- 31 - 
Seaber jilso investigated the possibility of making "use of the 
maximmi of the ultraviolet absorption curve to estimate the percentage of 
rotenone in the crude solvate. Sxtreme dilution was necessary, but never- 
theless results close to those by polarisation were obtained in some cases, 
although in others the values were too high. The method could not be used 
as a routine process for purity but might be useful in identifying and 
roughly estimating rotenone present in small amounts in mixtures. An 
attempt to estimate rotenone by this method in whole extracts of the root 
was not successful because of the effect of imp\irities. 
Jones (65_) in 1937 studied the crystallization -f the rotenone- 
carbon tetrachloride solvate from extracts and proposed a modifier! loro- 
cedure for thif. step in the rotenone determination. The proposed crystal- 
lisation procedure was as follows: 
The solvent-free extract from a 25-gii. sample of root is 
dissolved in 25 cc. of carbon tetrachloride, cooled in an ice 
bath, and seeded with crystals of rotenone-carbon- tetrachloride 
solvate. If only a small amount of crystalline m-.terial sepa- 
rates, an accurately weighed, amount of rotenone is add.ed, so 
that at least 1 gm. of pure rotenone is present. This extract 
and a wash solution having 0.27 gm. of rotenone for 100 cc. of 
carbon tetrachloride are maintained at 0° C. in an ice bath 
overnight. The crystals are then filtered in a tared Gooch 
crucible, washed with 6-10 cc. of ice-cold wash solution, and 
dried to constant weight at 40'^ C. One gram is treated with 10 
cc. of alcohol saturated witn rotenone at room tem.perature and 
set aside at this temperature for 4 hours. This material is 
filtered throu^gh a tared C-oocn crucible, washed with about 
5 cc. of alcohol saturated with rotenone at the same tempera- 
ture, and dried to constant weight at 105°. Corrections are 
made of 0.07 gm . for the rotenone dissolved by the 25 cc. of 
carbon tetrachloride used and also for any rotenone added. 
The purity of the solvate was found to deoend principally upon the propor- 
tion of rotenone to total extract and upon the proportion of solvent used 
in crystallization. As the so-called ".pure" rotenone obtained "oy the 
alconol treatment was' not entirely pure, and as its purity depended on 
the purity of the solvate from which it was prepared, it was desirable to 
obtain tne solvate in as nearly piare a form as possible. lor this reason 
the author preferred to crystallize the solvate from a larger proportion 
of solvent than that used by Cahn and Boam (l3) . '.'ore rapid conversion 
of the solvate to rotenone was obtained when 10 cc. of alcohol per frarn of 
solvate v;as used rather than the 5 cc. proposed by Cahn and Boam. 
A determination of the precision of replicate results on one sample 
of derris of about 4-percent rotenone content showed a standard deviation 
of i 0.05 percent. A' study was ma.de of trie accuracy of the proposed 
crystallization procedure when applied to extracts of both derris and 
cube roots v;ith various proportions of rotenone to total extract. The 
method used assumed that the ixonrotenone portion of the extract exerts only 
a retarding effect on the crystallization and has little or no actual sol- 
vent effect on the rotenone, an assumotion tnat was indicated by all the 
