Author Archives: William Pearce


Napier H-24 Sabre Aircraft Engine

By William Pearce

Aircraft engine designer Frank Bernard Halford believed that an engine using a multitude of small cylinders running at a relatively high rpm would be smaller, lighter, and just as powerful as an engine with fewer, large cylinders running at a lower rpm. Halford was contracted by the British engineering firm D. Napier & Son (Napier) in 1928 and built the Rapier I (E93) in 1929 and the Dagger I (E98) in 1933. Both of these air-cooled engines had a vertical H configuration, with the Rapier having 16-cylinders and the Dagger having 24-cylinders. Ultimately, the 539 cu in (8.83 L) Rapier VI (possibly E106) produced 395 hp (295 kW) at 4,000 rpm in 1936, and the 1,027 cu in (16.84 L) Dagger VIII (E110) produced 1,000 hp (746 kw) at 4,200 rpm in 1938.


The Napier Sabre’s block-like exterior hid the engine’s complicated internals of 24-cylinders, two crankshafts, sleeve-valves, and numerous drives. The Sabre VA seen here was the last variant to reach quantity production. (Napier/NPHT/IMechE image)

Back around 1930, Napier Chairman Montague Stanley Napier and the company’s Board of Directors sought to diversify into the diesel aircraft engine field. Montague Napier and Bill Nowlan laid out the design for a liquid-cooled, vertical H, 24-cylinder diesel engine that used sleeve valves. Given the Napier designation E101, the engine had a 5.0 in (127 mm) bore, a 4.75 in (121 mm) stroke, and a total displacement of 2,239 cu in (36.68 L). Montague Napier passed away on 22 January 1931, but Nowlan continued design work under the direction of George Shakespeare Wilkinson, Ronald Whitehair Vigers, and Ernest Chatterton. Wilkinson took out a patent for the sleeve drive (GB363850, application dated 7 January 1931), and Vigers took out patents for sealing rings on a plug-type cylinder head (GB390610, application dated 15 February 1932) and sleeve-valves (GB408768, application dated 24 January 1933). It appears the E101 diesel was abandoned around 1933. However, two- and six-cylinder test engines had been built to test the sleeve-drive mechanism and prove the validity of the entire design.

In 1935, Halford joined Napier’s Board of Directors, acting as the company’s Technical Director. Halford was disappointed that the Rapier and Dagger were not more successful. He decided to design a new, larger, 24-cylinder, H-configuration engine that would be capable of 2,000 hp (1,491 kW). The design for at least part of the new engine was based on the E101 diesel. As he had done with the Rapier, Halford showed his design to George Purvis Bulman, the Deputy Director of Engine Research and Development for the British Air Ministry. Bulman was aware that designers of fighter aircraft were interested in such an engine and was able to arrange financial support for Napier to develop the H-24 engine. Halford’s 2,000 hp (1,491 kW) engine was given the Napier designation E107 and became known as the Sabre.

Serious design work on the Sabre started in 1936. The spark-ignition engine had a similar layout to the E101 diesel—both being liquid-cooled H-24s with sleeve-valves and possessing the same bore and stroke. Liquid-cooling was selected to efficiently reject the heat that the compact 2,000 hp (1,491 kW) engine generated, and a mixture of 70 percent water and 30 percent ethylene-glycol would be used. The Air Ministry enabled the free flow of information between Napier, Halford, and Harry Ralph Ricardo—a British engine expert who had been researching sleeve-valve engines for quite some time. With the engine technology known in the early 1930s, a perception existed that the poppet-valve engine had reached its developmental peak. Sleeve-valves were seen as a way to extract more power out of internal combustion engines. The sleeve-valve offered large, unobstructed intake and exhaust ports, a definite advantage to achieve a full charge into the cylinder and complete scavenging of the exhaust when the engine is operating at high RPMs.


A drawing of a Sabre II, which was the main production variant. Note the two-sided supercharger impeller and the location of the supercharger clutch at the rear of the engine. The design of these components was changed for the Sabre IV and later variants. All accessories are mounted neatly atop the engine. (AEHS image)

The layout of the engine was finalized as a horizontal H-24. The Napier Sabre had a two-piece aluminum crankcase that was split vertically on the engine’s centerline. Sandwiched between the crankcase halves was an upper and lower crankshaft, each secured by seven main bearings. The center main bearing was larger than the rest, which resulted in an increased distance between the third and fourth cylinders in each bank. The crankshafts were phased at 180 degrees, and a cylinder for each crankshaft fired simultaneously. The single-piece, six-throw crankshafts were identical, and both rotated counterclockwise when viewed from the rear of the engine. Fork-and-blade connecting rods were used, with the forked rods serving the three front-left and three rear-right cylinders of the upper banks and the three front-right and three rear-left cylinders of the lower banks.

A 21-tooth spur gear on the front of each crankshaft meshed with two compound reduction gears, each with 31 teeth. A 17-tooth helical gear on the opposite side of each of the four compound reduction gears drove the 42-tooth propeller shaft counterclockwise. The drive setup created a double gear reduction, with the compound reduction gears operating at .6774 times crankshaft speed and the propeller shaft operating at .4048 times the speed of the compound reduction gears. The final gear reduction of the propeller shaft was .2742 crankshaft speed. A balance beam was mounted to the front of the two upper and the two lower compound reduction gears. A volute spring acted on each side of the beam to equally balance the tooth loading of the helical reduction gears on the propeller shaft. The forward ends of the compound reduction gears were supported by a gear carrier plate that was sandwiched between the crankshaft and the propeller shaft housing. The propeller shaft, balance beams, and volute springs were secured by the propeller shaft housing that bolted to the front of the engine.


Sectional view through a Sabre cylinder block showing the upper and lower cylinders paired by the sleeve-valve drive. Intake and exhaust passageways were cast into the cylinder block, and coolant flowed through the hollow cylinder head. Note that the sleeve extends quite a distance between the cylinder head and cylinder wall. Also note the supercharger torsion bar extending through the hollow sleeve-valve drive shaft. (AEHS image)

Attached to each side of the crankcase was a one-piece, aluminum cylinder block that consisted of an upper and a lower cylinder bank, each with six cylinders. With the exception of a few installed studs, the left and right cylinder blocks were interchangeable. A two-piece sleeve-valve drive shaft was mounted between each cylinder block and the crankcase, and it ran between the upper and lower cylinder banks. Each sleeve-valve drive shaft was driven at crankshaft speed through a layshaft by an upper compound reduction gear. The left and right sleeve-valve drive shafts each had six worm gears with 11 teeth, and each worm gear drove the sleeves for an upper and a lower cylinder pair via a 22-tooth worm wheel made from bronze. This setup enabled the sleeves to operate at half crankshaft speed (and half the speed of the sleeve-valve drive shaft). The worm wheels and their separate housings were mounted to the inner sides of the cylinder blocks. Each worm wheel had an upper and lower sleeve crank, which were phased at 180 degrees. Each sleeve crank drove a sleeve via a ball joint mounted on a lug on the outer bottom of the sleeve. The rotational movement of the sleeve crank caused the sleeve to reciprocate and oscillate in the cylinder bore. In addition, when the sleeve for the upper cylinder was rotating clockwise, the sleeve for the paired lower cylinder rotated counterclockwise. Due to the opposite rotation, the sleeves for the upper and lower cylinder banks had different (mirrored) port shapes. Each sleeve-valve drive shaft was supported by 14 bearings, with each of the six worm wheel housings incorporating two bearings.

Each sleeve-valve drive shaft was hollow and had a supercharger torsion bar running through its center. The two supercharger torsion bars acted on a compound supercharger gear at the rear of the engine. Via a fluid-actuated clutch, the two-speed supercharger was driven at 4.48 times crankshaft speed in low gear (often called moderate supercharging, MS) and 6.62 times crankshaft speed in high gear (often called full supercharging, FS). The supercharger’s centrifugal impeller was double-sided. Air was drawn in through a four-barrel updraft SU (Skinner’s Union) suction carburetor and fed into the impeller. The air and fuel mixture was distributed from the supercharger housing via one of four outlets to a cast aluminum manifold that ran along the outer side of each cylinder bank.

When ports in the sleeve-valve aligned with three intake ports cast into the cylinder, the air and fuel mixture was admitted into the cylinder. As the sleeve rotated and ascended, the ports closed. Two spark plugs mounted parallel to one another in the cylinder head ignited the mixture, initiating the power stroke. As the sleeve rotated back and descended, the cylinder’s two exhaust ports were uncovered to allow the gasses to escape between the upper and lower cylinder banks. The sleeve’s stroke was approximately 2.5 in (64 mm), and its full rotation was approximately 56 degrees (its rotary movement being approximately 28 degrees back and forth from center). Each sleeve had only four ports, one of which was used for both intake and exhaust. Valve timing had the intake ports opening 40 degrees before top dead center and closing 65 degrees after bottom dead center. The exhaust ports opened 65 degrees before bottom dead center and closed 40 degrees after top dead center. Intake and exhaust ports were simultaneously partially uncovered for 80 degrees of crankshaft rotation—the last 40 degrees of the exhaust stroke and the first 40 degrees of the intake stroke. Twelve exhaust ports were located in a single line on each side of the engine, and each ejector exhaust stack served two ports—one for an upper cylinder and one for a lower cylinder.


A Sabre engine being assembled. In the foreground are the individual cylinder heads with their sealing rings. In the row above the heads is a long, slim shaft that is the supercharger torsion bar. It passes through the two-piece sleeve-valve drive shaft. Further right are six sleeve-valve cranks, followed by their housings, and a set of 12 sleeves. The crank end of the sleeve is up, and note the helical grooves for oil control. Next is a row of pistons sitting inverted, each with rings and a piston pin. On the next row is a crankshaft being worked on and a set of six fork-and-blade connecting rods. Further to the right is another set of connecting rods that are already attached to the other crankshaft (out of frame). The lady furthest from the camera is working on the four compound reduction gears that will take power from the two crankshafts and deliver it to the propeller shaft, which is being held in a wooden fixture in front of her. On the far left, behind the ladies, is a Sabre cylinder block with numerous studs to attach the cylinder bank. Next is an upper accessory housing with some accessories attached. Last is a lower accessory housing with fuel, water (both external), and oil (internal) pumps.

The forged aluminum pistons were rather short with a minimal skirt, which was required for the engine’s relatively short stroke, use of sleeve-valves, and narrow width. Each flat-top piston had two compression rings above the piston pin, with one oil scraper ring below. The top ring was later tapered to prevent the buildup of carbon. The piston operated directly in the sleeve-valve, which was .09375 in (2.4 mm) thick and made from forged chrome-molybdenum steel. When the piston was at the bottom of its stroke, it was almost completely removed from the cylinder and supported only by the sleeve. The sleeves had a hardened belt on their inner diameter at the top of the piston stroke. Helical grooves inside the lower part of the sleeve helped prevent excessive oil accumulation on the sleeve walls. Oil was controlled further by an oil scraper fitted at the bottom of the sleeve between its outer diameter and the cylinder. The top of each cylinder was sealed by a cast aluminum cylinder head. The cylinder head acted as a plug atop the cylinder and was sealed against the sleeve by a compression ring. The top of the sleeve extended between the cylinder head and the cylinder wall. The cylinder head incorporated coolant passages that communicated with passages in the cylinder block. The engine had a compression ratio of 7.0 to 1.

The upper and lower crankshafts also respectively drove upper and lower auxiliary drive shafts. These auxiliary drive shafts were contained in their own separate housings which were respectively attached to the upper and lower sides of the assembled engine. The upper auxiliary drive shaft powered a vacuum pump, the propeller governor, two distributors, two magnetos, a generator, an air compressor, a hydraulic pump, and an oil pump for the supercharger. All of this equipment was mounted as compactly as possible to the top of the engine. The lower auxiliary drive shaft powered left and right coolant pumps, a fuel pump, and various oil pumps. The coolant and fuel pumps were mounted below the engine, while the oil pumps were internal. The coolant pumps provided a combined flow of 367 US gpm (306 Imp gpm / 1,389 L/min). Also mounted atop the engine and geared to the rear of the upper crankshaft was the Coffman combustion starter unit. The starter had a five-cartridge capacity.

The upper and lower cylinders were numbered 1–12, starting from the left rear and proceeding clockwise to the right rear. With the simultaneous firing of a cylinder for each crankshaft, the engine’s firing order was Top 1/Bottom 6, T9/B10, T5/B2, T12/B7, T3/B4, T8/B11, T6/B1, T10/B9, T2/B5, T7/B12, T4/B3, and T11/B8. Four mounting pads on the underside of the engine attached it to the support structure in the aircraft. The basic design of the Sabre enabled easy access for routine maintenance. Once the aircraft’s cowling was removed, crews had unobstructed access to all of the spark plugs on the sides of the engine and all accessories mounted atop the engine.


A Sabre IIB being pulled from a Typhoon IB. Note the coolant header tank at the front of the engine, the accessories packaged atop the engine, the two-into-one exhaust stacks, and the hydraulic supercharger clutch at the rear of the engine. The cylinder housing for the five-cartridge Coffman starter can be seen above the supercharger.

The Napier Sabre I (E107) engine had a 5.0 in (127 mm) bore and a 4.75 in (121 mm) stroke. With a bore diameter greater than the stroke length, the Sabre was an over-square engine. Each cylinder displaced 93.2 cu in (1.53 L), and the engine’s total displacement was 2,239 cu in (36.68 L). At 3,700 rpm, the Sabre I produced 2,050 hp (1,529 kW) at 2,500 ft (762 m) with 7 psi (.48 bar) of boost and 1,870 hp (1,394 kW) at 14,500 ft (4,420 m) with 8 psi (.55 bar) of boost. The engine was 81.1 in (2.06 m) long, 40.0 in (1.02 m) wide, and 51.1 in (1.30 m) tall. The Sabre I weighed 2,360 lb (5,203 kg).

Sabre development at Napier’s works in Acton, England progressed quickly, and single-, twin-, and six-cylinder test engines were all running by the end of 1936. The first of four 24-cylinder prototype engines was run on 23 November 1937, and the Air Ministry ordered six additional test engines by December. In January 1938, the Sabre passed initial acceptance tests with a rating of 1,350 hp (1,007 kW), and on 3 March, the Air Ministry ordered two Sabre-powered Hawker Typhoon fighter prototypes. Also in March, the engine passed a 50-hour test that included a peak output of 2,050 hp (1,529 kW). All ordered engines were completed by the end of 1938 and were running on test stands by February 1939. While testing continued, the Sabre I was first flown in a Fairey Battle on 31 May 1939, piloted by Chris Staniland. As installed in the Battle, the Sabre had a single exhaust manifold on each side of the engine that collected the exhaust from all 12 cylinders.

In July 1939, the Air Ministry ordered 100 production engines and material for another 100 engines. In August, the Sabre passed a type test with a rating of 1,800 hp (1,342 kW). On 8 October 1939, an order for 250 Typhoons was placed, and on 24 February 1940, the Typhoon prototype (P5212) made its first flight, piloted by Philip G. Lucas. Three four-into-one exhaust manifolds were originally installed on each side of the Typhoon’s Sabre, but these were quickly replaced by what would become the standard two-into-one exhaust stacks. In March 1940, Napier created its Flight Development Establishment at Luton, England for flight testing the Sabre and developing installations for the engine. By all accounts, the Sabre continued to perform well, although some vibration issues were experienced with the Typhoon. In June 1940, the engine passed a 100-hour type test with a maximum output of 2,050 hp (1,529 kW) at 3,700 rpm, making the Sabre the first engine to have a service rating over 2,000 hp (1,491 kW).


The installation of Sabre engines on the Fairly Battle (top) and Folland F.108 (bottom) were well executed. Two Battles and three Fo.108s were employed to test the Sabre, and these aircraft provided valuable information about the engine.

Since mid-1938, a plan was underway to use an uprated Sabre engine in a specially-designed aircraft for a speed record attempt. The special engine produced 2,450 hp (1,827 kW) at 3,800 rpm with 9.2 psi (.63 bar) of boost and was first run on 6 December 1939. Installed in the Napier-Heston Racer, the combination first flew on 12 June 1940, piloted by G. L. G. Richmond. Difficulties with the new engine and airframe resulted in a hard landing that damaged the aircraft beyond repair. The Sabre engine installed in the Napier-Heston Racer featured two six-into-one exhaust manifolds on each side of the engine.

Around November 1939, the Air Ministry ordered 500 examples of the Typhoon. This order was temporarily suspended due to the Battle of Britain but was reinstated in October 1940. At that time, Napier began work to produce additional Sabre engines for the Typhoon order, but production was still a very limited affair. These early engines were limited to 25 hours before being removed for major inspection. The first production Typhoon IA (R7576) flew on 27 May 1941, with other aircraft soon to follow. Nearly all Sabre I engines were used in Typhoon IAs.


With its 14 ft (4.27 m) three-blade propeller turning, this early Typhoon IB warms up its Sabre engine for a flight. The Typhoon IB had four 20 mm cannons, while the earlier IA had 12 .303 machine guns. At the center of the radiator is the open carburetor intake, which was later covered by a momentum air filter. Note the underwing identification/invasion stripes

Napier continued to develop the engine as the Sabre II, and the first production Sabre II was completed in January 1941. The Sabre II produced 2,090 hp (1,559 kW) at 3,700 rpm at 4,000 ft (1,219 m) with 7 psi (.48 bar) of boost and 1,735 hp (1,294 kW) at 17,000 ft (5,182 m). Sabre II engines were first installed in Typhoons on a trial basis in June 1941, and the engine was cleared for 50 hours between major inspections around this time. The Sabre II would ultimately replace the Sabre I in Typhoon IAs and IBs, and the Sabre I was phased out around October 1941. In addition to the Typhoon, the Sabre II also powered the Martin-Baker MB3 fighter, which made its first flight on 31 August 1942, and the Hawker Tempest V fighter prototype (HM595), which made its first flight on 2 September 1942, piloted by Lucas. The Tempest V was a new aircraft developed from the Typhoon.

The Folland Fo.108 was built to Air Ministry Specification 43/37 calling for an engine testbed aircraft. Three of the fixed-gear monoplanes were delivered to Napier’s Flight Development Establishment at Luton in 1941 and were initially fitted with Sabre II engines. The aircraft were to serve Napier for several years testing various versions of the Sabre engine. One of the Sabre-powered aircraft was lost on 14 September 1944.

The Sabre III was similar to the II but was intended for higher engine speeds. The Sabre III was selected for the Blackburn B-37 Firebrand carrier strike aircraft. At 4,000 rpm, the Sabre III had a takeoff rating of 2,250 hp (1,678 kW) and military ratings of 2,310 hp (1,723 kW) at 2,500 ft (762 m) with 9 psi (.62 bar) of boost and 1,920 hp (1,432 kW) at 16,000 ft (4,877 m). At 3,500 rpm, the engine had a normal rating of 1,890 hp (1,409 kW) at 5,000 ft (1,524 m) and 1,630 hp (1,215 kW) at 16,500 ft (5,029 m). The Firebrand (DD804) was first flown on 27 February 1942. However, with production priority going to the Typhoon, the Ministry of Aircraft Production decided to reengine the Firebrand with the Bristol Centaurus sleeve-valve radial engine. Only around 24 of the Sabre-powered versions were built.


The Blackburn Firebrand, was to be powered by the Sabre III. However, Sabre engine production was allocated to the Typhoon, and the Firebrand was reengined with the Bristol Centaurus. Pictured is DD815, the third Firebrand Mk I prototype.

With production engines in production airframes, Sabre reliability issues were soon encountered. After running for a few hours, sometimes not even passing initial tests, Sabre engines began to experience excessive oil consumption and sleeve-valves cracking, breaking, seizing or otherwise failing. Examinations of numerous engines found sleeves distorted or damaged. Since the Sabre’s main application was the Typhoon, it was that aircraft that suffered the most. To make matters worse, the Typhoon was experiencing its own issues with in-flight structural failures. Other aircraft suffered as well. On 12 September 1942, the Sabre II engine in the MB3 failed; the subsequent crash landing destroyed the prototype and killed the pilot, Valentine H. Baker.

The Sabre had performed admirably during testing, but the production engines were encountering issues at an alarming rate. The early engines were built and assembled by hand. Parts with small variances were matched to obtain the desired clearances and operation. This was a luxury that could not be afforded once the engine was mass produced. The sleeves were found to be .008 to .010 in (.203 to .254 mm) out of round. This caused the cascading failure of other components as the engine was operated. In addition, the piston was forming a ridge in the sleeve, leading to excessive wear and the eventual failure of the piston rings, piston, or sleeve.

Carbon build-up was causing issues with the lubrication system. While in flight, aeration of the oil resulted in a heavy mist of oil flowing from the breather and coating the cockpit, obscuring the pilot’s view. The Coffman cartridge starter caused other issues; its sudden jolt when starting the engine occasionally damaged sleeve-drive components, setting up their inevitable failure. Part of the starting issue was that the sudden rotation of the engine with a rich mixture washed away the oil film between the pistons and sleeves. Finally, service crews were misadjusting the boost controller, creating an over-boost situation that led to detonation in the cylinders and damaged engines.


The Tempest I was powered by the Sabre IV engine. At 472 mph (760 km/h), the aircraft was the fastest of the Tempest line. The Tempest I was rather elegant without the large chin radiator, and the wing radiators were similar to those that would be used on the Sabre VII-powered Fury.

Napier worked diligently to resolve the issues. A detergent-type oil was used to prevent the build up of carbon on internal components. A centrifugal oil separator was designed to deaerate the oil and was fitted to Sabre engines already installed in Typhoons. Changes were made to the starter drive, and a priming mixture of 70 percent fuel and 30 percent oil was utilized to maintain an oil film in the cylinders. The boost controllers were factory sealed, and severe repercussions were put in place for their unauthorized tampering.

The issues with sleeve distortion were the most serious and vexing. Methods were devised to measure the sleeve with special instruments via the spark plug hole. While this helped to prevent failures, it also caused the withdrawal of low-time engines as sleeves became distorted. To fix the issue, different sleeve materials were tried along with different processes of manufacture, but nothing seemed to work. The supply of Sabre engines fell behind the production of Typhoon aircraft, and engineless airframes sat useless at manufacturing facilities. The engine shortage was so severe that a good Sabre would be installed in a Typhoon to ferry the aircraft to a dispersal facility. The engine would then be removed, returned to the aircraft factory, and installed in another Typhoon to shuttle that aircraft away, repeating the process over and over.

In October 1941, Francis Rodwell ‘Rod’ Banks replaced Bulman, who was, at the time, the Director of Engine Production for the Ministry of Aircraft Production. Bulman was back in Engine Research and Development and continued to work with Halford and Napier to resolve issues with the Sabre. Banks suggested that Napier work with the Bristol Engine Company on a suitable sleeve for the Sabre. Bristol had been manufacturing radial sleeve-valve engines since 1932, and their Taurus engine had the same 5.0 in (127 mm) bore as the Sabre. Napier was apparently not interested in pursuing that possible solution, so Banks went directly to Bristol and had them machine a pair of sleeves for use in the Sabre two-cylinder test engine. The Bristol sleeves were made from centrifugally cast austenitic steel comprised of nickel, chromium, and tungsten. The sleeve was nitrided to increase its hardness and was not more than .0002 in (.005 mm) out of round. The Sabre two-cylinder test engine with the Bristol sleeves ran 120 hours without issue. Banks then had Bristol produce 48 sleeves for two complete 24-cylinder Sabre test engines. Bristol became unhappy with sharing its components and processes with a competitor, and Napier was still hesitant to utilize Bristol’s materials and techniques.


The Sabre VA had a one-sided supercharger impeller, a relocated supercharger clutch, and a two-barrel injection carburetor. These refinements were introduced on the Sabre IV. The Sabre VA powered the Tempest VI. (Napier/NPHT/IMechE image)

With the Air Ministry’s push, Napier was taken over by English Electric in December 1942. The new management was happy to accept any assistance from Bristol, and Bristol was now more willing than ever to lend support. A lack of support from the Napier board of directors had caused Halford to give a three-month notice of resignation, and he left in early 1943 to focus on turbojet engines at the de Havilland Engine Company. However, Halford continued consulting work on the Sabre for a time. Before his departure from Napier, Halford’s Sabre designs had progressed up to the Sabre V. Ernest Chatterton took over Sabre development after Halford’s departure. Through all this, Bulman continued to work with Napier, but the Ministry of Aircraft Production handed all responsibility for the Sabre engine to Banks in early 1943. To get engine production up to speed, Sundstrand centerless grinders made in the United States and destined for a Pratt & Whitney factory producing R-2800 C engines were rerouted to Napier’s Sabre production facility in Liverpool. While it is not entirely clear how Banks felt at the time, he later wondered what would have become of the Fairey Monarch H-24 engine if the Air Ministry and the Ministry of Aircraft Production had encouraged its development with the same financial and technological resources supplied for the Sabre.

