Chapter 5

Expansion and Crisis

As summer gave way to fall in 1962, the smooth progress that Project Gemini had enjoyed during its first half year roughened. Concern mounted over the steady expansion and rising costs of the project as a whole. Hopes for using much of Mercury’s technology in Gemini eroded. One system after another became the subject of full-scale development, rather than modification or simple transfer from Mercury. The scope of launch vehicle development likewise grew far beyond first expectations. Costs kept climbing until they collided with an unexpectedly restricted budget toward the year’s end.

These concerns were virtually unknown outside NASA. But the striking dual mission launched by the Soviet Union in August led some to wonder if the United States had any hope of flying the first rendezvous mission. Vostok III, piloted by Major A. G. Nikolayev, lifted off on 11 August, followed a day later by Lieutenant Colonel P. R. Popovich in Vostok IV. The two spacecraft came close enough to each other to spur some talk of rendezvous. With no means of maneuvering their spacecraft, however, the two cosmonauts could not match orbits or speeds. The Soviet Union had shown only that it could launch two spacecraft in quick succession, so there was still hope for the maneuverable Gemini spacecraft to achieve the first rendezvous, if it survived its troubles.¹

Changing Plans and Rising Costs

Preliminary cost estimates from Gemini contractors began reaching the Gemini Project Office in March 1962. These rough figures pointed toward a large but not yet clearly defined increase in the projected total cost of the program. Air Force Space Systems Division (SSD) discussed finances with the project office at the first launch vehicle coordination meeting on 1 March and furnished its first budget estimate for the program at a meeting in Houston later that month. Boosters now appeared likely to cost Project Gemini a good deal more than had been supposed. The development plan of December 1961 had assumed $113 million for modified Titan II launch vehicles. But the March 1962 figure was half again as much - something over $164 million.²

The statement of work for Titan II that SSD had received early in January called for more than the limited modifications first proposed. It required a malfunction detection system and other unspecified changes to improve the missile. Making sure that Titan II could safely launch manned spacecraft - referred to as manrating - was crucial, and it was going to cost money. A revised statement of work in mid-May 1962 spoke of “an adaptation of the Titan II ICBM,” rather than “a development of the present Titan II ICBM,” and spelled out the changes required in greater detail. They included not only a fully redundant malfunction detection system but also a backup flight control system; an electrical system with backup circuits for guidance, engine shutdown, and staging; inertial instead of radio guidance; and a new launch tracking system.³

The target vehicle likewise soon seemed to demand more than had first been expected. Even though Agena work was moving at a slower pace, by May the $88 million programmed for Atlas-Agena development in the December 1961 plan had climbed above $106 million.⁴

The project development plan had the Gemini spacecraft costing $240.5 million. This figure, like those for launch and target vehicles, could not have been more than an educated guess, with a natural bias toward guessing on the low side to make the program more palatable. But McDonnell’s first formal cost proposal for the Gemini spacecraft still came as something of a shock. The first step in negotiations between the project office and McDonnell to convert the letter contract of December 1961 into a definitive contract was a series of technical meetings in Houston between 19 April and 24 May 1962, to make sure that both sides agreed on plans and specifications.⁵ McDonnell’s “Gemini Spacecraft Cost and Delivery Proposal,” prepared for these meetings, raised the spacecraft ante to $391.6 million.⁶

This new and higher estimate was based in part on McDonnell’s more careful study of the cost of what the contract called for, in part on its enlarged view of what the program ought to include. The letter contract, for example, had mentioned the need for flight simulators and trainers as well as test spacecraft but included no specifics. A new feature of engineering development for Gemini was to be the use of a number of test articles - spacecraft built for early static and dynamic testing - for want of which Mercury had sometimes been delayed. GPO admitted that building them might slow spacecraft construction at first but believed that the data they provided would more than make up for the temporary setback.⁷ McDonnell proposed four boilerplate spacecraft (metal models designed to be used chiefly in escape and recovery system testing) and four static articles (non-flying spacecraft to be used in structural tests). McDonnell also proposed to add to Gemini a test program that it had worked out in Mercury. Known as “Project Orbit,” this entailed building an extra spacecraft and target docking adapter for an extended series of laboratory-simulated orbital missions “to investigate potential problems and to evaluate engineering changes generated during the life of the program.”⁸

A major part of crew training for Gemini depended on simulating in great detail every aspect of a mission, to expose the astronauts before they left the ground to anything they might meet during a flight. The basic device was a flight trainer, a precise duplicate of the real spacecraft, in which crews could fly a complete simulated mission from launch through touchdown, seeing through its windows what they would see in flight, hearing the noises - even feeling the vibrations - they could expect. There were to be two flight trainers, one in Houston and the other at Cape Canaveral, each hooked up to mission control and remote displays to form a complete mission simulator.

Three aspects of a mission were outside the scope of the flight trainers. One involved the forces imposed upon the astronauts by high acceleration during launch and by rapid deceleration during reentry. These stresses could be matched on a man-carrying centrifuge. The project office planned to use the one at the Naval Air Development Center in Johnsville, Pennsylvania, its gondola fitted out with a mockup of the inside of the spacecraft. Maneuvering in orbit to rendezvous was the second aspect. This was to be simulated by a translation and docking trainer, in which the crews would practice techniques of rendezvous and docking.⁹ The third, extended weightlessness, was then beyond human ingenuity to imitate.

