By 1966, the Air Force was considering developing its own lifting-body configuration to add to the program. To gain experience in engineering and flight planning useful later in developing and testing its own lifting body, the Air Force participated in the M2-F2 project. Heading up the Air Force's lifting-body effort was program manager Robert G. "Bob" Hoey, who had extensive experience with the X-15 and experimental flight testing. Air Force Captain John Durrett assisted with general engineering. In January 1970, after the X-15 program ended, program engineer Johnny Armstrong joined the Air Force's lifting-body team. Although the team was relatively young, it had considerable experience in experimental flight testing.
Hoey and Armstrong had worked together as Air Force flight-test engineers in the highly successful X-15 program. Before he became NASA director at the Flight Research Center in 1959, Paul Bikle had served as technical director for the Air Force Flight Test Center at Edwards AFB. Hoey, who had been at Edwards approximately twelve years, had a good relationship with NASA management, including Bikle. The success of the X-15 program made it easy for us at NASA to consider the Air Force's lifting-body team as "the experts." Bertha Ryan worked closely with Hoey and the rest of his team as the NASA stability and control engineer and aerodynamicist for the M2-F2. Excellent communication existed between the lifting-body teams, with the Air Force offices only about a mile down the road from those of NASA.
Hoey and his team modified an X-15 simulator to use for training pilots and planning the first 15 flights of the M2-F2, while we at NASA were bringing up our own M2-F2 simulator and changing computers. Hoey's team loaded its simulator with the M2-F2 data from the wind-tunnel tests. Before the first flight of the M2-F2, Milt Thompson spent many hours on the simulator, becoming well acquainted with the vehicle's stability limits, including the boundaries for pilot-induced oscillation (PIO) and roll-control reversal.
First Flight of the M2-F2
For its first glide flight on 12 July 1966, the M2-F2 was mated with the B-52 mothership, carried aloft, then launched on a north heading at 45,000 feet. The launch was very mild, Milt Thompson reported, with at most 28 degrees of right roll following launch. The flight plan called for two 90-degree turns to the left with a landing to the south on the lakebed's Runway 18. He made a simulated landing starting at 22,000 feet, coming level at 19,000 feet between the two 90-degree turns, firing the peroxide rocket during the landing simulation with no noticeable changes in attitude (orientation) with thrust.
Using the manual control to lower the interconnect ratio between the ailerons and rudder to 0.4 on the pushover at altitude, Milt felt that the vehicle's roll response was not great enough as he tried to begin the second 90-degree turn as planned at 16,000 feet and 190 knots. He increased the interconnect ratio to 0.6, in effect adding rudder as he began the final turn. During the turn's pushover, the M2-F2 developed an uncomfortable lateral-directional oscillation.
Milt tried to turn the interconnect ratio down, but, as he later said, "I turned it the wrong way just as I was turning final." Rather than decreasing it, he had accidentally increased it to 1.25. The oscillations increased to 90 degrees, the flight films showing the vehicle swinging madly from side to side. The view through the windshield inside the M2-F2, as captured on film by the camera behind Milt in the cockpit, showed a horizon rolling rapidly from vertical to vertical. Quickly realizing the error, Milt reduced the interconnect ratio back to 0.4, which decreased rudder. He took his hand off the control stick, and the oscillations damped out rapidly.
He reached a pre-flare speed of 280 knots at 1,200 feet altitude. At flare completion, speed was 240 knots. Landing gear was deployed at 218 knots, accompanied by mild pitch transient, or change in attitude. Milt landed the M2-F2, the vehicle touching down at the exact spot planned at 164 knots, then coasting 1.5 miles across the lakebed. Lasting not quite four minutes, the first flight of the M2-F2 appeared to be an unqualified success.
During the debriefing afterwards, Milt apologized for nearly losing control of the vehicle by moving the interconnect wheel the wrong way. Later, we found two errors had been made in the simulator. First, by employing the Air Force's X-15 simulator cockpit, we had inadvertently used the X-15's speed brake handle instead of the M2-F2 pilot's interconnect wheel. Second, the interconnect control direction was the reverse of what was in the actual aircraft. In short, Milt had been practicing with a simulator that did not represent the M2-F2, a serious foul-up that both we and Bob Hoey's Air Force simulator team found embarrassing. What might have been a disaster in the air was averted by Milt's quick adaptability and knowledge of the lifting body's characteristics. Realizing that the interconnect settings were incorrect, he had taken immediate corrective action.
One more error-this time, a minor one-was made during the M2-F2's first flight. Vic Horton had been onboard the B-52, his only task to turn on the 16-mm camera 10 seconds before launch to film the top of the M2-F2 as it fell away from the B-52. He forgot to turn on the camera. After the crew briefing for the second M2-F2 flight, Wen Painter and Berwin Kock presented Horton with a "Launch Panel Camera Switch Simulator." It was made out of a cardboard box and had a large lever marked CAMERA ON/OFF. As the crew laughed, Horton turned the lever to CAMERA ON. A banana rolled out. The crew howled with laughter. Horton grabbed the banana and threw it at Painter and Kock.
