Chapter 8

Gemini Rising

The faith that sustained Project Gemini’s managers and workers through the dark days of 1963 was not misplaced. Even before the year was over, some of the hardest problems had begun to yield. Gemini’s prospects were far brighter by the spring of 1964 than they had been in the fall of 1963. There was still much work to be done, and not every effort at problem-solving was crowned with success. The project that stood on the verge of proving itself in the spring of 1964 was not the same project that had begun two years and more before, nor even the same project that emerged from the budget and managerial crises of late 1962 and early 1963. But most of what its founders had set out to prove had survived, and what had been lost could be balanced with what had been gained.

On 1 November 1963, “Program” replaced “Project” in the title of the office that directed Gemini. This change reflected its responsibility for the program as a whole, and not merely for the spacecraft. Since that had been true from the outset, the new name did no more than underwrite a reality that already existed. MSC Director Robert Gilruth announced it as part of a major reorganization designed to strengthen both Gemini and Apollo now that Mercury was over.* Mercury’s manager, Kenneth Kleinknecht, joined Gemini as deputy manager under Charles Mathews. Kleinknecht brought with him about a third of his former staff.¹

On the same day, 1 November 1963, an important realignment of NASA Headquarters also went into effect, and for much the same reason: Project Mercury’s demise was a chance to reassess the agency’s management structure. James Webb, Hugh Dryden, and Robert Seamans had become dissatisfied with the November 1961 reorganization. Headquarters had failed to secure the strong program direction over Apollo that Webb had wanted. When hardware development problems continued to mount, with attendant escalating costs and slipping flight schedules, something very definitely had to be done. Moreover, having a program the size of Apollo, along with all the other programs NASA was pursuing, made it difficult for one man - Seamans in this case - to serve as “general manager” over day-to-day affairs. In 1961, Webb had needed decision makers at the program level, but in 1963 he needed this talent, armed with the proper authority, at the administration level to unify the agency, provide direction to the field centers, and lessen some of the autonomy the latter had held onto so tightly. The major change involved putting the field centers under Headquarters “Associate Administrators” for special activities - George Mueller for Manned Space Flight, Homer Newell for Space Science and Applications, and Raymond L. Bisplinghoff for Advanced Research and Technology - rather than under the Associate Administrator as they had been. Mueller, who had replaced Brainerd Holmes as chief of manned space flight, now took charge of both the program and the centers carrying it out - MSC, Marshall Space Flight Center, and Launch Operations Center. Mueller also set up a Gemini Program Office in Washington,** chiefly as a device to oversee Gemini and to bring together in a single group all those in NASA Headquarters whose work related to Gemini. William Schneider had taken over a tiny liaison office of seven people from Colonel Daniel D. McKee earlier in the year. Now he headed a program office seven times that size. Several months would elapse before the effects were felt in Houston.² In the meantime, some of Gemini’s most severe technical problems were at last beginning to respond to hard work in the field.

Titan II Makes the Grade, But Not Paraglider

What had been Project Gemini’s greatest concern - whether Titan II could function as a booster for manned space flight - was soonest laid to rest. Titan II Missile N-25 was launched 1 November 1963 from the Atlantic Missile Range, the 23rd in the series of test flights conducted by the Air Force Ballistic Systems Division (BSD). It furnished the first real proof that Titan II would do for Gemini. Missile N-25 was equipped with the standpipes on its oxidizer lines and mechanical accumulators on its fuel lines that the revised theory had predicted would suppress the severe length-wise bouncing (Pogo) that threatened Titan II’s role as a manned booster. The November flight proved it worked. The devices installed in fuel and oxidizer feedlines reduced Pogo to the lowest level ever in a Titan II flight, only one-ninth the force of gravity (±0.11g), and for the first time well below the ±0.25g that NASA insisted marked the upper limit for pilot safety.³

The Gemini Program Office had no way of forecasting that the next five months were to see the Titan II test flight program produce an unbroken string of successes. But, knowing that standpipe and accumulator had worked on Missile N-25, GPO inferred that the theory behind installing these devices had been confirmed and acted quickly, sure that the Pogo problem had been solved. On 6 November, GPO decided to procure several sets of the suppression devices for Gemini launch vehicles. The soundness of that action was soon confirmed. Titan II launches on 12 December 1963 and 15 January 1964 both carried the oscillation dampers and both met NASA standards. The 15 January flight, added at Aerospace urging, proved the devices effective even with reduced fuel-tank pressures. This was all the more heartening because raised tank pressures had lowered Pogo levels in some earlier missile flights.⁴

While Titan II was proving itself in flight, NASA and the Air Force completed their nearly year-long efforts under the aegis of the Gemini Program Planning Board to fix standards for the Gemini launch vehicle. NASA’s final statement, on 15 November 1963, rehearsed its long-stated demands: longitudinal oscillations during powered flight must be no greater than ±0.25g, incipient combustion instability must be eliminated, and all known design shortcomings and anomalies revealed in Titan II ground and flight tests must be corrected. On the same day, BSD and SSD (Space Systems Division) of the Air Force Systems Command issued a plan to prove in flight their program to reduce Pogo and improve engines. These two documents, along with the earlier Air Force plan for cleaning up Titan II problems, answered the board’s request of 11 October 1963 for data on which to base a formal Memorandum of Understanding between NASA and the Air Force.⁵

What NASA required and how the Air Force planned to respond were discussed for the last time at the board meeting of 3 December. The board accepted the NASA specifications as reasonable, the Air Force plans to resolve the problems and verify the results as technically feasible. Then the co-chairmen of the board, Brockway McMillan for the Department of Defense and Robert Seamans for NASA, signed the formal “Memorandum of Understanding on Certain Design Requirements for the Gemini Launch Vehicle.”⁶ No further managerial obstacles blocked the way to a man-rated Gemini launch vehicle.

