Chapter 9

A Taste of Success

While Gemini’s first spacecraft and launch vehicle were moving toward their mating on complex 19 at Cape Kennedy, the Gemini Program Office itself was coping with another kind of move. The permanent home of the Manned Spacecraft Center at Clear Lake, though not quite finished, was ready to be occupied. GPO began shifting its desks from the old Veterans Administration building in downtown Houston to the new campus-like setting near Clear Lake on 6 March 1964. Shortly after the transfer had been completed, Program Manager Charles Mathews announced a reorganization of GPO. Major changes reflected the growing stress on schedules and testing as Project Gemini poised on the verge of its first flight. Project Administration changed its name to Program Control.* Scott H. Simpkinson left Mathews’ staff to take charge of a new Test Operations Office dealing with reliability and quality assurance as well as test planning and evaluation.** Launch Vehicle Integration became Vehicles and Missions, divided into vehicle development and mission planning offices, plus a new integration office to keep tabs on spacecraft/launch vehicle and spacecraft/target interfaces.*** The Spacecraft Management Office simply changed its name to the Spacecraft Office.**** The Houston-based strength of the program office had now reached 117; GPO also maintained representatives at Martin in Baltimore and Lockheed in Sunnyvale, California, as well as resident manager’s offices at McDonnell in St. Louis and Kennedy Space Center at the Cape.# This was the organization that, with only minor changes, saw Project Gemini through to its end.¹ Before that happy end, however, there was the more immediate matter of Gemini-Titan 1.

The First Flight

By 3 March 1964, spacecraft and booster were at last together on launch complex 19 at Cape Kennedy. The series of tests that showed all booster systems were working had just been completed, and the spacecraft had been hung on a tripod in the “white room” atop the launch vehicle erector. This room, with its four levels and 4.5-tonne (5-ton) crane to hoist the spacecraft, was sealed off from the outside world and maintained at a constant temperature of 295 kelvins (72°F) and a constant relative humidity of 50 percent, to provide a controlled environment for the spacecraft and the upper stage of the booster. Next to the erector was an umbilical tower 31 meters high. Its seven booms supported 31 cables and lines to spacecraft and booster, feeding electrical power, propellants, and other needs until the moment of launch. Gemini-Titan I was scheduled to lift off on 28 March 1964.²

A premate systems test on 4 March confirmed the spacecraft ready for mating the next day, when the spacecraft-to-launch-vehicle adapter would be bolted to the booster’s upper stage. The effort was delayed briefly when a McDonnell worker dropped his wrench on the dome of the oxidizer tank just below the spacecraft. A plastic sheet protected the dome, but the impact produced a scratch 0.95 centimeter (0.375 inch) long and 0.0038 centimeter (0.0015 inch) deep in the steel surface, just 0.16 centimeter (0.64 inch) thick at the point of impact. The area was burnished to the depth of the scratch and tested to confirm that the metal was still solid.³

After the spacecraft and launch vehicle had been mechanically mated, they also had to be connected electrically. But first the booster’s status had to be checked in a combined systems test. That was slated for Sunday, 8 March, to be followed by three electronic-electrical interference tests between 9 and 13 March, to make sure there was no serious incompatibility. Minor problems delayed the booster combined systems test until Tuesday, and interference testing did not start until Thursday, 12 March.⁴

The first try at an interference test had to be scrubbed, and that cost another four days. On Monday, 16 March, however, the test went off without any trouble, prompting the crew to run through the second test at once. The attempt went awry through a procedural error. Another try, on Thursday, 19 March, brought bad news. Some amplifiers in the circuits that controlled the boosters tandem actuators (which shifted the engines to alter flight path) showed noisy outputs. A special dry run the next day produced the same problem, and the third interference test had to wait until the trouble was resolved. There was some question about how that was to be done, which was settled on Tuesday 24 March, when Martin troubleshooters pinpointed the problem - in the test equipment. Another test, on Wednesday, confirmed the finding. A conference that evening concluded that the data from the dry run the previous Friday met the intent, if not the precise format, of interference testing. The test equipment was removed that night.⁵

But the tests had taken almost two weeks longer than planned, forcing the launch to be postponed to 7 April 1964. Things now began to move more smoothly. On Friday, 27 March, a combined systems test and simulated flight produced no serious problems.⁶ The following Tuesday, 31 March, all the nonflight parts that GLV-1 had carried to the Cape were replaced and Pogo gear installed. GLV-1 was scheduled to have its tanks filled with propellants that night as part of a complete countdown exercise, the wet mock simulated launch.

At 9 p.m., as shift workers were clearing the area for the start of tanking, someone saw smoke pouring from a switch at the pad. A burnt-out transformer and switch motor forced the test to be suspended, since there were no spares on hand and the switch performed a crucial function. It automatically transferred the launch complex to auxiliary power if commercial power failed. Safety demanded that the launch area be deluged with water in case of propellant leak; a power loss would leave that system inoperable for about 30 minutes if the automatic switch were not working. Workmen found a spare transformer at 1:18 Wednesday morning and installed it, but a new motor was harder to locate. One was finally borrowed from the blockhouse since that system could be run by hand.⁷ But another day had been lost.

Propellant loading resumed just before 10 Wednesday night and finished four hours later. The countdown began at 5 o’clock Thursday morning, but now came weather trouble. The Cape was under an “atmospheric inversion,” a blanket of warm air above cooler air near the ground, which would block the upward dissipation of toxic fumes in case of accident. The count was held from 7 to 8:30, when the inversion started to break up. Ground crews then removed the propellant lines leading to the booster tanks and the count resumed. It followed its normal course until three minutes before launch, T-3, when a minor problem (quickly corrected) required the count to be recycled to T-5. Five minutes later, at half-past noon, the count reached T-0, the moment when the booster’s first-stage engine would have ignited in a real launch. The test was a complete success, free of spacecraft problems and marred only by a minor procedural error in the launch vehicle countdown. After a vibration test of GLV-1, the tanks were drained of propellants, a five-hour process finished at midnight.⁸

