366 PROCEEDINGS OF THE AMERICAN ACADEMY. 



result. The table below includes all of the five possible ratios obtainable 

 from our work, series I. and IV. being rejected because they were merely 

 preliminary : — 



(Series II.) 58.995 



(Series III.) 58.987 



(Series V. and VI.) 58.998 



(Series II. and V.) 58.994 



(Series III. and V.) 58.992 



Average .... 58.993 



This table, although giving an interesting statement of the possible 

 combinations, does not yield a fair average, — for Series V. is introduced 

 three times, Series II. and III. each twice, while Series VI., which is at 

 least as accurate as the others, appears only once. A fairer method 

 would probably be to avoid all hypotheses and combinations, and assign 

 to each of the four series equal weight, as follows : — ■ 



Series II. (uncorrected) Co = 58.995 



Series III. (uncorrected) Co = 58.987 



Series V. (corrected) Co = 58.992 



Series VI. (corrected) Co = 59.004 



Final Average . . . 58.995 



Obviously it makes but little difference which method we adopt : the 

 averages are essentially identical. The highest individual experimental 

 result among all these determinations was 59.021, and the lowest 58.955, 

 the average variation from the mean 58.995 being 0.012. Because 

 these results are less concordant than one could wish, and the conclusion 

 is somewhat less positive than that reached in the case of nickel, the 

 atomic weight of cobalt is being further studied by radically different 

 methods in this Laboratory. 



In a recent article,* Professor Winkler calls attention to some possible 

 errors, in the work upon both nickel and cobalt published last year. 

 That the disagreement between his results and ours is due to the metliods 

 employed by him in his work upon these two elements has been suf- 

 ficiently shown in the preceding paper upon the atomic weight of nickel. 

 In addition, however, his several suggestions concerning our own work 

 should obviously be reviewed and discussed in detail. 



His specific criticisms are four in number. First, he suggests that the 



* Zeitschr. Anorg. Chem., XVII. 236. 



