The Secret Sentry (15 page)

At one forty-eight a.m. GOT time, August 5, Herrick sent another message to Admiral Sharp at CINCPAC, which stated,

Certain that original ambush was bonafide. Details of action following present a confusing picture. Have interviewed witnesses
who made positive visual sightings of cockpit lights or similar passing near Maddox. Several reported torpedoes were probably
boats themselves which were observed to make several close passes on Maddox. Own ship screw noises on rudders may have accounted
for some. At present cannot even estimate number of boats involved. Turner Joy reports 2 torpedoes passed near her.
⁵⁴

Despite Herrick’s more upbeat and confident report, Sharp became worried about the strength of the evidence, or lack thereof,
regarding the purported engagement. The three after-action reports that Sharp had received from Herrick were far from definitive
and clearly indicated doubts about what had actually happened. When McNamara called Sharp at eight past four p.m. EDT (eight
past three a.m. GOT time, August 5), Sharp was forced to tell him that the latest messages from Herrick indicated “a little
doubt on just what exactly went on.” With the air strike preparations now nearing completion, this clearly was not what McNamara
wanted to hear. He told Sharp that the air strike execution order would remain in force (the aircraft were expected to launch
from their carriers in three hours), but ordered him to confirm that an attack had indeed taken place before the navy fighter-bombers
were launched.
⁵⁵

At four forty-seven p.m. EDT, McNamara met with the JCS “to marshal the evidence to overcome lack of a clear and convincing
evidence showing that an attack on the destroyer had in fact occurred.” Based on the information then available to CINCPAC,
Sharp concluded that an attack had taken place, an opinion that carried great weight with McNamara and the JCS. From Herrick’s
reports, which were a mixed bag at best, McNamara and the JCS were able to extract some evidence to support their belief that
the attack had occurred, including sightings of ship wakes by navy pilots; sonar reports of torpedoes being fired at the American
destroyers; a report from the captain of the
Turner
Joy
that his ship had been illuminated by what was believed to be a searchlight while taking automatic weapons fire; and the fact
that one of the destroyers had observed cockpit lights on an unidentified ship. Finally, and most important, there were a
number of SIGINT intercepts that appeared to buttress the case for an attack having occurred, the contents of which were apparently
briefed to McNamara and the JCS, though hard copies of the intercepts were not provided to those attending the meeting.
⁵⁶

Among the five evidentiary items then available indicating that an attack had taken place, the only two reliable pieces of
information were SIGINT reports from NSA. One was an intercept of a statement that a North Vietnamese patrol boat had shot
at U.S. aircraft. The other, received via teletype two hours earlier, at two thirty-three p.m., contained the text of a report
by an unidentified North Vietnamese command authority who stated that his forces had “shot down two planes in the battle area”
and that “we have sacrificed two ships and all the rest are okay.” At the end of the intercept was a report that “the enemy
ship could also have been damaged.”
⁵⁷

McNamara and the JCS knew from Herrick’s reports from the Gulf of Tonkin that there were numerous problems with the evidence
cited above. Admiral James Stockdale, then a navy pilot who flew from the
Ticonderoga
that night, later disputed the navy’s official position that pilots had seen the wakes of enemy torpedo boats and gun flashes.
A navy reconnaissance mission flown the morning after the supposed battle found no evidence of one, particularly oil slicks
or debris that would have supported the claim that the destroyers had sunk one or more of the attacking North Vietnamese ships.
The sonar evidence was highly dubious. Detailed examination of the reports of visual sightings turns up numerous inconsistencies
that in aggregate render these reports less than reliable, especially since they were “firmed up” after the JCS demanded conclusive
proof that an attack had taken place.
⁵⁸

The Fruit of the Poisoned Tree

This left the NSA intercepts as the sole remaining credible evidence to support McNamara and the navy’s contention that an
attack had taken place. A declassified NSA history notes, “The reliance on SIGINT even went to the extent of overruling the
commander on the scene. It was obvious to the president and his advisors that there really had been an attack— they had the
North Vietnamese messages to prove it.”
⁵⁹

