Walter F. Burke of McDonnell summarized the company's studies of the redesigned Mercury spacecraft for Space Task Group's senior staff. McDonnell had considered three configurations: (1) the minimum-change capsule, modified only to improve accessibility and handling, with an adapter added to carry such items as extra batteries; (2) a reconfigured capsule with an ejection seat installed and most of the equipment exterior to the pressure vessel on highly accessible pallets; and (3) a two-man capsule, similar to the reconfigured capsule except for the modification required for two rather than one-man operation. The capsule would be brought down on two Mercury-type main parachutes, the ejection seat serving as a redundant system. In evaluating the trajectory of the two-man capsule, McDonnell used Atlas Centaur booster performance data.

McDonnell submitted to Manned Spacecraft Center the detail specification of the Mercury Mark II spacecraft. A number of features closely resembled those of the Mercury spacecraft. Among these were the aerodynamic shape, tractor rocket escape tower, heatshield, impact bag to attenuate landing shock, and the spacecraft-launch vehicle adapter. Salient differences from the Mercury concept included housing many of the mission-sustaining components in an adapter that would be carried into orbit rather than being jettisoned following launch, bipropellant thrusters to effect orbital maneuvers, crew ejection seats for emergency use, onboard navigation system (inertial platform, computers, radar, etc.), and fuel cells as electrical power source in addition to silver-zinc batteries. The long-duration mission was viewed as being seven days.

NASA laid down guidelines for the development of the two-man spacecraft in a document included as Exhibit "A" in NASA's contract with McDonnell. The development program had five specific objectives: (1) performing Earth-orbital flights lasting up to 14 days, (2) determining the ability of man to function in a space environment during extended missions, (3) demonstrating rendezvous and docking with a target vehicle in Earth orbit as an operational technique, (4) developing simplified countdown procedures and techniques for the rendezvous mission compatible with spacecraft launch vehicle and target vehicle performance, and (5) making controlled land landing the primary recovery mode. The two-man spacecraft would retain the general aerodynamic shape and basic systems concepts of the Mercury spacecraft but would also include several important changes: increased size to accommodate two astronauts; ejection seats instead of the escape tower; an adapter, containing special equipment not needed for reentry and landing, to be left in orbit; housing of most system hardware outside the pressurized compartment for ease of access; modular systems design rather than integrated; spacecraft systems for orbital maneuvering and docking; and a system for controlled land landing. Target date for completing the program was October 1965.

Gemini Project Office (GPO) decided that seat ejection was to be initiated manually, with the proviso that the design must allow for the addition of automatic initiation if this should later become a requirement. Both seats had to eject simultaneously if either seat ejection system was energized. The ejection seat was to provide the crew a means of escaping from the Gemini spacecraft in an emergency while the launch vehicle was still on the launch pad, during the initial phase of powered flight (to about 60,000 feet), or in case of paraglider failure after reentry. In addition to the seat, the escape system included a hatch actuation system to open the hatches before ejection, a rocket catapult to propel the seat from the spacecraft, a personnel parachute system to sustain the astronaut after his separation from the seat, and survival equipment for the astronaut's use after landing. At a meeting on March 29, representatives of McDonnell, GPO, Life Systems Division, and Flight Crew Operations Division agreed that a group of specialists should get together periodically to monitor the development of the ejection seat, its related components, and the attendant testing. Although ejection seats had been widely used in military aircraft for years, Gemini requirements, notably for off-the-pad abort capability, were beyond the capabilities of existing flight-qualified systems. McDonnell awarded a $1.8 million subcontract to Weber Aircraft at Burbank, California, a division of Walter Kidde and Company, Inc, for the Gemini ejection seats on April 9; a $741,000 subcontract went to Rocker Power, Inc., Mesa, Arizona, on May 15 for the escape system rocket catapult.

