182 TRAXSACTIOXS 01" THE [maKCU 15, 



aiul the anomalous behavior of the mixed earths yields })heiu)m- 

 eiiu •' without precedent," ' 



About the same date as the later communication by Crookes, 

 Lecoq de Boisbaudran' published a method of obtaining what he 

 terms "reversion spectra," which is pi-actically the same in effect 

 as that of Crookes. Tlie French savant linds indications of two 

 new elements in certain brilliant lines, but Ci'ookes distinctly 

 warns us that 'inferences drawn from spectrum analysis per se 

 are liable to grave doubt,^' and "chemistry after all must be the 

 court of final appeal." Crookes' reflections on the sufficiency of 

 spectrum observations as criteria of the elementary cluiracter of 

 l)odies are justified by the experience of many, notably of Sorby, 

 whose pscudo-jargoniiim is well rememberccl. Tiiis difllculty 

 arises especially with absorption spectra, and neglect of tl)e warn- 

 ing given by vSorby has led several chemists into fruitless re- 

 searches. 



(5.) When DaltoUj the Manchester schoolmaster, added to 

 t!ie atomic theory of the Greeks the laws of definite and of mul- 

 tiple pro]iortions, he transformed an ''interesting intellectual 

 j)hiytliing " into an exact scientific theory ca})able of experimental 

 demonstration. The importance of ascertaining tlie atomic 

 weights of the elements with the utmost accui-acy has stimulated 

 chemists to apply to the problem their best endeavors ; and as tlie 

 metliods of analysis become more refined the determinations are 

 again and again repeated, every ascertainable and imaginable 

 source of error being carefully eliminated. Beside the exi)eri- 

 mental repetitions, the figures obtained by various observers have 

 recently been submitted to careful re-calculations by Clarke," in 

 .his country, and soon after by Lothar Meyer and 8eul)ert,^ 

 in Germany. Their labors give chemists the latest and most re- 

 liable constants. 



The prevailing, though partly unacknowledged, adherence to 

 Front's hypothesis, leading chemists to prefer whole numbers (or 

 at least eVen fractions) for the atomic weights, is liable to result 

 in confusion and perplexity. Stas demonstrated that the atomic 

 weight of oxygen is not quite sixteen times as great as that of 

 liydrogcn, but that when 11 = 1, = 15.96. The tendency to dis- 

 regard this difference of j-J-g- is unfortunate, since important errors 

 in calculations, based on organic analyses, might result there- 

 from. Lothar Meyer and Seubert show that in the analysis of 

 compounds of carbon and hydrogen, the error introduced by 

 making = 16 is greater tlian the errors of observation, and in 

 the analvsis of a body belonging to a homologous series doubts 

 might arise as to the identity of the body under examination." 

 Of course, the formula of a body is not determined by analytical 

 data alone; still, this liability to errors marks forcibly the desira- 