The compound of jurisdictional disputes and technological problems that had made the launch vehicle the single biggest question mark in the Gemini program until late in 1963 vanished almost overnight. By mid-January 1964, Titan II no longer seemed a concern. After the missile’s third success with Pogo suppression gear, on 15 January, Seamans was convinced “that the currently completed flight demonstrations of POGO fixes indicated a qualitative understanding of the problem and its solution and provided sufficient confidence to go ahead with the Gemini program.” Another sign of the times was the end of the weekly Titan II status reports Seamans had been getting from Air Force Systems Command because, “based on the successful resolution and flight verification of the axial oscillation fix (Pogo) on missiles N-25, -29, and -31, the primary requirement, for which this weekly report was originated, has been satisfied.”⁷

Pogo had not, of course, been the only problem, although it was the greatest. Still to be resolved was the potential instability of Titan II’s second-stage engine, which Aerojet-General had begun to tackle in October 1963 with Gemsip, the Gemini Stability Improvement Program, focused on working out a new design for the propellant injectors. Gemsip ended 18 months later with complete success, having cost the Air Force about $13 million. NASA spent $1.45 million to install the changes in the last six Gemini launch vehicles. The first six flew with the old-style injectors, which NASA later defended on the somewhat specious grounds that no instability had shown up in a Titan II flight. That was essentially a statistical argument of the kind earlier rejected as a basis for man-rating. NASA found a better reason for going on with the flight program. Aerojet engineers knew that any number of techniques might be used to reduce starting shocks, the major trigger for unstable burning. Very early in Gemsip, they found that a certain minimum pressure in the cartridges that started the motor eased the problem. Temperature conditioning - keeping the start-cartridge temperature above a critical value - proved even more effective. This was the finding that chiefly convinced NASA that Titan II’s second-stage engine was safe enough for manned missions, although only Aerojet’s redesigned injector finally provided a dynamically stable engine.⁸

NASA’s third concern about Titan II had been just how reliable some engine parts were. This was less a matter of design than of the general standards of manufacturing and quality control observed by Aerojet-General. The Air Force, however, saw potentially dangerous weaknesses in design that demanded the development of new parts, an effort that got under way in September 1963 as the Augmented Engine Improvement Program. NASA deemed improved engines nice but not vital (as damped Pogo and stable second-stage engines were) for Gemini. This was just as well, because the engine improvement program produced small results for the $11 million it cost the Air Force: some minor design shortcomings corrected, welding techniques improved, and better assembly methods adopted. NASA did buy one product of the program for Gemini, redundant shutdown circuitry, at a cost of $1.5 million. But the rest of the hardware developed under the program looked more risky than what it was intended to replace. The Air Force canceled the program in November 1964.⁹

Looking back, NASA officials had nothing but praise for the hard work put in by the Air Force and its contractors to man-rate Titan II for Gemini even while they were trying to prove it as a missile. As George Mueller reported to NASA Administrator James Webb:

In the broad view of this booster program where a military vehicle, the Titan II, was selected prior to its development and a program of man-rating carried out actually in parallel with the flight test and acceptance of the military versions, we have, I believe, a unique situation. It is unique not only in technical complexity but also in management relations and control. . . . [T]his collaboration between two demanding users has produced an unusually reliable military launch vehicle . . . [and] a man-rated launch vehicle with a remarkable record of success. . Configuration management is not a new term but the detailed application of the Air Force to the GLV [Gemini launch vehicle] development is a model of its kind and a significant contribution toward improved management of all major programs, in DOD and in NASA. We have seen major improvements in electrical circuit design, in electrical soldering and welding techniques, in assembly procedures and in test specification.¹⁰

This picture of a smoothly meshed team moving from success to success, although true enough for the last six months of the program, slighted the obstinate technical and managerial problems that had to be surmounted before the happy outcome was reached.

Even in retrospect, the record of Titan II research and development flights was spotty, especially in view of the high promise that had induced NASA to choose it for Gemini in the first place. Only 22 of the 32 flights that comprised the test program would have succeeded in launching a Gemini mission. Based on Titan II flight tests, in other words, every third or fourth Gemini mission would have been abortive; this does not include the Pogo that rattled missiles during first-stage flight without compromising Air Force test objectives. This picture was, nevertheless, far brighter than it had been in mid-1963 - half the 20 tests flown by 20 June would have been failures on Gemini. The concentration of all 10 unsuccessful flights in the earlier part of the program, however, may have held the greatest promise. The unbroken string of 12 nearly flawless flights that concluded the Titan II test program strongly implied that the missile’s problems had, in fact, been solved. With Pogo reduced to tolerable levels by techniques that accorded with theoretical analysis, the threat of combustion instability eased by an operational expedient, and a series of successes to show that other troublesome areas had been cleared up, Titan II could be judged man-rated in the early spring of 1964. This judgment seemed amply confirmed by Gemini-Titan 1, launched 8 April 1964,* the day before the last flight in the missile’s research and development test program and well before men were first scheduled to ride the Titan.¹¹

The striking vindication of Titan II in the final months of 1963 had no parallel in the paraglider program. Paraglider’s only chance to regain a place in Gemini hinged on the outcome of North American’s new series of deployment flight tests with the full-scale vehicle. A full-scale wing was to be uncased and inflated in midair, to prove it could support the vehicle in stable gliding and maneuvering under radio control. Each of the planned 20 tests was to end with the wing cut loose at 3,000 meters and the test vehicle landing by parachute. The parachute system was qualified on 3 December 1963, clearing the way for flight testing of the full-scale vehicle to begin on 22 January 1964. The first test did nothing to dispel doubts about paraglider; the second test, on 18 February, was also a failure.¹²

That same day, George Mueller told the House Subcommittee on Manned Space Flight that the paraglider “is not presently scheduled on the . . . Gemini spacecraft.”

"Will it be used at all in the Gemini program?” one of the Representatives wanted to know.

Mueller replied, “That will depend upon the development status of the paraglider which we will evaluate next spring. It will also depend upon the needs for a paraglider for precise landing of the Gemini spacecraft which we are developing now with the Air Force.”

Further probing revealed that paraglider could be ready for the tenth Gemini mission, particularly if the Department of Defense lent its support - this from George Low. But, he added, “we have no money included [in] 1965 or beyond for the paraglider under the assumption we will not go into production.”¹³

NASA’s public position was that, while land recovery appeared to be both desirable and feasible, it was riskier than water landing. Crew safety, the paramount concern, dictated the proven mode of safer landing for all 12 Gemini flights.¹⁴ The risks of land recovery were real enough, needless to say, but they had been just as real in 1961 when NASA decided to adopt land landing as a major Gemini objective. Toward the end of winter in early 1964, however, the means to that end, a paraglider landing system, had yet to achieve a level of performance great enough to rely on. After nearly three years of work, there was still no certain answer to the key paraglider problem - how to unship and inflate the wing from a two-tonne spacecraft plunging downward through the atmosphere. The risk that loomed so large early in 1964 was perhaps not so much land landing as paraglider landing.

