I recently picked up a copy of the 1909 to 1971
Racing Planes and Air Races series by Reed Kinert and it includes a few firsts. First of the firsts, is a solution to the confusion over "first airplane with a ducted spinner" that I mentioned in a
previous post. The answer comes by way of pushing the date back two years before all the other contestants. In September 1937, the Wittman D-12 Bonzo featured just such an innovation. (Source: Reed Kinert,
Racing Planes and Air Races: A Complete History, First, vol. 3, 8 vols. (Fallbrook, California: Aero Publishers, Inc., 1969) 71.)
It apparently started with an earlier version...

(Source:
The Dreamy Dodo)
...that later morphed into the 1937 racer.

(Source:
Free Flight Archive)
The book also claims that the Hughes H-1 was "the first aircraft with wing leading edge air intake ducts" and that the "long wing" specifically - as opposed to the shorter "speed wing" - "was the first to have power-driven retractable landing gear". (Source: Reed Kinert,
Racing Planes and Air Races: A Complete History, First, vol. 3, 8 vols. (Fallbrook, California: Aero Publishers, Inc., 1969) 75, 77.)

(Source:
Wikimedia Commons)
Then, as if perfectly placed to draw a connection between the D-12 and the modified A-17A, Fw 190 V1, and Me 209 V4, it also includes the following interesting claim:
Reed Kinert wrote:
And so ended another great era of air racing, cut short by World War II. During the 1920-1925 race era, commercial aircraft had copied the innovations introduced in military racers, then, after the 1926-29 racing "doldrums," the military designers copied unabashedly from the back-yard-built craft. The Curtiss Hawk model 75, a monoplane built in 1936 and exported to France early in the war, had the exact profile of the earlier Wedell-Williams. Later fitted with retractable gear, the Hawk became the Army P-36, and when fitted with an Allison V-12 engine it became the famous P-40. Designer Keith Rider's racers of 1932 featured a retractable landing gear that left enough of the wheels protruding to allow safe emergency landing. Seversky copied this idea into his P-35 pursuit; Boeing used it on their early 247-D airliners and on the B-17; and Douglas featured it on the famous DC-3.
(Source: Reed Kinert,
Racing Planes and Air Races: A Complete History, First, vol. 3, 8 vols. (Fallbrook, California: Aero Publishers, Inc., 1969) 87.)
I couldn't find a photo of the Rider R-1 with its gear retracted, so I had to settle for this model blueprint:

(Source:
Outerzone)
EDIT (21-04-27): By sheer coincidence, while reading through a different book today, I came across a statement today that was so relevant I just had to include it. It makes the exact counterpoint to the claim about the relevance of air racing to the development of military aircraft:
S. Paul Johnson wrote:
An argument is sometimes advanced that airplane racing, like horse racing, tends to "improve the breed.” Experience has indicated, however, that we get farther faster by depending on well planned, day-in-and-day-out scientific research than by looking for spectacular advances out of racing machines. The Navy came early to this conclusion. Its participation in air racing has been on a modest scale. In 1923 a Navy racer was built by the Curtiss Company for the Schneider Speed Race (an international contest established in 1913, finally abandoned in 1931). That year Lieutenant David Rittenhouse showed his heels to the pack with an average speed of just over 177 miles an hour. There was no contest in 1924, but in the 1925 Schneider Race, America won again. The machine was a Curtiss R-3 seaplane racer. It was flown by James H. Doolittle at an average speed of 232.57 miles per hour. After two straight wins, the U.S. Government officially gave up Schneider racing. It cost too much. The return was too small.
(Source: S. Paul Johnson,
Flying Fleets: A Graphic History of U.S. Naval Aviation (New York: Duell, Sloan and Pearce, 1941), 57-58.)
EDIT (21-10-17): Although it's not clear that it was entirely an intentional feature of the design, the other day I came across an advertisement that pushes the idea of partially extended wheels further back. In 1919, the Martin Aeroplane Factory (apparently unrelated to the company founded by Glenn L. Martin) in Elyria, Ohio was offering "[e]ngineering data and license" for a retractable landing gear system that appeared to result in protruding wheels:
Attachment:
Aerial Age 3 Febuary 1919 Page Second Cover.png [ 969.99 KiB | Viewed 3434 times ]
(Source:
Aerial Age via Internet Archive)
EDIT: (22-10-01): The other day I came across an example of what I thought was going to be an earlier instance of semi-retractable wheels than the edit above. However, it was only just now, when I went to write this edit, that I realized it was actually the same individual mentioned above! (The
source material even includes the same disclaimer about not being confused for the more famous Martin that I did!) Anyway, without further ado, the aircraft was the Martin K-III Kitten, designed by J.V. Martin. It even still exists in the collection of the National Air and Space Museum. Interestingly, the statement in the advertisement above that it should be useable "in any intermediate position between fully extended and fully retracted...in case the pilot should make a hurried descent" seems to confirm the claim that the partially retracted design was an intentional safety feature, and not just an incidental benefit.

(Source:
National Air and Space Museum)