On 8 December 1964, at Bunker Hill Air Force Base in Indiana, a Convair B-58 Hustler sat on the ramp going nowhere. Cranes were rigged alongside it. Then one of its three crew stations fired itself off the aeroplane on a rocket, climbed away on a column of smoke, and came down under a parachute.

It was a static test of the Stanley escape capsule, and it is the clearest photograph ever taken of a problem that still has no tidy answer: how do you get a crew out of an aircraft that is too big, too fast or too high for a man and a parachute?

For a single-seat fighter the answer has been settled since the 1950s. A seat, a cartridge, a rocket, one person. For bombers the US Air Force tried three completely different answers at the same time, on three aircraft that were all in service together, and only one of them is still flying.

The problem scales badly

An ejection seat solves one person at a time. That is fine in a fighter, where the crew sits in a row under a single canopy and everybody can be out in under two seconds. A large aircraft is a different engineering problem entirely. The crew is bigger, it is spread over two decks, some of it sits below the floor, and historically some of it sat facing backwards.

Three answers were tried. Encapsulate everybody in a single armoured compartment and throw the compartment clear. Give everybody a seat, but fire half of them through the roof and half through the floor. Or simply give everybody a seat and accept a small crew. The F-111 Aardvark is the type most people remember for the first approach, and its module is now a museum exhibit.

The capsule era

Der B-58 Hustler entered service in 1961 with ordinary ejection seats, and the US Air Force quickly concluded that leaving a Hustler at its design speed was close to unsurvivable. The National Museum of the US Air Force puts it plainly: ejection at very high speed proved extremely dangerous, so Stanley Aircraft developed an ejection capsule that was retrofitted into the aircraft in late 1962.

What Stanley built was not a seat with a lid. Pulling the handle closed two clamshell doors and sealed the occupant inside an airtight, pressurised shell fed by its own oxygen supply. Inside, he could ride the capsule out at twice the speed of sound and from as high as 70,000 feet, or he could stay sealed in it and simply wait until the aircraft descended to an altitude where he no longer needed a pressure cabin. The pilot’s capsule went further still: it contained a control stick and the controls needed to fly the aeroplane, so he could keep flying it from inside a sealed pod. If the capsule came down in water, manually operated flotation cells turned it into a life raft.

The same idea appeared on the XB-70 Valkyrie, whose crew sat in individually encapsulating seats. On 8 June 1966 the second XB-70 collided with NASA’s F-104N chase aircraft. North American test pilot Al White ejected in his capsule and, in NASA’s own words, received serious injuries in the process. Co-pilot Major Carl Cross was unable to eject and died in the crash, and NASA pilot Joe Walker died in the F-104. The encapsulation sequence has to retract the seat and close the shell before the rocket can fire, and that is exactly the step that can fail.

The capsules were proved the hard way. Between 1962 and 1964 the test programme ejected live animals at supersonic speed to see whether anything inside would survive. Two independent accounts agree on six bears and one chimpanzee across seven tests, the bears chosen because a young black bear is close to a man in size and weight. The best-documented run was on 21 March 1962, when a two-year-old American black bear was ejected at 35,000 feet at Mach 1.3 and descended for seven minutes and 49 seconds. He landed unhurt and was then put down so that his organs could be examined. He is generally recorded as the first living creature to survive a supersonic ejection.

Ten thousand pounds of answer

Capsules worked. They were also enormously heavy, and the US Air Force knew exactly how heavy. In 1988 Captain James A. Hubert of the Air Force Wright Aeronautical Laboratories published a paper arguing for a new, lighter capsule, and he opened by cataloguing everything wrong with the old ones.

His numbers are the reason the idea died. The F-111 capsule weighed 3,300 lb (1,500 kg) for a crew of two. The B-1 capsule weighed nearly 10,000 lb (4,500 kg) for a crew of six, which works out at roughly the same 1,700 lb per person. A single ejection seat of the kind Hubert was comparing against came in at 600 to 700 lb. You were paying two and a half times the weight, per head, for the privilege of a pressurised shell.