Training equipment and test articles together, increased in detail and enlarged in scope, came to just under $39 million in McDonnell’s cost proposal. McDonnell also needed money to cover its roles in mission planning and launch operations support and for spare parts and checkout gear, to name only some of the more costly items. And all this aside from the expense of developing and building the spacecraft ($242.7 million), which alone exceeded the December budget ($240.5 million).¹⁰ Even at that, McDonnell’s estimate was still little more than guesswork. Few of the company’s subcontractors had yet provided any hard financial data. The chiefs of procurement and financial management at MSC jointly deplored both the size of the McDonnell estimate and the lack of data on which it was based, a viewpoint that echoed Paul Purser’s marginal note on SSD’s interim financial plan for boosters in April 1962: “This is still up in the air. Attempts are being made to bring down these costs.” ¹¹

On 12 May 1962, in a review of Project Gemini for NASA Administrator James Webb, the Office of Manned Space Flight revealed for the first time the pattern of rising costs that was beginning to mark the program. Since the project development plan was issued, little more than five months earlier, Gemini’s expected cost had climbed from $529.5 million to $744.3 million.¹² Given the shaky data on which the new total depended, it could not be the last word. The program kept growing and technical problems began to appear, not all of them in areas where they had been expected.

Some Foreseeable Problems and a Surprise

As Project Gemini moved from design into testing during the spring and summer of 1962, problems multiplied, although not (with one exception) beyond what might be seen as the normal headaches of a large-scale research and development project. Those areas that demanded the longest step beyond current practice were those where trouble threatened. The paraglider program, with its early start, began running into marked delays in planning and design before the rest of Project Gemini. When actual testing began in May 1962, only two contract months remained to settle on the best design for a paraglider landing system.

The first task was qualifying an emergency parachute recovery system for the half-scale vehicle. North American began on 24 May with a successful drop test at the Naval Parachute Facility in El Centro, California, near the Mexican border. Two failures followed before a second success, on 20 June. What should have been the final drop to complete qualification failed on 26 June, when the vehicle’s electrical system shortcircuited. North American shuttled the vehicle 260 kilometers back to its plant in Downey for a closer look, which revealed a design flaw. The company reworked the test vehicle and returned it to El Centro for another try, on 10 July, with no better luck. This time the drogue designed to pull out the main parachute failed to do so. After another round trip to Downey for changes, everything worked on 4 September. GPO agreed with North American that the half-scale emergency landing system was now qualified. But two and a half months had been lost.

The full-scale emergency system proved even harder to qualify. First came design problems, then the parachutes were late in arriving. North American could not ship the test capsule to El Centro until 20 July. The Air Force’s 6511th Test Group, which ran the El Centro test range, demanded a special test to be certain the vehicle’s pyrotechnic devices were safe - that delayed the first qualification flight until 2 August. It was a success, but more delays followed - first bad weather, then the lack of a launch aircraft. The second drop, on 21 August, was marred by one of the three main parachutes breaking loose. Damage was only minor, as it was in the next test, on 7 September, when two parachutes failed. Efforts to correct this problem took over two months. On 15 November, some four months after the full-scale emergency recovery system was supposed to have been qualified, the fourth drop was a disaster. When all three parachutes failed, the test vehicle was destroyed as it hit the ground. Clearly the system could not be relied upon. GPO directed McDonnell to furnish North American with a boilerplate spacecraft for further tests at some later date.¹³

These problems, however disheartening, should not have cast any shadow on the concept of a paraglider. The emergency parachute systems were intended only to back up testing; they were not part of the Gemini landing system. Yet the pattern of delays, errors, and malfunctions that marked North American’s efforts to qualify the emergency system proved to be symptomatic of a lingering malaise. Paraglider advocates knew that the program would be made or broken, so far as Gemini was concerned, by the success or failure of flight testing, and time was limited. North American had been chosen over Ryan and Goodyear because of its first-rate job in testing the design during the summer of 1961.¹⁴ But on 28 November, scarcely a week after North American received word to go ahead with paraglider development, NASA notified the company that it had been selected as prime contractor for the Apollo spacecraft. The impact on paraglider was catastrophic. North American froze the number of engineers assigned to paraglider, then allowed even that group to decline. The quality of work suffered as well, becoming, in the opinion of one NASA engineer assigned to the program, “abysmal.”¹⁵

The pattern of trouble sketched in emergency system testing persisted when North American began testing the paraglider itself by flying half-scale models with wings inflated and deployed before they left the ground. Scheduled to begin in May 1962, these trials got under way in mid-August at Edwards Air Force Base, 100 kilometers north of Downey. North American’s first try, on 14 August, got nowhere. Because a plug pulled loose inside the capsule, the wing, which was tied down for takeoff, failed to release after a helicopter had towed it to the proper height. The wing released too soon in the second try, three days later, although the capsule did go brief it into a stable glide. North American also achieved a stable glide in the third flight, on 23 August, but an erroneous radio command caused the vehicle to come down too fast and suffer some damage in landing. The fourth flight was postponed twice, each time because someone forgot to charge the battery. Towed aloft on 17 September, the vehicle failed to release on command, voiding the test. Twice in a row, short circuits forced the contractor to call off the fifth flight test, the second time on 21 September.¹⁶

That same day, James Chamberlin, MSC Gemini Project Manager, ordered North American to halt flight tests of the half-scale paraglider. He expressed “growing concern” over “the repeated unsuccessful attempts of S&ID [North American’s Space and Information Systems Division] to conduct satisfactory predeployed half-scale paraglider tests.” Flights were not to resume until the contractor had reorganized its paraglider project and could spell out just what it intended to do about the test vehicle’s electronics and pyrotechnics and the company’s own checkout and inspection procedures.¹⁷

North American had already made some moves along the lines Chamberlin demanded. The paraglider effort was raised to the status of a major program, and George W. Jeffs was named Paraglider Program Manager on 1 September 1962. Norbert Witte, the former project manager, stayed on as Jeffs’ assistant.¹⁸ Jeffs was something of a corporate troubleshooter, and he had the respect of the NASA engineers working on paraglider.¹⁹ This augured well for the future, but, in the meantime, a fully successful flight test had yet to be performed.

