256 Mr. A. A. Michelson on the Visibility of 



point, the diffused light of a dull December day being quite 

 sufficient to act upon the selenium * . 



But I do not at present attach any serious importance to 

 such practical applications of these devices. I regard them 

 simply as affording somewhat attractive illustrations of the 

 effect of light upon the resistance of selenium. 



XXVIII. Visibility of Interference-Fringes in the Focus of a 

 Telescope. By Albert A. Michelson. f 



WHEN the angle subtended by an object viewed through 

 a telescope is less than that subtended by a light- 

 wave at a distance equal to the diameter of the objective, the 

 form of the object can no longer be inferred from that of the 

 image. Thus, if the object be a disk, a triangle, a point, or a 

 double star, the appearance in the telescope is nearly the 

 same. 



If, however, the objective is limited by a rectangular slit, 

 or better by two such, equal and parallel, then, as has been 

 shown in a former paper J, the visibility of the interference- 

 fringes is in general a periodic function of the ratio of a, the 

 angular magnitude of the source in the direction perpendicular 

 to the length of the slits, and a , the " limit of resolution." 



The period of this function, and thence — , may be found with 



great accuracy ; so that by annulling the greater portion of 

 the objective the accuracy of measurement of the angular 

 magnitude of a small or distant source may be increased from 

 ten to fifty times. As ordinarily understood, this increase of 

 u accuracy " would be at the cost of u definition 9S (which in 

 this sense is practically zero) ; but if by " definition ■" we 

 mean, not the closeness of the resemblance of the image to the 

 object, but the accuracy with which the form may be inferred, 

 then definition and accuracy are increased in about the same 

 proportion. 



In almost every case likely to arise in practice, the form of 

 the source is a circular disk ; and if the illumination over its 

 surface were uniform, the only problem to be solved would be 

 the measurement of its diameter. But in many cases the 

 distribution is anything but uniform. If the curve repre- 



* Experiments with the hell and the lamp were shown at the meeting, 

 t Communicated hy the Author. 



\ "On the Application of Interference Methods to Astronomical 

 Measurements " (Phil. Mag. July 1890). 