The Spacecraft Flight Readiness Review Board* convened Friday afternoon, 3 April, in the conference room of the Engineering and Operations Building, headquarters for MSC’s Florida Operations. A check of items left open from the preflight review of 18-19 February showed that everything had been taken care of except a circuit breaker not yet fully qualified. It was close enough, however, for McDonnell to certify it flightworthy, a judgment the board shared. Only two new problems had cropped up since the earlier review, both easily corrected. The board judged all systems ready for flight, pending the outcome of the final systems test, a simulated flight scheduled for 5 April. When the simulated flight went off without a hitch on Sunday, Spacecraft 1 was ready for its mission.⁹

Flight readiness of the launch vehicle was reviewed Saturday afternoon. The Air Force reported two problems, one of which turned out to be nonexistent. The other involved a missing report of the results of an analysis of a failure in the secondary autopilot. The report was still absent on the eve of flight, but a phone call confirmed that the problem had been analyzed. After the simulated flight on Sunday, Walter Williams convened the Mission Review Board. Spokesmen for every group involved in the mission reported everything ready - “all systems ‘go.’” At noon, Williams announced that NASA was “proceeding toward a launch not earlier than 11:00 a.m. Wednesday, April 8.”¹⁰

The final decision for launch came on Tuesday morning. At 7:30, 7 April, SSD’s Status Review Team for GLV-1 met, took a last look at the launch vehicle, and agreed it was ready to go. That recommendation was passed on to the Flight Safety Review Board at 9:00 a.m. The board approved GLV-1 for flight and committed it to launch, with lift-off set for 11 the next morning.¹¹

Preparations for the final countdown were already under way. The first part of the planned 390-minute split countdown started before dawn on Tuesday. That 60-minute segment ended at 5 a.m., when the count was held for 23½ hours to prepare the spacecraft for final countdown, install and hook up pyrotechnics, run some launch vehicle tests, and load propellants. GLV-1’s tanks were topped off at 4:10 Wednesday morning, with about 75 people from Martin, the Air Force, Aerojet-General, and Aerospace on hand. Thirty systems experts from McDonnell and MSC arrived at the blockhouse at 4:30.The hold ended right on time, an hour later, and final countdown began at 6 a.m. or T-300. No flaw marred the entire five-hour process.

One second after 11 o’clock Wednesday morning, 8 April 1964, the booster’s first-stage engine ignited. Of this one-second discrepancy, a joking Williams later remarked to a roomful of reporters, “There must be something wrong with the range clock.” Four seconds later, the 136-tonne (156-ton) vehicle lifted from the pad on that curiously lambent flame so distinctive of Titan II’s hypergolic propellants.¹²

Within moments, Gemini-Titan 1 vanished into the hot Florida sky, beyond reach of human senses but not electronic sensors. Telemetered data flowed back to mission controllers at the Cape, telling them that the launch was as nearly perfect as it looked. Two and a half minutes after liftoff, the 118 tonnes (130 tons) of propellants in its first stage exhausted after driving Gemini-Titan 1 64 kilometers high and 91 kilometers downrange, GLV-1’s first-stage engines cut off. The second-stage engine flared into life, and the four bolts that had held the two stages together exploded as they were designed to, cutting the spent first stage loose from the still-accelerating second stage and spacecraft. Five and a half minutes after launch, the second- stage motor stopped, its 27 tonnes (30 tons) of propellants gone. Now 1,000 kilometers downrange and 160 kilometers high, coasting at a speed of 7,888 meters (25,879 feet) per second, Gemini Spacecraft 1, with the second stage of GLV-1 still attached, was in orbit.¹³

Everything had gone beautifully. Purists might cavil at an excess 7 meters (24 feet) per second launch-vehicle speed that propelled the spacecraft into an orbit reaching out 320 kilometers instead of the programmed 299 kilometers. But they could scarcely deny the handsome achievement of the main goals - proving that the booster could do its job and that combined with the spacecraft its structure was sound. “There’s no question these objectives were met,” Walter Williams observed to the press shortly after launch.** The nearly flawless performance of the launch vehicle elated its sponsors, prompting one of them, Major General Ben Funk of SSD, to call it “just completely a storybook sort of flight.”¹⁴

The mission of Gemini-Titan 1 was much shorter than its actual trip. Only the first three orbits were part of the flight plan. When Spacecraft 1 passed over Cape Kennedy for the third time, about 4 hours and 50 minutes after launch, the first Gemini flight came to a formal close. The spacecraft had been expected to orbit Earth for three and a half days. Because of its slightly higher than planned orbit, it actually stayed up for nearly four days. During that time, the Manned Space Flight Network,*** a round-the- world system of tracking stations controlled from Goddard Space Flight Center in Maryland, followed the vehicle by radar. On Sunday, 12 April, during its 64th pass, the steadily slowing spacecraft plunged back into the atmosphere, ending its career in flames over the South Atlantic, midway between South America and Africa.¹⁵

NASA Associate Administrator Robert Seamans commended “the Air Force for its most successful Launch Vehicle Program.”¹⁶ So fine a performance of the first mission augured well for those to follow and surely enhanced the prospect that Gemini astronauts would be in orbit before the end of the year. But the glow of accomplishment soon faded before the hard work yet to be done. While the launch vehicle was now qualified for manned missions, the spacecraft was not. Despite the gratifying success of Gemini-Titan 1, and some real progress on troublesome spacecraft systems, there was no time to rest on laurels. The target vehicle for Gemini’s later missions was still a very large question mark, and Gemini’s chronic money woes were far from settled. For all of that, Gemini’s future in the spring of 1964 must have looked much brighter than it had only a few months earlier.

