwatow-class patrol boats pursued the enemy forforty-five nautical miles before giving up thechase.i'" Earlier, on 30 June, another patrol boathad taken potshots at two jet aircraft flying alongthe coast and claimed a hit.33

Tap SEeRETxeeMIHfl1'Xl Page 9

Cryptologic Quarterly laP SE6RElf(6SMINlfl*1

{~If~I) By early June, Hanoi's stepped-updefensive posture had registered in its radio traf­fic. On 8 June, NSA reported that the level ofNorth Vietnamese tactical radio communicationshad increased almost fourfold during the earlypart of June from the previous period in May,probably in reaction to attacks along its coast. Italso reported that DRVnaval patrols now seemedto cover its entire coastline." Clearly, Hanoi wasdetermined to defend itself resolutely. Whetheror not the Vietnamese believed that theAmericans were preparing for a larger war wasnot important. What was critical was that the sit­uation along North Vietnam's territorial watershad reached a near boil.35

eT~i/~t' The SIGINT support to OPLAN 34Astarted at almost the same time as the operationsbegan. Codenamed Kit Kat, the effort requiredthat the then current ceiling of 660 cryptologicpersonnel in South Vietnam had to be raised. InFebruary 1964, an increase of 130 personnel forKit Kat was approved by CINCPAC.36 The ABAmoved personnel from the Philippines to Phu Bai,and the Naval Security Group added coverage ofNorth Vietnamese naval communications to itsmission at San Miguel in the Philippines. The AirForce Security Service units at Monkey Mountainnear Danang increased their coverage of the com­munications of DRV navy and coastal surveil­lance posts. A small special SIGINT unit at TanSon Nhut Airbase, known as the Special SupportGroup (SSG), was formed in late February tocoordinate Kit Kat support between the interceptsites and the Studies and Observations Group.

(S//SI) A few last notes before we review theattacks. Itwill be necessary to limit the discussionto the role SIGINT played during the incident.Other evidential sources, such as that from theAmerican ships' own radar, sonar, and visualsightings, will be mentioned in passing simplybecause they are part of the story and cannot bealtogether ignored. However, the brunt of the fol­lowing discussion will center on the SIGINT evi­dence because of its critical role in convincing the

Johnson administration that the attack actuallyoccurred.

(S//~I) Besides the NSG detachment aboardthe Maddox (USN-467N), other SIGINT ele­ments that were involved in the events of the nextthree days included a Marine SIGINT detach­ment (USN-414T), collocated with the ArmySecurity Agency intercept site at Phu Bai (USM­626J), and the NSG site at San Miguel,Philippines (USN-27), which also had a MarineSIGINT contingent, but the latter was not desig­nated separately as was the Marine group at PhuBai. Itwould be the intercept and reporting by theMarine unit at Phu Bai and the navy site in thePhilippines which would prove critical to theevents in the Gulf of Tonkin.

-tSffSfjl

//

ret letnJ

(b) (1)

Page 10 lOP &I!GliIl!lIIG9MINT:I*1

(b)(3)-50 usc 403(b) (3)-18 usc 798(b) (3)-P.L. 86-36

19P SEGAElJ.lEi9MINlJRE1 Cryptologic Quarterly

authorities), and communications (communica­tions links among all operating elements andunits) are combined in military operations.

19P SEGAElNG9MINl.v11

~S//Sl) A large number of the reports by thevarious field sites and NSAwere issued contem­poraneously with the events themselves. A few ofthese would be cited in the various after-actionanalyses and postmortems that attended the Gulfof Tonkin. However, many more field transla­tions and reports based on the intercept duringthe period of the incidents would be issued as lateas two to four days after the crisis. The reason forthe apparent delay was that the request from NSAfor ALLintercept came only on 7 August.P"

~CiiSI) Because of the nature and enormousamount of the SIGINT evidence used here for thevery first time in discussing the Gulf of Tonkincrisis, we will need to present it in a format whichwill highlight that information. Rather than try toretell the story all at once and incorporate the newevidence into the narrative, which could be over­whelming, especially to those readers not inti­mately familiar with the events of 4 August, a dif­ferent tack will be used. We will break down theevents into their separate days. First, we willreview the details of the known engagement ofthe afternoon of 2 August. While there is no con­troversy surrounding this fight - at least there isno question that it occurred - there is an impor­tant point to draw from it: that is, the NorthVietnamese communications profile during anaval combat engagement was revealed. For easeof reference, we shall refer to this communica­tions profile as the "command and control com­munications and intelligence" system or C3I. Thisis a functional description used widely in theintelligence and defense communities to describethe process whereby the individual elements ofintelligence (information/ intelligence), com­mand and control (interaction by command

(b) (1)(b)(3)-50 USC 403(b) (3)-18 USC 798(b) (3)-P.L. 86-36

(U) After looking at the "uneventful" day of 3August, we will consider the "official" version ofthe engagement of 4 August. Although, as weprogress through the narrative, we will considerthe problems with the various other pieces of evi­dence which support the contention that anattack occurred, the emphasis will be on the SIG­INT "clinchers," that is, those reports that con­vinced the Johnson administration that an attackhad occurred. These items willbe presented whenand how they appeared to the participants.

(U) Finally, we will go back over the clinchingSIGINT "evidence" of 4 August and illustratewhat problems exist with the individual pieces.In this section, the entire scenario of what wasreported and, more importantly, what was notreported, will be considered. We will reviewclosely the technical problems with the two criti­cal SIGINTreports which prop up all of the otherevidence of an attack by the North Vietnamese. Inthis approach we will consider how the productwas developed and the serious problems in trans­lation, composition, and reporting of the infor­mation.

...(CIISI)One last item. For purposes of clarity,all time references will be marked either Zulutime ("Z,"or Greenwich Mean Time) or Golf ("G,"or Zulu +7), which is the time zone for the Gulf ofTonkin. While the actual time of the incidentswas in local, or Golf time, SIGINT reports wereissued in Zulu time. This is done because of theworldwide nature ofSIGINT reporting. The use ofZulu time allows for a consistent and universalbenchmark for analysts and recipients of theintelligence. To further confuse the issue, the U.S.Navy used Hotel time (Zulu +8) in all of its mes­sages, which is carried over into its history of theVietnam War. Then there are the events inWashington, D.C., and NSA HQ, Fort Meade,MD, which are in the Eastern time zone, or

Page 11

Cryptologic Quarterly l8P SEeREffieOMIN IJiX I

Romeo ("R," or Zulu-S hours). The latter timeswill be notated "EST"for Eastern Standard Time.All times will be in given in the military twenty­four-hour clock. So, all "P.M." times after 1200hours can be determined by subtracting 1200from the time: e.g., 1700 hours equals 5:00 P.M.

Also, it must be remembered that events in theGulf of Tonkin occurred west of the internationaldate line, so that certain events in the region wereoccurring the next day in terms of Washington'stime. For example, if something happened at1500 hours Zulu, it is reflected as 2200 hoursGolf,2300 hours Hotel and 1000 hours Romeo ofthe same day. However, a two-hour advance inZulu time, that is, 1700 hours on 4 August, means0000 hours Golf and 0100 hours Hotel time on 5August, while Washington will be 1200 hours on4 August. For ease of reference, the reader canobserve that there is a twelve-hour differencebetween Washington and the Gulf of Tonkin.

(U) Round One: The 2 August Battle

(~//SI) It all began with the fireworks of thenight of30/31 July 1964, when South Vietnamesecommandos struck at Hon Me Island (19°21'N,lOS"S6'E), located off the central coast of NorthVietnam. At first the commandos tried to landand attack a radar station, but were driven off.The raiders then stood offshore in their boats andpeppered the installation withmachine gun and small cannonfire. At the same time, two othercommando boats bombarded HonNgu Island (lS04S'N, 105°47'£)near the port of Vinh. During theattack, the Maddox had drawn offfrom the scene as required by itsorders to stay well out at sea dur­ing the night. On the morning of31 July, as the Maddox made forits patrol station near the coast,Captain Herrick observed theretreating commando boats(called "Nasties" after the manu­facturer of their boat, "Nast")

heading south. Communist communicationswere intercepted by the navy monitoring site inthe Philippines, which reported the vain attemptsby their patrol craft to catch the "enemy." 39

(~//SI) On the morning of 1August, the ASAsite at Phu Bai, Republic of Vietnam, monitored aDRV patrol boat, T-146, a Swatow-class patrolcraft communicating tracking data on theMaddox to another Swatow. At the time,between 0700G to 0730G (0030Z), the Maddoxwas located nine miles southeast of Hon MeIsland moving northeasterly. The Swatow-classpatrol craft was one of a group supplied by thePeople's Republic of China. It was a fairly largepatrol craft displacing sixty-seven tons. It had atop speed of forty-four knots and a cruising speedof twenty knots. It was armed with two 37-mil­limeter (mm) antiaircraft (AA) gun mounts, two20-mm AAmounts, and carried up to eight depthcharges. This armament limited the Swatow'srole to countering other small vessels. TheSwatow carried the Skin Head surface searchradar. The Swatows often worked in tandem withP-4 torpedo boats, acting as communicationsrelays between North Vietnamese naval com­mand centers and the P-4s, whose long-distancecommunications capability was limited. This wasa role that the Swatows filled all during the nextfew days' action.?"

CV) 5yyqtow-cbss pqhol boat

Page 12 lOP iI!GRl!lNS8MI..l ....)(1

laP 9EeRETNeeMINY,.,'l1 Cryptologic Quarterly

ES//SI) The T-146 patrol craft also orderedthe other craft to turn on its "equipment," whichprobably referred to its Skin Head radar.However, the Maddox did not intercept any emis­sions from the Swatow's radar. The NorthVietnamese boats referred to the track as the"enemy"; the equation of the term to the Maddoxwas made by Phu Bai.41

~~f/~I) Shortly after 2300G (1600Z) on 1August, the naval intercept site in the Philippinesreported that the DRV naval base at Ben Thuy(IS039'N, lOS

042'E)had informed an unidentified

entity, possibly the T-146 patrol boat, that it hadbeen "DECIDED TO FIGHT THE ENEMYTONIGHT [1 Group unreadable] WHEN YOURECEIVE DIRECTING ORDERS." The base alsoqueried the boat if it had received the "enemy's"position change from another naval entity, possi­bly an authority on Hon Matt Island (IS04S'N,105° S6'E).42 The Maddox was informed of thisintercept. A half hour after receiving the mostrecent report, Captain Herrick informed SeventhFleet and CINCPAC that he had terminated theDesoto mission because of indications of animminent attack and was now heading east out ofthe patrol area at ten knots. These indications ofan attack were from Vietnamese communicationsintercepted by the two field sites, as well as theNSG detachment aboard the Maddox.Throughout the rest of the day, these stationswould monitor the North Vietnamese ship-to­ship and ship-to-shore manual morse and voicecommunications nets. They intercepted the all­important vectoring information, the orders fromshore commands, and all the tactical communica­tions. However, the DRV boats made no hostilemoves against the Maddox that day.

ESIISI) Throughout the night of 1/2 August,according to the intercepted communist mes­sages, the North Vietnamese continued to trackthe destroyer as it remained east of Hon MeIsland, some twenty-five miles offshore. Still,nothing had happened that night, and so the

Maddox returned to its patrol line off the DRVcoast on 2 August.

(S//SI) During the early morning, theMaddox, which was heading along the northerntrack of its patrol area, was notified of furtherNorth Vietnamese tracking of its movements. TheNorth Vietnamese naval motor torpedo boatsquadron stationed at Port Wallut command wasreceiving the tracking. A coastal surveillanceradar station on Hon Me may have been orderedto begin tracking the destroyer "continuously." (Itis possible that this station had been inactive dur­ing the previous day so as to deny any informa­tion on its operation parameters from theAmerican monitoring effort.)43

(TS//SI) More ominously for the Maddox, thecommunists also had ordered P-4 patrol torpedoboats (MTB) and Swatow-class patrol boats tobegin concentrating near Hon Me Island later inthe morning.'" These patrol torpedo boats hadbeen supplied by the Soviet Union. The P-4 boatdisplaced twenty-five tons. Its top speed was fiftyknots; its cruising speed was thirty knots. It hadtwo twin 12.7-mm machine-gun mounts and twoeighteen-inch torpedo tubes. The P-4 boat alsocarried a Skin Head surface search radar. Thereporting from the American intercept sites con­strued the Vietnamese boat concentration nearHon Me as a prelude to an attack on theMaddox.45

ES//SI) NSA feared that an attack on theMaddox was in the offing. At 1002G (0302Z) on2 August, NSA sent an urgent message to a num­ber of commands and sites in the region warningof a possible attack. Included in this message wasCINCPACFLT, MACV, and the Commander, 7thFleet. Ironically, the Maddox was not on distribu­tion for this message; the DSU would havereceived the message, but it was not addressedeither. The gist of the message was simple:repeated attacks by "enemy vessels" on Hon MeIsland had led Hanoi to make preparations torepel any further assaults. NSA added that

tOP SIiQRElJieeMINTfIX I Page 13

Cryptologic Quarterly TO!' SEe'U!Tl1eOMINTllX I

(V) P-4

motor

"...THE INDICATEDSENSITIVIlY ON PARTOFDRV AS WELL AS THEIR INDICATED PREPA­RATION TO COUNTER, POS[SI]BLE THE DRVREACTION TO DESOTO PATROL MIGHT BEMORE SEVERE THAN WOULD BE OTHER­WISE BE ANTICIPATED." The problem with theMaddox not receiving these critical warningswould not be resolved until after the first attack. 46

(T5//8I) Shortly before noon, at 1144G(0444Z), the Marine SIGINT group attached tothe ASAsite at Phu Bai, RVN, intercepted a mes­sage from the T-142 Swatow-class patrol boat tothe DRV naval base at Port Wallut which statedthat "[WE] HAVE RECEIVED THE ORDERS.[T]I46 AND [T]142 DID USE [1 Group unread­able] HIGH SPEED TO GET TOGETHER [PAR­ALLEL] WITH ENEMY FOLLOWINGLAUNCHED TORPEDOES." 47 The Phu Bai sta­tion issued a Critic, short for a critical message,that alerted all relevant commands, and theMaddox, of the planned attack. In the samereport, the Phu Bai site added that four boats, T­142, T-146, T-166, and T-135, had been engagedin tracking and following an "enemy" which "isprobably the current Desoto mission." The finalparagraph of the message added that the DRVnaval facility in Port Wallut was acting as theshore-based "coordinator/director" for the sur­veillance of the probable Desoto vessel.48

(5//81) About a half hour later, at 1218G(0518Z), another Marine SIGINT detachmentstationed with the navy monitoring station at SanMiguel, Philippines, intercepted the same mes­sage. This later intercept is not unusual; it meantthat the Vietnamese were retransmitting the mes­sage to ensure its reception. However, this inter­cept was reported in a different manner than PhuBai's version. The second version was reported asa translation instead of a report. In essence, thismeant that the actual intercept was reported, andnot a restatement of its contents. Therefore, someinteresting items of intelligence, which weremissing from the first report, were included.

~h'5f7 First of all, the second version con­tained what is known as the "file time" of the DRVmessage, that is, the time when the message wasentered into a log prior to its transmission by theVietnamese radio operator on the T-142. In thiscase, a file time of 1113G was noted. This time ref­erence tells us that there was a half-hour delaybetween the receipt of the message from the orig­inator and the initial transmittal of the "attack"message (1144G/0444Z), as well as an hour's dif­ference in the second intercept (1218G/0518Z).The differences are interesting for two reasons.First of all, if the intercept times from bothAmerican sites reflect the beginning of the actualintercept of the Vietnamese transmission, thenthe half-hour difference suggests that the "attack"message was sent more than once. Why more

Page 14 yep SESRETNeeMIN=rJlX1

laP 9EeRE'Fr:e9MIN1:I*4 Cryptologic Quarterly

than once? It is possible that Port Wallut had notreceived the first transmission from T-142,although the reports from both Marine sitesimply that the message was received each time.secondly, the lag between the :file time and theactual transmission time by the Vietnamese, iffigured from the American time of intercept, sug­gests that the Vietnamese were having difficultiesin transmitting messages in a timely manner.This delay, as we shall see, becomes an importantelement in determining the DRV intentions.

(D) At about this time, the three torpedo boatshad arrived at Hon Me Island. The Maddox,which was steaming on a northeast heading awayfrom the island, had observed visually the arrivalof the three boats. Shortly afterwards, the twoSwatows were seen by the Maddox in the area ofHon Me. The five North Vietnamese boats nowwere concentrated at the island.

E~I/SI) The "attack" message was followed upby another message, this time from Port Wallut toT-146, which was intercepted at 1306G (0612Z)by the Marines in the Philippines. The messageinstructed T-146 (and probably T-142) to "LEAVE135AND TURN BACK TO [THE PATH] OF THEENEMY." The "135" that T-146 was told to leaveturns out not to have been an individual boat, asearlier reported by the Marines, but the squadrondesignator for the three P-4 torpedo boats whichwould take part in the upcoming attack. Thesethree boats made up the Section 3 of Squadron135.

(S//SH The five boats, which included the P-4boats, T-333, T-336, and T-339, departed HonMe Island at about 1300G, quite possibly on theirway to seek out the Maddox.49 Within the nexthour a set of apparently conflicting orders wassent to the Vietnamese boats. At 1409G (0709Z),Port Wallut notified both Swatow craft that the"enemy" was a large ship bearing 125 degrees(from My Duc?) at a distance of nineteen miles ata speed of eleven knots on a heading of twenty­seven degrees. This put the target on a north-by-

northeast heading, which matched that of theMaddox. The same message also included a gar­bled phrase to "THEN DETERMINE," but it isunclear what this phrase meant/" However,according to Edwin Moise, the North Vietnamesesaid that section 3 received its order to attack thedestroyer at 1350G.51 Since the :file time of themessage from Port Wallut was 1400G, this mayhave been the "attack" message.

(8//81) However, there is a complicating fac­tor. At 1403G (0703Z), just six minutes earlier,the site at San Miguel had copied a message fromHaiphong to the two Swatow patrol boats whichtold them to "ORDER 135 NOT TO MAKE WARBY DAY." Furthermore, the message added thatall of the boats were ordered to head towardsshore (though an intercept of the same transmis­sion by the Marines at Phu Bai ordered the boatsfirst to pretend to head towards shore), and thenreturn to Hon Me Island." Although this mes­sage was sent shortly after 1400G (0700Z), it con­tains a :file time of 1203G (0503Z). This meansthat this message, which constitutes an order torecall the boats, was originated some two hoursbefore the order to attack was transmitted! A sec­ond intercept of the same message added that T­146 was supposed to order the recall of Squadron135's torpedo boats.53 According to Moise, theNorth Vietnamese claimed that a recall order wassent after the attack message, but T-146 neverrelayed it in time. 54

(SiiSI) This conflict in orders by commandelements from Port Wallut and Haiphong indi­cates that there was a loss of control of the situa­tion. It further suggests that the DRV navalauthority in Haiphong had never wanted theattack to occur, at least not during the day, whenconditions were not favorable for surprising theMaddox. Since the boats continued their attackon the destroyer, it appears that the recall orderwas ignored. The deciding factor for theVietnamese boat commander may have been themuch earlier :file time of the recall order; theattack message with the more current file time

TO,. SEeltET/leOMINT/lXl Page 15

Cryptologic Quarterly lap SESRElNsaMINllI)(1

probably superseded everything else in his deci­sion.

