Code-breakers enabled U.S. spies to peer into the communications centers of the North Vietnamese, but then problems arose.
THROUGHOUT MUCH OF THE WAR, AMERICAN code-breakers intercepted North Vietnam- ese radio transmissions of planned actions along the Ho Chi Minh Trail—perhaps the National Security Agency’s greatest contribution during the war. President Lyndon B. Johnson certainly thought so. The NSA’s success excited Johnson and his national security adviser, Walt W. Rostow, when they were briefed on the program in early 1968. The American spy program, called the “Vinh Window” after a city that was one of the main interception points, was undoubtedly the most extensive penetration into North Vietnam’s secret communications that the United States achieved. Johnson and Rostow thought this intelligence would win the war. It’s not the NSA’s fault the war turned out differently.
Communications intelligence, or “Comint,” came to Vietnam early. American radio detachments worked in Indochina even during the French war in 1954, and once the insurgency reignited they came back quickly. In April 1961 John F. Kennedy’s administration ordered all U.S. intelligence agencies to establish a presence in South Vietnam and trade information with the Saigon government. Kennedy rejected plans to send U.S. combat troops to Vietnam in November 1961, but he accepted “covert annexes” that included increasing the NSA contingent. That December a U.S. Army Comint man became one of the first Americans killed in action. Specialist 4 James T. Davis had been working with members of a South Vietnamese radio team when their truck hit a land mine and guerrillas ambushed the survivors.
The Vinh Window operation principally involved the United States Comint units at Phu Bai and Da Nang, in the upper reaches of South Vietnam. Americans came to that area early in 1962, when a 50-man team arrived at Da Nang to support tactical air operations. The radio spooks conducted “hearability studies”—they listened in on radio transmissions to find the best place for interception antennas. Around Da Nang the clear preference was the nearby Monkey Mountain. At first the Americans employed only two radio receivers, Air Force Security Service specialists trained and focused on adversary air forces. But they had little to monitor aside from some Chinese activities in southern China and North Vietnamese transport flights.
General William Westmoreland, the U.S. commander in Vietnam, demanded more relevant intelligence, but those efforts encountered special difficulties. The effect of tropical temperatures and humidity on radio waves hampered the U.S. spies’ ability to record transmissions. In addition, radio beams on the frequencies the Americans were working followed a line-of-sight path, further cutting the interception range. Without the men or equipment for more stations, not to mention intercept locations near enemy transmitters, the radio spooks were challenged to find practical solutions for those problems.
In 1964 the Air Force stationed a Comint aircraft at Da Nang. The following year the intercept station at Da Nang became the 6924th Air Force Security Squadron, which grew to 30 intercept posts manned 24/7 and an aircraft component of “Commando Lance” RC-130 radio surveillance planes. In addition, a detachment of Marine radiomen previously in the Central Highlands moved to Phu Bai in 1963. Sailors of the Naval Security Group joined them.
The Marines went to Da Nang in 1965 and worked directly with the headquarters of the III Marine Amphibious Force, the high command for American troops throughout the region. The Army got into the act with its 8th Radio Research Unit, at Phu Bai from November 1964. All these Americans— and their South Vietnamese allies—increasingly focused on North Vietnamese and Viet Cong tactical communications and logistics traffic.
U.S. efforts to spy on North Vietnamese through radio intercepts faced one particularly vexing problem: For a long time Hanoi and the Viet Cong put few communications on the air. They did not use radios to direct engagements until the very end of 1964, at the battle of Binh Gia. Hanoi’s intelligence service, the Research Bureau, deliberately minimized its use of the ether; messages went primarily by courier. Most radios were inexpensive shortwave receivers, and radiomen simply listened in and wrote down encrypted Morse code transmissions from Radio Hanoi. The North Vietnamese maintained a handful of transmitters that were used when spy networks urgently needed to get a message up north. Hanoi’s People’s Army did not engage openly until 1965, and its radio equipment—a hodgepodge of French, Chinese, Japanese, Russian and American gear—proved increasingly difficult to maintain. In North Vietnam, most communications went through the old French telephone system, and landline communications could not be intercepted without physically putting a wiretap on cables inside enemy territory.
