Design and development
Lockheed designed the P-38 in response to a February 1937 specification from the
United States Army Air Corps. Circular Proposal X-608 was a set of aircraft performance goals authored by
Benjamin S. Kelsey and
Gordon P. Saville for a twin-engine, high-altitude "interceptor" having "the tactical mission of interception and attack of hostile aircraft at high altitude."
 In 1977, Kelsey recalled he and Saville drew up the specification using the word interceptor as a way to bypass the inflexible Army Air Corps requirement for pursuit aircraft to carry no more than 500 lb (230 kg) of armament including ammunition, as well as the restriction of single-seat aircraft to one engine. Kelsey was looking for a minimum of 1,000 lb (450 kg) of armament.
 Kelsey and Saville aimed to get a more capable fighter, better at dog-fighting and at high-altitude combat. Specifications called for a maximum airspeed of at least 360 mph (580 km/h) at altitude, and a climb to 20,000 ft (6,100 m) within six minutes,
 the toughest set of specifications USAAC had ever presented. The unbuilt
Vultee XP1015 was designed to the same requirement, but was not advanced enough to merit further investigation. A similar single-engine proposal was issued at the same time, Circular Proposal X-609, in response to which the
Bell P-39 Airacobra was designed.
 Both proposals required liquid-cooled
Allison V-1710 engines with turbo-superchargers and gave extra points for
tricycle landing gear.
P-38 armament, concentrated in the nose of the aircraft
The Lockheed design team, under the direction of
Hall Hibbard and
Clarence "Kelly" Johnson, considered a range of twin-engine configurations, including both engines in a central fuselage with push–pull propellers.
The eventual configuration was rare in terms of contemporary fighter aircraft design, with only the preceding
Fokker G.1, the contemporary
Focke-Wulf Fw 189 Luftwaffe reconnaissance aircraft, and the later
Northrop P-61 Black Widow night fighter having a similar
planform. The Lockheed team chose twin booms to accommodate the tail assembly, engines, and turbo-superchargers, with a central nacelle for the pilot and armament. The XP-38 gondola mockup was designed to mount two .50-caliber (12.7 mm)
M2 Browning machine guns with 200 rounds per gun (rpg), two .30-caliber (7.62 mm) Brownings with 500 rpg, and a T1 Army Ordnance 23 mm (.90 in)
autocannon with a rotary magazine as a substitute for the non-existent 25 mm
Hotchkiss aircraft autocannon specified by Kelsey and Saville.
 In the YP-38s, a
37 mm (1.46 in) M9 autocannon with 15 rounds replaced the T1.
 The 15 rounds were in three five-round clips, an unsatisfactory arrangement according to Kelsey, and the M9 did not perform reliably in flight. Further armament experiments from March to June 1941 resulted in the P-38E combat configuration of four M2 Browning machine guns, and one
Hispano 20 mm (.79 in) autocannon with 150 rounds.
Clustering all the armament in the nose was unusual in U.S. aircraft, which typically used wing-mounted guns with trajectories set up to crisscross at one or more points in a
convergence zone. Nose-mounted guns did not suffer from having their useful ranges limited by pattern convergence, meaning that good pilots could shoot much farther. A Lightning could reliably hit targets at any range up to 1,000 yd (910 m), whereas the wing guns of other fighters were optimized for a specific range.
rate of fire was about 650 rounds per minute for the 20×110 mm cannon round (130-gram shell) at a muzzle velocity of about 2,850 ft/s (870 m/s), and for the .50-caliber machine guns (51-gram rounds), about 850 rpm at 2,900 ft/s (880 m/s) velocity. Combined rate of fire was over 4,000 rpm with roughly every sixth projectile a 20 mm shell.
 The duration of sustained firing for the 20 mm cannon was approximately 14 seconds while the .50-caliber machine guns worked for 35 seconds if each magazine was fully loaded with 500 rounds, or for 21 seconds if 300 rounds were loaded to save weight for long distance flying.
The Lockheed design incorporated tricycle undercarriage and a bubble canopy, and featured two 1,000 hp (750 kW)
turbosupercharged 12-cylinder Allison V-1710 engines fitted with
counter-rotating propellers to eliminate the effect of engine
torque, with the turbochargers positioned behind the engines, the exhaust side of the units exposed along the dorsal surfaces of the booms.
