the mirrors on this side of the interferometer beam separated by an amount indicated by the distance the wedge has been shifted. The angle of the prism is about 10 degrees and a linear motion of the prism of 1 mm. corresponds to a path difference of 0.09 millimeter. If glass be subtracted in moving from the zero fringes to the reflector fringes, the mirrors must move closer together in order that both sets of fringes may appear in the eyepiece at the same time. The wedge is shifted by means of a rod; one turn of this rod moves the wedge 0.5 mm. and compensates for 0.045 mm. of air path. Fringes can be observed throughout one-third of a turn of this rod, corresponding to a path difference of about 26 light waves.

Having in mind the operation of the interferometer and the appearances to be expected, we will turn to the observations made. with the telescope in operation.

Several days were spent by the writer in November, 1920, in preparing the beam for operation, but as several important alterations were necessary, actual work was not begun until December. On December 13, with the outer mirrors at 10 feet separation the instrument was put in complete adjustment by observing ẞ Persei and y Orionis, both stars known to have diameters much smaller than can be measured with this instrument. This adjustment meant that, upon observing at the eyepiece, both the reference and the interferometer images were seen with fringes superimposed upon them. Upon turning to a Orionis the "zero" fringes were seen but no glimpse could be obtained of the interferometer fringes.

Turning to a Canis Minoris both sets of fringes were visible simultaneously, indicating that the instrument was in complete adjustment and that the disappearance of the fringes on a Orionis was real.

in

It may be thought that reliance cannot be placed upon a null measurement; there is no reason for this assumption as any strumental flexure or atmospheric disturbance requires but a very slight adjustment of the wedge, and at this time the seeing was very good. Dr. Anderson was present on this night and checked the writer's observations. Further observation of a Orionis was not attempted on the succeeding nights in December because the seeing was poor, as was indicated by a reduction in the visibility of the zero fringes; consequently observations were made on a Ceti, a Tauri,

and ẞ Geminorum, with the outer mirrors at 13 feet separation. In February, with mirrors approximately 16 feet and 19 feet apart, observations were made on a Orionis, a Tauri, ß Geminorum and ɑ Bootis. The seeing did not warrant drawing any definite conclusions except that fringes were seen at all points for ß Geminorum. This indicates that an interferometer with a base longer than 20 feet will be required to measure its diameter. Fringes were seen for a Tauri at 13 feet, at 19 feet, and in March at 14.5 feet, the visibility becoming less with increased separation of the mirrors. Additional measures will be made at points between 16 feet and 18 feet to see whether the fringes disappear as calculations indicate they should. For a Bootis the fringes were much reduced at 16 feet and could not be seen at 19 feet; the seeing was bad, however, and the observation indecisive.

Many stars have been used for checking the instrument; among them & Tauri, y Orionis, a Canis Minoris, a Geminorum, and Bootis. All have shown strong fringes at 19 feet.

η

Experience has shown throughout that better seeing is required for this work than was at first supposed, particularly when the mirrors are widely separated and the visibility of the fringes is approaching the point where they disappear. Change of seeing at these times will cause the fringes to flicker in and out, but a check is always at hand, for at the same time the visibility of the "zero" fringes is also reduced.

On some occasions in bad seeing the zero fringes will remain fixed, but the interferometer fringes will shift to the side of the image, probably because small sections of the wave front become inclined to the general wave front, due to varying atmospheric densities.

Having determined the distance at which the fringes vanish, we find the angular diameter of the object from the expression

1.221/b where a is the angular diameter in radians (206265′′), A is the effective wave-length (in cm.) of the star or that portion of the spectrum which is most predominant in forming the fringes seen by the eye of the observer and b is the distance apart of the mirrors (in cm.); Anderson has found, in connection with his work on Capella, that the effective wave-length of a solar type star is

5.5 X 10-5 cm. and it is assumed for a Orionis that the value of the wave-length is 5.75 X 10-5 cm., a true value for which must be found by direct experimental work. The value of b found for a Orionis is 121 inches (10 per cent.). The approximate value then for the angular diameter of a Orionis is .047". The agreement of this value with those obtained by calculation, which range from .031" to .051", is striking. If there is a falling off of intensity toward the limb, as in the case of the sun, Michelson finds this value would be increased by 17 per cent. Several determinations of the parallax have been made for a Orionis and from these its distance b may be found from the expression

where R93,000,000 miles, the distance of the earth from the sun, and is the value of the parallax in seconds of arc. Measures of the parallax thus far obtained are: Adams, .013′′; Yale, .032′′; Schlesinger, .016"; Yerkes, .022", the weighted mean of which is 020". From these values the distance is about 9.6 X 1014 miles. Knowing the distance and the angular value of the star, its linear diameter is found to be 218 X 10 miles. This value is not a definite figure but only an approximation; but in any case it means that the diameter of the star is several times the distance between the earth and the sun and several hundred times the diameter of the sun itself.

The work is being continued until the half dozen stars which calculations indicate as measurable with the twenty-foot beam have been investigated. Most of these are stars having late type spectra. In order to measure diameters of early type stars such as Sirius and Procyon a much longer base is needed. For this work an interferometer with mirror separations as great as 50 or 100 feet has been discussed but it is felt the present instrument should be used to its limit and many data accumulated, particularly regarding seeing conditions with the mirrors widely separated, before anything definite is attempted in the way of a larger instrument.

MOUNT WILSON OBSERVATORY,

PASADENA, CALIF.