q66 
Journal of Agricultural Research voi.xxiv, No ijt 
Apparently alcohol used in the quantity and manner prescribed in the 
method has very little, if any, effect on the results. The use of alcohol 
would be more likely to affect the results in the analysis of tobacco than 
in that of extracts. Experience has shown that it is not usually required 
for extracts and is not always necessary for tobacco. 
Alcohol has been found, however, to have a solvent action for extra¬ 
neous matter contained in the ether extract of some tobaccos if it is 
added with the acid and thoroughly shaken, as is done in the acid extrac¬ 
tion. For example, two of the samples in Table II, when alcohol was 
used in this manner, gave the following results: No. 80236 = 7.37 per cent 
and No. 80237 = 6.32 per cent nicotine. On the other hand, it has been 
repeatedly found that when as much as 5 cc. alcohol is added to an 
aliquot after the acid extraction is made, identical results are obtained 
as compared with another aliquot containing no alcohol. However, it 
should not be used except where it is necessary and then only in the 
manner prescribed. 
When a comparison was made of this method with Chapin’s silico- 
tungstic acid method, it was found that higher, and in some samples of 
tobacco abnormally higher, results were obtained by the latter method. 
Much better agreement, however, was shown with extracts. Steam dis¬ 
tillations showed that whereas not over 750 cc. of distillate was usually 
sufficient to obtain the nicotine from extracts, a much larger volume was 
generally required for tobacco, notwithstanding the fact that a smaller 
amount of nicotine might be present in the larger distillate. Again, it 
was found essential that the solution in the distilling flask should be 
maintained at about 15 to 25 cc. in order to distill over the nicotine in 
a minimum volume of distillate. If the volume of liquid greatly ex¬ 
ceeds this extreme difficulty is met with in distilling the nicotine, 
Especially from tobacco. For example, tests with negative results have 
been made for nicotine in a distillate coming from a comparatively 
large solution, although further concentration to the optimum volume 
showed its presence. Another interesting fact observed in the steam 
distillation was that the nicotine results obtained by its use, especially 
with tobacco, were largely influenced by the amount of alkali used in the 
distillation. For instance, it was found that where 2 gm. of sodium 
hydroxid was used in the distillation higher results were always obtained 
than when smaller amounts were employed, despite the fact that in the 
former case the nicotine usually came over in a smaller volume of dis¬ 
tillate. In all distillations the liquid in the distilling flask was alkaline, 
and sufficient alkali was present to liberate combined nicotine. It has 
been shown in this connection that the larger amount of alkali liberates 
some volatile compound other than nicotine’which comes over in the 
distillate and is precipitated by the reagent. This substance occurs in 
variable quantity in tobacco, and a distillate containing it generally has 
some odor and color. The compound does not give the characteristic 
test for nicotine, but forms with silicotungstic acid a dirty white precipi¬ 
tate which becomes pinkish on standing. Where this substance occurs 
in appreciable quantity it can be obtained by distilling with an excess of 
alkali (2 gm.) the exhausted tobacco obtained from the ether extraction, 
or, after a weak alkaline distillation is completed, a further addition of 
alkali up to the above amount and continued distillation will show its 
presence. There is no doubt but that the excess of alkali and heat used 
in the distillation acts to a greater or less extent on the tobacco residue 