Postscripts and Prospects

So bright, in fact, did the future seem that the long dormant idea of using the Gemini spacecraft for a lunar mission stirred again. George Mueller, NASA’s Associate Administrator for Manned Space Flight, had some reason to be concerned about the outlook for Project Apollo in the spring of 1964. Only a few months earlier, plans for manned flights using Saturn I had been canceled, leaving Gemini as the only possible system for manned orbital flights during the next two years or more. Mueller wanted to know if a Gemini lunar mission could be flown. If it could, then a contingency plan was to be prepared for a Gemini flight around the Moon in case Apollo suffered a serious setback. A review of past studies strongly suggested that the idea was feasible and that McDonnell should be asked to conduct a more detailed study.* ¹⁷

But that was not to be. During a tour of the plant in Louisiana where Saturn rockets were built, Wernher von Braun, Director of Marshall Space Flight Center, told a journalist that Gemini might be able to fly around the Moon, but only as “a possible project to salvage this country’s prestige if the manned lunar goal proves impossible.” Whether this was intended to squelch an Apollo rival, the effect might have been predicted. The same factors that had blocked the idea before still held. NASA had too much invested in Apollo - too much money, time, and prestige - to really think about Gemini to the Moon. Funds, in any case, were tight. On 8 June, Seamans told Mueller there would be no money for study contracts. “Any circumlunar mission studies relating to the use of Gemini will be confined to in-house study efforts.”** ¹⁸

But that was never more than a side issue. In mid-1964, the first task was still Project Gemini, however attractive the prospects of a more ambitious program might seem. The outstanding performance of Gemini-Titan 1 and the qualification of the Gemini launch vehicle were most cheering portents. When the Gemini Management Panel met a week after the mission, on 15 April, a comfortable optimism suffused the group. The current work schedule called for the second flight toward the end of August and the third in mid-November, with almost a four-week cushion in each instance to handle unforeseen problems.¹⁹

This bright outlook darkened in the late summer before a series of natural disasters. First lightning, then hurricanes, conspired to abuse the second Gemini launch vehicle on complex 19 at Cape Kennedy and to delay its flight long past the scheduled time. Even had the weather been perfect, however, McDonnell’s difficulties in getting Spacecraft 2 ready to fly might have compromised the schedule.

Late deliveries - notably of thruster systems from Rocketdyne and fuel-cell stacks from General Electric - had slowed construction of the spacecraft during 1963. Parts had failed tests that had to be passed before they could be installed in the spacecraft; modifications meant further delays. Spacecraft 2 could not begin its systems tests until 13 January 1964.²⁰

The Spacecraft 2 Design Engineering Inspection (DEI), earlier set for November 1963, had been postponed in the face of these delays until February 1964. MSC formed a permanent DEI board 31 January 1964 to make sure that the spacecraft as a whole and each of its parts would do what they were intended to that the spacecraft could, in fact, be expected to achieve its assigned objectives. Normally, the DEI for each spacecraft would fall between the end of manufacturing and the start of systems testing, but the DEI for Spacecraft 2 was a little late. The nine-member board convened at the McDonnell plant on 12 February.*** Also present for the two-day meeting were 50 experts from GPO and McDonnell, as well as another 50 observers from other MSC offices, NASA Headquarters, and the Air Force. The board looked over the hardware and studied the records to see that each part either matched design specifications or was the subject of a proper waiver. A long list of minor discrepancies ended up as 22 mandatory changes, 4 conditional, and 10 to be studied.²¹

The first phase of spacecraft systems tests went slowly, as problem after problem turned up; troubleshooting them, working out the required changes, and testing the results all took time, adding to the delays. By mid-April 1964, Spacecraft 2 had become the “pacing item” for the second Gemini mission, a dubious honor held by the launch vehicle before the first flight. Getting the spacecraft ready was now the crucial factor in meeting the scheduled launch date.²² This was not altogether a surprise. Spacecraft 1 had been little more than an instrumented shell, but GLV-1 had been a launch vehicle in every sense of the term. The Martin crews working on GLV-2 were going over ground they had already surveyed, but Spacecraft 2 was the first fully equipped ship to go through the McDonnell plant and its slow progress reflected its novel status.

After the modules of the spacecraft had been mated, the second phase of systems tests began, on 3 July. Further problems hampered testing into the next month.²³ Whatever delay might have resulted, however, became purely academic after mid-August, when Florida weather dealt the first of a series of time-consuming blows to GLV-2.

GLV-2 and the Elements

While spacecraft testing floundered past snag after snag, GLV-2 had been moving briskly through its test program despite some rough spots. At the outset, the second-stage oxidizer tank was found defective, and a new tank had to be built. Since the first-stage tanks were not yet ready, the delay was inconsequential. Martin-Baltimore received all four tanks from Denver on 12 July 1963. Engines were late in arriving from Aerojet-General, but testing went ahead with nonflight first-stage engines. By the end of January 1964, GLV-2 had completed its horizontal test program. Early the next month it was standing in the Vertical Test Facility; and, after two weeks of modification work, functional verification tests of subsystems began on 21 February.²⁴

GLV-2 finished these tests by 13 April, in roughly two thirds the time taken by the first booster. Another week saw it through electrical-electronic interference tests and three preliminary combined systems acceptance tests (CSAT), an effort that had cost GLV-1 over a month. The formal CSAT was run on 22 April with no trouble, and the results were approved by the Vehicle Acceptance Team the following week. The dummy engines still had to be replaced, which took a month. By mid-June, GLV-2 had been inspected and formally accepted for the Gemini program. Since spacecraft work was lagging, the booster’s transfer to the Cape was postponed so Martin crews in Baltimore could complete some of the modifications that would otherwise have been made by the Martin-Canaveral team.²⁵

Workmen loaded the booster aboard an Air Force C-133B aircraft on 10 July 1964. By noon the next day, both stages had been unloaded and secured. Working a two-shift, five-day week, Martin’s Cape crew expected to have GLV-2 ready for Spacecraft by mid-August. Everything proceeded routinely through July and into August, with only minor problems causing small delays. This was of no moment, since the spacecraft was still in St. Louis. Its shipment, scheduled for 1 August, had been postponed for three weeks; it could not now reach complex 19 before the first week in September. The Martin crew nevertheless prepared for the final test of the booster before its mating with the spacecraft and were almost through by 17 August.²⁶