(3//S0 At around 1400G, the Maddox's radardetected the approach from the southwest of thethree P-4 torpedo boats. Forewarned by the SIG­INT of the Vietnamese intentions to attack, theMaddox then started turning eastward, then tothe southeast and increased its speed from elevento twenty-five knots. The North Vietnamese boatsinitially may have missed the turn to the south­east by the Maddox. They probably had beenvisually tracking the American vessel." There isno SIGINT evidence that their Skin Head radarswere active, though the Vietnamese claimed theboats used it. Pictures from the action appear toshow the radar masts still upright and not low­ered in a combat position. By the time theVietnamese did react to the Maddox's change incourse, they found themselves in an unfavorableattack position. They were chasing the Maddoxfrom its rear starboard, that is, from the north­west, which meant it would take some time, evenwith a near twenty-knot advantage in speed, toachieve an optimal firing position for their torpe­do run - perhaps as long as thirty minutes beforethey could execute a turn on an attack heading.

By 1430G, Commander Ogier ordered theMaddox to go to general quarters.

(0) At about 1440G (07402) the Maddox senta flash precedence message to various commandsin the Pacific that she was being approached byhigh-speed craft with the intention of attackingwith torpedoes. Herrick announced that he wouldfire if necessary in self-defense." He also request­ed air cover from the carrier Ticonderoga, whichwas then 280 miles to the southeast. Four F-8ECrusaders from the carrier, already aloft, werevectored to the Maddox. The destroyer TurnerJoy (DD-9S1) was ordered to make best speed tothe Maddox.

(0) For the next twenty minutes, the chasecontinued. The Vietnamese boats inexorablyclosed the gap between themselves and thedestroyer. At lS00G, Captain Herrick orderedOgier's gun crews to open fire if the boatsapproached within ten thousand yards. At aboutlS0SG, the Maddox fired three rounds to warn offthe communist boats. This initial action wasnever reported by the Johnson administration,which insisted that the Vietnamese boats firedfirst.

(U) VSSTicondcroqi:

Page 16 lep SEeREliveeMIN1WX1

TeflSee"!Tf1COMINTfflCI Cryptologic Quarterly

twelve. However, the other Vietnamese boatswere unaware of what had happened and report­ed T-339 as sunk, and would continue to do so fordays afterwards.58

(V) 2 August naval action. Note the useofHotel time(Z+8/G+1). (Courtesy ofNaval Historical Center)

..... f iH

.... JM!..;.

(V) P-4 torpedo boat under fire trom Mil44ox. 2 August

(U) There would continue to beconfusion over losses for some time.The DRV claimed that two aircraft hadbeen shot down. In reality, one of thenavy'sjets had sustained wing damageduring its maneuvering for the attackand was escorted out of the area byanother jet. Both aircraft departed thearea under full power, the blackexhausts trailing from their enginesprobably appeared asbattle damage to theVietnamese sailors."The damaged navy jetwould be forced to landat Danang.

(U) A few minutes later theMaddox resumed fire. Through theshellfire, the DRVboats bore in on theMaddox. But their attacks were inef­fective. Within fifteen minutes ofMaddors first salvo,jets from the car­rier Ticonderoga had arrived andattacked the Vietnamese boats, leav­ing one dead in the water and theother two damaged. As for theMaddox, she was unscathed except fora single bullet hole from a Vietnamesemachine gun round.

(U) As for the attack­ing communist vessels,eventually all threestruggled back to theirbases. The one craft, T­339, thought to be deadin the water and claimedto have been sunk by theAmericans, and, inciden­tally, initially reportedsunkbytheVietnamese aswell, actually restartedits engines and managed to limp back to shore.On board were four dead and six woundedVietnamese sailors out of a crew complement of

lOP 8EGRE'fr','eeMINTiIX I Page 17

Cryptologic Quarterly 19P 6E6RET,.,.eeMINTinc1

Tep SEeREftteOMINTItX I

C5775I) At 1630G (2330Z), the Vietnamesepatrol boat, T-142, received orders to concentrateback at a location north of Hon Me Island, and tomake contact with another possible Swatow­class patrol boat, T-165. T-146 also receivedorders from Haiphong to send two boats out andhelp the P-4s of Squadron 135 to return/" Twodays later, on the afternoon of 4 August, T-146would report to Haiphong the damage to theboats during the attack. T-333 had been hit threetimes and suffered scattered damage to its waterpipes and lifeboat. Its auxiliary engine had beenhit and oil pressure was low, suggesting a leak.Still, the boat was assessed as being "lightly dam­aged." On the other hand, T-336 was described asbeing "heavily damaged with many holes." Its fueloil was contaminated, possibly by sea water, andthe barrel of one of its deck guns was ruined.f"The boat's crew had suffered at least two wound­ed as well. The status of both boats and T-333'screw is important to remember when we look atthe events of the later evening of 4 August.

(U) In Washington, the reaction to the attackwas relatively subdued. Since no Americans hadbeen hurt, President Johnson wanted the eventdownplayed while a stern note of protest was sentto the North Vietnamese. (Ironically, this mes­sage was the first diplomatic note ever sent toNorth Vietnam by the United States.) The presi­dent had said that we would not "run away"; yetwe were not going to ''be provocative." However,Hanoi was to be informed in no unambiguousterms that any more unprovoked actions wouldentail "grave consequences." 61

ES//SI) The lack of any reprisal was surpris­ing, especially since freedom of navigation wasone of the official reasons for the Desoto mis­sions. However, it is likely that there were miti­gating factors which caused Washington to pause.Secretary of Defense McNamara was incorrect toclaim that the Vietnamese had fired first. 62 At thesame time, the Johnson administration had seri­ously miscalculated the reaction by Hanoi to theOPLAN 34A missions. It had never considered

Page 18

that the communists might correlate the attackswith the presence of the American destroyer/"NSA,monitoring the increasing aggressiveness inDRV naval communications, had seen the possi­bility and had warned everyone, except theMaddox.

(B//BI) Furthermore, Washington, throughthe intercept of the DRV's naval communications,had seen the confused set of orders sent to theboats, which suggested that Hanoi had lost con­trol of the situation. McNamara would state, "Webelieved it possible that it had resulted from amiscalculation or an impulsive act of a local com­mander." 64 It seemed that everyone was trying todefuse the crisis.

15//5& DIRNSA, concerned about possibleaggressive reactions r I

lordered all flle sites in the region to-m-am....' ..,.ta....in.--"....e...lxtreme collection, processing, and

reporting vigilance on part of all with reportingaccomplished'•. lAW [in accordance with] estab­lished procedures and at precedence appropriateto activity, especially in regards t~ }eac­tion.,,65 A SIGINT Readiness Level Bravo Lanternwas declared..•• Under this readiness level, eight

field sites were...tasl.7:~SA. tf monitor for anyNorth Vietnamese reaction to thepatrols. The brunt 0 .. e mtercept and reportingwas handed to. the navy at San Miguel and theArmy and Marine missions at Phu BaL66

CU) The Pentagon was not going to waitaround for another incident to happen, either.Plans were put into motion to augment U.S.forces in the region, including deployment ofUnited States Air. Force combat aircraft to thePhilippines and the dispatch of the carrierConstellation to join the Ticonderoga. A seconddestroyer, the Turner Joy, already had been dis­patched to rendezvous with the Maddox. CINC­PAC ordered bothships back to the patrol area,seeing it "in our interest that we assert right offreedom of the seas." CINCPACFLT issued newrules of engagement for the next three days which

(b) (1)(b)(3)-50 USC 403(b) (3)-18 USC 798(b) (3)-P.L. 86-36

laP 6E6RET....eeMINT,.,"'1 Cryptologic Quarterly

(V) VSS Turner Joy (OD-951) in 1964

allowed both ships to approach the NorthVietnamese coast as close as eight nautical milesand four miles from its islands. The two destroy­ers were ordered to arrive at their daylight patrolpoint about one hour before dawn. One hourbefore sunset they were ordered to retire east outto sea during the night."

~Ifthe Pentagon brass was anxious to insertits ships into harm's way, Captain Herrick wasmore cautious. In an after-action report transmit­ted that evening, which reviewed the attack andthe successful American defense, he added awarning: the "DRV HAS C[AS]T DOWN THEGAUNTLET AND NO[W] CONSIDERS ITSELFATWARWITH US."He added a concern that theDRVs torpedo boats, especially at night, couldhide and then approach the destroyers with little

warning.6S He stated that the Maddox and theTurner Joy, with their five-inch guns and topspeed. of thirty-three knots, were inadequatelyarmed for defense against such boats. He sug­gested that the Desoto patrol would be safe onlywith a cruiser and continuous air cover. One lastitem was reported by Captain Herrick: theMaddox's long-range, air search radar (AN/SPS­40) was inoperative, and the fire control radar(AN/SPG-53) belonging to the USS Turner Joy,

which had just arrived toreinforce him, was out ofaction indefinitely."

(U) At the close of 2August, the NorthVietnamese boats werehiding in coastal waterscaring for their casual­ties and waiting fororders as to what to donext. The Maddox wasjoined by the TurnerJoy out at sea, and bothwere being replenishedwith ammunition and

supplies while under way.Theyhad been orderedto return to the coast at daybreak.

(8//SI) The SIGINT community could beproud of its efforts during the day. The field sitesand NSAhad intercepted, processed, and report­ed North Vietnamese naval communications insuch a rapid and clear way that everyone in thePacific command was aware of the approachingattack. It also had provided the information toWashington that suggested that Hanoi's grip onevents was less certain than was expected. At thesame time, by monitoring the DRVs naval com­munications, the cryptologists had developed apicture of the command and control elementsprior to an attack: extensive tracking by coastalobservation posts; the identification of a targetand the communication of an attack command;

and the use, if limited, of radars in locating thetarget. The Maddox had never been explicitlynamed as the target of the attack; in fact, therewas just the notation of an "enemy"; however, theanalysts at Phu Bai, San Miguel, and inside theDesoto hut had correlated the North Vietnamesetracking with the American ship. The Maddoxhad been fixedin the minds of the American cryp­tologists as an "enemy vessel" to the NorthVietnamese; they would be on the lookout forpossible new attacks. The question was, though,

TaPS!e~!T/teOIllIN IhXi Page 19

Cryptologic Quarterly leP SE6RETrr69MIN1WK1

was Hanoi spoiling for another round with theu.s. Navy?

(U) Interlude: Maneuvers andWatchfulness, 3 August

(U) On 3 August, President Johnson madepublic the instructions he had issued to the Navyearlier. He said that the patrols would continue inthe Gulf of Tonkin, that they would be reinforcedby another destroyer with combat aircraft over­head. He added that if attacked in internationalwaters, U.S. forces would attack any force withthe intention of not just driving it off, but of"destroying it."

(U) At the same time, the State Departmentpublicized the note it had sent Hanoi protestingthe attacks. It concluded with the words "TheUnited States Government expects that theauthorities of the regime in North Vietnam willbeunder no misapprehension as to the grave conse­quences which would inevitably result from anyfurther unprovoked military action against theUnited States forces." 70

(0) Despite the increased North Vietnamesevigilance and the observed sensitivity toAmerican and South Vietnamese naval activity inHanoi's territorial waters, COMUSMACV wentahead with an OPLAN34A mission scheduled forthe night of 3-4 August. In accordance with anearlier agreement, the Maddox and Turner Joywere advised to avoid sailing in the area boundedby the 17th and 18th parallels. A 34A missionagainst the radar site at Vinh Son (17°57'N,106°30'E), which involved a four-boat task group,set sail at 1510G (081OZ) on 3 August. At mid­night it shelled the radar station. One of the boatsbroke off and attacked a nearby security post andwas pursued for a short distance by a NorthVietnamese patrol craft.

(U) By mid-morning of 3 August the twodestroyers were heading to their patrol station,which was about 100 miles northwest of the new

34A mission area. They expected to be on stationby early afternoon. However, this location keptthem in the area of the island of Hon Me, whichwas the focus of DRV naval activity during theensuing day and night.

(S//S£) Meanwhile, the North Vietnamesewere concerned with the salvage of their damagedboats. Just past midnight on 3 August, T-142 andT-146 were in the area of Hon Me Island trying tocontact another Swatow, T-l65, as well as findthe missing boats from Squadron 135.At 0300G(2000Z), T-142 sent an after-action report to theT-146 (for relay to Port Wallut), which highlight­ed the previous afternoon's combat. It included achronology of the various actions the squadron'sboats carried out from 0935G to 1625Gwhen theyattacked the Maddox.71

(8//S£) Even by mid-afternoon of 3 August,naval headquarters in Haiphong still did notknow where the torpedo boats were and demand­ed that the Swatows inform it when they knewtheir situation." However, the SIGINT site at PhuBai misconstrued this search and salvage activityas a prelude to a potentially dangerous concen­tration of enemy boats. It issued a Critic at 1656G(0956Z), which placed six DRVpatrol and torpe­do boats near Hon Me lsland.P However, thereport was wrong in that it identified thesquadron reference "135" as a boat, as well aslocating the two torpedo boats, which, at the time,were still missing. The ominous concentration ofboats simply was not occurring. However, thisincident revealed how tense the situation hadbecome. It also illustrated a precedent by the fieldsite at Phu Bai for misinterpreting Hanoi's inten­tions.

(TSI/8I) Almost as soon as the two destroyersarrived on station south of Hon Me Island in earlyafternoon, they were shadowed by a DRV patrolboat which tracked them using its Skin Headradar." The tracking continued through theafternoon into early evening. The Haiphong navalauthority and the Swatow boats near Hon Me

Page 20 lOP &iQRETJJ69MINTrlK1

Tep SE6RETN6eMINTfRE1 Cryptologic Quarterly

exchanged position information on the twodestroyers as they moved from the north to southand back north on their patrol." At one point,another Swatow, T-379, erroneously identified asan 80-1 class subchaser, was ordered to go outand observe "different targets," which probablyreferred to the American ships." The NorthVietnamese also detected aircraft in the area ofthe Desoto patrol, though it is unclear from theirreport whose aircraft these were. However, thecommander, 7th Fleet, had ordered a continuouscombat air patrol accompanying the two destroy­ers. The navy jets flew their cover to the east ofthe Desoto position so as to avoid infringing onDRVair space."

-tSi/~ By early evening, Haiphong orderedT-142 to track the Desoto patrol. T-379, whichearlier had been instructed to observe the Desotopatrol, had sailed to Hon Ngu Island (ISo4S'N,105°47'E). It had arrived at 2250G (1550Z) andreported that the situation at sea was "peace­ful.,,78

iStT-142 took up a position to the north of thetwo destroyers and stayed with them, reportingthe location of the American ships to Haiphongeither directly to naval HQs or relaying reportsthrough T-146. Both U.S. ships reported beingfollowed from the north at a distance of thirty­eight miles by a DRV patrol craft using its SkinHead radar. By this time, 2252G (1552Z), theDesoto patrol was heading southeast out of thepatrol area as had been instructed earlier.79

Tracking of the destroyers ended soon after whenthey were out of range.

,",~{fsa Meanwhile, the main concern of theDRV navy was the recovery operation for theboats damaged during the 2 August attack. Latein the night of 3 August, Haiphong informed T­142 that the salvage tug Bach Dang would soonleave Haiphong (it was not clear from the inter­cept if the time of departure was OlOOG, 4August/1S00Z, 3 August) and head towards HonMe Island to tow T-333 and T-336 back to

Haiphong or Port Wallut, which was their unit'sbase. 80 It was expected that the tug would arriveat about noon on 4 August. Meanwhile, T-146 wasordered to stay with the two damaged boats fromSquadron 135 and report their position and sta­tus.

(0) So ended 3 August. That evening's 34Araid on Vinh Son was protested by Hanoi. In itscomplaint, it accused the two destroyers of partic­ipating in the raid. Although the DRVs owntracking of the two ships had ceased some hoursbefore, and they could not be certain of where theAmerican ships were, the Vietnamese hadinferred anyway that the Desoto ships wereinvolved. It may not have been the right conclu­sion, but the Vietnamese believed it. Washingtonstill did not think that Hanoi would act.

(U) Round 2: "Everything in Doubt"­The 4 August Action

-f.8IIStTAt 0600G (2300Z) on the morning ofAugust 4, the two destroyers turned westwardtowards the DRV coastline to begin their day'spatrol. By 1300G (0600Z) they returned to theirduty station off the coast of North Vietnam nearThanh Hoa (20

00S'N,

105°30'E), known as point"Delta," where they began to steam to the south­west along the Vietnamese coast. The air coverfrom the Ticonderoga again was overhead and tothe east. An hour later, the Maddox reported thatit had another shadow, this time fifteen miles tothe east. The identity of this shadow cannot bedetermined.

(S.,tfSI)- The North Vietnamese had beentracking the Americans. Haiphong informed T­142 at 1610G (091OZ) that they had located thedestroyers near 19°36'Nand 106°19'Etraveling ona southwest heading. However, this last positionof the two ships had been acquired by the NorthVietnamese some two and one-half hours earlierat 1345G (0645Z).81 At approximately 1600G(0900Z), following his operational directive from

19P 6E6RETNesMINT/nc I Page 21

Cryptologic Quarterly TQP SiGAET/J69MIN1WiE1

CINCPAC to be clear of the patrol area by dark,Herrick ordered the patrol to head due east.

fSIIS!) At 1115Z (1815G), the naval SIGINTdetachment aboard the Maddox received a Criticfrom the Marine SIGINT unit collocated with theASA at Phu Bai, which stated, "POSS DRVNAVAL OPERATIONS PLANNED AGAINSTTHE DESOTO PATROL TONITE 04 AUG.AMPUFYING DATA FOL." 82 Twenty-five min­utes later, Phu Bai issued a follow-up report at1140Z (1840G) which reported, "IMMINENTPLANS OF DRV NAVAL ACTION POSSIBLYAGAINST DESOTO MISSION.,,83 The reportwent on to add that three DRVboats, T-142, T­146, and T-333 had been ordered at 092'7Z(l627G), the time the message was intercepted byPhu Bai, to "make ready for military operationsthe night of 4 August." Although the report didnot specify the nature of the military operations,the Marines appear to have concluded that it wasan attack against the Desoto. The NSG detach­ment informed Herrick. Within an hour, at1240Z, he informed CINCPAC and other com­mands that he had received "INFO INDICATINGATTACK BY PGN P-4 IMMINENT. MY POSI­TION 19-1ON 107-00E. PROCEEDING SOUTIl­EAST." 84 At this point, the two ships were abouteighty to eighty-five nautical miles from the near­est DRVcoastline and began to head southeast attwenty knots.

~ A short time later, just after 1300Z(2000G), the Desoto vessels acquired their firstradar contacts. The Maddox reported that it haddetected "two skunks" (surface contacts) andthree ''bogies'' (air contacts) on its radars. Thesurface contacts were about forty to forty-fivemiles to the northeast of the two destroyers, put­ting them about 100-110 miles away from theVietnamese coast at sea, but very close to HainanIsland." (The appearance of aircraft returns(bogies) on the destroyer's radar has generallygone unremarked upon by various commenta­tors. Herrick speculated that these were terrainreturns. Whatever the case, these false "bogies"

suggest Maddox's air surveillance radar was stillmalfunctioning.) The Ticonderoga ordered thefour jets on CAPto cover the two ships. It scram­bled four more AlH Skyraiders. Within an hour,the aircraft were overhead.

ts1'1\t about 2045G (l345Z), Herrick reportedhe had lost the original surface contacts: they hadnever closed to less than twenty-seven miles fromhis own ships. At 2108G (l408Z), Maddox detect­ed another return - first identified as one boat,later thought to be several boats in a tight forma­tion - this time only fifteen miles away to thesouthwest, moving towards the destroyers at thir­ty knots. Nine minutes later, naval A-4 Skyhawksflying air cover were vectored towards the sup­posed boats. Although the pilots could see thewakes of the destroyers clearly, they could see noboats at the point the radar indicated. At 2131G(1431Z), this radar return disappeared."