Representatives of McDonnell, Weber Aircraft, Gemini Procurement Office, Life Systems Division, Gemini Project Office, and US Naval Ordnance Test Station, China Lake, California, concluded plans for development testing of the spacecraft ejection seat. Requirements peculiar to the Gemini spacecraft, in particular off-the-pad abort capability, caused the plan to stress testing from a stationary tower early in the test program. The purpose of these simulated off-the-pad ejection tests was to investigate the effects of varying the center of gravity on the trajectory of the ejected seat and to optimize the timing of the recovery sequence. Tower tests began July 2. They were to be followed by rocket sled ejection tests to investigate simultaneous ejection with open hatches at maximum dynamic pressure. Sled tests actually began on November 9, before tower tests had been completed.

Five ejections were completed by the first week of August. The tests revealed difficulties which led to two important design changes: the incorporation of a drogue-gun method of deploying the personnel parachute and the installation of a three-point restraint-harness-release system similar to those used in military aircraft. August 6-7 representatives of Manned Spacecraft Center and ejection system contractors met to review the status of ejection seat design and the development test program. They decided that off-the-pad ejection tests would not be resumed until ejection seat hardware reflected all major anticipated design features and the personnel parachute had been fully tested. Design changes were checked out in a series of bench and ground firings, concluding on August 30 with a successful inflight drop test of a seat and dummy. Off-the-pad testing resumed in September.

A study group formed at the Gemini mock-up review of August 15-16 met to review the ejection seat development program. McDonnell reported the successful completion of redesign and testing which cleared the way for resumption of off-the-pad developmental testing. McDonnell described the major outstanding design task as the determination of the dynamic center of gravity of the seat-man combination under expected acceleration profiles.

7 followed on September 20. Though primarily successful, these tests revealed some problems. The seat-structure thrust pad required reanalysis and redesign. Simulated off-the-pad testing was temporarily halted until a final configuration rocket catapult became available. A rocket motor test on January 4, 1963, demonstrated the structural integrity of the thrust-pad area, and simulated pad ejection tests resumed the following month.

Despite its designation, this test did not call for seats actually to be ejected. Its purpose was to provide data on the aerodynamic drag of the test vehicle and to prove the test vehicle's structural soundness in preparation for future escape system tests. The test vehicle, mounted by boilerplate spacecraft No. 3 (a welded steel mock-up of the Gemini spacecraft aerodynamically similar to the flight article), was a rocket-propelled sled running on tracks. Although test objectives were achieved, the boilerplate spacecraft was severly damaged when one of the sled motors broke loose and penetrated the heatshield, causing a fire which destroyed much instrumentation and equipment. Despite repairs required for the boilerplate and major modification or rebuilding of the sled, Gemini Project Office foresaw no delay in the sled test program.

A 10-percent fluctuating-pressure model of the Gemini spacecraft completed its exit configuration test program in the mach number range of 0.6 to 2.5, the region of maximum dynamic pressure. On January 15, 1963, a Gemini spacecraft dynamics stability model also completed its test program providing dynamic stability coefficients for the spacecraft reentry at mach numbers 3.0 to 10. These tests completed all the originally scheduled wind tunnel testing for Project Gemini; however, three additional test programs had been initiated. These included additional testing of the spacecraft 20-percent ejection seat model, testing of the astronaut ballute model to obtain data for design of the astronaut stabilization system, and testing of the rigid frame paraglider model to determine optimum sail configuration.

Simulated off-the-pad ejection test No. 8 was conducted at Naval Ordnance Test Station. Two dummies were ejected, and for the first time the test incorporated a ballute system. The ballute (for balloon + parachute) had been introduced as a device to stabilize the astronaut after ejection at high altitudes. Ejection seat and dummy separated satisfactorily and the personnel parachute deployed properly; but faults in the test equipment prevented the canopy from fully inflating. The ballute failed to inflate or release properly on either dummy. As a result, the parachute was redesigned to ensure more positive inflation at very low dynamic pressures. The redesigned chute was tested in a series of five entirely successful dummy drops during March.