 Counter-rotation was achieved by the use of "handed" engines, which meant the crankshaft of each engine turned in the opposite direction of its counterpart, a relatively easy task for
a modular-design aircraft powerplant as the V-1710.
The P-38 was the first American fighter to make extensive use of stainless steel and smooth, flush-riveted butt-jointed aluminum skin panels.
 It was also the first military airplane to fly faster than 400 mph (640 km/h) in level flight.
XP-38 and YP-38 prototypes
Lockheed won the competition on 23 June 1937 with its Model 22 and was contracted to build a prototype XP-38
 for US$163,000, though Lockheed's own costs on the prototype would add up to US$761,000.
 Construction began in July 1938, and the XP-38 first flew on 27 January 1939 at the hands of Ben Kelsey.
One of 13 YP-38s constructed
Kelsey then proposed a speed dash to
Wright Field on 11 February 1939 to relocate the aircraft for further testing.
General Henry "Hap" Arnold, commander of the USAAC, approved of the record attempt and recommended a cross-country flight to New York. The flight set a speed record by flying from California to New York in seven hours and two minutes, not counting two refueling stops,
 but the aircraft was downed by carburetor icing short of the
Mitchel Field runway in
Hempstead, New York and was wrecked. However, on the basis of the record flight, the Air Corps ordered 13 YP-38s on 27 April 1939 for US$134,284 each.
 (The "Y" in "YP" was the USAAC's designation for a
prototype, while the "X" in "XP" was for
experimental.) Lockheed's Chief
Tony LeVier angrily characterized the accident as an unnecessary publicity stunt,
 but according to Kelsey, the loss of the prototype, rather than hampering the program, sped the process by cutting short the initial test series.
 The success of the aircraft design contributed to Kelsey's promotion to captain in May 1939.
. Planes start at the back of the building on the far right (without wings, so that section of the line is narrower). When they reach the end of that line, they shift to the center line, grow wings, and move backward down this line. Upon reaching the end, they are then shifted to the line at the left, and progress forward to the end of the line.
Manufacture of YP-38s fell behind schedule, at least partly because of the need for mass-production suitability making them substantially different in construction from the prototype. Another factor was the sudden required expansion of Lockheed's facility in
Burbank, taking it from a specialized civilian firm dealing with small orders to a large government defense contractor making
Hudsons, and designing the
TWA. The first YP-38 was not completed until September 1940, with its maiden flight on 17 September.
 The 13th and final YP-38 was delivered to the Air Corps in June 1941; 12 aircraft were retained for flight testing and one for destructive stress testing. The YPs were substantially redesigned and differed greatly in detail from the hand-built XP-38. They were lighter and included changes in engine fit. The propeller rotation was reversed, with the blades spinning outward (away from the
cockpit) at the top of their arc, rather than inward as before. This improved the aircraft's stability as a gunnery platform.
High-speed compressibility problems
View of a P-38G cockpit. Note the yoke, rather than the more-usual stick.
Test flights revealed problems initially believed to be tail
flutter. During high-speed flight approaching Mach 0.68, especially during dives, the aircraft's tail would begin to shake violently and the nose would tuck under, steepening the dive. Once caught in this dive, the fighter would enter a high-speed
compressibility stall and the controls would lock up, leaving the pilot no option but to bail out (if possible) or remain with the aircraft until it got down to denser air, where he might have a chance to pull out. During a test flight in May 1941, USAAC Major Signa Gilkey managed to stay with a YP-38 in a compressibility lockup, riding it out until he recovered gradually using
 Lockheed engineers were very concerned at this limitation but first had to concentrate on filling the current order of aircraft.
In late June 1941, the Army Air Corps was renamed the
U.S. Army Air Forces (USAAF), and a total of 65 Lightnings were finished for the service by September 1941 with more on the way for the USAAF, the
Royal Air Force (RAF), and the Free French Air Force operating from England.