 Milt's Last Lifting-Body Flight
On 2 September 1966, Milt Thompson made his fifth flight on the M2-F2, his last lifting-body flight. He had decided to make a career change, moving into management with the Flight Research Center. NASA lost a superb research pilot when Milt Thompson retired from the cockpit, but we later reaped great benefit from his experience when he became chief of the research projects office at the Center in January 1967, responsible for all flight projects, including those of the X-15 and the lifting bodies. Milt never spoke publicly in those days about why he made the career change. Some surmised that he might have felt he had used up his "nine lives" in the close calls he had had as a pilot.
One of those close calls happened on 20 December 1962, about eight months before he flew the M2-F1. He was flying an F-104 chase aircraft. As he prepared for landing, he was lowering the flaps when the mechanical cross link between the right and left flaps broke. The flaps were stuck, one up and one down, and the F-104 started rolling. Somehow Milt managed to maintain altitude while the aircraft made a series of 360-degree rolls across the sky. He tried recycling the flaps and resetting the circuit breakers during the rolls, but to no avail.
As the F-104 continued to roll, Milt managed to steer it over the bombing range at Edwards AFB. Since the aircraft was only about 5,000 feet above the ground, Milt made a carefully timed ejection when the cockpit was pointed upward. He floated down in his parachute, landing safely on the bombing range.
The F-104 went down about two miles from where Milt had landed, the aircraft digging a huge black hole in the ground upon impact. Milt gathered up his parachute and walked half a mile along the edge of the bombing range to a road leading to the rocket test site on Leuhman Ridge. He stuck out his thumb and hitched a ride in a pickup truck that brought him back to the NASA building.
When he walked into the pilot's office, a full-scale search was already underway. Helicopters were landing at the crash site. No one had seen Milt eject or spotted his parachute descending. The assumption was that his body would be found in the wreckage of the F-104. The mood changed from heavy sadness to surprised relief when Milt walked into the office.
After he retired as a NASA pilot in 1966, Milt later made (to my knowledge) only one public statement about his career change, and he made it in his book, At The Edge of Space, published in 1992. There, he explains it was boredom, not fear, that led to his career change, saying that he had made up his mind and even discussed the career change with Bikle nearly two months earlier, before he began flying the M2-F2 in July.
"I felt that the exciting programs were winding down," he wrote, "and I could not see any new challenging programs coming up in the near future. I really enjoyed the challenge of an X-15 flight or a lifting-body flight, but I was getting bored with....
....the routine proficiency flying that was required between research flights. When a pilot gets bored with flying, it is time to quit." 1
Gentry Fast Forwards
By 12 October 1966, the M2-F2 had been flown ten times-five by Milt Thompson, two by NASA research pilot Bruce Peterson, and three by Air Force test pilot Don Sorlie. Sorlie also got into a PIO problem on his first flight in the M2-F2, but he had planned ahead of time what he would do if it happened, and he had sufficient altitude to execute a full recovery. After two more flights with no additional problems, Sorlie gave the okay for Air Force research pilot Jerry Gentry to fly the M2-F2.
Gentry's first flight in the M2-F2 on 12 October went smoothly according to flight plan from B-52 launch to just before touchdown. Then, the unexpected happened. At about 100 feet above the ground, mere seconds before touchdown, Gentry reached for the landing gear handle-and couldn't reach it. What happened next was the result of quick thinking. Within no more than five or six seconds, he loosened the shoulder harness, leaned forward, pulled the handle, tightened the shoulder harness, and continued with the landing.
 For a second time, the M2-F2 was saved from disaster by the quick thinking and skill of the pilot. Northrop had designed the cockpit dimensions to accommodate Milt Thompson and Bruce Peterson. No consideration had been given to the needs of smaller or shorter pilots, including arm span. A second error was a faulty preflight checkout procedure, for Gentry's inability to reach and pull the landing gear handle while secured in the shoulder harness should have been discovered then, not seconds before touchdown.
Gentry became the Air Force's chief lifting-body pilot on the M2-F2 and, later, the HL-10. With the retirement of Milt Thompson from research flying, there were now only two official lifting-body pilots, Gentry for the Air Force and Bruce Peterson for NASA. Before the first flight of the HL-10 in late December 1966, Peterson made two unpowered flights in the M2-F2. Between July and late December, four pilots-Milt Thompson, Bruce Peterson, Don Sorlie, and Jerry Gentry-had made a total of fourteen flights in the M2-F2.
Air Force/NASA Simulators
When the HL-10 arrived from Northrop, it was trucked to NASA Ames for wind-tunnel testing, as had been done with the M2-F2. The only difference was that data handling was even more automated with the HL-10 than it had been with the M2-F2, thanks to our and the wind-tunnel crew's greater experience and practice in testing the earlier lifting bodies. The HL-10 project was also better staffed with NASA personnel than the M2-F2 had been, the average flight-test experience being three to six years. However, while the M2-F2 team was made up of both NASA and Air Force research or analytical engineers, the HL-10 project was essentially a solo in engineering by NASA.
Bob Hoey wanted to maintain hands-on experience with the aerodynamics of the M2-F2, even after we had developed our own M2-F2 simulator, so he decided to keep the original M2-F2 simulation at the Air Force Flight Test Center. Later, the NASA team at the Flight Test Center concentrated mainly on the simulation of the HL-10.