Weight was not the only bill. Both capsules were structurally integral to the fuselage and used explosive shaped charges to cut themselves free, which meant, in Hubert’s words, that all capsule subsystems were accessible only through the skin on the fuselage. Routine refurbishment involved removing much of that skin and replacing every pyrotechnic component in the system. A maintenance plan that begins with taking the aeroplane apart is not a maintenance plan anybody wants twice.

Fifteen hundred feet above the desert

On 29 August 1984 the second B-1A prototype was flying minimum-control-speed tests out of Edwards. As the wings were swept forward, fuel should have been moved forward with them. It was not. The centre of gravity went beyond the aft limit, the aircraft pitched up and departed controlled flight.

The crew’s own rules required them to leave the aircraft no lower than 10,000 feet. They did not get out at 10,000 feet.

The capsule fired and the crew survived the separation. What failed was the landing. According to NASA, a malfunctioning mechanism meant the parachutes could not reposition the capsule so that it would touch down softly on its inflatable landing bag. Instead the module struck the ground at a steep nose-down angle, hurling the crew forward into the control panels at around 40 g.

Rockwell International’s chief test pilot, Tommie D. “Doug” Benefield, died when his seat tore away from the floor of the capsule. Reynolds, the aircraft commander, suffered serious back injuries. Flight test engineer Captain Otto J. Waniczek Jr., who was not wearing his helmet or harness, was also seriously hurt. Reynolds, who went on to command the Air Force Flight Test Center, has since used the accident as a teaching case, and his verdict on the cause is as blunt as it gets.

One thing worth being precise about, because it is easy to get backwards: the B-1B had already been specified with conventional ejection seats before this crash. The fourth B-1A prototype flew with individual seats, and the production B-1B carries four. The capsule was not abandoned because it failed in the desert in 1984. It was abandoned because of the weight, the cost and the pyrotechnic servicing burden, and the crash came afterwards.

Four up, two down

Der B-52 Stratofortress took the third path, and it is the strangest of the three. The G and H models carried six crew stations, each with its own hatch and its own ejection seat. The pilot, the copilot, the electronic warfare officer and the gunner sat on the upper deck and ejected upward. The navigator and the radar navigator sat on the lower deck, below and in front of them, and ejected downward, through the floor.

Firing a man at the ground is a reasonable thing to do if there is enough ground beneath him to allow a parachute to open. The question is how much. Published figures differ, and it is worth saying so rather than picking one: the minimum has been given as 200 feet, as 250 feet in level flight, and as 400 feet by Jay Lacklen, a former USAF B-52 pilot, who describes the navigators as needing four hundred feet of ground clearance and level flight to get one swing in the parachute before hitting the ground. Call it somewhere between 200 and 400 feet and treat anyone quoting a single number with caution.

Below that height the lower deck has no seat solution at all. What crews describe instead is a manual exit through the hole that the other downward seat has just left behind. Barry Schiff, riding in the B-52’s instructor seat, which has no ejection capability, was briefed to dive headfirst through one of the holes remaining after someone on the lower deck had ejected. It is not a procedure anybody could find in a technical order for us, but it is what the men who flew the aeroplane say they were told.

It has been done for real. On 3 September 1975 a B-52G called Exalt 15 lost its right wing after a fuel leak, near Aiken, South Carolina. The pilots ejected at roughly 120 degrees of roll. The electronic warfare officer and the radar navigator went after them; the radar navigator’s seat got him out of the aircraft but never separated from him, and he died. The navigator, Hector Marquez, was the last one out, and his account of it was published decades later in the US Air Force’s own Air and Space Power Journal.

The gunner is gone now. Strategic Air Command announced the deletion of the B-52 gunner in late 1991 and the last one flew on 1 October that year. The station itself survives as a sixth seat with no ejection capability, flown by anyone riding along. In the 2008 Guam B-52H accident the sixth man aboard was a flight surgeon in the Number 6 crew position. Whether the gunner’s seat hardware was physically pulled out of the fleet, we could not establish from any source and will not guess at.