Paraglider still had ardent defenders in NASA, and the decision to strike it from Gemini was not yet final.¹⁵ But NASA was ready to drop the paraglider, the more so since the system might still fly in another version of Gemini. In the spring of 1963, under the auspices of the Gemini Program Planning Board, the Air Force had begun laying the groundwork for its own Gemini program. Gemini B/Manned Orbital Laboratory (Gemini B/MOL). The Air Force X-20 orbital glider, still often called by its former name, Dyna-Soar, had been canceled in December 1963, a victim of low priorities and lagging development. Some X-20 funds were diverted to the new MOL program, which projected two men in a modified Gemini spacecraft launched by a Titan III. In orbit, the crew would transfer to a separately launched laboratory for two to four weeks, after which they would return in their spacecraft.¹⁶

Air Force planning had progressed far enough by January 1964 to require a formal agreement between NASA and the Air Force in the form of a memorandum signed by Seamans for NASA and Harold Brown for the Air Force.¹⁷ Although Gemini B/MOL would not be officially approved until August 1965 and design work was only beginning, NASA saw a chance to save paraglider. On 17 March 1964, George Mueller asked the Air Force for “an expression of the DOD interest in this capability,” whether for Gemini B/MOL or any other program. Six weeks later, having concluded that paraglider development had too many problems to warrant putting it in the new program, the Air Force discounted any prospect of joining in paraglider development and threw the problem back to NASA: “Should the NASA qualify and demonstrate the paraglider in the NASA Gemini program, consideration would be given to its application to the Gemini B/MOL.”¹⁸ By then, however, it was too late.

North American’s further efforts to fly the full-scale test vehicle produced a string of failures, each distinct in detail but united in a single root cause, “an inability to adequately predict the wing loads of flexible structure[s].” The fifth failure in a row, on 22 April, was the last straw. The next day, William Schneider, NASA Headquarters Gemini chief, informed George Mueller that he planned to transfer what was left of the paraglider program to Flight Research Center and to spend no more Gemini money. A week later, the program office in Houston began cutting back paraglider work and phasing the program out of Gemini. Early in May, GPO and North American agreed to run the rest of the flight-test program with the equipment and money already committed. Paraglider was dead as far as Gemini was concerned, although a public statement of its demise waited until 10 August.¹⁹

Ironically, North American achieved its first full-scale test vehicle success on 30 April, the day after phasing it out of Gemini began. In fact, the worst was over. Before the end of 1964, North American flew 19 more tests for a total of 25, 5 more than originally planned. By July, the deployment sequence was no longer giving much trouble, although a stable glide after the wing inflated was harder to manage. The last three flights, however, displayed the complete sequence without flaw.²⁰

The last full-scale test vehicle flight was on 1 December 1964. Two days later, NASA told North American there would be no more money for flight testing, but equipment on hand might be used, if the company cared to spend its own money. North American seized the chance to complete the other major portion of the May 1963 program - working out landing techniques with a piloted tow-test vehicle. Tow-testing had begun during the summer. On 29 July, a helicopter had towed the vehicle up to a height of a few hundred meters, around the test area, and back to a safe landing. A free flight followed on 7 August, but the vehicle went into a series of uncontrolled turns, forcing the pilot to bail out. North American attacked the problem with dispatch and came up with an altered wing design. On 19 December, a pilot flew the tow-test vehicle through the complete test to a safe landing.²¹

NASA had long since decided to dispense with paraglider for Gemini, however, and that was irrevocable.** ²² The system’s shortcomings, or at least North American’s slowness in coming up with answers, account chiefly for paraglider’s failure to survive in Gemini. But the immediate reason for the abrupt action in the last week of April 1964 to kill what remained of the Gemini paraglider may have had more to do with money than with technology.

Money and Management Problems Again

Gemini’s chronic budget ills were marked from time to time by acute episodes. The crisis of late 1962 had scarcely subsided before the project reeled under a new round of cost increases. By 8 March 1963, the program’s total price tag stood at just over $1 billion. NASA’s projected budget for fiscal year 1963 had been $232.8 million after the impact of reprogramming had been assessed; actual expenditures topped $289 million. The pattern repeated in fiscal year 1964, with a planned budget of $383.8 million exceeded by $35 million. By 2 March 1964, NASA expected to spend over $1.2 billion on the program.²³ These increases reflected, in part, Gemini’s changing scope and the technical problems that somehow proved harder to solve than anyone had expected. They also reflected, perhaps inevitably in so large and complex a program, mistakes, errors of judgment, and mismanagement, though Gemini appears to have suffered less from those ills than other programs of comparable size. Swelling costs were, for whatever reason, evident throughout the program.

NASA and McDonnell had finished negotiating the Gemini spacecraft contract in February 1963, settling on a total cost plus fixed fee of $456,650,062. This figure was not so firm as it then seemed. At the end of 1963, McDonnell estimated total spacecraft costs at upwards of $612 million. Something less than half the difference could be ascribed to approved changes in the program, as exemplified by the $2.7-million price for adding drogue stabilization to the parachute recovery system, though this change was itself prompted by development problems with reentry thrusters. Much of the balance derived from cost overruns on major Gemini subcontracts, with thrusters by Rocketdyne and fuel cells by General Electric the chief culprits. The new year brought no relief. In March 1964, when NASA estimated the total cost of Gemini at $1.2203 billion, the spacecraft accounted for $667.3 million.²⁴

Launch vehicle budgets were equally ephemeral. The billion-dollar estimate of March 1963 had included $240 million for the Gemini booster. As the year wore on, Air Force Space Systems Division found the situation “extremely fluid. Costs were constantly increasing and changes were being approved so fast it was difficult to keep track of them. . . . Engine problems were causing late deliveries and increasing costs.” When SSD completed its first comprehensive review of the Gemini budget in January 1964, it felt obliged to revise the cost upward to $296 million. Just two months later, after another hard look at launch vehicle costs, SSD claimed to need $324 million. This was the same month, March 1964, when NASA was counting the booster’s share of a $1.2-billion Gemini budget as $281 million. Toward the end of the month, Gilruth warned Major General Ben Funk, SSD Commander, that MSC’s 1964 booster money had been exhausted. With three months of the fiscal year still to go, the $46.9 million allotted looked as if it would fall $30 million short of expenses. Gilruth was much concerned about funding in the coming two years and asked Funk to take another look at his needs. Funk replied with an estimate of $332 million that included $75.3 million for fiscal year 1965, $8.4 million higher than NASA had planned.²⁵

Inexorably rising costs plagued target vehicle as well as launch vehicle development, and for much the same reasons: technical problems compounded by the fact that NASA and the Air Force simply did not agree on how a development program ought to be managed. NASA wanted more control than the Air Force thought wise to impose. NASA efforts to promote its view during late 1963 had availed little, and Mathews’ communications with SSD grew more caustic. On 5 February 1964, he scored Bell and Lockheed (and, by implication, SSD) for the sorry job being done on Agena engine development. Costs had “continued to increase even at this late date to a level far beyond that considered reasonable by this office.” The excuses offered were, in Mathews’ view, worthless:

The emphasis which BAC [Bell Aerosystems Company] has placed on the fact that the development effort was to be one of minimum cost has apparently led them to a belief that sound technical judgment was no longer required or that minimum cost eliminated its use. The GPO does not consider this argument valid or useful.