McDonnell and Weber Aircraft had completely redesigned the blackboard and mechanism linkage to obtain more reliable load paths and mechanism actuation, and to eliminate the 'add-on' character of the many features and capabilities introduced during seat development which contributed to the unsuccessful test in February. The new design was proved in a series of tests culminating in a preliminary ejection test on April 22. Test No. 9 was followed by test No. 9a on May 25. Both tests were completely successful. Test Nos. 10 and 11 (July 2, 16) completed the development phase of pad ejection testing. Both were dual ejection tests. No. 10 was completely unsuccessful, but No. 11 was marred by the failure of a seat recovery chute (not part of the spacecraft ejection system), resulting in major damage to the seat when it hit the ground.

At a Gemini Abort Panel meeting, McDonnell reported the possibility of dropping the mode 2 lower abort limit to 35,000 to 40,000 feet. McDonnell also presented computer data on studies using a combination of mode 2 and mode 1 for launch to T + 10-second aborts; during this period, mode 1 abort might not be adequate. Current Gemini abort modes: mode 1, ejection seats - from pad to 70,000 feet; mode 2, booster shutdown/retrosalvo - 70,000 to approximately 522,000 feet; mode 3, booster shutdown/normal separation - from approximately 522,000 feet until last few seconds of powered flight.

Sled test No. 2, the first dynamic dual-ejection test of the Gemini escape system, was run at China Lake. Both seats ejected and all systems functioned properly. The test was scheduled to be rerun, however, because the sled failed to attain high enough velocity. The purpose of sled tests in the ejection seat development program was to simulate various high-altitude abort situations. Sled test No. 3 was successfully run on August 9. Further tests were delayed while the ejection system was being redesigned. A modified egress kit was tested in two dummy drops on December 12, with no problems indicated. Gemini Project Office directed McDonnell to proceed with plans for the next sled test. Developmental sled testing on the escape system, incorporating the redesigned egress kit and a soft survival pack, resumed on January 16, 1964, with test No. 4; all systems functioned normally. Test No. 5, the planned repetition of test No. 2, brought developmental sled testing to an end on February 7.

A meeting was held to discuss ejection seat system problems. Of major concern was the ejection seat ballute that was planned to stabilize the astronaut after he ejected and separated from the seat. Wind tunnel test data had suggested two problem areas: the ballute was failing at supersonic speeds and was not opening at subsonic speeds. Increasing the diameter and lengthening the riser lines improved performance considerably. A major system change recommended at the meeting was the incorporation of provisions for automatic separation of the seat backboard and egress kit before touchdown; Gemini Project Office directed McDonnell to study the feasibility of this recommendation.

McDonnell began spacecraft pyrotechnic hatch firing tests, using boilerplate No. 3A, with a single-hatch firing test. The hatch opened and locked, but opening time was 350 milliseconds, 50 milliseconds over the allowable time. This test was followed, on February 10, by a dual-hatch firing test with satisfactory results. The boilerplate spacecraft was prepared for shipment to Weber Aircraft to be used in the qualification program of the ejection seat system.

This was a preliminary test to prove that hatches and hatch actuators would function properly under abort conditions; no ejection was attempted. The test was successful, and qualification testing proper began on July 1 with test No. 7. The test simulated conditions of maximum dynamic pressure following an abort from the powered phase of Gemini flight, the vehicle being positioned heatshield forward as in reentry. Both seats ejected and all systems functioned as designed. Further sled testing was delayed by slow delivery of pyrotechnics; sled test No. 8 was not run until November 5. This test revealed a structural deficiency in the ejection seat. When the feet of one of the dummies came out of the stirrups, the seat pitched over and yawed to the left, overloading the left side panel. The panel broke off, interrupting the sequence of the ejection system, and the seat and dummy never separated; both seat and dummy were destroyed when they hit the ground. Representatives of Manned Spacecraft Center and McDonnell met during the week of November 15 to consider revising the test program as a result of this failure. They decided to conduct test No. 9 under conditions approximating the most severe for which the ejection system was designed, in order to demonstrate the adequacy of the reworked seat structure. Test No. 9 was run on December 11, successfully demonstrating the entire ejection sequence and confirming the structural redesign. This brought the qualification sled test program to an end.

