DESCRIPTION OF APPARATUS AND METHODS. 2/ 



cross-section paper. In this manner many slight variations in the trans- 

 mission curve could be verified by repeating the observation before 

 setting for a different wave-length, and, in going through an absorp- 

 tion band, the spectrometer settings were made more frequent, thus 

 increasing the accuracy of its location. 



That this method has not been prejudicial against the results obtained 

 is shown in the transmission curves of the gases in which it was neces- 

 sary to find the transmission of the cell and the included gas, and then 

 deduct the absorption of the empty cell, in order to obtain the transmis- 

 sion through the gas itself. 



The manner of recording the observations is shown in Table III,^ 

 which also shows the peculiar distribution of the energy in the spectrum 

 of the " heater " used as a source of radiation. In the subsequent tables 

 only the " transmission " of each substance examined is given. 



At the beginning and at the conclusion of the exploration of each 

 compound the sodium lines were thrown upon the radiometer slit to 

 determine whether the apparatus was still in adjustment. 



SOURCES OF ERROR. 



It is a difficult matter to determine the accuracy attained in this work, 

 as well as the sources of greatest error. With the smaller spectrometer 

 and rock-salt prism it was not difficult to find one's work in agreement 

 with that of previous investigations on absorption spectra. But the 

 emission band of COo, which Paschen^ and others found at 4.40 ij., by 

 using a fluorite prism, which has about twice the dispersion of rock salt 

 in this region, was sometimes found at 4.42 ju,, or a difference of about 

 10" in arc. This was found to be due to the curvature of the slit, as 

 explained. With the large spectrometer, which gave about twice the 

 dispersion of the small one, and accidentally equivalent to the fluorite 

 dispersion of Paschen and of Ransohoff,^ the emission maximum came 

 exactly at 4.40 /* for both Bunsen-gas and Bunsen-acetylene flames. 

 This is the best defined, the most accurately known, and the most easily 

 obtained comparison line in the infra-red. The estimation of the degree 

 of accuracy of this band is based upon the number of observers who 

 have located it. The next in order of accuracy of location is the 3.43 ^ 

 absorption band found in compounds containing CHg-groups, by Julius,^ 

 Ransohoff, and others. This band is not quite so sharp as the COo 

 emission band. Using these as comparison bands the 6.86 /a band, found 

 so frequently in the present work, has, in the writer's estimation, been 



^Given at the end of the text. ^Ransohoff, loc. cit. 



Taschen, loc. cit. *Julius, loc. cit. 



