820 On the Atomic Weights of analogoits Elements. 



I believe that, with one exception, every well-defined group of 

 elements has been considered under these three classes. That 

 exception is in the case of arsenic and antimony, which are un- 

 questionably analogous, phosphorus being also closely allied to 

 them : the three are placed together in GmeHn's arrangement, 

 but I see no ratio between their numbers 31, 75, 129. Schroet- 

 ter's reduction of the equivalent of phosphorus has prevented 

 our considering antimony, ] 29, as four times that element. It 

 is, however, double tellui'ium, 64'2, which is in some respects an 

 analogous body. ,,p 



There are several elements, such as bismuth, which have no 

 veiy evident analogues; and of others which are similar we 

 cannot speak, because we are ignorant of their atomic weights. 

 These are pelopium and niobium, and yttrium erbium and 

 terbium. 



Alumina is usually classed with the earths ; but its compounds 

 are strictly isomorphous with those of the sesquioxides of the 

 iron group. The equivalent of aluminium, 13'7, happens to be 

 half of theirs. Again, glucina is certainly an analogous eartl?if|, 

 viewing it also as a sesquioxide, the equivalent of glucinum is 7, 

 which is half that of aluminium. 



Some of the properties of lead would ally it to the family of 

 the alkaline earths, but there is no apparent numerical relatioii't- 

 in other points it resembles silver, with which it is so generally 

 found; their respective equivalents are 103'7 and 108*1, but 

 neither the resemblance of chemical properties nor of number is 

 very close. I would rather not consider it an instance of the 

 general law. 



Whatever may be thought of some of the speculations towards 

 the close of this paper, these numerical relations are indisputable 

 facts. That we should frequently find relations between 56 

 numbers drawn at hazard from a range little exceeding 100 

 'might be predicted from the laws of probability, but that this 

 should be to a considerable extent coincident with chemical 

 relationship is not probable. Still more is it against all proba- 

 bility that, by mere chance, whenever, with one exception, close 

 analogy of properties exists, there exists also numerical relation- 

 ship ; and although we cannot now see the precise reason of this, 

 we can scarcely imagine that the intimate constitution of these 

 related elementary bodies will long remain an unfmitful field of 

 mvestigation. 



