GENERAL DISCUSSION OF THE SPECTRA. II3 



Spectroscopy it is remarked that such a search is delusive, and GriJn- 

 wald's mathematical spectrum analysis^ is cited as an example in which 

 all wave-lengths of the so-called compound spectrum of hydrogen can 

 be converted into corresponding wave-lengths of the water-vapor spec- 

 trum by multiplying by 0.5. Recent and more exact measurements on 

 these lines show that no such relations exist. 



Schuster- speaks of the iron spectrum, which has two lines which are 

 in the ratio of 2 to 3, while hydrogen has lines in the ratios of 20 : 27 : 37. 

 He demonstrates that in accordance with the Theory of Probability a 

 certain number of coincidences between lines of two spectra might be 

 expected to occur, even if the spectra be quite unrelated. Furthermore, 

 there appears to be a tendency for functions formed by two lines to 

 cluster around harmonic ratios, and, " most probably, some law hitherto 

 undiscovered exists which in special cases resolves itself into the law 

 of harmonic ratios." Of course, as is well known now, the nearest 

 approximation to such a law is Balmer's Law and the numerous other 

 convergent series formulae used by Kayser and Runge and others. 

 Nevertheless, in spite of these warnings, and fully realizing the danger 

 from lack of dispersion, experimental errors, etc., I venture to call 

 attention to certain marked absorption bands which occur so frequently 

 in positions which so closely fulfill this relation that it is necessary to 

 examine more fully into the probable significance. 



Abney and Festing (loc. cit.) found that compounds having CH3- 

 groups have a band at 0.74 ix and another between 0.907 and 0.942 /*, 

 while benzene and CH compounds have a band at 0.867 /*• Puccianti 

 (loc. cit.) found a band at 1.71 ix for all compounds in which the C atom 

 is joined directly to the H atom in the molecule. Aschkinass (loc. cit.) 

 found the absorption bands of water at 1.5 1 11, 3.06 /i, and 6.1 /£, while 

 Paschen (loc. cit.) found them at from 2.916 to 3.024 /x and at 6.06 fi, 

 which values are closely harmonic, Ransohoff found closely harmonic 

 bands for alcohols at 1.71 and 3.43 /a. 



In the present work, using a quartz prism, the first band occurs at 

 from 0.83 to 0.86 fx, while the second one oscillates between the values 

 1.66 /i, for the thiophene and 1.73 /u. for ethyl succinate. The next dis- 

 turbance is in the region of 3.4 ^, the maximum being at 3.25 /* for ben- 

 zene and 3.43 fi for compounds rich in CH2 or CHg groups. 



The next region where there is a constant recurrence of bands is at 

 6.75 /A for benzene and 6.86 n for other compounds rich in CH2 or CHg 



'Griinwald : Wien. Ber., 96 to loi, 1887 to 1892. 

 'Schuster : Proc. Roy. Soc, 31, p. 337, 1881. 