By November 1941, many of the initial assembly-line challenges had been met, which freed up time for the engineering team to tackle the problem of frozen controls in a dive. Lockheed had a few ideas for tests that would help them find an answer. The first solution tried was the fitting of spring-loaded servo tabs on the elevator trailing edge designed to aid the pilot when control yoke forces rose over 30 pounds-force (130 N), as would be expected in a high-speed dive. At that point, the tabs would begin to multiply the effort of the pilot's actions. The expert test pilot, 43-year-old
 Ralph Virden, was given a specific high-altitude test sequence to follow and was told to restrict his speed and fast maneuvering in denser air at low altitudes, since the new mechanism could exert tremendous leverage under those conditions. A note was taped to the instrument panel of the test craft underscoring this instruction. On 4 November 1941, Virden climbed into YP-38 #1 and completed the test sequence successfully, but 15 minutes later was seen in a steep dive followed by a high-G pullout. The tail unit of the aircraft failed at about 3,500 ft (1,000 m) during the high-speed dive recovery; Virden was killed in the subsequent crash. The Lockheed design office was justifiably upset, but their design engineers could only conclude that servo tabs were not the solution for loss of control in a dive. Lockheed still had to find the problem; the Army Air Forces personnel were sure it was flutter and ordered Lockheed to look more closely at the tail.
In 1941 flutter was a familiar engineering problem related to a too-flexible tail, but the P-38's
empennage was completely skinned in aluminum
[Note 2] rather than fabric and was quite rigid. At no time did the P-38 suffer from true flutter.
 To prove a point, one elevator and its vertical stabilizers were skinned with metal 63% thicker than standard, but the increase in rigidity made no difference in vibration. Army
Lieutenant Colonel Kenneth B. Wolfe (head of Army Production Engineering) asked Lockheed to try external mass balances above and below the elevator, though the P-38 already had large mass balances elegantly placed within each vertical stabilizer. Various configurations of external mass balances were equipped, and dangerously steep test flights were flown to document their performance. Explaining to Wolfe in Report No. 2414, Kelly Johnson wrote "the violence of the vibration was unchanged and the diving tendency was naturally the same for all conditions."
 The external mass balances did not help at all. Nonetheless, at Wolfe's insistence, the additional external balances were a feature of every P-38 built from then on.
P-38 pilot training manual compressibility chart shows speed limit vs. altitude.
Johnson said in his autobiography
 that he pleaded with
NACA to do model tests in its wind tunnel. They already had experience of models thrashing around violently at speeds approaching those requested and did not want to risk damaging their tunnel. Gen. Arnold, head of Army Air Forces, ordered them to run the tests, which were done up to Mach 0.74.
 The P-38's dive problem was revealed to be the
center of pressure moving back toward the tail when in high-speed airflow. The solution was to change the geometry of the wing's lower surface when diving in order to keep lift within bounds of the top of the wing. In February 1943, quick-acting dive flaps were tried and proven by Lockheed test pilots. The dive flaps were installed outboard of the engine nacelles, and in action they extended downward 35° in 1.5 seconds. The flaps did not act as a speed brake; they affected the pressure distribution in a way that retained the wing's lift.
Late in 1943, a few hundred dive flap field modification kits were assembled to give North African, European and Pacific P-38s a chance to withstand compressibility and expand their combat tactics. Unfortunately, these crucial flaps did not always reach their destination. In March 1944, 200 dive flap kits intended for
European Theater of Operations (ETO) P-38Js were destroyed in a mistaken
identification incident in which an RAF fighter shot down the
Douglas C-54 Skymaster (mistaken for an
Fw 200) taking the shipment to England. Back in Burbank, P-38Js coming off the assembly line in spring 1944 were towed out to the ramp and modified in the open air. The flaps were finally incorporated into the production line in June 1944 on the last 210 P-38Js. Despite testing having proved the dive flaps effective in improving tactical maneuvers, a 14-month delay in production limited their implementation, with only the final half of all Lightnings built having the dive flaps installed as an assembly-line sequence.
Johnson later recalled:
I broke an ulcer over compressibility on the P-38 because we flew into a speed range where no one had ever been before, and we had difficulty convincing people that it wasn't the funny-looking airplane itself, but a fundamental physical problem. We found out what happened when the Lightning shed its tail and we worked during the whole war to get 15 more kn [28 km/h] of speed out of the P-38. We saw compressibility as a brick wall for a long time. Then we learned how to get through it.
Buffeting was another early aerodynamic problem, difficult to distinguish from compressibility as both were reported by test pilots as "tail shake". Buffeting came about from airflow disturbances ahead of the tail; the airplane would shake at high speed. Leading edge wing slots were tried as were combinations of
filleting between the wing, cockpit and engine nacelles. Air tunnel test number 15 solved the buffeting completely and its fillet solution was fitted to every subsequent P-38 airframe. Fillet kits were sent out to every squadron flying Lightnings. The problem was traced to a 40% increase in air speed at the wing-fuselage junction where the thickness/chord ratio was highest. An airspeed of 500 mph (800 km/h) at 25,000 ft (7,600 m) could push airflow at the wing-fuselage junction close to the speed of sound. Filleting solved the buffeting problem for the P-38E and later models.
