Decembee 10, 1900] 



SCIENCE 



851 



materials is elaborately studied; the errors 

 due to occlusions of gases by metallic oxides, 

 and of water by crystallized salts, are pointed 

 out; and by attention to minutiae of this kind 

 the accuracy of the determinations has been 

 greatly increased. 



In general, with a few exceptions. Professor 

 Richards has confined himseH to one group of 

 methods, namely, the analysis, by known 

 processes, of metallic chlorides and bromides. 

 These, in nearly all instances, involve a knowl- 

 edge of the atomic weight of silver, through 

 which the atomic weights of the other ele- 

 ments are referred to that of the standard, 

 oxj-gen. That is, ratios are determined, from 

 which, with reference to silver as the experi- 

 mental standard, the other atomic weights are 

 computed. At first, the secondary standard 

 Ag = 107.93, established by Stas, was ac- 

 cepted ; latterly, however, it has been shown by 

 several authorities that Ag = 107.88 is nearer 

 the truth, and that the true value may even be 

 slightly lower. This change produces corre- 

 sponding changes in the other atomic weights ; 

 a condition of affairs which is not altogether 

 satisfactory. In most cases each atomic 

 weight determined by Richards is a function 

 of the atomic weights of silver, chlorine and 

 bromine, and these have been, in effect, three 

 variables. Theoretically they are constants, 

 but the values found for them have varied, 

 and the variations are far reaching in their 

 effects. The great exactness of Eichards's 

 work is in the measurement of definite ratios, 

 which, once established, form the basis upon 

 which our knowledge of the atomic weights 

 must stand. As the variations in the refer- 

 ence values diminish, the accuracy of our de- 

 ductions will increase. 



From one point of view it is well that the 

 Harvard chemists should have devoted them- 

 selves, not exclusively, but in great part, to 

 one group of methods. Those methods have 

 been perfected, their sources of error have 

 probably been reduced to a minimum, and the 

 measurements made with their aid leave little 

 to be desired. Considered more broadly, how- 

 ever, it is desirable that other, radically dif- 

 ferent methods should be developed with equal 



thoroughness. Not until that has been done, 

 not until closely agreeing determinations of 

 atomic weights have been made by several dis- 

 tinct reactions and processes, can we regard 

 these constants as sharply established. Work 

 of this sort, especially with reference to the 

 more fundamental atomic weights, is now 

 going on in several laboratories, among which 

 may be mentioned that of Guye, at Geneva. 

 Within the next ten years our knowledge of 

 the atomic weights is likely to be greatly in- 

 creased. Meanwhile, the work of Richards 

 and his colleagues must be assigned preem- 

 inence. 



F. W. Clarke 



Elemente der ExaMen Erhlichlceitslehre. By 

 W. JoHANNSEN. Deutsch wesentlich erweit- 

 erte Ausgabe in fiinfundzwanzig Vorles- 

 ungen. Jena, G. Fischer. Pp. vi + 515. 

 Gebunden, 10 Marks. 



The epoch in evolutionary study opened by 

 deVries's " Mutationstheorie " had been one 

 not only of experimentation, but also, fortu- 

 nately enough, of thoroughgoing analysis. We 

 had analysis of evolution in suflicient amount, 

 even ad nauseam, in the latter part of the 

 last century; but the newer speculations are 

 based on novel, experimentally acquired facts, 

 and the marvel of it is that they bear little re- 

 semblance to the conventional and orthodox 

 teachings which we accepted almost without 

 question a decade or two ago. It is to the 

 shame of biological science that it must be 

 acknowledged that it was long contented to 

 accept these speculations as fundamental 

 principles without testing them experiment- 

 ally. But all that is now happily by and the 

 era of framing hypotheses for the purpose 

 only of testing them is well launched. 



Of the old ideas, those grouped about varia- 

 tion have undergone, perhaps, the completest 

 analysis. And they needed it too, for if one 

 thing is clearer than another, it is that Dar- 

 win and his followers did not analyze the 

 phenomena of variation satisfactorily. It is 

 almost pathetic to see in his letters and books 

 how he fails to distinguish the fundamental 

 differences between fluctuating non-inherit- 



