566 PROCEEDINGS OF THE AMERICAN ACADEMY. 



possible only within recent years since the freezing and boiling point 

 methods for the determination of molecular weights have been known. 

 Then it is not possible to determine the vapor densities of these hydro- 

 carbons, because, as I have recently ascertained, even in vacuo small 

 quantities of the hydrocarbons such as are used in vapor density deter- 

 minations, undergo serious decomposition; and this occurs even in oils 

 that have been distilled many times over in hulk in vacuo. In the ear- 

 lier work, it was evidently assumed that a few distillations under atmos- 

 pheric pressure were sufficient to collect the individual hydrocarbons 

 within the limits of their boiling points, sufficient at least to afford 

 reliable data as to their composition. A glance at the small differences 

 in percentage composition is sufficient to show that it is not possible by 

 analysis alone of products even well purified to distinguish between 

 homologous members of a series, although such analysis may define the 

 series. 



But the chief difficulty is to obtain each hydrocarbon uncontaminated 

 by any admixture of its homologues or by products of decomposition. 

 This is well illustrated by the experience of MarkownikofF in separating 

 the hydrocarbons in the Russian oil, who found it impossible to collect 

 distillates closer than limits of five degrees on account of decomposi- 

 tion. That the same is true perhaps in a less degree in distillates from 

 Pennsylvania oil is evident whenever distillation is made of the higher 

 portions under atmospheric pressure. The rank odor is evidence of 

 cracking. Yet the constituents with higher boiling points are under the 

 influence of vapor tension as much as the constituents with lower boil- 

 ing points, and consequently require as prolonged distillation for com- 

 plete or approximately complete separation. In my experience it is 

 only possible to obtain even an approximate separation by exclusion of 

 air and depression of boiling points. 



In this manner the higher hydrocarbons may be distilled any number 

 of times with no appreciable decomposition. The only limit is the 

 patience of the operator. But the stability of these hydrocarbons is 

 evidently dependent on the influence of mass. Since as mentioned 

 above, while distillation of any considerable quantity of the oil may be 

 carried on indefinitely, a limited quantity cannot be volatilized even in 

 vacuo without decomposition. Many attempts to determine the vapor 

 density of the hydrocarbons in Pennsylvania and California petroleum 

 by volatilization in vacuo according to the method of Lunge and Neu- 

 berg have failed on account of cracking, even so far as the separation 

 of sooty carbon from the members with high boiling points. 



