68 REPORTS ON INVESTIGATIONS AND PROJECTS. 



motor or its connections. A bulb containing saturated ether vapor 

 is immersed in the water, which is heated by two small incandescent 

 lamps controlled through a relay. If the temperature rises too high 

 the current is cut off by the expansion of the ether vapor, which 

 moves a column of mercury and thus breaks contact in the relay 

 circuit. In the preliminary work on Arcturus described below, the 

 grating showed no' variations of temperature equal to o.i° C, as 

 measured by a thermometer immersed in mercury in contact with the 

 grating. Further improvements in this apparatus will undoubtedl)- 

 give all necessary constanc}- of temperature. 



An automatic arrangement is now being installed to control the 

 temperature of the entire spectrograph, as the variations of several 

 degrees from minimum night to maximum day temperature, which 

 occur in extreme cases, are greater than can be allowed. Under 

 ordinary circumstances the range within the room is very small and 

 probably inappreciable in effect. 



One photograph of the blue region of the first-order spectrum of 

 Arcturus required an exposure of fourteen hours on three successive 

 nights. Another, made when the mirrors were badly tarnished, 

 required twenty-four hours on five successive nights. The guiding 

 was done with the aid of electric slow motions attached to the mount- 

 ing of the 6o-foot concave mirror of the Snow telescope, supplemented 

 by cords (kept taut by weights) with which the cell can be sprung 

 slightly without danger of distorting the mirror. The good defini- 

 tion and the uniform succession of clear nights make Mount Wilson 

 an excellent place for such work. 



The photographs are good, but the}^ are not equal to solar spectra 

 taken in the same apparatus with short exposure. It is expected 

 that improved temperature control will remedy this difficulty. The 

 value of the spectra is well shown by the fact that the measures of 

 a large number of lines made by Mr. Adams show the linear error 

 of setting to be as small as in the case of the best Bruce spectrograph 

 plates. As the scale of the present photographs is nearly three times 

 as great, the increase of precision should be in the same ratio. 



These experiments with the Snow telescope should be of special 

 service in their bearing on the design and use of the large grating 

 spectrograph for the five-foot reflector. As this instrument will give 

 six times as much light as the Snow telescope, it is already evident 

 that valuable results may be expected from its application to high- 

 dispersion stellar spectroscopy. 