For what radio traffic there was, Hanoi used basic codes that had evolved from a system developed by the People’s Army Cryptographic Bureau as early as 1951. French radio intelligence had broken many variants of those codes, but ones introduced late in 1953 had stymied all French efforts to penetrate them. Although Hanoi’s codes were surprisingly good, the North Vietnamese lacked sophisticated encryption devices, and their codes were not impregnable. For example, the NSA was able to read North Vietnamese naval messages at the Gulf of Tonkin. Hanoi changed its land forces codes and its communications system in 1962, however, making them more difficult to crack. At this point the People’s Army introduced specific codes for messages to Laos and the forces running the Ho Chi Minh Trail. The North Vietnamese further improved their security in 1963, centralizing communications at their Hanoi transmission hub. From 1965 through 1967 they introduced radio techniques that required no responses, which protected subordinate stations from revealing themselves.
American radio spooks fell back on direction-finding and traffic analysis. In direction-finding, different interception stations take bearings on a message’s point of origin and triangulate the bearings to locate the transmitter. Traffic analysis plots the hierarchy of the network by observing which stations send orders, which ones report and what the number of messages indicates. Over time this record generates an increasingly precise picture. These methods became the backbone of U.S. Comint in Vietnam. Using them, the Americans were able to discover the expansion of the Ho Chi Minh Trail in 1964-65, the moment when the North Vietnamese sent troops across the Demilitarized Zone in 1966 and the movement of two infantry divisions toward Khe Sanh a year later. Some argue that after a certain point—in 1966 or 1967—the Comint was aware of every major action being planned by Hanoi.
But detailed knowledge of enemy intentions and strength remained shrouded in the encrypted messages. Westmoreland’s strategy of defeating his adversary through attrition depended on an accurate assessment of enemy strength, and a key factor in that calculation was the number of fresh troops Hanoi sent down the Ho Chi Minh Trail. American road-watch teams and aerial reconnaissance along the trail furnished some information, but much was of questionable accuracy. The big break came from the North Vietnamese themselves.
The A Shau Valley had long been an arena of confrontation. Hanoi strengthened its hold there in 1966, when it forced the evacuation of the A Shau Special Forces camp. The valley then became a base area and one terminus for the Ho Chi Minh Trail. In May 1967 monitors at Comint’s Da Nang–Phu Bai complex overheard the enemy in plain-language conversations that emanated nearby. Direction-finding and traffic analysis traced the North Vietnamese transmitters to the A Shau and linked them to the Rear Services Department of the People’s Army, responsible for operating the trail. Voice radio conversations had been rare in III Marine Amphibious Force sector—none had been detected until 1965—and had never been encountered on Rear Services networks. It turned out that Hanoi, seeking higher rates of data transmission, had begun switching to voice traffic. Most of those messages concerned air defenses, but for the first time they offered a look at the enemy’s playbook.
Five months later a 6924th Squadron plane out of Da Nang overheard similar voice radio from a location in the panhandle of North Vietnam. Orbiting over the Gulf of Tonkin, the Commando Lance recorded traffic in a simple voice cypher the code-breakers soon understood. More North Vietnamese transmitters joined this network, at points from Thanh Hoa to Vinh, until 31 had been detected. The network also belonged to the enemy’s Rear Services Department. Once the radio spies deciphered the messages they realized the discussions concerned the movement of supplies and reinforcements down the trail. Traffic analysis established that the enemy transmitters worked for binh trams, the sector commands along the trail, and their subordinate way stations, which were sites where convoys stopped and troops camped on their march south. This source of information soon became known as the “Vinh Window” because the Americans were working on traffic intercepted from that station when they first broke the enemy codes.
The NSA’s director, General Marshall S. Carter, directly controlled Comint in Southeast Asia and made the Vinh Window a top priority. The 6924th Squadron had four of the Commando Lance aircraft. Carter, based at NSA headquarters in Fort Meade, Md., persuaded the Air Force to divert RC-130s from units in Europe to increase the Commando Lance detachment at Da Nang to six planes, permitting Comint missions for 12 hours a day. But the RC-130 had one critical weakness: It was restricted to relatively short missions, which meant that more aircraft were needed to maintain the radio watch. Even a 12-hour watch meant putting every aircraft on the line, either to fly or to serve as a spare.