(U) Then at 2134G (l434Z) came the mostimportant radar contact of the entire incident.What appeared to be a single boat suddenlyappeared on the Maddox's radar screen east ofthe two destroyers at 9,800 yards and closing atnearly 40 knots. The Turner Joy detected anoth­er object approaching, but on a different heading,distance, and speed. According to Marolda andFitzgerald, the navy claimed that this was thesame return as the Maddox's.87At 2137G(l43'7Z)at a distance of 6,200 yards from the Desoto ves­sels, the return tracked by the Maddox appearedto make a sharp turn to the south. This maneuverwas interpreted by the Maddox combat informa­tion center as a tum after a torpedo run. If thiswas a torpedo launch, then it was an extraordi­narily desperate one. Hanoi's tactical specifica­tions for its P-4s called for torpedo launches atranges under 1,000 yards. At over 6,000 yards, itwas unlikely a torpedo launched at a moving tar­get could hit anything." The sonar operatoraboard the Maddox detected a noise spike on hisequipment, but did not report it as a torpedo. Thisconclusion was reached on the CIC. However, theTurner Joy never detected any torpedoes on its

Page 22 lOR SECRETJ.lCQMIN'J/RE4

Tap 9EeRET(ieaMIN'ff~1 Cryptologic Quarterly

sonar. Nor did it detect any torpedoes at all on itssonar that night.89

CD) For the next fifteen minutes all surfacecontacts were gone from the radars of the twodestroyers. Then, at 2201G (150lZ), more con­tacts were detected coming from the west. Nowthe thickest part of the naval action commenced.The two destroyers gyrated wildly in the darkwaters of the Gulf of Tonkin, the Turner Joy fir­ing over 300 rounds madly at swarms of attack­ing North Vietnamese boats - maybe as many asthirteen - and dodging over two dozen torpedoes.

(D) At 2140G (1440Z), Herrick informedCINCPACFLT that he had commenced firing onthe attacking PT boat. The Turner Joy had begunfiring at its return shortly before this. Bothdestroyers had a difficult time holding a radarlock on their targets. Within five minutes, thereturn on Maddox's radar, which was movingaway from the destroyers, disappeared from itsscreen at a distance of about 9,000 yards. The onethat the Turner Joy was tracking kept approach­ing, and at a distance of about 4,000 yards, it dis­appeared as well.9 0

19·

1R"

10'1'-

10" so

(D) It should be mentioned again that theradar returns from both ships were not continu­ous trackings. Rather, they were mostly flashingreturns, that is, they appeared on the scope, heldfor a few sweeps of the radar, then disappeared.Other targets would suddenly appear a few milesfrom the destroyers, hold for a while and then dis­appear. They came from all directions. As eachreturn was logged, it was assigned a target desig-

nator, a single letter. One offi­cer from the Turner Joydescribed the confusion of pro­liferating targets this way: "Wewere getting blotches on thethe radar screen - nothing realfirm, so we were whackingaway at general areas withproximity fuzes, hoping to getsomething." 91 A target wouldapparently be hit and then dis­appear as if it had completelyand instantaneously incinerat­ed in an explosion - contraryto what had happened twodays earlier when the NorthVietnamese PT boats wouldtake several hits but remainafloat afterwards. TheMaddox's main gun director

Another twenty-four star shells had been fired toilluminate the area and four or five depth chargeshad been dropped to ward off the pursuing boatsand the torpedoes. The Maddox vectored over­head aircraft to the surface contacts, but time and.again the aircraft reached the designated point,dropped flares, and reported they could not findany boats. By the time the attack was consideredover at 2335G (1635Z), Herrick reported twoenemy patrol boats sunk and another damaged.(The count of the damaged boats varied; Herrickbelieved that the DRVboats sank one of their ownaccidentally. It is not understood how he arrivedat this conclusion, except as a misinterpretationof the radar data which itself was of dubious qual­ity.)

0500H3 Aug

108·

Gulf ofTonkin

'07·

.....-_.-,0700H

(V) GulfofTonkin track,3-5 August 1964(Courtesy ofthe N;wal HistotiQI Center)

Thanh Hoe.

20·

-19'

I

"1...""

'8· i.\,<,,

1080 •

lOP &EGAE'FgeaMIN'fI'~ I Page 23

Cryptologic Quarterly 1'611 SEeRE1'l/eaMINTl/l1

maintained that the ship was never able toacquire any of the targets during the battle; he fig­ured he was shooting at the high swells broughton by the storms." Ironically, during all of thislatter action, the Maddox never fired a round; itsradar never acquired another target after the ini­tial one detected two hours earlier.93

(U) The sonar returns of the supposed torpe­do attacks were later determined to be a result ofthe high-speed maneuvering by both U.S. ships.As we saw above, the first "evidence" of a torpedolaunch by the enemy boats came from radar.When one of the radar tracks turned away to thesouth from a westerly heading, this was interpret­ed by the Americans as a torpedo launch. Thesonar rooms in both destroyers were then alertedto a possible torpedo attack. Four crewmenaboard the Turner Joy thought they saw a "whitestreak" in the water as the ship turned.?" Bothvessels had then gone into wild evasive maneu­vers to avoid the torpedoes that were thought tohave been launched against them. It was thishigh-speed gyrating by the American warshipsthrough the waters that created all of the addi­tional sonar reports of more torpedoes. Everytime one of the destroyers changed course, thesonar reported the distinctive high-speed soundsof torpedoes. Eventually, Herrick and the otherofficers realized what was happening: the rud­ders of the two ships had caused the high-speedreturns when they reflected the turbulence of theships' own propellers."

~Within an hour of the end of the attack,Herrick relayed his doubts about the attack in anafter-action report. After reviewing the number ofcontacts and possible sinkings, he stated,"ENTIRE ACTION LEAVES MANY DOUBTSEXCEPT FOR APPARENT ATTEMPTEDAMBUSH AT BEGINNING." 96 Herrick then sug­gested in the morning that there be a thorough airreconnaissance of the area for wreckage. In a fol­low-up message, Herrick added that the Maddox

had "NEVER POSITIVELY IDENTIFIED ABOAT AS SUCH."97

(U) Herrick's doubts did not sit well withWashington. Since the first Critic warning of theattack, which had arrived at 0740 EST,Washington had been following the action in theGulf of Tonkin. At 0925 EST, SecretaryMcNamara had called the president with thenews of the imminent attack. At 1000 EST theflash message from the destroyers that they wereunder attack reached the Pentagon. Within threehours after the attack ended, 1400 EST, PresidentJohnson had already approved a retaliatory strikeagainst North Vietnamese naval bases to be car­ried out at 1900 EST, 4 August (0700G, 5August).

(U) Precisely why President Johnson ordereda retaliatory strike so quickly is not totally clear,especially when there was conflicting evidence asto whether it had actually occurred. Johnson wasin the midst of a presidential campaign and hisopponent, Republican senator Barry Goldwaterfrom Arizona, a noted hawk, would have gainedin the race ifJohnson had hesitated or refused toretaliate. Johnson, even in his pose as a moderaterelative to Goldwater, could hardly appear weakbefore a public audience demanding a counter­strike." It also has been suggested that whenJohnson first learned of the possible attack, thatis, the first Critic issued by Phu Bai, he decided touse the warning as an excuse to get Congress topass what was soon to be known as the Gulf ofTonkin Resolution."

-tSTWhatever the president's own rationalefor ordering the air strike, he required immediateverification of the North Vietnamese attackbecause of the doubts that started to be openlyexpressed within the administration. At around1400 EST, Admiral Ulysses S. Sharp, CINC­PACFLT, called the Pentagon with the news that"a review of the action makes many reported con­tacts and torpedoes fired 'appear doubtful' "because of freak weather, over-eager sonar oper-

Page 24 Tall SI!CR:EnICOMIN IflX,

T9P SESRElt'r'eeMIN111)(1 Cryptoloqic Quarterly

ators, and the absence of visual sightings.l?"McNamara called Sharp, who added that therewas "a little doubt on just what exactly wenton.,,101 Messages buzzed back and forth betweenWashington and the Pacific, demanding informa­tion and then getting contradictory evidence ofthe attack. The Desoto mission reported thatexcept for possibly the first torpedo report at2159G (1459Z), all others were caused by reflec­tions off the two destroyers' screws.102 At thesame time, Herrick reported that the air coverfrom the two carriers was unable to locate the tar­gets because of poor weather. Yet the carrierTiconderoga transmitted its own evaluation inwhich the pilots had "REPORT[ED] NO VISUALSIGHTINGS OF ANY VESSELS OR WAKESOTHER THAN TURNER JOY AND M[ADDOX].WAKES FROM TURNER JOY AND M[ADDOX]VISIBLE FROM 2-3000 YARDS." 103 Crews fromthe two destroyers reported seeing nothing forcertain. One sailor thought he had seen flashes ofgunfire, but wasn't sure.

('fBI/51) Then, like a classic deus ex machina,along came a second SIGINT report that seemedto clinch the case for an attack. This report was atranslation issued by NSAon the 4th of August at1933Z (1433 EST in Washington) and was leapedupon by administration officials, especially thesecretary of defense, Robert McNamara, as directevidence of the attack. What this translationappeared to be was a sort of North Vietnameseafter-action report. An unidentified NorthVietnamese naval authority had been interceptedreporting that the DRVhad "SHOT DOWN TWOPLANESIN THE BATTLE AREA," and that "WEHAD SACRIFICED TWO SHIPS AND ALL THEREST ARE OKAY." It also added that "THEENEMYSHIP COULDALSOHAVEBEEN DAM­AGED." 104

(U) At 1640 EST, Admiral Sharp again calledMcNamara with more information on the attack.Just before 1700 EST, McNamara and the JCSmet to evaluate the evidence on the attack. Theyconcluded that it had occurred and that five fac-

tors were critical: "(I) The Turner Joy was illu­minated [by a searchlight] when fired on by auto­matic weapons; (2) One of the destroyersobserved cockpit [bridge] lights [of one of theDRVpatrol boats]; (3)APGM 142 had shot at twoU.S. aircraft (from COMINT); (4) A NorthVietnamese announcement that two of its boatswere 'sacrificed' (from COMINT); (5) AdmiralSharp's determination that there was indeed anattack." 105

(D) Of the five pieces of "evidence," two werefrom the same NSAproduct issued that afternoon(EST). If the two pieces of visual evidence - thesearchlight and cockpit light reports - were con­tentious, the SIGINT was, in the minds of the sec­retary of defense, the JCS, and the president, the"smoking gun" evidence needed to justify the airstrikes on North Vietnam.'?" So, at 0700G

(U) Burning North Vietnamese patrol boatilfter 5 August strike

T9P SESRE1NeeMINliV)(1 Page 25

Cryptologic Quarterly TepSEeREli\'eSMINJ.Y*1

(OOOOZ) on 5 August, CINCPAC received theorder to execute the retaliatory raid, codenamedPierce Arrow. At 1030G (0330Z), naval strike air­craft from Ticonderoga were launched. By earlyafternoon they hit several targets in the DRV,including almost all of its naval installations.

(U) The Silent Dogs: What theSIGINT Really Did (and Did Not)Report

(S/fSI) Events surrounding the apparent sec­ond attack had been driven almost exclusively bySIGINT. Herrick's personal doubts, the falsesonar readings, the confused radar returns, andthe pilots' reports, all subverted the validity of theattack reports. But not the SIGINT. For theJohnson administration, both reports - the initialCritic reporting the North Vietnamese prepara­tions for operations, and the after-action report ­acted as factual bookends, propping up the otherpieces of contentious evidence. The details of theattack, as contradictory as they were, could bemassaged or explained to fit the scenario set bythe SIGINT. For example, since there were noreported shootdowns of American aircraft thatnight, then the North Vietnamese report ofdowned U.S. planes must have resulted whenthey had confused illuminating flares for fallingaircraft.107

ESf/BI) However, there were many problemsspecific to the SIGINT information whichemerged almost as soon as it was being reported.In this section we will reconsider what happenedthat night using all of the relevant SIGINT. Wewill begin with the initial order to the Vietnameseboats ordering them to make ready for militaryoperations.

fS/";'8I) Exhibit A: The FirstAttack Message

(BffSf) The first product, the "attack" mes­sage, issued at 1115Z (1815G), reported only thefact that there was a possible DRV naval opera­tion planned against the Desoto patrol. At 1140Z

(1840G), this was followed up by a second reportfrom Phu Bai which contained a number ofdetails, such as that T-146 and T-333 were tocarry out military operations with T-142. Unlikethe messages of 2 August, there was no referenceto an "enemy," no tracking to equate to theDesoto patrol, or any indication of the nature ofthe operations to be carried out by the boats. Infact, the original intercepted message was onlythe first part of a larger message, the rest of whichwas not intercepted. So, what might have been inthe latter part is unknown, except that it mighthave amplified the meaning of the type of opera­tion the boats were involved in.

(BffSf) What made this intercept a Critic wasthe interpretation put to it by the Marine SIGINTsite at Phu Bai, which stated that this was an"OPERATION PLANNEDAGAINSTTHE DESO­TO PATROL." 108 The follow-up report from PhuBai amplified the original Critic and maintained,as well, that the attack was against the Desotomission.i'" When one considers the events of 2August, this interpretation was not totallyunfounded; one could see a reference to a militaryoperation being directed against the Americanwarships. However, the text of the intercept nevermentioned a target or any objective of the militaryoperation, or even the nature of the operation. Aswe shall see soon, not everyone who saw thisintercept jumped to the same conclusion that anattack against the American ships was beingplanned.

(S//SI) Another problem is that the decryptedVietnamese phrase for military operations, hanhquan, has an alternate meaning of "forced or longmarch or movement," which, in a nautical con­text, could refer to a voyage by both T-146 and T­333. As it turns out, this is the activity that theintercept was actually alluding to.

(SffSI) For at 1440Z, almost at the precisemoment that Herrick ordered his two destroyersto open fire on the approaching radar returns, thePhu Bai intercept site issued a spot report which

Page 26 l'QP &E6RETHeSMINT;tll

Z 0 lI41115Z ZiB

FIl USN 414T

'fO" !ECI'ETileOMINTi'1XI

SE6RET KIMBO

Cryptologic Quarterly

SF;CRET KIMBO

TO USN 46711

INFO!.......__...I

USM 27

aSAP"C REP VIETNAM (C)

DIRNSI\

S I!l eRE T KDlBO

DESOTO

I.. PDSS. ORV IIAVAL OPERAnO!! PLARIlI!:D AGA.IHST THI DESOTO PM'ROL

SBCRET KIMBO

~ phu Bai Ctitic alerting Desoto patrol topossible attack

stated that both DRVtorpedo craft, T-336 and T­333, the latter of which earlier had been reportedready to attack the Desoto patrol, were, in fact,being readied to be towed to either Haiphong orPort Wallut. This second report carried twosalient points: First, at 1946G (1246Z), SwatowT-142 reported to Haiphong that the tug BachDang was unable to return to port. T-142 alsoincluded the statement that if the ship [BachDang] "MET THE DESOTO MISSION, IT WASTO [AlVOIDTHEM." 110 Besides being a warningabout the Desoto ships, the message also impliedthat the North Vietnamese thought that thedestroyers were close enough to shore to be athreat to DRV vessels, whereas, at this time, theAmerican ships were far out at sea. In all proba­bility, the North Vietnamese had lost track of theAmerican destroyers (an issue which we will dis­cuss further on in this narrative).

liTo 145 & Cadre 101oal. [I U].- so that wben you have orders the

146 ca.D. .tow [4gr MJ the 336 back. 1.£ t.he t.a~1t truck hasn't yet

come to SUPPly you, then the 146 can transfer fuel to tM 333.

With regard to order., the 333 will carry out .1'l111tary operations

lndependentlt. vlt,h 146. (Continued)

{continuation unava11able~

4 AUlt ~927Z

US!f 414'1' inter'cepted. at .409Z7Z Da.te/time of file: 041558G

~~CIl:lIT KIMBO

1"57'7'S1t Translation ofthe Irrtercepted original oftheRafulck" order

ESf/Sf) The second point of the Phu Baireport was that at 2031G (1331Z) T-142 hadinformed an authority in Port Wallut that the tugwas towing the two craft from Squadron 135.Theanalysts at Phu Bai added this comment to theend of their report which read, "WITH THE MTB336 ADDED TO ITS STRING, IT AP[PE]ARSTIIAT T333 WIll. NOT PARTICIPATE IN ANYMILITARY OPERATIONS." So, the boats origi­nally reported being ready to attack the Desotopatrol, were incapable of even moving on theirown!

(SffSI) In fact, this attempted salvage of thetwo damaged torpedo boats would occupy theefforts of Hanoi's sailors for much of the night of4/5 August. The Vietnamese would try variousmethods of getting the two damaged P-4s to aport for repairs. During the 2300G hour, T-146was ordered by Haiphong to escort the BachDang as it returned to base. When that was com-

(b) (1)

(b)(3)-50 USC 403(b) (3)-P.L. 86-36

lap 5EeRE'ffle6MINTtIX I Page 27

Cryptologic Quarterly 'F9P 6E6REl:rJeeMIN'mcI

pleted, T-146 was ordered to Bay Chay, a pointnear Haiphong harbor.!" Shortly afterwards, T­142 informed Haiphong that the very busy T-146was now to tow T-336 back, but since the latterboat was short of fuel, the T-333, which was shortof oil but under tow from the Bach Dang, couldtransfer one to five tons of its fuel to its sister ves­sel.11Z At 1830Z on 4 August (0130G on 5 August),the navy monitoring site at San Miguel intercept­ed T-142's report to Haiphong that T-146 hadcompleted its preparations for the two torpedoboats by OlOOG 5 August (1800Z 4 AUgust).l13 So,in reality, none of the boats named in the originalattack Critic in fact participated in anything butsalvage efforts.

iStRemember, Captain Herrick did not knowthat the original Critic was really an interpreta­tion, and that there was no explicit reference to anattack on his ships. He accepted the Critic's con­tents as intercept of actual Vietnamese plans toattack his ships when he informed theTiconderoga task group commander of his deci­sion to leave the area. He added his own twist tothe report to include specifically the unsupportedamplification mentioning the involvement ofNorth Vietnamese P-4 torpedo boats when onlyone was mentioned as a potential participant inthe unidentified operations, and then only if itcould be refueled.i'"

(S//81) The possibility that, even if the inter­pretation was incorrect, the Marine Critic wasjustified in light of the events from two days ear­lier, does not stand up when we consider thatanother site, the navy intercept station at SanMiguel, Philippines, had translated the same"operations order," but reported it in a much dif­ferent fashion. The navy translated the sameintercept and then reported it at a Priority prece­dence, two levels below a Critic (or one levelabove Routine). The navy analysts titled thereport "REPLENISHMENT OF DRV NAVALVESSEL." The San Miguel report translated thecritical sentence as: "T146 SUPPLY FUEL FOR

THE 333 IN ORDER TO GIVE ORDERS TO PUTINTO OPERATION ((2 GR G)) WITH T146." 115

(SffSI) The difference (and correctness/incorrectness) between the translations is notimportant as much as the fact that San Miguelviewed the information as nothing more than therefueling of the damaged torpedo boats. This wasin line with an earlier intercept of a query fromHaiphong to T-142 asking ifT-333 had been refu­eled yet.116 Unfortunately, because the San Miguelversion was a lower precedence, it was releasedmuch later. In fact, it came out at 1838Z (0038G),some two hours after the destroyers had stoppedshooting.