North American reworked the half-scale vehicle in its plant, then shipped it back to Edwards Air Force Base on 15 October for another try. A bad ground transmitter stalled matters for a while but, on 23 October, the fifth test flight was a complete success.²⁰ Even with all its problems, the series of tests had met its main goal, showing that the paraglider was stable in free flight.²¹ But predeployed flight testing ended more than two months late, and the crucial deployment flight tests - spreading the paraglider wing in flight - had not even begun.

In the meantime, other problems were beginning to compete for the attention of the overworked project office. Like the paraglider, ejection seats had been a controversial innovation in manned spacecraft, and their development problems also gave critics an early opening. The reasons were much the same. Both systems were a long step beyond current practice, both presented test problems not clearly related to their final roles, and both were subject to changing requirements that imposed makeshift adjustments, further complicating matters.

Although ejection seats were widely used in military aircraft, they were designed to give pilots a chance to survive, not to guarantee that survival. Manned spacecraft levied more stringent demands. Most critical was the “off-the-pad abort mode.” Before liftoff, the spacecraft perched some 45 meters from the ground atop a shell filled with potentially explosive chemicals, the Titan launch vehicle. However rigorous the precautions, there was always the danger of some mischance setting it off. For a length of time that might stretch into hours before they were airborne, the crew would be aboard with no recourse, should that mishap occur, save their ejection seats. The Gemini seat had to be able to propel itself from a starting point 45 meters in the air in a trajectory stable enough to get clear of an exploding booster and high enough to allow parachutes to open. No existing seat could do that, and developing one that could was the crux of the Gemini effort.²²

McDonnell chose Rocket Power, Inc., of Mesa, Arizona, to supply the rocket catapult (or rocat) for the Gemini escape system.²³ For the seat itself, McDonnell turned to Weber Aircraft, of Burbank, California.²⁴ As luck would have it, the Naval Ordnance Test Station at China Lake in the middle of California’s Mojave Desert had earlier constructed a 45-meter tower for Sidewinder missile tests. This tower was admirably suited for simulated off-the-pad ejection (or, acronymically, Sope) tests.²⁵ Kenneth F. Hecht, who left the ordnance test station in January 1962 to take charge of Gemini escape and recovery systems, set up a special working group to oversee seat development and qualification.* He was convinced, and in this he was seconded by those who knew most about ejection seats, that the key problem was finding ways to control the relationship between the rocat’s line of thrust and the shifting center of gravity of the seat-man combination while the rocket was burning. Without this control, a trajectory of the proper height and stability could not be achieved. This was one of the reasons why Hecht insisted the tests be conducted with a dummy in the seat, rather than with a solid mass. He also knew that haste was vital, since the seat design could not be settled until the answers were in.²⁶

The first Sope test came off on schedule 2 July 1962, followed by four more over the next month. All produced their share of problems and mechanical failures, each dealt with as quickly as possible to get on with the next test. None of these mechanical problems much bothered Hecht and his colleagues, because they had their eyes on the dynamic problem of rocket thrust and center of gravity. They were concerned with ejection at this point, not the complete escape sequence through recovery, and thought they were close to solving that key problem.²⁷ From this viewpoint, the first five tests were a success. But if the goal were seen as a complete system with all parts working as they should in the final version, the tests left much to be desired. The seat seemed to be turning into a maze of makeshift fixes, and the personnel recovery parachute system (the crewman’s landing device) had failed twice.

At an extended meeting in Houston on 6 and 7 August, the total system viewpoint prevailed. Sope testing was halted until a complete design of the whole system was ready and the personnel parachute had been fully tested.²⁸

A month elapsed before McDonnell was able to report on 6 September that seat design and testing were complete, clearing the way for a new round of Sope trials. Tests on 12 and 26 September went well but highlighted a set of problems with the rocket motor. Some were functional and some structural, but all affected, however slightly, the direction of thrust and so made accurate control impossible. Testing stopped again, pending the availability of the rocat in its final form.²⁹ This delay was much prolonged, lasting well into 1963.

Other major Gemini systems seemed less troublesome. Through the summer and early fall of 1962, such problems as appeared could be, and were, regarded as nothing more than the routine hurdles in a large program. One possible exception was the fuel cell, which, like paraglider and ejection seats, was new to manned spacecraft and had aroused some debate, at least in its General Electric version.

The basic source of electrical power in the spacecraft was to be batteries. The weight of ordinary batteries, however, became prohibitive as missions increased in length. Something more was needed, and the choice was fuel cells. That choice was resolved in January 1962. After analyzing the merits and defects of competing approaches, Robert Cohen of MSC strongly recommended the General Electric fuel cell as lighter, simpler, and more generally suited to Gemini needs than other designs he had investigated.³⁰

In a fuel cell, hydrogen and oxygen react to produce water and heat. The unique feature of the General Electric design was its use of a solid ion-exchanging membrane in which electrolyte and water were chemically bound; most other cells diffused gases into a liquid electrolyte. A separate stream of coolant condensed the water produced at the cell, then removed it through a series of wicks to keep the reaction going at a constant rate. This used little of the cell’s own power, in contrast to the gas-diffusion cells that required a complex self-powered process of flushing with hydrogen, condensation, and centrifuging to remove the water produced. General Electric had devoted intense research to the design since 1959 and had already set up a fuel-cell facility, the Direct Energy Conversion Operation in West Lynn, Massachusetts.³¹ McDonnell shared Cohen’s view and formally recommended General Electric for a subcontract, to which NASA agreed.³²

Nonetheless, in early 1962 the General Electric fuel cell was still no more than a laboratory device, however promising.³³ NASA Headquarters was looking into fuel cells for Apollo, which raised some questions about Gemini’s choice of General Electric. The Office of Manned Space Flight’s survey of General Electric alleged that the company was understaffed, slow in getting started, and unlikely to meet Gemini schedules - all this in addition to what seemed to be an untested and questionable design concept.³⁴ Cohen responded to these charges for GPO. He saw no reason to doubt that General Electric would meet its commitments: the company was adding to its staff and improving its effort, which had only begun with an order from McDonnell two and a half weeks earlier. More important, the much tested General Electric design was at least as far along as any other and was inherently simpler to boot.³⁵ That settled the issue.