The Air Force came up with an alternative: RC-135 “Rivet Card” jets of the 82nd Reconnaissance Squadron. Crammed with Comint and stationed at Yokota Air Base in Japan to monitor the Soviet Union, the planes were redeployed to Kadena Air Base on Japan’s Okinawa island and added the Gulf of Tonkin to their assignments. Each Rivet Card jet, a specialized version of the Boeing 707, could spend 19 hours in the air. If one plane spent 12 of those hours orbiting in the gulf, two planes could keep a 24-hour watch. Starting in early 1968 the Rivet Card entirely replaced the RC-130 on the gulf orbit. Commando Lance planes then shifted to mounting watch in Laos over the Ho Chi Minh Trail.
Interception of the Vinh Window voice radios opened a stunning panorama. During the five months preceding the Tet Offensive the radio spies identified 50 infiltration groups of People’s Army soldiers, most confirmed by other sources.
The data quickly assisted U.S. air operations, which suddenly could narrow their targets to the points where North Vietnamese troops were expected. When targeters lacked specific intelligence, they just aimed at general interdiction and bombed the Hanoi-Haiphong region; as the data collection improved however, U.S. bombers put more emphasis on other places. The Rolling Thunder air assault against the Hanoi-Haiphong area peaked in October 1967 at 36 percent of U.S. combat sorties. At that time 61 percent of U.S. sorties were aimed at the North Vietnamese panhandle. A month later the attacks against the infiltration routes rose to 75 percent of Rolling Thunder strike missions, and in December to 83 percent. Meanwhile, action against the Hanoi-Haiphong zone dropped to 15 percent of the total. Over the same period the Seventh/ Thirteenth Air Force doubled and then more than redoubled the sorties flown over the Ho Chi Minh Trail in Laos.
In December 1967 nearly 6,000 combat sorties were sent against Vietnamese panhandle targets and an equal number were directed at south Laos, but barely 800 sorties were flown against Hanoi-Haiphong. Even more telling, over half of Rolling Thunder combat missions (57 percent) went against the sector south of Vinh and just above the DMZ. The air generals increased their concentration on that area, known as “Route Package 1,” until Rolling Thunder devoted to it more than four times the effort being directed at Hanoi-Haiphong. During Tet and afterward, air commanders maintained this high-priority effort.
Hanoi changed its method for identifying infiltration groups from a three- to a four-digit system in February 1968. The radio spies quickly adapted. Even better, the code-breakers discovered the first digit stood for the destination of the force. Additionally, radio intercepts concerning requisitions for supplies at the binh trams helped U.S. troops estimate the size of infiltration groups. The interrogation of prisoners and defectors then permitted a check—and further refinement of force projections. Sometimes the North Vietnamese gave away that information directly. A typical example is a report that Art McCafferty, director of the White House Situation Room, furnished to Johnson security adviser Rostow on June 10, 1968. In it he observed that the number of infiltration groups on the trail had diminished over the preceding month from 18 to seven, but the projection of People’s Army reinforcements en route or arrived had grown from 151,000 to 177,000 men.
The quality of the information improved steadily. When General Creighton Abrams took over from Westmoreland in spring 1968, the Vinh Window intelligence became a staple at his command briefings. That August he learned that 124,000 infiltrators had entered the pipeline. Abrams and his intelligence chiefs speculated on the accuracy of the projections. At another typical briefing, in November 1968, the discussion included the tonnages of specific types of supplies the North Vietnamese had moved the previous week. In June 1969 Abrams’ intelligence chief told him captives had identified only two infiltration groups that had not appeared in Vinh Window traffic. The data could become quite granular. In June 1970 Abrams learned that only 39 of 116 men in an infiltration unit had arrived with their group. The rest were dead, exhausted or sick, or had stayed behind at binh trams along the trail. At a May 1971 briefing, intelligence analysts reported a binh tram’s notice that half its vehicles were out of commission.