(SffSI) The quandary created by the reportsabout the salvage operations is this: If the origi­nal suspect vessels, the two Swatow-class patroland two damaged P-4 torpedo boats, were notparticipating in the anticipated "attack" againstthe Desoto patrol, then who exactly was going toattack? No other messages had been interceptedwhich suggested that any other DRV boats werehanded the mission of attacking the Americandestroyers. In fact, there was no intercept at allwhich hinted at an attack; nothing at all like whathad been intercepted on 2 August. So, if the orig­inal culprits were involved in salvage operations,then just what was going on in the Gulf ofTonkin?

-f:B1ISlJ. For NSA and the rest of the SIGINTparticipants, the second Phu Bai report shouldhave acted as a brake to any further reportingabout an attack. It directly contradicted the inter­pretation - remember, it was an interpretationonly - contained in the initial Critic whichclaimed an attack was being prepared. At thispoint, all the SIGINT community could accurate­ly state was that there was no signals intelligencereflecting a planned or ongoing attack against theDesoto mission.

ffS//SI) Except this is not what happened.The second Phu Bai report was not used to report

Page 28 'F9P SIiSRE;HG9MIN'FJRE1

1'011 SEeRE'fr\-eeMIN1WiE1 Cryptologic Quarterly

what was going on in the Gulf of Tonkin. Instead,the problem posed by the second Phu Bai reportwas handled in a curious manner. Late on 4August, Washington (050130Z August 1964),NSA issued a Gulf of Tonkin situation reportwhich covered the events of 4 to 5 August. At theend of the report, NSA added these interestingsentences: "ALTHOUGH INTI1AL MESSAGESINDICATED THAT THE T142, T146, AND T333WOULD BE INVOLVED IN THE ATIACK ...SUBSEQUENT MESSAGES [not further identi­fied in the report - a curious lapse by NSAwhichwe will address in detail later] SUGGESTTHATNONE OF THESE [BOATS] WAS INVOLVED.REPORTSFROMTHE MADDOX THAT IT WASUNDER ATIACK SOME SEVENIY NAUTICALMILES NORTHEASTOF THE NAVAL BASE ATQUANG KHE SUGGEST THAT NAVAL UNITSSUBORDINATE TO THE SOUTHERN FLEETCOMMAND ... WERE INVOLVED...." 117

fTSiiSI) However, the effort to find "culprits"only compounded the errors: the only boatsknown to be stationed permanently at Quang Khewere Swatow-c1ass patrol boats which did notcarry torpedoes.f" All P-4 torpedo boats stagedfrom Port Wallut far northwest of the action.Accusing the Swatow craft of participating in theattack was no "solution"; in fact, it only added tothe confusion. In reality, though, this statementby NSAwas a vain attempt to cover the problemof the contradictory report from Phu Bai. It wasnothing but speculation - ignorant speculation atthat. Furthermore, this summary report still didnot address the issue of the total lack of interceptof any North Vietnamese attack command andcontrol communications.

(U) Fingering the Swatows as the culpritsonly made the "attack" scenario more improbablefor another reason. The distance from Quang Khenaval base (li46'N, 106°29'E) to the reportedfirst radar plot by the Maddox, forty to forty-fivenautical miles northeast of its position, is about120 nautical miles. However, this distance shouldnot be construed as a "straight line" dash from

Quang Khe. Because the DRVboats were "detect­ed" coming from the east, they would have had totravel in a long arc northward and then southeastaround the American destroyers which werespeeding to the southeast. Also, remember thatthe Maddox and Turner Joy did not "detect"these boats until they approached from the east,so the route to the north of the American destroy­ers had to be at a distance sufficient to avoid dis­covery by radar. This lengthens to a distance ofaround 180 nautical miles. Since the "attackorder" was issued at 1115Z and the initial radarplot was at 1336Z(and we are presuming that thepostulated boats left at the exact time of the firstintercept, or were soon under way at the time),then the boats would have had to have been trav­eling at a speed of nearly seventy miles per hour(about 110kph) to have been where the Maddoxfirst detected them - at a rate some 58 percenthigher than the Swatow's known top speed!

(D) The only other base from which the"attack" could have been staged was Port Wallut,which was the base for the P-4 Squadron 135.Thedistance from Port Wallut (2t"13'N, 107°34'E) tothe initial point of detection by the Desoto radarsis about 140 nautical miles. However, the sameproblem exists here as for Quang Khe, though notquite as extreme, for the P-4s. The scenario pre­sumes that they would have been moving at a lit­tle less than seventy miles per hour, or a good 40percent higher than the boat's listed maximumspeed.

~IfSIr-Another possibility to consider whenlooking at the "attack message" is that there wassome other activity to which the "military opera­tions" (if that is the interpretation one couldhave) might have referred. In fact, there wassomething else going on that night of 4/5 Augustwhich is seldom mentioned in the public record:a maritime OPLAN-34A mission was, in fact,moving northward along the DRVcoastline at thetime when the American destroyers were shoot­ing away at those radar returns. The Marolda andFitzgerald history of the U.S. Navy in Vietnam

'fell SEeRE'fHeeMIN'f:Rt1 Page 29

Cryptologic Quarterly Tep&E6RETH6eMINl1AE1

fails to mention the ongoing 34A mission. OfficialWashington as well never mentioned this 34Amission. In classified hearings in February 1968,Secretary of Defense McNamara never men­tioned this mission, claiming that the last oneprior to the 4 August attack occurred on the nightof 3-4 August. Obviously, if the 34A mission ofthe night of 4-5 August were known at the time, itwould have undercut Washington's claim thatnothing else was happening that night whichmight have provoked Hanoi.

(D) This 34A mission had been scheduledback at the end of July by COMDSMACV, whichthen had informed Washington of the missionsplanned for all of August. This particular foray'smain objective was the shelling of the island ofHon Matt. It is not certain when this mission left:Danang, though it was normal for the boats todepart in the late afternoon to take advantage ofdarkness by the time they reached the DRVcoast­line. So a departure time between 1500G and1600G (0900Z) would not be too far off.

(S/JSI~ At 2316G (l616Z) the Marine missionat Phu Bai intercepted a message from the DRVnaval HQ in Haiphong to T-142 that six enemyraiding vessels had been located somewheresouth of Thanh Hoa (20

000JN,105°30'E). (The

actual position is confusing due to a garble in thetext transmitted from Phu Bai. Neither the timeof the enemy boats' position nor their course isclear.)"" This intercept occurred only a few min­utes before the JCS approved an urgent recallorder from CINCPACFLT for the 34A mission tobe discontinued and return to Danang immedi­ately.120 It is possible that the Kit Kat support ele­ment may have passed this intelligence to theMACV/SOG, which in tum began the recall.

ES/tSI) In light ofwhat finally transpired withT-142 and the two P-4 torpedo boats, it seemsthat they were not part of an defensive planagainst the raiders. That this Swatow receivedthe message about the raiders does not seem oddin light of the fact that T-142 seems to have served

as some sort of radio relay for other boats or as acommunications guard vessel for all DRV navaloperations: a majority of intercepted messagesduring the period seem to have been sent to orthrough T-142. From other intercepts, we knowthat at least another Swatow, T-379 , was nearHon Matt; two others, T-130 and T-132, werenear Hon Me Island; and T-165 had deployed, aswell. If the DRV was planning to attack the 34Araiders on 4 August, these craft: would have beenthe logical ones to use because of their substantialdeck gun armament. However, no other commu­nications activity related to any other Swatowpatrol craft was intercepted that night. So itremains uncertain what, if anything, Hanoi wasplanning to do to fend off the 34A mission of 4August.

(S"j'-/Slj Exhibit B: The Lack ofVietnameseCommand, Control, Communications, andIntelligence

(S//Sf) To our initial question as to who wasinvolved in the apparent attack of the twoAmerican destroyers, we must add a corollaryquestion: How did the North Vietnamese carryout the "attack"; that is, how were the boats con­trolled and vectored to the American ships? Ifwerecall the three elements of the command, con­trol, communications and intelligence (C3I)observed during the previous two days' activities- communications from Haiphong and PortWallut, relayed through the Swatow-class boats;the relay of tracking information on the Americanships; and the use ofthe Skin Head surface searchradar - then we have another serious problemwith the engagement of the night of 4 Augustbecause none of these elements was present dur­ing the so-called attack.

(SI/SI) During the entire day of 4 August,most of the communications intercepted fromeither DRV naval command entities in PortWallut or Haiphong either were directed to thecraft: involved in the salvage and recovery of thetwo Squadron 135 torpedo boats, or else were

Page 30 lQP SESRETo'o169MINTJRE1

lQP 6E6RElHeaMINTfflC: I Cryptologic Quarterly

relays of tracking reports of the Desoto patrol,and those latter messages were exchanged withT-142, which was involved in the ongoing recov­ery operations. The only other messages whichwere intercepted contained orders for otherSwatow-class patrol boats to move to positionsalong the coast: T-130 and T-132were ordered toHon Me Island, while T-165 was ordered to leaveHaiphong at 1448G (0748Z) and move to theentrance of an unspecified bay.121

~S//SI) During the 2 August attack, therewere elements of high-level control from thenaval commands at Port Wallut and Haiphong,both of whom sent orders and tracking reports tothe attacking boats. The Swatows, principally T­142, acted as a communications relay between thetorpedo boats and the onshore commands. Themessages were transmitted using high frequencymanual morse communications which wereintercepted throughout the day, even during thefighting. Finally, there were sporadic boat-to-boatVHF, tactical voice communications which theintercept positions aboard the Maddox's hutcould intercept, at least until the destroyer acti­vated its fire control radars, which interfered withthe navy's monitoring.

~S//SI) However, not one of these elementswas detected during the night of 4 August. Tryingto find more evidence of the purported attack,NSAhad queried the NSG detachment aboard theMaddox on 6 August to supply urgently all inter­cept that "PROVIDES PROOF OF DRVATTACKON FOUR AUGUST UPON U.S. NAVAL VES­SELS."122 Within five hours came the dishearten­ing reply from the DSU. There was no manualmorse intercept to prove the DRV attack of 4August. Furthermore, voice intercept was nil,except for signal checks between two unidentifiedstations.l 23

-(Si/SITThe tracking messages locating theDesoto patrol ships had been intercepted by theAmericans early in the day of 4 August. However,the last credible position of the American ships

was passed at 1610G (0910Z) from Haiphong toT-142. The position, 19°36'N,106°19'E, was fairlyclose to the Desoto patrol's position at the time.This was just about two hours before Herrickordered his ships to head east in reaction to thePhu Bai CritiC.I24 However, it should be pointedout that this position report was sent to the T-142,which was involved in the salvage of the two tor­pedo boats. There is no evidence that the T-142relayed it to any other boat or command.

1St/Sa- One more position report on theDesoto patrol was sent from Port Wallut to aprobable vessel at 2246G (1546Z), which wasabout an hour after the supposed engagementhad begun. This position report might seem asrelated to the action, except for two problems.First of all, the report located the American shipsthirty-five nautical miles east of Hon Matt Island,which places the destroyers some eighty nauticalmiles northwest of where they actually were at thetime! In addition, the report does not carry thetime associated with the Americans' position.(The reported location suggests, however, at leastfrom the track the Desoto patrol took that night,that this position report was about four to fivehours old.) So, this information could hardly beused by any North Vietnamese boats intending toattack the Americans. Secondly, the messageincludes an order (or advisory) to the recipient tomaintain a continuous communications watchwith an unidentified entity, as well as to "go closeto shore." 125 This latter command seems to behardly intended for boats looking to attack theAmerican ships; rather it appears suited for theboats involved in the salvage operations or theother patrol boats spread out along the DRVcoast.

~S//SI) The issue of DRV tracking of theDesoto patrol is important. For in September1964 NSA would release a report on Vietnamesecoastal radar operations during the period. In thisreport, NSA would contend that active trackingby the coastal observation posts equipped withcoastal surveillance radars would indicate hostile

lap SEeRE'FNeaMINl'/tX1 Page 31

Cryptologic Quarterly Tap5E6RETr'r'eaMINTvr9E1

intentions by Hanoi. The report pointed out thatthere was no tracking of the Craig earlier inMarch!26 This was not quite true: the DRVwasaware of the location of the destroyer, but its timeoff the Vietnamese coast was quite short so thetracking was spotty.

ES//SI) The same report also pointed out thatthe Maddox was under "constant" radar surveil­lance before it came under attack on 2 August.However, the report then ducks the issue of theobserved sporadic tracking by the NorthVietnamese on 4 August with the claim that "Theevidence is still inconclusive in light of the virtualabsence of trackings on 3-4 August before the sec­ond attack.,,127 The evidence would never befound. The final report from the DSU aboard theMaddox showed only occasional coastal trackingfrom shore stations and North Vietnamese boatson 4 August. And it had ended by mid-after­noon.f"

~Finally, the Americans detected no SkinHead emissions during the "attack" on 4 August.Keep in mind that during 3 August the DRVboatsthat shadowed the Desoto patrol used their SkinHead surface search radars, and that these emis­sions were detected by the ELINT position in theintercept hut aboard the Maddox. These signalswere also intercepted during the morning andearly afternoon of 4 August.f"

ESfjSI) While it is true that no NorthVietnamese radar emissions were detected dur­ing the 2 August attack on the Maddox, it must beremembered that this attack occurred in the day­time under nearly ideal conditions.P" Yet, theDRV boats had initial difficulty visually locatingand then following the Maddox. What we areconfronted with in the second "attack" is theproposition that the North Vietnamese boatsthemselves, which the Thrner Joy and Maddoxdetected using only their radars, could find theAmericans so far out at sea (over 100 nauticalmiles), in heavy swells (three to six feet), at night,with a low cloud cover, without using their

radars. Even if the North Vietnamese had theequipment to receive the American radar pulses,this information would have given them only acrude bearing on which to track. They could notdetermine distance, speed, or anything else withwhich to plot any sort of torpedo attack.P'Besides that, how could they even begin to trackthe American ships when the latest valid positionwasalmost five hours old!

(U) In the Sherlock Holmes story "SilverBlaze,"the great Victorian detective and his assis­tant, Dr. Watson, are confronted with the para­dox of a crime which cannot be proven to havehappened. In the story there is this exchange:

Is there any point to which you wish to draw my

attention'?

To the curious incident of the dog in the night­

time.

The dog did nothing in the night-time.

That was the curious incident, remarked

Sherlock Holmes.132

ES/lSI) And so it is with the 4 August inci­dent: there were no DRV naval communicationsor radar emissions which were normally associat­ed with a naval engagement. Just two days prior,the Americans had an opportunity to observeVietnamese naval communications during theattack on the Maddox. Among other things, theyhad seen that the Vietnamese had difficulties insetting up and maintaining control of an attack,as the incident with the conflicting orders illus­trated. And so there should have been a generousamount of intercept of any communicationswhich would have supported the claims of the twoAmerican destroyers.

€S/fSI) Yet, nothing as much as a single barkwas intercepted. As Holmes would come to con­clude that no crime was committed, so we mustconclude that, since U.S.SIGINT never intercept­ed anything associated with an attack, none everoccurred. And the contention that all possiblecommunications and emissions reflecting an

Page 32 Tap5E6RETNeaMINT/r9E1

After the 135 ho!..d a,1':"'8edy started to repcrl to yO"'.J. we shot

DIY lAVAL 5'fITY REPORI' LOSSES po CtA;MS TWO EII!JlIYAlfJSiMPT gO'I !KMll

b) (1)

\b)(3)-50 USC 403

Cb)\(3)-PoLo 86-36

""'\'::,\\\ Cryptologlc QUarterlY

!MMEDTA'TE

2242G

2/0;val1'I)~·64

sr .:l2lZ!l):;.stl ReP

4 Aug: 64

\SECRET KIMBOOBCAR,N!C!'<Bl .......Pl."." .M....

IU/:] ,

.WS; L..........U COK7THPLT. U511-414TU85 ...46'. I.. ",~SAPAC RIP PtiIL •

VietnUlBse I 61-1329...64(Morth) ---

lep SEeRE1WeeMINMl

attack might have gone unheard can be dis­missed. As Gerrel Moore, the officer-in-charge ofthe DSU on board the Maddox, observed: "I can'tbelieve that somebody wouldn't have picked upsomething." 133 Finally, a review of the DSU inter­cept log for 4 August showed no variation inVietnamese communications procedures whichcould suggest that any change or changes, such asnew operating frequencies, callsigns, or proce­dures, were implementedjust for the "attack" thatcould elude American intercept.f"

~own twe:- eneTlY plares 1:"_ the batt.~e ar ea , and (I:"." o::.her pjan 'Was

(8//81) ,Exhibit C: The '~er-Action"Report damaged. ife aac:rifH:ed two ships and all the :-est bore c.ka.y. 1h~

:,atiliilt epr r i t a very [l.Jqh a.r;d we are 9tllrt.lng out on tne hcn t anc.

(SI/S1) With there being no SIGINf evidenceof an attack, and the rest of the evidence fromvisual, radar, and sonar sources so unsupportive,we are left with attempting to explain the inter­cept of late 4 August, which was interpreted as anafter-action report. Remember, it was this inter­cept which was so critical to McNamara's con­tention that an attack had occurred - two of thefivepieces of his list of "convincing" evidence. Yet,when we look closely at the intercept, there arefour major problems with the assertion that it wasa report on the supposed engagement from just afew hours earlier on 4 August. The translation,"TlO-64," issued by NSA at 1933Z on 4 August(0233G, 5 August) is shown on this page.

~~//~I) The first difficulty with the intercept isthat it does not resemble an after-action report ofthe type which had been intercepted early on 3August by the Marine element at Phu Bai. Thatintercept, sent by T-142 to T-146 and the PortWallut HQ of Squadron 135,contained a chronol­ogy of events beginning at 0925G on 2 Augustwhen T-146 met the three boats from Squadron135 and guided them to Hon Me Island. Thereport noted that the attack against theAmericans began at 1525G,and that by 1625G,allthe boats had received the orders to break. off theattack. 135

l ar e ...ai:iD9 to] eeeca vc aaeJgnment. Men axe w~ry t:'enf..i..~iel"i":. eecause

't-hey theJr.~.lve8 sew t.he- enemy p)ar.es 81J\k.. The eneny !Il"1ip could

.:.lsl.""· have beer. damaqed. uepcr-c this info:rmatlor. back t o the un r t

.c vaJ. to IJInbi.L~.t-a th~_·:

SECItET K1MBO

(5",'51) The supposed Vietnamese communistnaval qftet-adion report

ing of the aircraft. There is no mention of any par­ticipating boats or units, except to mention thattwo were "SACRIFICED ... ANDALLTHE RESTARE OKAY." The only sense of when anythinghappened comes with the beginning phrase,"AFTERTHE 135 HAD ALREADY STARTEDTOREPORT TO YOU:' In fact, the entire reportseems incoherent, not the type one expected tosee sent by an officer on the scene, as had beenintercepted on 3 August. It rambles, mixingmorale boosting statements with seemingly repe­titious references to planes being shot down andthen seeing them "sink."

