•26 



RESUME. 



Below are given tlie collected results of these observations. In the second 

 column will be found the corresponding wave-leugtbs according to Angström and 

 in the third their differences. 





:\[ 



A 



A-M 



B 



68G6.8-2 



6867.0 



.18 



C 



6r)61.G-2 



6562.1 



.48 



D, 



.'5894.78 



5895.17 



.39 



D, 



5888.G8 



5889.09 



.41 



E 



5269.13 



5269.59 



.46 



h 



5182.40 



5183.10 



.70 



I., 



5171.47 



5172.16 



.69 



I'. 



5167.73 



5168.48 



.75 



k 



5166.16 



5166.88 



.72 



F 



4860.16 



4860.72 



..56 



G 



4307.19 



4307.25 



.06 



It will be seen that the difference in no case amounts to a whole \niit. 

 Much of it. indeed, might be destroved, bv a more accurate knowledge of the 

 grating space in each case. It will be observed that a sudden increase in the 

 differences occurs when the gi'oup of h lines is reached. I think this finds an ex- 

 planation in the iact, referred to in gi\'ing the observations on /;,, tliat the setting 

 was made upon what appeare<l to be a well defined part of the line, tntrards the 

 side of lertft wave hvfith. As the other three lines are referred to this, it is to be 

 expected that all will be affected alike and this is the case. It seems likely that 

 Augström set iipun the middle of (he line. The iiart of the line referred to in 

 these observations seems so well defined that I prefer to retain that as the point 

 of departm-e rather tlian to reduce to the middle of the line, although the 

 uniformity of differences would be greatly increased by such a reduction. I 

 know of no reason for the small difference in the results for G. other than the 

 difficulty of making measurements in that part of the sj^ectrum whicli would 

 doubtless affect both sets of observations more or less. If Augström had used 

 the same part of ^i for a jioint of reference that has been adopted here, I am 

 convinced that for all of that portion of the spectrum in whicli the best measure- 

 ments are possible, the series of differences would have been very nearly uniform 

 and easily accounted for by an ineonsistancy in the deteriuiuations of the grating 

 space. As before stated, the ratios of these numbers ought to be independent of 

 the value of the grating space and in order that a comparison may be made 

 upon that basis, ratios have been computed for both series, using the value of D, 

 as a unit. Thev are as follows. 



