SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. IO7 



0.423/* this corresponds to o.o8/x and at 0.81 3/i, it corresponds to 0.25/x. 

 However, the ends of these spectrum intervals were much weakened 

 by obstructions within the radiometer itself. So the spectrum that 

 could be seen in the eyepiece of the radiometer appeared only about 

 three times as high as the radiometer vanes. The other parts of the 

 spectrum, which might have scattered extraneous light onto the radi- 

 ometer, were lost in part by overrunning the mirror f, and those re- 





29fk 



32ft. 



Fig. 2. — Diagram of the optical path. The star beam focused on the slit at o 

 is made parallel by the quartz lens h. Thence it is reflected by the mirror c 

 through the prism d onto the mirror e, thence to the mirrors / and g and 

 through the quartz lens h to the mirror i which reflects it to focus on the 

 radiometer vane at /. 



maining were lost in large part by overrunning the mirror g and lens 

 h. Such fragments of these extraneous spectral rays as fell on the 

 mirror / and the mirror g were scattered over the surface of these 

 mirrors at such angles that they must have been almost entirely lost 

 from the observed wave-length intervals falling on the radiometer 

 vanes. In place of a slit at the Coude focus we used a round aperture 

 1.5 millimeters in diameter, so as to avoid mainly variable losses of 

 light with changes of atmospheric "seeing." 



The radiometer itself has been sufficiently described in Publication 

 3808. Abbot constructed a brass gadget with which he was able to 