ES/fSI) In the 4 August translation, there is nochronology associated with the supposed down-

~S//SI) Secondly, there is a problem with thetranslation of a critical passage: "WE SACRI­FICED rwo SHIPS AND ALL THE REST ARE

19P SE6RE1:;:eeMINTinc: I Page 33

Cryptologic Quarterly lap SE6RE1N6aMIN1:RE:1

lap SE6REt::e6MIN'fmC1

OKAY." Unfortunately, the original, decryptedVietnamese language version of the message can­not be located in the NSAArchives. Also, a possi­ble original translation of the entire message Corpart of it), numbered "T162-64" and issued by thenavy site at San Miguel, cannot be found in theNSA Archives file of that site's 1964 translations.Without either document, we are left with theconjecture of what Vietnamese words were seenby the navy analysts and linguists at San Migueland their counterparts at NSA.

(S//Si) However, from the existing records,what we do know is that the translation finallyissued by NSA was not what was initially report­ed by San Miguel. At lSS0Z (22S0G) on 4 August,when the American destroyers were shootingaway at those radar returns, San Miguel inter­cepted a message which it identified as being sentfrom T-142 to an unidentified entity at My Due(19°S2'N, lOSoS7E). In total, the report, num­bered "R38," read:

WE SHOT ATTIVO ENEMYAIRPlANES AND

AT LEASTONE "VAS DAMAGED. WE SACRl­

FreED TIVO COMRADES BUT ALL ARE

BRAVE AND RECOGNIZE OUR OBLIGA­'nON. 136

(U) How the translation changed from "com­rades" in the San Miguel version to "boats" in theNSA version is unknown. Edwin Moise, in hisstudy of the Tonkin Gulf, suggests that aVietnamese sentence to the effect of losing twocomrades could hardly be construed to mean twoships: "HAl DONG CHI HY SINH" or "HAlDONG CHI BI HY SINH" are possibleVietnamese phrases which could be translated to"sacrificing two comrades.,,137 The Vietnameseword for boat, "TAU," had been seen in earlierintercepted messages. This would be consistent,since Hanoi's messages usually shortened theword to just the letter "T" from where the sameletter designators for Hanoi's boats comes from,such as "T-142,""T-146," etc.

Page 34

....(gIIS~ A possible argument that there was agarble in the encryption of the message whichcould have led to confusion does not hold.

CU) There is an additional point of interest:President Johnson in his memoirs noted that"The North Vietnamese skipper reported that hisunit had 'sacrificed two comrades'. " Our expertssaid that this meant either two enemy boats ortwo men in the attack group." 139 (My italics in allcases.) This is an interesting admission, for it sug­gests, and rather strongly, that even the day thatthe NSAtranslation was issued, the intercept wasconsidered, at best, ambiguous in its meaning.Why NSAopted for ''boats'' instead of "comrades"in its final translation is not clear, especially if thedifference was enough to tell the president.

(S11S!) The third problem is with the time ofthe intercept and the file time listed on the NSAtranslation. The file time, 2242G (lS42Z), is bare­ly one hour after the Turner Joy and Maddoxopened fire on the first radar returns. As we sawwith the messages from 2 August, this entry is thetime that the Vietnamese communications centerCor a radio operatorr' assigned to the messagewhen it arrived ready for transmission, which, asit turns out, in this .case took another eight min­utes to complete. /If we allow any time for the

(b) (1)(b)(3)-50 USC 403(b) (3)-18 USC 798(b) (3)-P.L. 86-36

TOil S!e~E'I.,.,eOMINT;nt1 Cryptologic Quarterly

message's drafting, coordination, and encryption(remember, this is a manual system with threecharts), then the actual time of the composition ofthe message must be pushed back close to thebeginning of the so-called engagement. Even ifwe are generous with our appreciation of the skillof the Vietnamese communications personnel inencrypting the message, we still have to concedesome time to get the message from compositionto transmission. The more time we allow for thisprocess, then the closer its origin comes up to thetime that the destroyers first opened fire. In thatcase, then, the intercept cannot be considered anafter-action report of the events currently occur­ring at sea in the Gulf of Tonkin.

(S//61) The question of the time of origin forthe information in the Vietnamese message getseven more suspect when we consider the identi­ties of the Vietnamese who may have sent andreceived it. The NSA translation carries the twocallwords "TRA" and "LAP" as unidentified.Actually, this is not true. San Miguel, in itsreports, identified the transmitting station,known by the covername "TRA," as the T-142patrol boat. The receiving station, "LAP," wasidentified as a shore station at My Due, possiblythe coastal observation post which earlier hadtracked the American ShipS.l40

(S//SI) In reality, these equations probablywere incorrect. The probable identities for thecovernames had been known for some time; it isjust that San Miguel confused them. "TRA" hadbeen associated with a DRV naval HQs inHaiphong as recently as 2 August. "LAP" hadbeen identified with T-142 on 30 July.141

However, the exact identities are not important.What is critical is that Haiphong could not haveoriginated the information in the intercept; it hadto come from some other source. Another stationhad to compose a report, encrypt, and transmitthe information to Haiphong before it could, inturn, send its message. This means that the veryfirst version of this "after-action" report probably

was composed at or before the time the twodestroyers opened fire!

ESIISI) The message file time, 2242G (l542Z)and the intercept time, 1550Z (2250G), shouldhave been highlighted in the translation. Thesetimes would have indicated that the interceptcould not have been construed as an after-actionreport. Neither critical time element was noted inthe translation. Instead, it seems that the timeNSAreleased the translation, 1933Z(or 1955Zif ithad been relayed), was the critical element. Thatthe translation was issued some two and one-halfhours after the incident was over probably wasthe reason it was interpreted by its Washingtonrecipients as a North Vietnamese after-actionreport.

-f:St151) The translation as issued is hardlyhelpful in providing a useful background toexplain its significance. The title, "DRV NAVALEN1TIY REPORTSLOSSES AND CLAIMS TWOENEMY AIRCRAFf SHOT DOWN," does notindicate any context for the translation. Thatbeing so, it would not be difficult to infer that thetranslation referred to the recently ended combataction. So, it just hung there waiting for someoneto claim it, and the Johnson administrationjumped on it. Remember, this translation arrivedin Washington midway in the afternoon of 4August just at the time that the administrationwas trying to resolve the doubts about the attackthat Captain Herrick had reported. And, as wehave seen, it was to be the answer to all of the lin­gering doubts as to the validity of the attack. NSAitself would use the translation to support thecontention that there had been a second attack aswell, quoting excerpts from it in several Gulf ofTonkin Summary reports issued from 4 to 6August. The problem with the file and intercepttimes is a critical one, and it reflects a failure bythe analysts who issued the translation to drawattention to them.

(51751) Yet, it is the fourth problem with thetranslation which is the most troublesome: that

T9P 6E6RETHe9MINf'RE1 Page 35

Cryptologic Quarterly T9P SEeRETNe9MINTiRE1

is, specifically, how it was put together. It wasmentioned above that the original intercept of thetranslation was missing from NSA files on theGulf of Tonkin. We also mentioned that the pos­sible English translation of the entire or part ofthe intercept, "T162-64," issued by San Miguel,was missing. This situation is odd since crucialearlier and original intercepts, such as the "attackmessage" and several tracking reports, wereavailable and placed in the allegedly "complete"NSA chronology of the attacks, the latter docu­ment of which we will discuss shortly. But neitherthe original intercept nor the translations fromSan Miguel are in the chronology. It would seemthat they should be there to buttress the validityof the all-important "after-action" report.14 2

However, they are not; therein lies the problem.

ESiiSI) For only four minutes (1554Z) afterSan Miguel reported the transmission about "sac­rificing two comrades," it published the followingintercept from T-142 to My Duc:

((3 GR G)} nm NEWS [BECAU[S]E] THEY

DID CONTINUOUSLY SEE WITH THEIR

OVV'N EYES ENEMY AIRCRAFf FALL INTO

THE SEA. ENEMY VESSEL PERHAPS IS

DA:.\1AGED. REPORT THIS NEWS TO THE

MOBILIZED UNIT. 143

ESiISI) Ifwe take the two intercepts from SanMiguel in the sequence in which they were moni­tored and put them together, we have construct­ed, with the addition of some transitional words,the so-called "after-action" translation, "T-lO,"issued by NSA at 1933Z on 4 August. Since themessages were transmitted by the Vietnamese inthis sequence, both spoke of aircraft, and weretransmitted shortly after one another with little orno interval, it probably was not difficult to con­flate the two as parts of the same message.

(S/iffft However, are these two interceptsreally parts of the same message? The answerturns out to be no. This is because the Englishtranslation of the second intercept exists. San

Miguel transmitted it to NSA on 8 August as partof the post-erisis review. It carried an importantitem - the Vietnamese-assigned message filenumber, "NR24," which indicates that the secondintercept was a separate message after all, andnot part of the first interceptl''"

ES//SI) So, if we look at the NSA translation,"TlO," specifically beginning at the phrase"BECAUSE THEY THEMSELVES SAW...." tothe end, what we actually are looking at is a sepa­rate North Vietnamese message. The reason fortwo messages is easy to explain. The second one isreporting what the Vietnamese observed of the 4August action from either one of their boats nearthe coast, or coastal installations.r" What theVietnamese actually saw was either the flaresdropped by the carrier Ticonderoga's aircraft toilluminate the DRV boats they were told werethere by the two destroyers, or any of the fifty orso starshells fired by the two American ships toilluminate targets. Note that the second interceptreports only that "ENEMYAIRCRAFTFALLINGINTO THE SEA." There is no mention by theVietnamese of shooting at them, as we wouldexpect if it were an report after an engagementwith the Americans as there is in the first inter­cept. In the same fashion, the flashes from thedestroyers' guns and shells exploding observedfrom over the horizon must have suggested to theVietnamese that one of the American ships hadbeen hit. San Miguel's analysts recognized thatthe second intercept dealt with that evening'sactions. San Miguel, then, reported it first at1632Z, while the first intercept about "sacrificingcomrades" was reported later at 1646Z.

ESI/SI) If we again look at the first interceptfrom San Miguel, we note that the Vietnameseclaim they shot at two planes and damaged one.This happens to be in line with their later claimsfrom the action on 2 August. Additionally, theloss of two comrades probably refers to the casu­alties suffered by T-336 from the same day'sfighting. l46 (Keep in mind that the whereaboutsand condition of T-339 were unknown to the

Page 36 Tell 8I:eR:I:TiiCOMINT,/x I

'f611 SEeRElNeeMINl:Rl1 Cryptologic Quarterly

DRV command as late as 4 August. It was stillconsidered sunk.)

(U) The congruence of the NSA and the SanMiguel reports has been noted elsewhere. InEdwin Moise's book on the Tonkin Gulf, he dis­cusses the resemblance between a "longer" mes­sage and a "shorter" one he had received fromNSA in response to a FOIA request. Since he hadreceived heavily redacted versions of "TlO" and"R38" and "R39" from San Miguel, it was difficultfor him to determine the critical fact that the tworeports from the Philippines were issued beforethe NSA translation. However, he did catch thesimilarity among them, especially the phrasesabout the downed planes.t"

ES/isn This finding that San Miguel hadissued two separate reports, which probably hadbeen conflated into a single translation by NSA,may explain the description by PresidentJohnson of the discussions with the so-calledtechnical experts at the White House the after­noon of the attack. The major point that Johnsonrelated was the explanation that the expression"sacrificing two comrades" could have meant twoenemy boats or two men. The fact that this issuewas brought up strongly suggests that the reportsfrom San Miguel probably were circulatingamong intelligence and defense officials, and thatquestions were being raised as to which versionwas correct, the boats or the comrades. But it isstill not clear from this incident what the sourcewas of the NSA version which claimed that twoboats were lost instead of two men. As we statedearlier, without the original Vietnamese text, weare left with conjecture. However, with the greatdivergence between the reports issued by SanMiguel and NSA,attention must fall primarily onthe actions of the NSA analysts. Why did theychange San Miguel's original translation?

ES//SI) This analysis of the NSAtranslation ofthe so-called after-action report may appearexcessive. Yet it is warranted because of the cru­cial role played by it in convincing the Johnson

administration of the validity of the claim that thetwo destroyers indeed had been attacked by theNorth Vietnamese. The critical analysis of thetranslation has revealed several problems withthe text itself, the context and timing of the inter­cept, that is, whether it was really related to theattack, and finally, the circumstances of the origi­nal analysis of the intercept.

(SIISI) If the results of this analysis of thetranslation were not enough to make one suspectits validity, the difficulties with the documentarysource record undermine it all the more. For thesources we do not have, that is, the missing tech­nical supplements and the translation, "T162,"leave us with a serious gap: we have only the twofield reports and single NSA English translation.The differences between the field version and theone published by NSA are too large to ignore;depending on which translation one accepts, thepossible interpretations of the incident of 4August are either that nothing happened or thatthere was an attack.

(U) Exhibit D: A Matter of'Certaintu

(U) A question remains, What were the cir­cumstances surrounding the issuance of this lasttranslation? The answer is that we do not exactlyknow the details of how it was put together.However, we do have some clues as to the envi­ronment in which the analysis reporting by NSAwas done.

ES//SI) After the 2 August attack, the analyticdivision concerned with the North Vietnameseproblem, B26, had established an informal twen­ty-four-hour watch center to handle the srGINTreporting from the Gulf of Tonkin. A relativelysmall team, perhaps fewer than ten, of analysts,linguists, and supervisory personnel, staffed thecenter. Unfortunately, there were what can becalled "environmental pressures" on the staff.Notably, a crisis atmosphere surrounding every­one and everything, which, combined withtwelve- to sixteen-hour days, probably led to seri-

19p 8E6RE'FJ)eeMINlfl~1 Page 37

Cryptologic Quarterly IUP SECRET/leO_INTi'" I

ous problems of pressure and fatigue. There wasalso the problem that the linguists available wererelativelyinexperienced, some being barely a yearor two removed from language school. Besidesjust reviewingthe field intercept, people from thiscrisis cell also briefed the Pentagon and NationalSecurity Council.r'"

(SIISI) It appears that there was little in thewayof control or interaction between this cell andsenior NSA leadership. The director, NSA,General Blake, was out of town at the time. Thevarious briefings at the Pentagon, and possiblythe White House, were handled by mid-levelmanagers and staffers operating out of the crisiscell and NSA liaison positions in the Pentagonand the White House. In fact, for the most part, itseems that senior NSA leadership stayed out ofthe proceedings, exercising little control or over­sight.149

(D) That there might have been a lot of pres­sure on the NSA people to produce "proof' isquite likely. Regarding that charged period, RayCline, the former CIA deputy director, recalledthat "Everybody was demanding the sigint (sig­nals intelligence; intercepts); they wanted itquick, they didn't want anybody to take any timeto analyze it." 150 It was certainly a crisis moment.We know from the chronology mentioned earlier,that the translation of the "after-action" reportarrived about two hours after the time that thefirst news of Captain Herrick's doubts about theaction had arrived in Washington. Also, as wehave seen, McNamara's evidence contained atleast two points from the NSAtranslation. Ofthis,there is little to doubt. However, it remains aquestion as to whether the analysts and man­agers in NSA were certain of the second attack.

(S//SItIt has been reported in other historiesthat the NSA analyst (or analysts) who actuallydecrypted and translated the intercepts weredoubtful of the second incident from the verybeginning, believing that the message referred to

the 2 August attack,ISI Furthermore, a review oforal histories suggests that in the watch centerthere was a sort of division between those whowere certain the second attack occurred, whichwas composed of mid-level management, and theanalysts who were not so sure.1S2

(S//SI) Actually, the doubters were not asskeptical about the reality of the attack as muchas they as were uncertain as how to label theintercept about the Vietnamese shooting at/downthe aircraft. Was it related to what was happeningin the GulfofTonkin? As one linguist recalled, theproblem came down to "Was this, or was thisnot?" The deciding element for the analysts wasthe fact that the intercept time (1550Zor 1559Z)of the "after action" intercept coincided with the

'time frame of the attack on the two destroyers: ananalytic "coin toss" was made, and the translationwent out which wasinterpreted as supporting thevalidity of the second attack153 There was noexplicit connection between the intercept andevents: it was inferred from the coincidence ofthe time of the intercept and the time of the ongo­ing "attack" Also implicit in this decision was alack of confidence concerning the validity of theinformation; it could not stand by itself as the evi­dence, at least in the minds of the analysts.

(D) On such small things as a mental "cointoss," then, does history often turn.

(SI/SI) As to the nature of the translation,according to the same linguist, reportedly therewere no enforced "word changes" in this report(or any others which were issued), though argu­ments over translation "styles" did occur. Thesearguments were over the rendering of the transla­tions from the Vietnamese original "into suitableEnglish." 154

C'fSi/SI) This analysis by coin flip left thedoor open for follow-up reports which moreopenly supported the notion of an attack Barelysix hours after it issued the "after-action" transla­tion, NSAreleased its first summary report of the

Page 38 Tell S!e~ElnCUMINtnx1

TOP S!CIt!TltCOMiN IIIX' Cryptologic Quarterly

action. This summary contained quotes from theearlier after-action translation. These quoteswere placed in summary in such a way as to sub­stantiate collateral radar, sonar, and visual infor­mation from the Desoto patrol. On 6 August twomore summaries were released by NSA whichcarried more SIGINT which the Agency assertedsupported the second attack scenario. Publicly, atleast, and probably from the very beginning, NSAsupported the Johnson administration's claim fora second attack.155 These reports are important inunderstanding the post-attack position taken byNSA.

-tB/lffft As for the doubts about the secondattack among the analysts at NSA, it appears thatnone of them were ever publicized during thebriefings with officials at the Defense Depart­ment. Or, if they were mentioned, they weredownplayed. In fact, it seems that the NSA posi­tion was a fairly straightforward one: that thesecond attack occurred.P" So firm was NSA'sposition, that one previous NSA historian hassuggested that this allowed President Johnson toshift the blame for the final decision from himselfto the "experts" who had assured him of thestrength of the evidence from the SIGINT.157

(D) Officially, everyone else in Washingtonsupported the notion that there had been anattack. Later statements by various intelligenceand Defense Department officials suggest thatthere was a large group who simply did notbelieve that the attack had happened or that theevidence even pointed to an attack. Many high­ranking officials from CIA, the Department ofState, and the Pentagon could not see the evi­dence assembled by McNamara as supporting aVietnamese attack. Some of them were skeptical(or claim to have been so) from almost the begin­ning of the incident. This group of doubtersincluded the then U.S. Army's deputy chief ofstaff for military operations, General BrucePalmer Jr., Ray Cline, the CIA's deputy directorfor intelligence, the heads of the Department ofState's Intelligence and Far Eastern Divisions, as

well as a host of staffers on the National SecurityCouncil and in the Defense Department, who, inyears to come, would become notable: DanielEllsberg, Alvid Friedman, and Alexander Haig.