For a period of time, there were two M2-F2 simulators, one at the Air Force and one at NASA. Even though both simulators used the same wind-tunnel data, the way in which the data was processed and interpreted by the computers within the simulators was different. Once a week, I compared the technical results from both simulators. Generally, the simulators gave the same results. However, now and then, slight differences would appear in the results, followed by lively discussions of which were correct. I felt this was a healthy activity, especially when both simulations concluded that the M2-F2 was safe to fly and when neither set of results required alteration in the vehicle's control settings, stability augmentation system gains, or flight procedures. Joe Weil, my boss and head of NASA's research division, felt uneasy about the lively discussions, seeing them as discord. He basically felt that if there was only one....
...M2-F2 simulator, the Air Force and NASA lifting-body teams would work together even more harmoniously.
New Lifting-Body Project Engineer
By 1966, I was finding my job as lifting-body project engineer more a job of managing people and solving their problems than of directing a technical effort. Once again, as I had in 1965, I found myself facing a career decision.
Over the years, I had worked with Garrison "Gary" Layton on several NASA programs. On our own time, we also had helped one another in our common hobby, flying  experimental radio-controlled model airplanes. As I grew more concerned at how far I was getting away from technical engineering and into management, Gary Layton mentioned that he would like the opportunity to take over as the lifting-body project engineer so that I could have the opportunity to get back to the type of work I loved, especially developing some ideas I was having for remotely controlled vehicles.
Layton and I went to our bosses, Paul Bikle and Joe Weil, to get their approval for Layton to take over as lifting-body project engineer. Once the change was approved in 1967, I became at once involved in a continuing series of about 20 unpiloted vehicle programs at the Flight Research Center until my retirement from NASA in 1985. The unpiloted, or remotely piloted, vehicle programs appealed especially to me....
 ....because they were easy to keep small and innovative and they involved conducting experiments of higher risk.
I have always felt that my talent with people is as a catalyst, a person who can help get individual team members launched creatively in different directions of exploration, especially when the venture is into new and uncharted territory. My talents at NASA seemed best used in small programs of no more than 10-15 people, the larger programs soon becoming complex matters of management and bureaucracy best left to those with talents in those areas.
Developing the Mini-Sniffer became my favorite remotely-piloted vehicle (RPV) program after I left involvement in the lifting-body program as its project engineer. Designed to fly to the 90,000-foot altitude in earth's atmosphere, the Mini-Sniffer weighed only 196 pounds, had a lightweight 22-foot wing span, and was primarily developed to be used as an aerial reconnaissance vehicle in the Martian atmosphere. Folded into a spacecraft and flown to the planet Mars, the little airplane would be unfolded on a parachute and propelled in the Martian atmosphere by a rear-propeller-powered engine fueled by hydrazine, a rocket propellant that does not need oxygen to burn and that would work well in Mars' mostly carbon-dioxide atmosphere. Using a cruise-missile guidance system, the Mini-Sniffer is capable of flying 3,000 miles over Mars through canyons at low altitudes as well as performing a vertical soft landing using hydrazine rockets in the same fashion as the Mars Viking spacecraft.
NASA's HL-10 Team
Operations engineer Herb Anderson headed the 13-member HL-10 hardware team that included crew chief Charles W. Russell; mechanics Art Anderson, John W. "Bill" Lovett, and William "Bill" Mersereau; aircraft electricians Dave Garcia and Albert B. "Al" Harris; instrumentation engineer William D. Clifton; instrumentation technician Richard L. Blair; operations systems engineers Andrew "Jack" Cates and George Sitterle; and inspectors Bill Link and John Reeves. The HL-10 11-member analytical team consisted of aerodynamicist Georgene Laub; systems engineers John Edwards, Berwin Kock, and Wen Painter; stability and control engineers Robert W. "Bob" Kempel and Larry Strutz; simulation engineers Don Bacon, Larry Caw, and Lowell Greenfield; and two members of the United States Army, Lieutenants Pat Haney and Jerry Shimp.
Bob Kempel assumed the leading role in the analysis of the stability and control characteristics of the HL-10, taking over the analytical role previously performed by Ken Iliff and Larry Taylor. In developing the control laws, Kempel worked hand-in-hand with the NASA Langley wind-tunnel team and the Northrop aircraft designers. Kempel had watched the evolution of the M2-F2 configuration, and he was aware of the vehicle's marginal lateral-directional control characteristics. He swore that he would do everything he could to make the HL-10 the best flying lifting-body.
"We were the neophytes," Kempel recalled later of the tension surrounding the first flight of the HL-10. The team preparing the HL-10 simulation had only three to six years of experience. Still "untried and unproven," to use Kempel's words, the HL-10 team wasn't really a full-fledged team yet. "We were a group of individuals working as individuals toward a common goal," Kempel said. "Our approach to completing our tasks was not necessarily lacking in quality but, rather, lacking in experience." 2
Pilots who "flew" the HL-10 real-time simulator found the vehicle's handling and lift-to-drag ratio suspiciously good, compared to those of the M2-F2. Others-including Paul Bikle, the Air Force's M2-F2 team, and NASA project manager John  McTique- were equally skeptical of the HL-10's simulation results. However, the simulator showed the HL-10 to be much more stable and generally much easier to handle than the M2-F2, besides having a better lift-to-drag ratio.
