The “present arrangement,” Wheelon wrote, “has been neither a happy nor productive one. External program control has frustrated many CIA activities or forced their development outside the terms of the agreement. Everyone who is aware of the NRO situation is properly concerned about it, and many believe that the present arrangement is basically unworkable.” 73
Wheelon then considered several alternatives for managing the national reconnaissance effort. Responsibility could be assigned to a single agency (NASA, the Air Force, the CIA), which would handle all aspects of the effort—from payload design to launch, operations, and recovery. That alternative was not viable, Wheelon argued, because although the CIA should have significant responsibility in regard to the procurement, tasking, and operation of reconnaissance systems, it needed the Air Force to conduct the launch, tracking, and recovery operations.74
Wheelon contended that since launch, tracking, and recovery were clearly DOD functions, and there was agreement at the highest level in DOD that targeting and orbit selection should be handled by the CIA, the only remaining question was who should develop payloads. He noted two alternatives—assigning the CIA the task of developing all payloads, on the grounds that it had to be done secretly and because the design should be responsive to national intelligence needs, or assigning the task to the Air Force. The second alternative was unacceptable because “it would give the Air Force complete control over all satellite reconnaissance,” and “its success would depend on continuing, faithful Air Force responsiveness to truly national intelligence needs.”75
A third possibility was to divide the task. Indeed, Wheelon proposed that there be “an orderly assignment of satellite payload development to the various agencies”—possibly with the Navy handling SIGINT payloads, the Army geodetic and mapping payloads, the Air Force high-resolution photographic systems, and the CIA search systems. Assignments would be made by the DCI and Deputy Secretary of Defense jointly.76
A fourth possibility, which Wheelon rejected, would be to assign the basic research role for satellite systems to the CIA, and leave development and procurement to the Air Force. The problem with such an arrangement, he argued, was that the aerospace industry was responsive to procurement agencies that had a large number of dollars at its disposal. It would be unreasonable to expect the aerospace companies to give their best efforts to a group with only a few million dollars to spend, when a half billion dollars would be available from the procurement agencies. In addition, it would be “unreasonable to expect the development and procurement agency to have deep, continuing enthusiasm for another’s concepts and become ‘a loving foster parent.’”77
All-out competition represented a fifth alternative, which Wheelon rejected because it “would be difficult to keep such a competition orderly, especially with a limited technical and industrial base in which to establish such a competition.” He concluded by noting that he hoped the CIA proposal would be accepted, but if that were not possible, “the assignment of all reconnaissance payloads to CIA is the only way to preserve dedication of these satellite collection systems to national intelligence needs.”78
In early summer, before much work was done on the ultimate agreement, the NRO staff learned that McMillan would be leaving in a few months. (He had in fact been fired by Vance.) Vance and John Bross reached agreement on August 6, and the resulting document was signed by Raborn and Vance on August 13, 1965. Vance apparently relied on the advice of Fubini, who may have been its principal author, in accepting the agreement. It incorporated several concepts he had discussed with various members of the NRO staff in the preceding weeks. Final details were worked out by Vance and Raborn.79
The agreement assigned responsibilities to the Secretary of Defense, DCI, and NRO and formally established a National Reconnaissance Program Executive Committee (NRPEC). The Secretary was to have “the ultimate responsibility for the management and operation of the NRO and the NRP,” choose the Director, concur in the choice of the Deputy Director, and review and have the final power to approve the NRP budget. The Secretary also was empowered to make decisions when the executive committee could not reach agreement.80 The DCI was to establish collection priorities and requirements for targeting NRP operations, determine frequency of coverage, review the results obtained by the NRP and recommend steps for improving its results if necessary, serve on the executive committee, review and approve the NRP budget, and provide security policy guidance.81