The fault was as much Lockheed’s as Bell’s. Mathews believed that -

the costs quoted by BAC and submitted by LMSC [Lockheed Missiles& Space Company] are excessive or unjustified in many areas. Moreover, these costs have increased and are continuing to increase with apparently little financial hazard to BAC and only after-the-fact recognition by LMSC. . . . GPO must express dissatisfaction with LMSC and BAC management of these programs.²⁶

Engine development costs were only part of the problem. The first “firm” budget for the Gemini Atlas-Agena program was ready in September 1963. SSD projected a total cost of $103 million, with Agena’s share as $57.5 million. By March 1964, NASA was prepared to spend $137 million for the program, $93 million on Agena alone. The $37 million programmed for Agena in fiscal year 1964 was almost exhausted, although that figure was $2.4 million higher than Lockheed had, in September, claimed to need. Mathews termed the situation “critical” and demanded a complete explanation in writing for the discrepancy between current costs and the September projections. GPO once again saw, in “the contractor’s frequent increases in the estimated costs,” signs of “a serious need for improvement by the contractor in proper planning and cost control.” Mathews warned SSD and Lockheed that “lack of adequate cost control places this program in real jeopardy.”²⁷

Ironically, at the same time that Mathews was urging SSD to get Lockheed under control, the contractor was finding that it needed still another $2.5 million in 1964 funds, a request that was duly passed along to GPO on 4 April. Lieutenant Colonel Mark E. Rivers, Jr., who had just replaced Major Charles Wurster as chief of Gemini-Agena engineering for SSD, saw signs of sloppy management in the new Lockheed request, which appeared to be based on small changes that had piled up unnoticed over several months.²⁸

This, then, was the setting in April 1964 when North American, for the fifth time in a row, failed to deploy the paraglider wing in flight. Mounting costs in all phases of Gemini development had stretched the 1964 budget to the breaking point, and the trend was still upward. Paraglider had been budgeted for $16.4 million in 1964, but that would be the last of the money. Keeping paraglider meant finding new funding or cutting back other parts of the program. In the money budget as in the weight budget, once paraglider’s status became doubtful, its place was preempted. Against this confluence of forces - technical, operational, and budgetary - paraglider could not stand.

Whether the target vehicle program could survive was also a question. In late April, budget pressures forced Mathews to discuss with his staff some desperate measures. Paraglider, Atlas-Agena, and even one of the planned Gemini missions were on the chopping block. Once again, however, MSC was able to reprogram funds to save the full 12-flight program and, via Agena, the rendezvous objective, if not paraglider and land landing.²⁹ One of the factors that may have made Agena’s place in Gemini shaky in April 1964 was a new round of technical problems that had cropped up earlier in the month.

Bell’s efforts to complete development testing of Gemini-Agena propulsion systems during 1963 had produced spotty results and many delays, which had, in turn, postponed the start of preliminary flight rating tests of these systems. Scheduled to begin in June 1963, testing of the main engine had been put off until January 1964 but began only on 6 February. Still another two weeks elapsed before the secondary system began its tests on 17 February. Both programs soon ran into trouble.³⁰

Main engine testing proceeded with only minor problems through the first week in April. In the following week, however, the test program encountered what proved to be a six-week delay when the test unit’s fuel and oxidizer start tanks failed. These tanks were stainless steel canisters with bellows inside them to push the propellants that started the main engine. Visible lengthwise cracks in their outer shells allowed the gas that was supposed to force the propellants from the tanks to escape. The steel in the shells had corroded. Tanks with a new heat-treated steel shell replaced the defective tanks, and testing resumed in May. But the tests, which should have ended in April, ran into late June. Alarmed by the threat of increased cost such a failure implied, GPO demanded a complete written account of the causes and effects, a point of special concern being “indications that subcontractors may have failed to process materials in a manner essential to the proper operation of components being developed.”³¹

Agena’s secondary propulsion system, like the main engine, started preliminary flight-rating tests smoothly, then ran into trouble early in April. Failure of a propellant valve, however, imposed only a minor delay. A harder problem emerged later in the month during high temperature firing, when the wall of a thrust chamber burned through after 354 seconds. While well beyond the 200 seconds regarded as the system’s longest useful life in orbit, it fell below the specified time of 400 seconds. Bell installed a new thrust chamber and finished the tests - in mid-August instead of the scheduled mid-June. The failure, however, needed to be explained, and that meant more tests. Bell planned a series of six tests over two weeks, beginning early in September. Test-cell problems hampered the work, which did not end until mid-November and then after only four tests. The four were, however, enough to spot the problem - elevated propellant temperatures - and to show that it would not affect the system’s performance in orbit.³²

Bell’s slow progress in its test program delayed Lockheed’s testing. Because of the scope of changes in propulsion systems required to adapt the standard Agena D for Gemini, Lockheed planned a series of static firings using an Agena skeleton fitted out with propulsion and propellant systems at its Santa Cruz Test Base in California. Lockheed received the propulsion systems from Bell in February and March and had the test assembly at Santa Cruz by the end of March. Checkout problems and Bell’s cracked start tanks in April held up the testing. Lockheed returned the main-engine start tanks to Bell, but they were not replaced until mid-May. Other minor problems delayed the first firing until 16 June. Once under way, however, the test program moved quickly to an end on 7 August 1964 with no further mishaps. Post-test analysis confirmed that the propulsion systems had come through in fine shape.³³

In the meantime, doubts about the Agena’s ability to perform its mission had been growing. On 15 April 1964, SSD suggested flying a non-rendezvous Gemini-Agena mission to bolster confidence. GPO dismissed this scheme but accepted an alternative recommendation that one target vehicle be assigned the role of development test vehicle. This would be helpful for troubleshooting malfunctions and testing changes and would also allow further development testing, should the need arise. The plan was approved in May and the first Gemini Agena target vehicle, GATV-5001, was to be the test vehicle. AD-71, the first standard Agena D for Project Gemini, had been accepted by the Air Force on 30 April and transferred to the final assembly area at the Lockheed plant, where it was being converted to GATV-5001. Despite its new role, GATV-5001 was expected to remain in flight status until GPO decided otherwise, although GATV-5002 was now tentatively scheduled for the first rendezvous mission. GATV-5001 was not likely to fly unless GPO later opted for a non-rendezvous mission. So GPO canceled one of the eight Atlas boosters then under contract as Agena launch vehicles, saving the program $2.15 million.³⁴

A Set of Breakthroughs

The three spacecraft systems that had caused the most trouble in 1963 - escape, fuel cell, and thruster each enjoyed a sharp change of fortune as the year turned. Problems that had resisted the best efforts of NASA and contractor engineers for so many months suddenly yielded. All the answers were not in yet, but by the spring of 1964 the prospect that any of these systems might fail to meet Gemini needs had largely vanished.