"We always had a difficult time convincing the pilots that we really did know what we were doing," Kempel said. "Before flight they remained skeptical. Our desire, of course, was to have simulations somewhat pessimistic rather than the other way around. We did not want to foster overconfidence."
It wasn't easy instilling even minimal confidence as "the new kids on the block," recalled Kempel. "Managers would pass us in the corridors and shake their heads." The comment most often heard was, "It can't be that good!"3 The team's work continued, nevertheless, kept on track by Gary Layton. Despite the team's lack of assurance, all objectives were met in preparation for the first flight of the HL-10.
HL-10's Maiden Flight
Shortly before Christmas, the HL-10 team convinced Paul Bikle and the rest of NASA and Air Force management that it was ready for the first glide flight. Two captive flights of the HL-10 on the B-52 followed, allowing the team to practice going through check lists and control-room procedure, as well as correct anomalies that appeared in hardware or procedure.
On 21 December, the HL-10 was positioned beneath the B-52's right wing, lifted into position, and attached. Preflight checks were completed. However, the flight was aborted later that day due to an electrical tip-fin flap failure. Since only the subsonic configuration would be flown initially and the flaps would not be moved outboard for the first flight, the wiring was disconnected and stowed.
All preparations for the first free-flight of the HL-10 were completed early the next day, 22 December. Strapped into the cockpit, project pilot Bruce Peterson completed the preflight checks. The canopy was lowered once all ground preparations had been completed. The B-52 taxied to Edwards' main runway, Runway 4. The take-off was smooth. The flight plan called for a launch point about three miles east of the eastern shore of Rogers Dry Lake, abeam of lakebed Runway 18, almost directly over the Air Force's Rocket Propulsion Test Site (now known as the Phillips Laboratory). Launch heading was to be to the north with two left turns. The ground track looked much like a typical lefthand pattern with the launch on the downwind leg, then a base leg, a turn to final, and a final approach to landing on Runway 18.
At 10:30:50 a.m. PST, the HL-10 was launched from the B-52 at 45,000 feet and at an airspeed of 195 miles per hour. Actual launch proved to be very similar to simulator predictions. Although airplane trim was much as expected, Peterson sensed what he described as a high-frequency buffet in pitch and somewhat in roll, later  specifically identified as a "limit cycle"-that is, a rapidly increasing oscillation of a control surface that occurs when the sensitivity (or "gain") of the automatic stabilization system is too high. As speed increased, the limit cycles got noticeably worse. During the first left turn, Peterson noticed that the sensitivity of the pitch stick was excessively high. As the flight progressed, the limit cycles increased in amplitude, and it became obvious that the longitudinal stick was excessively sensitive.
Throughout the flight, Peterson and systems engineer Wen Painter were in constant communication through flight controller John Manke, making gain changes in the vehicle's stability augmentation system (SAS). During the somewhat premature landing, the SAS gains were set at the lowest rate possible without being shut off. Pitch problems masked the difficulties in the roll axis. Peterson initiated the landing flare at approximately 370 miles per hour (mph) with touchdown at about 322 mph, or about 35 mph faster than anticipated. The first flight of the HL-10 had lasted 189 seconds-that is, three minutes and nine seconds from launch to touchdown-with an....
 ...average descent rate of nearly 14,000 feet per minute. Following Painter's requests for adjustments in SAS gains, Peterson had done an excellent job of flying and landing the marginally controllable HL-10.
Peterson remained greatly concerned about the pitch sensitivity and limit cycles. To be precise, a limit cycle is a condition in a feedback control system that produces the uncontrollable oscillation of a control surface due to closed-loop phase lag that, in turn, results from excessive lag in the system (called "hysteresis"), accumulated free play of mechanical linkages, and power actuator non-linearity. The amplitude of the cycle increases with each augmentation to airspeed and system gain setting.
The particular limit cycle that occurred during the first flight of the HL-10 was a 2.75 Hz oscillation (0.4g peak-to-peak) feeding through the gyro-driven SAS. Primarily the problem was in the pitch axis, although it also affected the roll axis. The problem was more severe during the final third of the flight, despite the fact that the SAS gain had been reduced from 0.6 to 0.2 deg/deg/sec. Afterwards, for the entire first HL-10 flight, Peterson gave the pitch axis a Cooper-Harper pilot rating of 4, a rating indicating that deficiencies warrant improvement and are not satisfactory without improvement.
The flight proved to be a large disappointment for the HL-10 team. It seemed to confirm the opinion of others who had said that the team didn't know what it was doing. The team's morale was at low ebb, the flight results quite poor in comparison with the expected results of preflight simulation and analysis.
After the holidays, as 1967 began, team members concluded that if they fixed the stick sensitivity and lowered the SAS gains, they could probably try another flight. There was, however, one lone dissenter in the group. Systems engineer Wen Painter was not convinced that the team completely understood all of the problems.