In the spring of 1943, some 1,250 engines had accumulated a total of 12,000 hours of testing and 40,000 hours of service use, and the Sabre’s service life was extended from 25 hours to 250 hours between major inspections. With Sabre reliability issues resolved and production resuming, development of the engine continued. The Sabre IV incorporated a two-barrel Hobson-RAE injection carburetor and a revised supercharger with a single-sided impeller. The supercharger clutches were updated and relocated from the extreme rear of the supercharger to between the supercharger and the engine. Revised gears turned the impeller at 4.68 times crankshaft speed in low gear and 5.83 times crankshaft speed in high gear. The Sabre IV produced 2,240 hp (1,670kW) at 4,000 rpm at 8,000 ft (2,438 m) with 9 psi (.62 bar) of boost. The engine was selected for the Tempest I, the prototype of which was initially ordered on 18 November 1941, followed by an order for 400 production aircraft in August 1942. The Tempest I featured a streamlined nose and its radiator and oil cooler were installed in the wing’s leading edge. The prototype Tempest I (HM599) was first flown on 24 February 1943, piloted by Lucas, and would go on to record a speed of 472 mph (760 km/h) at 18,000 ft (5,486 m) in September 1943. However, delays and development issues with the Sabre IV engine led to the Tempest I order being converted to Sabre IIA and IIB-powered Tempest Vs.

The Sabre IIA (E115) was a refinement of the Sabre II and had been developed in mid-1943. The engine had a modified oil system and used dynamically-balanced crankshafts. The Sabre IIA had a takeoff rating of 1,995 hp (1,488 kW) at 3,750 rpm with 7 psi (.48 bar) of boost. At 3,750 rpm and 9 psi (.62 bar) of boost, the engine had a military rating of 2,235 hp (1,667 kW) at 2,500 ft (762 m) and 1,880 hp (1,402 m) at 15,250 ft (4,648 m). At 3,700 rpm and 7 psi (.48 bar) of boost, the engine had a normal rating of 2,065 hp (1,540 kW) at 4,750 ft (1,448 m) and 1,735 hp (1,294 kW) at 17,000 ft (5,182 m). Fuel consumption at cruise power was .46 lb/hp/hr (280 g/kW/h). Starting around August 1943, Sabre IIA engines were incorporated into production Typhoon IB and Tempest V Series I aircraft.


Cutaway drawing of a Sabre VA illustrating the engine’s propeller reduction gears and sleeve-valve drive. Note the upper and lower accessory drives, the slight fore-and-aft angling of the spark plugs, and the single-sided supercharger impeller. (Napier/NPHT/IMechE images)

In 1944, prototypes of the Sabre IIB (E107A) became available. Compared to the Sabre IIA, the IIB used a different carburetor, had a modified boost controller, and was cleared for additional engine speed. The Sabre IIB had a takeoff rating of 2,010 hp (1,499 kW) at 3,850 rpm with 7 psi (.48 bar) of boost. At 3,850 rpm with 11 psi (.76 bar) of boost, the engine had a military rating of 2,400 hp (1,790 kW) at sea level, 2,615 hp (1,950 kW) at 2,500 ft (762 m), and 2,045 hp (1,525 kW) at 13,750 ft (4,191 m). The Sabre IIB had the same normal rating as the IIA. The engine was used in later Typhoon IBs and was the main Sabre version to power the Tempest V Series II.

The Sabre IIC (E107B) was a similar to the IIB but with new supercharger gears. The impeller turned at 4.73 times crankshaft speed in low gear and at 6.26 times crankshaft speed in high gear. The engine had a takeoff rating of 2,065 hp (1,540 kW) at 3,850 rpm. At the same engine speed and with 11 psi (.76 bar) of boost, the military power rating was 2,400 hp (1,790 kW) at 2,000 ft (610 m) and 2,045 hp (1,525 kW) at 13,750 ft (4,191 m). The Sabre IIC was used in some late production examples of the Tempest V, including those converted as target tugs in 1948.

The Sabre V (E107C) was developed from the IV with an updated carburetor. Linkages were incorporated to allow one lever to control the engine’s throttle and the propeller’s pitch along with automatic boost and mixture control, but this system could be overridden by the pilot. The spark plugs were repositioned, although it is not clear if this change was made on the Sabre V or the Sabre VA engine. Rather than being parallel, as in earlier Sabre engines, the electrode of the front spark plug was angled forward, and the electrode of the rear spark plug was angled back. The engine produced 2,420 hp (1,805 kW) at 3,750 rpm at 4,250 ft (1,295 m) with 15 psi (1.0 bar) of boost. The Sabre V was tested in the Tempest I, and the combination was first flown on 8 February 1944 by Bill Humble. On 12 February, an order for 700 Sabre V-powered Tempest Is was issued. This order was later reduced to 300 examples, and then converted to the Sabre V-powered Tempest VI in May. The prototype Tempest VI (HM595 again) made its first flight on 9 May 1944, piloted by Humble. Cooling the more powerful engine in warmer climates required modifications to be incorporated into the Tempest VI, including a larger chin radiator and a secondary oil cooler in the wing. Carburetor inlets were also relocated to the wing’s leading edge. Otherwise, the aircraft was similar to the Tempest V.


A Tempest V Series I (top) and Tempest VI (bottom). The Tempest V Series I had Hispano Mk II cannons with long barrels that protruded from the wing’s leading edge. The Tempest V Series II and other Tempests had Hispano Mk V cannons with short barrels. The Sabre VA-powered Tempest VI (bottom) has an enlarged chin radiator, an oil cooler in the wing, and carburetor inlets in both wing roots.

The Sabre VA was essentially the production version of the Sabre V. The Sabre VA had a takeoff rating of 2,300 hp (1,715 kW) at 3,850 rpm with 12 psi (.83 bar) of boost. The engine’s military rating at 3,850 rpm with 15 psi (1.0 bar) of boost was 2,600 hp (1,939 kW) at 2,500 ft (762 m) and 2,300 hp (1,715 kW) at 13,750 ft (4,191 m). At 3,650 rpm, the Sabre VA had a normal rating of 2,165 hp (1,614 kW) at 6,750 ft (2,057 m) and 1,930 hp (1,439 kW) at 18,000 ft (5,486 m). Cruise power at 3,250 rpm was 1,715 hp (1,279 kW) at 6,750 ft (2,057 m) and 1,565 hp (1,167 kW) at 14,250 ft (4,343 m). Fuel consumption at cruise power was .50 lb/hp/hr (304 g/kW/h). The engine was 82.2 in (2.10 m) long, 40.0 in (1.02 m) wide, and 46.0 in (1.17 m) tall. The Sabre VA weighed 2,500 lb (1,134 kg). Starting around March 1946, the engine was the powerplant for production Tempest VI aircraft.

The Sabre VI was the same engine as the Sabre VA, but it incorporated an annular nose radiator and provisions for a cooling fan, all packaged in a tight-fitting cowling. The cooling fan rotated clockwise, the opposite direction from the propeller. The intent of the engine and cooling system combination was to produce a complete low-drag installation package that would cool the engine sufficiently for use in tropical climates. The radiator incorporated cooling elements for both engine coolant and oil. Napier and Hawker experimented with annular radiators using various Sabre IIB engines installed on a Typhoon IB (R8694) and a Tempest V (EJ518). In 1944, the Sabre VI with an annular radiator was test flown on a Tempest V (NV768). Numerous changes to the annular radiator and its cowling eventually led to the development of a ducted spinner, which was installed on NV768. The aircraft continued to test annular radiators through 1948. While the annular radiator added 180 lb (82 kg), it decreased drag by eight percent and improved the Tempest’s top speed by 12 mph (19 km/h). Two Sabre VI engines, each with an annular radiator and a cooling fan, were installed on a Vickers Warwick C Mk III (HG248) twin-engine transport. With the Sabre engines, the Warwick’s top speed was limited to 300 mph (483 km/h) due to its fabric covering. This was still about 75 mph (121 km/h) faster than the aircraft’s original design speed. Most of the annular radiator testing was conducted at Napier’s Flight Development Establishment at Luton. While some of the ducted spinner research was applied to the Napier Naiad turboprop, none of the work was applied to production piston engines.


The Sabre VI incorporated an annular radiator and provisions for an engine-driven cooling fan. Tempest V NV768 was used to test a number of different spinner and annular radiator cowling configurations with the Sabre VI. The aircraft is seen here with a large ducted spinner. The configuration slightly improved NV768’s performance over that of a standard Tempest. (Napier/NPHT/IMechE image)

The Sabre VII carried the Napier designation E121 and was essentially a VA engine strengthened to endure higher outputs. The engine was fitted with water/methanol (anti-detonant) injection that sprayed into the supercharger via an annular manifold. The mixture used was 40 percent water and 60 percent methanol. The water/methanol injection lowered the engine’s tendency toward detonation and allowed for more power to be produced. The supercharger housing was reworked for the water/methanol injection, and the cylinder heads were modified to accommodate two compression rings. Individual ejector exhaust stacks were fitted, replacing the two-into-one stacks previously used on most Sabre engines.

Initially, the Sabre VII had a takeoff rating of 3,000 hp (2,237 kW) at 3,850 rpm with water/methanol injection and 17.25 psi (1.19 bar) of boost. This was later increased to 3,500 hp (2,610 kW) at the same rpm with 20 psi (1.38 bar) of boost. The engine’s military rating at 3,850 rpm with 17.25 psi (1.19 bar) of boost and water/methanol injection was 3,055 hp (2,278 kW) at 2,500 ft (762 m) and 2,820 hp (2,103 kW) at 12,500 ft (3,810 m). The water/methanol injection flow rate was 76 US gph (66 Imp gph / 300 L/min) at takeoff, 78 US gph (65 Imp gph / 295 L/min) at military power in low supercharger, and 122 US gph (102 Imp gph / 464 L/min) at military power with high supercharger. The water/methanol flow rates corresponded to 30 percent of the fuel flow at low supercharger and 45 percent of the fuel flow at high supercharger. The Sabre VII’s fuel flow was 284 US gph (235 Imp gph / 1,068 L/min) at takeoff, 287 US gph (239 Imp gph / 1,087 L/min) at military power in low supercharger, and 289 US gph (241 Imp gph / 1,096 L/min) at military power with high supercharger. At 3,700 rpm and 10.5 psi (.73 bar) of boost, the Sabre VII had a normal rating of 2,235 hp (1,667 kW) at 8,500 ft (2,591 m) and 1,975 hp (1,473 kW) at 18,250 ft (5,563 m). Cruise power at 3,250 rpm was 1,750 hp (1,305 kW) at 8,500 ft (2,591 m) for a fuel consumption of .45 lb/hp/hr (274 g/kW/h), and 1,600 hp (1,193 kW) at 17,000 ft (5,182 m) for a fuel consumption of .51 lb/hp/hr (310 g/kW/h). The engine was 83.0 in (2.11 m) long, 40.0 in (1.02 m) wide, and 47.2 in (1.20 m) tall. The Sabre VII weighed 2,540 lb (1,152 kg). Some sources state that a Sabre VII engine achieved an output of 4,000 hp (2,983 kW) and was run at 3,750 hp (2,796 kW) for a prolonged period without issues during testing.


A Vickers Warwick C Mk III (HG248) was used to test the installation of the Sabre VI engine with an annular radiator and an engine-driven cooling fan. Note that the fan rotates in the opposite direction from the propeller and that the lower cowling folds down level to be used as a work platform. The rear four exhaust ejectors were replaced with elongated stacks to prevent excessive heat build-up on the wing’s leading edge. (Napier/NPHT/IMechE image)

The Sabre VII was intended to power the Hawker Fury Mk I, of which 200 were ordered in August 1944. Shifting priorities at the end of the war all but cancelled the aircraft, and only two prototypes were built. The first prototype (LA610) made its initial Sabre VII-powered flight on 3 April 1946. This aircraft would go on to record a speed of 483 mph (777 km/h) at 18,500 ft (5,639 m) and 422 mph (679 km/h) at sea level. The Sabre VII was also test-flown on a Tempest V or VI in mid-1946, but additional details have not been found. This aircraft had the larger radiator and wing root carburetor inlets of the Tempest VI, but it did not have the additional oil cooler in the wing.

The Sabre VIII carried the Napier designation E122 and was based on the Sabre VII. The engine incorporated contra-rotating propellers and a two-stage supercharger. Four aftercoolers were to be installed—one on each induction runner leading from the supercharger housing to the intake manifold attached to the cylinder bank. Although some sources say the Sabre VIII was built, it appears to have remained an unbuilt project. The engine was forecasted to have a military rating of 3,350 hp (2,498 kW) and be capable of 25 psi (1.72 bar) of boost.


A Sabre VII with its revised supercharger housing that accommodated water/methanol injection. The injection controller is mounted just above the supercharger housing. The Sabre VII ultimately produced 3,500 hp (2,610 kW) at 3,850 rpm with 20 psi (1.38 bar) of boost and was installed in the Hawker Fury Mk 1 prototype. (Napier/NPHT/IMechE image)

Production of the Sabre was halted shortly after the end of World War II with approximately 5,000 engines produced. Starting in October 1939, Napier worked to establish a shadow factory in Liverpool to produce Sabre engines. The first engine, a Sabre II, was completed at this factory in February 1942. The Liverpool site manufactured around 3,500 II, IIA, IIB, and VA engines, with the remaining 1,500 engines, including all prototypes, coming from Napier’s Acton works. With Sabre development at an end, Napier focused on their next aircraft engine, the two-stroke diesel/turbine compounded Nomad.

A number of engine designs based on the Sabre were considered, but most stayed as projects, and none progressed beyond cylinder testing. The E109 of 1939 was half of a Sabre, with 12-cylinders and a single crankshaft. It would have displaced 1,119 cu in (18.34 L). The E113 of 1940 was a fuel-injected, two-stroke, uniflow, Sabre-type test engine intended for increased engine speed and boost. The design concept originated with Harry Ricardo, and a two-cylinder test engine was built in 1942. Reportedly, the test engine was so loud that people on the street had to cover their ears as they passed by Napier’s works in Acton. The E120 of 1942 was a 32-cylinder Sabre consisting of four banks of eight cylinders. It would have displaced 2,985 cu in (48.91 L). The E123 of 1943 was a complete 24-cylinder, fuel-injected, two-stroke Sabre based on the E113 test engine. It had a forecasted output of 4,000 hp (2,983 kW) but was never built.

Although the Sabre was proposed for many projects that never left the drawing board and powered a few prototypes, the engine’s main applications were the 109 Typhoon IAs, 3,208 Typhoon IBs, 801 Tempest Vs, and 142 Tempest VIs produced during World War II. After the initial production difficulties, which were quite severe, the engine served with distinction. The Sabre could be difficult to start, and it was advisable to use a remote heater to pre-heat the coolant and oil in cold temperatures. Sleeve trouble came back with Typhoons stationed around Normandy, France in the summer of 1944. Fine dust particles from the soil were getting into the engines and causing excessive sleeve wear. A momentum air filter developed by Napier cured the trouble. The filter was designed and test flown the same day of its original request, and all the Typhoons in France were fitted with a filter within a week. Production of the Sabre was an expensive affair, with each horsepower costing four to five times that of the Rolls-Royce Merlin. However, Typhoons and Tempests played an important role in attacking German forces on the ground and countering V-1 flying bombs. Around a dozen Sabre engines survive and are on display in museums or held in private collections. As of 2020, there are no running Sabre engines, but efforts are underway to create running examples to power Typhoon and Tempest aircraft under restoration.


General arrangement drawing of the unbuilt Sabre VIII (E122). The engine featured a two-stage supercharger and contra-rotating propellers. It was forecasted to produce 3,350 hp (2,498 kW).

Major Piston Aero Engines of World War II by Victor Bingham (2001)
Allied Aircraft Piston Engines of World War II by Graham White (1995)
Aircraft Engines Volume Two by A. W. Judge (1947)
By Precision Into Power by Alan Vessey (2007)
An Account of Partnership – Industry, Government and the Aero Engine by George Bulman and edited by Mike Neale (2002)
I Kept no Diary by F. R. (Rod) Banks (1978)
Boxkite to Jet — the remarkable career of Frank B Halford by Douglas R Taylor (1999)
The Napier Way by Bryan ‘Bob’ Boyle (2000)
The Hawker Typhoon and Tempest by Francis K. Mason (1988)
Hawker Typhoon, Tempest and Sea Fury by Kev Darling (2003)
Tempest: Hawker’s Outstanding Piston-Engined Fighter by Tony Buttler (2011)
Hawker Aircraft since 1920 by Francis K. Mason (1991)
Blackburn Aircraft since 1909 by A. J. Jackson (1968/1989)
Aircraft Engines of the World 1945 by Paul H. Wilkinson (1945)
Aircraft Engines of the World 1946 by Paul H. Wilkinson (1946)
– “The Napier Sabre Engine Parts 1–3” by J. A. Oates, Aircraft Production Volume 6, Numbers 66–68 (April–June 1944) via The Aircraft Engine Historical Society
– “Napier Sabre II” by F. C. Sheffield, Flight (23 March 1944)
– “Napier Sabre VII” Flight (22 November 1945)
– “Napier Flight Development” Flight (25 July 1946)
Jane’s All the World’s Aircraft 1945/46 by Leonard Bridgman (1946)
Jane’s All the World’s Aircraft 1947 by Leonard Bridgman (1947)
Jane’s All the World’s Aircraft 1948 by Leonard Bridgman (1948)


Napier H-24 Dagger Aircraft Engine

By William Pearce

In 1928, independent aircraft engine designer Frank Bernard Halford was contracted by D. Napier & Son (Napier) to design aircraft engines with a displacement between 404.09 and 718.37 cu in (6.62 and 11.77 L). Halford’s first designs for Napier were the H-16 Rapier (Napier designation E93) of 1929 followed by the inverted I-6 Javelin (Napier designation E97) of 1931.


The Napier Dagger I air-cooled H-24 with its downdraft carburetor and propeller shaft in line with the engine’s centerline. “Napier Halford” can be seen on the upper camshaft housing. Note the two engine mounts on the side of the crankcase and third mount on the accessory housing. (Napier/NPHT/IMechE image)

Around 1932, Halford and Napier reached a new agreement, and the design of engines larger than the 718.37 cu in (11.77 L) limit were initiated. The first of these designs was a 24-cylinder development of the Rapier with an enlarged bore and elongated stroke. This engine was named the Dagger, and it carried the Napier designation E98. The engine was also called the Napier-Halford Dagger. Like the Rapier, the air-cooled Dagger was a high-revving aircraft engine with numerous small cylinders and minimal frontal area. Halford’s belief was that a smaller engine running at higher speeds would produce the same power as a larger engine running at slower speeds.

The Dagger had a vertical H configuration with four cylinder banks, each with six cylinders. The two-piece aluminum crankcase was split horizontally at its center. The two crankcase halves supported left and right crankshafts via seven main bearings each. An eighth crankshaft bearing was located in the gear reduction housing. Each one-piece, six-throw crankshaft served one vertical and one inverted bank of cylinders. The crankshafts were phased at 30 degrees with power strokes occurring sequentially between the two crankshafts. The connecting rods were of the fork-and-blade type, with the forked rods serving the upper front three cylinders on the left side of the engine and the upper rear three cylinders on the right side of the engine. Spur gears at the front of each crankshaft meshed with a larger gear mounted to the propeller shaft, which turned at .372 crankshaft speed. When viewed from the rear, both crankshafts rotated clockwise, and the propeller shaft rotated counterclockwise.


A Dagger II engine preserved and in storage as part of the Smithsonian National Air and Space Museum. The engine appears complete with its upper and lower air ducts as well as the baffling around the cylinders. At one time, this particular Dagger II belonged to the US Navy. The engine data plate says “Halford-Napier Dagger.” (NASM image)

The individual cylinders were made from forged steel barrels with cast aluminum heads. The heads for each cylinder bank were first installed to a common camshaft housing and then drawn down on the cylinder barrels via four studs protruding from the crankcase around each cylinder opening. An aluminum sealing ring was sandwich between the cylinder head and barrel. The cylinders had a 7.75 to 1 compression ratio, and each cylinder had a single intake and a single sodium-cooled exhaust valve. The intake port was on the inner side of the cylinder, and the exhaust port was on the outer side. The valves for each cylinder bank were actuated via rockers and tappets by a single overhead camshaft. The self-adjusting hydraulic valve tappets were designed by Halford. Each camshaft was driven via a vertical shaft and bevel gears from the rear of the engine.

Each cylinder had one spark plug mounted on its outer side and another mounted on its inner side. The spark plugs were fired by two magnetos mounted to and driven from the gear reduction housing. An accessory drive case was mounted to the back of the engine. A shaft extending back from the propeller shaft powered the accessory drive case. Driven from the accessory case were the camshafts, supercharger, generator, oil and fuel pumps, and various accessories. The single-speed supercharger drew in air through a downdraft carburetor and compressed the air and fuel mixture with a centrifugal impeller. The air and fuel mixture exited the supercharger housing via upper and lower passageways in the crankcase. These passageways were located between the upper and lower cylinder banks, and each had six outlets. A T-shaped manifold that was attached to each induction passageway outlet delivered the air and fuel mixture to two cylinders, one on each bank.


A Dagger III with individual exhaust stacks and many components chromed and polished to perfection for display purposes. Note the “Napier Halford” placard on the upper camshaft housing. (Napier/NPHT/IMechE image)

For engine cooling, air was ducted between the upper and lower cylinders. Baffles directed the air’s flow through the cylinders’ integral cooling fins and to the outer side of the cylinder banks. The cooling air exited via a cowl flap on each side of the aircraft and behind the engine. Two engine mounting pads were incorporated into the crankcase on each side of the engine. Two integral pads on each side of the rear accessory case were used together to form a third engine mount.

The Napier Dagger I (E98) had a 3.8125 in (96.8 mm) bore and a 3.75 in (95.3 mm) stroke. Each cylinder displaced 42.8 cu in (.70 L), and the Dagger’s total displacement was 1,027 cu in (16.84 L). The engine had a maximum output of 705 hp (526 kW) at 4,000 rpm at 12,000 ft (3,658 m). At 3,500 rpm, the Dagger I had a normal output of 630 hp (470 kW) at 10,000 ft (3,048 m) and produced 570 hp (425 kW) at sea level. The engine was 80 in (2.03 m) long, 22.5 in (.57 m) wide, and 45.125 (1.15 m) tall. The Dagger I weighed 1,280 lb (581 kg).


Front view of a Dagger III illustrates the engine’s two 24-cylinder distributors mounted under the propeller shaft and the 300 ft (91 m) or so of ignition cables. Just visible between the upper cylinder banks is the T-shaped manifold delivering air to the first two cylinders. (Napier/NPHT/IMechE image)

As engine design was underway, a two-cylinder test engine representing a Dagger’s upper and lower cylinder pair was built and tested. A complete 24-cylinder engine followed and was first run around early 1933. The Dagger I was installed in a two-seat light bomber biplane Hawker Hart (K2434) to serve as a testbed for the engine. The Dagger-powered Hart made its first flight on 17 December 1933. The engine experienced vibration and reliability issues and was later replaced with a Dagger II.

Napier continued to develop the Dagger engine line. Dagger E104 was a test engine with its bore enlarged to 4 in (102 mm). This increased the engine’s displacement by 104 cu in (1.70 L) to 1,131 cu in (18.53 L). It appears the E104 was built up using components from a Dagger I, but the engine never entered production.

The Dagger II was a refined Dagger I with additional supercharging for higher altitudes. The engine had a maximum rating of 760 hp (567 kW) at 4,000 rpm at 12,250 ft (3,734 m) with 1.5 psi (.10 bar) of boost, a normal rating of 695 hp (518 kW) at 3,500 rpm at 10,000 ft (3,048 m) with 1.5 psi (.10 bar) of boost, and a takeoff rating of 710 hp (529 kW) at 3,500 rpm with 3.0 psi (.21 bar) of boost. Fuel consumption at cruise power was .420 lb/hp/hr (255 g/kW/h). The Dagger II weighed 1,305 lb (592 kg). The engine was first run around early 1934 and passed a 100-hour type test on 18 June 1934. The Dagger II made its first flight in Hawker Hart K2434 in January 1935. Like the Dagger I, the Dagger II needed further work before the engine could enter production.


The Dagger VIII incorporated many changes from the previous Dagger engines and was capable of 1,000 hp (746 kw). Note the propeller shaft’s position has been raised above the engine’s centerline. (Napier/NPHT/IMechE image)

The Dagger III (E105) was a moderately supercharged version of the Dagger II. The engine had a maximum output of 805 hp (600 kW) at 4,000 rpm at 5,000 ft (1,524 m) with 2.25 psi (.15 bar) of boost, a normal output of 725 hp (541 kW) at 3,500 rpm at 3,500 ft (1,067 m) with 2.25 psi (.15 bar) of boost, and a takeoff output of 755 hp (563 kW) at 3,500 rpm with 3.5 psi (.24 bar) of boost. Fuel consumption at cruise power was approximately .448 lb/hp/hr (273 g/kW/h). Hawker Hart K2434 again served as a testbed and first flew with the Dagger III around September 1935. The improved engine was found to be reliable and was selected for the Hawker Hector, a two-seat liaison biplane. Hart K2434 was used to develop the engine cowling and installation for the Hector, and the Dagger III entered production in 1936. The Hector was first flown on 14 February 1936, and 179 examples were built. By June 1937, the Dagger III had completed a 100-hour test run at 4,000 rpm. Its initial output was record as 850 hp (634 kW). The engine was also selected for the Martin-Baker MB2 monoplane fighter, which made its first flight on 3 August 1938, but only the prototype was built. The Hector served in World War II, but the aircraft required extra maintenance due to its tight cowling and problematic Dagger III engine and was never a favorite of ground crews.