But we now know that Johnson and McNamara got it badly wrong in their headlong rush to launch the retaliatory air strikes.
The former head of the State Department’s Bureau of Intelligence and Research (INR), Dr. Ray Cline, recalled that NSA fed
the White House and the Defense Department raw intercepts, which were analyzed and evaluated by civilian officials and military
commanders with little or no background in intelligence, much less SIGINT analysis. At no point were the SIGINT specialists
at NSA called upon to provide the benefit of their deep knowledge of North Vietnamese communications, nor were CIA intelligence
analysts called upon to provide an assessment of the intelligence concerning the alleged August 4 naval engagement. Cline
later told an interviewer, “Everybody was demanding the SIGINT; they wanted it quick, they didn’t want anybody to take any
time to analyze it.”
⁶⁰

McNamara’s proceeding solely on the basis of
his
analysis of the available SIGINT may go down in history as one of the most serious mistakes made by a senior U.S. government
official. He ended up seeing what he wanted to believe. Like a future secretary of defensenamed Donald Rumsfeld, the intellectually
gifted McNamara made no secret of the fact that he thought he was a better intelligence analyst than the men and women at
the CIA who had done it all their adult lives, a situation exacerbated by his intense distrust of intelligence professionals
in general. In another interview, Cline said, “I of coursenever had a lot of faith in Bob McNamara’s judgment about intelligence.
I think, like many policy makers, he was too persuaded of his own ability to analyze things correctly and he didn’t feel that
intelligence officers were very likely to tell him anything he didn’t already know. Now, this is a congenital disease among
high-level policy makers.”
⁶¹

If McNamara and the JCS had taken the time to look long and hard at the intercepts on the afternoon of August 4, 1964, maybe
history would be different, because there were some significant problems with the intercepts if they were to be taken as the
most conclusive proof that an attack had occurred that night.

For example, a halfway decent SIGINT analyst looking at the scanty evidence would have immediately noticed that there were
no intercepts of North Vietnamese radio traffic or radar emissions, such as one would expect to find during the course of
a heated naval battle, and such as had been intercepted by NSA during the first Gulf of Tonkin battle two days earlier. For
the August 4 “Phantom Battle,” there were no comparable intercepts to be found anywhere.
⁶²
Former NSA officials indicated that the traffic analysis reports produced by B Group at NSA headquarters at Fort Meade after
the battle showed only routine radio activity within the North Vietnamesenavy radio grid on the night of August 4. North Vietnamesenaval
traffic showed a heightened state of alert along the coastline, almost certainly because of the continuing OPLAN 34A raids,
but the NSA analysts could find no indications of any spike in radio traffic that would have been indicative of combat activity
by North Vietnamesenaval units.
⁶³

In the absence of any other reliable SIGINT information, the only piece of tangible evidence left was the report by the unidentified
North Vietnamese command authority, which McNamara thought was an after-action report on the August 4 naval battle. The substance
of the NSA translation is this:

We shot down two planes in the battle area, and one other plane was damaged. We sacrificed two ships and all the rest are
okay. The combat spirit is very high and we are starting out on the hunt and [are waiting to] receive assignment. Men are
very confident because they themselves saw the enemy planes sink. The enemy ship could also have been damaged.
⁶⁴

But in fact the NSA translation does not reflect what the navy listening post at San Miguel intercepted. In fact, the San
Miguel intercept reads as follows:

We shot at two enemy airplanes and at least one was damaged. We sacrificed two comrades but all are brave and recognize our
obligation.
⁶⁵

It would seem that some unidentified person or persons in the reporting unit of B Group, for reasons we can only speculate
about, not only changed the wording of the translation and, in doing so, the import and meaning of the text, but also changed
the call signs used by the North Vietnamese transmitter and recipient and reformatted the message to include material not
contained in the original intercept. Sadly, the section of the NSA historian’s report on how this could conceivably have happened
at Fort Meade was redacted by the NSA FOIA office. But more important, the intercept could not have been an after-action report
because it was intercepted only an hour after the destroyer
Turner Joy
opened fire, and the “battle” raged for another two and a half hours. The only reason McNamara thought it was an after-action
report was because he got it off the teletype from Fort Meade two and a half hours after the battle in the Gulf of Tonkin
was over. Apparently McNamara did not bother to look at the times contained in the intercept itself.
⁶⁶

