Agriculture Reference
In-Depth Information
Table 5.2 (not included in this topic) shows ISTA tolerances for germination tests on the same or dif-
ferent submitted sample when tests are made in the same or different laboratory on 400 seeds (two-way
test at 2.5% signiicance level). Table 5.3 (not included in this topic) shows ISTA tolerances for the same,
except these are for one-way tests at the 5% signiicance level. Either of these two tables may be used,
depending on the philosophy of law enforcement and the level of signiicance desired. In a practical sense,
use of Table 5.2 will result in only 1.25% rejection rate (of truthfully labeled seed lots), since only those
lots germinating below the labeled rate will be of concern. Consequently, this tolerance favors the seed
producer since (theoretically) only 1.25 of all correctly labeled seed lots should be declared (falsely) to be
out of tolerance and the chance of Type-II error would increase. Table 13.6 is similar to Table 13J, AOSA
germination tolerances which are philosophically balanced between the interests of both producer labeling
the seed and the consumer. Use of this tolerance should lead (theoretically) only to a 5% rejection of truth-
fully labeled seed lots; thus, the interests of both the producer and the consumer are addressed.
ToLErAnCES for THE TETrAzoLIuM TEST
Both AOSA and ISTA rules have tolerance tables for the tetrazolium (TZ) test. The AOSA has the follow-
ing tables:
Table 13.7 (Table 13L, AOSA). Maximum tolerance values for comparing two tetrazolium tests from
the same or different submitted samples tested in the same laboratory (2-way test at
P = 0.05
).
Table 13.8 (Table 13M, AOSA). Maximum tolerance values for comparing two tetrazolium tests from
different submitted samples tested in different laboratories to determine if a second test is signiicantly
poorer than a labeled value of a irst test (1-way test at
P = 0.05
).
Table 13.9 (Table 13N, AOSA). Maximum tolerance values for comparing two tetrazolium tests from
the same or different submitted samples tested in different laboratories to determine if one test is signii-
cantly better or poorer than another test (2-way test at
P = 0.05
).
Table 13.7 is appropriate for internal laboratory training and quality assurance purposes. Table 13.8 is
appropriate for seed law enforcement purposes. Table 13.9 is appropriate for proiciency and referee stud-
ies. All three tables are based on Kruse (2005), “How to establish tolerance tables for speciic seed testing
stations.” Seventh ISTA Seminar on statistics, University of Hohenheim, Stuttgart, Germany.
In addition, Table 13I in the AOSA is the “Maximum tolerance values between two and four replicates
of 100 seeds in a tetrazolium test (2-way test at P = 0.025),” adapted from Table G1 (Miles, 1963).
The ISTA rules has two tolerance tables for the TZ test: Table 6.1. Tolerances for tetrazolium viability
tests on the same or different submitted sample when tests are made in the same laboratory each on 400
seeds (two-way test at 2.5% signiicance level); and Table 6.2. Tolerances for tetrazolium viability tests on
two different submitted samples in different laboratories each on 400 seeds (one-way test at 5% signiicance
level).
Examples
(From AOSA rules)
A new analyst found a 200-seed TZ test of sugar beet to be 89%. An experienced analyst at the
same seed lab found the results of another 200-seed subsample from the same submitted sample
to be 78%. Are these two results within tolerance?
The average of the two TZ tests = (89 + 78)/2 = 84% (83.5 % rounded)
Difference between the two TZ results = 89 - 78 = 11%
In Table 13.7, Column D, for 84% average of two TZ tests of 200 seeds each, the tolerance is 8.
The difference between the two test results is more than the tolerance values; therefore the differ-