(U) Yet, despite doubts, people in the intelli­gence and defense communities kept theirsilence. As much as anything else, it was anawareness that President Johnson would brookno uncertainty that could undermine his position.Faced with this attitude, Ray Cline was quoted assaying: "... we knew it was bum dope that wewere getting from the Seventh F1eet,but we weretold only to give the facts with no elaboration onthe nature of the evidence. Everyone knew howvolatile LBJ was. He did not like to deal withuncertainties." 158

(S//SI) And there were plenty of people inNSA and the cryptologic community who doubt­ed that the SIGINT was convincing evidence of anattack. Notable among these were the chief of BGroup, who seems to have been skeptical fromthe morning of 5 August, and the NSA PacificRepresentative (NSAPAC),who sent a message toDIRNSA listing his doubts after reviewing aCINCPACstudy of the affair. 159

('FSI/SI) With all of the doubters about theattack, whether they were uncertain from thebeginning, or saw the problems with the "evi­dence" later on, it is surprising that what emergedfrom various intelligence and Defense Depart­ment studies of the 4 August event were evenmore confirmations that the attack had occurred.Within weeks of the event, teams from the navycommands in the Pacific region, CINCPAC andSeventh Fleet, conducted reviews which verifiedthe attack. A Defense Department team arrived inmid-August and conducted interviews of thepilots and the crews of the destroyers. They foundstrong evidence for the attack as well.160 TheJoint Reconnaissance Center issued a chronologyof events, while ASA Pacific Headquarters con­ducted a critique of the reporting by Phu Bai dur­ing SIGINT Readiness Bravo Lantern, the

19P SE6REl::e9MINlJRE1 Page 39

Cryptologic Quarterly TOP S!CRET11eOMINTI'IXt

enhanced SIGINT coverage ordered during theGulf of Tonkin crisis.161Both documents support­ed the idea of a second attack.

(S//SI) Exhibit E: And Some More SilentDogs

~S//SI) Various elements of the NavalSecurity Group, which oversaw and provided themanning for the Desoto missions, issued reportson the incidents in the Gulf of Tonkin which werestrangely reticent about the evidence of the attackon the night of the 4th. For example, in the reportissued by the commanding officer of NSGdetach­ment aboard the Maddox, two and one-halfpagesare devoted to SIGINTreflections of the 2 Augustattack. The follow-up air strikes of S August war­rant another half page. Yet the statement summa­rizing the SIGINTactivity of 4 August is renderedin just in one sentence:

H. On 4 August infonnationreceived from USN

4141'and USM626.J [Phu Bai] indicated a pos­

sible attack on the Desoto ships by the DRVnaval vessels!62

-{871§f) A report from the director, NavalSecurity Group Pacific, of 24 August was similar.Twelveparagraphs of the message are devoted inrecounting the SIGINT detail of the 2 Augustattacks. The recounting of the "attack" of 4August was done in a short entry of two para­graphs, the first of which contained the informa­tion that T-142 was "again shadowing" the U.S.ships. It also refers to "moderately heavy track­ing" by two DRV tracking sites at Thanh Hoa(20

000'N,

lOSo30'E) and Hon En (lSolS'N,106°09'E)." The site at Than Hoa would havetracked the two ships early on 4 August, but theattack was several hours later. When Hon Entracked the ships is unknown. The second para­graph mentions only the two reports from PhuBai, stating that they indicated "a possibleattack." 163

~S//Sf) Further evidence, and perhaps one ofthe strongest pieces available indicating that noattack had happened, came from the NorthVietnamese themselves. On 6 August, an uniden­tified DRVnaval entity, possibly the naval HQ atPort Wallut, transmitted to an unidentified sta­tion a recap of the previous combat with theAmericans. It summarized the events of 2 Augustand mentioned their boats fighting the "Americanwarship." It also recounted that their naval andair defense forces had shot down some Americanwarplanes on S August and had captured oneAmerican pilot alive. Yet, there is no mention ofanything occurring on the night of 4 August inthis recap.164 The absence of any reference to 4August cannot be attributed to North Vietnameseembarrassment over the results of the "action";they lost heavilyon both 2 and SAugust. The onlyconclusion that this intercept points to is thatthere was no attack on the night of 4 August.

(Sf/Sf) Oddly, this last intercept has neverbeen used in any evaluation of the Gulf of Tonkinincidents. Understandably, those evaluationshave tended to rely on the evidence from the timeperiod of the incidents themselves. Surely, aNorth Vietnamese accounting of the operationsfor the previous three days would have been con­sidered as part of the body of evidence concerningthe attack. Yet it was not used, although NSAsummaries issued on the same day were. Wasthat because the intercept says nothing about anattack on 4 August?

(Sf/Sf) Maintaining the Line: TheNSA Summary Reports and the"Del Lang Chronology"

~SffSI) As the field sites submitted theirreports on what intercept they did or did nothave, as in the case of the NSG element aboardthe Maddox, and the analysts had the luxury oftime to review all of the SIGINT,the various eval­uations they produced continued to reflect theofficial position that the second attack hadoccurred. The most important early response

Page 40 IUPSECRETt,COMINTiIX1

TOil S!CIt!TiJlCOMINTtIX I Cryptologic Quarterly

from Fort Meade was a series of summary reportsissued between Sand 7 August. It is these reportswhich make up first official NSA judgment onwhat happened. Because of this, they deserve aclose look, since they establish the tone and formfor the later chronology, which became, in a way,the final NSA statement on what had happened.

(T8/fSI) NSAissued five summary and situa­tion reports after the incident, beginning early onS August. Of the five, numbers "R01" through"ROS," the pertinent ones are the first three, espe­cially the first and third. These three reportsexplicitly state that the 4 August attack occurred.Report "ROl" notes that the reports from thedestroyer that it had sunk two torpedo boats werelater "confirmed by a DRV message which stated'that we had sacrificed two ships and the rest areokay'." 165 Where this idea that two boats weresunk came from is hard to say. NSA received allmessages from the Desoto patrol via the JCS. Allthrough the afternoon of 4 August, the destroyersreported at first that three boats had been sunk,then later changed it to one sunk and one, possi­bly two, damaged.f" The second post-incidentreport, known as "Gulf of Tonkin SIGINTSituation Report No. I," included the statement"following the 4 August attack. "

ffS1/SI) It was the third report that was themost open in supporting the idea of the secondattack. It was stated in the lead sentence of thereport that "This report is a summary of thoseDRV naval communications during the period I­SAugust which demonstrate irrefutably that DRVnaval boats did, in fact, engage in preplannedcombat against U.S. destroyers patrolling ininternational waters." 167

(TSiISI) However, the confident tone of thethird report is belied by its thin layer of evidence.And this problem was noticed by some of itsrecipients. Late on the afternoon of 6 August, aDIA representative queried NSA if additionalSIGINT was available from the 4 August incident.He reported that Secretary McNamara was not

satisfied with the contents of this third summaryreport, "that it was insufficient for his purposes."In reviewing the SIGINT from the incident, it wasdiscovered that there was a large gap with nointercept - specifically, the time leading up to theactual attack. Based on this discovery, urgentmessages were sent to the field sites requesting allintercept.P" And, as we have seen, the field siteshad nothing else to add.

ffSi/sn There are problems with the way thisseries of reports portrays the information inthem. For example, the first report mentions thesalvage operations of the two damaged DRVtor­pedo boats which had been discussed earlier.However, unlike what we discovered, the summa­ry does not go on to report that these operationscontinued into the time of the attack as reportedby the marines at Phu BaL The authors of thethird report tried to address this with the specu­lation that the attacking boats might have comefrom Quang Khe or some other base in the DRVSouthern Command.P" But this has already beenshown to be wrong since the distance traveled forthe boats to have attacked from the east could nothave been accomplished because of the limita­tions of the boats' speed.

(T8/f8I) Perhaps the most serious problem,though, is the lack of any citation of sourcereports which made up the summaries. This is acritical point, since the information referred to inthe summaries is coming from already published,serialized NSA and field site reports and transla­tions. The very lack of notes is odd since this typeof summary reporting required that source notesbe included. It seems that if the Agency wasattempting to build a case demonstrating that anattack had occurred, then the source reports andtranslations which substantiated the positionwould have been included. However, this was notthe case. In fact, there were cases in which infor­mation used in the summaries as evidence, was,in fact, not related at all, or impossible to verify.

1'6" SECRE1'IICOMINTiIX I Page 41

Cryptologic Quarterly TOil S!el'!T/ICOMINT/IX I

eTSt/SI) For example, the first summary,"ROl," issued early on 5 August, contained thissection which strongly suggests that the Desotopatrol was surveyed by DRV aircraft. The entryread:

During 3 August, DRV NavalCommunications reflected the trackingand shadowing of the two destroyersthroughout the day; this activity wasreported by both destroyers. They werealso apparently shadowed by two presum­ably DRV aircraft. A DRV merchant shipadvised its shipping office in Haiphongthat 'two bombers' would 'fly' in the direc­tion of the ship and investigate. No furtheridentification of the aircraft ..las provid-

d. 170e .

ffSiiSI) This entry was lifted from a SanMiguel report on DRV merchant shipping. In it, asingle North Vietnamese merchant ship, theThong Nhat, reported that two single-propelleraircraft (chong chongs), and not bombers, wereflying to investigate the ship, presumably a refer­ence to itself.'?' Hanoi's aircraft inventory con­tained two single-prop planes - the AN-2 (Colt),a small transport biplane and the YAK-18 (Max)trainer - both of which were unsuitable to mar­itime patrols. Since the report never specified thenationality of the aircraft, it is likely that theywere American A-IH single propeller fighterbombers from the Ticonderoga.

ESIISI) At the time of the intercept, 1018Z on3 August, the Desoto patrol was some sixty milesto the south of the Thong Nhat; it seems reason­able that the Desoto combat air patrol would havegone to investigate the North Vietnamesefreighter.'?" A few hours after the Thong Nhatreported the aircraft, the Haiphong shippingoffice transmitted an urgent message to threeDRV merchant ships to "take precautions againstenemy airplanes and ships." 173

ff'Si/51j In addition, the third report, "R03,"refers to intercept at 1054Z on 4 August that theDRV was trying to keep "activities under cover"when it was claimed that it had intercepted amessage with the sentence "YOU CANNOT PUB­UCIZE THE SITUATION OF THE BOATS OFFLOTILIA 135 TO THE BACH DANG." 174 Whois sending this message, and to whom, is notmentioned in the summary. To date, the source ofthis sentence has not been found; its context, thecorrectness of the translation, or even its correla­tion to the attack, cannot be determined.

(TSIISO Report "R03" also carried anothercurious entry supporting the idea of an attack.This read "KROAI HAD MET THE ENEMY."Over the ensuing years this entry bothered peopleresearching the incident. No one could find theoriginal intercept, and no one could seem toexplain it.175 No wonder. The sentence was arewrite of a San Miguel intercept. The originalintercept was of a message from Haiphong to T­146, which originally read: "WHEN ((YOU))MEET THE ENEMY T333 MUST MOBILIZE."Since the local time of the intercept is 0211G(20HZ) on 5 August, the reference to meeting theenemy has nothing to do with the prior evening'sincident. In fact, the tense of the original transla­tion suggests that this was a message anticipatinga possible future clash with the Americans, and itwas expected that torpedo boat T-333 had to beready to defend itself.176 The name "KHOAI" wasseen in other intercepts over the prior two days,including the infamous "military operations" oneof early 4 August. In reality, "KROAI" probablywas Le Duy Khoai, the commander of Squadron135. That he, the commanding officer, accompa­nied Section 3 in its attack against the Maddox on2 August, and stayed on to supervise the recoveryoperations of his two damaged boats, was stan­dard procedure for DRV naval officers.'?"

(TS/ISI) The main NSA effort at producing arecord of the events of 2-5 August 1964 centeredon a joint postmortem with the DefenseIntelligence Agency, begun in late August 1964

Page 42 lep 9EeRElrreeMINl,'RE1

TOP S!ellt!TltCOMINTmC 1 Cryptologic Quarterly

and released on 14 October 1964. What was pro­duced was a chronology of events which support­ed the contention that there had been a secondattack. The NSAversion of the chronology stayedwithin the cryptologic community with a verynarrow distribution totalling ten recipients.Later, after the second Gulf of Tonkin "incident"of 19 September 1964, a second volume wasadded to cover that event. l 78

-ffSJ/ST)This chronology, specifically the vol­ume titled "Chronology of Events of 2-5 August inthe Gulf of Tonkin," was bound in a black binderand came to be popularly referred to as the "DelLang Chronology," named after the B Group staffofficer, Lieutenant Colonel Delmar Lang, USAF,who compiled it. Colonel Lang was a veterancryptologic staffofficer with a great deal ofliaisonexperience with various SIGINT missions in Asia,starting with work during the Korean War. Hewould be instrumental later in implementing var­ious SIGINT support efforts for Rolling Thunderand Linebacker air campaigns. The chronology heproduced solidified the official position that theattack had occurred. In the introduction, Langclaimed it to be as complete as far as the SIGINTinvolvement necessitated. The SIGINT materialincluded product reports, translations, andselected messages between NSAand various fieldsites and liaison offices. The chronology alsomade heavy use of non-SIGINT sources, in thiscase messages from the Desoto patrol, CINCPAC,and the JCS. The chronology was arranged withan introductory time line which highlightedevents between 2 to 5 August, followed by thedocuments which were notated with "tabs" num­bered sequentially and cross-referenced in theintroductlon.F"

~S//SI) Like the summaries discussed above,there are serious problems in the chronology withboth the amount and subject matter of the SIG­INT evidence and the way it is presented. Forexample, in reference to the 4 August incident,the chronology makes use of only six SIGINTproducts (not counting the summaries which

were a review of published product) as evidence.Now, we have been referring to a large number ofthese products about the 4 August "attack"throughout this article. All told, between 3 and 6August, fifty-nine SIGINT products can be identi­fied as being relevant to that purported attack,that is, containing information related in someway to it. These include serialized reports, trans­lations, critics, follow-ups to the Critics, and tech­nical supplements. The fifty-nine productsinclude status reports on the North Vietnameseboats, DRV tracking of the Desoto patrol fromcoastal observation posts and boats, salvage oper­ations of the damaged boats originally thought tobe involved, DRV boat movement and locationreports, and intelligence reports. So the six prod­ucts used in the chronology constitute a bit morethan 10 percent of the total available.

(Sf/Sf) Now, the introduction to the chronol­ogy refers to using "representative samples ofDIRNSA's COMINT reporting of the activitiesdirectly and indirectly related to the situation ofthe activities in the Gulf of Tonkin." 180 Howmerely six out of fifty-nine is "representative" isdifficult to understand. Furthermore, these sixreports are the only ones which can be construedto demonstrate an aggressive intent on the part ofHanoi's navy. They include a 3 August report of aconcentration of DRV vessels near Hon MeIsland, the three Critics and follow-ups concern­ing the "attack" being planned for the night of 4August, the translation of the so-called "after­action" report, and an early 5 August messagereporting DRV plans for combat operations onthe night of 5 August, which turned out to berelated to the ongoing salvage operations.f"

EBI/SI) None of the other fifty-three productswere included in the chronology. These includeall of the ones that have been cited earlier in thisarticle, and which demonstrated that no attackwas planned, or proved that the NorthVietnamese did not know the location of theAmerican destroyers, or indicated that the sal­vage operations were the primary activity of

T6PSEeRET(ie6MINT,.,iE1 Page 43

Cryptologic Quarterly Tep SEeRE'f/ICOIlIINTnxi

Hanoi's navy, or the outright statements in someintercept for the DRVboats to stay away from theAmericans. These products were available at thetime of the composition of the chronology. Yetwhy they were not included is unknown.Obviously, their absence leaves the reader withthe impression of Hanoi's overt aggressionagainst the American ships.

(Sf/Sf) The way the material is presented isalso curious. Almost all of the SIGINT productincluded for both 2 and 4 August has attached thereproduction of the original intercept of the DRVnavy's messages: that is, the cipher and itsdecrypted Vietnamese text. This allows the read­er to see the unfolding of the SIGINT process,from intercept to report.

(S//81) However, there is one glaring excep­tion to this: the 4 August translation of the so­called "after-action" report used by SecretaryMcNamara and President Johnson as primaryevidence of the attack. In fact, only the translationis included, and it is there only as "a sample."Considering the importance attached to it by theadministration, as we saw earlier, this is a veryodd way of presenting this piece of critical evi­dence. It would seem that the NSA originators ofthe chronology would have added the completeVietnamese cipher and text to bolster the case foran attack. Yet the translation stands alone. Sincewe know that the intercept used to produce thetranslation currently is missing, might we ask ifthey were already "missing" shortly after the inci­dent itself?

(S//Sf) Finally, the chronology does notaddress the problem of the total lack of NorthVietnamese C3I related to the supposed 4 Augustattack. Not surprisingly, there are samples of theC3I from the 2 August attack. Yet, aside from theso-called "attack" message and the purported"after-action" report, there is nothing. We havecommented on this before. The argument that thematerial may not have been available in earlyAugust might have had some slight relevance.

The chronology might have been the vehicle foraddressing this shortcoming. However, fully twomonths later, there is still nothing included of theenemy's C3I - the huge gap is not addressed,much less explained, by NSA

(S/fSI) Over the years, the chronology cameto be the source book for responses to congres­sional inquiries into the Gulf of Tonkin incidents.That is, the other 90 percent of related SIGINTproduct was not offered to any congressionalinvestigating committees. Instead, the chronolo­gybecame, by virtue of its position as an "official"report, the only source for background on theGulf of Tonkin incidents.

('fSffSo-The first investigation came in early1968 when the Senate Foreign RelationsCommittee, under the chairmanship of SenatorWilliam Fulbright, who had steered the Gulf ofTonkin Resolution through the Senate, openedhearings on the incident. Secretary of DefenseRobert McNamara was called in to testify. Priorto his testimony, he requested that the pertinentCOMINT on the incidents be given to him. TheNSA and the Defense Intelligence Agency werereluctant to have the SIGINT used; both agencieswere fearful that the exposure would compromisethe then current capabilities against the NorthVietnamese.P" Ultimately, Secretary McNamarawas given the contents of the chronology, as wasthe Senate committee as well. The resulting hear­ings did nothing to clear up the confusion.McNamara argued for the attack, citing the vari­ous SIGINT reports, though he seemed to mix upwhat was in them, and left observers sometimesconfused. l83 Many senators, looking at the samechronology, remained skeptical.

-(S/fSIr In August 1975, the Senate SelectCommittee on Intelligence, under the chairman­ship of Senator Frank Church of Idaho,approached NSA about the Gulf of Tonkin inci­dent. The committee's interest, though, may nothave been in establishing the validity of the inci­dent; their attention was focused on information

Page 44 IUPSECR!TiiCOMINT,.,K1

IUP SECRET/ICOMINTttX I Cryptologic Quarterly

concerning the covert OPLAN 34A and Desotomissions, and what exactly was being done byboth operations. NSA's response to the ChurchCommittee's request was similar to that ofFulbright's: limited release of materials from thechronology. In fact, NSA was concerned that theChurch Committee get exactly what Fulbrighthad recetved.P" Again, the chronology of theevents of 2 to 4 August was the source used formaterial to be released. Interestingly, a major fig­ure in these latter deliberations on what to releaseto the Senate was the then-retired, former deputydirector of NSA, Dr. Louis Tordella. He hadadvised the NSA staff as to what to release andhold back. Curiously, one of the few things heldback was a similar chronology of the events ofmid-September 1964, in which another Desotopatrol claimed it had been attacked.

