72 



REPORT OF THE SECRETARY. 



VI. Sources of error inherent in the bolometric method. 



1. From tardy action of the galvanometer, clue to electrostatic causes, damping of 

 the needle, etc., causing displacement of deflections. 

 May be eliminated by reversing the direction of motion of the circle between 

 different runs, as has already been pointed out, and heretofore practiced. 



As a general summing up of these investigations it may be said that if the steps 

 now being taken are successful in their outcome, it will be possible to determine the 

 difference of deviation between the lines appearing on the holographs and the A line 

 in the visible spectrum, with an error not exceeding 1 second of arc. 



The number of lines which may be theoretically discriminated with the process 

 and apparatus employed is very large. With a perfect prism of the size used, 

 and a galvanometer of somewhat greater sensitrveness, more than a thousand might, 

 if they exist, be discovered within the region in which we are now working. But 

 this supposes all minor tremors in the holographic curves to be eliminated, and it is 

 certain that this has not yet been done, and that very many of the lines shown in 

 maps exhibited in illustration of the process will prove to have been due to such 

 causes. The chief aim at the present stage of the investigation is to give a number 

 of lines with their positions in the spectrum of a 60° rock salt prism, which may be 

 regarded as "standards," and as many more as may be considered to have been 

 verified beyond reasonable doubt. Of these it is believed that at least 2G0 can be 

 presented iu the forthcoming publication. 



B. General Spectrobolographic Work. 



In the following table is given the number of photographs of various kinds taken 

 at the observatory during the year, not including instrumental records taken to show 

 the magnitude of the accidental galvanometer deflections : 



Date. 



1895. 

 July 



August 



September 



October 



November 



December 



1896. 



January 



February 



March 



April 



May 



June 



Total 



holographs. 



Cylindxics. 



209 



Other 

 photographs. 



96 



Grand total, 362. 



Full comparator measurements on curves of April, 1894, and March, 1896, have been 

 made by three observers after the manner already described, aud the results have 

 been reduced to deviation. The curves measured were taken with coarse bolometer 

 at a rapid rate of speed, and the finer lines were thus eliminated. About 80 promi- 

 nent or "standard" lines were verified in each of the sets of curves measured. 



It has been repeatedly remarked in these reports that the most satisfactory results 

 can not be obtained in the site at present occupied by the observatory within the 



