Albert Michelson measured the diameters of Jupiter's moons by interferometry in the early 1890's in order to test the technique. Using the 12-inch refractor at California's Lick Observatory he covered the lens with a large screen that had two holes ("pick-off points") at separations of up to 12 inches to simulate different baselines. The diameters of the satellites were determined to be on the order of one arc second, which is approximately correct.
By 1919, knowing that the diameters of stars appeared much smaller because of their great distances from Earth, Michelson and Francis Pease designed and built the 20-foot Stellar Interferometer. Here the 20 feet refers to the maximum baseline of the pick-off mirrors. It was installed atop the 100-inch telescope (shown above), the largest telescope in the world at the time. The maximum baseline possible was 20 feet because anything longer would hit the interior of the dome as the telescope moved from one part of the sky to another.
The design of this instrument and the perforated screen used at Lick Observatory were the idea of the French scientist Hippolyte Fizeau (1819-1896), who was also the first person to measure the speed of light in a laboratory. In effect, the 100-inch telescope was being used as a large and stable optical bench capable of being pointed at different parts of the sky. The 100-inch telescope has a long history of being used in this way -- the natural guide star adaptive optics instrument on the 100-inch telescope has an optical bench bolted to the telescope at Cassegrain focus and an artificial guide star adaptive optics system is being developed at its coude focus.
To observe with the 20-foot interferometer, the telescope was pointed at the target star and the two flat mirrors were moved inwards as far as possible (thereby giving the shortest baseline and the brightest fringes). The goal was to adjust the light path lengths so that the band of brightest fringes would be carried to the eyepiece of the observer. It was easier to adjust the light path lengths and move the fringes to a fixed eyepiece than to move the observer to a fixed fringe location. This concept was also used in later interferometers.
The fringes would be located manually by an observer. Once located, the relative fringe brightness was noted. Then the outer two mirrors would be moved apart and the measurement repeated. Eventually the fringes would disappear and the star would be "resolved". Doing this manually was very tedious and time consuming, which was a problem since the newly-commissioned 100-inch telescope in great demand.
The diameters of only seven stars were successfully measured, all red giant stars: Betelgeuse, Arcturus, Antares, Aldebaran, Ras Algethi (Alpha Herculis), Scheat (Beta Pegasi), and Mira. The very close binary star Capella was also observed in December 1919 J. Anderson and elements of its orbit were successfully measured. However, many other smaller stars could not be resolved and it soon became apparent that baselines longer than 20 feet would be required.