Rear view of a Dagger VIII highlighting the engine’s supercharger housing that conceals a two-sided impeller. The updraft carburetor can be seen on the right side of the engine. (Napier/NPHT/IMechE image)

In 1937, Dagger E108 incorporated several major changes. The engine had a double-entry, two-sided supercharger impeller for increased boost and incorporated an updraft carburetor. The propeller gear reduction housing was redesigned to accommodate a controllable-pitch propeller and moved up approximately 3.5 in (90 mm) above the engine’s centerline. The raised propeller shaft enabled the use of a larger diameter propeller. The relocation of the propeller shaft and redesign of the gear reduction housing resulted in the accessory drive shaft being powered by the left crankshaft, and the right crankshaft drove the magnetos and distributors mounted to the nose case. The propeller gear reduction was lowered to .308. New cylinders were designed with finer and more numerous cooling fins. Cylinder compression ratio was decreased slightly to 7.5 to 1. A single mounting pad on each side of the accessory case replaced the two pads previously used. Dagger E108 produced 935 hp (697 kW) at 4,100 rpm at 9,750 ft (2,972 m), and the engine was developed further as the Dagger VIII.

For the Dagger VIII, Napier developed a nose cowling with air ducts between the upper and lower cylinders. This was done in an attempt to make sure that the engine, once installed in an aircraft, was properly cooled. The Dagger VIII (E110) was first run in 1938 and had a maximum output of 1,000 hp (746 kw) at 4,200 rpm at 8,750 ft (2,667 m) with 5.0 psi (3.4 bar) of boost. The engine was rated at 925 hp (690 kW) at 4,000 rpm at 9,000 ft (2,743 m) with 4.0 psi (.28 bar) of boost and 955 hp (712 kW) for takeoff at 4,200 rpm with 6.0 psi (.41 bar) of boost. Its cruising output was 830 hp (619 kW) at 3,600 rpm at 7,000 ft (2,134 m) with 3.5 psi (.24 bar) of boost. Fuel consumption at cruise power was .461 lb/hp/hr (280 g/kW/h). The Dagger VIII was 73.9 in (1.88 m) long, 26.8 in (.62 m) wide, and 47.8 in (1.21 m) tall. The engine weighed 1,390 lb (630 kg).


A Hawker Hector with its Dagger III was the most successful application of the engine in an airframe. However, maintenance crews did not like the engine or its tight cowling.

In March 1937, the Dagger VIII was selected for what would become the Handley Page HP.53 Hereford I, a twin-engine medium bomber monoplane. The Hereford was simply a Dagger-powered HP.52 Hampden, and 100 examples were ordered in August 1937. The selection of the Dagger engine was more out of necessity than desirability. With all the other orders coming in during the scramble to rearm in the late 1930s, an alternative powerplant was desired to substitute for the standard Bristol Pegasus engines in the Hampden. The Hereford prototype (L7271) made its first on 8 October 1938. Cooling issues were encountered during flight trials, and the cowlings were modified and redesigned several times. The first production Hereford I (L6002) first flew on 17 May 1939. Persistent issues with the Dagger engines resulted in most of the 100 Herefords ordered being finished with Pegasus engines, since Pegasus production was able to keep up with demand. The few Herefords that retained their Dagger engines were used mostly as trainers. The Dagger VIII was also installed in Fairey Battle K9240 for engine tests. The Dagger VIII-powered battle made its first flight in November 1938.

The last of the Dagger line was the E112. This was an enlarged Dagger with a 4.0625 in (103 mm) bore, a 3.9375 in (100 mm) stroke, and a total displacement of 1,225 cu in (20.07 L). The E112 engine design dated from around 1939 and may have been a development of E104. It does not appear that the E112 was ever built.


The first Handley Page HP.52 Hereford I production aircraft (L6002) with its Dagger VIII engines. The cowling was similar to that developed for the Rapier. Note the carburetor intake under the engine and the cooling air exit door on the side of the rear cowling.

Like the Rapier, cooling the Dagger engine was difficult while the aircraft was on the ground. Cylinder head temperatures would often reach their upper limit before oil temperatures reached their lower limit. The result was that an aircraft would take off with oil temperature too low. This affected the oil’s ability to flow and led to the failure of various internal engine components. The Dagger did not achieve a level of success that warranted the engine’s mass production. However, what production there was of the Rapier and Dagger was enough to keep Napier going. The British Air Ministry was somewhat sympathetic to the powerful, compact, high-revving, small-frontal-area aircraft engine concept and continued to support Napier and Halford. By 1939, Napier was fully focused on developing the 2,000 hp (1,491 kW) Sabre engine for the war in Europe. While the air-cooled Dagger H-24 may have contributed to the knowledgebase upon which the liquid-cooled Sabre H-24 was built, the engines were very different. A Dagger II is preserved and in storage as part of the Smithsonian National Air and Space Museum. One Dagger VIII is on display at the Royal Air Force Museum in London, England and another is part of the Science Museum’s collection at Wroughton, England.


A Dagger VIII engine preserved and on display at the Royal Air Force Museum in London, England. Note the baffles on the cylinders to direct the flow of cooling air through the fins. (Nimbus227 image via Wikimedia Commons)

Aero Engines Vol. II by Various Authors (1939)
British Piston Aero-Engines and Their Aircraft by Alec Lumsden (2003)
By Precision Into Power by Alan Vessey (2007)
Boxkite to Jet — the remarkable career of Frank B Halford by Douglas R Taylor (1999)
Aircraft Engines Volume Two by A. W. Judge (1947)
Jane’s All the World’s Aircraft 1935 by C. G. Grey (1935)
Jane’s All the World’s Aircraft 1939 by C. G. Grey (1939)
Aerosphere 1939 by Glenn D. Angle (1940)
Aircraft Engines of the World 1941 by Paul H. Wilkinson (1941)
An Account of Partnership – Industry, Government and the Aero Engine by George Bulman and edited by Mike Neale (2002)
Hawker Aircraft since 1920 by Francis K. Mason (1991)
Fairey Aircraft since 1918 by H. A. Taylor (1974/1988)
Handley Page Aircraft since 1907 by C. H. Barnes (1976)
– “The Napier-Halford Daggers” Flight (11 July 1935)
– “Accent on the Aspirate” Flight (10 June 1937)
– “The Napier Dagger VIII” Flight (12 January 1939)


Napier H-16 Rapier Aircraft Engine

By William Pearce

Frank Bernard Halford had been an aircraft engine designer since World War I. In 1923, he established himself as a for-hire consultant to design aircraft engines for established manufacturers. By 1927, Halford had designed a new high-revving aircraft engine with numerous small cylinders and minimal frontal area. Halford’s belief was that a smaller engine running at a faster speed would produce the same power as a larger engine running at a slower speed. The new engine design was a vertical H with four cylinder banks, each with four individual cylinders.


The Napier Rapier I with its intake and exhaust ports mounted on opposite sides of the cylinder, Note the magnetos mounted to the rear of the engine and the external oil line on the crankcase.

Halford showed the design to George Purvis Bulman, the Chief Inspector (of engines) for the British Ministry of Munitions. Bulman was impressed with the design and knew that the British engineering firm D. Napier & Son (Napier) was in search of a new product. Napier’s Lion W-12 aircraft engine was designed 10 years previous, and the company had stopped producing automobiles in 1924. Napier wanted to pursue the development of new aircraft engines but felt that its current in-house design department did not have the needed experience.

Bulman introduced Halford to George Pate, Napier’s Production Chief Engineer. With the blessing of Napier’s board of directors and its chairman, Montague Stanley Napier, Halford was contracted in 1928 to design aircraft engines for Napier. One stipulation was that the engines must fall between a displacement of 404.09 and 718.37 cu in (6.62 and 11.77 L) to not conflict with any of Halford’s projects with other companies. Halford immediately began detailed design work on the H-16 engine, which would eventually be known as the Rapier. The engine is often referred to as the Napier-Halford Rapier.


Rear and front views of the Rapier I. On the left, the upper “Y” intake pipe can be seen behind the spark plug wires. On the right, the intake manifolds can be seen atop the inner side of the cylinder banks, just under the valve rocker housings.

Much of Halford’s previous aircraft engine experience was with air-cooled cylinders, and the 16-cylinder Rapier was no different. An Air-cooled engine was lighter and less complex than a liquid-cooled engine. The Rapier had a two-piece aluminum crankcase that was split horizontally at its center. The left and right crankshafts were supported between the two crankcase halves via five main bearings each. Each one-piece, four-throw crankshaft served one vertical and one inverted bank of cylinders. The crankshafts were phased at 180 degrees (some sources say 90 degrees, and it may be that the Rapier I was so phased and that later engines were at 180 degrees). Power strokes occurred simultaneously for both crankshafts. The connecting rod attached to each crankpin was a master rod with an articulating rod mounted to its end cap. When viewed from the rear, master rods served the upper left and lower right cylinder banks. Spur gears at the front of each crankshaft meshed with a larger gear that was mounted to the propeller shaft, which turned at .390 crankshaft speed. When viewed from the rear, both crankshafts rotated clockwise, and the propeller shaft rotated counterclockwise.

The air-cooled cylinders were made of aluminum heads that were screwed and shrunk onto forged steel barrels. Each cylinder was mounted to the crankcase via four studs. The cylinders had a 6.0 to 1 compression ratio, and each cylinder had a single intake and a single exhaust valve. The intake port was on the inner side of the cylinder, and the exhaust port was on the outer side. The valves for each set of eight upper and lower cylinders were actuated by a single camshaft via pushrods and rockers. Each camshaft was located between its respective set of cylinders (upper and lower). Each cylinder had one spark plug mounted on its outer side and another mounted on its inner side.


The Havilland DH.77 prototype fighter monoplane was initially powered by a Rapier I engine, but a Rapier II was later installed. Note the individual exhaust stacks and the machine gun installed on the side of the aircraft.

An accessory drive case was mounted to the back of the engine. A shaft extending back from the propeller shaft powered the accessory drive gears. Driven from the accessory case were the camshafts, magnetos, supercharger, generator, and various accessories. The engine’s two magnetos were mounted to the rear of the accessory case, and each magneto fired one of the cylinder’s two spark plugs. The single-speed supercharger drew in air through an updraft carburetor and compressed the air and fuel mixture with a centrifugal impeller. The air and fuel mixture exited the top and bottom of the supercharger housing into a Y pipe that distributed the charge to each cylinder via a manifold that ran along the inner side of each cylinder bank. A hand crank or an air starter was used to start the engine.

Napier developed a cowling for the Rapier so that the engine could be installed as a complete package. The cowling was narrow in form and had large upper and lower scoops. For engine cooling, air was ducted between the upper and lower cylinders. Baffles directed the air’s flow through the cylinders’ integral cooling fins and to the outer side of the cylinder banks. The cooling air exited via a cowl flap on each side of the aircraft and behind the engine.


The Rapier II had a revised cylinder with intake and exhaust ports on its outer sides. The supercharger housing was also modified with four outlets serving individual intake manifolds for each cylinder bank. Note the crankcase’s horizontal parting line.

The Napier Rapier I was designated by Napier as the E93. The engine had a 3.5 in (88.9 mm) bore and a 3.5 in (88.9 mm) stroke. Each cylinder displaced 33.7 cu in (.55 L), and the Rapier’s total displacement was 539 cu in (8.83 L). At sea level, the engine had a maximum output of 350 hp (261 kW) at 3,900 rpm and a normal output of 300 hp (224 kW) at 3,500 rpm. The Rapier I was 54 in (1.37 m) long, 21 in (.53 m) wide, and 35 (.90 m) tall. The engine weighed 620 lb (281 kg).

The Rapier I was first run around the start of 1929 and was mainly a developmental engine. The engine was installed in the de Havilland DH.77 (J9771) prototype fighter monoplane, which made its first flight on 11 July 1929. Although the aircraft exhibited good qualities, it was not selected for production. After completing its evaluation, the DH.77 was used to accumulate 100 hours of engine tests until December 1932. A Rapier II engine (see below) was then installed with a modified cowling. Engine development continued until the summer of 1934, when the aircraft was scrapped. The Rapier I was also installed in a Bristol Bulldog TM (K3183) biplane trainer around 1933. The aircraft served as the Rapier I test bed to evaluate the engine and cowling in a wind tunnel and in flight. Bulldog TM (K3183) kept its Rapier powerplant until 1938, when it was used to test another engine.


The Rapier IV was very similar to the Rapier II but with decreased supercharging. The baffles helped direct cooling air through the cylinder’s fins. Note the magneto mounted vertically from the accessory case.

The Rapier II was a development of the Rapier I with the supercharger’s impeller geared at a higher speed to improve the engine’s performance at altitude. New cylinders were used that had the intake and exhaust ports both located on the outer side of the cylinder. The induction system was revised with four outlets from the supercharger that distributed the air and fuel mixture via separate manifolds to each cylinder bank. The accessory case was also updated with the magnetos mounted vertically.

The Rapier II carried the Napier designation E95 and was first run in 1932. At 10,000 ft (3,048 m), the Rapier II had a maximum output of 355 hp (265 kW) at 3,900 rpm and a normal output of 305 hp (227 kW) at 3,500 rpm. The engine was 55.25 in (1.40 m) long, 20.75 in (.53 m) wide, and 35.25 (.90 m) tall. The engine weighed 710 lb (322 kg). As mentioned above, the engine was installed in the DH.77 prototype, which flew in this configuration in early 1933.


The Rapier VI had a revised, magnesium crankcase, a separate gear reduction housing, and used a downdraft carburetor. Otherwise, its structure was similar to that of the Rapier IV.

The Rapier IV was similar to the Rapier II, but it generated maximum power at low altitude due to revised supercharger gearing. At sea level, the Rapier IV had a maximum output of 385 hp (287 kW) at 3,900 rpm and a normal output of 340 hp (254 kW) at 3,500 rpm. The Rapier IV was 52.0 in (1.32 m) long, 21 in (.53 m) wide, and 37.7 in (0.96 m) tall. The engine weighed 726 lb (329 kg). The Rapier IV was first run in 1933, and Napier purchased an Airspeed Courier AS.5C (G-ACNZ) touring aircraft to serve as an engine testbed that same year. The AS.5C with its Rapier IV engine was first flown in June 1934. The aircraft was used as a demonstrator for a few years. By 1937, the engine had been replaced, and the aircraft was sold. Prior to AS.5C’s delivery, two Rapier IV engines were installed in a Saro A.19/1A Cloud (G-ABCJ) amphibious transport. The A.19/1A was the first testbed for the Rapier IV. The aircraft was loaned to Jersey Airways in August 1935 and withdrawn from service in December 1936.

The Rapier V was a further development of the Rapier line. Changes consisted of a magnesium crankcase, a separate updated gear reduction housing, fork-and-blade connecting rods, and an increased compression ratio of 7.0 to 1. The forked rods were in the rear lower cylinders, second from rear upper cylinders, second from front lower cylinders, and front upper cylinders. The induction system was revised to accommodate a downdraft carburetor. The engine was given the Napier designation E100 and was first run in around 1934. At 10,000 ft (3,048 m), the Rapier V had a maximum output of 380 hp (283 kW) at 4,000 rpm and a normal output of 340 hp (254 kW) at 3,650 rpm. Fuel consumption at cruise power was approximately .429 lb/hp/hr (261 g/kW/h) at 240 hp (179 kW) and 3,300 rpm. The Rapier V was 57.37 in (1.46 m) long, 23.37 in (.59 m) wide, and 36.0 in (.91 m) tall. The engine weighed 720 lb (326 kg). Four of the engines were installed in the Short S.20 Mercury (G-ADHJ) seaplane, which first flew on 5 September 1937. These engines were replaced with Rapier VIs in June 1938.


Front and rear views of the Rapier VI. Internally, the engine used fork-and-blade connecting rods and had a cylinder compression ratio of 7.0 to 1. It was the most powerful of the Rapier engines.

The Rapier VI (possibly E106) was similar to the Rapier V, but with decreased supercharging. The Rapier VI had a maximum rating of 395 hp (295 kW) at 4,000 rpm at 6,000 ft (1,829 m); a normal rating of 370 hp (276 kW) at 3,650 rpm at 4,750 ft (1,448 m); and a takeoff rating of 365 hp (272 kW) at 3,500 rpm at sea level. Fuel consumption at cruise power was approximately .412 lb/hp/hr (251 g/kW/h) at 310 hp (231 kW) and 3,500 rpm. The engine was 56.6 in (1.44 m) long, 22.4 in (.57 m) wide, and 36.0 in (.91 m) tall. The Rapier IV weighed 713 lb (313 kg). The engine was first installed in the Fairey Seafox reconnaissance float plane, which made its first flight on 27 May 1936. Early issues were experienced with engine cooling, but ultimately 66 Seafoxes were built, making it the most successful Rapier application. The Seafox was withdrawn from service in 1943. The Rapier IV was also installed in the Blackburn H.S.T.10 transport, the development of which was halted in 1936, before the aircraft was completed.


The Fairey Seafox reconnaissance float plane was powered by the Rapier VI engine, and 66 examples of the aircraft were built.

As previously mentioned, four Rapier VI engines were installed in the Short S.20 Mercury in June 1938. When the S.20 was mounted atop the Short S.21 Maia, the pair formed the Short-Mayo Composite, which was envisioned to provide long-range transport service. After being hoisted aloft by the Short S.21 Maia on 21 July 1938, the S.20 separated and later completed the first commercial, non-stop East-to-West transatlantic flight by a heavier-than-air machine. The Maia-Mercury composite went on to establish a seaplane distance record, covering 6,045 miles (9,728 km) between 6 and 8 October 1938. The Mercury and Maia made several flights until commercial operations were suspended due to World War II.

Cooling the Rapier engine was particularly difficult while the aircraft was on the ground. The uncuffed propellers did not provide sufficient airflow to effectively cool the engine, especially the rear cylinders. This issue was never fully resolved. In the early 1930s, Napier and Halford were working on the development of other aircraft engines, which would ultimately lead to the air-cooled Dagger H-24 and liquid-cooled Sabre H-24. By mid-1935, resources at Napier were wearing thin, and the decision was made to discontinue Rapier development so that efforts could be concentrated on other projects. Rapier production continued until around 1937. One Rapier VI engine was preserved and is on display at the Shuttleworth Collection in Bedfordshire, England.


The Short S.20 Mercury (top) and Short S.21 Maia (bottom) seaplane composite. Although originally fitted with four Rapier V engines, the Mercury had Rapier VIs installed for its service flights. The Maia was powered by four nine-cylinder Bristol Pegasus radial engines.

– “The Napier Rapier” Flight (14 March 1935)
British Piston Aero-Engines and their Aircraft by Alec Lumsden (2003)
By Precision Into Power by Alan Vessey (2007)
Boxkite to Jet — the remarkable career of Frank B Halford by Douglas R Taylor (1999)
Aircraft Engines Volume Two by A. W. Judge (1947)
Jane’s All the World’s Aircraft 1931 by C. G. Grey (1931)
Jane’s All the World’s Aircraft 1934 by C. G. Grey (1934)
Jane’s All the World’s Aircraft 1936 by C. G. Grey (1936)
Aerosphere 1939 by Glenn D. Angle (1940)
An Account of Partnership – Industry, Government and the Aero Engine by George Bulman and edited by Mike Neale (2002)
Aircraft Engines of the World 1941 by Paul H. Wilkinson (1941)
Bristol Aircraft since 1910 by C. H. Barnes (1964/1994)
De Havilland Aircraft since 1909 by A. J. Jackson (1987)
Airspeed Aircraft since 1931 by H. A. Taylor (1970)
Saunders and Saro Aircraft since 1917 by Peter Jackson (1988)
Shorts Aircraft since 1900 by C. H. Barnes (1989)
Fairey Aircraft since 1918 by H. A. Taylor (1974/1988)
Blackburn Aircraft since 1909 by A. J. Jackson (1968/1989)


Piaggio P.7 / Piaggio-Pegna Pc 7 Schneider Racer

By William Pearce

Giovanni Pegna was an Italian aeronautical engineer who started to design racing seaplanes and other aircraft in the early 1920s. Partnering with Count Giovanni Bonmartini, the pair formed Pegna-Bonmartini in 1922 to bring some of Pegna’s aircraft designs to life. Pegna was particularly interested in designing a racing seaplane for the Schneider Trophy Contest. Pegna-Bonmartini was short lived, as it was bought out by Piaggio & C. SpA (Piaggio) in 1923, when the latter company decided to start designing its own aircraft. Pegna was appointed head aircraft designer for Piaggio.


Giovanni Pegna’s previous racing seaplane designs. The engine and propeller of the Pc 1 pivoted up to clear the water for takeoff, landing, and while operating on the water’s surface. The Pc 2 and Pc 3 were fairly conventional designs but were advanced for their 1923 time period. The Pc 4 had tandem engines in a push/pull configuration and a single, central float. Wing floats would have been incorporated into the design. The Pc 5 and Pc 6 both used a retractable hull that was extended for takeoff and landing. The Pc 6 also had tandem engines in a push/pull configuration.

Pegna’s racing seaplane designs focused on minimizing the aircraft’s frontal area. Some of the designs used floats, while others incorporated a flying boat hull. Construction of the Pc 3 was started by Piaggio in 1923. The “Pc” in the aircraft’s designation stood for Pegna Corsa (Race), and this aircraft most likely carried the Piaggio designation P.5. The Pc 3 was a fairly conventional, single-engine monoplane utilizing two floats, but the aircraft was never finished.


The Schneider Trophy Contest inspired a number of extraordinary designs, but the Piaggio P.7 / Pegna-Piaggio Pc.7 was the most radical to be built. Its hydrovanes were much smaller and lighter than floats, offering the aircraft a distinct advantage if it could get airborne. Note the water rudder behind the water propeller.

In 1927, Pegna was asked by the Ministero dell’Aeronautica (Italian Air Ministry) to design a racing seaplane for the 1929 Schneider Trophy Contest. After studying three designs (Pc 4 through Pc 6), Pegna became increasingly focused on utilizing a central float that would be extended to support the aircraft on water and retracted while the aircraft was in the air. However, the complexity and estimated weight of the float and its retraction mechanism, combined with the unknown aerodynamic forces during retraction and extension, made the design impractical. Pegna returned to the drawing board and, aided by Giuseppe Gabrielli, designed the Pc 7, which was also known as the Piaggio P.7. On 24 March 1928, the Italian Air Ministry ordered two examples of the P.7 and assigned them serial numbers (Matricola Militare) MM126 and MM127.

After experiments in a water tank, Pegna finalized the aircraft’s design. The Piaggio P.7 (Piaggio-Pegna Pc 7) had a watertight fuselage that sat in the water up to the shoulder-mounted wings when the aircraft was at rest. A two-blade propeller at the front of the aircraft was just above the waterline. The engine was located just forward of the wing and drove the propeller via a shaft. A second shaft extended behind the engine to a water propeller positioned in a skeg under the tail. Clutches on both shafts allowed the front propeller or the water propeller to be decoupled from the engine. When the front propeller was decoupled, it would come to rest in a horizontal position. For takeoff, the engine would power the water propeller with the front propeller stationary. As the aircraft gained speed, the front would rise about 10 degrees out of the water by the hydrodynamic forces imparted on two hydrovanes extending below the fuselage and by a third hydrovane located in front of the water propeller. With the front propeller clear of the water, engine power was diverted from the water propeller to the front air propeller. The front propeller would continue the aircraft’s acceleration until enough speed was gained to lift off from the water’s surface.


A view of the P.7’s internal layout. A and B are the drive shaft clutches. C is the lever that engages and disengages the air propeller; when disengaged, it locks the propeller in a horizontal position and closes the main carburetor inlets. D is the lever that engages and disengages the water propeller; when disengaged, it feathers the water propeller. E is not recorded, but it appears to be a bulkhead and support for the propeller shaft. F is a rubber diaphragm operated by the air propeller lever that seals the propeller shaft when the air propeller was disengaged.

The P.7’s airframe was made mostly of wood with some metal components. The aircraft was skinned with two layers of plywood with a waterproof fabric sandwiched between the layers. Two watertight compartments were sealed into the fuselage, and the vertical and horizontal stabilizers were watertight. A single fuel tank was positioned in the fuselage under the wing and between the engine and cockpit. The one-piece wing had three main spars and was mounted atop the fuselage. Two legs extended below the fuselage, and each supported a planing surface. The planing surfaces, including the one on the tail, were inclined approximately three degrees compared to the aircraft while in level flight. The relative angle would increase as the aircraft was landed with a slight tail-down configuration. A water rudder extended below the fuselage directly under the aircraft’s tail. The movement of the water rudder and normal rudder were linked.


The nearly complete P.7 without its engine or hydrovanes. The original carburetor inlets are visible on the side of the aircraft. Note the pipes for the surface radiators on the wings.