But that Monday a severe thunderstorm pounded Cape Kennedy. About half an hour before midnight, lightning struck complex 19. There was no visible damage to the blockhouse, erector, or rocket, but that proved nothing about the status of the electrical and electronic gear. Whether GLV-2 was fit to fly was a real question. NASA labeled the event an “electromagnetic incident” and demanded a thorough investigation. Inspectors from Martin, Aerospace, and the 6555th Aerospace Test Wing found no signs of any physical damage, but they did locate a number of failed parts, mostly in the ground support equipment. This suggested that the complex had not taken a direct hit but rather had suffered the electromagnetic effects, or induced static charges, of a nearby lightning strike. A test order issued on 20 August set the task: To “re-establish confidence in all [launch vehicle], AGE, . . . and Facility Systems, and to determine that all degraded equipment is replaced and appropriate reverification tests are successfully completed.” The next day, Gemini manager Mathews flew in from Houston for an “Incident Status Meeting.” A three-man steering committee was appointed to oversee the efforts of Air Force, Aerospace, and Martin work crews.* ²⁷

Two weeks seemed ample to put things back in order. Most subsystems would have to be retested, and all booster systems, test equipment, and facilities would have to be checked out. Any equipment that might have been affected had to be repaired or replaced. After some consultation, NASA agreed that no airborne units with semiconductors ought to be retained. Once new units were installed, testing could begin again as though the vehicle had just arrived at the Cape.²⁸

Before the work was finished, however, Hurricane Cleo belied the forecasts and brushed the Cape on Thursday, 27 August. The Martin crew had time to get the second stage down and under cover, but the first stage remained upright, lashed in place with the erector lowered. Cleo’s winds were well below the upper limit that the booster was designed to withstand. With the weather still bad on Friday, the second stage stayed in storage over the weekend. On Monday, the Air Force was getting ready to launch its first Titan IIIA from the next complex, which hampered work on pad 19 for most of the day. By 3 o’clock the next morning, however, the Martin crew had stage II back in place atop the first stage. Further work was delayed by the countdown on the nearby pad, which ended at 10 a.m., Tuesday, when the Titan IIIA blasted off. GLV-2’s repeat of subsystems functional verification tests began on Thursday, 3 September.²⁹

By then, MSC was just about ready to give up on GLV-2. The Center proposed dropping it from the program and moving each of the other launch vehicles up a notch. GLV-3 would launch Spacecraft 2, and the flight program would lose one mission. The Air Force, strongly seconded by the launch vehicle contractors, urged NASA to stick with GLV-2. A thorough review of the effects of both lightning and hurricane, the measures taken to counter them, and the test results had convinced the Air Force and its contractors that GLV-2 was still as sound as ever. Their case was solid enough to convert the skeptics. An Air Force spokesman concluded: “Based on technical considerations, Martin-Marietta Corporation, Aerojet-General Corporation, [and] Aerospace Corporation recommend fly GLV-2. In addition, SSD has reviewed cost and schedule considerations and concludes fly GLV-2.” NASA agreed, and the work went on.³⁰

Testing had scarcely begun, however, before Nature intervened a third time. Cleo had struck only a glancing blow, but Hurricane Dora was aiming straight for the Cape. As Dora approached on 8 September, Martin workers raced to get both stages of GLV-2 down and safely under cover in a hangar. Wednesday was a day of waiting as Dora passed by. On Thursday, Dora was no longer a threat, but Hurricane Ethel was heading for the Cape and due to arrive by the weekend. GLV-2 stayed under wraps. By Monday, 14 September, the danger was past, and GLV-2 was back in place before the end of the day. The rest of the week was largely given over to replacing semiconductor units and to a thorough inspection of booster and launch complex. Testing resumed after the weekend, on 21 September.³¹

That was the day Spacecraft 2 finally arrived at the Cape. The second phase of systems testing at St. Louis had lasted through August and into September, with frequent interruptions for the receipt and installation of a number of pieces of flight equipment.A simulated flight on 15 September completed testing. A Spacecraft Acceptance Review Board headed by Charles Mathews had already gone over the spacecraft to make sure it was ready for the final simulation.** The board met again on 17 September and decided that Spacecraft 2 was now ready for delivery. It was shipped to Florida the following Monday, 21 September.³²

GLV-2’s misfortunes during August and September 1964 forced NASA to forego its goal of a manned Gemini flight before the end of the year, as a rueful Mathews informed the Gemini Management Panel on 29 September. The second flight was now set for mid-November 1964, the third for the end of January 1965. There seemed no need to alter planned dates for the later Gemini missions, although the schedules would have to be tightened. Once again, Gemini’s slowness was highlighted by a Russian first. On 12 October, the Soviet Union orbited Voskhod I. The three-man crew flew in a “shirtsleeve” environment (flight coveralls rather than space suits) and all remained in the spacecraft to a land landing (previously only Yuri Gagarin was believed to have stayed with his vehicle until it landed, the others leaving the spacecraft and coming down by parachute).³³

GLV-2 began an expected two weeks of subsystems tests on 21 September, with the combined systems test that preceded spacecraft mating scheduled for 6 October. Spacecraft 2 should have taken only 11 working days in the hangar area before it joined the booster at the launch complex on 25 October. Once again, however, work on the booster went smoothly, but the spacecraft lagged. GLV-2 completed subsystems tests and the premate test on schedule. In another week the launch vehicle finished electrical-electronic interference tests, the last step before it was ready to receive the spacecraft. While the launch vehicle was being tested, so was the worldwide tracking network. From 9 to 16 October, Goddard and MSC put the tracking stations through their paces.*** ³⁴

The spacecraft, however, had yet to arrive at the pad. Work had gone well enough the first week, but trouble cropped up in getting the thrusters ready for a static firing test. After firing, the system had to be flushed and purged, another delay. By 10 October, Spacecraft 2 was already eight days behind schedule; it lost another two days while pyrotechnics were installed. Spacecraft 2 was ten days late when it reached complex 19 on Sunday, 18 October, and settled in the tripod in the white room an hour before noon.³⁵

Attempts to run the spacecraft premate systems test brought new problems. As one was solved, another appeared; and it was 27 October before the test was complete. The final step before the spacecraft was joined to the launch vehicle was a premate simulated flight, run in two parts. Despite more than one discrepancy revealed by the test, the spacecraft was mechanically mated to its booster by noon Thursday, 5 November.