Escape system development trials had come to a halt in August 1963 as the system went through another series of design changes and some of its key parts, particularly pyrotechnics, remained hard to get. Active testing resumed on 22 November, with the first in a projected series of about 30 drops of the ballute, which had been added to the crew parachutes for the sake of high-altitude stability. The first 10 tests, which involved both men and dummies and used a ballute 91 centimeters (36 inches) in diameter attached by a single riser, ended on 9 January 1964. In each case, the subject spun too rapidly on the riser.* This was solved by raising the ballute diameter to 122 centimeters (48 inches) and using two-point suspension. Fourteen more drops over the next few weeks, the last on 5 February, confirmed the changes, and the ballute was ready for its qualification tests.³⁵

Only two days later, sled-ejection development trials also came to an end. Testing had resumed with the fourth run, on 16 January, and ended with the fifth, on 7 February. Everything worked in both tests. Since simulated off-the-pad and ballute development tests had already been completed, the successful 7 February test brought the development phase of escape-system testing to a close.³⁶ Neither fuel cell nor thruster was so far advanced.

Fuel-cell production had stopped in late November 1963, as a pair of GE task groups sought to resolve the system’s stubborn engineering and manufacturing problems. Within six weeks they had finished their work, which furnished the basis for turning the program around. Everyone involved in the fuel-cell program gathered in Houston on 27 January 1964 to review development status and decide what to do about it. All agreed that the system needed redesigning. The current PB2 model was to be discontinued; the units already built were to be used for limited testing and to be carried in Spacecraft 2 to gather data and help qualify the reactant supply system. All future cells were to be the new P3 design, and they were to be installed in every spacecraft beginning with the fifth.³⁷

Major changes in the new model reflected the narrow technical nature of the problems: dams (or baffles) were added to improve hydrogen distribution; the water collection wick was removed from each cell; and the orifice of the hydrogen feed tube of each unit was restricted so that any stoppage caused by water clogging could be cleared. Other changes included adding Teflon to the electrode to cut the loss of active material from the membrane and an anti-oxidant to the membrane to slow the rate of polystyrene breakdown. Tests had also suggested that the crucial problem of short operating life might respond to reduced temperatures. When further tests confirmed this finding, the coolant supplied to fuel cells was adjusted for lower temperatures.³⁸

Although fuel-cell problems were largely technical, GE decided the program could be better managed. It reorganized the Direct Energy Conversion Operation to work solely on the Gemini fuel-cell program. Roy Mushrush, the new manager, had a background as corporate troubleshooter for GE. He arrived on the scene with a blank check on the company’s resources for whatever help he needed. Mushrush was seconded by Frank T. O’Brien as Gemini manager. Both men impressed a NASA visitor with their enthusiasm, and morale throughout the plant remained high despite the shakeup.³⁹

The fuel-cell program was still a question mark, and no one could be fully certain that the system would be ready in time for Gemini. But in the early spring of 1964, the program’s technical and managerial problems seemed to have been taken in hand, and prospects were a good deal brighter than they had been. By the end of May, GE had finished switching to the P3 design and had started a broad test program.⁴⁰

Rocketdyne’s thruster development program was also turning a corner. So far, attempts to improve performance had been little more than stopgaps, centered chiefly on cutting the engines’ thermal load by dropping the ratio of oxidizer to fuel. But lower working temperatures and longer engine life were being achieved at the expense of combustion efficiency and specific impulse. This was one of three major topics discussed at a review of thruster problems in Houston on 23 December 1963. Rocketdyne was directed to cut the current oxidizer to fuel ratio of 1:1.3 still further, if that could be done without harm to good starting and stable burning.

Study of another expedient was also approved: shifting the sidefiring thrusters to align them more closely with the spacecraft center of gravity and so reduce demands on the smaller attitude thrusters in holding spacecraft attitude during lateral moves. Development of this small engine was the least hopeful aspect of thruster work - no one really understood what its design ought to include, and tests produced large and hard-to-explain variations. No attitude thruster had yet shown itself able to fire through a complete mission duty cycle without failure.

A third decision, of greatest impact on grew out of the 23 December review. André Meyer, chief of GPO administration, had been urging a change in the design of the ablation material lining the thrust chamber. A newly developed parallel-laminate material showed promise as an answer to thruster-life problems. Meyer wanted the laminates oriented nearly parallel to the motor housing, instead of perpendicular as before. His efforts to convince both McDonnell and Rocketdyne to make this change had been resisted because of its expense, but now, strongly backed by MSC Director Robert Gilruth, the idea was accepted and an engine to test the concept was ordered built.⁴¹

The thruster picture brightened perceptibly over the next month. Further tests confirmed that reduced oxidizer-to-fuel ratios prolonged engine life, bringing the maneuvering thrusters within sight of their required mission duty cycles. The performance of the smaller attitude thrusters also improved, though not as much. By mid-January 1964, NASA Headquarters felt sanguine about the prospects for Gemini’s big thrusters but saw little hope for so happy an outcome to the development of the smaller thrusters. There was strong support for a study of a radiation cooled engine as a backup.⁴²

Meanwhile, Rocketdyne’s efforts during the last two months of 1963 to work out the basic problems of small ablative engines had also borne fruit. A search through the files uncovered a research report on the problem of heat flux in small engines and an answer in the technique of “boundary-layer cooling.” The injector of a maneuvering thruster was modified to spray about a quarter of its fuel down the walls of the thrust chamber before firing. On 25 January 1964, Rocketdyne tested the engine through its full mission duty cycle without failure, its liner charring only to a depth of little more than a centimeter (one-half inch). A second thruster produced the same results. Since the lining of the flight weight engine was twice that thick, the margin seemed ample. Buoyed by these results, GPO, after a meeting at the McDonnell plant in St. Louis on 13-14 February, ordered McDonnell to have boundary-layer cooling designed into the larger thrusters in time for Spacecraft 5.⁴³

The smaller attitude thrusters did not respond as well to boundary-layer cooling, although it helped. A modified injector, combined with an oxidizer-to-fuel ratio of 0.7:1, allowed one small engine to survive a 570-second firing on 15 February with some of its liner intact; in earlier tests with the same ratio but without the injector, the liner had not lasted beyond 380 seconds. Two flight-weight engines with the new injector and lower ratio lasted for 435 and 543 seconds. Another change made these results look even better. Canting the lateral engines to direct the thrust vector closer to spacecraft center of gravity (as suggested at the 23 December meeting) was shown to reduce the thruster life needed to less than 400 seconds.⁴⁴