Originally, the P.7 was to be powered by a 1,000 hp (746 kW) FIAT AS.5 V-12 engine. For reasons that have not been found, the engine was switched to an Isotta Fraschini Asso 500 V-12 that produced 800 hp (597 kW) at 2,600 rpm. Isotta Fraschini fully supported the P.7 project, and Giustino Cattaneo, the Asso 500’s designer, redesigned the engine with a rear drive for the water propeller. In addition, new cylinder heads were designed with the exhaust ports on the inner, Vee side of the engine. As originally designed, the Asso 500 had intake and exhaust ports on the outer sides of the engine. Having the open exhaust ports on the side of the fuselage would lead to water intrusion when the aircraft was at rest on the surface. Relocating the exhaust ports to vent out the top of the fuselage resolved this issue. The cylinder heads were most likely the same or very similar to those that Cattaneo had designed for the Savoia-Marchetti S.65 Schneider racer. Cattaneo and Isotta Fraschini also designed at least some of the P.7’s drive systems. Surface radiators on the wings cooled the engine’s water coolant, and engine oil was cooled by a surface radiator on the sides and bottom of the aircraft’s nose.

The cockpit was situated low in the aircraft’s fuselage and between the wing’s trailing edge and the tail. Two levers on the left side of the cockpit controlled the engine’s output to the air and water propellers. One lever engaged and disengaged the air propeller. When engaged, the main carburetor inlets at the front of the aircraft were automatically opened. When disengaged, the carburetor inlets were closed, a rubber seal was pressed against the front of the propeller shaft, and the propeller was slowed and subsequently locked in a horizontal position. The carburetor inlets were originally located on the sides of the aircraft by the engine but were moved to above the nose. When the carburetor inlets were closed, the engine drew in air from the cockpit. When the water propeller’s lever was disengaged, the blades were feathered to offer as little aerodynamic resistance as possible.


The completed P.7 supported by a hoist illustrates the aircraft’s sleek design. The pilot sat quite far aft, and landings would have been a challenge.

Six air propellers were ordered for testing on the P.7. They varied in diameter and profile. Three were made from steel with a ground-adjustable pitch, and the other three were made from duralumin, and each had a different fixed pitch. One of the steel air propellers was designed by Pegna. Originally, the adjustable-pitch water propeller was made from duralumin components, but testing resulted in a switch to a steel hub with duralumin blades. The Piaggio P.7 had a 22 ft 2 in (6.76 m) wingspan, was 29 ft 1 in (8.86 m) long, and was 8 ft (2.45 m) tall. It had a maximum speed of 373 mph (600 km/h) and a landing speed of 103 mph (165 km/h). The aircraft weighed 3,122 lb (1,416 kg) empty and 3,717 lb (1,686 kg) fully loaded.

The design of the complex and unique aircraft delayed its completion. It appears that the first aircraft, MM126, was completed and sent to Desenzano before the Schneider Trophy Contest was held in September 1929, but there was not enough time to test the P.7 before the race. Both P.7 aircraft were finished by late October 1929, which is when testing began. Most pilots of the Italian Reparto Alta Velocità (High Speed Unit) were not interested in testing the radical machine. However, Tommaso Dal Molin was up to the task. Testing occurred on Lake Garda, just off from Desenzano, home of the Reparto Alta Velocità.


The P.7 on Lake Garda for tests. A simple structure connected to hardpoints above the wing was used to raise and lower the aircraft out of the water. More so that most Schneider Trophy racers, the P.7 could only be operated on calm waters.

Using the water propeller, Dal Molin in MM126 was able to raise the nose of the aircraft to a sufficient height to engage the air propeller, but this was not done. The P.7 was unstable planing on the water, and issues were experienced with the clutch for the water propeller. Oil on the clutch caused it to slip, resulting in a loss of power to the water propeller. In addition, the sudden cavitation of the main hydrovanes while planing caused a loss of buoyancy, which resulted in the P.7 suddenly and violently settling back on the water’s surface. Because of the issues, it seems that tests were conducted over only a few days.

There was no cover to easily access the clutch. The needed repairs would require substantial disassembly of the aircraft. By this time, the Air Ministry and Piaggio showed little interested in the P.7, but Pegna wanted to continue its development. Some of the changes Pegna had in mind were adjustable hydrovanes and cooling the engine oil with water rather than using a surface radiator. However, it appears that the repairs were never made. MM126 was stored at Desenzano for a time but was destroyed after a few years. MM127 was taken to Guidonia Montecelio, near Rome, for testing in a water tank to improve the aircraft’s hydrovanes. The aircraft was eventually abandoned, and it is not clear if any tests were ever conducted. MM127, along with other aircraft, was destroyed in 1944—a casualty of World War II.


The P.7 surrounded by contemporaries at Desenzano. At left is the Macchi M.39. At right is the Savoia-Marchetti S.65. The Macchi M.52’s wing is in the foreground. Note the P.7’s exhaust stacks protruding above the engine.

Some Ideas on Racing Seaplanes (Technical Memorandums National Advisory Committee for Aeronautics No. 691) by Giovanni Pegna (November 1932) 31.4 MB
Schneider Trophy Seaplanes and Flying Boats by Ralph Pegram (2012)
MC 72 & Coppa Schneider Vol. 2 by Igino Coggi (1984)
Schneider Trophy Aircraft 1913–1931 by Derek N. James (1981)
Volare Avanti by Paolo Gavazzi (2000)
Jane’s All the World’s Aircraft 1932 by C. G. Grey (1932)


Cobb Railton Land Speed Record Car

By William Pearce

John Rhodes Cobb was a fur trader who turned to auto racing and setting endurance records in his Napier-Railton car. The Napier-Railton was designed by Reid Antony Railton, head engineer at Thomson & Taylor. Run by Ken Thomson and Ken Taylor, the company was located at the Brooklands raceway in Surrey, England and specialized in designing and building race cars.


John Cobb and the Railton streak across the Bonneville Salt Flats in 1947. The car was the first to go over 400 mph (644 km/h).

Around October 1935, Cobb approached Railton and Taylor about designing a Land Speed Record (LSR) car. At the time, a new record had just been set on 3 September 1935 by Malcolm Campbell. For the record, Campbell ran his Campbell-Railton-Rolls-Royce Blue Bird car at 301.129 mph (484.620 km/h) on the Bonneville Salt Flats in Utah. After the record, Campbell retired from attempting any further LSRs. Railton had done much of the design work on Campbell’s car, and Cobb did not care much for Campbell. What Cobb offered Railton was the freedom to design a LSR car from scratch. All of Railton’s work with Campbell was redesigning and modifying a car that was originally built in 1926.

Cobb made slow, deliberate steps toward his goals, and his work on the LSR car would be no different. It was not until early 1937 that Railton and Ralph Beauchamp began serious design work on the car. At the same time, Cobb’s friend and fellow record-breaker George Eyston began the construction of his own LSR car, Thunderbolt. Eyston’s huge car was powered by two Rolls-Royce R engines and needed eight wheels to distribute its immense weight. While similar in concept and designed to achieve the same goal, Railton’s LSR car design would stand in stark contrast to the Thunderbolt. Railton’s LSR design carried the Thomson & Taylor designation Project Q-5000. Cobb named the car Railton in honor of its designer.

While Cobb was financially well-off, he did not have unlimited funds for an LSR car. Railton wanted to design the car using existing technology and keep its proportions within the limits suitable for four wheels. Railton also felt that four-wheel drive was necessary. Having the front and rear wheels driven independently by their own engine circumvented many challenges and simplified the overall design. The choice to use two Napier Lion W-12 engines was an easy one. Railton had experience with the engine when he first worked on Campbell’s Blue Bird in 1930. The Lion was also selected to power Cobb’s Napier-Railton, and Thomson & Taylor had much experience with the engine type, as they converted them for marine use.


Rear view of the Railton shortly after its completion in 1938. Once the one-piece body was quickly removed, nearly all of the car’s components were accessible. The large water tank is on the left, and the air brake can be seen forward of the rear tires.

Originally designed in 1917, the Lion was a 12-cylinder aircraft engine with three banks of four cylinders. The center bank extended vertically from the crankcase, with the left and right banks angled at 60 degrees from the center bank. Two supercharged Racing Lion VIID engines were available for Cobb’s LSR car. Built in 1929, the engines had been used by Marion Barbara (Joe) Carstairs to power her Estelle IV motorboat. The Lion VIID was the same type of engine Campbell had used to power his Blue Bird in 1931 and 1932. The modified engines produced 1,480 hp (1,104 kW) at 3,600 rpm during tests, but would only produce 1,250 hp (932 kW) at Bonneville’s 4,200-ft (1,280-m) elevation. Carstairs gave both Lion VIID engines to Cobb. Incidentally, Carstairs had funded Campbell’s purchase of two Lion VIID engines in 1930 for his Blue Bird.

After the basic design of the car’s body was determined by wind tunnel tests, Railton focused on filling the body with the needed equipment. The Railton’s frame was a single central boxed girder made from high-strength steel and perforated with large lightening holes. The girder was 11 in (279 mm) wide and varied between 8 and 12 in (203 to 305 mm) tall. When viewed from above, the girder took the shape of a flattened S. Mounted above the front and rear of the girder were the front and rear axles. The cockpit was mounted in front of the front axle on cantilevered supports that extended from the girder. The central part of the girder was angled seven degrees across the car’s centerline. Staggered outriggers extended from each side of the girder to support a Lion engine. The engines were installed 10 degrees off the car’s centerline. The front engine was offset to the right and drove the rear wheels, and the rear engine was offset to the left and drove the front wheels.

Each engine drove a three-speed transmission without a conventional clutch or flywheel. Gear changes were made carefully and with the aid of an overrunning clutch device with locking dogs. Linkages were synchronized so that the single throttle pedal operated both engines, the single clutch pedal unlocked both clutches, and the single gearshift lever operated both transmissions. Each driveshaft also incorporated an 11 in (279 mm) drum brake with hydraulically actuated shoes contracting on its outer diameter. The drums were water-cooled, utilizing the same coolant as the engines. Just forward of the rear wheels was a pneumatic airbrake. Its operation could be linked to the brake pedal so that it deployed vertically as the brake was pressed.


Front view of the Railton on the Salt Flats in 1938. The open covers at the bottom of the car allowed access for two of the body’s eight mounts. Note that the air brake has been removed, as Cobb found the driveshaft brakes more than adequate.

The front axle featured a differential and independent wishbone suspension. The rear axle was narrower than the front and had a solid housing with no differential. The axles’ final drive ratio was 1.35. A combination coil spring and shock absorber controlled the suspension’s movement at each wheel. Forward of the left engine was a 90 US gal (75 Imp gal / 341 L) water tank for engine cooling. The tank was filled with ice, and delivered water to the engines. The Railton had no radiator, and the heated water was purged after passing through the engines. Behind the right engine was a 22 US gal (18 Imp gal / 82 L) fuel tank and an 18 US gal (15 Imp gal / 68 L) oil tank.

The Railton was entirely encased by its streamlined body. The body was designed to not create any lift. Wind tunnel experiments and calculations indicated that the nose of the car would need to be lifted 12 in (305 mm) before aerodynamic lift overcame the car’s weight. The maximum expected lift on the Bonneville Salt Flats was 3 in (76 mm). The one-piece upper body was made of aluminum panels welded and riveted to aluminum supports. The body weighed approximately 450 lb (204 kg) and was designed to be quickly removed to allow access to the entire vehicle for servicing. The 44 x 7.75 in (1,118 x 197 mm) Dunlop tires were mounted on 31 x 7 in (787 x 178 mm) steel wheels and were concealed beneath humps protruding above the body’s upper surface. A square opening covered the cockpit, which was sealed by an aluminum cover with a bulge and a small windscreen for the driver’s head. Two cockpit covers were built, one with an open top and one with a closed top. The open top version was discarded shortly after arriving at Bonneville.

The car’s body could be lowered in place over the seated driver, or the driver could enter the cockpit with the body in place via the opening. However, an overhanging structure to the cockpit opening was needed to support the driver if the body was in place. An undershield covered the underside of the chassis. The body was secured to the car’s frame at eight points and attached to the undershiled via approximately 36 Dzus fasteners. Exhaust from the upper cylinder bank of each engine exited via a manifold protruding above the body. Exhaust from each engine’s left and right cylinder banks exited via a manifold protruding from the underside of the car. The inboard exhaust passed though the girder frame. All exhaust manifolds were directed to the rear. The Railton was 28 ft long (8.53 m), 8 ft (2.44 m) wide, and 4 ft 3 in (1.30 m) tall. The car’s wheelbase was 13 ft 6 in (4.11 m). The front axle had a track of 5 ft 6 in (1.68 m) and the rear track was 3 ft 6 in (1.07 m). The Railton weighed 6,280 lb (2,849 kg).


The Railton being prepared at Bonneville in 1939. The fuel tank has been relocated to the car’s port side, and a large ice tank has been added at the back of the car. The man by the body is painting the Gilmore Red Lion on the nose of the car.

On 5 April 1938, the nearly-complete Railton was debuted for the press. The car was missing its wheel covers, but the craftsmanship involved in its construction and the vehicle’s purpose were evident. Attending the event was Eyston, who, in his Thunderbolt car, had established a new LSR of 311.42 mph (501.18 km/h) over the mile (1.6 km) and 312.20 mph (502.44 km/h) over the km (.6 mi) on 19 November 1937. The Railton was first displayed to the public on 18 April at Brooklands. There were no suitable places in Britain to test the car, so once it was completed, it was packed up and sent to the United States at the end of July.

When Cobb, his team, and the Railton arrived on the Bonneville Salt Flats, Eyston and Thunderbolt had been there for a few weeks. The weather had been bad, and Eyston had not been able to make any record attempts. The course was shortened to about 10 miles (16 km) because of the poor conditions. For starting, first gear was engaged, and the Railton was pushed by a truck to about 20 mph (32 km/h), at which point the magnetos were energized to start the engines. Cobb began testing the Railton, including a first shakedown run up to around 250 mph (402 km/h) without the car’s body. Initial test runs with the body resulted in deformations caused by air pressure pushing on specific areas at the rear of the body. Also, hot exhaust from the center cylinder banks damaged the top of the aluminum body. The body was straightened and reinforced, and an asbestos-lined steel shield was added behind the upper exhaust stacks. On 20 August 1938, conditions had improved, and Cobb took the Railton out for a serious test run. The peak speed was 300 mph (483 km/h) and the Railton averaged 270 mph (435 km/h) over the mile (1.6 km).

On 25 August 1938, the camera timing equipment failed to record Eyston in the Thunderbolt on what would have been a record-breaking run. The failure was caused by a lack of contrast between the car and the background. As a result, both Thunderbolt and Railton were partially painted black to improve contrast. On 27 August, Eyston in the Thunderbolt established a new LSR at 345.49 mph (556.01 km/h) for the mile (1.6 km) and 345.21 mph (555.56 km/h) for the km (.6 mi).


Cobb and the Railton making a run on the Salt Flats in 1939. The trip that year was quite successful, but the start of World War II overshadowed the records.

On 30 August 1938, Cobb made a record attempt. The Railton’s quick acceleration caused the tires to spin, subsequently damaging them, and the attempt was aborted. Even so, Cobb reached 325 mph (523 km/h). More work was done while the surface of the Salt Flats continued to improve. Cobb had found that the driveshaft friction brakes were sufficient to stop the car, and the airbrake was removed. A record attempt was made on 12 September, but issues with shifting the car resulted in a speed of 342.50 mph (551 km/h). With the knowledge and experienced gained by all the previous runs, another record attempt was made on 15 September. Cobb made his run north and covered the mile (1.6 km) at an average of 353.29 mph (568.57 km/h). The body was quickly removed, and the tires were changed during the turnaround. On the return south, the Railton averaged 347.16 mph (558.70 km/h). Cobb and the Railton were successful and set new records of 350.20 mph (356.59 km/h) over the mile (1.6 km) and 350.10 mph (563.43 km/h) over the km (.6 mi).

Eyston and his team had been modifying Thunderbolt for even more speed in case Cobb got the record. On 16 September 1938, one day after Cobb’s record run, Eyston and Thunderbolt made another attempt. The runs established a new LSR at an average of 357.50 mph (575.34 km/h) for the mile (1.6 km) and 357.34 mph (575.08 km/h) for the km (.6 mi).

Cobb and Railton knew their car was capable of more speed. They also learned a lot from its first outing and had a number of modifications in mind. The decision was made to not push the Railton for higher speeds, but to return to England, modify the car, and return to Bonneville in 1939, when conditions might be even better.


Cobb sits in the bodyless Railton in 1947. This image illustrates the tight fit under the body of the two Lion engines, various tanks, and other components. The twin belts, pulley, and shaft of the anti-stalling device can be seen between the cockpit and rear engine, which drove the front wheels.

Back in England, the Railton’s frame was modified to prevent its deflection by engine torque, and the suspension was upgraded. The cooling system was revised by incorporating a new 90 US gal (75 Imp gal / 341 L) tank for ice between and behind the car’s rear wheels. A new 22 US gal (18 Imp gal / 21 L) water tank with an additional header tank of about 6 US gal (5 Imp gal / 23 L) replaced the fuel tank on the right side of the car. The fuel tank was relocated to the left side of the car where the old water tank used to be. For the new cooling system, a thermostat controlled the flow of ice water from the ice tank to the water tank. Water from the water tank flowed to the engines. The total-loss system did not circulate water back to the tank, but vented the heated water out of the car. An opening was added at the front of the car that ducted air to the front engine. The engines’ supercharger gears were changed to increase impeller speed and provide additional boost. The Gilmore Oil Company of California was brought on as a major sponsor for the 1939 record attempt, and the car was often referred to as the Railton Red Lion for that year. Gilmore’s mascot/logo was a red lion, and the company had a line of Red Lion Gasoline.

Cobb, his team, and the Railton were back at the Bonneville Salt Flats in mid-August 1939. The salt was in good condition, and Cobb would have a course of about 13 miles (21 km) for the record attempt. On 17 August, a single run north was made at 352.94 mph (568.00 km/h). A tire tread had separated, and some adjustments to the car were needed. The baffling in the coolant header tank was subsequently modified, and the car was put back into good working order. On 22 August, an attempt was made, and speeds for the run north were recorded at 369.23 mph (594.22 km/h) for the mile (1.6 km) and 365.57 mph (588.33 km/h) for the km (.6 mi). On the return south, the left engine powering the front axle acted up, and the run was aborted. Adjustments were made to the carburetors, and another run was planned for the following day.

On 23 August 1939, the car was prepared, and Cobb set off in the early morning. The run north was covered at 370.75 mph (596.66 km/h) through the mile (1.6 km) and 367.92 mph (592.11 km/h) through the km (.6 mi). The car was back on the course in 25 minutes, after changing all four tires and adding fuel, oil, and water. On the run south, the Railton averaged 366.97 mph (590.85 km/h) over the mile (1.6 km) and 371.59 mph (598.02 km/h) over the km (.6 mi). The average of the runs were new LSRs at 368.86 mph (593.62 km/h) for the mile (1.6 km) and 369.74 mph (595.04 km/h) for the km (.6 mi). Cobb had exceeded six miles a minute, and a tachograph recording unit in the car indicated the peak speed was 380 mph (612 km/h).


While the body could be lifted by six men, many hands make light work. The oil tank is just forward of the rear wheel, followed by the relocated (in 1939) water and header tank. Many Dzus fasteners used to secure the body can be seen on the undershield. Note the very forward position of the driver

The Railton had performed so well that the decision was made to attempt longer distance records, and the car and the course were subsequently reconfigured. On 26 August 1939, Cobb and the Railton set new speed records covering 5 km (3.1 mi) at 326.66 mph (525.71 km/h), 5 miles (8.0 km) at 302.20 mph (486.34 km/h / timing equipment issues made this speed unofficial), 10 km (6.2 mi) at 283.01 mph (455.46 km/h), and 10 miles (16 mi) at 270.35 mph (435.09 km/h). Since the runs were made on the 13-mile (21-km) course, Cobb applied the brakes before exiting the longer, timed sections.

When the team had set off for the United States, Europe was in an unstable state and seemingly headed toward war. On 3 September 1939, as the team returned to England after their successful record runs, Britain declared war on Germany after the latter’s invasion of Poland on 1 September. Against such a backdrop, record setting became insignificant and irrelevant. During the war, the Railton was placed in storage, and Cobb served as a pilot with the Air Transport Auxiliary. But there was still some unfinished business, as Cobb knew the Railton was capable of more speed.

Toward the end of 1945, Cobb had the Railton removed from storage and sent to the Thomson & Taylor shop to be put in working order. Since the engines did not have a flywheel, they had a tendency to rev down and stall out during gear changes. Such an occurrence essentially brought a record run to an end. While the car was being worked on, Railton, who was now living in the United States, had a device fitted to both engines to prevent the stalls. The device was essentially a shaft that connected the engine to its drive line via a belt-driven overrunning clutch. If the engine speed dropped below one-seventh that of the drive line, the shaft turned by the drive line would keep the engine running. Other modifications were additional ducting to feed air from the opening at the front of the body to both engines and changing the final drive gears for high speed. New fuels allowed the engines to operate up to 4,000 rpm, and the pair produced a combined 3,300 hp (2,461 kW). The work on the Railton was performed under the ever-watchful eye of Ken Taylor. The Gilmore Oil Company, a major sponsor from 1939, had been bought out by the Socony-Vacuum Oil Company, which marketed its products under the “Mobil” name. The company agreed to sponsor Cobb’s efforts in 1947, and the car became the Railton Mobil Special.


A serious Cobb peers out the windscreen of the Railton. The slits forward of the canopy brought in air to the cockpit. A steel and asbestos panel behind the upper exhaust stacks protected the car’s body from heat damage.

The restored Railton was displayed before the press in late June 1947 and departed for Bonneville in July. The salt flats and the course were found to be in poor condition, and the Railton’s engines ran roughly. It took some time to resolve carburation issues and make the engines run right. One of the engines was later damaged during a test run. A camshaft was shipped from England to repair the Lion. When the engine issues had been resolved, the ice tank was punctured during a test run. After the tank was repaired, everything was finally in order for a test run on 14 September. The run north was timed at 375.32 mph (604.02 km/h). However, the rough course had caused the aluminum body to crack, necessitating yet more repairs.

On 16 September 1947, the wind had picked up considerably and the course was still less than ideal, but the car was ready. Cobb decided to make a record attempt. Setting off to the south, Cobb shifted into second gear at around 120 mph (193 km/h) and hit third at around 250 mph (402 km/h). The Railton shot through the measured mile (1.6 km) at 385.645 mph (620.635 km/h). The tires were changed and fluids refilled. On the run north, Cobb covered the mile (1.6 km) at 403.136 mph (648.785 km/h). The two-way average of the runs was a new LSR at 394.197 mph (634.399 km/h). And so it was that a 47-year-old man in a 10-year-old car with 20-year-old engines established a new LSR. It had taken quite a bit of effort to set the record in 1947, but Cobb and the team were confident the car could break 400 mph (644 km/h) on both runs if the course were a little better and the wind a little less. The Railton had left the measured mile (1.6 km) at about 410 mph (660 km) and was still accelerating. Plans were started to make another attempt the next day, but a serious rainstorm ended any hope for further runs.

LSRs were big news in the late 1920s and early 1930s. By 1947, and with no challengers on the horizon, Cobb breaking his own record was not nearly as sensational as previous LSRs. Cobb decided not to race the Railton again unless his record was broken. The LSR remained Cobb’s long after his tragic death on 29 September 1952, when his Crusader jet boat disintegrated during a water speed record attempt at over 206 mph (332 km/h). Cobb did make at least one demonstration of the Railton at Silverstone Circuit in England on 20 August 1949. In 1953, the Railton was sold by Cobb’s estate to the Dunlop Rubber Company, which donated it to the Museum of Science and Industry in Birmingham in July 1955. The car was displayed in the United States in 1954 (New York) and 1962 (San Francisco), and at the Brussels World’s Fair in 1958. In September 2001, the Railton was moved to the Thinktank, Birmingham Science Museum, where the car is currently on display.


The Railton on the wide expanses of the Salt Flats in 1947. The various exhaust manifolds can be seen above and below the body. Note the two streams of water pouring out the underside of the car from the total-loss cooling system.

Essentially, Cobb and the Railton held the LSR for 25 years*—from 1939 until Donald Campbell went 403.10 mph (648.73 km/h) in the turboshaft-powered Bluebird CN7 on 17 July 1964. Cobb’s record represented the end of an era, as later speed machines used jet engines to push them along. But, the LSR for the class of piston-powered, wheel-driven cars is still the goal for many racers. On 9 September 1960, Micky Thompson made one run at 406.60 mph (654.36 km/h) in the Challenger 1 before a failed transmission aborted his return. Bob Summers went 409.277 mph (658.667 km/h) in Goldenrod on 12 November 1965, a speed that was not bettered until 21 August 1991, when Al Teague averaged 409.986 mph in Spirit of ’76. Tom Burkland in the Burkland 411 Streamliner achieved 415.896 mph (669.319 km/h) on 26 September 2008. On 17 September 2012, George Poteet in Speed Demon averaged 439.024 mph (706.541 km/h) over the mile (1.6 km). In a car originally built by his father in 1968, Danny Thompson averaged 448.757 mph (722.204 km/h) in Challenger 2 on 12 August 2018. On 13 August 2020, Poteet in Speed Demon took back the record, averaging 470.016 mph (756.417 km/h) over the mile (1.6 km).

