180 



PROCEEDINGS OF THE AMERICAN ACADEMY 



Lake Superior Copper. 



The average of these two series is 63.450, with greatest variations 

 of +.002 and — .003, and a probable error of ±.0006. The average 

 of the results calculated from the weight of silver dried at 150° is 

 63.436, while the results published in the preceding paper gave 

 63.437. 



The complete concordance of these results with each other, and 

 with the previous value above referred to, would point strongly to the 

 following conclusions. 



First, that the copper used in each case was absolutely free from 

 metallic alloy ; for manifestly the three entirely different samples would 

 be likely to contain different impurities, or at least different amounts 

 of the same impurity. The copper was tested for sulphur with the 

 greatest care by solution in nitric acid and treatment with baric chlo- 

 ride, and no trace of cloudiness was perceptible. That the copper was 

 absolutely free from impurity is not contended ; only that it did not 

 contain a weighable amount of impurity in one gram, the amount 

 used in each experiment. It is manifest that attempts to purify the 

 copper beyond this limit would be labor thrown away, and would pro- 

 duce no effect upon the atomic weight. For example, one tenth of a 

 milligram is a very large amount of foreign material to suppose exist- 

 ing in a gram of copper purified with such care ; but this large amount 

 would only change the atomic weight five units in the third decimal 

 place, — a quantity which is of no consequence when the atomic weight 

 is in doubt three units in the first decimal place. 



Another and still more positive conclusion reached by these results 

 is that the atomic weight of copper is a constant quantity with refer- 

 ence to nitric acid and silver. If copper had a variable atomic weight, 

 it would surely appear in specimens taken from such widely different 