After the mating Martin conducted tanking exercises on the launch vehicle to check calibration, to see whether or not the launch crew could load the tanks accurately with the equipment on hand, and to train for launch loading. The Martin crew found some differences between the data gathered from calibration and what they thought they had loaded. This led to a series of tanking exercises throughout the program and set up “a new family of people, called the Wednesday Evening Tanking Society and the Thursday Evening Tanking Society - the WETS and the TETS.”³⁶

The troubled course of testing and checkout now smoothed. Over the next month, any problems that showed up were handled quickly, as Gemini 2 ticked off the milestones on its way to a 9 December launch: electrical interface integrated validation, 9 November; joint guidance and control test, 12 November; joint combined systems test after electrical mating, 17 November; wet mock simulated launch, 24 November; spacecraft final systems test, 28 November; simulated flight test, 3 December; and launch precount, 7 December.³⁷

Setback and Success

Loading propellants aboard GLV-2 began in earnest on Tuesday, 8 December, an hour before midnight and finished shortly after three o’clock in the morning. The final countdown started an hour later. It went smoothly, though not quite so smoothly as the first Gemini countdown - there were three holds for a total of 41 minutes. The count reached zero at 11:41 Wednesday morning, and the first-stage engines ignited. One second later, a signal from the master operations control set shut down the engine. Flight controllers in the Cape control center observed that the launch vehicle had lost hydraulic pressure in its primary control system and had switched over from primary to secondary guidance and control. Within the blockhouse, technicians began to power down the spacecraft and, at three minutes before noon, Flight Director Christopher Kraft officially canceled the flight.³⁸

The proximate cause of the shutdown was the command from the master operations control set, an automatic response to an automatic function - the switchover from primary to secondary flight control during the 3.2 seconds between ignition and liftoff. After the engines ignited, the launch vehicle remained bolted to the stand until thrust built up to 70 percent of maximum. During that time, a switchover in the control system was an automatic shutdown order. The GLV-2 switchover followed automatically when the booster’s malfunction detection system sensed the pressure drop in the primary hydraulic system. GLV-2, in other words, spotted its own hydraulic failure, responded by switching over to its secondary system, and then, because it was still on the ground, commanded its engine to shut off.

Having saved itself, GLV-2 stood poised on the pad - a giant question mark. Why had its primary control system failed? The answer was quick in coming. Unexpectedly high pressure in one of the hydraulic lines had burst the aluminum housing of a servovalve, letting the hydraulic fluid leak out. This valve controlled one of the booster’s four tandem actuators, the devices that moved the thrust chambers to steer the vehicle in flight. Why the valve housing had failed was a lesson in the folly of unneeded “improvement.” At some time during development, someone had decided that the walls of the housing were twice as thick as they needed to be; a third of a centimeter of aluminum was ample to meet design pressures. No one, however, thought to test the actual pressure the housing would have to withstand, nor was any impulse test, as such, included in system qualification. More likely than not, one or another Titan II had suffered the same sort of hard start, but the stouter housings that remained standard in the missile could survive such a pulse while the lighter structural shell in the Gemini booster could not.³⁹

When GLV-2 shut down, Spacecraft 2 posed something of a problem. Launch crews knew what to do with a ready-to-go booster, since they dealt with one after the mock launch that was a regular feature of launch vehicle checkout. There was no comparable background for the spacecraft, however, and that led to some hasty improvisation. Aside from its propellants, the spacecraft fairly bristled with pyrotechnic devices, all armed for flight. Should one of them explode, the results might be catastrophic.

Draining the booster of propellants took first priority, so Wednesday had passed and Thursday was well along before the main part of spacecraft “safing” was complete. One particularly ticklish operation remained, however - pulling the pyrotechnics from the isolation valves that barred propellants from the spacecraft thrusters until time to fire. The problem was complicated by the fact that the explosive cartridge was not a replaceable unit, and the whole valve assembly had to come out. But this might allow propellants to reach the thrusters or to spill their highly noxious chemicals over the workers. The makeshift answer was to freeze the propellant lines. After one or two false starts - no one was quite sure how to do the freezing - copper tubing was wrapped around the lines (which were packed in dry ice), liquid nitrogen was run through the tubing, and the whole thing was sprayed with CO₂.* That worked, and the valve assemblies were replaced over the weekend.⁴⁰

There was really not much that could be done with the spacecraft over the next few weeks besides making sure it remained in flight status, and nothing much could be done with the launch vehicle until new actuators arrived.** A product of Moog Servocontrols, Inc., the tandem actuators had been taken back to the vendor’s plant in East Aurora, New York, for extensive tests. Then the actuators had gone to Martin-Baltimore for further testing. The lightweight servovalves had to be redesigned. Work was further curtailed by the holidays. A messenger reached the Cape with the four new parts on 6 January 1965. They were installed at once and testing resumed, focused mainly on the flight control system. The new round of launch preparations went quickly; by Thursday, 14 January, the last major test was complete. Reviews of spacecraft and launch vehicle gave both a clean bill of health, and launch was set for 9 o’clock Tuesday morning, 19 January.⁴¹

The countdown began two hours past midnight. It was almost flawless, although it did produce one disappointment. Spacecraft 2 had been slated to carry six fuel-cell stacks of the old model P2B, left over after the design had been updated early in 1964. Despite their known defects, flight testing them with the reactant supply system seemed like a good idea, but only on a “non-interference with flight” basis and with a dummy load, since electrical power would actually be supplied by battery. The six stacks assigned to Spacecraft 2 had behaved erratically since they were first installed in St. Louis. When they acted up during the abortive countdown on 9 December and threatened to delay the launch, they were scratched from the mission. Only one stack proved to be still operable; it was activated on 18 December, then shut off and left alone until the next launch attempt. An hour and a half after the countdown started on 19 January, hydrogen intake to the stack was blocked by a stuck valve. Two hours of work left troubleshooters faced with breaking the spacecraft wiring to correct the problem. Since that would have meant a hold in the countdown, the attempt to activate the stack was called off, and the fuel cells were not operated on Gemini 2.⁴² Aside from the fuel-cell problem, the countdown produced only the most minor anomalies an preplanned two-minute hold.