(S//SI) Gulf of Tonkin Redox:The 18 September "Attack"

(57751) In an interesting and ironic repeat ofthe Gulf of Tonkin incidents, on 18 September1964 another Desoto patrol would undergo thesame experience as the Maddox and Turner Joy.In this incident, two destroyers, the USSMorton(DD 948) and the USS Richard S. Edwards (DD950), were assigned a Desoto mission for mid­September. The ships began their operations on16 September. The North Vietnamese knewalmost from the start that the two vessels wouldbe in the area and were tracking it. The DRVnaval authorities also ordered their ships andposts to be on alert and to be aware for "provoca­tions" by the Americans.i'"

(S/lSI) North Vietnamese tracking of the twodestroyers held through the 17thand into the 18thof September. At 1738G (1038Z) on 18September, a message was passed from anunidentified DRVnaval authority that ordered allships to take precautions against possible SouthVietnamese maritime commandos who mighttake advantage of the presence of the Americanships in the area to launch an attack. The North

Vietnamese ships were also ordered to "avoidprovocation" and to disperse and camouflage.P"

(0) At about 1729G(1029Z), the two destroy­ers acquired radar contacts following them. Bothships began to maneuver and increase speed toclear the apparent vessels trailing them. Aboutforty-five minutes later, the Morton fired a warn­ing shot at one of the contacts. By this time, theAmericans counted on their radar scopes fiveships trailing them. However, the warning shotdid not deter the threatening vessels. About tenminutes later, both ships opened fire. For aboutthe next hour, both American ships engaged thecontacts as they appeared on their radar screens.Oddly, at no time did the contacts return any fire,nor did they launch any torpedoes. Even morecurious, only one of the enemy ships ever closedfaster than twenty-three knots. In fact, the con­tacts pretty much matched the speeds of thedestroyers. Meanwhile, the Morton and Edwardsfired almost 300 rounds at the contacts andclaimed to sink as many as five of the vessels(there were now more than the original five con­tacts) which had been menacing them.

-tSTThe JCS ordered a search, to begin thenext morning, of the area for debris to confirmthe attacks. At the same time, plans were putunder way for another retaliatory strike againstthe DRV. More air force and navy aircraft weredispatched to the region to reinforce the proposedstrikes. 187 Yet, nothing happened. The area wassearched, but no debris nor even an oil slick wasfound. The JCS continued to request data on theattacks from all the intelligence and combat com­mands. Yet even by the 19th there still was noconcrete evidence of an attack.188

(TaffS!) Available SIGINT information indi­cated that the North Vietnamese were well awareof the presence of the two destroyers, butremained in a defensive posture. The DRV waslooking to react to a possible maritime raid by theSouth Vietnamese, but there were no reflections

lOP SE6RETNeOMIN II/X' Page 45

Cryptologic Quarterly TOil S!CIl!'f/teOMINT/IX I

of any hostile intent against the two destroyers.P?In fact, on 20 September NSA corrected a Criticby San Miguel which claimed that the DRV wasplanning to attack the Desoto patrol that evening.Fort Meade pointed out that the interceptedinformation could apply equally to an attack onSouth Vietnamese "raiders." 190

CD) By the end of 20 September, the issueremained unresolved. The Edwards and Mortonwere ordered to return to the nearby carrier taskgroup, and the Desoto missions were indefinitelysuspended, and, in fact, except for an occasionaltraining cruise, they were never carried outagain.i'"

CU) In certain histories of the Indochina War,it has been fashionable to maintain that, in thefinal accounting, whether or not there was anattack on U.S. Navy destroyers on 4 August in theGulf of Tonkin may not have mattered at all. TheJohnson administration had been looking for away to expand America's role in South Vietnam.In June 1964, two months before the Augustattacks, a resolution had been prepared byWilliam Bundy, assistant secretary of state for FarEastern Affairs, which would give the presidentthe right to commit U.S. forces to the defense ofany nation in Southeast Asia threatened by com­munist aggression or subversion. Furthermore,the draft resolution gave Johnson both the dis­cretion to determine the extent of the threat and,by virtue of this evaluation, the leeway to definewhat forces and actions were necessary to count­er it. At first, the resolution was planned to be putbefore the Senate as soon as possible. ButPresident Johnson demurred, fearing that itwould ruin the image of moderation he had beencultivating for the presidential election inNovember. The draft resolution was quietlyshelved until another opportunity could comealong.192

Page 46

* * * *

(U) The Johnson administration used the 4August incident to ride the resurrected resolu­tion, now popularly referred to as the Tonkin GulfResolution, through the Senate, with only twodissenting votes. It was portrayed as a moderat­ing measure "calculated to prevent the spread ofwar." 193 However, President Johnson now hadthe legal cover to use whatever military force hewanted. When he heard of its passage by bothhouses, he laughed and told an aide that the res­olution "was like Grandma's nightshirt. It coverseverything." 194

CU) Yet, even with the resolution in his pock­et, President Johnson ignored the similarSeptember Gulf of Tonkin "incident," and did notorder a retaliation against North Vietnam. Itwould take another communist attack onAmerican forces, the strike at the American baseat Pleiku in February 1965, to make Washingtonescalate the war a further step, this time initiatingthe Rolling Thunder air campaign.i'"

CU) The problem, of course, was the nature ofthe provocation which made possible the passageof the resolution. If the resolution had been tiedto the naval action of the afternoon of 2 August, orto the communist bombing of the officers' quar­ters in Saigon on Christmas Eve 1964, or even tothe VC sapper attack on the air base at Bien Hoaon 1 November 1964, then the administration atleast would have had an actual incident uponwhich to base support for it. Then any reconsid­eration of the resolution would have centeredsolely on it and not the incident on which it wasbased.

(D) Unfortunately, the administration choseto hang the rationale for expanding its war-mak­ing franchise in Southeast Asia on an incidentwhich could not stand up to any kind of objectiveexamination of the full documentation. So, aseventually happened in 1968, when the Gulf ofTonkin Resolution came to be reviewed, the inci­dent that it was based on also came under scruti­ny. When the events of 4 August were revealed to

TOil S!Cft!'fNeOMIN'f/1X1

iUP SEeft:~;f;teeMIU;'VX1 Cryptologic Quarterly

days later: "Hell, those damn, stupid sailors werejust shooting at flying fish."

CU) After recounting all of the events and ana­lyzing the sources, the remaining task for the his­torian is to attempt to characterize them, to offera summation or a judgment that will place thenarrative into a coherent framework. But beforethat can be done, it is necessary to review whathas been presented. In this way we can consideragain what we have learned about the events inearly August.

O~I/~I) We have seen that the Gulf of Tonkinincidents occurred as a result of the congruenceof the Desoto patrols and the maritime comman­do raids along the North Vietnamese coast car­ried out under OPLAN34A. In the period leadingup to the Maddox mission, the DRV had beenreacting with increasing force to the OPLAN34Aattacks. Monitoring Hanoi's more aggressiveresponse to the raids, NSAhad warned the majorcommands in the region of the potential dangerfor the Desoto patrols, but the warning did notregister. The decision makers in Washington

****

CD) In this article we have done somethingquite apart from most Agency histories: Usingvirtually hitherto untouched material from a vari­ety of sources, we have told a radically differentversion of an important event in cryptologic his­tory which, in turn, had a critical effect on thecourse of American history. In doing so, a greatdeal of unfamiliar ground, in terms of sourcematerial, had to be covered, and the new infor­mation could not be presented in a typical, his­torical narrative format. Instead, we had topainstakingly analyze a series of documentswhich were quite important if we were to graspwhat happened on 4 August 1964. Admittedly,this was a difficult task, but it was necessary if wewere to be as comprehensive as possible in ouranalysis of what happened.

have been based on very thin evidence, it concur­rently was demonstrated that the Johnsonadministration had indulged in a very selectiveuse of information. If the administration had notlied exactly, it had not been exactly honest withthe public, or, for that matter, even honest withinits own deliberations. The question no longer wasabout the appropriateness of the resolution, butthe basic honesty of the administration. It wouldcast a pall on an already distrusted Johnson pres­idency. As Senator Barry Goldwater, who had runagainst Johnson in the 1964 presidential election,bitterly noted years later in 1972,"I had no reasonto believe that Mr. Johnson's account of the grav­ity existing in the Gulf of Tonkin was not legiti­mate." 196

CD) As for the Tonkin Gulf incident itself,President Johnson summed it up best just a few

(V) A Doug\;.\5 A-4- Skyh<lwk qt1:ack plane

qtapults f1-om a carrier in the GulFoFTonkin

eluting 'lt1:'lck operations in August 1964.

TOil 8I!e~~T,1'eOMiNTlIX I Page 47

Cryptologic Quarterly TOP 8E6RrFHeeMINTtlX I

believed that Hanoi would not see the two mis­sions as related.

ES//SI) On 2 August, the SIGINT system per­formed admirably when it provided sufficientwarning to the Maddox to allow it to defend itselfagainst the attack by the three DRV torpedoboats. At the same time, the American cryptolo­gists were able to observe the DRVnaval C3I sys­tem in action. From this they should have devel­oped a profile from which further timely warn­ings could be deduced. During 3 August, bothsides maintained a distant watchfulness, thoughtensions remained high - high enough, perhaps,for the field site at Phu Bai to confuse salvageoperations around the island of Hon Me for a pre­attack concentration of forces.

(S//SI) The 4 August incident began in theearly afternoon due to an analytic error by theMarine contingent at Phu Bai. This mistake set inthe minds of the crew of the two destroyers theidea that they shortly would be attacked. This wasan error of interpretation by the Marine unit atPhu Bai, a mistake, as we have seen, which wasnot committed by the navy site at San Miguel.Nor was the Critic transmitted by Phu Bai ques­tioned or corrected at NSA.This may have been inline with an unspoken policy of not second-guess­ing field sites since they were "closer" to theaction. However, under Critic procedures, PhuBai had to supply the technical information uponwhich it based its alert. When the discrepancybetween what the intercept actually said and whatthe Marine detachment reported became known,NSA should have cautioned the recipients of theCritic. However, this did not happen.

-(8f/SttThree hours later, at almost the samemoment that the American destroyers openedfire on the approaching radar return, Phu Baiissued another report which stated that the spe­cific boats, which had been identified as beingreadied for an attack, in reality, were to be towedto Haiphong for repairs. This salvage operationwould be the subject of several more reports dur-

ing the rest of the evening of 4 August. Since noother boats were referenced in the original"attack" message, the cryptologists at NSAfoundthemselves without any SIGINT evidence sup­porting the reports of an ambush. The Phu Baireports had effectively cancelled out the originalCritic. However, the response by NSA was tocounter the SIGINT evidence with an unfoundedspeculation that the boats the Desoto patrolthought were attacking it came from Quang Khe.And it has been demonstrated how impossiblethis scenario was.

ES//SI) It also has been established that noneof the C3I associated with DRVnaval attack of 2August was present on 4 August. Aside from spo­radic North Vietnamese coastal tracking, whichended hours before the two destroyers turnedeast, there was no intercept to suggest the NorthVietnamese had anything more than the usualinterest in the two ships. Nor, for that matter, wasthere any intercept of any DRV naval communi­cations which suggested in any manner that anattack was planned, much less that one actuallywas occurring. In fact, Hanoi seemed more inter­ested in warning its boats of the patrol's presence,viewing the Americans as a threat to its navy. Forthe cryptologic community, this lack of any attackC3I is one of the most critical points of the Gulf ofTonkin crisis. Yet, NSAnever addressed the issuein any reports or activity summaries it publishedconcerning the 4 August incident.

ESffSi) Instead, NSA would issue summarieswith scattered tidbits of contentious and unrefer­enced intercept ("Khoai had met the enemy" andthe purported aerial tracking) to support thenotion that an attack had been planned and thatit had been carried out. The extensive amount ofSIGINT evidence that contradicted both the ini­tial attack order and the notion that any NorthVietnamese boats were involved in any "militaryoperations," other than salvage of the two dam­aged torpedo boats, was either misrepresented orexcluded from all NSA produced post-incidentsummaries, reports, or chronologies. NSA's fail-

Page48 lOPSEGIt!'fiteeMINf....K1

TO~ SI!CIltI!T/ICOMINT/IX I Cryptologic Quarterly

ure to deal with both issues, the lack of any attackC3I and the contradictory SIGINT, especiallyduring the critical hours leading up to the retalia­tory air strikes of 5 August, remains its most glar­ing shortcoming in this incident.

~81ISI) We have seen as well the many tech­nical problems with the supposed "after-action"translation. This product, upon which the admin­istration based so much of its case, appears tohave been the result of an analytic error of com­bining two separate messages, each dealing withseparate incidents, into a single translation.There were more problems, such as the actualtranslation of the term "comrades" and how itwas rendered into "boats" by NSA Here, the ana­lytic problems mix with those discovered aboutthe available records: the original decryptedVietnamese text, and an important translationfrom San Miguel cannot be located. Consideringthe importance of this translation to the adminis­tration's case, the fact that the original text cannotbe found (and was not used as early as October1964) is unusual. That these original texts andtranslation are the only missing papers in the SanMiguel reports allows for suspicion to shade anyfurther discourses.

EStlSI) What we are confronted with is thesame conundrum that confronted the NSA ana­lysts at the time. We have discussed earlier that,for the most part, the NSApersonnel in the crisiscenter who reported the second Gulf of Tonkinincident believed that it had occurred. The prob­lem for them was the SIGINT evidence. The evi­dence that supported the contention that anattack had occurred was scarce and nowhere asstrong as would have been wanted. The over­whelming body of reports, if used, would havetold the story that no attack had happened. So aconscious effort ensued to demonstrate that theattack occurred.

(8//81) The exact "how" and "why" for thiseffort to provide only the SIGINT that supportedthe claim of an attack remain unknown. There are

no "smoking gun" memoranda or notes buried inthe files that outline any plan or state a justifica­tion. Instead, the paper record speaks for itself onwhat happened: what few product (six) wereactually used, and how 90 percent of them werekept out of the chronology; how contradictorySIGINT evidence was answered both with specu­lation and fragments lifted from context; how thecomplete lack of Vietnamese C3I was notaddressed; and, finally, how critical originalVietnamese text and subsequent product were nolonger available. From this evidence, one can eas­ily deduce the deliberate nature of these actions.And this observation makes sense, for there was apurpose to them: This was an active effort tomake SIGINT fit the claim of what happened dur­ing the evening of 4 August in the Gulf of Tonkin.

~Si/SfjThe question why the NSA personnelhandled the product the way they did will proba­bly never be answered. The notion that they wereunder "pressure" to deliver the story that theadministration wanted simply cannot be support­ed. If the participants are to be believed, and theywere adamant in asserting this, they did not bendto the desires of administration officials. Also,such "environmental" factors as overworked cri­sis center personnel and lack of experienced lin­guists are, for the most part, not relevant whenconsidering the entire period of the crisis and fol­low-up. As we have seen, the efforts to ensure thatthe only SIGINT publicized would be that whichsupported the contention that an attack hadoccurred continued long after the crisis hadpassed. While the product initially issued on the 4August incident may be contentious, thin, andmistaken, what was issued in the Gulf of Tonkinsummaries beginning late on 4 August was delib­erately skewed to support the notion that therehad been an attack. What was placed in the offi­cial chronology was even more selective. That theNSApersonnel believed that the attack happenedand rationalized the contradictory evidence awayis probably all that is necessary to know in orderto understand what was done. They walked alonein their counsels.

lap SI!ClltET1lCOMIN Inx, Page 49

Cryptologic Quarterly(b) (3)-P.L. 86-36

TOil SI!!CI\ET/teOMINTfIX I

(U)Notes1. (U) James S. Olson and Randy Roberts, Where

the Domino Fell: America and Vietnam 1945 to 1990(New York: St Martin's Press, 1991),118.

2. (U) James Bond and Sybil Stockdale, In Loveand War (New York: Harper & Row, 1984), 19.

3. (U) Robert S. McNamara, In Retrospect: TheTragedy and Lessons of Vietnam (New York: TimeBooks, 1995), 134-135.However, in the introduction tothe paperback edition of In Retrospect, McNamarapulls back a bit from his traditional certainty.

4. (U) Edward J. Marolda and Oscar P.Fitzgerald, The United States Navy and the VietnamConflict, Vol II: From Military Assistance to Combat,1959-1965 (Washington, D.C.: Naval HistoricalCenter, 1986), 440-442.

5. (U) Edwin Moise, Tonkin Gulf and theEscalation of the Vietnam War (Chapel Hill, NC:University of North Carolina Press, 1996), 199-200.

6. ('FS/fSlj tJ'he Gulf of TonkinIncident," Cryptolog, February-March 1975, 8-10;"Tonkin Gulf: The Untold Story of the 'PhantomBattle'that Led to War," U.S.News and World Report,July 23,1984, 84,-fSf/StrThe source of this quote wasa 20 January 1972 meeting between the DeputyDirector NSA, Dr. Louis Tordella, and the Chief ofStaff, Senate Foreign Relations Committee, CarlMarcy. In that meeting Dr. Tordella conceded that theintercept of 4 August could refer to the 2 Augustattack. Dr. Tordella could not produce the "original"version of the report in question. "Reading file of theDeputy Director," July-December 1971, NCA ACC#25853 and similar file, January-July 1972, NCAACC#25854.

7. (U) Marolda, 442.8. (U) McNamara, 134; Lyndon B. Johnson,

Vantage Point: Perspectives of the Presidency 1963­1969 (New York: Holt Rinehart and Winston, 1971),114-115.

9. (U) McNamara, 130.10. (U) Marolda, 394-395.11. (U) Ibid., 395.12. (U) McNamara, 130.13. (U) Marolda, 398.14. (U) Ibid., 404.

15. ('FBI/SIr-William Gerhard, In the Shadow ofWar (To the Gulf of Tonkin), Cryptologic HistorySeries, Southeast Asia, June 1969, National SecurityAgency, 51.

16. fF8//8I) OIC USN-467 Yankee. "SIGINTOperations Report, Desoto February/March Patrol,Submission of," 23 August 1961, NSG Archives, Box 7file 32316. Direct Support Operations Pacific 1964;2/0/VRN/TOl-64, 021943Z March 1964.

17. (TSffSI) Gerhard, 51. Naval DSUs used theSIGADUSN-467 as a generic designator for their mis­sions. Each specific patrol received a letter suffix for itsduration. The next mission would receive the subse­quent letter in an alphabetic sequence.

18. -(S/f81r- HQ NSAPAC to DIRNSA, 262020ZAugust 1964.

19.a:SffSftUSN-27, "Standard Operating Proce­dures for Support of August 1964 Desoto Patrol," 28July 1964.

20.1StCINCPAC, 140203Z July 1964.21.iS11bid.22. fftlffSI) See 3/0/VRK/T24-60, 20 July 1960

and 3/0/VRM/TlO-60, 28 July 1960, among others.23. 2/G9/VRM/R28-64, 10 June 1964, inter alia;

see Secret Army, Secret War by S. Tourison for a com­plete recounting of these doomed missions.

24. (S~ NSACSS MACV~ ~Tiger,6f~: :~:25. (8/f811 Ibid., Appendix A-I.26...(Sf/8t) Ibid., A-2.

27. fSffSf1Ibid., A-3. per CIA

28. (U) Harold P. Ford, CIA and the VietnamPolicymakers: Three Episodes, 1962-1968 (Center forthe Study of Intelligence, Central Intelligence Agency,1998),46.

29. (U) McNamara, 133; Moise, 22.30. (U) Olson and Roberts, 117.31. (U) Moise, 25; Young, 115.32.-(8ffS~ 2/Q/VHN/TI3-64, 281006Z July

1964.33. (8ftSI) 2/Q/VHN/T76-64, 010652Z July

1964.34.-(8ffSt12/0jVHN/R03-64, 8 June 1964.35. (U) Moise, 25.36. ~/f8trGerhard, 112; also, see DIRNSA

08006/02, 021615ZAugust 1964.

