Wave-length of Light. 557 



diameters — is by no means an easy matter, and at the same 

 time a powerful and symmetrica] illumination is absolutely 

 necessary for the most accurate work, particularly in the case 

 of rather small grating-spaces. I had been thoroughly dis- 

 satisfied with the illumination previously used — a lamp at a 

 suitable distance — and now made a radical change. A three- 

 candle power electric lamp was attached directly to the 

 microscope just below the eyepiece and about a foot above the 

 objects measured. A small mirror carried by an arm screwed 

 to the objective reflected the beam into the Tolles illuminator. 

 A glass bulb filled with water surrounded the light and 

 served the double purpose of stopping radiation and partially 

 condensing t^e beam upon the mirror above mentioned. 



I am aware that such an arrangement is somewhat revolu- 

 tionary, and it was only after a careful trial that I convinced 

 myself that the heat from so near a source was not injurious. 

 In the first place, it should be noted that the lamp is only 

 used for a few moments at a time and at intervals long com- 

 pared with the time of observation. Thus the very minute 

 heat-wave that reaches the bar through the bulb of water can- 

 not possibly produce a perceptible rise of temperature during 

 the time of an observation, while during the intervals it is 

 completely dissipated. 



As an experimental fact, no heating effect whatever is 

 sensible even after a whole day's observations. To show at 

 once this fact, and the general character of an average series 

 of comparisons I subjoin ten comparisons of DnijS^ with a 

 certain decimetre on glass, made at intervals of about three- 

 quarters of an hour on two successive days. The figures are 

 taken directly from my note-book. 



Date. 









8. 



T= 



June 1, 1887 . . . D mi S a 2 = 



= G+21-3 



17-4 



)> 







V 



+ 21-6 



17-4 



V 







■)■) 



+ 22-1 



17-5 



)} 







J? 



+ 22-1 



17-5 



» 







JJ 



+ 20-8 



17-5 



V 







JJ 



+ 20-1 



17-5 



June 2, 1887 







V 



+ 21-4 



17-0 



» 







It 



+ 21-0 



17-0 









■>■> 



+ 21-0 

 + 21-0 



17-0 

 17-1 



The temperature was given by a thermometer in contact 

 with S a 2 , and 18 of the micrometer equalled K 28. In a 

 comparison of two standards with such unequal coefficients of 

 expansion as glass and speculum metal, the evil effects of 



Phil, Mag. S. 5. Vol. 25. No. 156. May 1888. 2 B 



