Construction and test of the flying mock-up, for delivery in 360 days, were approved by the Secretary of War on October 3, 1941. The aircraft bore a distinct resemblance to its smaller predecessor, the N-1M. With two 260-hp Menasco C6S-4 engines buried in the wing, the 7000-pound N-9M was designed for an endurance of 3.2 hours with 100 gallons of fuel, and had a design ceiling of 21,500 feet. Controls consisted of elevons, rudders, and trim tabs. An hydraulically operated, retractable tricycle type landing gear was provided, and supplemented by a fourth retractable wheel extending from the trailing edge just aft of the canopy to protect the propellers in the event of an extremely tail-low landing.

Cockpit space was at a premium, with various pieces of test equipment, switches, and gauges using up what little space was available. Seat and rudder pedals were not adjustable, quite unusual for an airplane of that size and manoeuvrability. As in the N-1M, the usual control stick was replaced by a control column and wheel, an awkward and cumbersome arrangement for the size of the cockpit, but appropriate for its role as a bomber mock-up.

The first of the four N-9Ms had unusual spilt flap rudders whose hinge lines were orientated fore and aft. The The aircraft was designed for a speed of 257 mph at 7000 feet

The first flight of the N-9M-1 was on December 27, 1942, three months later than specified by contract. Over the next five months there were 45 flights. With a few exceptions, most were terminated by mechanical failures of one sort or another, with the Menasco engines the primary source of problems. Consequently, very little data relative to drag, stability, and control were obtained. After about 22.5 hours of accumulated flight time, the N-9M crashed approximately 12 miles west of Muroc Army Air Base on May 19, 1943. The pilot, Max Constant, was killed as he attempted to recover the aircraft from a right-hand, 60 degree nose-down spin. Apparently, Constant made every attempt to stop the spin, since the left hand spin chute had been deployed, and the flaps appeared to be partially lowered. There was definite indication that the pilot had attempted to leave the aircraft; the cockpit enclosure had been released, the propeller brakes had been applied, and Constant had unfastened his safety belt. Whether he was prevented from bailing out by lack of time or by some unknown physical circumstance was not determined.

The remains of N-9M-1 after Max Constant's fatal accident on May 19, 1943. The view is from in front of the aircraft, looking aft. The two propeller shaft housings are evident, with the cockpit in the middle, canopy missing, antenna mast at the rear.

It appeared that the N-9M might have dangerous spin characteristics that could not be controlled by normal use of the control surfaces; perhaps the spin chutes were inadequate in size or improperly placed on the airplane for quick recovery. Further investigation in the NACA spin tunnel was required.

The loss of the first N-9M was followed a month later by the first flight of N-9M-2 on June 24, 1943. A minor setback was experienced when the cockpit canopy came off in flight and, despite a successful landing, slight damage to a landing gear door, radio mast, and yaw meter resulted. Nonetheless, by mid September 1943, the first reliable drag data were being obtained on N-9M-2; they indicated the drag of the XB-35 at cruising speed would be approximately 7 to 12 percent greater than that estimated from wind tunnel test results.

The N-9M-2 generally exhibited satisfactory longitudinal and lateral stability and control in high speed and cruising ranges. Some difficulty in obtaining satisfactory directional control was experienced. Of more concern, however, was the occurrence of severe reversal of elevator control forces at high lift coefficients. Due to premature separation of the airflow over the top surface of the wing as the stall was approached, flow conditions over the elevons caused them to trail upward producing a force reversal on the elevator control. This may have been the reason for the difficulty which led to the crash of the first N-9M.

By April 1944, the N-9M-2 had completed 33 flights for a total flying time of 23 hours. Data obtained on the flights were constructive and contributed to performance predictions for the XB-35, then in the design phase.

On June 28, 1944, Army pilots flew acceptance flights on the N-9MA. This model contained practically all of the design features that were to be used in the XB-35 aircraft. In addition to the split trailing edge and pitch control flaps that served as devices for increasing lift and drag, a leading edge slot was installed near the tip of each wing to lessen stall tendencies at high angles of attack. Flight characteristics of the N-9MA gave firm indications that the characteristics of the XB-35 bomber would also be satisfactory. Ground handling was excellent; the aircraft was very manoeuvrable and responsive in turns; control was good, although elevator control was quite sensitive. Flying characteristics in rough air, previously reported as unsatisfactory for the first two N-9Ms, were satisfactory, due to new rudders and hydraulically powered elevons.

By October 1944, the authorized flight test program for the N-9MA had been completed. The aircraft had flown about 50 flights, and stability, control, and flying qualities were judged to be generally satisfactory. Several inadvertent half-turn spins had been experienced and recovery had been accomplished without difficulty. Northrop test pilot Alex Papana, who along with Harry Crosby had experienced some interesting rides in Northrop's experimental MX-334 glider, had even engaged in some extra-curricular aerobatics in the N-9MA, in one instance accomplishing four loops in quick succession.

With the completion of the test program, the N-9MA was used by the Army to familiarize pilots with the craft. At about the same time, the fourth and last of the series, the N-9MB, was nearing completion. It would eventually be used to flight test the latest in configuration and improved systems that would be incorporated into the XB-35. As in N-9M-1, space for a passenger was created behind the pilot by removing a fuel tank.

The N-9MB, with more powerful 300-hp Franklin engines, provided yeoman service in the flight test program, producing invaluable data that contributed directly to the engineering of the bomber program. The production of the B-35 had become bogged down by a shortage of engineers, design problems, conflicts with other production programs, and excessive requirements for production engineering. The cumulative effect was postponement of the first flight date. World War II ended, the program dragged on, and the N-9Ms continued to do their jobs.

In May 1946, an Army test pilot assigned to Colonel Albert Boyd's Flight Test Section at Muroc summarized his observations and impressions of his one-flight evaluation of the N-9MB as follows:

'An hour's flight is hardly a fair basis for drawing decisive conclusions. However, the airplane flew surprisingly well, was more stable and handled far better than most would expect. It would take a few more hours practice to make good takeoffs and get the proper coordination on turns. But the technique could be mastered without too much difficulty. It serves its purpose well as a flying model.'

The pilot's name was Captain Glen W. Edwards. Almost two months later the first XB-35 made its initial flight, beginning a protracted flight test program during which Edwards would lose his life.