Continuing to analyze the results of the first flight, Painter argued against another attempted flight, despite the fact that Bruce Peterson had convinced Bikle that the team should try again. Their confidence shaken by the first flight's results, the team gave in to Painter. Bikle backed Painter fully, saying that if Painter didn't sign the ship's book-that is, okay the flight-there would be no flight. Following Painter's suggestion, the team initiated an in-depth unified analysis of the data from the first flight. Very subtly this effort would mold them over time into a real team of proven experience.
Two serious problems identified even before touchdown were substantiated in post-flight analysis: large amplitude limit cycles in the pitch SAS and extreme sensitivity in the longitudinal stick.
The problem with limit cycles apparently was caused by higher-than-predicted elevon control effectiveness and feedback of a 2.75 Hz limit-cycle oscillation through  the SAS. The solution involved using lower SAS gains and modifying the structural resonance 22 Hz mode lead-lag filter that had been installed before the first flight. The modification consisted of a lead-lag network in the SAS electronics and a notch filter, a device that removes a nuisance frequency while having relatively little effect on lower and higher frequencies.
The problem with longitudinal stick sensitivity was relatively simple to solve with a basic gearing modification. On the first flight, the stick gearing of 6.9 deg/inch of elevon proved to be much too sensitive. The nonlinear gearing used in flights 10-37 was approximately 3.5 deg/inch in the elevon range for landing-or about half of what it had been during the first flight. This type of problem is easy to miss when all preparations for flight are made on a fixed-base engineering simulator, a "safe" environment that is relatively relaxed for the pilots who know that if anything goes wrong, they can simply reset the computers. Furthermore, the trim characteristics of a new aircraft are not known precisely. Stick sensitivity, whether longitudinal or lateral, has always been difficult to determine in fixed-base simulations. Pilots always want a very responsive aircraft.
A third problem proved more elusive, not apparent to the pilot or test team during the initial post-flight analysis: lack of longitudinal or lateral-directional control at some portions of the flight. Peterson had realized during the first flight that something wasn't right at high gains and consequently had flown a faster landing approach. Understanding and resolving this problem would require more thorough flight investigation and the assistance of NASA Langley, grounding the HL-10 for fifteen months.
In-Depth Flight Investigation
Wen Painter had insisted that even more analysis needed to be done to find out why lateral control was good sometimes and almost totally lacking at other times, so Bob Kempel launched an in-depth investigation. The assumption before the first flight of the HL-10, according to Kempel, had been that the simulation generated from wind-tunnel test results, an analog computerized mathematical model of the HL-10, was relatively accurate in representing the actual flight vehicle. The expectation, then, was that if flight-recorded control inputs were fed into the computerized model, the dynamics (or motions) of the simulator should be similar to those of the actual vehicle-a technique used for years to validate aerodynamic data by actual flight data. Ideally, the simulation matches the flight exactly; however, such perfection is rarely realized. When the simulation and flight data don't match, aerodynamic parameters are adjusted to duplicate as closely as possible the flight motions. In this way, engineers can then determine how wind-tunnel aerodynamics differ from flight and, perhaps, even why they differ.
The first engineering task in the in-depth flight investigation involved selecting twelve specific maneuvers from five to fifteen seconds in duration from the flight....
....results. Next, the engineers tried to match these maneuvers with those generated by computer, a good match being one in which the computer solution overlays all parameters recorded during flight within the specified time interval and there is little difference between the flight maneuver and the computer generation. However, there were no good matches and only seven found to be acceptable. The other five maneuvers were impossible to match by model. Kempel and the team determined that the computer solutions didn't even remotely resemble the actual flight response of the HL-10. They concluded that they must not have been using an accurate mathematical model, leading them to examine once more the actual flight data.
We decided to play the entire flight-recorded data back through the ground station, the team this time selecting parameters that would be grouped together. The team  selected three families of specific data-accelerations, angular rates next to the control inputs, and information from control surface strain gauges. We then traced out these groupings as a function of time. The new approach gave the team the capability of looking at eight channels of data on each strip-chart. What we found was quite revealing.
The inexperience of the team had shown in how it had earlier arranged the control-room strip-charts for the initial post-flight analysis. Real-time data hadn't been arranged in the best logical manner for accurate assessment of data families. With the data re-arranged, the team found that, although of different parameters, each of the traces generally moved with the appropriate responses indicating the vehicle's motion. However, during certain portions of the flight, some of the traces would become blurry or fuzzy, especially the control surface strain gauges when a higher frequency disturbance occurred. When the data was lined up on a common time interval, many data traces displayed similar phenomena.
A second but related discovery was that there had been two significant intervals when Bruce Peterson had commanded significant amounts of aileron, only to have the vehicle not respond until the angle of attack was reduced. Peterson was disturbed enough by the vehicle's response to control input that Kempel and the team decided to investigate it further. What they found was that each time the problem occurred, the angle of attack was above the range of 11 to 13 degrees, and that as the angle of attack decreased through this range, the ailerons suddenly became very effective, producing significant amounts (30 to 45 degrees per second) of roll angular rate.
When the team computer-matched these two time intervals, the initial part of each response would not match. However, as the angle of attack was reduced to the point that the ailerons became effective, the mathematical model began to match the flight data. But why?
