In the 1930s, the Royal Air Force was interested primarily in twin-engine bombers. These designs put limited demands on engine production and maintenance, both of which were already stretched with the introduction of so many new types into service. Power limitations were so serious that the British invested heavily in the development of huge engines in the 2,000 horsepower (1,500 kW) class in order to improve performance. In the late 1930s, none were ready for production. The U.S. and USSR were developing bombers with four smaller engines, which proved to have excellent range and fair lifting capacity, so in 1936 the RAF also decided to investigate the feasibility of the four-engined bomber.
The Air Ministry Specification B.12/36 had several requirements. The bomb load was to be a maximum of 14,000 lb (6,350 kg) carried to a range of 2,000 miles (3218 km) or a lesser payload of 8,000 lb (3,629 kg) to 3,000 miles (4,800 km) (incredibly demanding for the era). It had to cruise at 230 or more mph at 15,000 ft (4,600 m) and have three gun turrets (in nose, amidships and rear) for defence. The aircraft should also be able to be used as a troop transport for 24 soldiers, and be able to use catapult assistance for take off. The idea was that it would fly troops to far corners of the British Empire and then support them with bombing. To help with this task as well as ease production, it needed to be able to be broken down into parts, for transport by train. Since it could be operating from limited "back country" airfields, it needed to lift off from a 500 ft (150 m) runway and be able to clear 50 ft (15 m) trees at the end, a specification most small aircraft would have a problem with today.
Initially left out of those asked to tender designs, Shorts were included because they already had similar designs in hand and they had ample design staff and production facilities. Shorts were producing several four-engined flying boat designs of the required size and created their S.29 by removing the lower deck and boat hull of the S.25 Sunderland. The new S.29 design was largely identical otherwise: the wings and controls were the same, construction was identical and it even retained the slight upward bend at the rear of the fuselage, originally intended to keep the Sunderland's tail clear of sea spray.
In October 1936, the S.29 was low down on the short list of designs considered and the Supermarine Type 317 was ordered in prototype form in January 1937. However it was decided that an alternative design to Supermarine was needed for insurance and that Shorts should build it as they had experience with four-engined aircraft. The original design had been criticized when considered and in February 1937 the Air Ministry suggested modifications to the original Short design, including considering the use of the Bristol Hercules radial engine as an alternative to the Napier Dagger inline, increasing service ceiling (28,000 ft) and reducing the wingspan. Shorts accepted this large amount of redesign. The project had added importance due to the death of Supermarine's designer, Reginald Mitchell, causing doubt in the Air Ministry. The S.29 used the Sunderland's 114 ft (35 m) wing and it had to be reduced to less than 100 ft (30 m), the same limit as that imposed on the P.13/36 designs (Handley Page Halifax and Avro Manchester). In order to get the needed lift from a shorter span and excess weight, the redesigned wing was thickened and reshaped. It is often said that the wingspan was limited to 100 ft so the aircraft would fit into existing hangars but the maximum hangar opening was 112 ft (34 m) and the specification required outdoor servicing. "The wing span was limited by the Air Ministry to 100 ft" The limitation was actually to force the designer to keep overall weight down.
In June 1937 the S.29 was accepted as the second string for the Supermarine 316 and formally ordered in October.
Shorts built a half scale version as the S.31 (also known internally as the M4 – the title on the tailfin), powered by four Pobjoy Niagara engines, which first flew on 19 September 1938, piloted by Shorts' Chief Test Pilot J. Lankester Parker. Everyone was happy with the design, except that the take off run was thought to be too long. Fixing this required that the angle of the wing to be increased for take off. If the wing itself was modified, the aircraft would be flying nose down while cruising (as in the Armstrong Whitworth Whitley). Shorts lengthened the undercarriage struts to tilt the nose up on take-off, leading to its spindly gear which in turn contributed to many take off and landing accidents. The Short S.31 was scrapped after a take off accident at RAF Stradishall, Suffolk in February 1944.