*Or 24 years if Craig Breedlove’s 407.447 mph (655.722 km/h) run in Spirit of America on 5 August 1963 is considered. At the time, the record for the three-wheel, jet-powered, non-wheel-driven Spirit of America was not officially recognized.

Note: Spirit of ’76 and Burkland 411 Streamliner both used supercharged engines, while Goldenrod was normally aspirated. Goldenrod’s speed record for a piston-powered, normally aspirated, wheel-driven car stood for 45 years until 21 September 2010, when Charles Nearburg in Spirit of Rett achieved 414.316 mph (666.777 km/h).


The Railton on display at the Thinktank, Birmingham Science Museum. Although fitting, the name “Dunlop” was never painted on the car while it was breaking records. (Geni image via Wikimedia Commons)

This article is part of an ongoing series detailing Absolute Land Speed Record Cars.

Reid Railton: Man of Speed by Karl Ludvigsen (2018)
Napier: The First to Wear the Green by David Venables (1998)
The Land Speed Record 1920-1929 by R. M. Clarke (2000)
The Land Speed Record 1930-1962 by R. M. Clarke (2000)
The Fast Set by Charles Jennings (2004)
The John Cobb Story by S. C. H. Davis (1953)
Napier: Lions at Large 1916 – 2016 by Alan F. Vessey (2016)


Eyston Thunderbolt Land Speed Record Car

By William Pearce

In 1935, Englishman George Edward Thomas Eyston traveled for the first time to the Bonneville Salt Flats in Utah, United States. At Bonneville on 3 September 1935, Eyston was able to witness Malcolm Campbell setting a Land Speed Record (LSR) in the last of his Blue Bird cars at 301.473 mph (485.174 km/h) covering one km (.6 mi) and 301.129 mph (484.620 km/h) covering one mile (1.6 km). In August, Eyston set a number of endurance records in his car, Speed of the Wind, which had been specially designed and built by Eyston and his partner, Ernest Arthur Douglas Eldridge. Eyston and Speed of the Wind set more records in 1936 and intended to return to Bonneville in 1937.


George Eyston in the Thunderbolt on the Bonneville Salt Flats in 1937. Note the short headrest fairing and the clearance bulges in the body panels above the wheels and engine.

In early 1937, Eyston and Eldridge began the design of an LSR car to break Campbell’s record and achieve a speed greater than 350 mph (565 km/h). Campbell’s last Blue Bird car was powered by a Rolls-Royce R engine and was supported by six wheels. It was quickly decided that the new LSR car would use two engines. After consulting with Dunlop, it was determined that the car’s weight necessitated the use of eight wheels. The car was primarily designed by Eyston. Eldridge contracted pneumonia returning from Bonneville after Speed of the Wind’s successful runs in 1936. Eldridge passed away on 27 October 1937, one day before the new LSR car was first run at Bonneville.

The new vehicle’s design progressed quickly, and Eyston and Eldridge were able to use their many contacts to obtain the parts needed for the car. However, major sponsors for their effort were in short supply. Eyston had become acquainted with many people at Rolls-Royce while he was building the Kestrel-powered Speed of the Wind. Eyston was able to acquire two Rolls-Royce R engines, the same type that Campbell had used to set several LSRs and had been used to set water and air speed records. Some sources state that Eyston wanted Merlin engines. However, since the Merlin was in an early production stage at the time and only produced around 1,030 hp (768 kW), this seems unlikely. Although heavier, the R engine had proven its reliability and produced twice the power of the then-current Merlin. With the power plant decided, serious work began on the new LSR car, which was later named Thunderbolt.


This top view of the Thunderbolt under construction illustrates the car’s layout. Note the track difference of the two front sets of wheels. The header water tank for each engine can be seen by the cockpit. The exhaust for the inner banks is at the center of the car. Only a single wheel is mounted on each side of the rear axle. The airbrakes are deployed and being skinned.

Thunderbolt was built to be strong and heavy. Two frame rails, 18 in (457 mm) tall at their maximum, ran the entire length of the car and supported all major components. A radiator was positioned at the front of the car and had an elongated, eight-sided opening. Behind the radiator was a splitter that directed air that had passed through the radiator either up above the car or below it. Both upper and lower air exits were positioned between the front-most wheels, which did not have brakes. Another set of front wheels with a slightly wider track were positioned behind the first set. This configuration allowed for a more streamlined nose and ensured the second set of front wheels would not ride in the ruts created by the first set of wheels. Drive shafts connected to the second set of wheels were equipped with hydraulically actuated disc brakes at their inner ends, which were supported by the main frame rails. This arrangement provided friction braking for the front of the car. All four front wheels were used to steer the Thunderbolt and featured independent suspension actuated by transverse leaf springs.

Behind the second set of front wheels was the cockpit. While the cockpit was open, the driver sat behind a windscreen. A headrest extended behind the cockpit and was faired into the car’s body. The main frame rails ran on both sides of the cockpit. The Rolls-Royce R engines were positioned behind the cockpit and outside of the main frame rails. A supporting frame extended outward from each of the main frame rails and cradled the engines. The placement of the engines added weight on the rear axle and helped improve adhesion to prevent wheel spin.

A header water tank was positioned in front of each engine, above the supercharger and alongside the cockpit. The intake scoops for the engines protruded above the car’s body and were located just forward of the cockpit. The intake duct ran under the header tank and to the supercharger. For the outer cylinder banks, individual exhaust stacks protruded from the car’s sides. Exhaust from each inner cylinder bank was collected by a manifold and directed out the upper-center of the car with the outlets protruding above the car’s body. Between the engines and the main frame rails was a 22 US gal (18 Imp gal / 82 L) oil tank and a 48 US gal (40 Imp gal / 182 L) fuel tank.


Nearing completion at Bean Industries, the Thunderbolt goes through an engine test run. Compressed air was used to start the engines. A cover is installed on only one of the engine intake scoops. Note the eight-sided radiator. (Getty image)

The output shaft of each engine was coupled to the three-speed transmission through friction plate clutches. However, the final drive gear was locked in using dog clutches to prevent slip. To accommodate the two engines, the transmission was 5 ft 6 in (1.68 m) wide and located on the Thunderbolt’s centerline. The transmission alone was 2,469 lb (1,120 kg) and contained 54 US gallons (45 Imp gal / 205 L) of gear oil, which was water cooled. It was connected to the drive wheels via a bevel gear. Two wheels were mounted to the drive axle on each side of the car, and each side used independent transverse leaf spring suspension. The drive shaft extended behind the rear axle and supported a hydraulically actuated disc brake, providing friction braking for the rear of the car. Behind the rear wheels were hydraulically controlled air brakes.

A light metal frame surrounded the car and served to secure Thunderbolt’s aluminum body panels. The body was designed by French aerodynamicist Jean Andreau. The body panels were made by Birmal Boat, Ltd and covered the car except for the radiator and cockpit openings. The panel fitment to the frame was so tight that bulges were needed to provide clearance for each wheel and for the valve covers of the engines’ outer cylinder banks. Most of the panels were designed to be quickly removed, especially those covering the wheels and engines. The Thunderbolt’s body tapered back behind the rear wheels, and a vertical tail was fixed at the extreme rear.


Thunderbolt at Bonneville in 1937. Note the fixed vertical tail. The dual rear wheels are visible, as is the outline of the retracted air brake. Eyston did not use the air brakes in 1937.

Thunderbolt was 30 ft 5 in (9.27 m) long, 7 ft 2 in (2.18 m) wide, and weighed 13,900 lb (6,305 kg). All eight tires were made by Dunlop and were 7.75 in (197 mm) wide and 44 in (1,118 mm) tall. The tires were mounted on wheels that were 7 in (178 mm) wide and 31 in (787 mm) tall and had a steel center and an aluminum rim. Each tire and wheel weighed around 210 lb (95 kg). At speed, the tires rotated 45 times each second and became 1.5 in (38 mm) taller. Reportedly, the R engines used in Thunderbolt had been derated for reliability and produced around 2,000 hp (1,491 kW). At power, the engines consumed 9.6 US gallons (8 imp gal / 36.4 L) of fuel each minute. Engines R25 and R27 were used in Thunderbolt. R25 was previously used in the Supermarine S.6B S1596 to set an absolute speed record of 379.05 mph (610.02 km/h) on 13 September 1931. Engine R27 had previously powered S.6B S1595 to set a new absolute speed record of 407.5 mph (655.1 km/h) on 29 September 1931. Additionally, Campbell had loaned one of his spare engines, R17, to Eyston.

Design and construction of Thunderbolt took about seven months, but the actual assembly of the car took only around six weeks. The car was built by Bean Industries, Ltd in Tipton, England. Reportedly, as the car was nearing completion, a public relations representative from Castrol wanted a name for the car. Eyston said to name it whatever he liked, and the representative from Castrol decided on Thunderbolt. Eyston, Speed of the Wind, and the untested Thunderbolt left for Bonneville in late August 1937 and arrived on 3 October. Upon arrival, the weather was poor, and the Salt Flats were in a sorry state. Time was needed for everything to improve, and that was time Eyston needed to finalize Thunderbolt.


This view of the Thunderbolt being serviced in 1937 with body panels removed shows the car’s inner frame. Part of the engine’s supporting cradle can be seen just under the engine. Note the coolant lines extending above the front axles.

Before getting behind the wheel of Thunderbolt, the fastest Eyston had ever driven was around 170 mph (275 km/h). On 28 October 1937, Eyston made his first test of Thunderbolt and decided to push the car. The run north was clocked at 309.6 mph (498.3 km/h), about eight mph (13 km/h) above the existing record set by Campbell. On the southbound return, the dog clutches were damaged by the engines running out of sync.

While Thunderbolt was being repaired, Eyston turned his attention to Speed of the Wind. On 3 November 1937, Eyston and co-driver Albert W. Denly set a new 12-hour record at 163.68 mph (263.42 km/h) and covered 2,000 miles (3,219 km) at an average speed of 163.75 mph (263.35 km/h). On 6 November 1937, Eyston made another run in Thunderbolt. Hoping to spare the clutches, Eyston utilized another vehicle to push start Thunderbolt and averaged 310.69 mph (500.01 km/h) on the northbound run. However, the dog clutches again failed on the southbound return.


Thunderbolt in 1938 with its new nose with rounded radiator opening, new intake scoops, and an extended tail. The headrest fairing has been extended back to the exhaust stacks, and the panels covering the wheels no longer have bulges. Barely visible are the shutters for radiator air exit on the car’s upper body between the first set of front wheels.

The clutch system underwent a modest redesign, and new parts were made. Some of the clutch redesign and new parts were made by Leo Goossen and Fred Offenhauser in Los Angeles, California. Eyston and the repaired Thunderbolt made another record run on 19 November 1937, with bad weather soon to close in. On the run north, Eyston shifted into second at 100 mph (161 km/h) and third at 200 mph 322 km/h). He covered the mile (1.6 km) at 305.34 mph (491.40 km/h) and the km (.6 mi) at 305.59 mph (491.80 km/h). It took just 16 minutes for Thunderbolt to be refueled and prepared for the return run with new tires. On the southbound leg, speed averages were 317.74 mph (511.35 km/h) for the mile (1.6 km) and 319.11 mph (513.56 km/h) for the km (.6 mi). Eyston’s goggles had gotten caught by the slipstream, and he had to grab them with one had while steering with the other at over 315 mph (507 km/h). All the effort had been enough—Eyston and the Thunderbolt set a new LSR of 311.42 mph (501.18 km/h) over the mile (1.6 km) and 312.20 mph (502.44 km/h) over the km (.6 mi).

In 1938, Thunderbolt was modified to improve its performance. The radiator inlet was extended slightly and rounded, with vanes added to help direct airflow. Shutters were added to the air exit to help regulate flow through the radiator. The engine intake scoops were enlarged, extended forward, and raised above the car’s body. The cockpit was enclosed by a rearward sliding canopy, and a respirator system was added. The respirator brought in fresh air from the front of the car. The headrest fairing was extended back to the center exhaust stacks. Exhaust manifolds replaced the individual stacks for the outer cylinder banks. New larger body panels without individual bulges over the wheels were installed. The vertical tail was decreased in size and modified so that it could be removed. The rear body of the car was extended for better streamlining. Coil springs replaced the heavy leaf springs used in the suspension. With all the modifications, Thunderbolt was lightened to about 12,000 lb (5,443 kg) and lengthened to about 35 ft (10.67 m).


After the timing camera failed to trigger in 1938, Thunderbolt had its sides painted black with matte paint to add contrast with the bright landscape. It was in this configuration that the car set its second LSR.

Eyston faced a challenger in 1938 in the form of John Rhodes Cobb and his LSR machine, the Railton. The Railton was designed by Reid Railton, powered by two Napier Lion engines, and much smaller and lighter than Thunderbolt. Eyston had intended to make his LSR runs in July, before Cobb arrived at Bonneville. However, bad weather and water on the course delayed any attempts until late August, by which time Cobb had arrived. On a test run at about 270 mph (435 km/h), smoke filled the cockpit due to an issue with the friction brakes. Eyston could hardly see and struggled to keep the car on the course. He felt that the respirator prevented asphyxiation and probably saved his life. Eyston decided to rely on the air brakes until the Thunderbolt slowed to 180 mph (290 km/h).

On 24 August 1938, Eyston averaged a blistering speed of 347.16 mph (558.70 km/h) on the northbound run. The return run may have been even faster, but the timing equipment malfunctioned and did not record a speed. The camera failed to trigger, most likely due to the lack of contrast with the silver car, white salt, and bright background. On the run, salt spray from the damp course was flung off the drive wheels and damaged the aluminum body panels above the wheels.


To beat Cobb, Thunderbolt’s vertical tail was removed and its nose faired over. This image shows the car with its tail removed. Also visible are the sliding canopy and the exhaust manifolds for the outer cylinder banks—all added for 1938. Note that the car’s sides are no longer black.

Repairs were made, and the sides of Thunderbolt were hastily painted matte black for contrast. On 27 August 1938, Eyston made another northward run and averaged 347.49 mph (559.23 km/h) over the mile (1.6 km) and 346.80 mph (558.12 km/h) over the km (.6 mi). The return south covered the mile at 343.51 mph (552.83 km/h) and the km at 344.15 mph (533.86 km/h). With those speeds, Eyston had established a new LSR at 345.49 mph (556.01 km/h) for the mile (1.6 km) and 345.21 mph (555.56 km/h) for the km (.6 mi). However, not to be outdone, Cobb bettered those marks on 15 September 1938, averaging 350.20 mph (356.59 km/h) over the mile (1.6 km) and 350.10 mph (563.43 km/h) over the km (.6 mi).

Eyston had been preparing Thunderbolt for more speed in case Cobb took the record. The Thunderbolt’s radiator was replaced with a water tank. The radiator intake and air exit between the front tires were faired over. With the runs lasting mere seconds, the water would not completely boil over. Two small scoops, perhaps to cool the front brake, were added behind the new nose. Small bulges for the first set of front wheels were added to the body panels. The car’s vertical tail was removed. The black paint that had been hastily applied was removed, and a back high-contrast section that incorporated a yellow circle was painted on the panel covering the rear set of front wheels. On 16 September 1938, one day after Cobb took the record, Eyston and Thunderbolt made another attempt. Northward, the mile (1.6 km) was covered at 356.44 mph (573.63 km/h) and the km (.6 mi) at 355.06 mph (571.41 km/h). Traveling south, the speed was 358.57 mph (577.06 km/h) for the mile (1.6 km) and 359.64 mph (578.78 km/h) for the km (.6 mi). The runs established a new LSR at an average of 357.50 mph (575.34 km/h) for the mile (1.6 km) and 357.34 mph (575.08 km/h) for the km (.6 mi). Eyston reported no stability issues in the tailless car, but said that the lack of a radiator caused the cockpit to get quite hot on the return run as the water boiled off.


Eyston and Thunderbolt setting their third LSR. The black, high-contrast section by the second front wheel is visible. Note the lack of a vertical tail.

Wanting to break the 360-mph (580-km/h) mark and go faster than six miles (9.7 km) per minute, Eyston took Thunderbolt out again on 21 September 1938. Just entering the measured mile (1.6 km) on the run north at over 360 mph (580 km/h), the cover for the right rear wheels broke free. As the cover tore loose, it damaged the two right rear tires and caused them to destroy themselves. Unsure of the issues, Eyston kept the throttle down through the mile (1.6 km), which was only about 10 seconds. After the mile (1.6 km), Thunderbolt skidded to a stop three miles (5 km) short of the course’s end with its right rear corner dragging. The car was too damaged to be repaired at Bonneville. Despite the damage and extra resistance through the measured mile (1.6 km), the car’s average speed was recorded as 349.85 mph (563.03 km/h).

While Cobb returned to Bonneville in August 1939 and set new LSRs at 368.86 mph (593.62 km/h) for the mile (1.6 km) and 369.74 mph (595.04 km/h) for the km (.6 mi), Eyston decided to take some time off from LSRs. Thunderbolt had been repaired, and Eyston knew it was capable of more speed, but not much more. Rather than racing again, Thunderbolt went on a world tour and was displayed at the New York World Fair in mid-1939. The car was subsequently sent to New Zealand and displayed in the British Pavilion at the Centennial Exhibition. After the Exhibition ended in May 1940, Thunderbolt and some other exhibits were stored at the Exhibition site in Rongotai, near Wellington. The outbreak of World War II put other priorities ahead of the exhibits. Extra space at the Exhibition site was used to store wool and several aircraft and aircraft engines. On 25 September 1946, the wool spontaneously ignited, and the blaze spread quickly amongst the 27,000 bales of wool in storage. Everything in the building, including Thunderbolt, was consumed by the fire. Reportedly, the remains of Thunderbolt were still located near the site as late as December 1956. The engines had been removed before the car was on display and were preserved. Engines R25 and R27 are respectively on display at the Royal Air Force Museum at Hendon and the London Science Museum.


Thunderbolt at the New York World’s Fair in 1939. This image illustrates the car in its final record-setting configuration. Note the covered nose, small scoops behind the nose, and the small bulges above the front set of wheels. The black section on the car’s side had a yellow circle at its center. The cockpit canopy and outer cylinder bank manifold are also visible. At this point, the R engines had been removed and mockups installed in their place. The Union Jack on the nose (and the rear body at one point) was added after the final record run.

This article is part of an ongoing series detailing Absolute Land Speed Record Cars.

The Land Speed Record 1920-1929 by R. M. Clarke (2000)
Fastest on Earth by Captain G. E. T. Eyston (1939)
Reid Railton: Man of Speed by Karl Ludvigsen (2018)
Land Speed Record by Cyril Posthumus and David Tremayne (1971/1985)
The Fast Set by Charles Jennings (2004)
Leap into Legend by Steve Holter (2003)
– “An Interview with Capt. G. E. T. Eyston” by William Boddy, Motor Sport (October 1974)
– “Thunderbolt Damaged in Speed Trail” San Pedro News Pilot (21 September 1938)


Pennsylvania Railroad 4-4-4-4 T1 Locomotive

By William Pearce

In the late 1930s, Baldwin Locomotive Works (Baldwin) of Eddystone, Pennsylvania sought a partner to support the design of an experimental, rigid-frame, duplex, 4-4-4-4 locomotive. With this wheel arrangement, the engine would have a four-wheel leading truck, two independent sets of four-wheel drivers, and a four-wheel trailing truck. As a duplex engine, each of the four-wheel drivers would be powered by a pair of separate cylinders. Baldwin’s Chief Engineer Ralph P. Johnson believed the newly designed engine would be capable of improved efficiency that would rival diesel locomotives, which were just beginning to outperform steam. Compared to an articulated locomotive, a rigid-frame duplex arrangement created a comparatively light engine well-suited for high speeds. In addition, having four smaller cylinders with a reduced piston speed decreased wear and maintenance compared to two larger, harder-working cylinders as used in a standard locomotive layout, such as a 4-8-4. If not well-balanced, the reciprocating and revolving forces of the drive wheels on powerful two-cylinder locomotives could actually damage the track, an issue that was alleviated with a four-cylinder duplex.


The T1 prototype, engine No. 6110, shortly after its completion by Baldwin in April 1942. The taper for the pointed nose extended much farther back than on the production engines. The front of the locomotive was enclosed with skirting, and casings extended the length of the engine, covering the top of the drive wheels. Note the gold accents and lettering.

Baldwin had just collaborated with the Pennsylvania Railroad (PRR) in creating the S1, which was finished in January 1939. The S1 was an experimental, rigid-frame, duplex locomotive with a 6-4-4-6 wheel arrangement. Designed to haul a 1,200-ton (1,089-t) passenger train at 100 mph (161 km/h), the very long S1 was PRR’s experimental trial with a duplex locomotive, and the company was interested in Baldwin’s new design. On 26 June 1940, PRR ordered two prototypes of Baldwin’s engine, but specified that it needed to use poppet valves and that the second prototype would be fitted with a booster engine on its trailing truck. PRR designated the prototype engines as the T1 class, and gave them engine numbers 6110 and 6111. Incidentally, the T1 prototypes were ordered before the S1 had entered regular service.

Starting in 1938, PRR had been experimenting with poppet valves in an effort to improve efficiency and increase power compared to the typical piston spool valve. In a standard Walschaerts valve gear, a piston spool valve was mounted in a valve chest above the double-acting cylinder. The spool valve slid back and forth, allowing steam to enter one side of the double-acting cylinder while simultaneously opening the other side to exhaust the previous steam charge. The steam flowed from the center of the valve chest into the front of the cylinder, pushing the piston back to the rear of the cylinder. The valve then slid rearward to direct steam into the rear part of the cylinder and allow the front part of the cylinder to exhaust. Steam entering the rear part of the cylinder pushed the piston forward, returning it to its original position. The efficiency of the design was limited since the admission and exhaust were both controlled by the single piston spool valve.


A glimpse inside of the cab of No. 6110 reveals the complex and utilitarian controls of even the most advanced steam engine. Image the heat, wind, soot, vibration, and sound of the locomotive under full steam at 100 mph (161 km/h).

With the Type A poppet valve system made by Franklin Railway Services Inc (Franklin), separate intake (admission) and exhaust valves opened to respectively allow the fresh steam charge into the cylinder and expel the spent charge after it acted on the piston. The head of these valves resembled a spoked wheel, the “hub” of which was mounted to the valve stem. When closed, the upper and lower rims of the head sealed against two separate seats. When open, steam flowed around the head and also flowed nearly unobstructed through the “spoked wheel” center of the head. The poppet admission and exhaust valves on the locomotive were independently controlled, allowing for different timing of when the valves opened and different durations of how long the valves were open. This flexibility enabled the most efficient flow of steam throughout all the various speeds the engine was traveling. PRR had installed poppet valves on a 4-6-2 K4 (No. 5399) locomotive and recorded an increase in power while operating at 80 mph (129 km/h) and above of over 40 percent.

The PRR T1 was a duplex locomotive that utilized a 4-4-4-4 wheel arrangement and was designed to haul 880 trailing tons (798-t) at 100 mph (161 km/h). PRR envisioned using the engines to haul express passenger trains on the 713-mile (1,147-km) route between Harrisburg, Pennsylvania and Chicago, Illinois. PRR anticipated that the T1 would replace its aging fleet of K4 engines.

The T1 used a four-wheel leading truck with 36 in (.91 m) wheels positioned at the front of the engine under the smokebox. A set of four 80 in (2.03 m) drive wheels followed, trailed by another nearly-identical set of four drive wheels. A four-wheel trailing truck with 42 in (1.07 m) wheels was positioned at the rear of the engine under the cab. To aid traction, sand carried in sand boxes could be deposited on the rails just ahead of the front drive wheels of each set. The two trucks and two sets of drive wheels were mounted in roller bearings to a single-piece frame bed made of cast steel by General Steel Castings in St Louis, Missouri. The cylinders and their valve chests were integrally cast with the frame, which was over 60 ft (18.29 m) long.


The second T1 prototype, No. 6111, displaying its unique styling done by Raymond Loewy. This engine was equipped with a booster engine, which was not included on any of the production locomotives.

The T1 was made up of a 68 ft 2.5 in (20.79 m) long engine and a 53 ft 9.5 in (16.40 m) long tender that carried the locomotive’s coal and water. This gave the complete engine an overall length of 122 ft 10 in (37.43 m). The Class 180 P 76 tender was supported by two eight-wheel trucks with 36 in (.91 m) wheels. It carried 82,000 lb (37,195 kg) of coal in a front compartment and 19,500 gallons (73,816 L) of water in a rear compartment. When combined with the 497,200 lb (225,526 kg) engine, the 433,000 lb (196,406 kg) tender gave the T1 a total weight of 930,200 lb (421,932 kg). The locomotive was 15 ft 6 in (4.72 m) tall and 11 ft 1 in (3.38 m) wide.