As Kempel recalls, "We began to think that a massive flow separation was possible over the upper aft portion of the vehicle at the higher angles of attack, causing the control surfaces to lose a large percentage of their effectiveness. . . . This flow separation can be likened to the sudden loss of lift and increase in drag of a conventional wing as AOA [angle of attack] is increased and the wing stalls. As the AOA was decreased, the airflow would suddenly reattach and the controls would behave in their normal fashion. The more we looked at the data, the more plausible this theory seemed; although the wind-tunnel data did not indicate a problem to the degree that we had experienced in flight. The data also indicated a significant loss of lift-to-drag ratio above Mach numbers of 0.5 and AOA of 12 degrees. This finding further convinced us that the problem was caused by massive flow separation."
At this point, Kempel and his team decided to share their preliminary findings with the NASA Langley engineers since, as Kempel said, the HL-10 was "their  baby." The Langley team agreed to do more wind-tunnel tests immediately, using the 0.063-scale, 16-inch-long HL-10 model. According to Kempel, the Langley team's decision seemed "almost unbelievable because wind-tunnel schedules are made at least one year and, many times, [several] years in advance."4 As the Langley team urged them to do, Kempel and his team packed their data and bags and traveled to NASA Langley to work jointly on the situation.
Bob Kempel, Berwin Kock, Gary Layton, and Wen Painter of the Flight Research Center gathered around a table with Langley's Bill Kemp, Linwood (Wayne) McKinny, Bob Taylor, and Tommy Toll in the building housing Langley's 7-by-10-foot high-speed wind tunnel. Kempel and his team, after presenting their data, theorized that the problem was caused by massive flow separation. Bob Taylor jumped up from his chair, angrily slammed his mechanical pencil to the floor, and let loose with a string of oaths. After he calmed down, Taylor said that he had earlier thought that this would be a problem. He had had a gut feeling that the flow separation seen by the Langley team on the wind-tunnel model would be worse in flight, and he was upset with himself for not following his instincts as an aerodynamicist and adding preventative measures to the HL-10 design before the vehicle was built.
The discussion then turned to what could be done now. The Langley team agreed to give the problem its immediate attention, assuming responsibility for coming up with a remedy. Kempel and his team left Langley more aware than they had been earlier of why they were having a lateral control problem in flying the HL-10. They agreed that, until Langley came up with a solution, the HL-10 would not be flown. While they waited for word from Langley, they busied themselves with solving the problems they had determined earlier (stick sensitivity and limit cycles), enlisting the help of Northrop in designing the electronic notch filter for eliminating the limit-cycle mode from feeding back through the flight control system.
HL-10 as "Hangar Queen"
The HL-10 was a "hangar queen" for the next 15 months, grounded after its first flight three days before Christmas 1966. During this time, flight safety began receiving more attention, to some extent due to the near crashes and temporary losses of control with the other lifting bodies. Adherence to flight schedules took a second priority to flight safety, benefiting the HL-10 program. Bob Kempel was given free license to work without a time restraint in leading the effort to fix the vehicle's control problems.
Throughout the winter and spring of 1967, members of the NASA Langley team continued to work on correcting the flow-separation problem, coordinating their efforts with those of Kempel and his team at the Flight Research Center. The Langley team came up with two possible ways to fix the problem, both modifications concentrating  on changes to the outboard vertical fins. The first proposed modification involved thickening and cambering the inside of the fins. The second proposed slightly extending and cambering the leading edges. Langley ran a full set of wind-tunnel tests on both proposed modification, sending the resulting data to Kempel and his team. Although the Langley team members gave their assessment of the wind-tunnel results, they left the decision of which modification to use up to Kempel and the team at the Flight Research Center.
Kempel recalls that once he had the preliminary data from these wind-tunnel tests, he initiated his own extensive evaluation of the data. "'Preliminary data,'" Kempel clarifies, "was the wind-tunnel guys' way of telling us that they had worked most of their magic in data reduction, but that they still were not going to say that this was the last word."5 Kempel plotted all of the data from digital listings by hand. Although engineers today use computer plotting routines to do what Kempel in 1967 had to do by hand, the approach made him and other team members intensely familiar with the data, for the extensive process of hand-plotting meant they had to live with the data day in and day out.
During the summer of 1967, Kempel plotted all of the data for both proposed modifications as a function of angle of attack for constant Mach numbers. He made all plot scales uniform to ease comparisons, plotting thousands of points in this way. Once the data was lined up and compared, Kempel found there were some subtle but significant differences between the Langley wind-tunnel data and the data set generated by the HL-10 simulator at the Flight Research Center.
As Kempel explains it, "Some non-linearities in the original data were not present" in the Langley data. He hypothesized that "if these non-linearities indicated flow separation, then the lack of these would indicate no flow separation or separation to a lesser degree."6 Based on that theory, Kempel backed using the second modification proposed by Langley. He presented his hypothesis to his boss, aerodynamicist Hal Walker, and then to the management at the Flight Research Center. With their agreement and the concurrence of the NASA Langley team, Kempel and his team began making arrangements for the modification of the HL-10.
In the early autumn of 1967, Northrop Norair was contracted to design and install the modification that would be the final configuration change to the HL-10. Northrop and NASA decided that the modification would involve a fiberglass glove, backed by a metal structure. Work on the glove continued through the autumn and winter of 1967.
