- 4 - 
Kriikoff and Smith (80) in 1937 reDorted that, if a lot of crude 
Loncho carpus roots is a mixture of different species, no sample will 
accurately represent the entire lot. In sampling large, untaled lets of 
a single species, fairly accurate results were obtained by making a care- 
ful ocular estimate of the weight of small, medium, and large roots in 
a given lot and taking a corresponding sample. For smaller lots, such 
as roots from a single plant, the different sizes were sorted out, weighed, 
and the sample was taken accordingly. This method would be impractical 
for sampling baled shipments as they arrive in this country. In the pre- 
paration of samples of fresh roots for analysis more uniform results were 
obtained by dr-ying them to about 12 percent moisture before reducing them 
to slices or shavings. 
Guillaume and Herve (53) in 1939 emphasized the importance of obtain- 
ing a representative sample in the analysis of plants containing rotenone. 
They suggested that a sample of at least 1 kg. of root be taken for p,Tinding. 
The ground material should be mixed and sifted and then mixed again at the 
moment of sampling. 
The Puerto Rico Experiment Station ( 123 ) recently reported that the 
already difficult problem of sampling baled derris root was further com- 
plicated by the variation in rotenone and total-extract content of roots 
of the same diameter found in worK at this station. It was suggested that 
a representative sample of the baled product could be obtained by using a 
small tube with a rotary cutter at the sampling end. Such a tool would 
make it possible to take a fairly large number of small samples and thus 
remove from the bale the minimum amount of root material required for a 
reliable analysis. 
Information on sajnpling was recently furnished by four large import- 
ers of derris and cube root: McCormick and Company, S. B. Penick and 
Company, John Powell and Company, Inc., and Derris, Incorporated (private 
communications). These firms agreed on the matter of sajr.pling about 10 
percent of the shipment, although S. B. Penick and Company stated that 
frequently half, or even all, the bales in a shipment were sampled. John 
Powell and Company stated that where possible they preferred to grind at 
least one-tenth of the shipment. Two sources described the methods employed 
in removing samples of whole root. Derris, Incorporated, stated that 10 
percent of the bales were opened and small bundles of root picked out at 
random from various sections of each bale. This material was ground, mixed, 
and a l-pound sample drawn for testinrf^;;. McCormick and Company removed from 
each of the bales chosen three small bundles of root, and from each bundle 
chopped sections about 4 inches long from the fine ends, the large ends, 
and the middle. These sections were used to make a composite sample for 
analysis. 
Discussion 
The sampling of derris and cube roots for analysis is verj-- important. 
The proper sampling of whole root is extremely difficult, and it is doubted 
if any method is entirely satisfactory. In addition to differences in 
rotenone and total-extract content of different rrots in a shipment due to 
differences in plant source, it has also been shown that fine and coarse 