Another issue with the P-38 arose from its unique design feature of outwardly rotating (at the "tops" of the propeller arcs) counter-rotating propellers. Losing one of two engines in any twin-engine non-
centerline thrust aircraft on takeoff creates sudden drag, yawing the nose toward the dead engine and rolling the wingtip down on the side of the dead engine. Normal training in flying twin-engine aircraft when losing an engine on takeoff is to push the remaining engine to full throttle to maintain airspeed; if a pilot did that in the P-38, regardless of which engine had failed, the resulting engine torque and
p-factor force produced a sudden uncontrollable yawing roll, and the aircraft would flip over and hit the ground. Eventually, procedures were taught to allow a pilot to deal with the situation by reducing power on the running engine, feathering the prop on the failed engine, and then increasing power gradually until the aircraft was in stable flight. Single-engine takeoffs were possible, though not with a full fuel and ammunition load.
 This same design feature was present from its earliest days on both the Luftwaffe twin-engine
Henschel Hs 129 ground-attack aircraft, and from the fourth prototype onwards of the otherwise troubled
Heinkel He 177A heavy bomber.
The engines were unusually quiet because the
muffled by the
General Electric turbo-superchargers on the twin
 There were early problems with cockpit temperature regulation; pilots were often too hot in the tropical sun as the canopy could not be fully opened without severe buffeting and were often too cold in northern Europe and at high altitude, as the distance of the engines from the cockpit prevented easy heat transfer. Later variants received modifications (such as electrically heated flight suits) to solve these problems.
On 20 September 1939, before the YP-38s had been built and flight tested, the USAAF ordered 66 initial production P-38 Lightnings, 30 of which were delivered to the USAAF in mid-1941, but not all these aircraft were armed. The unarmed aircraft were subsequently fitted with four .50 in (12.7 mm) machine guns (instead of the two .50 in/12.7 mm and two .30 in/7.62 mm of their predecessors) and a 37 mm (1.46 in) cannon. They also had armored glass, cockpit armor and
fluorescent cockpit controls.
 One was completed with a pressurized cabin on an experimental basis and designated XP-38A.
 Due to reports the USAAF was receiving from Europe, the remaining 36 in the batch were upgraded with small improvements such as
self-sealing fuel tanks and enhanced armor protection to make them combat-capable. The USAAF specified that these 36 aircraft were to be designated P-38D. As a result, there never were any P-38Bs or P-38Cs. The P-38D's main role was to work out bugs and give the USAAF experience with handling the type.
In March 1940, the
French and the
British, through the
Anglo-French Purchasing Committee, ordered a total of 667 P-38s for US$100M,
 designated Model 322F for the French and Model 322B for the British. The aircraft would be a variant of the P-38E. The overseas Allies wished for complete commonality of Allison engines with the large numbers of
Curtiss P-40 Tomahawks both nations had on order, and thus ordered the Model 322 twin right-handed engines instead of counter-rotating ones and without turbo-superchargers.
[Note 3] Performance was supposed to be 400 mph (640 km/h) at 16,900 ft (5,200 m).
 After the fall of France in June 1940, the British took over the entire order and gave the aircraft the
service name "Lightning." By June 1941, the War Ministry had cause to reconsider their earlier aircraft specifications based on experience gathered in the
Battle of Britain and
 British displeasure with the Lockheed order came to the fore in July, and on 5 August 1941 they modified the contract such that 143 aircraft would be delivered as previously ordered, to be known as "Lightning (Mark) I," and 524 would be upgraded to US-standard P-38E specifications with a top speed of 415 mph (668 km/h) at 20,000 ft (6,100 m) guaranteed, to be called "Lightning II" for British service.
 Later that summer an RAF test pilot reported back from Burbank with a poor assessment of the "tail flutter" situation, and the British cancelled all but three of the 143 Lightning Is.
 As a loss of approximately US$15M was involved, Lockheed reviewed their contracts and decided to hold the British to the original order. Negotiations grew bitter and stalled.
