REED RELATION BETWEEN VALENCE & ATOMIC WEIGHT. G'] 1 



dental, and that their atomic weights are affected by a negative 

 error of about one unit. We sliall class these elements, therefore, 

 along with S« and C^ as exceptions whose atomic weights are 

 too low. The correctious required are 



Atomic Weiji;ht. Corrections. 



Wj 85.251 -f 1.749 



Y 89.8 + 1.2 



C6 94. + 1. 



W) 94. + 1. 



Sk 117.698 + 2.302 



Cs 132.58 + 2.42 



The deviations of some of these might reasonably be attributed 

 to experimental errors. But I prefer to consider them all excep- 

 tions, and would even add A.y, Br, I and Sb to the list, though they 

 appear to conform regularly to the law. It will be remembered 

 that the curves pass tlirough the characteristics of A5, Br, I and 

 S/^, instead of their well marked maxima and minima, as we 

 should naturally expect. To bring the maxima and minima of 

 these elements on the curves in their regular places would require 

 the following corrections : 



Atomic Weight. Corrections. 



As' 74.918 + 4.082 



Br 79.768 " + 3.232 



I 126.557 ^ 4- 4.443 



S6 122.46 -f 4.54 



Such large deviations cannot in these elements be attributed to 

 ordinary experimental errors. They may, however, all be refer- 

 red to a common cause, viz., the presence of unknoxvn elements 

 ■which tend to rednce the apparent atomic iveights of those ele- 

 ments with which they are associated and with which they have 

 been hitJierto confoundid* 



The unknown elements to which I refer are such as would fill 

 the vacant "harmonics" in the seven groups. 



In the C/ group, for instance, we find two vacancies between 

 C/ and Br, two between Br and I, and one below F, represented 

 on Plates I. and IL by ^j, ^.^, ^3, ^^, and y ^. Their atomic 

 weights are 3, 51, 67, 99, and 115, respectively. 



If any or all of these elements were mixed with I, its apparent 



* Since the above was written D/ has been found to be a mixture of two elements. See 

 note on Dj. 