As development got under way, General Electric began to run into problems that seemed to suggest that theory had outpaced practice. The most serious in mid-1962 was how to achieve a satisfactory bond between cell membrane and frame. Solving these problems appeared more likely to tighten the schedules than to threaten the program as a whole. In any case, the worst appeared to be over by the end of August.³⁶

During the last half of 1962, the paraglider’s troubles probably posed the greatest threat to an approved Gemini objective, that of land landing, although ejection seats and, to a lesser extent, fuel cells were also worrisome. The paraglider was a major new system that demanded a large-scale effort. Ejection seats and fuel cells, though not so novel, were still major innovations in manned space flight. In all three cases, the novelty of the application and the advance beyond current practice imposed a greater development effort than required for other Gemini systems. Given that fact, the problems should have come as no surprise. A quite unexpected source of trouble loomed in another quarter. The suitability of Titan II as a launch vehicle for manned space flight came into question.

Responsibility for developing the Titan II missile belonged to the Ballistic Systems Division (BSD), like SSD a part of Air Force Systems Command. Titan II research and development test flights began on 16 March 1962, with a launch from the Atlantic Missile Range in Florida. In its first flight, Titan II displayed a disquieting characteristic. A minute and a half after it lifted off, while the first-stage engine was still firing, the missile began to vibrate lengthwise like an accordion about 11 times a second for roughly 30 seconds. This was not likely to bother a missile too much, but it implied real trouble for a launch vehicle with a manned payload. The steady acceleration of a booster like Titan II pressed a crewman to his couch with about two and a half times the force of gravity at that point in a normal flight. Bouncing at an extra two and a half gravities (+ 2.5g) could badly hamper a pilot’s efforts to respond to an emergency, a matter of special concern in Gemini since the crew played so large a role in flying the spacecraft.³⁷

Titan II’s longitudinal oscillations quickly acquired the nickname “pogo stick,” soon simply Pogo. Its cause remained unclear, how to get rid of it a matter of guesswork. By July, Pogo was becoming a regular topic at MSC’s weekly senior staff meetings, and BSD had formed a special Committee for Investigation of Missile Oscillations.** ³⁸ The problem turned out to be surprisingly easy to solve for the missile: higher pressure in the first-stage fuel tank cut Pogo in half during the fourth test flight, on 25 July, although nobody was quite sure why.³⁹

There were some ideas, however. Martin engineers thought the culprit might be oscillating pressure in propellant feedlines, analogous to the chugging of water in pipes, or “water hammer.” This suggested the use of something like the surge tanks familiar as devices to stabilize pressure in the flow lines of hydroelectric plants and pumping stations. Martin proposed to install a surge-suppression standpipe in the oxidizer line of a later Titan II. MSC endorsed the plan, and BSD agreed. By the end of August, GPO was cautiously optimistic. The lowered Pogo level of plus or minus 1.25g achieved in the fourth Titan II test flight was still too high for manned space flight, but the water hammer analogy at least suggested an answer.⁴⁰

GPO was also watching another problem. In two of its first four test flights, Titan II’s second-stage engine failed to reach full thrust. The causes appeared to be different in each case and unrelated to one another. Just how serious this might be could not be foreseen. Much depended upon whether or not it recurred, and GPO adopted a wait-and-see stance.⁴¹

Project Gemini’s technical problems in the summer and fall of 1962 might have aroused more concern if a far more serious threat had not intruded. The financial structure of the program began to totter. Two circumstances combined to produce a major crisis. On one hand, Gemini contractors were spending money at a much faster rate than the project office had expected. On the other, Congress was slow to approve NASA’s appropriation for fiscal year 1963, which restricted the funds available to Gemini. However serious development problems might be, or become, they could always be resolved if there were enough money. But now the question was how to spread limited funds over an ever more costly program.

The Budget Crisis

The pattern of program growth and cost increase revealed during the spring of 1962 persisted, and with the same shortage of dependable information. To NASA’s repeated pleas for more funding data, McDonnell regularly denied that any existed. In mid-July 1962, three months after its first budget proposal, the company could still not provide a detailed forecast of program costs because “cost projections from suppliers and subcontractors are currently unavailable as purchase order values continue to change and negotiated costs have not been established.”⁴² In August, when MSC and McDonnell began working out the final terms of the spacecraft contract, the contractor proposed a startling total of $498.8 million, double NASA’s first estimate in December 1961 and more than $100 million higher than the company’s own April 1962 proposal.⁴³ Hard negotiation brought the new figure down to $464.1 million,⁴⁴ but efforts to agree on a final price were suspended before the end of August because the whole Gemini program was in trouble.

