OF ARTS AND SCIENCES. 385 



method with important modifications. In Foucault's experiment, the 

 deflection of the light produced by the revolving mirror was too 

 small for the most accurate measurement. Mr. Michelson placed the 

 revolving mirror 500 feet from the slit (which was 10 times the dis- 

 tance in Foucault's experiment) and obtained a deflection 20 times 

 as great, although the mirror made only 128 turns in a second. With 

 apparatus comparatively crude, he obtained for the velocity of licht 

 186.500, with a probable error of 300 miles. This preliminary ex- 

 periment, made in the laboratory of the Naval Academy in May, 

 1878, indicated the directions in which improvements must be made 

 in order to insure greater accuracy. The distance from the slit to 

 the revolving mirror must be increased, the mirror must revolve at 

 least 250 times a second, and the lens for economizing the light must 

 have a large surface and a focal length of about 150 feet. With the 

 aid of $2,000 from a private source, Mr. Michelson was able to carry 

 out his ideas on a liberal scale. 



His new experiments were made in the summer of 1879. The 

 revolving mirror, made by Alvan Clark and Sons, was moved by a 

 turbine wheel. Its rapidity of revolution was measuied by optical 

 comparison with an electric fork which made about 128 vibrations 

 a second, the precise value being accurately measured by reference 

 to one of Kcinig's standard forks. The velocity generally given to 

 the mirror was about 256 turns a second. The distance between the 

 revolving and the fixed mirror was 1,986.26 feet. The light from 

 the moving mirror was concentrated on the fixed mirror by a lens 

 8 inches in diameter, with a focal length of 150 feet. These im- 

 provements on Foucault's arrangement were so advantageous that 

 Mr. Michelson obtained, even with a smaller speed in the revolving 

 mirror, an angle of separation between the outgoing and returning 

 rays of light so great that the inclined plate of glass in front of the 

 micrometer was not necessary ; the head of the observer not shutting 

 off the light. The mean result of 100 observations taken on 18 

 different days made the velocity of light 186,313 miles per second; 

 with a probable error of 30 miles. 



In 1882. at the request of Professor Newcomb, Mr. Michelson made 

 a redetermination of the velocity of light at the Case Institute, in Cleve- 

 land, Ohio, by the method already described, with some modifications. 

 The space traversed by the light in going and returning between the 

 two mirrors was 4,099 feet. Two slight errors in the reduction of his 

 former work were corrected in this. The velocity deduced from 563 

 new observations was 186,278 miles, with a probable error of 37 miles. 



VOL. XXIV. (n. S. XVI.) 25 



