116 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



From these absurdly varying results, Clarke * computes the value 

 58.932 as the most probable atomic w^eight, while Seubertf selects 59.6, 

 and Ostwald J 59.0, as the true value. Of course, many of the figures 

 above may be thrown out at once ; but the evidence which remains after 

 this first elimination is still too vague for anything like scientific cer- 

 tainty. It is not the purpose of the present work to discuss the accuracy 

 of these bygone investigations ; for such discussions are apt to be un- 

 convincing. Inattention to a single matter of detail may easily over- 

 throw all the value of quantitative labor ; a post-mortem examination is 

 not always able to detect the more subtle poisons. Our object is rather 

 to accumulate a mass of data which we thoroughly understand, and then to 

 interpret these data with regard to the probable chemical errors which may 

 have crept into them. The present paper is only the first one of an inves- 

 tigation which we hope to continue until we have solved the problem. 



The advantages to be gained by a simultaneous investigation of nickel 

 and cobalt are obvious. Although the similarity of the two metals is 

 usually overrated, it is still convenient to consider them together. Evi- 

 dently much time may be saved in the preparation of materials if two 

 similar researches are conducted side by side ; and evidently the experi- 

 ence gained in one may be immediately helpful in the other. Fortunately, 

 cobalt, like nickel, yields a bromide which is readily obtained anhydrous 



* Loc. cit. 



t Zeit. Anorg. Cliem., XIII. 229. 



i Lelirbuch, Vol. I. p. 79. 