Other costs were also on the upswing during the summer and early fall of 1962, though not as spectacularly as those for the spacecraft. SSD’s March 1962 figure of $164 million for the launch vehicles topped $170 million by September.⁴⁵ Less than a month later, SSD submitted to NASA a formal revised budget of $172.61 million.⁴⁶ Word reached MSC in July that the Atlas-Agena for Gemini now had a price tag of $12.3 million over its earlier total,⁴⁷ and this despite the fact that NASA had deleted the three spares to cut the number of Atlas-Agenas on order from 11 to 8.⁴⁸ A special briefing for NASA Administrator Webb on 28 September revealed that Project Gemini might cost as much as $925 million before it was over, 25 percent higher than Webb had been told in May it was going to cost and 75 percent more than MSC’s first estimate.⁴⁹

Such fast-rising costs would have been bad enough at any time. Now they presaged disaster, since Congress had not yet acted on NASA’s appropriation for fiscal year 1963 (which began on 1 July 1962). Without an approved money bill, NASA was compelled to carry on under a joint congressional resolution that provided enough money to support projects at roughly the same level they had enjoyed the year before but not enough to cover increases.⁵⁰ Gemini’s status was all the more threatened because it had not even appeared in the 1962 budget. NASA had found enough money to get Gemini started, but that was a makeshift that could not support an ongoing program.

The bill that authorized NASA’s funds was signed into law on 14 August, but the bill to appropriate that money was yet to come. Congressional action on NASA’s 1963 appropriation was not completed until 25 September. The figure was $3,774,115,000, $113,161,000 less than NASA had asked for and $70,000,000 under the total authorized in August.⁵¹

This delay prevented the Office of Manned Space Flight in Washington from giving MSC the normal authority to spend money on the basis of the full year’s budget. Instead, that authority was being granted on a month-to-month basis.⁵² Monthly funding brought anguished complaints from contractors, as expenses constantly threatened to outstrip the resources available to pay for them. By October, MSC was being bombarded with telegrams, each with urgent demands for full and quicker funding.⁵³

Lack of an appropriation also prevented NASA from adopting a final financial operating plan (FOP) for fiscal year 1963. Each center prepared an annual FOP to be approved by NASA Headquarters for allotting funds at the start of the fiscal year.⁵⁴ To meet the impending crisis, Associate Administrator Seamans imposed a ceiling of $1.51 billion on NASA research and development expenditures for the coming year. By this time, however, estimated funding needs for this purpose had already exceeded the figure first presented to Congress and now stood at $1.91 billion. Manned space flight chief Brainerd Holmes warned Seamans that current schedules could only be met by a supplemental appropriation from Congress.

In the meantime, Holmes directed MSC to prepare two separate fiscal-year 1963 FOPs: one staying within the Seamans-imposed ceiling, the other geared to actual needs. For Gemini, this meant a limit of $234.1 million against a needed $299 million. Holmes predicted a severe setback to program schedules if the smaller budget prevailed: a three-month delay in the first launch and in the first long-duration flight, an extra ten-month wait for the initial rendezvous mission, and no paraglider before the third flight.⁵⁵

Hopes for meeting the higher budget were dashed when President Kennedy rejected NASA’s case for extra funding. Holmes notified MSC on 9 October that its funds for fiscal year 1963 would be limited to $660.1 million. He directed the center to prepare new schedules to reflect this limit, voicing the somewhat forlorn hope that the unavoidable delay of several months might be made good if “later developments make it possible for the Administrator to obtain a FY 63 supplemental.”⁵⁶

The new ceiling was $27 million less than MSC had planned for under the earlier Seamans ceiling. The situation was now critical. Already tight at the level of $687 million, a budget of $660 million was nearly crippling. And Project Gemini bore the full brunt. Upon first hearing of the newly reduced budget, MSC planned to split the $27 million cut between Gemini and Apollo. Washington, however, ordered Gemini to take all the losses. Wesley L. Hjornevik, MSC’s Assistant Director for Administration, evaluating the situation for the senior staff on 19 October, saw no way out of this dilemma except to curtail Gemini sharply. “It appears", he glumly remarked, “that the consequent reduction to Gemini can only come by dropping paraglider, Agena, and all rendezvous equipment.”

Further complicating matters was the rate at which Gemini was piling up costs, a rate much higher than expected. Hjornevik pointed out that the program seemed to be costing $15 million a month, rather than the planned $11 million.⁵⁷ A budget memorandum that reviewed Gemini funding during the first quarter of fiscal year 1963 described as “an area of growing concern and one which can no longer be left unattended” the speed at which costs for spacecraft, paraglider, launch vehicle, and target vehicle were growing. The FOP could not “support acceleration of cost rates so projected by these contractors. Unless appropriate direction is given to the contractors to restrict this buildup or a Gemini reprogramming action is effected immediately then funding difficulties will commence during the second quarter.”⁵⁸

Reprogramming Gemini

The project office had already moved to reprogram Gemini, to alter the course of the program and compel the contractors to conform to the newly limited budget. Reprogramming was much more drastic in some areas than in others. Paraglider escaped almost untouched. McDonnell’s spacecraft effort took some trimming but remained much what it had been. The launch and target vehicle programs, the Air Force portion of Gemini, endured the most far-reaching changes. Plans for testing the Gemini launch vehicle were sharply cut back. Target vehicle testing was even more drastically curtailed; for some months, in fact, whether Agena still had a Gemini role was an open question.