The first S.29, now given the service name "Stirling" after the Scottish city, flew on 14 May 1939 with four Bristol Hercules II radial engines. Upon landing one of the brakes locked, causing it to slew off the runway and collapse the landing gear. A redesign added much stronger and heavier struts on the second prototype. On its first sortie two months later, one of the engines failed on take off but the aircraft landed easily. From then on, the record improved and service production started in August 1940 at Shorts' Rochester factory. The area, which included a number of major aviation firms, was heavily bombed in the opening days of the Battle of Britain, including one famous low-level raid by a group of Dornier Do 17s. A number of completed Stirlings were destroyed on the ground and the factories were heavily damaged, setting back production by almost a year. Some production was moved to Austin Aero's factory at Cofton Hackett just south of Birmingham and the factory there eventually produced nearly 150 Stirlings. From this point on, the Belfast factory became increasingly important as it was thought to be well beyond the range of German bombers. However, Belfast and the aircraft factory were subjected to German aircraft bombing during Easter week of 1941. To meet the increased requirement for its aircraft during the war, satellite factories near Belfast were operated at Aldergrove and Maghaberry, producing 232 Stirlings between them. In 1940, bombing damaged Supermarine's factory at Woolston and the incomplete Type 316 prototypes. The 316 was cancelled in November 1940 leaving the Stirling as the only B.12/36 design.
Although smaller than the US and Soviet experimental designs, the Stirling had considerably more power and far better payload/range than anything then flying. The massive 14,000 lb (6.25 long tons, 6,340 kg) bomb load put it in a class of its own, double that of any other bomber. It was larger than the Handley Page Halifax and the Avro Lancaster which replaced it but both of these were originally designed to have twin engines. The Stirling was the only British bomber of the period to see service that had been designed from the start with four engines; the Avro Lancaster was a re-engined Avro Manchester while the Halifax was planned to be powered by twin Vulture engines but was re-designed to use four Merlins in 1937, as the problems with the Vulture engines became clear.
The design had nose and tail turrets (the latter was notable for the wide angles of fire) and included a retractable ventral ("dustbin") turret just behind the bomb-bay. This proved almost useless due to cramped conditions, with the added distraction that the turret tended to drop and hit the ground when taxiing over bumps. It was removed almost from the start and temporarily replaced by beam hatches mounting pairs of machine guns, until a twin-gun dorsal turret could be provided. This turret also had problems; it had a metal back fitted with an escape hatch which turned out to be almost impossible to use. The later Stirling Mk.III used a fully glazed turret (the same FN.50 as in Lancaster) that had more room and an improved view. Later Stirlings could also carry an improved, low-drag remotely controlled FN.64 ventral turret.
Attention was paid to reducing drag – all rivets were flush headed and panels joggled to avoid edges – but camouflage paint probably negated the benefit. The wing was fitted with Gouge flaps similar to those of the flying boats.
The first few Mk.Is had Hercules II engines but the majority had 1,500 hp (1,100 kW) Hercules XIs. The Mk.III, introduced in 1943, was similar with the exception of the new dorsal turret and the improved 1,635 hp (1,200 kW) Hercules VI or XVI engines, which improved maximum speed from 255 to 270 mph (410 to 435 km/h).
Even before the Stirling went into production, Short had improved on the initial design with the S.34 in an effort to meet specification B.1/39. It would have been powered by four Bristol Hercules 17 SM engines, optimised for high-altitude flight. The new design featured longer span wings and a revised fuselage able to carry dorsal and ventral power-operated turrets each fitted with four 20 mm Hispano cannons; despite the obvious gains in performance and capability, the Air Ministry was not interested.
In 1941, Short proposed an improved version of the Stirling, optimistically called "The Super Stirling" in the company's annals. This Stirling would feature a wing span of 135 ft 9 in (41.38 m) and four Bristol Centaurus radials and a maximum takeoff weight of 104,000 lb (47,174 kg). The performance estimates included a 300 mph (483 km/h) speed and a 4,000 mile (6,437 km) range with a weapons load of 10,000 lb (4,536 kg) over 2,300 miles or 23,500 lb over 1,000 miles. Defensive armament was 10 0.5 inch machine guns in three turrets. It was initially accepted for under Specification B.8/41 (written to cover it) and two prototypes were ordered but the C-in-C of Bomber Command Arthur Harris felt that, while it would be a better aircraft, actual production would be slower and that effort would be better spent on giving the Stirling improved Hercules engines for a higher ceiling. Shorts were told in May 1942 that the Air Ministry would not be continuing the project and in August Shorts decided to terminate work.