An HT type mechanical stoker auger transported coal from the tender to the engine’s firebox. The firebox was 138 in (3.51 m) long and 96 in (2.44 m) wide. Coal was burned in the firebox at around 2,000 °F (1,093 °C). Heat from the firebox flowed through the boiler via 184 tubes that were 2.25 in (57.2 mm) in diameter and 69 flues that were 5.5 in (139.7 mm) in diameter. Each of the tubes and flues was 18 ft (5.50 m) long. The 253 tubes and flues would stretch for 4,554 ft (1,388 m) if laid end to end. The boiler was made from approximately 1 in (254 mm) thick nickel steel. After passing through the tubes, the soot, embers, smoke, and heat from the burning coal flowed into a smokebox at the front of the engine and was subsequently vented into the atmosphere via dual vertical stacks that were approximately 20 in (508 mm) in diameter. Spent steam from the cylinders was directed through the smokebox and helped create the draft that drew air into the firebox, through the tubes, and out the stacks.


Altoona-built No. 5518 looking fairly fresh from the factory with its original front and skirting. The styled skirting was a holdover from the prototypes and was later removed to facilitate maintenance.

The tubes, flues, and firebox of the T1 had a combined evaporative surface area of 4,218 sq ft (391.9 sq m). Heat radiating from these surfaces turned water in the boiler to steam and built up a working pressure of 300 psi (20.7 bar). With a temperature of over 420 °F (215 °C), the wet, saturated steam was collected from a steam dome above the boiler. The steam then flowed to the Type A superheater, which had a surface area of 1,430 sq ft (132.8 sq m). From the superheater, small superheater elements (tubes) took the wet steam back into the flues. The steam inside the superheater elements was heated well above its saturation value and converted to dry, superheated steam. The superheater elements delivered the dry steam to the steam chamber in the superheater.

Mounted horizontally in a steam chest above each end of each cylinder were two 5.0 in (127 mm) admission valves and two 6.0 (152 mm) in exhaust valves, giving the T1 32 valves in total. All the valves for each cylinder were controlled by an oscillating camshaft mounted transversely above the center of the cylinder. The camshaft lifted the admission valve 1.0 in (25 mm) and the exhaust valve 1.25 in (32 mm). The admission valves allowed steam to enter the front side of the double-acting cylinder and fill its 7,965 cu in (130.5 L) volume, pushing the 19.75 in (558.8 mm) diameter piston back 26 in (660.4 mm) to the rear of the cylinder. The exhaust valves at the front of the cylinder opened to let out the spent charge while the admission valves at the rear of the cylinder let in a fresh charge. The steam then pushed the piston forward to its original position. The cylinder had a smaller return volume of 7,557 cu in (123.8 L) because the 4.5 in (114 mm) diameter piston rod occupied some space. The piston rod extended straight back from the cylinder and was attached to the connecting rod via a crosshead. The connecting rod linked the piston rod to the rear driving wheel in the two-wheel set on each side of the engine. Here, the connecting rod was attached to the coupling rod, which connected the two driving-wheels together. The reciprocating parts for each four-wheel driving set were supported with roller bearings and weighed 1,992 lb (904 kg). An 88-point forced lubrication system was included to keep the locomotive’s moving parts in good working order.


No, 5518 later in life than the above image. The skirting at the front of the engine has been removed, and stairs have replaced the foot and hand holds. The number plate keystone was moved from the front grille to under the headlight, and a new light was added to the grille. Note the shape of the valve chests above the cylinders. The two admission valves were positioned above, and were flanked by, the exhaust valves.

The T1 engine developed around 6,550 indicated hp (4,884 kW) at 85 mph (137 km/h), with a maximum recorded output of 6,665 hp (4,970 kW). The engine had a maximum tractive effort of some 64,650 lbf (287.58 kN) based on an 85 percent efficiency factor. Without any slip, each rotation of the drive wheels moved the engine 20 ft 11 in (6.38 m). At 100 mph (161 km/h), each drive wheel rotated 420 times a minute, and each double-acting piston made 840 strokes. This resulted in roughly 15,091 cu ft (427.33 cu m) of steam passing through the T1’s four cylinders every minute.

The Franklin booster engine fitted to engine No 6111, the second prototype, consisted of two steam-operated cylinders that powered the trailing truck’s rear wheels. The unit was mounted to the rear of the trailing truck and was typically used to help start the locomotive from a standstill, assist with low-speed operation, and provide additional power up grades. The booster engine added 11,200 lb (5,080 kg) to the locomotive’s weight but provided an additional 13,500 lbf (60.05 kN) of tractive effort.

The exterior of the T1 was styled by industrial designer Raymond Loewy. Cladding encased the locomotive and tapered to a wedge at the front of the engine. Casings that concealed the top of the driving wheels covered the sides of the engine. The locomotive was finished in a dark Brunswick green (Dark Green Locomotive Enamel) with gold accents and lettering. Engine No. 6110 was completed in April 1942 with 6111 following in May. The T1 prototypes underwent a series of tests, one of which measured the engine’s machine efficiency at 93 percent, and another indicating more than 6,000 hp (4,474 kW) for all speeds above 55 mph (89 km/h). After successfully passing the tests, PRR pressed the engines into service, but only on a limited basis. The engines had no trouble averaging more than 100 mph (161 km/h) over portions of their route between Harrisburg and Chicago. By April 1944, No. 6110 had accumulated 120,000 miles (193,121 km), but 6111 had traveled less. No. 6110 could produce 4,100 drawbar hp (3,057 kW) at 100 mph (161 km/h) and outperform a 5,400 total hp (4,027 kW) four-unit diesel at all speeds above 26 mph (42 km/h). However, that was just performance and did not consider maintenance or crew costs.


Just completed by Baldwin, No. 5526’s nearly black Brunswick green paint shines on a bright day in November 1945.

The PRR was sufficiently impressed by the T1’s performance that they ordered 50 examples in February 1945. Production was split evenly between Baldwin and PRR’s Juniata Locomotive Shops in Altoona, Pennsylvania. Engine number assignments were 5500–5524 for Altoona and 5525–5549 for Baldwin. The production version of the T1 had a flatter prow and shorter casings that exposed the drive wheels. As production continued, the casing was trimmed back farther, and the locomotive’s nose was made more utilitarian, with stairs replacing the hand and foot grips. The suspension was revised on the production T1s in an attempt to reduce the engine’s proclivity for wheel slip. At 502,200 lb (227,794 kg), the production T1 weighed an additional 5,000 lb (2,268 kg). Production T1s were not fitted with a booster engine, which cut 15 in (381 mm) off the engine’s length, reducing it to 66 ft 11.5 in (20.41 m). However, the tender gained 15 in (381 mm), making it 55 ft .5 in (16.78 m) long and leaving the T1’s overall length unchanged. Altoona was responsible for manufacturing all 50 tenders. The tender was modified as the Class 180 P 84, carrying an additional 3,200 lb (1,451 kg) of coal with a 300-gallon (1,136-L) reduction of water. The tender’s total weight increased by 9,500 lb (4,309 kg) to 442,500 lb (200,715 kg). Combined with the heavier engine, the production T1’s total weight was 944,700 lb (428,509 kg).

Altoona and Baldwin both finished their first production T1s in November 1945. Altoona completed eight of the engines by the end of the year, and Baldwin built five. The remaining 37 engines, 17 from Altoona and 20 from Baldwin, were delivered in 1946. The last Altoona-built T1, No. 5524, was completed in June 1946, and it ended up as the last steam locomotive built at the works. The last Baldwin T1 was delivered in August 1946. Each engine, without its tender, cost around $250,000.

When the T1 was running well, it was fast and smooth. The engine was a free steamer—it could run full throttle and maintain boiler pressure, but it also ran dirty. In service, the locomotive quickly covered itself with soot and grime. The T1 had no issues operating above 100 mph (161 km/h), and one engine pulled 1,150 tons (1,043-t) at that speed. However, with the T1 now in service, its tendency toward wheel slip became more of an issue. Wheel slip was encountered with the prototypes, but the situation was apparently much worse with numerous T1 engines in service. More than likely, the prototypes were carefully operated by more-experienced crews, which minimized any tendency for wheel slip. However, wheel slip was a regular occurrence with the production engines operated in normal service by crews transitioning from the forgiving K4. Some T1s were modified to deposit sand in front of all drive wheels in an effort to minimize wheel slip. Skilled engineers avoided slip with the application of sand and smooth, careful throttle movements until the locomotive was above 25 mph (40 km/h).


Another image of No, 5526 in front of the Baldwin works. Compared to the prototype T1s, the nose of the production engines was more blunt with less taper, and much of the side casing was omitted.

The worst wheel slip was encountered at speed when the engine would pass over some type of irregularity on the track, including moisture. The front set of drivers would slip, then catch. As soon as they caught, the rear set would slip, and then catch. This would create an imbalance and cause the front drivers to slip again, repeating the whole process. At 80 mph (129 km/h), the slip was very unsettling, and the crew had to cut power and reduce speed to stop the oscillating front-rear driver slippage. Suspension changes helped tame the T1’s wheel slip at higher speeds.

The wheel slip could also damage or break the engine’s poppet valves. Maintenance and repair of the valves and their control and drive boxes proved to be very difficult. Much of the drive system was inaccessible unless the engine was over a maintenance pit. Beyond the wheel slip, the valves began to fail in an unpredictable manner. Franklin had guaranteed the valves for continuous operation at 100 mph (161 km/h) and short bursts up to 125 mph (201 km/h). After inspecting every valve and scrutinizing numerous maintenance records, Franklin was no closer to discovering what was causing the failures. However, the majority of the valve failures occurred over a high-speed section of rail between Crestline, Ohio and Fort Wayne, Indiana. Franklin sent an observer to secretly ride the route for a month and document the train’s activity. The observer’s log detailed some remarkable findings; the T1s were often operated in excess of 130 mph (209 km/h) to make up time. One train was clocked at 142 mph (229 km/h) over several miles. Even if this one calculation was done in error, the numerous times the T1 was calculated at over 130 mph (209 km/h) could not all be mistakes. The speedometer in the cab of the locomotive stopped at 120 mph (193 km/h).


Baldwin-built No. 5533 was delivered in January 1946. As seen in this manufacturer photo, it lacks the polish applied to No. 5526 two months previous. Note that the front cylinder’s piston rod was much longer than that of the rear cylinder.

Franklin management decided the best course of action was to not inform PRR that their engineers were regularly overspeeding the trains and operating them beyond the guaranteed limits of the valves. Rather, the company decided to find a better metal that would allow the valves to endure the higher speeds and would also make the valve immune to damage from wheel slip. Franklin management felt that a more material would better serve any railroad interested in utilizing poppet valves. Although various materials were evaluated and numerous valve redesigns were considered, no solution was found. The Franklin poppet valves were simply prone to failure above 130 mph (209 km/h).

In fall 1946, T1 engine Nos. 5511 and 5539 were loaned to the Chesapeake & Ohio Railway (C&O) for trials. While C&O ultimately did not purchase any of the engines, they noted that the T1 handled well, particularly at higher speeds, and was able to make up time between stops. Most interesting, C&O did not feel that the T1 had any excessive tendency toward wheel slip.

In 1947, engine No. 6111 had its cylinders lined, which reduced the bore by 1.0 in (25 mm) to 18.75 in (476 mm). The modification was done to reduce the engine’s tractive effort and subsequently reduce wheel slippage. Seven or eight additional T1s were later modified with the cylinder liners. Also in 1947, PRR reported a net loss for the 1946 year, which was the first time in the company’s history that it did not turn a profit.


No. 5534 seen early in its career with the original front. However, the engine has a good layer of soot and dirt. Note that the tender is not marked.

In early 1948, PRR was actively converting its locomotive fleet to diesel power. In July 1948, T1 No. 5550 was fitted with Franklin Type B rotary cam valves. This change was done solely as an experiment to test the Type B unit, which was simpler and easier to maintain than the original Type A oscillating cam system. This experiment was not meant to solve the issues of valves breaking, and no other T1s were modified with the Type B unit.

In July 1949, engine No. 5547 had its Franklin Type A oscillating cam poppet valves replaced with a conventional Walschaerts valve gear. The engine was subsequently reclassified as T1a, but it was too little, too late for the T1 and PRR’s steam engines. By the end of 1949, most of the T1s had been withdrawn from service, with all of them being dropped from PRR’s roster by the end of 1953. Scrapping of the engines began in 1951, with the last T1 going under the torch in 1956. While the T1 was in standard service, engines regularly racked up over 8,000 miles (12,875 km) per month. However, steam locomotives could not match the reliability of diesel engines or their comparatively low maintenance and crew costs.

In 2013, the Pennsylvania Railroad T1 Steam Locomotive Trust (T1 Trust) was founded to build a new PRR T1 locomotive, No. 5550. Since its inception up to mid-2020, the T1 Trust has acquired or completed 34 percent of the new engine and its tender and has numerous other parts and components on order. It is the intention of the T1 Trust to complete No. 5550 by 2030 and to make the engine available for special excursion service. The T1 Trust also hopes to use No. 5550 for an attempt to break the world land speed record for a steam locomotive, which was set by the British LNER (London and North Eastern Railway) Class A4 4468 Mallard at 125.88 mph (202.58 km/h) on 3 July 1938.


An incredibly dirty engine No. 5528 sits unused in a railyard covered with soot and grime. The T1 was known to run dirty, but this engine appears to be neglected. Note the rolling stock positioned on the track immediately before the T1 and that wedges are jammed behind the engine’s rear set of drive wheels. Being cut up for scrap was the unglamorous end for all 52 T1 locomotives.

Loco Profile 24: Pennsylvania Duplexii by Brian Reed (June 1972)
Pennsy Power: Steam and Electric Locomotives of the Pennsylvania Railroad, 1900-1957 by Alvin F. Staufer (1962)
American Steam Locomotives: Design and Development, 1880–1960 by William L. Withuhn (2019)


Mitsubishi [Ha-43] (A20 / Ha-211 / MK9) Aircraft Engine

By William Pearce

In 1916, the Internal Combustion Engine Section, Machinery Works (Nainenki-ka Zokisho) of the Mitsubishi Shipbuilding Company Ltd (Mitsubishi Zosen KK) was formed to build aircraft engines. A number of licenses to build engines in Japan were acquired from various European engine manufacturers. Initially, the engines were of the Vee type. The aircraft engine works was renamed Mitsubishi Aircraft Company Ltd (Mitsubishi Hokuki KK) in 1928. In the late 1920s, licenses were acquired to produce the five-cylinder Armstrong Siddeley Mongoose and the nine-cylinder Pratt & Whitney R-1690 Hornet air-cooled radial engines.


Front and side views of the Mitsubishi [Ha-43] (A/20 / Ha-211 / MK9). The engine performed well but was underdeveloped. Its development and production were slowed by bombing raids and materiel shortages. The engine powered two of Japan’s best next-generation fighters, the A7M2 and Ki-83. While the aircraft were excellent, the war was already lost.

In 1929, Mitsubishi built the first aircraft engine of its own design. Carrying the Mitsubishi designation A1, the engine was a two-row, 14-cylinder, air-cooled radial of 700 hp (522 kW). This engine was followed in 1930 by the A2, a 320 hp (237 kW) nine-cylinder radial. A larger 600 hp (477 kW) nine-cylinder engine, the A3, was also built the same year. None of these early engines were particularly successful, and only a small number were built: one A1, 14 A2s, and one A3. However, Mitsubishi learned many valuable lessons that it applied to its next engine, the A4 Kinsei.

The two-row, 14-cylinder A4 was developed in 1932 and was initially rated at 650 hp (485 kW). The A4 had a 5.51 in (140 mm) bore, a 5.91 in (150 mm) stroke, and a total displacement of 1,973 cu in (32.33 L). In 1934, Mitsubishi consolidated its subsidiaries and became Mitsubishi Heavy Industries Ltd (Mitsubishi Jukogyo KK). Also in 1934, an upgraded version of the A4 engine was developed as the 830 hp (619 kW) A8 Kinsei. The Kinsei was under continual development through World War II, and numerous versions of the engine were produced. Ultimately, the last variants were capable of 1,500 hp (1,119 kW), and production of all Kinsei engines totaled approximately 15,325 units.

In mid-1941, Mitsubishi began work on an 18-cylinder engine that carried the company designation A20. The engine was intended to be lightweight and produce 2,200 hp (1,641 kW). The A20 design was developed from the Kinsei, although the 18-cylinder A20 really only shared its bore and stroke with the 14-cylinder engine—it is not even clear if the pistons were interchangeable. The team at Mitsubishi designing the A20 engine were Kazuo Sasaki—main engine section; Kazuo Inoue, Ding Kakuda, and Mitsukuni Kada—supercharger and auxiliary equipment; Katsukawa Kurokawa—propeller gear reduction; Shigeta Aso—engine cooling; Shuichi Sugihara—fuel injection system, and Shin Nakano—turbosupercharger. The A20 eventually carried the Imperial Japanese Army (IJA) designation Ha-211, the Imperial Japanese Navy (IJN) designation MK9, and the joint designation [Ha-43]. For simplicity, the joint designation will primarily be used. However, few sources agree on the engine’s various sub-type designations, and there is some doubt regarding their accuracy.


The mockup of the Tachikawa Ki-94-I illustrated the aircraft unorthodox configuration. With its two [Ha-43] engines, the fighter had an estimated top speed of 485 mph (781 km/h). However, its complexity led to its cancellation and the pursuit of a more conventional design.

The Mitsubishi [Ha-43] had two rows of nine cylinders mounted to an aluminum crankcase. The crankcase was formed by three sections. Each section was split vertically through the centerline of a cylinder row, with the middle section split between both the front and rear cylinder rows. Each crankshaft section contained a main bearing to support the built-up, three-piece crankshaft. An additional main bearing was contained in the front accessory drive. The cylinders were made up of a steel barrel screwed and shrunk into a cast aluminum head. Each cylinder had one intake valve and one sodium-cooled exhaust valve. The valves were actuated by separate rockers and pushrods. Unlike the Kinsei engine, the [Ha-43] did not have all of its pushrods at the front of the engine. The [Ha-43] had a front cam ring that drove the pushrods for the front cylinders, and a rear cam ring that did the same for the rear cylinders. When viewed from the rear, the cylinder’s intake port was on the right side, and the exhaust port was on the left. Sheet metal baffles attached to the cylinder head helped direct the flow of cooling air through the cylinder’s fins. Cylinder numbering proceeded clockwise around the engine when viewed from the rear. The vertical cylinder atop the second row was No. 1 Rear, and the inverted cylinder under the front row was No. 1 Front.

At the front of the engine was the propeller gear reduction and the magneto drive. Planetary gear reduction turned the propeller shaft clockwise at .472 times crankshaft speed. Each of the two magnetos mounted atop the gear reduction fired one of the two spark plugs mounted in each cylinder. One spark plug was located on the front side of the cylinder and the other was on the rear side. A 14-blade cooling fan was driven by the propeller shaft and mounted in front of the gear reduction. Not all [Ha-43] engines had a cooling fan. At the rear of the engine was an accessory and supercharger section. The single-stage, two-speed, centrifugal supercharger was mechanically driven by the crankshaft. Individual intake runners extended from the supercharger housing to each cylinder. The intake and exhaust from the front cylinders passed between the rear cylinders, with the exhaust running above the intake runners. The supercharger’s inlet was directly behind the second row of cylinder. Behind the inlet was a fuel distribution pump that directed fuel to an injector installed by the inlet port of each cylinder.

The 18-cylinder [Ha-43] had a 5.51 in (140 mm) bore a 5.91 in (150 mm) stroke, and displaced 2,536 cu in (41.56 L). The basic engine with its 7.0 to 1 compression ratio and single-stage, two-speed supercharger produced 2,200 hp (1,641 kW) at 2,900 rpm and 10.1 psi (.69 bar) of boost for takeoff. Military power was 2,050 hp (1,527 kW) at 3,281 ft (1,000 m) in low gear and 1,820 hp (1,357 kW) at 21,654 ft (6,600 m) in high gear. Both power ratings were produced at 2,800 rpm and 8.1 psi (.56 bar) of boost. Anti-detonation (water) injection was available, but it is not clear at what point it was used—most likely for military power and above. The engine was 48 in (1.23 m) in diameter, 82 in (2.09 m) long, and weighed 2,161 lb (980 kg).


The high-altitude Tachikawa Ki-74 was built around a pressure cabin for high-altitude flight. The aircraft most likely has [Ha-43] engines with a 14-blade cooling fan. The [Ha-42] engine had a 10-blade cooling fan. The exhaust from the turbosupercharger can be seen on the right side of the image.

[Ha-43] design work was completed in October 1941. The first engine was built at the Mitsubishi No. 2 Engine Works (Mitsubishi Dai Ni Hatsudoki Seisakusho), which was located in Nagoya and developed experimental engines, and was finished in February 1942. As the [Ha-43] was being tested, Mitsubishi proposed in April 1942 to use the engine for its new A7M fighter. The first [Ha-43] engine for the IJA was completed in August 1942. In September 1942, the IJN selected the 2,000 hp (1,491 kW) Nakajima [Ha-45] engine for the A7M1 and many of its other high-powered fighter projects under development. This setback inevitably slowed development of the [Ha-43]. At the time, there were no applications for the engine, with the IJA feeling it was too powerful and the IJN selecting the Nakajima engine. Two more [Ha-43] engines, one each for the IJA and IJN were completed in November 1942.

Mitsubishi continued development at a slow pace, hampered in part by difficulties with designing turbine wheels for the engine’s remote turbosupercharger. It was not until June 1943 that the [Ha-43] passed operational tests and began to be selected for installation on several aircraft types and not just projects. The first [Ha-43]-powered aircraft to fly was the third prototype of the Tachikawa Ki-70. The Ki-70 was a twin-engine reconnaissance aircraft with a glazed nose and twin tails. Originally powered by two 1,900 hp (1,417 kW) Mitsubishi [Ha-42] engines, the aircraft’s performance was lacking, and the third prototype was built with two turbosupercharged [Ha-43] 12 (IJA Ha-211-IRu) engines. The [Ha-43] 12 produced 2,200 hp (1,641 kW) for takeoff; 1,930 hp (1,439 kW) at 16,404 ft (5,000 m); and 1,750 hp (1,305 kW) at 31,170 ft (9,500 m). First flying in late 1943, the [Ha-43] 12-powered aircraft still underperformed, and the engines were unreliable. Development of the Ki-70 was abandoned.


The Mitsubishi A7M2 Reppu (Strong Gale) with its [Ha-43] 11 engine did not have a cooling fan like the A7M1. As a result, the cowling was redesigned with a larger opening and scoops for the engine intake (top) and oil cooler (lower). Note that the individual exhaust stacks were grouped together, mostly in pairs.

In 1943, Tachikawa designed the tandem-engine, twin-boom Ki-94-I (originally Ki-94) fighter powered by two [Ha-43] 12 (IJA Ha-211-IRu) engines. The cockpit was positioned between the two engines, which were mounted in a push-pull configuration in the short fuselage that sat atop the aircraft’s wing. The front and rear engines both turned four-blade propellers. The front propeller was 10 ft 10 in (3.3 m) in diameter, and the rear was 11 ft 2 in (3.4 m) in diameter. After a mockup was inspected in October 1943, the design was judged to be too unorthodox and complex. This resulted in a complete redesign to a more conventional single engine aircraft, the Ki-84-II, which was powered by a 2,400 hp (1,790 kW) Nakajima [Ha-44] engine.

In early 1944, two [Ha-43] 12 (IJA Ha-211-I) engines were installed in the Tachikawa Ki-74, a pressurized, high-altitude, long-range reconnaissance bomber with a conventional taildragger layout. With only the mechanical two-speed supercharger, the [Ha-43] 12 produced 2,200 hp (1,641 kW) for takeoff; 2,020 hp (1,506 kW) at 3,281 ft (1,000 m) in low gear; and 1,800 hp (1,342 kW) at 16,404 ft (5,000 m) in high gear. The Ki-74 made its first flight in March 1944, and turbosupercharged [Ha-43] 12 (IJA Ha-211-IRu) engines were installed in the second and third prototypes. The turbosupercharger was located behind the engine on the outer side of the nacelle and improved the aircraft’s performance at altitude. However, the [Ha-43] engines were still under development and suffered from reliability and vibration issues. Subsequent Ki-74 aircraft used larger and less-powerful Mitsubishi [Ha-42] engines.


Like the A7M2, the Mitsubishi Ki-83 also did not use a cooling fan on its [Ha-43] engine. However, the Ki-83 did have a turbosupercharger which helped it achieve its very impressive performance of at least 438 mph (705 km/h) at 29,530 ft (9,000 m). Note the sheet-metal baffles on the cylinder heads.