Realignment of McDonnell’s work began first. Spokesmen from McDonnell and its subcontractors met in Houston at MSC on 24-26 August and again on 6-8 September. They agreed to limit the scope of development for some spacecraft systems and ground equipment.⁵⁹ But MSC Director Gilruth told Walter Burke, McDonnell’s spacecraft chief, not to do anything right away. When Gilruth talked to Burke on 8 September, the financial situation was still fluid enough to warn against too-hasty action. By the end of the month, however, prospects for any quick easing of the money crisis were fading. Burke flew to Houston to see Gilruth and Chamberlin on 28 September. Gilruth told Burke to carry out the earlier agreement on the revised scope of the program. Burke set his staff to work that same day on the necessary paperwork, wiring the subcontractors formal notice of their altered responsibilities and drawing up the required purchase order changes.⁶⁰

Reprogramming at McDonnell in St. Louis was mainly a matter of making some adjustments. The company cut back its own and its subcontractors’ quality assurance and reliability programs, reduced the number of published reports, decreased the number of spare parts to be maintained, trimmed the amount of engineering data and support required of subcontractors, and limited its support at Cape Canaveral. The net result of these changes was to slice $26 million from the $464 million that McDonnell thought its part of the project would cost, bringing the total down to $438.2 million.⁶¹

The largest savings in spacecraft development were to come through lessened testing by subcontractors. Teams from GPO spent much of October on two-day trips to major spacecraft subcontractors.* At each plant, they reviewed in detail the effect of various forms of systems failures, plans for qualification and reliability testing, and test facilities required. In general, they agreed that reliability testing could be sharply curtailed at the expense of slightly increased qualification testing. Qualification tests ensured that something worked; they usually preceded reliability tests, which made sure that something worked consistently. Assured reliability could thus be gained from augmented qualification tests.⁶² Concerned by the way the program had grown, GPO also asked McDonnell for prompt notice of any future action that might affect contract costs or schedules.⁶³

Spacecraft reprogramming was largely complete by mid-October, but the project office saw some further trimming possible in McDonnell’s test program. After a review of its plans for structural tests of the spacecraft, the contractor concluded that one of the four programmed static articles might be dispensed with, and GPO agreed.⁶⁴ The project office also suggested that Project Orbit might be canceled, a view McDonnell opposed. The dispute was eventually resolved with Orbit restricted to testing the spacecraft’s heat balance and renamed “spacecraft thermal qualification test.”⁶⁵

Another casualty of Gemini’s financial straits was a lately revived lunar landing scheme. This time the impetus had come from NASA Headquarters in the person of Joseph F. Shea, newly appointed Deputy Director for Systems in the Office of Manned S ace Flight. Shea wanted McDonnell to study using a Gemini spacecraft as a lunar logistics and rescue vehicle, a possibility also under study during that summer by the Space Technology Laboratories.⁶⁶ The eight-week McDonnell effort explored the concept of a two-man command module, evaluated using a Gemini spacecraft to land two men on the lunar surface, and looked at the design changes needed for such a mission.⁶⁷ Meanwhile, GPO computed the cost of buying extra spacecraft.⁶⁸ McDonnell submitted its findings to NASA Headquarters in November 1962.⁶⁹ Whatever chance the scheme may have had, however, vanished in the wake of Gemini’s money problems, and the idea once again came to nothing.⁷⁰

With the spacecraft taken care of by mid-October, the project office turned to launch vehicle programming. Limited funds compelled GPO to restrict 1963 costs to $59.28 million, some $10 million below its earlier plan and $18 million less than the $77.5 million SSD now claimed to need.⁷¹ Chamberlin wired Richard Dineen, SSD’s chief of Launch Vehicle Development, on 19 October to apprise him of the new funding limits. GPO believed that Gemini’s major goals might still be met despite shortage of funds. The key was a sharp cutback in testing, especially where it involved repeated engine firing.⁷² To Dineen, these changes seemed drastic, and he asked Chamberlin for a fuller explanation.⁷³ Chamberlin insisted that there was no hope of more than $59.28 million for 1963, which meant the planned test program had to be reduced and, in part, canceled. He asked Dineen for an early meeting to decide how to put these changes into effect.⁷⁴ SSD still objected.⁷⁵ Chamberlin persisted, wiring Dineen on 16 November that a meeting to review the launch vehicle test program was urgent and “should take precedence over other SSD/Aerospace/Martin/Aerojet Gemini commitments.”⁷⁶ The meeting finally convened on 27 November.

The proposed changes were indeed drastic. The revised engine program called for only 34 test firings, less than a fifth of the number originally planned. This would yield all the data needed at a saving of several million dollars, if effort were focused on thorough development and qualification to make sure each part worked and would keep on working.⁷⁷ Sound engineering, in other words, made reliability a natural product of development and qualification. SSD and its contractors could scarcely quarrel with this view, but they tended to see reliability in more statistical term - a part was reliable if it failed no more than some very small percentage of the times it was tested. The issue was not merely philosophical. Proving reliability statistically meant more tests, more equipment, and more money.

What was true for engines was also true for other parts of the launch vehicle. Martin’s reliability program was budgeted for $2.7 million, but the GPO approach, by concentrating dollars on qualification rather than on reliability testing, could cut that figure in half.⁷⁸ Further study convinced Chamberlin that most of the planned prelaunch firings of the complete launch vehicle could also be safely discarded, and they were.⁷⁹

NASA’s budget crisis in the fall of 1962 never posed any real danger to Project Gemini itself. Work on spacecraft and launch vehicle was simply adjusted to meet an unexpected funding squeeze. Whether the Gemini that emerged from reprogramming would be the same project that had been planned, however, was another question. Tight money threatened to deprive Gemini of its chief objective, the development of orbital rendezvous techniques. For several months the role of Atlas-Agena in the program was in jeopardy, as NASA Headquarters debated dropping it, cutting it back, or keeping it with whatever slippage restricted funding entailed. The choice was not made any easier by the complex management structure of the target vehicle program. Two organizations, Marshall and SSD, stood between GPO and Lockheed, Agena’s builder.

