Results obtained in a Recalculation of the Atomic Weights. 101 



The opposition of sign is apparent only, and relates to the dif- 

 ferent methods of measurement adopted in the two cases. In 

 (65) the primary and secondary disturbances are represented 

 by A/Kj but in (67) by the magnetic function c. If we express 

 the solution in the second case in terms of the electric func- 

 tion g, we shall find (see 13) that the ratio of c to g changes 

 sign when we pass from the primary light propagated along 

 — x to the secondary light propagated along -f x. The actual 

 ratio of amplitudes in the two cases is thus (k /2 + Jc 2 )/2k 2 , or 

 (K / + K)/2K. Unless the difference between K7 and K be 

 neglected, the two components of unpolarized light are scat- 

 tered along this direction in different proportions, that compo- 

 nent preponderating in which the electric displacement is 

 parallel to the axis of the cylinder. The secondary light is 

 therefore partially polarized in the plane perpendicular to the 

 axis. 



June 1881. 



XI. An Abstract "of the Results obtained in a Recalculation 

 of the Atomic Weights. By Frank Wigglesworth 

 Clarke, S.B., Professor of Chemistry in the University of 

 Cincinnati* . 



DURING the past three years I have been engaged upon 

 a recalculation of all the atomic-weight determina- 

 tions which have been published from the time of Berzelius's 

 earlier investigations down to the present date. My purpose 

 has been to reduce all similar series of experiments to com- 

 mon standards, to calculate the probable error of each series, 

 to combine the results into general means, and then to deduce 

 the atomic weights in such a way that each value should 

 represent a fair average of all the trustworthy estimations. 

 In other words, I have sought to bring together all the vast 

 number of scattered details, and to derive from them a more 

 consistent table of atomic weights than has hitherto been 

 found in chemical literature. My complete work will appear 

 in due time as a separate volume ; my present intention is to 

 give merely a summary of my methods, and my conclusions. 

 Taking hydrogen as unity, I necessarily began with the 

 ratio between it and oxygen. This ratio has been deter- 

 mined accurately in only two ways : — first, by the synthesis of 

 water over copper oxide ; and secondly, from the relative 

 density of the two gases. Ignoring earlier inexact experi- 

 ments, we may consider only the data furnished by Dumas, 

 by Erdmann and Marchand, and by Eegnault. From Pumns's 

 nineteen syntheses of water we get for oxygen values ranging 

 * Communicated by the Author. 



