58 BULLETIN 1003, U. S. DEPARTMENT OF AGRICULTURE. 
chanical agitation with a paddle-wheel stirring device was the first 
resort. It was subsequently found, however, that heating over a 
flame in a distilling flask fitted with return-flow condenser is equally 
- effective and much simpler in execution. This method of treatment 
thoroughly emulsifies the oil and caustic solution, giving the intimacy 
of contact desired, while the inverted condenser continually returns 
the aldehyde bodies to the action of the alkali until they have been 
changed to the nonvolatile products previously discussed. The in- 
verted condenser is then replaced by a Hempel column and the con- 
tents of the flask distilled with steam, yielding from the start a tur- 
pentine of standard requirements. 
Steam distillation is admirably adapted to the production of tur- 
pentine of uniform quality, because it affords a simple means of con- 
trol, in that the ratio of oil to water in the distillate is an index of 
the composition of the turpentine (12). This is a gradually dimin- 
ishing ratio in proportion as the oil contains less pinene and corre- 
spondingly more of the higher-boiling pine oils... For any observed 
oil-to-water ratio, however, the turpentine has a definite composition, 
as indicated by its density, refractive index, distillation-temperature - 
limit, etc. This, of course, follows from the law of relative vapor 
pressure of immiscible liquids. Its application as a simple and re- 
markably accurate means by which to judge the composition of the 
turpentine at any time during the distillation, however, has not 
been given the consideration it merits (12) as a means of standardiz- 
ing the output of commercial plants. Properly used, the oil-to-water 
ratio makes possible the production of turpentine having a constant, 
predetermined composition, any consignment of which will be prac- 
tically the same as a preceding or subsequent shipment. 
Following up preliminary observations, based on. the considera- 
tions set forth, a series of experiments was conducted to determine: 
(a) The relative efficiency of caustic soda, carbonate of soda, and 
milk of lime as refining agents; (0) the proportion of alkali to crude 
oil and concentration of the alkali solution giving the best results; 
(c) the time necessary for the reactions set up by the alkali treat- 
ment to produce its full effect; (d) the effect of drawing off the alkali 
after treatment and washing the oil with water before distilling; 
‘ (e) the effect of passing a current of air through the oil during treat- 
ment with alkali. arr | 
In carrying out these experiments 500 cc., taken from a large com- 
posite sample of crude western yellow-pine turpentine, were used in 
each test. The turpentine fraction proper was continued to where 
the ratio of oil to water was 4 to 6, beyond which the proportion 
changes rapidly, and a second turpentine fraction collected between 
the 4 to 6 and 3 to 7 ratios. The distillation was continued for the 
