540 



SCIENCE 



[N. S. Vol. XXXIV. No. 878 



gases. These gases are displaced, after 

 partial cooling, first by nitrogen, and then 

 by pure dry air, and the boat is pushed 

 past the stopper into the weighing-bottle, 

 the stopper being then forced into place, 

 and the substance thus shut up in an en- 

 tirely dry atmosphere. The weighing-bot- 

 tle may now be removed, placed in an ordi- 

 nary desiccator and weighed at leisure. 

 The substance is really dry, and its weight 

 has definite significance. 



Mention may be made also of another 

 instrument, which likewise has greatly 

 facilitated the recent work at Harvard, 

 namely, the " nephelometer. ' '" With the 

 nephelometer, minute traces of suspended 

 precipitate may be approximately deter- 

 mined from the brightness of the light they 

 reflect. The construction is very simple. 

 Two test-tubes, near together and slightly 

 inclined toward one another, are arranged 

 so as to be partly shielded from a bright 

 source of light by sliding screens. The 

 tubes are observed from above through two 

 thin prisms, which bring their images to- 

 gether and produce an appearance resem- 

 bling that in the familiar half-shadow 

 polarimeter. The unknown quantity of 

 dissolved substance is precipitated as a 

 faint opalescence in one tube by means of 

 suitable reagents; and a known amount, 

 treated in exactly the same way, is pre- 

 pared in the other. Each precipitate re- 

 flects the light; the tubes appear faintly 

 luminous. If the tubes show like tints to 

 the eye when the screens are similarly 

 placed, the precipitates may be presumed 

 to be equal in amount. In case of inequal- 

 ity of appearance, the changed positions of 

 the screens necessary to produce equality of 

 tint give a fairly accurate guide as to the 

 relative quantities of precipitate in the two 



' Eichards, Zeitsch. anorg. Chem., 1895, 8, p. 

 269; Richards and Wells, Amer. Chem. J., 1904, 

 31, p. 235; Richards, iUd., 1906, 35, p. 510. 



tubes. Traces of substance, which are too 

 attenuated to be caught on any ordinary 

 filter, may thus be estimated. 



The two errors obviated by these simple 

 devices — namely, the presence of residual 

 water and the loss of traces of precipitate, 

 respectively — have perhaps ruined more 

 previous investigations than any other two 

 causes, unless the inclusion of foreign sub- 

 stances by precipitates may be ranked as 

 an equal vitiating effect. But these are 

 merely details ; the scope and method of the 

 recent work on this subject at Harvard (in 

 the course of which thirty atomic weights 

 have been redetermined) may be seen in 

 their full bearing only in the original 

 papers.'' 



That the atomic weights may be con- 

 nected by precise mathematical equations 

 seems highly probable; but although many 

 interesting attempts have been made to 

 solve the problem,* the exact nature of such 

 relationships has not yet been discovered. 

 No attempt which takes liberties with the 

 more certain of the observed values is 

 worthy of much respect. It seems to me 

 that the discovery of the ultimate general- 

 ization is not likely to occur until many 

 atomic weights have been determined with 

 the greatest accuracy. No trouble being 

 too great to attain this end, the Harvard 

 work will be continued indefinitely, and 

 attempts will be made to improve its 

 quality, for the discovery of an exact 



' An important part in these researches has been 

 taken by G. P. Baxter, and many able students 

 also have assisted the author in the work. A com- 

 plete bibliography is given in Fubl. Carnegie Inst, 

 of Washington, 1910, No. 125, p. 91. Most of 

 the papers are reprinted in full in a volume en- 

 titled, " Experimentelle Untersuehungen iiber 

 Atomgewichte, ' ' by the author and his collabora- 

 tors (Hamburg, 1909). The Carnegie Institution 

 of Washington has generously subsidized the work 

 in recent years. 



" See especially Eydberg, Zeitsch. anorg. Chem., 

 1897, 14, p. 66. 



