INVESTIGATION WITH A ROCK-SALT PRISM. 43 



to 78 per cent at 12/1,. Excepting regions of selective absorption, the 

 gases, for this length of cell, were usually perfectly transparent, so that 

 windows made from plates of rock salt split from the natural crystal 

 were more serviceable and cheaper than highly polished ones, which 

 would have transmitted about 80 per cent of the energy when new, but 

 would have soon become tarnished with moisture. 



Many of the gases show but few absorption bands for the length of 

 cell used. Whether more exist, which are very weak and can be 

 detected only in layers of great thickness, is an unanswered question. 

 Rubens and Aschkinass,^ using a column of CO2 gas 65 cm. long, found 

 the same number of bands as for one 5 cm. long. On the other hand, 

 Langley's curves of the atmospheric absorption show a great com- 

 plexity, which can not be attributed to a mere variety of gases. 



GASES STUDIED. 



Methane. CHi. 

 (Cell 6.3 cm. ; barom., 75.0 cm. ; temp., 22° ; fig. 12. See end of text.) 



The methane used in this work was made by heating a mixture of 

 sodium acetate and soda lime, and washing it thoroughly in pipettes of 

 potassium hydrate and of fuming sulphuric acid, which removes the 

 unsaturated hydrocarbons, like ethylene. The latter is also formed 

 when methane is made by this method. Since methane boils at — 160° 

 and ethylene at — 105°, it is easy to apply the method of fractional lique- 

 faction and fractional distillation. This was done as an additional pre- 

 caution, but no change in the intensity or position of the transmission 

 maxima could be detected. The slight shift of 0.02 /x at 3.4 /* is due to 

 resetting of the prism, and will be noticed in ether, which was also one 

 of the first gases studied. For curve b the gas was purified by lique- 

 faction. 



The spectrum of methane is distinguished by two large absorption 

 bands at 3.31 fx. and y.y [i, and a smaller one at 2.35 fi. 



'Methane was investigated by Angstrom, who found the same number 

 of bands, but since his calibration is wrong beyond 5 /* no comparison 

 can be made. 



The effect produced by substituting four CI atoms for the four H 

 atoms, and thus forming carbon tetrachloride, is very striking. Here 

 we have an entirely different spectrum, with no bands in the region of 

 3 /A to 3.5 ^, characteristic to carbohydrates. In fact, we find no bands 

 until we arrive at 6.5 fi and its prominent harmonic at 13 fi. 



iRubens & Aschkinass : Ann. der Physik, 64, p. 584, il 