As Kempel recalls, "In the NASA hangar, Northrop's Fred Erb shed his normal working attire-a suit-and donned coveralls to assist in the installation of the fiberglass glove. He was a senior-level engineer with over 25 years with Northrop, rolling....
....up his sleeves and getting his hands dirty."7 "A real engineer!", Kempel might have added.
By the spring of 1968, the HL-10 was nearly ready to end its stay as a hangar queen, with vehicle preparation then in its final stages. Changes in the configuration, flight controls, and internal systems were already finished.
Gentry, Peterson, and the M2-F2
Meanwhile, following Jerry Gentry's flight on 14 November 1966, the M2-F2 was grounded five and a half months so that the LR-11 rocket-propulsion system could be installed by the lifting body's team under the leadership of Meryl DeGeer. Gentry made four glide tests in the M2-F2 by 2 May 1967, conducting research maneuvers to define the vehicle's aerodynamic characteristics and preparing for planned rocket-powered supersonic flights. Having flown the M2-F2 successfully several times, Gentry was by this time firmly established as an experienced lifting-body pilot, soon becoming the Air Force's most active pilot in the joint NASA-Air Force lifting-body program.
 A key member of the M2-F2 team, Gentry knew each crew member personally. Practical jokes abounded between them, and Gentry never once let anyone forget that he represented the Air Force on the project. During his early flights in 1966, he had told the crew that he hated the zinc-chromate yellow-green color of the insides of the lifting bodies. Afterwards, during one of his flights, his flightline car, a 1954 Ford, was "borrowed" long enough to be painted entirely in zinc-chromate yellow-green at the NASA paint shop.
In retaliation, Gentry and his Air Force cronies would slip over to NASA during the early morning hours and paste a large Air Force sign on the side of the HL-10, which originally had no markings indicating Air Force involvement in the program. When the NASA crew members arrived and saw the Air Force sign, they would promptly remove it. Later, they had the last word, decorating Gentry's yellow-green Ford by pasting large "flower power" decals all over it, the decals then popularly in use mainly by the era's "flower children."
By the winter and spring of 1966-1967, the two official lifting-body pilots-the Air Force's Jerry Gentry and NASA's Bruce Peterson-were doing alternate flights in the lifting bodies. Since Peterson had flown the HL-10 for its maiden flight on 22 December 1966, it was Gentry's turn to fly the M2-F2 on 2 May 1967 for its first flight with the rocket system installed. On this glide flight, his fifth in the M2-F2, Gentry reported that the weight increase from the installed rocket system had not changed the vehicle's control characteristics. However, he also confirmed what Milt Thompson and Bruce Peterson had reported on their previous flights: that if the M2-F2 is not flown properly, loss of roll control can occur quickly.
In September of 1966, during the symposium of the Society of Experimental Test Pilots, Bruce Peterson had given a detailed description of the M2-F2's lateral control characteristics. Maneuverability "was not appreciably affected" as yaw and roll damper gains were reduced to zero during the first 180 degrees of approach on the fifth flight, he said. However, he felt at the time that "abrupt aileron or rudder inputs could readily induce Dutch roll oscillations"; and these "could be continuous and could seriously hamper the pilot in holding a bank angle." His strategy was to "nudge" the M2-F2 to the desired bank angle by using small lateral control inputs.
"Acceptable lateral control is achieved only by means of aileron-rudder interconnect since the adverse yaw due to aileron at most flight conditions results in roll reversal," he said.
Crash of the M2-F2
On 10 May 1967, eight days after Gentry's glide flight, it was Bruce Peterson's turn for a glide flight in the M2-F2 with the rocket system installed. It had been eight months since Peterson's last six-minute glide flight in the lifting body, and this would be his third M2-F2 flight.
All went well during the beginning of Peterson's flight on 10 May. He launched away from the B-52 at 44,000 feet, heading to the north, flying east of Rogers Dry Lake, and descended at a steep angle to 7,000 feet. Then, as he flew with a very low angle of attack, the M2-F2 began a Dutch roll motion, rolling from side to side at over 200 degrees per second. Peterson increased the angle of attack by raising the nose. The oscillations stopped, but now the M2-F2 was pointed away from its intended flight path. Realizing that he was too low to reach the planned landing site on lakebed Runway 18, Peterson was rapidly sinking toward a section of the lakebed that lacked the visual runway reference markings needed to accurately estimate height above the lakebed.
At this moment, a rescue helicopter suddenly appeared in front of the M2-F2, distracting Peterson who was still stunned and disoriented from the earlier Dutch roll motions. He radioed, "Get that chopper out of the way." A few seconds later, he radioed, "That chopper's going to get me." NASA pilot John Manke, flying...
....chase in an F-5D, assured Peterson that he was now clear of the helicopter, which had chugged off out of Peterson's flight path. 9
Trying to buy time to complete the flare, Peterson fired the landing rockets. The M2-F2 flared nicely. He lowered the landing gear, only one-and-a-half seconds being needed for the M2-F2's gear to go from up and locked to down and locked. But time had run out. The sudden appearance of the helicopter likely had distracted Peterson enough that he began lowering the landing gear half a second too late.