At four minutes after 9 Tuesday morning, Gemini 2 began the last unmanned flight in the Gemini program. GLV-2 hurled the spacecraft 3,430 kilometers across the South Atlantic through an arc that peaked 160 kilometers above the ocean’s surface. The spacecraft endured the most severe heating Gemini was ever likely to meet as it plunged back into the atmosphere, its heat protection proved, its structural integrity uncompromised, and all systems working. It dropped into the South Atlantic on its parachute about 18 minutes after launch, bobbing in the water for an hour and a half until it was picked up by the U.S. Navy’s aircraft carrier Lake Champlain.⁴³

Some small question marks dotted the mission, but overall it looked quite good. The postflight news conference was a scene of quiet jubilation, with pats on the back for everyone involved. Nothing earth-shaking turned up in the detailed study of the recovered spacecraft - only minor scratches, chars, corrosion from exposure to sea water, just about what might have been expected - nothing that would in any way militate against the forthcoming launch of Gemini 3, the first to carry men aloft.⁴⁴

Down to the Wire

While most eyes had been focused on Gemini 2 at Cape Kennedy, work on still-to-be-resolved development problems continued elsewhere. Two spacecraft systems indispensable for Gemini’s first manned mission - thrusters and ejection seats - remained question marks through most of 1964, and a third - fuel cells - though not slated for Gemini 3, was as yet unqualified. What may have been the largest question of all centered on the Gemini Agena, which throughout 1964 fell further behind schedule.

In April 1964, Rocketdyne seemed at last to have solved its major problems in developing workable thrusters for Gemini, but misgivings persisted. When the Jet Propulsion Laboratory approached Rocketdyne about developing a small engine for the Surveyor spacecraft, Mathews protested. He argued that the company was still a year away from having the Gemini orbital attitude and maneuvering system and reentry control system on a sound footing, and that the main reason the work had improved was the belief that it would get no more NASA small-engine contracts until Gemini work was almost done. Workloads in the California plant were heavy, as shown by the large demands for overtime, and the original $30-million contract had ballooned to over $74 million, of which almost $36 million was an overrun.

Despite the enormous infusion of effort and money, Rocketdyne had failed to maintain schedules and deliveries. Engines for Spacecraft 2, for example, due in February 1963, arrived on in January 1964, and “the delivered products leave much to be desired.” Mathews thought it “quite evident that all three interested parties, the Gemini Program Office, the Surveyor Program, and Rocketdyne, will benefit through the selection of a vendor other than Rocketdyne,” since the added work could only hamper Gemini without contributing much to Surveyor.⁴⁵

This concern was echoed by manned space flight chief George Mueller;* in a memorandum to his counterpart in the Office of Space Sciences, which had charge of the Surveyor program, he urged that Rocketdyne be denied the contract. MSC Director Gilruth also acted, setting up a special committee to survey Rocketdyne’s Gemini program. After hearing some harsh committee findings on 5 August 1964, Rocketdyne’s president promised that whatever NASA wanted would be done. Gilruth sent him a long list of recommendations a week later. Some changes were already under way even while the committee was meeting, and more followed, including a reorganization of Rocketdyne’s Space Engine Division.⁴⁶

Among the recommendations was a full-scale NASA audit of Rocketdyne’s business management practices and Space Engine Division operations. It was a large undertaking, and a report was not ready until April 1965. Its findings revealed a badly managed program. Having “grossly underestimated the magnitude and complexities” of its Gemini subcontract, Rocketdyne had been slow to set up a sound organization. As a result, budgets were poorly controlled “and operations were inefficient,” producing “significant cost overruns and delays.” Not only had outright overruns very nearly doubled the cost of the program, but, of the 358 engines that should have been delivered by November 1964 under the original contract terms, only 167 had actually been received. Frequent personnel changes at top levels reflected the program’s weak management, as did the company’s complete inability to provide records showing the reasons for technical problems, what action they prompted, or what impact each problem had on costs and deliveries. The auditors recommended “that Rocketdyne’s fee under the Gemini subcontract be adjusted.”⁴⁷

When this report was released in the spring of 1965, the worst was already over. Rocketdyne’s performance had, in fact, begun to improve markedly in mid-1964, although as late as October Gilruth still thought an alternative source for thrusters might be a good idea. McDonnell received the first long-duration attitude maneuvering thrusters in October 1964, just five months after the new design had been released to production. By the time the audit report was issued, both the attitude and reentry control systems had been fully qualified in their Spacecraft 3 version. How greatly things had changed was shown most clearly when the long-life thrusters, not expected to be ready before Spacecraft 5, were actually installed in Spacecraft 4.⁴⁸

Qualification of the Gemini escape system, like that of the spacecraft rocket systems, was essential before astronauts could be committed to a mission. Rapid progress early in 1964, which saw the development test program concluded, augured well, as did a good start on dynamic proof-testing. A preliminary sled-ejection test on 4 June 1964, to see if hatches and hatch actuators functioned properly under abort conditions, went off without a hitch. Qualification testing began on 1 July with a sled run to simulate conditions of maximum dynamic pressure after an abort during the powered phase of launch vehicle flight. Once again, everything worked.⁴⁹

The same problem that had delayed development testing, one that had little to do with seat design, again brought the test program to a halt. Some of the pyrotechnic devices on which escape-system operation depended failed to arrive. The result was a four-month gap after the July run. In the meantime, NASA had decided to go ahead with a new test series. Sled and tower tests had been the only dynamic simulations planned for the system. Neither, however, could show the system working through its entire sequence as in a high-altitude abort. That became the purpose of a plan to eject the system from a high-flying F-106, worked out at a meeting between NASA, McDonnell, Weber Aircraft (the maker of the system), and the 6511th Test Group at El Centro, California, on 12 June. The first test, intended merely to show that the seat would work with the airplane, was set for September with the F-106 on the ground. Two flights, using production escape systems, were to follow, with the whole series to be finished in a month. Once again, however, lack of pyrotechnics caused delays. Enterprising engineers borrowed some from the ejection seat in North American’s paraglider tow test vehicle, enabling them to run the ground test on 15 October. But nothing more could be done for three months.⁵⁰

Enough pyrotechnics were on hand for another sled run on 5 November, which revealed a flaw in seat design. An instant after it had been ejected, one of the seats suffered a structural failure of its armrest and side panel that stopped the separation and recovery sequence. Seat and dummy smashed into the ground, strewing wreckage for 140 meters along the track. The hard question now was whether or not the test program had to be revised. The answer was no, provided the reworked seat structure performed well in a test approximating the most severe conditions for which the system was designed. In a sled run on 11 December, it did just that. The system came through with flying colors, bringing that part of the qualification program to an end.⁵¹

It was perhaps just as well that Gemini 2 had been so long delayed. By the end of 1964, only one of the four major parts of escape-system qualification had been completed. Still to be conducted were simulated off-the-pad ejection (Sope), personnel parachute, and high altitude ejection tests. All three resumed in January 1965, when pyrotechnics at last began to arrive.

