GENERAL DISCUSSION OF THE SPECTRA. IO5 



shift noticed in methyl and ethyl iodide, which had been examined on 

 different dates, several months intervening, was found not to exist, 

 showing an instrumental error. This method of testing a series of 

 compounds at one region of the spectrum on the same day is the only 

 way to be certain of slight differences in wave-length. With the 25 

 pure distillates of petroleum, already mentioned, belonging to the series 

 CnHgn-a, CnHgn, and CnH2n+2, a final test was applied to this perplexing 

 question. The absorption spectra of two of these, octane (CgHig) and 

 tetracosane (CgiHgo), are to be noticed in figs. 63 and 69. 



In these, as well as in the intermediate ones, no shifting could be 

 detected, although the greatest efforts were made to do so. This was 

 not a little surprising, for, according to the measurements on the alco- 

 hols shown by Schonn (loc. cit.) in the visible spectrum, a shift for this 

 greater number of CH, groups should have occurred. For, even if we 

 assume that the shifting is least for the infra-red and increases as we 

 approach the ultra-violet, unless the total shift for this increase of 16 

 CH,-groups [octane, C8Hi8=CH3(CH2)6CH3; tetracosane, £^,11^^= 

 CH3(CH2)22CH3] is less than o.oi /i, it is safe to assume that no shifting 

 occurred. At least, if there is a shifting, it is less than o.oi fi, which is 

 much smaller than one would anticipate from observations in other 

 regions of the spectrum. 



A shift of O.OI fi at 3.43 fi is 8" of arc on the spectrometer circle, and 

 at 6.86 fi it is 12" of arc for the small spectrometer, while on the larger 

 apparatus the values are twice as great, viz, 16" and 24" of arc, so that 

 it would have been impossible to escape detection, especially in the case 

 of such deep, well-defined bands as these. It is to be remembered that 

 a separate examination was made of the 3.43 fi and the 6.86 /x bands for 

 CigHge and C24H48, using the large spectrometer. This was done to 

 avoid the possibility of a shifting while examining the whole spectrum, 

 and, instead, only 7 spectrometer settings, at intervals of i', were made 

 for each of these two bands. First, the zero of the instrument was 

 determined by means of the sodium flame ; then the 3.43 /t region was 

 examined for CigHge, the examination being from right to left, say ; the 

 C24H48 was then examined going backward, from left to right, and then 

 from right to left, after which the zero was again tested by means of the 

 sodium flame. The 6.86 /* band was examined in the same manner, but 

 in neither case could a shifting of the band be detected, as will be noticed 

 in the curves (fig. 74) and tabulated data. These two bands are so sharp 

 that if a shifting of o.oi fi had occurred it could have been detected. It 

 is true that these bands vary by o.oi fj. for some of the other oils, but that 

 is to be ascribed to errors in reading the curves, as well as to the fact 