The NRP Executive Committee established by the agreement would consist of the DCI, Deputy Secretary of Defense, and Special Assistant to the President for Science and Technology. The DNRO was to sit with the committee but in a nonvoting capacity (a provision from an earlier draft and reinserted by Raborn that eliminated the DOD draft proposal he received that would have made the DNRO a voting member of the NRPEC).82
The committee was to recommend to the Secretary of Defense the “appropriate level of effort for the NRP,” approve or modify the consolidated NRP and its budget, and approve the allocation of responsibility and the corresponding funds for research and exploratory development for new systems. It was instructed to ensure that funds would be adequate to pursue a vigorous research and development program involving both CIA and DOD.83
The executive committee was to assign development of sensors to the agency best equipped to handle the task, while all other engineering development tasks—such as design of the spacecraft, reentry vehicles, and boosters—were assigned to the Air Force, with the proviso that development had to proceed on a coordinated basis to ensure “optimum system development in support of intelligence requirements.” At Raborn’s suggestion, the agreement also included the provision that “To optimize the primary objective of systems development, design requirement of the sensors will be given priority in their integration within the spacecraft and reentry vehicles.”84
The Director of the NRO would manage the NRO and execute the NRP “subject to the direction and control of the Secretary of Defense and the guidance of the Executive Committee.” His authority to initiate, improve, modify, redirect, or terminate all research and development programs in the NRP would be subject to review by the NRPEC. He could demand that all agencies keep him informed about all programs undertaken as part of the NRP. An annex to the agreement specified assignments of four optical-sensor subsystems to specific agencies. The CIA was assigned responsibility for development of CORONA improvements and development of the new sensor for a new search system once the concept for the full system was selected.85
Former CIA and NRO officials concurred on at least one matter—the agreement significantly reduced the independent authority of the NRO director. They disagreed on its wisdom. To McMillan, it was “a victory for the wrong guy.” Many at the NRO may have shared that conclusion, but at the CIA there was a different perspective. The DCI didn’t “really have a job under the agreement,” in Wheelon’s view, but the agreement was “a triumph of people who cared about the program,” and it “provided adult supervision” for the DNRO. Subordination of the DNRO to the three senior officials who made up the executive committee meant his decisions were subject to review, and he “could not act unilaterally.” The intent was “to stop adventurism on the part of the DNRO.” Frank Buzard, a member of the NRO staff at the time, later noted that “the creation of the ExCom certainly tied the hands of the DNRO as far as new systems were concerned. In any case there was peace in the valley for a while after it was issued.”86 But that peace would not prevent intense competition between the CIA and Air Force elements of the NRO over the rights to build new generations of imagery and signals intelligence systems.
On September 9, 1965, with the new agreement in place, Wheelon sent a memorandum to Deputy DCI Richard Helms again requesting approval for the creation of an Office of Special Projects. Wheelon argued that in view of the August NRO agreement, which reaffirmed CIA responsibility as a participant and assigned to CIA definite program areas, it was time to implement the planned organization
. The office was established on September 15, and John Crowley, who had joined the agency about a year earlier as CORONA program manager, was chosen as the first chief of the new office.87
Also on September 15, Raborn designated Huntington D. Sheldon, a graduate of Eton and Yale College and former head of the Office of Current Intelligence, as Director of Reconnaissance, CIA. Sheldon was responsible to Wheelon for the activities of the Office of Special Projects (OSP) and related activities within the science and technology directorate. He was to provide the DNRO with a single point of contact with the CIA for all reconnaissance programs, and as a CIA history of OSP noted, “the assignment of Mr. Sheldon was in the nature of adding a diplomatic negotiator to balance the aggressiveness of the DS&T in handling NRP matters.” It was also, Wheelon recalled, a way of “stiff-arming the NRO”—highlighted by Sheldon’s title as “Director of Reconnaissance, CIA” rather than “Director, Program B, NRO.”88
THE NEXT GENERATION