 Everything changed after the 7 December, 1941
attack on Pearl Harbor after which the United States government seized some 40 of the Model 322s for
West Coast defense;
 subsequently all British Lightnings were delivered to the USAAF starting in January 1942. The USAAF lent the RAF three of the aircraft, which were delivered by sea in March 1942
 and were test flown no earlier than May
Cunliffe-Owen Aircraft Swaythling, the
Aeroplane and Armament Experimental Establishment and the
Royal Aircraft Establishment.
 The A&AEE example was unarmed, lacked turbochargers and restricted to 300 mph (480 km/h); though the undercarriage was praised and flight on one engine described as comfortable.
 These three were subsequently returned to the USAAF; one in December 1942 and the others in July 1943.
 Of the remaining 140 Lightning Is, 19 were not modified and were designated by the USAAF as RP-322-I ('R' for 'Restricted', because non-counter-rotating propellers were considered more dangerous on takeoff), while 121 were converted to non-turbo-supercharged counter-rotating V-1710F-2 engines and designated P-322-II. All 121 were used as
advanced trainers; a few were still serving that role in 1945.
 A few RP-322s were later used as test modification platforms such as for smoke-laying canisters. The RP-322 was a fairly fast aircraft below 16,000 ft (4,900 m) and well-behaved as a trainer.
One result of the failed British/French order was to give the aircraft its name. Lockheed had originally dubbed the aircraft
Atalanta from Greek mythology in the company tradition of naming planes after mythological and celestial figures, but the RAF name won out.
The strategic bombing proponents within the USAAF, called the
Bomber Mafia by their ideological opponents, had established in the early 1930s a policy against research to create long-range fighters, which they thought would not be practical; this kind of research was not to compete for bomber resources. Aircraft manufacturers understood that they would not be rewarded if they installed subsystems on their fighters to enable them to carry
drop tanks to provide more fuel for extended range. Lieutenant Kelsey, acting against this policy, risked his career in late 1941 when he convinced Lockheed to incorporate such subsystems in the P-38E model, without putting his request in writing. It is possible that Kelsey was responding to Colonel
George William Goddard's observation that the US sorely needed a high-speed, long-range
photo reconnaissance plane. Along with a
change order specifying some P-38Es be produced without guns but with photo reconnaissance cameras, to be designated the F-4-1-LO, Lockheed began working out the problems of drop tank design and incorporation. After the attack on Pearl Harbor, eventually about 100 P-38Es were sent to a modification center near
Dallas, Texas, or to the new Lockheed assembly plant B-6 (today the
Burbank Airport), to be fitted with four K-17
aerial photography cameras. All of these aircraft were also modified to be able to carry drop tanks. P-38Fs were modified as well. Every Lightning from the P-38G onward was drop tank-capable off the assembly line.
In March 1942, General Arnold made an off-hand comment that the US could avoid the
German U-boat menace by flying fighters to the UK (rather than packing them onto ships).
President Roosevelt pressed the point, emphasizing his interest in the solution. Arnold was likely aware of the flying radius extension work being done on the P-38, which by this time had seen success with small drop tanks in the range of 150 to 165 US gal (570 to 620 L), the difference in capacity being the result of subcontractor production variation. Arnold ordered further tests with larger drop tanks in the range of 300 to 310 US gal (1,100 to 1,200 L); the results were reported by Kelsey as providing the P-38 with a 2,500-mile (4,000 km) ferrying range.
 Because of available supply, the smaller drop tanks were used to fly Lightnings to the UK, the plan called
Led by two
Boeing B-17 Flying Fortresses, the first seven P-38s, each carrying two small drop tanks, left
Presque Isle Army Air Field on June 23, 1942 for
RAF Heathfield in Scotland. Their first refueling stop was made in far northeast Canada at
Goose Bay. The second stop was a rough airstrip in Greenland called
Bluie West One, and the third refueling stop was in Iceland at
Keflavik. Other P-38s followed this route with some lost in mishaps, usually due to poor weather, low visibility, radio difficulties and navigational errors. Nearly 200 of the P-38Fs (and a few modified Es) were successfully flown across the Atlantic in July–August 1942, making the P-38 the first USAAF fighter to reach Britain and the first fighter ever to be delivered across the Atlantic under its own power.
 Kelsey himself piloted one of the Lightnings, landing in Scotland on July 25.