A new interagency intelligence committee had recently been established under the supervision of Air Force Colonel Earl McFarland, called the Guided Missiles Intelligence Committee, or “Gimmick,” for short. Dulles had McFarland look into Lanphier’s claims. Over the next couple of weeks, GMIC hunted but found nothing. One reasonable hypothesis it came up with was that Lanphier’s sources in Air Force Intelligence were counting a lot of intermediate- and medium-range missile tests, in addition to ICBM tests. The U.S. had a radar in Turkey that looked out across the Black Sea toward the Caspian Sea. Both Soviet missile test ranges—one of which tested ICBMs, the other IRBMs and MRBMs—were within view of this radar.
In any event, another meeting was held with Dulles, Symington and Lanphier in mid-August. McFarland was also present and reported there was nothing to substantiate Colonel Lanphier’s report; that, as Dulles had told him on August 6, the Soviets had fired only six ICBMs, four of which were believed to have landed in the target area.
To Lanphier, it didn’t add up. If the Soviets were going to deploy 100 ICBMs by 1959 or 1960, much less 500 by 1961 or 1962, then they had to have fired more than six test missiles by August of 1958. There were lead times involved. In Convair’s experience, a missile had to be tested at least twenty times before it could be declared operationally ready and reliable; then there was the additional time it would take to transport the missiles to their bases, set up launching sites, command-control centers and so forth. If the testing data were correct, then the National Intelligence Estimate must be wrong.
Over the next several months, Dulles and his staff reached the conclusion, hesitantly but inexorably, that the estimate must indeed be wrong. There simply were not very many more Soviet ICBM tests being conducted. It was the dilemma of negative evidence again, but they had waited a long time now and the evidence was still negative. Yet, as in the bomber-gap period, the estimate they had was all there was. If the CIA and the ONE denied its validity, where would they find the data for a new one?
Moreover, over the past year, much more had been learned about the technology of missiles. The scientific analysts realized, to a much greater degree than before, how complicated it was to set up an operational missile site. Before, they had just considered the task to be one of building and deploying missiles; now they realized that the support equipment—the launchers, the communications system and the like—was much bulkier, more complex, more time-consuming to set up. They realized that even if the Soviets had a lot of missiles, they might not have so many of them on launchers, ready to be fired in the event of war.
Then there were the U-2 photographs that were coming back. The U-2 was a super-secret program. Outside the intelligence community, only slightly more than a handful of Pentagon, White House and State Department officials knew of its existence. Certainly nobody in Congress was aware of it. The plane flew at 80,000 feet, was “armed” with a very long range lens camera with remarkably good resolution (developed by Edwin Land, inventor of the Polaroid), and had been making spy sorties across the Russian border since 1956.
The interagency Guided Missiles Intelligence Committee had developed criteria on where to look for ICBMs: for example, it figured they would have to sit not very far off the tracks of the Soviet Union’s huge railroad systems, the only network that could move the missiles around. But, even with the U-2, there were some uncertainties. The plane had not yet been flown over all the area around the tracks. More particularly, it had not yet reached Plesetsk, in northern Russia, where the Soviets had been test firing (and perhaps getting ready to deploy) their ICBMs.
The end of the year was approaching; the negative evidence was still negative. The NIE for that year was delayed, deliberately, the analysts racking their brains, going over the data again and again, looking for something that might be interpreted as a positive sign of more ICBM testing, some ICBM deployment. But there was nothing.
Finally, on February 9, 1960, two months late, the NIE was released. It was a hodgepodge. It offered no consensus, and the bottoms of the pages were filled with dissenting footnotes signed by the intelligence agencies of the various services. The date by which the Soviets could have 500 ICBMs was pushed back to mid-1963, perhaps even further back than that. They would have only 50 ICBMs by mid-1960, only 35 of them on launchers. By mid-1961, they would have between 175 and 270 missiles, 140 to 200 of them on launchers. By mid-1962, they would have 325 to 400, with 250 to 350 on launchers. By mid-1963, 450 to 650, with 350 to 450—still fewer than 500—on launchers.
The differences in the numbers reflected the differences between the CIA, which picked the lower numbers, and the Air Force, which estimated the higher numbers. And in the footnotes were the Army and Navy intelligence services’ dissents, which—using the same data available to the CIA and Air Force Intelligence—arrived at still lower numbers. The Soviets, they said, would have only 50 missiles by mid-1961, only 125 by mid-1962 and 200 by mid-1963.
At this point, very few in the CIA or the Air Force were willing to take these extraordinarily low estimates seriously. For one thing, the politics of the situation seemed clear: the Army and the Navy competed against the Air Force for scarce budgetary resources; if the Soviets had only a few ICBMs in the works, that would deny the Air Force its chief rationale for building several thousand ICBMs and would, thus, leave more for the non-nuclear forces of the Army and the Navy.
Second, Navy Intelligence was automatically suspect. Keith Brewer, head of ONI, the Office of Naval Intelligence, had not believed the Soviets had set off an atomic bomb, and for many years after the fact. The Navy was always estimating, since that time, that the Soviets had only about one-fifth the fissionable material that the rest of the intelligence community was estimating. Brewer had worked at the Oak Ridge nuclear laboratory in Tennessee during the war, and simply could not believe that any other nation, especially the Soviet Union, had the collective brains and know-how to do what he and his associates had done.
There was a third reason why the Army-Navy numbers were rejected, and this was most critical. With the Air Force numbers, the Soviets could still damage or destroy most of the American SAC bases by mid-1961. The CIA numbers were only slightly less pessimistic, pushing the danger date back to late 1961. Whatever the fine differences, SAC still seemed dangerously vulnerable.
By this time, for all their earlier objections, top Air Force officers had come to accept the assumptions about SAC vulnerability laid out by the Wohlstetter-Hoffman R-290 report from RAND and by the Gaither Report. In the few years since, Air Force Chief of Staff General Tommy White and the new SAC Commander, General Tommy Power (LeMay left SAC around the time of the Gaither Committee and came to the Pentagon to become Vice Chief of Staff), had put in motion several programs on dispersal of bases and airborne alert of the bombers themselves, all with the purpose of reducing their vulnerability to attack.
Yet the Air Force was compelled to take these steps only after realizing it would be in its interest to do so. If the policy-makers were assuming that a certain percentage of the planes would get destroyed on the ground, that meant still more bombers for the Air Force—to allow for the attrition and still be able to fulfill the “military requirements.” And if SAC bombers were up in the air flying around all the time, that yielded two bonuses: higher morale for the pilots, who loved to fly, and a better chance of getting “modernized” bombers sooner, since already-deployed ones will be worn out much sooner. In short, some of the R-290 and Gaither recommendations provided perfect intellectual rationales for a more steadily funded and larger Strategic Air Command.