On August 25, 1963, only weeks after Wheelon had assumed command of the Directorate of Science and Technology, the first of the KH-4A CORONA spacecraft blasted off from Vandenberg Air Force Base. The primary difference between the KH-4A and its immediate predecessor was not in terms of resolution—the new camera produced photographs with resolutions in the 9–25 foot range, a trivial improvement on the 10–25 foot resolution of its predecessor. Rather, the new cameras carried a greater film load—enough to fill two reentry vehicles. Missions could be extended to fifteen days in contrast to a maximum of seven for the KH-4, and a greater number of targets could be photographed.89
Fifteen KH-4A missions were flown through the end of 1964. In early 1965, Eugene Fubini and John Crowley agreed that studies should be made concerning the weaknesses and limitations of the KH-4A system. On June 29, after completion of the studies, formulation of recommendations, and a CIA–Air Force Office of Special Projects briefing, McMillan approved development of an improved version of CORONA—which would be launched in September 1967 as the KH-4B; it improved CORONA’s resolution to approximately six feet.90
By that time, Wheelon had completed his stay at the CIA and returned to private industry. In addition to RHYOLITE, his legacy included a technically demanding and ambitious program to develop a next-generation search system. Because the DS&T first began to explore the possibility of a follow-on system in fall 1963 and started early development work in 1964, the questions of what, if any, system to develop and whether the CIA should do the developing became a major issue in the NRO-CIA battles of 1963–1965. It would, in the words of one former NRO staffer, turn into a “real donny-brook.”91
In July 1964, McMillan had instructed the Itek Corporation to stop work on what was to be a follow-on to CORONA—the M-2—and to concentrate on improving the capability of the existing systems. His instructions followed a report from the Panel for Future Satellite Reconnaissance Operations, whose members included Edward Purcell, the chairman, Richard Garwin, Edwin Land, and NPIC director Arthur Lundahl.92
The panel had been briefed by various contractors and military and governmental personnel as to what systems were in design development or under consideration. James Reber, chairman of the interagency Committee on Overhead Reconnaissance (COMOR), which selected targets for the spy satellites, discussed the latest COMOR requirements. Lundahl covered the relationship between the resolution of an image and a pho-tointerpreter’s ability to extract intelligence from the photo. There was no difference, the CIA’s chief photointerpreter told them, in the intelligence that could be derived from photographs with resolution of ten feet than of five feet. The Purcell group concluded that development of a new search system with resolution between four and six feet was not justified.93
McMillan’s emphasis on improving CORONA was also based on the findings of the reconnaissance panel. Its report had noted that the KH-4 system operated at its ultimate photographic capability only about 10 percent of the time, partially as a result of recognized factors. The panel observed that “it seems entirely feasible to bring most of these factors under control so that one could count on peak resolution from the . . . system on 90% of the exposed film.” The consequence would be “an enormous gain in information acquisition.”94
But the vision of least one member of the Purcell panel went beyond five-foot resolution. Edwin Land concluded that a system was needed that covered as much territory as CORONA but with the resolution of GAMBIT—the Air Force’s new high-resolution satellite whose images had a resolution of eighteen inches. A study by Wheelon’s Systems Analysis Staff noted there were no plans for developing such a system.95
A high-resolution search system would address one of Wheelon’s concerns—that photointerpreters were “drowning in data” and finding it extraordinarily difficult to detect new items of interest with CORONA photography. They were failing to find the needles in the haystack. With too much to look at, they were “getting bleary-eyed.”96
In October 1963, Wheelon had established the Satellite Photography Working Group “to explore the whole range of engineering and physical limitations for satellite photography.” Chaired by Stanford physicist Sidney Drell, with ORD deputy director Robert Chapman serving as their CIA contact, the group was to look at possible ways to improve CORONA and set guidelines for the development of new systems. According to Wheelon, that effort marked the “resurgence of CIA activity in the satellite business.”97