By mid-March 1964, thruster development and qualification appeared likely to be completed in time, though without much leeway to handle any new problems and with performance that was still marginal. In April, that status was transformed. Thrust chambers lined with laminated ablative material oriented almost parallel (at an angle of only 6 degrees) to the motor housing achieved dramatically better performance. The first modified attitude thruster endured 2,100 seconds of burning without failure on 14 April, a fourfold increase over the best prior test. And the next day, a maneuver thruster with boundary-layer cooling and the 6-degree wrap fired for 1,960 seconds, the test ending only when fuel was exhausted. Just as striking was the first test of a lateral thruster with the new wrap: 3,049 seconds of firing time without failure. George F. MacDougall, Jr., Deputy Manager of Program Control in GPO, reported the results to the MSC senior staff as “a major breakthrough.”⁴⁵

Convinced that the answer had been found, GPO lost no time. Within two days after the first tests of the small and large thrusters, McDonnell and Rocketdyne had orders to replace 90-degree with 6-degree wraps in all thrusters and to see that the new thrusters were’ installed in the orbital attitude and maneuvering systems of all spacecraft beginning with the fifth and in the reentry control systems of all spacecraft as soon as possible. By 1 May, however, Spacecraft 5 looked too early for a complete set of new engines. Instead, all its attitude thrusters would have the modified injector and 6-degree wrap, but only the aft-firing maneuvering engines would feature the new design. The less critical lateral- and radial-firing engines would be the old model. All thruster designs were now frozen, with further testing limited strictly to qualification.⁴⁶

Rocketdyne was by no means home free, but the worst of the spacecraft propulsion systems’ technical problems did appear to be over by the spring of 1964. The fuel cell also seemed to be in good shape. Gemini’s escape system, already through its development test program, may have looked best of all. As later events were to show, the promise was not quite that easy to fulfill. But none of these three most stubborn systems was slated for the first Gemini spacecraft which McDonnell had been building in its St. Louis plant.

Toward Gemini-Titan 1

The primary objective of the first Gemini mission, as it emerged from the revised flight program of April 1963, was to prove the Titan II able to launch the Gemini spacecraft and put it into orbit within the constraints imposed by manned space flight. To gather and report data were the spacecraft’s main functions. Spacecraft 1 was, therefore, unique among the products of the Gemini assembly line in St. Louis in being largely without standard spacecraft systems. For the most part, it carried dummy equipment and ballast to match normal weight, center of activity, and moment of inertia. Structurally, however, Spacecraft 1 differed from later models in only one important respect. Since mission plans did not call for the spacecraft to be recovered, the heat-shield simply completed the structure. Four large holes bored in the ablative material ensured the total destruction of the spacecraft when it plunged back into the atmosphere.

Working equipment was mounted on two special pallets (much like the “crewman simulator” used in Project Mercury) located where the crew would be in later flights. Spacecraft 1 carried two active Gemini systems: a C-band radar transponder and related gear to help ground radar keep track of the spacecraft, and three telemetry transmitters to return data to Earth. Data were to be gathered by a set of special instruments that measured pressure, vibration, acceleration, temperature, and structural loads.⁴⁷

McDonnell began testing Spacecraft 1 on 5 July 1963, with plans to have it at Cape Canaveral by mid-August. The first phase of spacecraft systems tests centered on making sure that each working piece of equipment functioned properly. Many parts did not, bringing testing to a halt on 21 July. The instrumentation pallets had several defects, especially in their electrical circuits and in their response to vibration. Other problems included a transmitter and a radar beacon that had to be returned to their makers to correct out of specification performance. With these matters taken care of, testing resumed on 5 August and proceeded smoothly to the end of the first phase on 21 August.⁴⁸

Four days later, McDonnell workmen mated the major spacecraft modules. The now fully assembled vehicle was ready for the second phase of systems tests, checking its overall working and the compatibility between the mated sections. It was now slated to arrive at the Cape on 20 September. During the first half of the month, tests alternated with leftover manufacturing tasks, which slowed things down, but not seriously. All systems performed well during the last half of the month, as the spacecraft was vibrated to simulate a launch, then transferred to the altitude chamber for simulated flight tests under orbital conditions. A complete integrated systems test on 30 September concluded the testing.⁴⁹

A good share of the program office and a sampling of the rest of NASA were on hand the next day to watch Spacecraft 1 as it rolled out of the test area in the McDonnell plant. Throughout the morning, McDonnell experts lectured their NASA guests on the spacecraft, the status of each of its parts, and the results of testing. After lunch, the NASA party retired behind closed doors to ponder the fate of the spacecraft. The McDonnell staff gathered late in the afternoon to hear the decision. Spacecraft 1 had been accepted for shipment to the Cape.⁵⁰

When it arrived on 4 October, it entered a new round of testing. GPO had decided early in the program that Gemini preflight checkout would conform to the Mercury pattern, even though the two-man spacecraft had been designed to render that kind of repeated testing unnecessary. Plans called for the spacecraft to be broken down to its major modules, each of which was retested to the subsystem level. After being put back together again and passing a series of integrated tests culminating in a simulated flight, the spacecraft was to be transferred from the industrial area to the launch complex.⁵¹

Spacecraft 1, lacking most of Gemini’s normal systems, was much easier to check out than later models; by the evening of 12 February 1964, the task was finished. The next step was a formal Preflight Readiness Review of spacecraft status, both physical and functional. Gemini Manager Charles Mathews and a team of engineers from Houston and Cape Kennedy* conducted the review on 18-19 February, finding nothing that would prevent the spacecraft from being moved to the launch complex nor that seemed likely to delay the launch.⁵²

The launch vehicle was not ready for mating, so Spacecraft 1 waited until 3 March before its transfer to complex 19. While the spacecraft waited, minor work continued, especially on the spacecraft shingles. These beryllium shingles were part of the heat protection structure and covered the external surfaces of the two forward modules - the rendezvous and recovery canister and the reentry control system. A fully acceptable fit was not, in fact, achieved until after the spacecraft had been mated to the launch vehicle.⁵³

Building and testing the first Gemini launch vehicle was not as easy as getting the spacecraft ready, because GLV-1 had the same role as the later boosters in the program. Just as McDonnell had been building spacecraft despite hard-to-resolve problems in some spacecraft systems, the Baltimore division of Martin- Marietta had been building launch vehicles for Gemini, even during the long months when the Air Force and its contractors were struggling to make Titan II reliable.**

Titan II was built around its propellant tanks, one for fuel and one for oxidizer in both the first and second stages. Martin’s Denver division, which held the missile contract, provided the tanks for Gemini boosters as well and shipped the set for GLV-1 to Baltimore in October 1962. After a lengthy series of tests, with special attention to welded joints to be sure they were both strong enough and leakproof, the tanks were ready for formal inspection in mid-February 1963.*** Only three passed. The second-stage oxidizer tank was cracked. It was returned to Denver and replaced by the tank intended for GLV-2, which reached Baltimore on 1 March.⁵⁴