Before the gear locked, while it was still half-deployed, the M2-F2 hit the lakebed. The weight of the vehicle pushed against the pneumatic actuators, and the landing gear was pushed back up into the vehicle. The round shape of the vehicle's bottom did not lend itself to landing minus landing gear. The result was more like a log rolling than a slide-out on a flat bottom. (By contrast, the shape of the HL-10 likely would have lent itself readily to a gear-up landing, had one been required. Langley engineers had even given serious thought to eliminating the HL-10's landing gear for spacecraft recovery.)
 As the M2-F2 contacted the ground, the vehicle's telemetry antennae were sheared off. As this happened, I and the other engineers in the control room watched the needles on instrumentation meters flick to null. Startled, we looked up at the video monitor in time to see the M2-F2, as if in a horrible nightmare, flipping end over end on the lakebed at over 250 miles per hour. It flipped six times, bouncing 80 feet in the air, before coming to rest on its flat back, minus its canopy, main gear, and right vertical fin. The M2-F2 sustained so much damage that one would have been hard pressed to identify it visually as the same vehicle.
By all odds, Peterson could have been expected to have died in the crash. He was seriously injured. Assistant crew chief Jay King quickly crawled under the M2-F2 to shut off the hydraulic and electrical system. He found Peterson trying to remove his helmet. King unstrapped him and helped him out of the vehicle. Peterson was rushed to the base hospital at Edwards for emergency care. Afterwards, he was transferred first to the hospital at March Air Force Base near Riverside, California, and later, to UCLA's University Hospital in Los Angeles.
The heavy metal cage-like structure around the cockpit-ironically, added to the M2-F2 by its NASA/Northrop designers simply to provide ballast and save their pride-was mainly what saved Peterson's life. Even with this added protection, his oxygen mask was ripped off as his head made contact with the lakebed. Each time the vehicle rolled, a stream of high-velocity lakebed clay hammered at Peterson's face. He suffered a fractured skull, severe facial injuries, a broken hand, and serious damage to his right eye. He underwent restorative surgery on his face during the ensuing months; however, he later lost the vision in the injured eye from a staphylococcus infection.
He returned to the NASA Flight Research Center as a project engineer on the CV-990, F-8 Digital Fly-By-Wire, and F-8 Supercritical Wing. He continued to fly in a limited way on the CV-990 and F-111 and eventually became the Director of Safety, Reliability, and Quality Assurance. He also continued to fly as a Marine reservist. Later, he left NASA to serve as a safety officer at Northrop in the flight tests of the B-2 bomber and other aircraft.
About two years after the crash of the M2-F2, the popular television series The Six-Million-Dollar Man began its six years of weekly programming, using NASA ground-video footage of the crash as a lead-in to each episode. The producers of the television series capitalized on Peterson's misfortune by inventing a "bionic man" (played by Lee Majors) who had missing body parts replaced with bionic devices. Colonel Steve Austin, the fictional television character played by Majors, had, like Peterson, also lost an eye in the crash.
As can happen only in Hollywood, the fictional Austin gained a bionic eye with super powers. The television show also multiplied the injuries of Austin beyond those suffered in real life by Peterson, giving him two bionic legs and a  bionic arm that provided him with super power and speed. Nevertheless, NASA pilot Bruce Peterson is the real-life model on which The Six-Million-Dollar Man is based. Due to the popularity of this television series, it's possible that as many Americans viewed the crash of the M2-F2 on television as later viewed the first televised NASA shuttle landings.
The crash of the M2-F2 was the only serious accident that occurred during the twelve-and-a-half years of flight-testing eight different lifting bodies.10 Because of the popularity of the television program The Six-Million-Dollar Man, most people are more familiar with the solitary serious accident that occurred during the lifting-body program than they are with its extensive record of otherwise accident-free success.
1 Thompson, At the Edge of Space, p. 276.
2 Robert W. Kempel, Weneth D. Painter, and Milton O. Thompson, Developing and Flight Testing the HL-10 Lifting Body: A Precursor to the Space Shuttle (Washington, D.C.: NASA Reference Publication 1332, 1994), pp. 21-22. Since Kempel was the principal author of this paper, to avoid convoluted phraseology the narrative treats the words in it as his.
3 Ibid., p. 21.
4 Ibid., p. 26. for quotations 2 and 3.
5 Ibid., p. 27.
6 Ibid., pp. 27-28.
7 Ibid., p. 28.
8 Quotations in two paragraphs above from Bruce Peterson's comments in Milton O. Thompson, Bruce A. Peterson, and Jerauld R. Gentry, "Lifting Body Flight Test Program," Society of Experimental Test Pilots, Technical Review (September 1966): 4-5.
9 Quotations from Hallion, On the Frontier, p. 159; Wilkinson, "Legacy of the Lifting Body," pp. 57-60, but Dale Reed was watching the whole episode on a TV monitor from the control room, as pointed out below in the narrative, so he heard the comments first hand.
10 These were the M2-F1, M2-F2, M2-F3, HL-10, HL-10 modified, X-24A, X-24B, and the Hyper 3. For those of these vehicles not yet introduced into the narrative, see the succeeding chapters.