﻿22tt E. C. Pickering — Ultra Violet Spectrum. 



which these maps may be compared. The improved apparatus 

 and enlarged scale of Professor Rowland's work rendered it 

 probable that its accidental errors would be unimportant in the 

 present comparison, especially as the measures are only to be 

 carried to hundredths of a unit. As a check upon its sys- 

 tematic errors a comparison has been made with the investiga- 

 tions of Dr. Miiller and Dr. Kempf in the Publicationen des 

 Astrophysikalischen Observatoriums zu Potsdam, V. Three 

 hundred lines were measured between the limits of 686 and 

 389. About two hundred of these are contained in the region 

 covered by Professor Rowland's map. A comparison of a 

 mimber of them indicate that the accidental errors are very 

 small. The width of the bands or groups of lines into which 

 many of the lines are resolved renders it largely a matter of 

 judgment where the center should be taken. The systematic 

 differences are about O'Ol, or more strictly the wave-lengths 

 according to Rowland exceed those of Miiller and Kempf by 

 about one eighty thousandth paft. The region 588 to 590 

 occurs on four of the strips of Rowland's map. The positions 

 of forty-five of the lines contained in this region were estimated 

 on all four strips and indicated corrections of +'0007, +'0007, 

 — *0009 and —'0004 respectively. These quantities are proba- 

 bly due to accident. They are inappreciable by the method of 

 measurement employed, as they would not exceed one five 

 hundredth of an inch on the map. 



The specimen of Dr. Draper's map with which the comparison 

 was made was not a proof impression, but was taken from a copy 

 of this Journal. The positions of seventy-six lines were read, 

 estimating the divisions to tenths by the aid of a reading glass 

 having a focal length of about 10 cm . Preference was given to 

 well defined lines, a few bands with hazy edges being omitted. 

 The corresponding positions of these lines were then read from 

 Professor Rowland's map, estimating the tenths by the unaided 

 eye. The differences were arranged in groups as shown in 

 table I, according to their wave-length. Each group extends 

 over one unit, the middle points having the values given in 

 the first column of the table. The number of lines contained 

 in the group is given in the second column, and the mean of 

 the residuals, Draper minus Rowland, in the third column. 

 The fourth column gives the results of a similar comparison 

 with the map of Cornu. The same lines were used except that 

 having a wave-length of 393*76, which is not given on Cornu's 

 map. Some uncertainty exists regarding three or four of the 

 other lines, where the engraving does not agree with the photo- 

 graph. No large residuals are thus introduced, since lines in 

 the right place are selected even if the intensity is incorrectly 

 represented. 



