The aircraft of the 1920s were anything but streamlined by modern standards. Landing gears were not retractable, and the engines themselves, particularly the finned cylinders, were largely exposed for better air cooling. The large drag penalties of these awkward protuberances were unappreciated at first. The actual magnitude of this problem was uncovered almost accidentally.
In 1917 William F. Durand had published one the first NACA Reports (Technical Report 17) describing his propeller research in a 5.5-foot wind....
....tunnel at Stanford University in California. When NACA later tried to correlate Durand's wind tunnel data with the results of its own flight tests, large discrepancies appeared. Why did the isolated propeller tests at Stanford disagree with Langley tests of propeller-plus-fuselage ?
To resolve this problem, George Lewis, NACA Director of Research, decided to build a special wind tunnel for propeller research. He proposed that it be big enough to test the actual fuselages and their engines along with the propellers. This was a radical proposal because it meant going from the customary tunnels 5 feet in diameter to one on the order of 20 feet in diameter. Nevertheless, the propeller/ fuselage puzzle had to be resolved. Design of the new tunnel commenced in the spring of 1925. The new Propeller Research Tunnel (PRT) went into operation in July 1927. It was a giant, with two 1000-horsepower diesel submarine engines (courtesy of the U.S. Navy) turning a 27-foot, 8- bladed propeller. The test section air velocity was only 110 mph, but, with a 20-foot stream of air to play with, the entire fuselage with operating engine and propeller could be tested. The results of research with the PRT were far reaching and, in one instance, most surprising to the experts.
The PRT demonstrated almost at once that exposed landing gears contributed up to 40 percent of fuselage drag. Engineers quickly went to work and designed retractable landing gears-surely a simple solution, but one that did not come about until the  real magnitude of the landing gear drag penalty was appreciated. Second, the PRT demonstrated that multiengined aircraft perform best when the engines and their nacelles are in line with the wing- chord plane. These findings did much to shape the DC-3 transport and the B-17 and B-24 bombers of World War II.
Most startling of all was the discovery that the protruding cylinders typical of the air-cooled engines of the 1920s contributed almost one-third the drag of the entire fuselage. Aircraft designers had let the cylinders and their cooling fins stick out in the airstream for maximum cooling, but now it was apparent that the drag penalty was too high. The cylinders had to be covered with a streamlined cowling.
After a systematic study by Fred Weick of many engine cowlings in the PRT, the famous NACA Cowl was born in late 1928, less than 18 months after the new tunnel was placed in operation. Not only did the cowl reduce drag dramatically, but engine cooling was improved as well, confounding that engineering intuition that had stuck the finned cylinders directly in the external airstream.