446 



iGIENCE. 



[N. S. Vol. XI. No. 273. 



lute truth, there is, nevertheless, infinite 

 satisfaction in the consciousness that we 

 may approach nearer and nearer to the 

 goal. The experiments of Ramsay and Ray- 

 leigh, however, lead us also to the knowl- 

 edge that with the perfection of our methods, 

 there may follow also the discovery of wholly 

 new facts ; and so investigation and experi- 

 mentation are always worth while. They 

 also bring as a reward answers even more 

 numerous than the questions proposed. In 

 other words, as in other fields, so also in 

 natural science, one advance seldom takes 

 place alone. In this sense it is not just 

 that we should so frequently speak of nota- 

 ble discoveries being made by accident ; for 

 we thus minimize the labors and the deserts 

 of the experimenter in the minds of others 

 who are not familiar with his work. 



We pass now to our second example. 

 Methods which have for their aim the 

 qualitative detection of elements are, other 

 things being equal, considered more valu- 

 able according as they are more sensitive, 

 that is according as they lead one to his 

 desired end with the use of the least possi- 

 ble quantity of substance. In this respect 

 it seems to be almost universally accepted 

 that the smallest quantity of substance can 

 be recognized, not by special experimental 

 apparatus, but directly through one of our 

 senses — that of smell. On the one hand 

 this assumption rests upon the observation 

 of E. Fischer and Penzoldt, that the limit 

 of perception of the odor of mercaptan is 

 reached when one four hundred and sixty- 

 millionth part of a milligram of the sub- 

 stance is present. On the other hand the 

 work of Kirchhofi" and Bunsen shows that 

 by spectrum analysis one fourteen millionth 

 of a milligram of sodium can be recognized 

 with certainty. Bj' this it is not to be un- 

 derstood that Kirchhoif and Bunsen sub- 

 mitted this quantity as the smallest amount 

 of a given element which will sufiSce to give 

 a spectrum recognizable with certainty, but 



they merely put forth this statement to 

 show the superiority of the methods of 

 spectrum analysis in general. My own ob- 

 servations, the details of which do not be- 

 long here, have at all events convinced me 

 that it is easy to so extend in spectrum 

 analj'sis the limit of sensitiveness that we 

 must to-day look upon spectrum analysis 

 as by far the most delicate of all analytical 

 methods. It is possible for us to compare 

 the smallest amount of detectible substance 

 with the hypothetical mass of the molecule, 

 that is with the smallest quantity of sub- 

 stance that is capable of existing. In this 

 way we easily reach the perhaps surprising 

 fact that the least quantity of mercaptan 

 perceptible to the senses contains about 

 2x10"" molecules. Hence, even of the most 

 penetrating odors, it is necessary for twenty 

 billions of their molecules to bombard the 

 olfactory nerves, before these carry the sen- 

 sation to the brain ; they utterly ignore all 

 smaller quantities. In ordinary analytical 

 methods the quantity of substance is nat- 

 urally incomparably greater. Without un- 

 dertaking to weary you with figures I will 

 in general and briefly note that by micro- 

 chemical methods it is possible to detect 

 about one ten-thousandth of a milligram, 

 while in ordinary test tube reactions, it is 

 desirable to have at least several milligrams 

 of the substance present. 



And now we come to the consideration 

 of the last of the problems mentioned. The 

 processes by means of which we decompose 

 the mixtures which nature furnishes us, 

 into their individual constituents, that is 

 obtain from them chemically pure com- 

 pounds, are of many kinds. We make use 

 of variations in solubility and volatility, of 

 the possibilities of crystallization and diffu- 

 sion, of chemical, electrical and magnetic 

 relations. These are, in a single word, 

 really gradual differences, which can be used 

 so much the less perfectly as the bodies to 

 be separated resemble each other in phys- 