Representatives of NASA, McDonnell, Weber Aircraft, and Air Force 6511th Test Group met to define the basic objectives of a program to demonstrate the functional reliability of the Gemini personnel recovery system under simulated operational conditions. Such a program had been suggested at a coordination meeting on the ejection seat system on October 30, 1963. The planned program called for the recovery system to be ejected from an F-106 aircraft, beginning with a static ground test in September, to demonstrate compatibility between the recovery system and the aircraft. Two full system tests, using a production configuration recovery system, would complete the program in about a month. The program was delayed by the unavailability of pyrotechnics. The static ground test was successfully conducted October 15, using pyrotechnics from the paraglider tow test vehicle (TTV) seat. The TTV seat pyrotechnics were adequate to demonstrate system/aircraft compatibility but lacked certain items required for full system test. Full system testing accordingly did not begin until January 28, 1965.

Flight Crew Support Division objected to McDonnell procedures for conducting ejection seat sled tests because they were not adequate to give confidence in manned use of the seats. The dummies were being rigged with extreme restraint-harness tensions and highly torqued joints which could not be achieved with human subjects. McDonnell was requested to review the situation and prepare a report for Gemini Program Office.

After a long delay because pyrotechnics were not available, simulated off-the-pad ejection (SOPE) qualification testing resumed with SOPE No. 12. Performance of the left seat was completely satisfactory, but the right seat rocket catapult fired prematurely because the right hatch actuator malfunctioned. The seat collided with the hatch and failed to leave the test vehicle. All hatch actuators were modified to preclude repetition of this failure. After being tested, the redesigned hatch actuators were used in SOPE No. 13 on February 12. The test was successful, and all systems functioned properly. This portion of the qualification test program came to a successful conclusion with SOPE No. 14 on March 6. The complete ejection system functioned as designed, and all equipment was recovered in excellent condition.

This was the first ejection in flight to demonstrate the functional reliability of the Gemini personnel recovery system. The recovery system was ejected from an F-106 at an altitude of 15,000 feet and a speed of mach 0.72. Original plans had called for an ejection at 20,000 feet, but the altitude was lowered because of a change in the Gemini mission ground rules for mode 1 abort. Both seat and dummy were recovered without incident. The program ended on February 12 with HAET No. 3, although the dummy's parachute did not deploy. An aneroid device responsible for initiating chute deployment failed, as did an identical device on February 17 during qualification tests of the personnel parachute. These failures led to redesign of the aneroid, but since the failure could not be attributed to HAET conditions, Gemini Program Office did not consider repeating HAET necessary. All other systems functioned properly in the test, which was conducted from an altitude of 40,000 feet and at a speed of mach 1.7.

The ballute failed to deploy because of a malfunction of the aneroid device responsible for initiating ballute deployment. The identical malfunction had occurred during the high-altitude ejection test on February 12. These two failures prompted a design review of the ballute deployment mechanism. The aneroid was modified, and the qualification test program for the personnel parachute was realigned. In place of the remaining 23 low-altitude live jump tests, 10 high-altitude dummy drops using the complete personnel parachute system (including the ballute), followed by five high-altitude live jumps, would complete the program. The 10 dummy drops were conducted March 2-5 at altitudes from 12,000 to 18,000 feet and at speeds from 130 to 140 knots indicated air speed (KIAS). All sequences functioned normally in all tests but one: in that one, the ballute failed to leave its deployment bag (corrected by eliminating the bag closure pin from the design) and the backboard and egress kit failed to separate (resolved by instituting a special inspection procedure). The five live jumps were conducted March 8-13 at altitudes from 15,000 to 31,000 feet and at a speed of 130 KIAS. Again all test were successful but one, in which the ballute failed to deploy. After a free fall to 9200 feet, the subject punched the manual override, actuating the personnel parachute. This series completed qualification of the personnel parachute and also of the overall Gemini escape system.