In sampling small shipments it was proposed thatl pound of roots "be 
withdrawn at random from a tals of 200 po^inds. In the cas« of a lar^^e 
shipment it was suggested taat 10 perceiit of the "bales "be sani-nled and 5 
percent "by v/ei:;ht of each of these "bales "be removed. The roots v;ere to 
be chopped into short lengths and quartered until 1 poi.ind remained. 
Cahn and Bo am (13) in 1955, in an article on the determination 
of rotenona, made the following: statement in regard to sampling: 
"Derris root should pass entirely through a 50-mesa sieve 
"before it is sampled. Eoot, as ordinarily f;round, frequently con- 
tains a proportion of coarse, spicular material, through which 
the pov/der (often extremely fine) tends to filter, thus ceusinf:; 
an undue amount of coarse particles to "be present in' tne top 
layers. If the a-'^ount of sioicular material is considerable, 
the whole baton should "be sieved- through a 50-mesh sieve, ard 
the samples taken from proportionate aT.cunts of tne fine and 
coarse n'aT^erial . " 
Gahji (JL2) again in 1936 stressed the necessity of proper sampling. 
He found that, of 10 "bales analysed cut of a consignment of 40 "bales 
from the same estate, the total extract ranged from 15 to P5 percent. 
The only satisf actor;.'- method of sa.mplin.e derris root was said to "be to 
grind it first, mix i^ thorough'.ly , arc then anal'-ee the mixed fii^e powder. 
Gecrgi (Sl) in 1937 described a method used in sampling kiln-dried 
root during the process of "baling. Vrnen half the quantity required for 
a "bale had "been loaded into the press, a"bout c. pounds of the root v/as 
taken at random from the heap and set a,side as a saraple from that "bule. 
When 20 camples, say 40 rounds, ha,d ."ccimulated, th^^ bulk sample of root 
was mixed, spread cut in a thin layer .on the floor, an'-"' qua^rtered until 
a"bout 1 pound remained. This amoimt was cut into 1-inch lengths and 
quartered until about 3 ounces remained. This quantity was further cut 
into 1/4-inch lengths and ground in the la"bcratory mill for ar^alysis. 
In one ca.se 25 samples covering 497 "bales v/ere analysed. 
In 1937 Levallois (62) stated that the sampling of rotenone- 
"bea^ring roots was very delicate, as the rotenone ■ content of a sample 
depended on the a'Dunn^nce of long fibers. Progressive 5:rinding of roots 
gave fractions with rotenone contents as follov/s: Fria"tle parts and 
short fi'ijers 6, mediiim parts 4 to 4.5, and long ii"bers 3 to 4 percent. 
Tnis author did not take samples of derris or cube at random, but 
classed root in three or four categoriea according to size and sampled 
accordingly. From each lot he took an aliquot part of l/lO to 1/20 and 
chopped, up these parts. From each pa.rt he took a new fraction. These 
fractions miide up the samples for analysis. 
Chevalier and Chevalier (20) recommended that, since large and small 
roots differ in rotenone and total-extract conte'nt, samples taken for 
analysis should have a'bout the same proportion of large and small roots as 
the whole lot. 