Two seats, two people

Then there is the simplest answer of all, which is to need fewer seats. The B-2 Spirit flies with a crew of two, and it has two ACES II zero-zero ejection seats. That is the whole escape system. The US Air Force’s own accident board for the 2021 Whiteman mishap notes a crew of two compared with the B-1B’s four and the B-52’s five, and the pattern of the escape system follows directly from that number.

ACES II stands for Advanced Concept Ejection Seat, designed by Douglas Aircraft and now a Collins Aerospace product. Its published escape envelope runs from zero airspeed and zero altitude out to 600 knots equivalent airspeed, and a sequencer picks one of three deployment modes automatically from the conditions it senses at the moment of firing: a near-instant parachute at low speed, a drogue-first sequence at high speed, and a delayed deployment at high altitude that keeps the occupant attached to the seat and its emergency oxygen until he has fallen into thinner, slower conditions. The B-1B’s four seats are linked: in automatic mode, one crew member pulling the handle fires all four in a timed sequence designed to stop them colliding, in about two seconds.

It works. When a B-2 crashed on take-off at Andersen Air Force Base in Guam in February 2008, both crew members ejected safely. NASA’s own safety case study records the pilot with minor injuries and the co-pilot with spinal compression fractures. Nobody died.

And the airliner has nothing

The question that follows from all of this is the one every passenger eventually asks: why does a 180-seat airliner have no escape system whatsoever?

Start with the weight, because the numbers are already on the table. Six hundred to seven hundred pounds per person for a seat. Multiply by 180. You have added more than fifty tonnes to an aircraft that would then need 180 jettisonable hatches cut into the crown of a pressurised fuselage, one directly above each seat, each with its own thruster. There is no version of that aeroplane that flies.

The better argument is the one Boeing publishes every year without meaning it as an argument. In its Statistical Summary of Commercial Jet Airplane Accidents, Boeing breaks fatal accidents down by phase of flight. For 2016 to 2025, take-off and initial climb account for 22 per cent of fatal accidents and final approach and landing for 48 per cent. Landing alone produces 35 per cent of fatal accidents while occupying about 1 per cent of flight time. Cruise is 57 per cent of the time in the air and 10 per cent of the fatal accidents.

Roughly seven fatal accidents in ten happen in the few minutes near the ground where there is neither the altitude nor the seconds to use an escape system, and the long quiet hours at 35,000 feet, where a capsule or a seat could theoretically do something useful, are where almost nothing goes wrong. An escape system for airliners would be a fifty-tonne solution aimed at the part of the flight that is already safe.

Which leaves the bombers where they started. The B-52 still flies with its crew split between a deck that fires upward and a deck that fires downward, an arrangement designed when the aircraft was new and the Soviet Union was young. The B-1B and the B-2 fly on ACES II, the same seat as the F-15, Die F-16 und die A-10, because in the end the cheapest way to save a bomber crew turned out to be to build a bomber that needs a smaller one.

Captain Hubert was not writing an obituary in 1988. He was writing a proposal, for a capsule light enough to be worth having. Nearly forty years later, the seat under a B-2 pilot is still ACES II.

Sources: National Museum of the United States Air Force, B-58 Escape Capsule and XB-70 Valkyrie fact sheets; NASA Dryden X-Press, “Safety Lessons”; Capt James A. Hubert, Air Force Wright Aeronautical Laboratories, “Light Weight Escape Capsule for Fighter Aircraft”, NASA SOAR proceedings, 1988; US Air Force B-52, B-1B and B-2 fact sheets; USAF Accident Investigation Board report, B-2A 89-0129, 14 September 2021; NASA Safety Center case study, 2008 B-2 mishap; Hector Marquez in Air and Space Power Journal, Fall 2021, Air University Press; Peter Grier, “The B-52 Gunners”, Air Force Magazine; Boeing, Statistical Summary of Commercial Jet Airplane Accidents, Worldwide Operations 1959-2025; The Ejection Site.