By 21 May, the first Gemini launch vehicle was fully assembled and ready to begin testing as a unit. A check for wiring continuity revealed a short circuit in the second stage where a wire’s insulation had been cut through by a defective clamp. When inspectors found several other clamps with the same defect, every one of the more than 1,500 wiring-harness clamps in GLV-1 was removed, all wiring inspected, and a new set of clamps installed.⁵⁵

When electrical continuity had been confirmed, the first stage was erected in Martin’s new Vertical Test Facility on 2 June, the second a week later. This facility was a tower 56 meters high, adjoined to a three-story blockhouse fitted with test and checkout equipment, or AGE****, matching the AGE at complex 19 in Florida that would later ready GLV-1 for launch. The tower and blockhouse inside the Martin plant were designed to provide test data and to be compared with data gathered during checkout at the Cape.⁵⁶

The first phase of the test program, subsystem functional verification to make sure that each of the vehicle’s subsystems was working, began on 16 June. These tests went more slowly than planned. For one thing, the second stage had been late going up, partly because of electrical problems and partly because its engine arrived late. For another, minor troubles cropped up - hydraulic tubing that was not fully cleaned, solder flux that had boiled from a pinhole in a joint and gummed a gyroscope. By the end of June, subsystem testing had fallen about two weeks behind schedule, a source of concern but as yet no threat to the launch planned for December 1963. The functional verification tests lasted until late July, when a review of the data by SSD and the Aerospace Corporation found GLV-1 ready for the next phase of testing.⁵⁷

GLV-1 began combined systems tests on 31 July with a series of tests designed to uncover any interference between the vehicle’s several electrical and electronic systems. Five systems failed to meet standards after the first round of testing. Efforts to correct the problems - mainly by adding filters and grounds to Age and airborne circuits - produced results, though slowly. Only after the sixth test, on 5 September, was all interference cleared up. The launch vehicle’s last hurdle was a combined systems acceptance test (CSAT), which included a complete launch countdown, simulated engine start, liftoff, and flight, and ended with the simulated injection of the spacecraft into orbit. After several practice runs in conjunction with the electrical-electronic interference testing, Martin conducted the formal CSAT on 6 September, then presented both the data and the vehicle to the Air Force on 11 September for acceptance.⁵⁸

For the next week and a half, the Vehicle Acceptance Team, headed by SSD’s Colonel Richard Dineen, met at the Martin plant in Baltimore. SSD, NASA, and Aerospace inspectors explored the vehicle and studied its manufacturing and test records. This detailed inspection disclosed severe contamination of electrical connectors throughout, as well as a broken idler gear in the turbopump. These defects, plus the fact that 42 major components had yet to achieve documented flight status, forced the team to reject GLV-1. Failing to pass this type of inspection on the first try was not unusual, but it meant another long delay before GLV-1 reached the launch site.⁵⁹

SSD and Aerospace members of Dineen’s team also conducted a First Article Configuration Inspection (FACI) of GLV-1, with far more encouraging results. FACI had been a standard Air Force procedure since June 1962, a kind of audit of the actual product - as compared to engineering design - to provide a baseline for later products under the same contract. No SSD launch vehicle had ever made the grade on its first try, but GLV-1 did. Such defects as contaminated electrical connectors or broken gears, which barred its acceptance for Gemini, did not reflect discrepancies between design and product.⁶⁰

No sooner was the inspection over than Martin technicians began to set things right. Armed with magnifying glasses, they searched every one of the 350 electrical connectors aboard GLV-1 for traces of contamination and found 180 needing to be cleaned or replaced. All flight control equipment that had produced transient malfunctions during CSAT was removed and analyzed. Defective units were replaced and wiring harnesses reinstalled. At the same time, Martin tried to complete documentation of failure analyses and qualification of flight hardware. This extensive reworking of GLV-1 invalidated most of the earlier test results. Martin’s plan for an informal retest of problem areas only was rejected in favor of a full-scale repetition of CSAT. Subsystems testing and a preliminary acceptance test were finished by 2 October.⁶¹

The second formal acceptance test of GLV-1 ran on 4 October, uncovering little that needed to be corrected. Dineen’s team reconvened at Baltimore on 9 October and took only two days to complete its work and decide that GLV-1 could be shipped to the Cape. The team was scarcely enthusiastic about the vehicle. Much work remained to be done on GLV-1, but it could be done at the Cape, and there at least GLV-1 could be helping to check out the launch complex itself.⁶²

On 26 October 1963, GLV-1’s two stages, each strapped to an eight-wheeled trailer, were towed to the Martin Airport, next to the plant, and rolled through the rear loading door of a huge C-133B cargo aircraft provided by the Military Air Transport Service. A four-hour flight brought the two stages to Florida. Still on their trailers, they were rolled from the aircraft into the hands of Joseph M. Verlander’s Martin-Canaveral crew, who towed them to Hangar H to be unpacked, inspected, and fitted with the gear (such as lifting rings) required to erect them. There they remained, under guard, over the weekend. On Monday morning, 28 October, the trailer bearing the first stage reached complex 19.

At the launch complex, the Martin crew trundled the first stage up the long ramp to the launch vehicle erector, which rested on its side parallel to the deck of the test stand. The trailer rolled through the large door (the roof when the erector was standing) and stopped a meter and a half (five feet) from the other end. The crew secured the stage, removed the trailer, and closed the roof-door. A 150-horsepower electric motor then winched the 127-tonne (140-ton) erector upright, a process that took several hours. The trailer-borne second stage arrived at the launch pad a day later. Ordinarily, the next step was mounting the second stage on the first, but GLV-1 was slated for a special static firing test in mid-December, the sequenced compatibility firing of both stages. So stage II was placed in the second-stage erector, a smaller structure used only for checkout or static firings, and the two stages were cabled together. After checking to be sure there was no interference, Verlander’s team applied electrical power to the two stages standing side by side on 13 November.⁶³

Work at the Cape on GLV-1 was already a week behind schedule. Problems in Baltimore had pushed the launch date from December 1963 to February 1964. Another two-month delay now threatened. Mathews announced himself “greatly concerned with the present situation regarding the Gemini Program at the Atlantic Missile Range.” Four distinct groups - SSD, the Air Force’s 6555th Aerospace Test Wing (in charge of all Cape launches), Martin-Baltimore, and Martin-Canaveral - were testing and checking out the launch vehicle, with no formal understanding on how responsibilities were to be divided among them. Clarification was not long in coming; but meanwhile matters had become so confused that two distinct Launch Test Directives had surfaced. To make things worse, NASA people at the Cape complained about lack of access to technical data from the contractors. Poorly meshed working groups compounded other problems - a time-consuming review of the official work plan, procurement snags, and, most serious, questions of compatibility between booster and AGE - which extended the planned number of working days to get GLV-1 ready for launch from 86 to 118. By 22 November 1963, Mathews had to tell Seamans that even the already late 28 February 1964 launch date was likely to drop back to 1 April although GPO was working hard to improve the prospect.⁶⁴