Word of tight budgets and a need to cut costs had reached Marshall’s Agena Project Office by early October 1962 but was slower getting to SSD.⁸⁰ The first firm notice that the Atlas-Agena program was to endure something more than a routine economy drive came on 23 October, when Chamberlin wired Friedrich Duerr, Agena systems manager at Marshall, “to reshape and reschedule the Atlas-Agena to conform to budget limitations. MSFC is further directed to establish accounting procedures and funds expended monitoring procedures to assure that Agena development is prosecuted within the established fund limitations.”

GPO had just completed a detailed study of changes that might be made in the Agena program to keep costs within budget limits. It concluded that $16.7 million could be sliced from the 1963 Atlas-Agena budget, dropping it from $27 million to $10.3 million.Chamberlin presented Duerr with the $10.3 million figure as a funding limit for fiscal year 1963, as part of an overall goal to reduce the cost of development by a third. For Agena, like Titan II, the savings were to be found mainly in less engine test firing and more built-in reliability. But Agena faced sterner sanction - no more money and all work stopped until reprogramming was complete.⁸¹

Duerr passed the word to the Air Force,⁸² although, as he informed Chamberlin, GPO’s view of the savings that might be achieved was “optimistic” and the changes could only mean a long delay in the development program.⁸³ Reprogramming began with a meeting in Houston on 25 October to discuss plans and schedules. What reliability meant emerged as the central issue, just as it did for Titan II. A second meeting, to agree on a specific plan, was set for 2 November.⁸⁴

Before that meeting convened, however, the real need for Agena in the Gemini program was called into question. In mid-1962, NASA had decided in favor of the lunar orbit rendezvous scheme for the Apollo lunar landing. That tentative decision was confirmed on 24 October by the findings of a manned lunar landing comparison study.⁸⁵ At a meeting of the Manned Space Flight Management Council six days later, Holmes raised the issue of Gemini objectives in light of this decision. Shea reviewed Gemini’s aims and claimed “that all of these objectives appear to be possible of achievement without use of the Agena in the program.” MSC Director Gilruth disagreed, and an inconclusive debate over the fate of Agena followed. Although he knew that time was running out, Holmes asked Gilruth to study the matter further.⁸⁶

Meanwhile, the second reprogramming session convened at the Lockheed plant in Sunnyvale, California. The monthly spending rate under the Gemini-Agena contract had reached $2 million during October. The limit for November, however, was fixed at $650,000, and Lockheed was instructed to stay within it. Lockheed spokesmen protested, claiming that Bell Aerosystems, the engine subcontractor, could not produce engines for an October 1964 launch if funds were so restricted. Chamberlin told them they had no choice - they must find ways to stay within the fixed limits. Lockheed had a week to provide a rough cost estimate for the revised program to SSD, which would turn its findings over to Marshall’s Agena Project Office, which, in turn, would pass its findings up the line to GPO. A final meeting to coordinate the changes was scheduled for 20 November.⁸⁷

Duerr reminded Chamberlin that limited funding was bound to cost time, perhaps as much as 14 months, in Agena development. Extra money - $12.7 million instead of $10.3 million for the current fiscal year would hold the loss to a less painful five and a half months.⁸⁸ But even at that, it would still be “a maximum risk program. That is to say that the target vehicle program has been minimized and no allowance is made for contingencies that may arise which would adversely affect costs and schedules.”⁸⁹ Chamberlin knew as well as anyone that time was being traded for money, but his hands were tied. A financial operating plan for 1963 had yet to be approved. Whether Agena could even be kept in the Gemini program - and not the precise level of funding - was the crucial question.

At a meeting of MSC’s senior staff on 9 November, Chamberlin strongly objected to Shea’s claims at the Management Council meeting on 30 October. Shea, and others in NASA Headquarters, believed that rendezvous goals might be met by using a “piggyback” rendezvous package, carried aloft in the adapter section of the spacecraft and then ejected in orbit to serve as a stable but non-maneuverable target. Chamberlin dismissed the piggyback technique as inherently limited in contrast to the stabilized and maneuverable Agena. He also believed that the package would be far heavier than its proponents claimed. André Meyer, chief of GPO administration, figured its weight at 180 kilograms, twice the Headquarters estimate. If that were true, it could mean the end of paraglider. Meyer thought the package would cost as much as Agena, although without the problems and expenses of separate launches.⁹⁰

MSC had been thinking along similar, but much more modest, lines. A study issued on 28 March 1962 had concluded that a piggyback rendezvous target could provide useful data. A month later, McDonnell had suggested testing the spacecraft rendezvous radar and maneuvering systems on an early Gemini flight with what it called a “Rendezvous Evaluation Pod (REP).” This was a small battery-powered module with a radar transponder, radar beacon, and flashing light, the whole package weighing about 30 kilograms and designed to give the pilots a chance to practice terminal rendezvous maneuvers with their spacecraft. In June, MSC had told McDonnell to go ahead with design and development. The REP would be carried on the second and third Gemini flights. Planning was largely complete by the end of 1962, with Westinghouse, the rendezvous radar subcontractor, responsible for components and McDonnell for the package and its ejection.⁹¹ This, however, amounted to little more than an experiment, intended to prepare for, not supplant, the Agena rendezvous missions.

