Agriculture Reference
In-Depth Information
Figure 13.2. Noxious weed tolerances follow the Poisson distribution, which is non-symmetric with a posi-
tive skew. As the population mean (i.e., number of noxious weed seeds) increases, the distribution becomes
nearly symmetric (normal) (from Elias et al., 2000).
assumption, the tolerance limits can be obtained based on the number of noxious weed seeds in the original
test, or labeled rate.
Finally, in case of noxious weed seeds (or incidental foreign seed contamination), the one-sided toler-
ance limit in sampling from the Poisson distribution is used (Dodge and Canield, 1971; Elias et al., 2000).
AoSA Tolerances for noxious Weed Seeds
The AOSA tolerances for noxious weed seed are found in Table 13F, AOSA rules and are adapted from
Elias et al., 2000. Table 13F, AOSA (Table 13.3 in this topic) is mostly used for regulatory purposes to
determine if a second test has signiicantly more noxious weed seed than that stated on a label or found in
a irst test. Tolerance values in Table 13F are based on a one-way test at the 5% probability level and are
not for tests made on two different samples drawn from the same seed lot tested in the same or in different
laboratories. The tolerances in Table 13F are determined by entering the number of noxious weed seeds
listed on the label in column X; and column Y shows the maximum tolerated difference allowed between
the number of noxious weed seeds listed on the label and the number of noxious weed seeds found in the
second test. If the number of the noxious weed seeds found in the second test is equal to or less than the
number in column Y, then the labeled value and the second test results are within tolerance.
ISTA Tolerances for determining the Maximum difference in the numbers of other Seeds by number
The ISTA rules contain two tolerance tables for what are considered “other seed” counts when occurring
as contamination in seed lots. These are shown in Tables 4.1 and 4.2 of the ISTA rules, 2010. Table 4.1
shows tolerances for determination of other seeds by number when tests are made on the same or different
submitted sample in the same or different laboratory. These tolerances are based on two-sided tests at the
5% signiicance level; thus, they can be expected to lead to incorrect decisions 5% of the time, 2.5% on the
high side and 2.5% on the low side. Both the irst and second tests are given an equal chance of being cor-
rect; thus, they are averaged for use in selecting the correct tolerance. These tolerances were taken directly
from Table F1b in Miles (1963).