In one move to help resolve management problems, Mathews united the several coordination panels that had been dealing with Titan II and related areas into a single Gemini Launch Vehicle Coordination Committee with six standing panels.# All panels were to meet at the same time every third week, then report to the parent committee, which would decide what action was to be taken. That should mean no more delays caused by uncertain authority, duplicated effort, or conflicting decisions.⁶⁵ Mathews and GPO launch vehicle manager Willis Mitchell also took steps to make good some of the time already lost. The Martin crew switched from two 8-hour to two 12-hour shifts a day. Checkout problems persisted, however, and the scheduled sequenced firing slipped from 20 December 1963 to 3 January 1964. Although a February launch of GLV-1 seemed out of the question, Mathews still hoped to launch by 17 March.⁶⁶

But the problems refused to end. The combined systems test scheduled for l3 December was twice postponed and finally completed on New Year’s Eve. Lack of compatibility between the booster and its support systems in complex 19, as well as a faulty turbopump assembly that had to be returned to Aerojet-General, were the major causes of delay. Next was the so-called wet mock simulated flight test, a complete countdown that included filling the propellant tanks; it was voided on 3 January by procedural errors after propellants had already been loaded. The test was called off two and a half hours before the simulated launch, although the count went on until T-30 (30 minutes before launch) to see if any other problems turned up and to give the operations crew some practice. Another try, on 7 January, was a success.

The countdown for sequenced compatibility firing was now set to begin, but a three and a half hour delay was imposed by contaminated oxidizer. Then, during the countdown, a malfunctioning first-stage propellant valve caused the test to be called off 20 minutes before firing. A second try, on 14 January, had to be canceled because unusually cool weather had chilled the engine start cartridges below the 275 kelvins (35°F) specified as the lower limit be Aerojet-General to prevent combustion instability. At last, on 21 January, the third attempt overcame some minor problems and delays to show the whole sequence of fueling, countdown, ignition and shutoff commands, guidance control, and telemetry. First-stage engines fired for 30 seconds and cut off. The second-stage ignited and fired for 30 seconds, halted by radio signal from the ground computer as in real flight. Sequenced compatibility firing proved that the engines delivered the required thrust and gimbaled properly. This static firing, the only one performed on a Gemini launch vehicle, met all prelaunch standards.⁶⁷

With static firing finally out of the way, the ground crew could now begin getting the booster ready for the spacecraft. That meant putting the second stage on top of the first, which was scheduled for 27 January. But post-firing cleanup found a defective rotor in one of the turbopump assemblies. Shipped to the West Coast for repair, it returned to the Cape on 29 January. Then a missing seal held up its reinstallation until 7 February.

The launch crew did not wait for the new seal; the turbopump assembly could be put back in the second stage after it was erected. On 31 January, they removed the stage from the small erector and secured it in the launch vehicle erector, which was then winched upright. The upper stage was gently lowered onto the first, and the two were bolted together. GLV-1 had assumed its final form. Before the spacecraft could be mated to the booster, there were still subsystem functional verification tests (like those done earlier in Baltimore) to be conducted. Although these tests were supposed to start on 14 February, lack of spare parts and questions about failure analyses imposed another week’s delay. Once testing began on 21 February, however, it went smoothly to verify the launch vehicle’s readiness for full systems testing by3 March.

On that day, Spacecraft 1 arrived at the launch complex to be installed in the spacecraft erector support assembly in a controlled-access “white room” atop the launch vehicle erector.⁶⁸

Tightening Launch Schedules

The revised flight program of April 1963 had projected the first manned mission, Gemini 3, for October 1964. But as 1964 approached, that prospect was dimming. The first Gemini flight was held up by the late delivery and protracted testing of its booster, and Spacecraft 2 was falling behind schedule at the McDonnell plant. Efforts to install spacecraft test and checkout equipment at the launch site in Florida moved slowly enough to suggest that time might be too short there as well. The already certain delay of the first mission, added to the all-too-likely chance that the second would also be late, made the prospects for launching Gemini 3 in 1964 look poor.⁶⁹

At a meeting on 13 November 1963, the Gemini Management Panel* decided that the program’s current schedule needed rethinking. The key question was just how much spacecraft and booster testing had to be repeated at the Cape to ensure a successful mission. Two panel members, MSC Gemini Program Manager Charles Mathews and Space Systems Division launch vehicle chief Richard Dineen, set up an ad hoc study of work plans and schedules aimed at seeing men in orbit via Gemini before the end of 1964. Mathews reported the findings to the panel at its next meeting, 13 December 1963. Gemini 3 could he launched in November 1964 by cutting down spacecraft testing at the Cape that merely repeated work already performed in St. Louis and by better integrating the entire checkout effort. Launch-vehicle testing was already fairly well meshed between Baltimore and the Cape and needed only to be smoothed out.⁷⁰

Spacecraft checkout procedures were altered sharply “to get a complete working spacecraft out of the McDonnell plant.” All testing in St. Louis, along with whatever manufacturing tasks were left after systems testing began, was to be modeled on Cape practice. This meant that the McDonnell test crew had to be retrained. John J. Williams, Assistant Manager for Gemini of MSC Florida Operations,** took a Launch Preparation Group of 200 people, drawn from both NASA and McDonnell, to spend nearly nine months in St. Louis. They thoroughly revamped the testing process, training the St. Louis crew and actually checking out the second and third Gemini spacecraft. About half the group returned to the Cape with Spacecraft 2 in September 1964, and the rest stayed until Spacecraft 3 was ready in January 1965. The retrained McDonnell crew took over when Spacecraft 4 began systems testing. Basic to the new process was cutting down on repeated testing. Once a subsystem had been tested, it would take its proper place in the spacecraft and stay there. No longer was the spacecraft to be taken apart after it reached the Cape, tested, and put together again. Systems were to be rechecked, of course, but only as part of the complete spacecraft, not as individual pieces.⁷¹

The booster offered fewer problems in meeting Gemini schedules. Aside from efforts to speed up work on GLV-1, already at the Cape, the only major step was to strike flight readiness firing from the test program planned for the first three launch vehicles. With spacecraft checkout streamlined and booster testing smoothed out, GPO looked forward to getting back in step with the April 1963 schedule, even though the first flight was now going to be about three months late. The eight months that had been allowed between the first two flights was cut to five, with Gemini 2 only a month behind schedule, in August instead of July 1964. By then keeping to the three months between later flights, the first manned mission could be launched in November, a month late, but still in 1964.⁷²