On 16 November, Wesley Hjornevik, chief of MSC administration, reported to the senior staff that a financial operating plan for fiscal year 1963 had finally been approved. Agena funding, however, had been withheld.⁹² Target vehicle reprogramming went ahead, with the final meeting on 20 November in Houston. Lockheed’s new program was accepted. The major changes made reliability demonstration part of development and qualification testing, cut engine development testing to the bone, and trimmed production lead times to keep down 1963 expenses. This last meant chiefly that Lockheed was to work at a reduced level through the rest of calendar year 1962, then return to full effort on 2 January 1963. The program would be four months late, but its total cost could be as low as $44.1 million, $32.7 million less than estimated before reprogramming began.⁹³

Gilruth outlined the revised Atlas-Agena plans to the Management Council on 27 November, with a sharp reminder that “it is very critical that a decision as to the inclusion of the Atlas-Agena in the program is reached soon if the Agena target schedule is to be maintained.” Holmes promised a ruling by 10 December.⁹⁴ Not only had the fate of Agena become a matter of public speculation, but lack of funds threatened to stop the target vehicle even before anything was decided.⁹⁵

The decision came early but turned out to be only a stopgap: $900,000 for another month. This brought the total for fiscal year1963 to $4.9 million; the balance of the planned $10.3 million for Atlas-Agena remained in abeyance.⁹⁶ Shea, who had proposed dropping Agena from Gemini, told a reporter that NASA was thinking about several alternatives to simplify the rendezvous concept, with a decision due shortly. He gave Agena only a 50-50 chance of staying in the program.⁹⁷ Agena’s fate was in the hands of a NASA-wide committee, which Shea himself headed. A thorough investigation, bolstered by the well-informed and forceful case presented by James Rose, the GPO member, decided the committee in favor of Agena. A wire from Washington on 21 December authorized MSC to spend the full $10.3 million needed for the reprogrammed Agena in fiscal year 1963.⁹⁸

MSC also took over management of the Gemini Agena program. NASA decided to transfer all its Agena programs from Marshall so that that Center could focus on the Saturn launch vehicle for Apollo. Lewis Research Center in Cleveland, Ohio, assumed control of all NASA Agena programs except Gemini, which went to MSC.⁹⁹ MSC, now dealing directly with SSD,¹⁰⁰ took formal charge of the Gemini Atlas-Agena program on 14 January 1963, with active advice from the Marshall office for the next month and a half.¹⁰¹ Lockheed and SSD also adjusted their management relationships. The Gemini manager at Lockheed, Herbert Ballard, moved up a notch; he now reported directly to the head of Lockheed’s Medium Space Vehicles Programs. SSD followed suit by upping the rank of its program manager from captain to major; and Major Charles A. Wurster took over the reins.¹⁰²

Since the only function for Atlas in Project Gemini was launching the target, its fate waited on Agena’s. But Atlas, too, suffered in NASA’s fall budget crisis. On 25 July 1962, NASA Associate Administrator Seamans had agreed to support Air Force development of a standard Atlas launch vehicle, SLV3.¹⁰³ By the time the Department of Defense had drafted a formal Memorandum of Agreement and forwarded it to NASA on 21 August, NASA’s funding outlook had so deteriorated that it could no longer contribute to the program. Seamans restated NASA’s interest in SLV-3 development but declined to commit the roughly $10 million that was to have been its share of the cost.¹⁰⁴

Reprogramming raised the possibility of using surplus Atlas boosters from the Mercury program in Gemini. Chamberlin asked SSD for an opinion. A report to the Atlas-Agena reprogramming meeting of 20 November was favorable. Chamberlin then asked the Atlas contractor, General Dynamics/Astronautics, for a formal proposal.¹⁰⁵ The results made conversion look promising economically. Three converted Mercury boosters could be had for a net cost of $3.364 million, as opposed to $5.4 million for three new standard Atlases.¹⁰⁶ But by the time those figures were submitted on 13 February 1963, Gemini’s budget crisis was over, and NASA was back in the standard Atlas development program. In December, Seamans had formally committed NASA to pay its $10 million share.¹⁰⁷

The Prospect for 1963

With reprogramming completed, Gemini’s prospects looked reasonably bright as 1962 gave way to 1963. The crisis through which the program passed in the last quarter of 1962 was monetary, not technical. That crisis weathered, the technical problems looked less menacing as well. In his report to the Management Council on 18 December, Gilruth noted that Gemini still had a number of technical problems, but all, he judged, “are being actively pursued and none appear to be unresolvable.”¹⁰⁸

Gemini had lost time, though. The new Gemini program was chiefly a response to budget limits imposed from outside, compounded by sharply rising costs. Its immediate goal was cutting back expenses during the current fiscal year, and this meant slowing down the program. But a longer program, despite the curtailed and streamlined development that emerged from Gemini’s fall crisis, was likely to cost more in the long run. Whether the total cost of the program would really rise, and how much, could only be answered with the passage of time.

The effects of reprogramming on Gemini schedules were easier to define. Gemini was going to lose four months. The new date for the first launch was December instead of August 1963. It was now an unmanned suborbital qualification test. McDonnell had proposed in July 1962 an extra mission that it called Flight No. 0, a suborbital shot to precede the first planned mission. On 20 July, Burke and Chamberlin agreed to replace the planned unmanned orbital flight with the suborbital flight as the first mission (a slightly revised version of the Mission 0 plan). It was to be a ballistic test to investigate spacecraft heat protection, to integrate launch vehicle and spacecraft preflight and launch operations, and to obtain data on spacecraft structure and systems.¹⁰⁹ All other launch dates were set back four months. The second flight - manned orbital qualification followed the first by three months, in March 1964, with the rest of the missions coming every two months until the 12th and last, now scheduled for November 1965.¹¹⁰

By December 1962, everything seemed to be under control again. But while the project office and MSC were wrestling with the hard tasks of fitting development work to the limited money available, NASA Headquarters found itself fending off quite a different threat - perhaps the least expected of all. The Department of Defense was making gestures toward taking over Project Gemini.