Page 50 16P 9EeRElo\'69MINio'RU

"(b) (3)-P.L. 86-36

l'9P SE6RET,.,oeOMINT/IX I Cryptologic Quarterly

021628Z,DIRNSA,

57. (D) Moise, 84.58. (D) Ibid., 86; (SffSI) 2/Gll/VHN/RlO-64

040850Z August 1964.59. (SffSI) 2/GlljVHN/R08-64, 031925Z.

60. fSff8tr2/GUjVHN/RlO-64, 040850ZAugust1964.

61. (D) Marolda, 421.62. (D) Moise, 88; Marolda, 415.63. (D) McNamara, 140-141.64.-(TSt Marshall Wright, et al., "The Vietnam

Information Group. Presidential Decisions: The Gulf

ofTonkin Attacks of August 1964," Lyndon B.JohnsonLibrary, 1 November 1968, CCH Series VIII Box 13.

65. ~TSf/SI) DlRNSA. "DRV Vessels AttackDesoto Patrol in Gulf of Tonkin," B205/243-64020947Z August 1964.

66. ~/;,SI)3/0~ -64.

67.~CINCPACFLT, 021104ZAugust 1964.68;-fStcru 72.1.2, 021443Z (Genser).69.-tStDesoto OPSUM, 021443Z August 1964.70.~Wright, 12.

71. (SffSI) 2/Q/VHN/T126-64, 030450Z August1964.

72. ~~1/8l) 2/Q/VHN/T131-64, 050834Z August1964.

73. ~~/fSl) 2/GlljVHN/R07-64 August 1964.74.-fB1July/August Desoto SITSUM (3) 030745Z

August 1964 GENSER.75. (SttSf) 2/Q/VHN/R32-64 031008Z August

1964.76. (SffSOC) 2/Q/YHN/T144-64 080234Z August

1964.77. (SOSt) 2/Q/VHN/T142-64 080228Z August

1964; (D) Marolda, 42l.78. ('FS/fSIt OIC USN-467N, July-August 1964

Desoto Patrol srGINT Observations Report," USNSGASer 0003 23 August 1964.

79. (D) Marolda, 423;-kBt CTG 72.1 "July-August

Desoto SITSDM," 031405Z August 1964; 4S//St)'2/Q/VHN/R32-64 031008ZAugust 1964.

80. ~SffSI) 2/Q/YHN/T146-64 080314Z August

1964.81. -EB17Sf) 2/Q/YHN/T151-64 080324Z August

1964.

37. ('FSf/SOOralH~tOry interview withl I_____INavy enlisted Vietnamese/linguist, 22

December 1987, NSA OH 33-87, 8; Oral-History inter­view with Lieutenant General Gordon-A, Blake DSAF,

Director NSA, 1962-1965,5 June 1972, NSA OH -72, 5June 1972, 3-4.

38. €+SffSij DIRNSA, 070118ZAugust 1964.39. -fBf/8fT 2/Q/YHNj:R24-64, 310922Z July

1964.40. (TSNSI)I pH 33-87.41. ~SffSI) 2/Gll/VHN/ROl-64 Spot Report,

01l635Z August 1964; (TS/fSI) OIC USN 467November. "July-August 1964 Desoto Patrol SIGINTOperations Report," Serial 0003, 23 August 1964.

42. -(SHSI)... 2/Q/YHN/T26-64, 0l924Z August1964; 2/Q/VHN/T137-64, 080216Z August 1964.

43. (SNSI) 2/Q/VHN/T130-64, 050827Z August1964.

44. (SNSI) 2/Q/YHN/R27-64, 012152Z August1964; FLWP Nr.1 to 2/Q/YHN/R27-64.

45. (sttS!) 2/GlljVHN/R02-64, 020745ZAugust1964.

46. (S/fSI) DIRNSA. "Possible Planned Attack byDRV Navy on Desoto Patrol," B205/981-64, 020302ZAugust 1964; NSA Command Center Record of Events.2 August 1964. The navy intercept sit~,-- _

'-- !had the responsibility for relaying Criticomm

messages to the DSU aboard the Maddox. However,the initial Critic for 2 August wasiNOT passed untilmuch later. NCAACC# 45582, H04-030l-4.

47.~/fSl1CRfTIC,USN-27to DlRNSA, 020444ZAugust 1964; 2/GlljVHN/R02-64, 020745Z August1964.

48. tstf.ffl Ibid.49.~) 2/Gll/VHN/R06-64, 022127Z August

1964.50. (SffSI) 2(Q/YHN/T135-64 050950Z August

1964.51. (D) Moise, 73.52. (SlISI) 2/Q/VHN/T134-64 050948Z August

1964.53,-EB11~ 2/Gll/R03-64, 020822ZAugust 1964.54.(D) Moise, 73.55. (U) Ibid., 74.56. ~//Srr DIRNSA, 020947Z August 1964,

08004/02.

(b) (1)(b)(3)-50 USC 403(b) (3)-P.L. 86-36

leP 9EeREliJleOMINTttX I

(b) (1)(b)(3)-50 USC 403(b) (3)-18 USC 798(b) (3)-P.L. 86-36 Page 51

Cryptologic Quarterly lOP SECRETttCOMINTItX I

TeflSECRE IhCOMINTJIX1

82. (~/ISI) Critic USN-414T, 041115Z August

1964.

83. (81/Sf) 2/Gll/YHN/R11-64, 041140Z, August1964.

84.~CI1J 77.12, 041240Z.85. Ol//~I) An interesting speculation emerged

after Captain Herrick's initial radar contact placed thesuspect boats so far to the northeast of his positionthat these craft could have been Chinese Communistnaval vessels on patrol from nearby Hainan Island. As

86. (U) Moise, 123.

87. (U) Moise, 127; Marolda, 429.88. (U) Moise, 71.

89. (U) Ibid., 126.

90. (U) Ibid., 127.

91. (U) Joseph C. Goulden, Truth is the FirstCasualty (New York: James B. Adler Inc., 1969),146.

92. (U) Ibid., 143.93. (U) Moise, 129, 160.

94. (U) Goulden, 144.95. (U) Moise, 144.96. (S) Desoto Action SITREP, Z 041754Z August

1964.97.~ CTU 77.12 "Attack on Desoto Final

SITREP," 042158 August 1964, CCH Series VII. Box

13.98. (U) Moise, 211-212; Robert Divine, Since 1945:

Politics and Diplomacy in Recent American History(New York: Alfred Knoff, 1985 (Third edition)), 139.

99. (U) Moise, 209; John Schulzinger, A TimeforWar: The United States and Vietnam, 1941-1975(New York: Oxford University Press, 1997), 151.

100. fSt-"Memorandum for Mr. Bundy," 23, 10

August 1964 from LBJ Library, CCH Series VIII, Box13, Gulf of Tonkin, "Vietnam War Records."

101~Ibid.

102. ~ CI1J 77.12 Desoto, Final SITREP, Z

042158Z August 1964.

Page 52

103.i8)"CfG 77.5, Desoto Patrol SITREP Nr 8., Z

041928Z August 1964.

104.-(5/181) 2/0/YHN/TIO-64, 041559Z August1964; also, portions quoted in 3/0/YHN/ROl-64.

105. tBt-"Memorandum for Mr. Bundy," 24;McNamara, 134.

106. (U) Lyndon B. Johnson, 114-115.

107. ff'SIISI) 3/0/YHN/ROI-64, 5 August 1964.108.i81/~Ij USN-414T, 041115Z August 1964.109~{{SI)"USM-626J, 041140ZAugust 1964.

110. (81181) 2/G11/YHN/RI3-64, 041440ZAugust

1964. It is possible that this intercept was the missing

part of the message sent earlier and construed as the

"attack" order.

111. (81181) 2/Q/YHN/TI63-64, 090328Z August1964.

112. (S{{SI) 2/Gll/YHN/R17-64, 041820ZAugust1964.

113. (Sii8I) 2/Q/YHN/R42-64, 041900Z August

1964.114.iStCI1J 77.12 Z 041240Z August 1964.

115. (~//~I) 2/Q/YHN/TI28-64 041838Z August

1964.116.~) Ibid.

117.(+S//~I) 3/0/YHN/ROl-64, 5 August 1964.118. (U) Marolda, 408; bomb damage assessments

from 5 August indicate that some P-4s may have been

at Quang Khe. However, there is no SIGINT reflectionto support any activity during this period by P-4s inthe Southern Naval Command.

119. (8//8I) 2/G11/YHN/R16-64, 041746ZAugust1964.

120.~ JCS Z 0 041628Z August 1964, CCH

Series VI.HH.24.10.

121.-€S/fSI) 2/G11/YHN/R15-64, 041745Z August

1964.122. (SHSI) DIRNSA, 070118ZAugust 1964.

123.(8Yi81) US-467N, 070615Z, August 1964.124. (Sl/SI) 2/Q/YHN/T151-64, 080324Z August

1964. NCAssx» 45359Z.125.-(81181) 2/Q/YHN/T152-64, 080440ZAugust

1964, NCA ACC# 45359Z.126. (SIISI) 2/0/YHN/R11-64, Spot Report 3

September 1964.127.-(S11SI) Ibid.128. ts77SI) OlC, USN-467N.

(b) (1)(b)(3)-50 USC 403(b) (3)-P.L. 86-36(b) (3)-18 USC 798

TOil SI!C~I!TiI'COMINTiIX 1 Cryptologic Quarterly

129. (~lI~t) Ibid.130. (~//SI) COMNAVFORJAPAN. "Preliminary

All Source Evaluation Maddox Patrol," 120230ZOctober 1964.

131.(D) Moise, 163-164.132. (D) Sir Arthur Conan Doyle, The Complete

Sherlock Holmes (Garden City, New York: Doubleday& Company, 1930), 347.

133. CD) Moise, 201.134. CD) OIC DSN-467N.135. (S//SI) 2/Q/VRN/T126-64, 030450Z August

1964; attached worksheet 1609-64, DSN-414T. NCA

ACC# 45359Z, DSN-27 1964 translations. Also, see2/G11/VRN/R06-64, 022127ZAugust 1964.

136. (S//St)' 2/Q/VRN/R39-64, 041646Z August1964. This serialized report was sent at a precedenceC"ZZ") reserved for Critics. This was a technical errorand did not affect warning or timeliness. However, itmakes tracking down the reports more difficult.

137. (D) Moise, 200.

138s,f~::I) B26 COMINT Technical Report#009-65,l __ F9 March 1965,CCH Series .NN.

139. CD) Moise, 200; Lyndon B. Johnson, 114.fTSf/SI) In the matter of who the "experts" were thatPresident Johnson is referring to, it appears that theywere members of the White House IntelligenceAdvisory Staff. According to one source, they exam­ined rall available intelligence having even the mostremote relevance [to the Gulf of Tonkin]." Accordingto this source, SIGINT alone provided "positive evi­dence of DRV premeditation." Whether this evidence

refers to the 2 or 4 August incidents is unclear. It isalso not certain if all "relevant intelligence" was pur­sued. Furthermore, it is not clear if any of the "experts"were Vietnamese linguists, or if they were being

prompted by the analysts at NSA. Source: NSAMemorandum for the Record: "Interview with Mr.Arthur McCafferty, White House Staff, on the use ofSIGINT in Shaping White House Decisions onSoutheast Asia," CCH Series XII.NN., undated.

140.~ 2/Q/VRN/R39-64.141."t~//S6 For identity of LAP, see 2/Q/VRN/

TI23-64; for TRA see 2/QfVHN/TI34, 135-64.Callwords/eovernames were used to designate units,entities, and individuals. It was not uncommon to see

(b) (1)(b)(3)-50 USC 403(b) (3)-18 USC 798(b) (3)-P.L. 86-36

a particular entity, such as T-142, addressed with asmany as three callwords over this period. However,these callwords equated to any number of differingentities that would have been aboard.

142. fSf/SI) The missing intercept would havearrived in the form of so-called technical supplementsto the San Miguel reports "38" and "39." Generallythese supplements were sent anywhere from fifteen toforty-five minutes after the report was issued. Theyincluded the original Vietnamese text. These supple­ments were sent to a small audience of SIGINT-pro­

ducing elements. The supplements probably werewhat the B26 crisis center used in generating the after­action report, "T-lO."

(8//81) As a general practice, once the technicalsupplements were received, they were attached to theoriginal reports. A review of the NSA archival file con­taining the San Miguel reports issued in 1964 revealedthat reports in the preceding and following series havetheir supplements attached, while the two reports inquestions stand alone, sanssupplements.

143.~ 2/Q/VRN/T163-64, 080522Z August1964; as reported in 2/Q/VRN/R38-64, 041632ZAugust 1964. The text of "R38" was "at 041554ZSwatow-class PGM T-142 reported to My Due(19°52'N, 105°57'E) that an enemy aircraft wasobserved falling into the sea. Enemy vessel perhapswounded." The translation quoted on page 34 was ofthe full text that would have appeared in the missingTechnical Supplement.

144. €~h'~I) Ibid.145. CD) Moise, 106.

146. fSffSI) 2/Gll/VHN/RlO-64, 040850ZAugust 1964.

147. (D) Moise, 199-200.

148. ~/fSI) Oral History interview wiuiI ~3. Oral History interview withLie....u~te-na-n~t ....

General Gordon Blake, 4.149. en;l/sI) Blake Oral History, 9.150. (U) Moise, 197.151.-ffS//~ Johnson, 522.152. ~ffS~ IOral History; Oral History

Interview with Milt Zaslow, OH 17-93.

153. (S//SI~ IOral History.154.~)lbid;

(b) (3)-P.L. 86-36

Page 53

Cryptologic Quarterly 'Oil SE6RElJ:e9MINT{R(4 (b) (3)-P.L. 86-36

155. (TSIISI) See 3/0jVHN/ROl-64, Spot Report4 August 1964; 3/0jVHN/R02-64, 060J02Z August1964; and 3/0jVHN/R03-64, 061604Z August 1964.

156. ('fS11SIlj IOral History, 8; ZaslowOral History, 33-35.

157. ('F8f(81) Blake Oral History, 12.158. (U) Moise, 241-243.159. E8f(8I) HQ NSAPAC 080105Z August 1964;

William Gerhard, Untitled Notes on Gulf of Tonkin,CCH Series XII.NN (1970).

160. (U) Marolda, 443.161. ('F8ff8I) DIRNSA File, "Gulf of Tonkin, Both

Incidents," CCH Series VI.24.6.162. ('F8f(8I) OIC USN-467N.163. E~//Sl) DIRNAVSECGRUPAC 241832Z

August 1963.164. ESIISI) 2/0jVHN/Tll-64, 061656Z August

1964. A curious archival sidenote to this translation.When this historian reviewed the original NCA acces­sion of NSA-issued translations of DRV naval inter­cept, this one translation was missing. A copy wasfound in the files on the Tonkin Gulf incidents held bythe Center for Cryptologic History. A copy was thenreplaced in the NCAfile.

165. fTSIfSI) 3/0jVHN/ROl-64, 050130ZAugust1964 NCAACC# 45359Z.

166.~ See the following: JCS messages, 041610ZAugust 1964; 041830Z August 1964; 041811ZAugust1964; and 041754ZAugust 1964, Desoto Action Sitrep.

167. (TSf(SI) Spot Report 3/0/VHN/R03-64August 1964 NCAACC# 45359Z.

168. ffS11S1 NSA Command Center record ofEvents, 6 August 1964, NCAACC# 45582.

169. ('FSf(SI) 3/0jVHN/ROl-64.170. fTSIISI) 3/0/YHN/ROl-64, 5 August 1964.

171. is//Sij 2/QfVHS/R36-64 031212Z August1964; USN-27 Tech Supplement to same, 031236ZAugust 1964.

172. (S/fSH Ibid. Also, Desoto SITSUM 5,030745Z Genser August 1964; and 3/0jVHS/R38-64DRVMERSUM 3-9 August 1964.

173.~) Ibid.174. (XS/fSij Spot Report 3/0jVHN/#03-64, 6

August 1964.175. (TSngI) Blake Oral History, 7.

176. (SIISI) 2/QjVHN/T154-64, 080328Z August1964.

177.(U) Moise, 78.178.-(SI1St') DIRNSA File, Gulf of Tonkin, Both

Incidents, CCH Series VI.24.6.179. ETSf(SI) Lt. Col. Delmar Lang USAF,

"Chronology of Events of 2-5 August 1964 in the Gulfof Tonkin," 14 October 1964, CCHSeries VI.HH.24.10.

180. ffBff81) Ibid., paragraph 8.181. -ET8f/SI)- They are, in order:

2/GlljVHN/R07-64, 030956Z August 1964; USN­414T Critic, 041115Z August 1964; 2/GlljVHN/Rll­64, 041140Z August 1964; USN-414T Translation,unnumbered (time of intercept 040927Z August1964); 2/0jVHN/T1O-64, 041955Z August 1964; and2/GlljVHN/Rl8-64, 050435Z August 1964.

182. fFSIfSI) Memorandum for the Secretary ofDefense, "Release of COMINT Pertaining to Gulf ofTonkin Incidents of 2 and 4 August 1964," SI-TS­61/PL-4, 13 December 1967, CCH Series VIII, Box 13,Gulf of Tonkin Incident.

183. (TSltSI) Blake Oral History, 12. (U)Washington Post, 25 February 1968, "McNamaraDescribes Gulf of Tonkin Incident." At separate timesSecretary McNamara refers to four, then nineCOMINT Reports in his testimony.

184. (U) Note, BillGerhard to Mr. Lowman, et aI.,Subject: "Following for information concerningrequest/response," 3 September 1975, CCH SeriesVIII, Box 13, Gulf of Tonkin Incident.

185;ofB118n 2/QljVHN/RlO-64, 170515Z Septem­ber 1964. NCAACC# 45349Z.

186. ($;fffftj 2/Ql/VHN/RI8-64, 180920ZSeptem-ber 1964.

187.~CS, 190309Z September 1964.

188.{S1JCS, 190536Z September 1964.189. fffli/~I) 8205, "Chronology of Events of 18­

20 September 1964 in the Gulf of Tonkin," 14 January1965,4. After the second incident in September, thedifference in the NSAreporting, that is, the lack of evi­dence for an attack, was noted quickly by the PFIAB.This led the Board to order an investigation, which, intum, led to the development of the chronologies of thetwo attacks. Blake Oral History, 10-12, 14.

190. t'fSIISI) Ibid., 5.190...ffB11SY) Ibid., 5.

Page 54 TepSEeRET1ICOMINT/1X1

TOP SEe~!TtleeMlttT:tJ(1 Cryptologic Quarterly

191. (U) Marolda and Fitzgerald, 462.192. (D) Olson and Roberts, 117; Schulzinger, 148.193. (D) Kamow, 376.194. (D) Olson and Roberts, 120.195. (U) The Gulf of Tonkin Resolution was

repealed in May 1970. Ironically, the initiative for the

repeal originated with Senator Robert Dole of Kansas.196. (D) Congressional Record, Volume 118, Part

3, 92nd Congress, 2nd Session, 3-14 February 1972,3313.

(UNFOUO} Mr. Hanyok is a senior historianwith the Center for Cryptologic History(E05). He worked in the NSAICSS

Archives from 1992 to 1994 and in theNational SI6INT Operations Centerfrom 1990 to 1992. Mr. Hanyok hasalso served as a collection officer in 6

, Group (1976-1979), in the COMSECDoctrine organization (1979-1982), asa Traffic Analysis intern (1982-1984),and as an analyst in A Group (1984­1990). He holds the title of master in thetechnical track program of the IntelligenceAnalysis Career field.

rep 6EeRETflCOMIN IIIX, Page 55