July 28, 192 1] 



NATURE 



687 



the beams, compensated by a plane parallel plate 

 in the other beam. The girder is capable of rota- 

 tion, so as to test whether the fringes vanish in 

 all position-angles, thus excluding the possibility 

 of the vanishing being due to a double source. 

 The observation calls for a high degree of ex- 

 perimental skill, as all who have used an inter- 

 ferometer will realise, and after shifting the 

 mirrors it is a matter of considerable difficulty to 

 find the fringes again. 



With a base line of 20 ft. it should be possible 

 to observe the disappearance of the fringes in 

 the case of stars the angular diameters of which 

 exceed about o''o2. When the telescope was 

 pointed on Vega the fringes did not disappear 

 even when the two adjustable mirrors were at 

 their maximum separation, indicating that the 

 angular diameter of Vega is less than this 

 amount. In the case of Betelgeuse the fringes 

 disappeared when the separation of the mirrors 

 was 10 ft. Adopting as the mean wave-length of 



Fig. 3. — Diagram illustrating the paths of the two interfering beams. 



the light 5500 angstroms, the value of \/Z is 

 o"-037. Assuming that there is no darkening 

 towards the limb, the angular diameter of Betel- 

 geuse will be 1-22 xo"-037 or o"-046. It is prob- 

 able that the supposition of darkening according 

 to the same law as for the sun would be nearer to 

 the truth, and the angular diameter would then 

 be i-33xo"o37 or o"o49. To this extent the 

 deduced angular diameter is uncertain. More re- 

 cently the fringes were found to vanish in the 

 case of Arcturus when the mirrors were 19 ft. 

 apart, corresponding to an angular diameter of 

 o"o24. This is almost the limiting angular dia- 

 meter which can be measured with the present 

 apparatus. 



The number of stars the angular diameters of 

 which exceed o"o2 is probably not very great; 

 according to Eddington's estimate they are to 

 be found amongst K-type stars of zero visual 

 magnitude, or M-type stars of zero to third magni- 

 tude. The appearance of the fringes in the case 

 of Vega with the maximum separation of the 

 mirrors was so distinct, however, that it Is con- 

 NO. 2700, VOL. 107] 



sidered that interference would be obtained with 

 a base line of 100 ft. or more ; if the mechanical 

 difficulties can be overcome, the application of 

 the method will no doubt be extended to base 

 lines greater than 20 ft. 



The determination of the angular diameters of 

 stars is possible only with the aid of a very large 

 instrument. For the measurement of the separa- 

 tions and position-angles of close double stars, the 

 method can be employed with comparatively small 

 instruments, and it is in this field that the method 

 possesses the greatest possibilities. Its advan- 

 tages as compared with the use of a filar micro- 

 meter are considerable. Adopting the late Lord 

 Rayleigh's criterion for resolving power, a double 

 star will appear just separated in a telescope if the 

 central image of one component falls on the first 

 diffraction ring of the other ; since with a tele- 

 scope of aperture d the distance between the 

 central image and first diffraction ring is i-22/\/d, 

 it follows that this is the smallest angular separa- 

 tion of the components of a double star for which 

 the star will appear double. But with the inter- 

 ference method, as has already been stated, the 

 fringes disappear provided that the distance apart 

 of the slits is .\/2o, a being the angular separation 

 of the components, the fringes due to one star 

 then falling exactly between those due to the 

 other. The least separation which can be ob- 

 served by this method, corresponding to a dis- 

 tance apart of the slits equal to the aperture, d, 

 is therefore A./2J, and it will be seen that the 

 method increases the resolving power of a tele- 

 scope in the ratio of about 244 to i. Expressing 

 d in inches, the normal angular limit of resolution 

 is approximately s" jd; with the loo-in. telescope 

 a separation of o"o5 should be just observable 

 with normal methods of observation, the inter- 

 ference method reducing this to o''o2. That this 

 increase in resolving power is actually obtained 

 is proved by the observations of Capella at Mount 

 Wilson. This star was known to be a spectro- 

 scopic binary, but visual methods had failed to 

 separate the components, though it was estab- 

 lished that their separation could not exceed 

 o"-o6; it therefore provided an interesting 

 test object. The disappearance of the fringes 

 was easily observed with the loo-in. tele- 

 scope ; the distance apart of the slits when 

 this occurs determines the separation, and their 

 orientation the position angle of the binary. 

 Observations secured on six nights enabled 

 an orbit to be computed. The obsei ved 

 distances and position-angles, together with the 

 residuals from the values deduced from the com- 

 puted orbit, are as follows : — 



Date 



1919 Dec. 30 



1920 Feb. 13 



„ 14 



Mar. 15 

 .April 23 



Distance Residual Position-angle Residual 



00418 000000 — — 



00458 —000003 '^'^ +04 



00451 -fooooo4 10 00 



00443 000000 3564 — 0-9 



00505 000000 2420 —0-4 



— — (not stated) -02 