The project was approved by McCone and Gilpatric during an October 22 meeting, and on November 5, a letter from Wheelon to McMillan provided a detailed outline of the group’s agenda and requested NRO funding. McMillan responded on November 18, noting that establishment of the working group “affords an excellent opportunity to achieve a more basic understanding of the reasons for the variations in quality and resolution we have experienced to date with the CORONA system,” although he discouraged an analysis of systems still under development. He agreed to provide NRO funds to cover the costs associated with outside consultants.98
Wheelon asked Drell’s group, which included Rod Scott of Perkin- Elmer and two representatives from Itek, to determine how much CORONA’s resolution could be improved and how much intelligence could be extracted from wide-area photos of higher and higher resolution. In an attempt to determine the intelligence value of increasingly sharp images, the group degraded aerial reconnaissance photographs to five different levels of resolution and gave them to photointerpreters at NPIC to see how much intelligence could be extracted at each level.99
Drell and his colleagues concluded that CORONA had been pushed about as far as it could be, and that to achieve significantly better resolution would require a new system. Meanwhile, NPIC’s photointerpreters demonstrated that a wide-area system with resolution of two feet would dramatically improve their ability to spot new facilities and extract intelligence about them. As a whole, the exercise resulted in the realization that, according to Wheelon, “something a lot better was needed.”100
But McMillan’s willingness to fund a research effort did not mean a willingness to fund a satellite to be developed on the basis of the group’s findings and recommendations. Wheelon later recalled that “from the outset McMillan did everything in his power” to stop that program, including refusing to provide funds.101 But McMillan’s decision did not prevent McCone from authorizing the use of CIA funds for the same project. With the Land Reconnaissance Panel and PFIAB suggesting that the expenditure of $10 million would be worthwhile to investigate the feasibility of a new wide-area, high-resolution system, McCone approved the funding.102
DS&T representatives talked to both Itek and Perkin-Elmer about the possibility of working on the program. In February 1964, using personnel from various offices and staffs within the science and technology directorate, the systems analysis group began a study in conjunction with Itek, whose ideas were preferred to those of Perkin-Elmer, to determine the feasibility and potential intelligence value of using several individual sensors or combination of sensors in
a satellite system. The study led to a camera design believed to be capable of producing high-resolution over a wide swath.103
It was not until June that McMillan discovered the CIA effort, code-named FULCRUM, which the agency had concealed not only from the Soviet Union but from the NRO as well. To pursue the project beyond the initial phase would require NRP funding. Wheelon initially proposed a six-month design effort. At the beginning, a project office of five to seven people, reporting directly to Wheelon, would be established within the CIA and be responsible for system engineering and technical direction. The proposal, according to an NRO history, was “precise, carefully detailed, seemingly quite accurate, technologically conservative, and—on the whole—exceptionally well constructed.”104
But McMillan believed that to approve the proposal would enable the CIA to establish “an independent capability for full-scale development of space systems,” even though their feasibility had yet to be determined. To establish such a capability, the CIA would have to recruit a substantial technical establishment. Not surprisingly, the NRO director was thoroughly opposed to the idea.105
McMillan also believed that he had Fubini’s support. The deputy director of defense research and engineering had observed that over the past two years, no committees had recommended a new search system. Also, Fubini had technical reservations about whether the high-speed film flow envisioned in the FULCRUM system was attainable. He also argued that proceeding toward a new broad-coverage system was unwise while the causes of CORONA’s variable performance remained unknown.106
McMillan attempted to head off any fait accompli by turning McNamara’s attention to the matter. With Fubini’s support and Vance’s approval, he submitted a McNamara-to-McCone memorandum for the Secretary’s signature, but in the end it was revised and signed by Vance. It proposed that the CIA be authorized to do only those tests needed to establish FULCRUM’s feasibility while the NRO simultaneously undertook comparative studies. By January 1965, Vance suggested, a determination of development desirability and a selection of a system should be possible. He added, “At that time we can discuss the assignment of responsibilities for development and operational employment.”107
The Wizards of Langley Page 17