218 Love and Smcid 



—.89 - 



—.43 - 



.08 - 



— .U - 



m - 



—.20 - 

 .21 - 

 .17 - 

 —.07 - 

 —.03 - 

 —.23 - 

 —.25 - 

 —.22 - 

 .37 - 

 —.08 - 

 .37 - 

 .10 - 

 —.10 - 

 .55 - 

 .16 - 

 .04 - 

 .22 - 

 —.25 - 

 —.04 - 

 —.06 - 

 —.62 - 

 In spite of the larue variation> 

 errors of x and r are for the most part small. The probable 

 error of a single observation of x is 0.22 mm. and of r 2.05 per 

 cent. A few larf>e errors of course appear, but as stated before, 

 they are most likely due to errors of reading. 



It remains to apply Ekholm's two-constant formula to the 

 same observations, and see whether any improvement is 

 effected in the agreement between observation and calculation. 

 The values of /; and A, as determined by least squares from the 

 63 observations, are 0.9974 and 0.0007062 respectively. That 

 is, Yj is indistinguishal)le from unity to the order of accuracy 

 which can be attained in such observations, and A is within 