Vinh Window intercepts give us new perspective on “Igloo White,” the electronic battlefield the United States created along the trail. Igloo White used strings of sensors to detect movement, a command center in Thailand to track the big picture and aircraft to attack the targets. As a result of the Vinh Window, the Seventh/Thirteenth Air Force knew when to expect infiltration groups to enter the battle arena. From 1968 to 1972 the Air Force mounted eight “Commando Hunt” campaigns focused especially on the infiltration routes. The rain of destruction peaked in 1969 when 433,000 tons of bombs fell on the trail.
Maintaining the Vinh Window became difficult, how- ever. The most pernicious problem was the sheer mass of information. Not all of the intercepts had intelligence value, but some offered opportunities for immediate action. Efforts to pinpoint those opportunities included instant transcription and preliminary analysis aboard the aircraft so that only the most important information would be forwarded to ground stations. But not much room could be spared for linguists and analysts on the planes—the Rivet Card RC-135s were already crowded with 35 radiomen and flight crew—so the onboard analysis scheme proved impractical.
Analysis at Phu Bai, where most translation occurred, also became more challenging as the volume of intercepts mushroomed. The Army and Navy detachments there lacked enough Vietnamese linguists to do the job, and a backlog soon developed. In 1969, when Navy Vice Adm. Noel Gayler took over the NSA, he tried to transfer as many American linguists as possible to Phu Bai for Vinh Window work. Their places at other stations could be filled by South Vietnamese recruited for American Comint units under a program known as Dancer, which had begun a few years earlier. Some Dancer linguists even made it to Phu Bai. But the NSA and its service affiliates could never be sure that some Dancers weren’t really enemy agents, so even very skilled ones had to prove themselves time and again. In addition, once the Saigon government declared general mobilization in 1968, even South Vietnamese working for the Americans were subject to call-up for their country’s army. That caused more headaches as the NSA tried to protect its Vietnamese workforce. The Vinh Window backlog persisted.
There were also problems with the intelligence itself. Hanoi did not number its groups sequentially. This led to endless speculation over “gap” groups: Had the radio spies failed to intercept relevant messages or did those infiltration groups even exist? Should intelligence include numbers for gap groups in its estimates? U.S. intelligence conducted many studies to solve this mystery but never cleared it up. Even so, the Vinh Window provided far more information than American intelligence had ever had.
Meanwhile Hanoi, unable to preserve the security of its communications, did the next best thing and tried to interfere with the communications collectors. The Rivet Card planes endlessly orbiting over the Tonkin Gulf could be tracked on radar, and the North Vietnamese occasionally sent MiG-21 fighters after them. The MiGs would go supersonic as they approached, make a single pass firing everything they had and hightail it for home. The Americans countered with fighter escorts from aircraft carriers on Yankee Station in the gulf. Observing the Rivet Cards’ flight patterns, the North Vietnamese took advantage of times when escorts broke off to refuel, occasionally sending pairs of MiGs to strike them. The Americans responded with new tactics. Spy planes retreated to the southern end of their orbits to pick up a couple of fighters that sheltered within their radar image. The Rivet Cards then flew close to Haiphong to tease the enemy. MiGs scrambled, only to be destroyed by the fighter jets. After losing several MiGs, the North Vietnamese stopped chasing the spy planes. No Rivet Card was ever flamed on a Vietnam Comint mission. Rough weather frequently imperiling landings and takeoffs at Kadena were a greater concern than the enemy.
In 1972 the Vinh Window detected Hanoi’s deployment of most of its general reserves. The Air Force crafted a fresh aerial effort, “Island Tree,” to block the reinforcements. The Rivet Cards added a new orbit—one over Laos—to record the flood of North Vietnamese message traffic. But the intelligence bonanza did not blunt the enemy’s march. Hanoi struck at Easter and maintained its offensive through the year. Washington negotiated a cease-fire, signed at Paris in January 1973, and left the Vietnam War. Ultimately, the Vinh Window had not been quite the miracle Johnson anticipated.
John Prados is a historian based in Washington, D.C. His seven books on the Southeast Asian conflict include Vietnam: The History of an Unwinnable War, 1945-1975, a Pulitzer Prize nominee.
Originally published in the August 2014 issue of Vietnam. To subscribe, click here.