The Rush to Battle

In retrospect, it is clear that everyone in the White House was in a hurry to act, and nobody seemed to want to take the time
to scrutinize the evidence that was available to see if it justified going to war. After reviewing the intelligence material
for all of two full minutes, Secretary McNamara and the JCS agreed that the evidence, in their opinion, clearly indicated
that an attack had taken place in the Gulf of Tonkin on the night of August 4. At five nineteen p.m. EDT (four nineteen a.m.
GOT time, August 5), without waiting for additional information from Captain Herrick in the gulf or conducting a detailed
assessment of the COMINT intercepts, McNamara ordered that the air strikes be launched within two and a half hours.
⁶⁷

At CINCPAC headquarters in Hawaii, a harried Admiral Sharp was still trying to figure out what had happened in the gulf from
Herrick and the commander of the Seventh Fleet in Japan when McNamara’s strike execute order arrived on his desk. Finally,
at about five p.m. EDT, Sharp was given the COMINT intercepts described above. After quickly scanning them with his intelligence
staff, at five twenty-three p.m. EDT Sharp telephoned General David Burchinal at the Pentagon and told him that the intercept
concerning the “sacrifice of two ships” had convinced him that the attack had taken place. Sharp told Burchinal that the intercept
“. . . pins it down better than anything so far.” Burchinal asked Sharp, “Indicates that [the North Vietnamese] were out there
on business, huh?” Sharp’s response was “Oh, yes. Very definitely.” Burchinal agreed with Sharp’s assessment, despite the
fact that he had not yet seen the intercepts that Sharp was referencing. The only “hot” item that Burchinal had to pass on
to Sharp from the Washington end was that McNamara was “satisfied with the evidence.”
⁶⁸

At five thirty-four p.m. EDT, Sharp sent a FLASH-precedence message to Herrick demanding a categorical and unambiguous answer
as to whether he could “confirm absolutely” that the attack had taken place and that two North Vietnamese vessels had been
sunk during the engagement.
⁶⁹

While Sharp was waiting for a reply from the Gulf of Tonkin, a FLASH-precedence message from NSA arrived in the Pentagon communications
center. A report based on intercepted Chinese air force radio traffic, it ominously stated that the Chinese were in the process
of sending a unit of MiG fighters from an air base in southern China to the North Vietnamese airfield at Dien Bien Phu.
⁷⁰

Twenty minutes later, Herrick sent Sharp a radio message containing a qualified answer to his inquiries:

Turner Joy claims sinking one craft and damage to another with gunfire. Damaged boat returned gunfire— no hits. Turner Joy
and other personnel observed bursts and black smoke from hits on this boat.This boat illuminated Turner Joy and his return
fire was observed and heard by T.J. personnel. Maddox scored no known hits and never positively identified a boat as such.

  The first boat to close Maddox probably fired torpedo at Maddox which was heard but not seen. All subsequent Maddox torpedo
reports are doubtful in that it is suspected the sonarman was hearing the ship’s own propeller beat reflected off rudders
during course changes (weaving). Turner Joy detected 2 torpedo runs on her, one of which was sighted visually passed down
port side 3 to 5 hundred yards.

  Weather was overcast with limited visibility. There were no stars or moon resulting in almost total darkness throughout action.
⁷¹

Herrick’s report was filled with so many inconsistencies that it served only to further muddy the waters, rather than clear
them up. Herrick knew when he sent it that his report conflicted with a message sent by the captain of the
Turner Joy
, which claimed to have sunk one enemy vessel and damaged another. But in sum, Herrick told Sharp that based on the information
available to him, he believed that the attack had taken place, subject to the qualifications contained in the body of his
report, but that he would investigate further and provide more conclusive proof if he could. After reading Herrick’s message,
at six p.m. EDT Sharp again called McNamara to tell him that Herrick now was convinced that the attack had taken place, but
that there remained serious questions as to whether the engagement had, putting in jeopardy the retaliatory air strike.
⁷²