First to get under way, on 11 January, was parachute testing. Four dummy drops and 12 live jumps from low altitudes over the next month turned up only minor problems. High-altitude testing followed.⁵² In the meantime, On 16 January (a year and a half after Sope development tests ended) Sope qualification testing began. Shortage of pyrotechnics had again been the chief culprit in the delay. The first try failed. One seat worked, but the catapult on the right-hand seat fired too soon and exploded when the seat jammed against the still partly closed hatch. Almost a month passed while all hatch actuators were modified and the results checked out. Both the redesigned actuators and the escape system proved themselves in flawless Sope tests on 12 February and 6 March.⁵³

High-altitude ejection was the last test program to resume but the first to finish. Nothing went wrong in the first test, an ejection at 4,780 meters at mach 0.65 on 28 January. Two weeks later, however, in a test at 12,000 meters at mach 1.7, the aneroid device that was supposed to trigger parachute deployment failed, although everything else worked. That device also failed to deploy the ballute on 17 February, in the first high-altitude live jump, forcing McDonnell and Weber engineers to redesign the aneroid-controlled firing mechanism. Although the aircraft ejection test did not have to be repeated, since being ejected from the F-106 did not cause the failure, the parachute test program did have to be revised. That meant an extra 10 dummy drops and 5 live jumps, which began on 2 March. The final jump, on 13 March, qualified the personnel parachute system and completed the qualification of the Gemini escape system as a whole.⁵⁴ And not a moment too soon. The launch of the third Gemini mission, the first to carry a human cargo, was only days away.

The demand for fuel cells was not so pressing in late 1964 as for thrusters and ejection seats, since Spacecraft 3 and 4 were already being converted to battery power as a result of earlier problems. GE’s redesigned fuel cell, the P3, had not at first lived up to its promise. Test sections performed erratically, their outputs tending to decay under load and their lives falling far short of requirements. This prompted NASA Headquarters to ask GPO on 10 July to provide a backup battery-power module in case fuel cells were not ready for the fifth Gemini mission. This was a drastic step, since Gemini 5 was slated for seven days; a battery installation to handle so long a mission meant a severe weight penalty and a narrow limit on what might be achieved during the flight. One of the main reasons for putting fuel cells in Gemini had been to ease constraints on such lengthy missions. GPO directed McDonnell to work out with Eagle-Picher, the battery subcontractor, a plan for a backup system.⁵⁵

Early in August, GPO enlarged the scope of the study, asking McDonnell to cover the effects of substituting batteries for fuel cells in all two-day rendezvous missions, of using Agena-supplied power in a combined long-duration and rendezvous mission, and of such changes on the fuel-cell program itself. McDonnell found the feat possible but costly, especially in weight. At a meeting on 14 August, Mathews and Burke decided to provide Spacecraft 5 with a combined system of batteries for the peak loads and fuel cells for basic power needs. If most of the experiments planned for the mission were discarded, Spacecraft 5 would only weigh 30 kilograms more with its battery-augmented system. NASA Headquarters sanctioned the change on 1 October.⁵⁶

The combined system reflected GE’s success, finally, in pinpointing the sources of fuel-cell shortcomings. GE engineers found that the life of test stacks declined as electrical load and the temperature of reactants rose. The greater the load - the amperage drawn from the stack - or the higher the inlet temperature, the shorter the stack’s life. With a constant load, a change of only 17 kelvins (30°F) in reactant temperature - 313 kelvins (103°F) instead of 330 kelvins (133°F) - more than doubled stack life, from 125 to 290 hours. Holding the temperature constant and varying the load produced similar results. With batteries to handle peak loads, a major factor in truncated fuel-cell life might have been countered.⁵⁷

These findings were based only on analysis of prior test data. Now GE revised its test program to see what effect lowered inlet temperatures and reduced loads actually had on test stacks. The results confirmed the premise. Two test units under a steady three-ampere load with reactants at 297 kelvins (75°F) lasted 1,100 and 800 hours. Further tests produced equally encouraging results at various levels of load and temperature under normal and abnormal conditions. All difficulties were not yet out of the way, but those that remained were largely matters of detail.⁵⁸

Concern about “the rapidly rising costs of the General Electric fuel cell development program, coupled with the lagging development,” persisted for a while; but, significantly, that worry was expressed in a memorandum never sent.⁵⁹ The Gemini Program Office in Houston retained some doubts about fuel-cell prospects through the early fall of 1964, urging NASA Headquarters to allow batteries to replace fuel cells in Spacecraft 6 to ensure meeting the prime objective of that mission, rendezvous with an Agena target vehicle. Headquarters demurred until 6 November, but then granted the change.⁶⁰

That decision stood, Spacecraft 6 eventually flying with battery power. In the meantime, however, the response of fuel-cell test units to lower temperatures was so marked during late summer and early fall as to convince both NASA and its contractors that the power system for Spacecraft 5 need not be augmented by batteries. That change was therefore canceled on 18 December 1964. The Gemini fuel cell completed its basic qualification test program in May 1965, three months before it flew in the fifth Gemini mission.⁶¹

Agena was still further down the line, and its lagging pace showed no signs of speeding up during 1964. Project Gemini received its first Agena D at the end of April 1964, but nearly five months passed before it was converted into GATV-5001, the first Gemini Agena Target Vehicle. Lockheed completed that effort on 24 September and transferred the vehicle to the systems test complex, where cabling it up for preliminary vehicle systems tests began the next day. Not too surprisingly, testing did not run smoothly.