In the summer of 1944, Mitsubishi was given permission to install a [Ha-43] 11 (IJN MK9A, similar to the [Ha-43] 12) engine in an A7M1 airframe, creating the A7M2. The Mitsubishi A7M Reppu (Strong Gale) was a carrier-based fighter intended to replace the A6M Zero. The A7M1 prototypes had underperformed with the 2,000 hp (1,491 kW) Nakajima [Ha-45] engine selected by the IJN. The [Ha-43]’s installation in the A7M2 was conventional, and the aircraft made its first flight on 13 October 1944. Performance met expectations, and the A7M2 was ordered into production. Subsequently, manufacturing of the [Ha-43] started to ramp up, with 13 engines being built in March 1945. The following month, [Ha-43] 11 production was sanctioned at the Mitsubishi No. 4 Engine Works (Mitsubishi Yon Hatsudoki Seisakusho) in Nagoya. On 1 May 1945, Mitsubishi No. 18 Engine Works (Mitsubishi Dai Juhachi Hatsudoki Seisakusho) was established in Fukui city to build [Ha-43] 11 engines for the IJN, while the No. 4 Engine Works would build engines for the IJA. As events played out, only seven or eight A7M2s were built by the end of the war, the No. 18 Engine Works never produced a complete engine, and bombing raids prevented the March 1945 [Ha-43] production numbers from ever being eclipsed.

Further developments of the A7M were planned, such as the A7M3 powered by a [Ha-43] 31 (IJN MK9C) engine with a single-stage, three-speed mechanical supercharger. The [Ha-43] 31 produced 2,250 hp (1,678 kW) for takeoff; 2,000 hp (1,491 kW) at 5,906 ft (1,800 m) in low gear; 1,800 hp (1,342 kW) at 16,404 ft (5,000 m) in medium gear; and 1,660 hp (1,238 kW) at 28,543 ft (8,700 m) in high gear. The three-speed supercharger added about 5.4 in (138 mm) to the engine’s length and 88 lb (40 kg) to the engine’s weight, increasing the respective totals to 87 in (2.22 m) and 2,249 lb (1,020 kg). The A7M3-J would incorporate the [Ha-43] 11 engine with a turbosupercharger installed under the cockpit to produce 2,200 hp (1,641 kW) for takeoff; 2,130 hp (1,588 kW) at 22,310 ft (6,800 m); and 1,920 hp (1,432 kW) at 33,793 ft (10,300 m). While the A7M2 did not have a cooling fan, one was used in the A7M3 and A7M3-J designs.


The turbosupercharger installed in the Ki-83’s left engine nacelle. The large duct on the right was for the exhaust after it passed through the turbosupercharger. The outlet at the end of the nacelle was from the wastegate. Both were positioned to provided additional thrust. The Ki-83 had a ceiling of 41,535 ft (12,660 m).

In the fall of 1944, two [Ha-43] 12 (IJA Ha-211-IRu) engines were installed in the Mitsubishi Ki-83. The Ki-83 was a twin-engine heavy fighter with a conventional taildragger layout. A turbosupercharger was placed in the rear of each engine nacelle. Fresh air would enter the turbocharger near the rear of the nacelle on the outboard side, be compressed, and then flow to the engine through an air box in the upper nacelle. The engine’s exhaust was expelled from the turbocharger on the inboard side of the nacelle, and a wastegate was positioned at the end of the nacelle. The exhaust arrangement provided some additional thrust. Each engine turned an 11 ft 6 in (3.5 m) diameter, four-blade propeller. The Ki-83 made its first flight on 18 November 1944, but with the main focus on single-engine interceptors, only one was built before the Japanese surrender.

In April 1945, a [Ha-43] 42 (IJN MK9D) was installed in the Kyushu J7W1 Shinden (Magnificent Lightning), an unconventional pusher fighter with a canard layout. The [Ha-43] 42 had two-stage supercharging, with the first stage made up by a pair of transversely-mounted centrifugal impellers, one on each side of the engine. The shaft of these impellers was joined to the engine by a continuously variable coupling. The output from each of the first stage impellers joined together as they fed the normal, two-speed supercharger mounted to the rear of the engine and geared to the crankshaft. The [Ha-43] 42 produced 2,030 hp (1,514 kW) at 2,900 rpm with 9.7 psi (.67 bar) of boost for takeoff. Military power at 2,800 rpm and 5.8 psi (.40 bar) of boost was 1,850 hp (1,380 kW) at 6,562 ft (2,000 m) in low gear and 1,660 hp (1,238 kW) at 27,559 ft (8,400 m) in high gear. An extension shaft approximately 29.5 in (750 mm) long extended back from the engine to a remote propeller reduction gear box. The gear reduction turned the 11 ft 2 in (3.40 m), six-blade propeller at .412 times crankshaft speed and also drove a 12-blade cooling fan that was 2 ft 11 in (900 mm) in diameter.


The [Ha-43] 42 (IJN MK9D) installed in the Kyushu J7W1 Shinden, pictured while the aircraft was in storage at the Smithsonian National Air and Space Museum’s Paul E. Garber facility. The front of the aircraft is on the left. One of the two transversely-mounted, first-stage superchargers can be seen left of the engine, and the ducts from both superchargers can be seen joining together as they feed the mechanically-driven supercharger at the rear of the engine. Note that the exhaust stacks are flowing toward the front of the engine (rear of the aircraft).

Since the engine was mounted with the propeller shaft toward the rear of the aircraft, it incorporated new cylinders with the exhaust port on the side opposite of the intake port. The intake port faced toward the supercharger (front of the aircraft), and the exhaust port faced toward the propeller (rear of the aircraft). The engine’s individual exhaust pipes were used to help the flow of air through the cowling and oil coolers. After flowing through the oil cooler on each side of the aircraft, air was mixed with the exhaust from four cylinders and ejected out a slit on the side of the fuselage just before the spinner. The ejector exhaust helped draw air through the oil coolers. The same was true for the exhaust from the lower six cylinders, which was ducted into an augmenter that helped draw cooling air through the engine cowling and out an outlet under the spinner. The exhaust from the remaining four cylinders, which were located on the top of the engine, exited via two outlets arranged atop the cowling to generate thrust.

The J7W1 made its first flight on 3 August 1945. The third J7W1 was planned to have a [Ha-43] 43 engine that used a single impeller for its first-stage, continuously variable supercharger and produced an additional 130 hp (97 kW) for takeoff. Production J7W1 aircraft would be powered by a 2,250 hp (1,678 kW) [Ha-43] 51 engine with a single-stage, three-speed, mechanical supercharger replacing the two-stage setup with the continuously variable first stage. The engine would turn a four-blade propeller, 11 ft 6 in or 11 ft 10 in (3.5 m or 3.6 m) in diameter. However, only the first J7W1 was completed by war’s end.


The [Ha-43] 11 engine with cooling fan in storage as part of the Smithsonian National Air and Space Museum’s collection. Note the rust on the steel cylinder barrels. The spark plug wires are disconnected and desiccant plugs have been installed to help preserve the engine. (Tom Fey image)

In January 1945, construction commenced on the Mansyu Ki-98 (or Manshu Ki-98), a twin-boom pusher fighter with tricycle undercarriage. A single, turbosupercharged [Ha-43] 12 (IJA Ha-211-IRu) engine turning an 11 ft 10 in (3.6 m) four-blade propeller would power the aircraft. With the exception of the turbosupercharger, the installation was similar to that of the J7W1 with an extension shaft and remote propeller gear reduction. The prototype was ready for assembly when it was destroyed in August 1945 to prevent its capture by Soviet forces.

In addition to the aircraft listed above, the [Ha-43] was selected to power a number of aircraft projects that were not built. Plans were initiated to use the [Ha-43] to repower a number of different production aircraft that used the 2,000 hp (1,491 kW) Nakajima [Ha-45]. However, none of these retrofit redesigns were carried out before the end of the war. From 1942 to 1945, the production run of the [Ha-43] amounted to only 77 engines, and it was not fully developed by the end of the war.

At least three [Ha-43] engine survive, and all three are held by the Smithsonian National Air and Space Museum. One engine does not have a cooling fan and is probably a [Ha-43] 11 for a A7M2. The second engine is a [Ha-43] 11 with a cooling fan. The third engine is a [Ha-43] 42 still installed in the J7W1 prototype. All of the engines are in storage and not on display.


The fanless [Ha-43] 11 engine held by the Smithsonian National Air and Space Museum. The fuel distribution pump with its 18 lines can be seen atop the rear of the engine. The small-diameter lines appear to be made of copper.

Japanese Aero-Engines 1910 – 1945 by Mike Goodwin and Peter Starkings (2017)
Japanese Secret Projects by Edwin M. Dyer III (2009)
Japanese Secret Projects 2 by Edwin M. Dyer III (2014)
Japanese Aircraft of the Pacific War by René J. Francillon (1979/2000)
The History of Mitsubishi Aero-Engines 1915–1945 by Matsuoka Hisamitsu and Nakanishi Masayoshi (2005)
– “Mitsubishi Heavy Industries, LTD” The United States Strategic Bombing Survey, Corporation Report No. I (June 1947)
– “Design Details of the Mitsubishi Kinsei Engine” by W. G. Ovens, Aviation (August 1942)


Kyushu J7W1 Shinden Interceptor Fighter

By William Pearce

Masayoshi Tsuruno (also spelled Masaoki) was a member of the Imperial Japanese Navy’s (IJN) Aviation Research Department. Around 1940, Tsuruno first began to investigate designs of a pusher aircraft with a canard layout. Tsuruno’s research led him to believe that such a configuration would enable an aircraft to achieve a very high level of performance. In addition, the basic configuration could be easily adapted to turbojet power if such an engine became available.


Kyushu J7W1 Shinden was an unorthodox fighter designed to intercept US bombers at high speed and high altitude. Although just two were completed, it was the only canard aircraft ordered into production during World War II. Exhaust from two cylinders flowed out the two ejector slits atop the engine cowling.

In early 1943, the IJN issued 18-Shi Otsu specification calling for a land-based fighter capable of intercepting enemy bombers. The aircraft should achieve 460 mph (740 km/h) at 28,543 ft (8,700 m), reach 26,247 ft (8,000 m) in 10.5 minutes, have a service ceiling of 39,370 ft (12,000 m), and carry four 30 mm cannons. Tsuruno worked up a design for such an aircraft and submitted it to the IJN. The IJN liked the design but was hesitant to move forward with the radical, untested configuration. Tsuruno was able to work with the First Naval Air Technical Depot (Dai-Ichi Kaigun Koku Gijitsusho) at Yokosuka to develop a proof of concept, designated MXY6.

The Yokosuka MXY6 was a glider of all wooden construction possessing a canard layout with fixed tricycle landing gear. The aircraft featured a foreplane with elevators mounted to its nose for pitch control. The swept wings were mounted to the rear fuselage, and each wing had a vertical stabilizer with a rudder mounted near its mid-point. Three of the gliders were built by Chigasaki Industry Ltd (Chigasaki Seizo KK). Piloted by Tsuruno, the MXY6’s first flight was made in January 1944. Later, one of the gliders was fitted with a 22 hp (16 kW) Nippon Hainenki Semi 11 [Ha-90] engine turning a wooden, fixed-pitch, two-blade propeller. The engine was not intended make the MXY6 fully operational under its own power, but it would enable the aircraft to sustain flight and prolong its glide. The MXY6’s flight tests indicated that Tsuruno’s design was sound. The aircraft handled well at low speeds and resisted stalling. Based on the positive preliminary tests of the MXY6, the IJN decided to proceed with Tsuruno’s 18-Shi Otsu design in February 1944. The aircraft would be built by the Kyushu Airplane Company (Kyushu Hikoki KK), and it was designated J7W1 Shinden (Magnificent Lightning).


One of the Yokosuka MXY6 gliders that survived to the end of the war and was found by US forces. The glider validated the basic configuration that was later applied to the J7W1.

Kyushu Airplane Company was founded in October 1943 as a subsidiary of the Watanabe Iron Works Ltd (Watanabe Tekkosho KK). Kyushu was selected as the manufacturer because it had both workers and production facilities that were available. Kyushu had no experience designing high-performance fighter aircraft, but the company would be aided by Tsuruno and the First Naval Air Technical Depot. An official order for the J7W1 was issued in June 1944, with the prototype’s first flight expected in January 1945.

The Kyushu J7W1 Shinden used the same layout as the MXY6, having a canard configuration with a swept, rear-mounted wing and tricycle undercarriage. The aircraft consisted of an aluminum airframe covered by aluminum panels, forming a monocoque structure. Depending on location, the panels were either flush riveted or spot welded in place. The control surfaces were skinned with aluminum. The foreplane had two spars and was mounted to the extreme nose of the aircraft at a one-degree angle of incidence. A leading-edge slat was deployed with the flaps. On the foreplane’s trailing edge was a two-section flap. The first section acted as a traditional flap that extended 26 degrees. The second section on the trailing edge acted as an elevator.

Mounted in the fuselage between the foreplanes were four 30 mm Type 5 cannons, each with 60 rounds per gun. Each cannon was 7 ft 2 in (2.19 m) long and weighed 154 lb (70 kg). The cannons were slightly staggered to allow for clearance of their respective feed belts and keep the fuselage as narrow as possible. A compartment under the cannons collected the spent shell casings because of concerns that they would strike the propeller if they were ejected from the aircraft. Two 7.9 mm machine guns with 75 rounds per gun were planned for the very front of the nose and could be used for either training or target ranging. As ranging guns, they would help ensure that the cannon shells hit the intended target and not waste the limited ammunition supply. No armament was fitted to the prototype, and ballast weight was used to simulate the cannons.


The wheels under the vertical stabilizers were added after the aircraft’s first flight attempt ended with bent propeller blades. Note the long landing gear’s relatively short wheel base.

Behind the cannons was the single-seat cockpit, which was covered by a rearward-sliding glazed canopy. The pilot was protected by 2.76 in (70 mm) of armored glass in the front windscreen and a .63 in (16 mm) bulkhead by the cannons. Passageways ran on both sides of the aircraft between the cockpit and outer skin. Flight controls, hydraulic lines, and wiring ran in these passageways, which were accessible via removable outer skin panels. Under and slightly behind the cockpit was a 106-gallon (400-L) self-sealing fuel tank made of .87 in (22 mm) thick rubber.

Directly behind the cockpit was a 44-gallon (165-L) oil tank, followed by a Mitsubishi [Ha-43] 42 (IJN designation MK9D) engine. The [Ha-43] was a two-row, 18-cylinder, air-cooled engine. The [Ha-43] 42 had two-stage supercharging, with the first stage made up by a pair of transversely-mounted centrifugal impellers, one on each side of the engine. The shaft of these impellers was joined to the engine by a continuously variable coupling. The output from each of the first stage impellers joined together as they fed the second stage, two-speed supercharger mounted to the rear of the engine and geared to the crankshaft. As installed in the J7W1, the engine produced 2,030 hp (1,514 kW) at 2,900 rpm with 9.7 psi (.67 bar) of boost for takeoff. Military power at 2,800 rpm and 5.8 psi (.40 bar) of boost was 1,850 hp (1,380 kW) at 6,562 ft (2,000 m) in low gear and 1,660 hp (1,238 kW) at 27,559 ft (8,400 m) in high gear.


The prototype was unarmed, but four 30 mm cannons, each capable of firing 500 rounds per minute, were to be mounted in the nose. The projectile from each 30 mm shell weighed 12.3 oz / 5,401 grains (350 g).

The engine was mounted in the center of the fuselage and atop the wingbox. An extension shaft approximately 29.5 in (750 mm) long extended back from the engine to a remote propeller reduction gear box. The extension shaft passed through an extended housing that was mounted between the engine and the propeller gear reduction. The gear reduction turned the propeller at .412 times crankshaft speed and also drove a 12-blade cooling fan that was 2 ft 11 in (900 mm) in diameter. A screen was placed in front of the fan to prevent any debris from exiting the rear of the aircraft and hitting either the fan or propeller. Mounted to the propeller shaft was a 11 ft 2 in (3.40 m) diameter, metal, six-blade, constant-speed, VDM (Vereinigte Deutsche Metallwerke)-type propeller built by Sumitomo Metal Industries Ltd, Propeller Division (Sumitomo Kinzoku Kogyo KK, Puropera Seizosho). The propeller had approximately 29 in (740 mm) of ground clearance with the aircraft resting on all of its landing gear. If bailing out of the aircraft was needed, the pilot could detonate an explosive cord that would sever the propeller and gear reduction.

Cooling air for the [Ha-43] engine was taken in via an oblique inlet mounted on each side of the fuselage just behind the cockpit. Flaps at the inlet’s opening were raised to decrease the flow of cooling air to the engine. Cooling air entered the inlets, passed through the fins on the engine’s cylinders, traveled along the outside of the extension shaft housing, passed through the cooling fan, and exited around the spinner or an outlet under the rear of the aircraft. Two intakes, one on each side of the aircraft, were mounted to the cooling inlet. These intakes ducted induction air through the cooling air duct and directly into the transversely mounted superchargers.


The Mitsubishi [Ha-43] 42 engine installed in the J7W1 as seen post-war. The front of the aircraft is on the left. One of the two transversely-mounted, first-stage superchargers can be seen left of the engine. The oil cooler duct is in place and blocking the view of the extension shaft to the right of the engine. On the wing is the middle panel of the supercharger’s inlet scoop.

On each side of the fuselage directly behind the induction scoop was an inlet for an oil cooler. For each of the two oil coolers, after air passed through the cooler, it was mixed with the exhaust of four cylinders and ejected out a slit on the side of the fuselage just before the spinner. The ejector exhaust was used to help draw air through the oil coolers. The same philosophy applied to the exhaust from six cylinders on the bottom of the engine. These were ducted into an augmenter that helped draw cooling air through the cowling and out an outlet under the spinner. The exhaust from the remaining four cylinders, which were located on the top of the engine, exited via two outlets arranged atop the cowling to generate thrust.

The leading edge of the J7W1’s wing was swept back 20 degrees, and the trailing edge was swept back six degrees. The wings were mounted with no incidence angle. The inner wing from the wingbox to the rudder had 2.5 degrees of dihedral, and the outer wing from the rudder to the tip had zero dihedral. The structure of each wing was formed with three spars. The front spar ran along the wing’s leading edge. The center, main spar was swept back 14.5 degrees and ran in front of the main landing gear wells. A rear spar was swept forward 3.5 degrees and ran from the wingbox to just behind the main gear mount. A vertical stabilizer extended above and below the rear spar. The vertical stabilizer was mounted at approximately the midpoint of each wing and extended past the wing’s trailing edge. Initially, nothing was mounted under the vertical stabilizers, but a wheel was later added under each stabilizer to prevent propeller ground strikes. A rudder ran the entire 7 ft 3 in (2.20 m) height of each vertical stabilizer. Each wing housed a 53-gallon (200-L) fuel tank and a 20-gallon (75-L) anti-detonation fluid (water/methanol) tank for injection into the engine. Split flaps were positioned along the trailing edge of the wing between the vertical stabilizer and the fuselage. The flaps on the main wing extended 20 degrees. Two hardpoints under each outer wing could accommodate 66 or 132 lb (30 or 60 kg) bombs.


Rear view of the J7W1 showing its six-blade propeller and the engine’s 12-blade cooling fan in the rear of the cowling. The exhaust augmenter outlet can be seen on the bottom of the cowling. Note the rudders extending the entire height of the vertical stabilizers.

When deployed, the legs of the main gear were angled forward more than the nose gear. This effectively extended the nose gear and caused the aircraft to sit five-degrees nose-high while on the ground. This stance minimized the rotation needed to achieve liftoff, which is very important in the pusher aircraft. The main gear was mounted forward of the vertical stabilizers. The swiveling but non-steerable nose gear retracted forward, and the main gear retracted inward. Gear retraction and extension were powered hydraulically. At approximately 5 ft 11 in (1.8 m) long, the landing gear was quite tall to allow clearance for the propeller. The gear had a fairly wide track of 15 ft (4.56 m), but the wheelbase was short at only 10 ft 2 in (3.11 m). The short wheelbase combined with the tall gear legs and the aircraft’s high center of gravity could have given the J7W1 undesirable ground handling characteristics.

The J7W1 had a 36 ft 5 in (11.11 m) wingspan, was 32 ft (9.76 m) long, and was 12 ft 10 in (3.92 m) tall. The aircraft had a top speed of 466 mph (750 km/h) at 28,543 ft (8,700 m), a cruising speed of 276 mph (444 km/h), and a stalling speed of 107 mph (172 km/h). The J7W1 could climb to 26,247 ft (8,000 m) in 10 minutes and 40 seconds and had a 39,370 ft (12,000 m) service ceiling. The aircraft had an empty weight of 7,639 lb (3,465 kg), a normal weight of 10,864 lb (4,928 kg), and a maximum weight of 11,526 lb (5,228 kg). Cruising at 9,843 ft (3,000 m) gave the J7W1 a 528-mile (850-km) range. The aircraft was stressed for a maximum speed of 575 mph (926 km/h) and 7 Gs.


The various ducts on the side of the J7W1 are illustrated in this image. The flaps to reduce cooling air can be seen just before the oblique inlet on the side of the aircraft. The smaller scoop that fed air into the supercharger is mounted to the outside of the cooling air inlet. The oil cooler inlet can be seen just behind the tapered fairing for the induction scoop.

While the prototype was still under construction, the IJN ordered the J7W1 into production in May 1944 to counter the imminent threat of American bombing raids with the Boeing B-29 Superfortress. Ultimately, the production schedule called for Kyushu to produce 30 aircraft per month, and the Nakajima Aircraft Company, Ltd (Nakajima Hikoki KK) would build 120 units per month. In June 1944, the United States Army Air Force began conducting bombing raids against Japan using the B-29. To intercept these bombers and disrupt these raids were the exact purposes for which the J7W1 was designed. In September 1944, a mockup of the J7W1 was inspected by the IJN, and wind tunnel tests of a scale model had yielded positive results.

The J7W1 was built at Kyushu’s Zasshonokuma Plant, near Fukuoka city. The airframe was nearing completion in January 1945, when the first flight was originally scheduled to be conducted. Bombing raids delayed delivery of the [Ha-43] 42 engine, which finally arrived in April. The J7W1 was finally completed on 10 June and was subsequently disassembled and moved to Mushiroda Airfield (now Fukuoka Airport) in Fukuoka city on 15 June. Reassembled, the aircraft was inspected on 19 June, but bombing raids caused some delays. Ground tests were soon conducted and indicated a tendency for the engine to overheat due to a lack of cooling airflow. Tsuruno attempted the first flight in July, but as the J7W1 began to take flight, the engine’s torque induced a roll to the right. The aircraft’s nose went high and caused the propeller tips to strike the ground, bending the tips back.


Following World War II, the J7W1 was repaired and then painted before the aircraft was shipped to the United States. The new panels are easily seen in this image prior to the aircraft being repainted. Note that there is no cockpit glass.

The J7W1 was repaired, and the second prototype’s propeller was installed. A tailwheel from a Kyushu K11W Shiragiku (White Chrysanthemum) trainer was added under each vertical stabilizer so that during an over-rotation, a propeller strike would not occur again. Yoshitaka Miyaishi took over the flight tests and started over with ground runs to assess the aircraft’s handling. The J7W1 made its first flight on 3 August 1945. Liftoff occurred at 126 mph (204 km/h), and the aircraft was not flown above 1,312 ft (400 m). The speed did not exceed 161 mph (259 km/h), and the flight lasted under 15 minutes, with the aircraft landing at 115 mph (185 km/h). The J7W1’s tendency to roll to the right persisted and needed much left aileron input to correct, but the aircraft behaved reasonably well otherwise. Two further flights were made on 6 and 8 August, each about 15 minutes in length. The aircraft’s basic handling was evaluated, and the landing gear was never retracted during the tests. The roll to the right was made worse with the flaps deployed and the engine producing more torque to maintain airspeed. The J7W1 exhibited a tendency for its nose to pitch down, which was countered by a steady pull on the control stick. The engine, extension shaft, and remote gear reduction caused some vibration issues.

Modifications were contemplated to neutralize the engine’s torque reaction and correct the aircraft’s handling. A proposition was made to increase the foreplane’s angle of incidence to three degrees and change the main wing’s flap deployment to 30 degrees. In addition, the oil cooler needed to be improved. It was decided that speed tests would be initiated on the aircraft’s next flight, scheduled for 17 August. However, all work was stopped with the Japanese surrender on 15 August, and much of the aircraft’s documentation was burned on 16 August.


The J7W1 on display in Japan after it was repaired and painted. The inlet for the right oil cooler can be seen just behind the induction scoop, and the oil cooler’s exit can be seen right before the propeller. Note that the flaps are partially deployed.

At the end of the war, the second J7W1 was nearly complete and waiting on its [Ha-43] 42 engine, and the third aircraft was under construction. No other examples were completed to any meaningful level. The third J7W1 was planned to have the three-degree foreplane angle of incidence and a [Ha-43] 43 engine that produced an additional 130 hp (97 kW) for takeoff. This engine would have a single impeller for its first-stage, continuously-variable supercharger. The intake for the engine was moved to the inside of the J7W1’s cooling air inlets. The fourth and later aircraft would incorporate the changes from the third and also have a four-blade propeller 11 ft 6 in or 11 ft 10 in (3.5 m or 3.6 m) in diameter. The four-blade propeller had wider blades, was easier to manufacture, and was intended to cure some of the J7W1’s tendency to roll to the right. Beginning with the eighth aircraft, a 2,250 hp (1678 kW) [Ha-43] 51 engine would be installed. The [Ha-43] 51 had a single-stage, three-speed, mechanical supercharger instead of two-stage supercharging with a continuously-variable first stage.

