OF ORGANIC SUBSTANCES ON THE RAYS OF THE SPECTRUM. 
267 
propionate, and butyrate. Even in this case two specimens of the same substance, 
with precisely the same boiling point, vary slightly in diactinic quality. Unless very 
large quantities of liquid are operated upon, substances cannot be purified by fractional 
distillation, and even under most favourable circumstances distillation must be several 
times repeated. r l lie remarkably short spectrum transmitted by amyl-butyrate was 
due evidently to some impurity, since the behaviour of the substance on dilution with 
alcohol was quite in accordance with this view. The generally transparent character of 
the alcohols and ethereal salts is remarkable. 
Normal Hydrocarbons. 
Only three of these—hexane, heptane, and octane—have been examined ; they were 
specimens kindly lent us by Prof. Schorlemmer, and believed to be perfectly pure. 
Three series of photographs were taken of them, but they yielded spectra in no way 
corresponding to their difference in composition, from which we are inclined to believe 
they were impure (see Diagram No. 2). 
Compound Ammonias. 
A series of five of these bodies was examined, the solutions containing 33 per cent, 
of the substances in the case of ethylamine and methylamine. It is believed that, 
with the exception of ethylamine, they contained some impurity, since the methyl- 
amine and climethylamine yielded shorter spectra than the ethylamine. These were 
commercial samples, and most probably were prepared from ammonia obtained from 
gas-liquor. M. Soret has shown that commercial ammonia, even after many recrystal¬ 
lisations as sulphate, still shows an absorption band, due to some impurity which is 
constantly present; it is therefore scarcely likely that pure products would be obtained 
unless specially prepared ammonia be used for the purpose. We considered that the 
volcanic ammonia, which is free from some of the impurities present in that made from 
gas-liquor, might be perfectly diactinic. Three separate samples, each measuring half 
a gallon, were examined, with the result that all the rays beyond 17 Cd were cut off. 
Even when so dilute that only one volume of ammonia was contained in 64 of water, 
all traces of absorption had not been destroyed (see Diagram No. 5, Plate 25). This 
absorption is not due to nitric acid, because M. Soret has shown that the absorption 
band due to that substance occurs between 12 Cd and 17 Cd. 
Conclusions. 
(1.) The normal alcohols of the series UTU-iOH, are remarkable for transparency 
to the ultra-violet rays of the spectrum, pure methylic alcohol being nearly as much 
so as water, 
2 m 2 