The hardest and most stubborn problems centered in Agena’s command and communication (C&C) system - the electronic devices for tracking the vehicle, monitoring its subsystems, and passing commands to the vehicle in orbit. Because of Gemini’s unique demand for rendezvous and docking, Lockheed had to design and prove a new C&C system for the Gemini Agena. The new design struck GPO as very good, a judgment confirmed by a special consultant group from Stanford Research Institute, which recommended only minor changes. During testing in October, however, parts of the system started acting up. Troubleshooting got GATV-5001 through its testing, but it seemed all too likely that the C&C system suffered from basic defects in its mechanical and electronic design. The question became, as Mathews later recalled, “Should we live with what we had, or should we back off and completely redesign the configuration?” When the problems persisted, the Air Force insisted on redesign, and Lockheed finally initiated a “Ten Point Plan for C&C Equipment” in February 1965.⁶²

In the meantime, GATV-5001 had emerged from its preliminary tests in November 1964 and gone to Lockheed’s Santa Cruz Test Base for a round of captive-firing tests. First, however, the target docking adapter had to be installed. This was the unit, built by McDonnell but carried aloft by Lockheed’s Agena, to which the spacecraft would attach. When Lockheed workers hoisted the adapter into the test stand and tried to mate it with the Agena, they found it did not fit. After some struggling, they managed to get the two physically hooked together, but the wiring failed to match. The captive firing had to be postponed until January.⁶³

The test on 20 January 1965 simulated a full two-week mission. It included related firings of both primary and secondary propulsion systems, with operational data transmitted to telemetry stations at the test site and at Lockheed’s Sunnyvale plant. The propulsion systems worked well, but the C&C system again had problems. One part, the programmer time accumulator, jumped erratically, picking up almost eight extra weeks. Shipped back to Sunnyvale on 1 February, GATV-5001 lost three weeks while Lockheed tried to fix the capricious timer. A makeshift fix allowed GATV-5001 to move on to the next phase, electromagnetic and radio-frequency interference tests, while engineers continued their efforts to diagnose and cure the jumping timer. By 23 February, when the interference tests began, GATV-5001 was more than a month behind schedule.⁶⁴

Interference tests ended 9 March, but the vehicle stayed in the anechoic chamber for another week while Lockheed checked out its answer to the erratic timer and to a telemetry synchronization problem that had also cropped up. On 18 March, GATV-5001 moved to the systems test complex for a planned six days of “minor” modifications: filters were to be installed in the command controller (another part of the C&C system) and the forward auxiliary rack (which supported the target docking adapter and housed most of the C&C gear) was to be aligned. These two tasks proved to be more than minor. The first eventually required a complete redesign, the second extensive machining. The result was another lost month. By the end of March, GATV-5001 was 66 days behind schedule.⁶⁵

Final systems testing got under way on 9 April and ended with a simulated flight on 6 May. On 27 May, the Air Force and Aerospace team found GATV-5001 formally unacceptable for Gemini, since FACI (first article configuration inspection) from 10 to 26 May had shown that it was not flightworthy. SSD took the vehicle anyway, but conditionally. Lockheed was expected to correct all defects; some were merely matters of paperwork, but others, like propulsion and C&C systems qualification, were major efforts. GATV-5001 was then flown to the Cape on 29 May, to be used as a development test vehicle.⁶⁶

In the meantime, the first Atlas booster for Gemini had joined the program on 1 December in San Diego. It had then been shipped by truck to Cape Kennedy, a six-day trip. It was erected on complex 14 a week later, to help in checking out the launch pad and ground support equipment. Finished with that by 11 February, the Atlas was moved to a hangar, there to be modified and stored until GATV-5002 arrived.⁶⁷

A Vote of Confidence

On Tuesday afternoon, just a few hours after the launch of Gemini 2, the program received another vote of confidence. Although the second launch had been long delayed, the nature of the delays in no way cast doubts on Gemini itself; NASA and its contractors decided that Gemini missions should he launched at two-month intervals, instead of the three-month cycle then planned.

In September 1964, the Air Force had not only convinced NASA that GLV-2 ought to fly, but also proposed to speed up the program by launching every two months. Although the Vertical Test Facility at Martin-Baltimore had been designed to handle two launch vehicles at once, only one of these test cells was working. The Air Force suggested opening the second cell to speed up launch vehicle deliveries. SSD Commander Funk assured his Gemini colleagues that the Cape crew could handle launches only 60 days apart.

LeRoy E. Day, Headquarters Gemini Test Director, took charge of a task force to canvass spacecraft, launch vehicle, and target vehicle contractors about the practicality of the plan. A two-month study convinced Day and his group that it could be done. Although NASA’s checkout crew at Cape Kennedy expressed a measure of skepticism based on their experiences in Project Mercury and the opening stages of Gemini, the Gemini Program Office had more faith. GPO had, in fact, been thinking of less time between launches when it imposed revised test and checkout procedures in St. Louis and at the Cape early in 1964. When Day presented his findings to Gemini’s top echelon on 19 January 1965, they bought the plan and wanted it put into effect by the fifth mission. This vote of confidence in Gemini was founded on a technological judgment, and in that sense it was fully justified. Later events were to show that fitting astronaut training into the shorter schedule was a harder task, although it produced no problems that could not he surmounted.⁶⁸

As 1965 dawned, Project Gemini had cleared most of the hurdles in its path. The past year had seen its last serious development problems overcome. Agena was perhaps not as far along as it should be, but there was plenty of talent at hand to put that in order. The repeated setbacks suffered by GLV-2 could only be seen as acts of God, not defects in technology. That could not be said of its failure on 9 December, but little more than a month of hard work was needed to put matters right. The second Gemini mission, on 19 January 1965, almost matched the first, on 8 April 1964, in the quality of performance. Gemini’s spacecraft and launch vehicle had been proved. All that remained, the last hurdle, was sending men aloft. Although the publicly scheduled date for Gemini 3 was the second quarter of 1965, Charles Mathews told the Gemini Management Panel shortly after the flight of Gemini 2 that late March looked like a good bet.⁶⁹