The second and third J7W1 were both destroyed following the Japanese surrender. The first prototype, with around 45 minutes of flight time, was captured by US Marines and found to have all of the cockpit glass removed and some body panels damaged, possibly from a typhoon. For many years, it was thought that the first prototype was destroyed and that the second aircraft was captured by US forces, but this was later found to be incorrect. Under US orders, the aircraft was repaired and repainted while still in Japan. Most pictures of the J7W1 are immediately after the repairs have been made or shortly after it was painted. In almost all of the pictures, the cockpit glass is missing. In October 1945, the J7W1 was disassembled and shipped to the United States.


Six US Servicemen and four Japanese dignitaries pose next to the J7W1. Masayoshi Tsuruno, the aircraft’s designer, is the fourth from the left. The men give a good indication of the aircraft’s tall stance and overall size.

The surviving J7W1 was assigned ‘Foreign Evaluation’ FE-326 (later T2-326), and attempts were made to bring the aircraft to a flightworthy status. It is believed that most of this work, including new cockpit glass and installing several American flight instruments, was conducted in mid-1946 at Middletown Air Depot (now Harrisburg International Airport) in Pennsylvania. In September 1946, the aircraft was moved to the Orchard Field Airport (now O’Hare Airport) Special Depot in Park Ridge, Illinois. Instructions indicated that the J7W1 could be made airworthy if an overhauled engine was found, but this never occurred and the aircraft was not flown in the United States. The J7W1 was transferred to the Smithsonian National Air and Space Museum in 1960. The aircraft is preserved in a disassembled and unrestored state, with the [Ha-43] 42 engine still installed in the fuselage. Amazingly, video of the aircraft’s aborted first flight attempt and eventual first flight can be found on YouTube.

Around 2016, a full-size model of the J7W1 was built by Hitoshi Sakamoto. The model was on special display at the Yoichi Space Museum in Hokkaido, but it is not known if it is still there.

A turbojet version of the aircraft had been considered from the start, but a suitable powerplant had not been built in Japan by the close of the war. Designated J7W2 Shinden-Kai, the jet aircraft most likely would have had shorter landing gear, with additional fuel tanks in the wings occupying the space formerly used by the longer gear. There is no indication that the J7W2 had progressed beyond the preliminary design phase before the war’s end.


Today, J7W1 is disassembled but fairly complete. However, the years of storage have led to many bent and dented parts. The aircraft was long stored in the Smithsonian National Air and Space Museum’s Paul E. Garber facility, but the cockpit and foreplanes are on display at the Steven F. Udvar-Hazy Center in Chantilly, Virginia. (NASM image)

Zoukei-mura Concept Note SWS No. 1 J7W1 Imperial Japanese Navy Fighter Aircraft Shin Den by Hideyuki Shigete (2010)
Japanese Secret Projects by Edwin M. Dyer III (2009)
Japanese Aircraft of the Pacific War by René J. Francillon (1979/2000)
– “Kyushu Airplane Company” The United States Strategic Bombing Survey, Corporation Report No. XV (February 1947)
Encyclopedia of Japanese Aircraft 1900–1945 Vol. 4: Kawasaki by Tadashi Nozawa (1966)
The XPlanes of Imperial Japanese Army & Navy 1924–45 by Shigeru Nohara (1999)
War Prizes by Phil Butler (1994/1998)


McDonnell Aircraft Corporation XP-67 Fighter

By William Pearce

On 20 February 1940, the Army Air Corps (AAC) issued Request for Data R40-C that sought designs of new fighter aircraft capable of 450 mph (724 km/h), with 525 mph (845 km/h) listed as desirable. The AAC encouraged aircraft manufacturers to propose unconventional designs. The McDonnell Aircraft Corporation proposed four variants of its highly-streamlined Model 1 (often called Model I), the company’s first design. Each of the four Model 1 variants were powered by a different engine, and all the engines produced over 2,000 hp (1,491 kW). The Model 1’s engine was buried in the fuselage and drove wing-mounted pusher propellers via extensions shafts and right-angle gear boxes. Although not selected for R40-C, the AAC did purchase engineering data and a wind tunnel model of the design powered by an Allison V-4320 engine.


The McDonnell Model 2 as originally proposed was similar to the Model 1 but with Continental XI-1430 engines mounted under the wings. This configuration was found to create excessive drag.

McDonnell worked with the AAC to refine the Model 1 design and submitted the Model 2 (often called Model II) on 30 June 1940. The Model 2 had a crew of two, and two wing-mounted Continental XI-1430 engines replaced the single engine in the fuselage. The aircraft retained the basic shape of the Model 1’s fuselage and wings, but the engines were initially mounted directly under the wings in a tractor configuration. The engine mounting was changed as a result of wind tunnel tests. The new configuration was to mount the engine forward of the wing with a nacelle that housed a turbosupercharger extending back past the wing’s trailing edge. The nacelle was mounted mid-wing, and this design minimized drag. To further reduce drag, the Model 2 design was modified to incorporate fairings that blended the fuselage and engine nacelles to the wings. In addition, the design had the pilot as the sole occupant. The single-seat, blended design was called the Model 2A (often called Model IIA), and it was submitted to the AAC on 24 April 1941.

On 5 May 1941, McDonnell submitted preliminary specifications of the Model 2A to the AAC. Under these specifications, the aircraft had a wingspan of 55 ft (16.8 m), a length of 42 ft 3 in (12.9 m), and a height of 14 ft 9 in (4.5 m). The Model 2A had a calculated speed of 500 mph (805 km/h) at 35,000 ft (10,668 km), 472 mph (760 km/h) at 25,000 ft (7,620 m), and 384 mph (618 km) at 5,000 ft (1,524 m). The aircraft would climb to 25,000 ft (7,620 m) in 9 minutes and have a service ceiling of 41,500 ft (12,649 m). At a cruising speed of 316 mph (509 km/h), maximum range was 2,400 miles (3,862 km) with 760 gallons (2,877 L) of internal fuel. The Model 2A had an empty weight of 13,953 lb (6,329 kg), a gross weight of 18,600 lb (8,437 kg), and a maximum weight of 21,480 lb (9,743 kg).


The Model 2 was revised with the engines mounted forward of the wings with streamlined nacelles mounted mid-wing. This produced a more attractive aircraft, very similar to the Model 1. However, the relation to the XP-67 is clear.

McDonnell continued to work with the AAC to refine the design of the Model 2A. On 30 September 1941, the Army Air Force (AAF—the AAC was renamed in June 1941) issued a contract to McDonnell to build two prototypes of the Model 2A interceptor pursuit fighter as the XP-67. The aircraft was assigned Materiel Experimental code MX-127. The first aircraft was scheduled to be delivered on 29 April 1943, with the second example delivered six months later on 29 October 1943. The XP-67 had a fairly conventional layout for a single-seat, twin-engine aircraft with tricycle undercarriage. What was not conventional was the extensive blending of the fuselage and engines nacelles to the aircraft’s wings to maintain true airfoil sections throughout the entire aircraft. The end result was a streamlined appearance.

The XP-67 was constructed of an aluminum frame with aluminum skin that formed a monocoque structure. All control surfaces consisted of a fabric covered aluminum frame, although aluminum skinning was later proposed for production aircraft. Effort was expended to keep the XP-67’s surface smooth and make everything flush. Initially, a door on the left side of the pressurized cockpit was to allow access. However, pressurization was dropped on the prototype, and a glazed, rearward-sliding canopy was used.

The wings had two spars, a dihedral of five degrees, and consisted of inner and outer wing sections. The outer wing section extended from the engine nacelle and was removable. Split flaps were located between the nacelle and fuselage. A small split flap existed on the outer side of the engine nacelle. The outer wing section’s trailing edge was occupied by an aileron. The ailerons drooped 15 degrees with deployment of the flaps, which had a maximum deployment of 45 degrees. However, it does not appear that the drooping ailerons were ever installed on the prototype. No hardpoints existed under the wings for bombs or drop tanks.


The Model 2A as originally proposed in May 1941 was essentially the latest Model 2 design but with large fairings that blended the fuselage and engine nacelles to the wing. This design was contracted as the XP-67.

Mounted to each wing was a liquid-cooled, Continental XI-1430 inverted V-12 engine. Initially, clockwise-rotating (right-handed) XI-1430-1 engines were to be used. In June 1942, the engines were switched to an XI-1430-17 installed on the right wing (clockwise, right-handed rotation) and an XI-1430-19 installed on the left wing (counterclockwise, left-handed rotation). Each engine of the first prototype turned a cuffed, four-blade Curtiss Electric constant-speed propeller that was 10 ft 6 in (3.2 m) in diameter. However, the cuffs were installed after the first aircraft was completed. In April 1943, McDonnell proposed installing Curtiss Electric contra-rotating propellers on the second XP-67 prototype, noting that such a change would increase the aircraft’s speed by 7–10 mph (11–16 km/h) and climb rate by 400 fpm (2.0 m/s).

The engine nacelle extended back from each engine and housed a General Electric D-23 turbosupercharger. Engine exhaust was directed straight back from the nacelle to gain some thrust. Initially, it was proposed that each engine would have a coolant radiator located in the fuselage. This was changed to each engine having two coolant radiators housed in the engine nacelle and located directly under the rear of the engine. The engine nacelles were blended into the wing, and several intakes were incorporated into the wing’s leading edge. For both engine nacelles, the intakes closest to the nacelle passed air to a cooling jacket around the exhaust manifold. The center intake directed air through the two coolant radiators per engine and to the turbosupercharger. The intakes farthest from the engine each led to an oil cooler.

An oil tank in each wing held 26 gallons (98 L) for each engine. The aircraft’s normal fuel load was 282 gallons (1,067 L), but 478 gallons (1,809 L) of additional fuel could be housed in the aircraft’s four fuel tanks located in the fuselage and wing. This brought the XP-67’s total fuel capacity to 760 gallons (2,877 L). The aircraft’s tricycle landing gear was hydraulically-powered and fully retractable. The nose wheel was swiveled, but was not steerable, and folded back into the fuselage. The main gear was mounted just inboard of the engine nacelles and folded inward. In early 1942, the AAF requested that the main gear fold into the engine nacelle, necessitating a complete redesign of the nacelles to accommodate the rearward retracting main wheels. The horizontal stabilizer had 9.55 degrees of dihedral and was mid-mounted to the aircraft’s vertical stabilizer. Like the outer wing panels, the tail was detachable for transporting the aircraft by ground. The XP-67 airframe was stressed for +8 and -4 Gs and had a diving limit of 604 mph (972 km/h) indicated.


An XI-1430-17 with a GE D-23 turbosupercharger installed in the McDonnell XP-67 wing section for tests at the Langley Memorial Aeronautical Laboratory in September 1943. The tests were conducted to evaluate the cooling ducts of the XP-67’s radical blended design. The top image illustrates the unusual ducting of the XP-67’s nacelles, which were duplicated on the opposite side. Closest to the spinner is the exhaust manifold cooling air duct. The large middle duct was for the coolant radiator and engine intake. The outer duct was for the oil cooler. The bottom image shows the turbosupercharger, which was installed so that the exhaust provided additional thrust. Note the radiator cooling air exit duct on the landing gear door and the cuffed propellers. (LMAL images)

The XP-67’s armament changed as the aircraft was developed. Initially, the aircraft would have four 20 mm cannons with 166 rounds per gun and six .50-cal machine guns with 500 rounds per gun. The cannons would be installed on the sides of the cockpit, just behind the pilot. The machine guns were to be installed just behind the cannons. On 5 August 1941, the AAF requested that two 37 mm cannons be installed in place of two 20 mm cannons. By 16 August, the armament was revised again to six 37 mm cannons with 45 rounds per gun and no other guns. The 37 mm cannons were installed in the blended-wing’s leading edge between the cockpit and engine nacelle. The three cannons on each side of the fuselage were outside of the propeller arc. On 20 October, it was suggested that the aircraft’s design should incorporate provisions to replace four of the 37 mm cannons with four 20 mm cannons. On 8 November, it was decided that the first aircraft would have six 37 mm cannons, and the second aircraft would have two 37 mm and four 20 mm cannons.

Extensive wind tunnel tests were conducted on various XP-67 models throughout 1942 and 1943. These tests led to many minor changes in the aircraft. Much of this testing was focused on the extensive fairing used to blend the wing and fuselage. The cooling system was also carefully scrutinized with many minor changes taking place to the cooling ducts. A full-size mockup of the XP-67 was inspected in mid-April 1942, which led to more changes. The most significant changes were lengthening the aircraft’s nose by 15 in (381 mm) and changing the flight control actuation system from push-pull rods to cables. In May, a fuselage section was built to test fire the 37 mm cannons. The tests proved satisfactory, but McDonnell redesigned the 37 mm cannon installation in October, necessitating another mockup and more tests. The new 37mm cannon installation mockup successfully passed its tests in March 1944, but the armament was never installed in the prototype. On 17 June 1942, the decision was made to finish the prototype without a pressurized cockpit. In April 1943, there were discussions of cancelling the XP-67, but the aircraft was seen as a good way to test the experimental wing blending, cannon armament, and XI-1430 engines.


The McDonnell XP-67 nearly complete in mid-November 1943. Even though the nacelle’s duct design was found to be insufficient in the wind tunnel tests, the aircraft was not modified with a new design until later. Note the covered ports for 37 mm cannons on each side of the cockpit and that the propellers do not have their cuffs installed.

McDonnell had built a full-scale XP-67 engine nacelle for testing the XI-1430 engine installation. Tests were conducted by McDonnell starting in May 1943. After accumulating almost 27 hours of operation, the rig was sent to the National Advisory Committee for Aeronautics (NACA) at the Langley Memorial Aeronautical Laboratory (LMAL, now Langley Research Center) in Virginia. The NACA added about 17.5 hours to the engine conducting tests in August and September to analyze the installation’s effectiveness for cooling the coolant, oil, and intercooler. The tests indicated that the cooling system was insufficient. The nacelle with revised ducts was then shipped to Wright Field in Dayton, Ohio in October 1943. Wright field added another 6.5 hours to the engine, bringing the total to 51 hours. The new ducts proved satisfactory, reducing the drag of the ducts by 25 percent and improving cooling by 200 percent. However, excessive vibrations occurred between the engine and its mounting structure, necessitating a more rigid mount. McDonnell was allowed to proceeded with testing the first XP-67, although the prototype would not be changed until after its first flight when additional changes beyond the cooling system would most likely need to take place. Wind tunnel tests had indicated that the horizontal stabilizer would need to be raised by 12 in (305 mm) to improve stability. McDonnell was instructed to stop work on the second prototype until successful flight tests of the first aircraft had been conducted.

Serial number 42-11677 was given to the first XP-67, and serial number 42-11678 was given to the second prototype. Unofficially, the XP-67 was given the name ‘Moonbat’ or just ‘Bat,’ but it does not appear that an official name was ever bestowed upon the aircraft. With all the design changes since the XP-67 was initially contracted, the aircraft’s specifications had changed. The wingspan remained at 55 ft (16.8 m), but the length increased 2 ft 6 in (.8 m) to 44 ft 9 in (13.6 m), and the height increased 1 ft (.3 m) to 15 ft 9 in (4.8 m). The standard fuel load remained at 280 gallons (1,060 L), but the additional fuel load decreased by 25 gallons (95 L) to 455 gallons (1,722 L), giving a total maximum internal fuel load of 735 gallons (2,782 L). The XP-67’s weight had increased by 3,792 lb (1,720 kg), resulting in an empty weight of 17,745 lb (8,049 kg), a gross weight of 22,114 lb (10,031 kg), and a maximum weight of 24,836 lb (11,265 kg). A reduction in performance accompanied the weight increase, resulting in an estimated speed of 448 mph (720 km/h) at 25,000 ft (7,620 m), which was a 24 mph (39 km/h) reduction, and 367 mph (591 km/h) at sea level. The time to climb to 25,000 ft (7,620 m) was increased by nearly five minutes to 14.8 minutes, and the service ceiling decreased 4,100 ft (1,250 m) to 37,400 ft (11,400 m). The XP-67’s cruising speed decreased 46 mph (74 km/h) to 270 mph (435 km/h), but maximum range was little changed at 2,385 miles (3,838 km) with 735 gallons (2,782 L) of fuel.


The completed XP-67 with revised nacelle cooling ducts and after the horizontal stabilizer was raised 12 in (305 mm). The most noticeable duct modification was to the exhaust manifold cooling intake, which was changed to a scoop. Note that the propellers rotated in opposite directions.

On 1 December 1943, the XP-67 had its XI-1430 engines installed and was ready for ground tests. However, both engines caught fire and damaged the aircraft on 8 December. The fires were caused by issues with the exhaust manifolds. The XP-67 was repaired and made its first flight on 6 January 1944, taking off from Scott Field in Belleville, Illinois. The flight was nearly two years later than the anticipated first flight when the XP-67 contract was originally issued. Test pilot Ed E. Elliott had to cut the flight to just six minutes due to both turbosuperchargers overheating, which resulted in small fires. During the short flight, the XP-67 exhibited good handling characteristics.

The aircraft was again repaired, with the second and third flights occurring on 26 and 28 January 1944. On 1 February, the aircraft’s fourth flight was cut short due to a main bearing failure on the left engine caused by an unintentional overspeed of the engine. The cockpit canopy also detached during the flight. While the XP-67 was down for repairs and new XI-1430 engines, the horizontal stabilizer was raised 12 in (305 mm). The cooling ducts in the engine nacelles were also modified, with the most noticeable being the exhaust shroud inlet, which was changed to more of a scoop. The updated aircraft flew again on 23 March 1944 and demonstrated improved stability, but one turbosupercharger failed at 10,000 ft (3,048 m).

In April 1944, it was reported that the engines were running too cool. The closed main gear door formed part of the air duct aft of the radiator. However, the gear doors did not seal tightly and caused an excessive amount of air to exit the duct. This resulted in too much air passing through the radiator and reducing the engine temperature below ideal levels. McDonnell was allowed to install a thermostat on the prototype to help control coolant temperatures but was also told that such issues would not be acceptable on production aircraft. Around this same time, construction of the second prototype was allowed to proceed with the exception of parts that would be affected by an engine change.


The unusual planform of the XP-67 is illustrated in this view. The two ports in the middle of each nacelle were the forward exit for the exhaust manifold cooling air. The rear exit is denoted by the white staining at the end of the nacelle. The outer wing section was detachable just outside of the nacelle.

In May 1944, three AAF pilots flew the XP-67 and reported that the XI-1430 engines ran rough and seemed underpowered. Tests indicated that at normal power, the engines were only delivering 1,060 hp (790 kW), well below the expected 1,350 hp (1,007 kW). The XP-67 was noted for having high control forces at high speeds, exhibiting a Dutch roll indicating some directional instability, and not making a good gun platform. The maximum speed with the engines delivering 1,600 hp (1,193 kW) at 3,200 rpm was 357 mph (574 km/h) at 10,000 ft (3,048 m) and 393 mph (632 km/h) at 20,000 ft (6,096 m). From these values and other tests, McDonnell calculated that the XP-67 could attain 405 mph (652 km/h) at 25,000 ft (7,620 m) at the same power setting. Takeoff speed was 130 mph (209 km/h); the clean stall speed was 118 mph (190 km/h) with buffeting starting at 140 mph (225 km/h); and the aircraft had a high landing speed of 120 mph (193 km/h). In general, the XP-67 was found to be inferior to other fighters currently in production.

McDonnell got permission to install contra-rotating propellers on the first prototype when the engines were ready, and they were expected in June 1944. No information has been found indicating that the contra-rotating versions of the XI-1430 were delivered. In June, it was decided to install 11 ft (3.4 m) diameter four-blade Aeroproducts propellers rather than contra-rotating propellers. However, tests would continue with the Curtiss propellers until the Aeroproducts were ready. It was also noted that the XP-67 had experienced no engine fires since its fourth flight, and the aircraft had completed about 50 flights without any serious issues.


The limited flight trials of the XP-67 indicated the aircraft handled fairly well. It was noted as underpowered and slightly unstable. Overall, visibility was said to be poor, with the engine and fairing blocking most of the view to the side and rear. Formation flying would have been difficult, as the pilot was unable to see their wingtips.

In July 1944, some in the AAF felt that the XP-67 program was expensive and served no purpose. However, others felt that the aircraft was a unique platform that would allow the testing of the six 37 mm cannons. In addition, the possibility existed to install 12 .50-cal machine guns or eight 20 mm cannons. The aircraft was seen as a good test machine, even if its performance fell below what was originally specified. It was decided to complete tests on the current aircraft to assess the blended design and then consider the possibility of armament trials.

McDonnell had long sought to change the aircraft’s engines. On 19 January 1944, McDonnell proposed discarding the XI-1430s for the second prototype and using either two-stage Allison V-1710 or Rolls-Royce Merlin RM 14SM (100-series prototype) piston engines. In addition, each engine nacelle would house a Westinghouse 9.5 (J32) turbojet behind the piston engine. The mixed-power proposal was brought up again on 16 March 1944, now using an Allison V-1710-199 (F32R) piston engine and either a Rolls-Royce W2B/37 turbojet or a GE I-20 (J39) turbojet in the nacelle. With mixed power plants, the aircraft had an estimated top speed at sea level of 500 mph (805 km/h). The engine issue was discussed again in July 1944, with McDonnell now suggesting a Rolls-Royce Merlin RM 14SM piston engine paired with a GE I-20 (J39) turbojet in each nacelle. However, AAF felt that the aircraft would need a complete redesign to incorporate different piston engines with turbojets.

Since its initial design in May 1941, there were suggestions of using a modified version of the XP-67 for photo reconnaissance. In April 1942, McDonnell suggested that the aircraft’s range could be extended to 4,000 miles (6,437 km) at a cruise speed of 200 mph (322 km/h), which would be a 20-hour flight. For this, two of the 37 mm cannons would need to be omitted and six additional fuel tanks installed along with 280 lb (127 kg) of ballast in the nose. With the extra tanks, the aircraft’s internal fuel capacity was 1,290 gallons (4,883 L). This concept was not pursued at the time, but the range extension was considered later for a photo-recon role.


A model of the XP-67E with its bubble canopy and mixed piston / turbojet power plants. It is not clear what engines (if any) are intended to be depicted by the model, but the nacelles were extended back to house the jet engine (LMAL image).

By July 1944, it was believed that a photo-recon version of the XP-67 would have inferior performance compared to the Lockheed F-5 (P-38). However, a mixed-power version of the aircraft was seen as a possible candidate as a photo-recon aircraft. The XP-67E was designed for the photo-recon role, and it incorporated mixed power, additional internal fuel tanks, and provisions for two 150-gallon (568-L) drop tanks mounted under the aircraft’s center section. In the XP-67E design, the engine nacelles were extended back to house the GE I-20 (J39) turbojet engine. Cameras were installed in the aft fuselage, and the XP-67E was unarmed. The fuselage was mostly unchanged, but the cockpit was enclosed in a rearward-sliding bubble canopy.

The XP-67 prototype had been undergoing modifications and repairs through August 1944. Perhaps the most major change was alerting the wing dihedral from 5 degrees to 7 degrees in an attempt to increase stability. The aircraft was ready to resume flight tests in early September. On 6 September 1944, the exhaust valve rocker of the No. 1 cylinder in the XP-67’s right engine broke while the aircraft was in flight at 10,000 ft (3,048 m). Exhaust gases unable to escape the cylinder backed up into the intake manifold and caused it to fail, resulting in a fire. The fire was first noticed at 3,000 ft (914 m) as the aircraft was preparing to land. Test pilot Elliott was able to land the XP-67 and stopped it to limit the flames from spreading. However, the brake failed after Elliott exited the aircraft, and wind turned the XP-67 so that the flames blew toward the fuselage. The XP-67 was nearly burned in half and damaged beyond repair. The aircraft had a total flight time of 43 hours. This event effectively killed the XP-67 project and the XP-67E photo-recon proposal. The entire program was suspended seven days later on 13 September, and on 24 October, McDonnell was notified that the XP-67 contract was cancelled. A formal Notice of Cancellation followed on 27 October 1944. The second prototype was about 15 percent complete and was subsequently scrapped. The total cost of the XP-67 program was approximately $4,733,476.92.


The XP-67 after the fire on 6 September 1944. Once on the ground, the fire from the right engine spread to the rear fuselage and left nacelle. The rear fuselage was nearly burned through and collapsed to the ground. An inglorious end to both the XP-67 and XI-1430 programs.

Interceptor Pursuit Airplane Twin Engine Type XP Preliminary Specifications by McDonnell Aircraft Corporation (5 May 1941)
Memorandum Report on XP-67 Airplane, AAF No, 42-11677 by Osmond J. Ritland (19 May 1944)
Final Report on the XP-67 Airplane by John F Aldridge Jr. (31 January 1946)
Case History of XP-67 Airplane by Historical Division, Air Materiel Command (23 July 1946)
USAF Fighters of World War Two Volume Three by Michael O’Leary (1986)
U.S. Experimental & Prototype Aircraft Projects: Fighters 1939-1945 by Bill Norton (2008)