Agriculture Reference
In-Depth Information
of experimental error are eliminated. However, in reality, most seed lots are not completely homogenous,
thus subsamples are taken and combined in hopes of achieving a composite sample which is representative
of the quality of the entire lot. Heterogeneity of a lot is relected in the variation in test results on samples
drawn from different containers (bags), or spatial locations from within the lot. Although seed lot variability
may be reduced by thorough mixing, heterogeneity can be introduced through handling, storing, condi-
tioning and the differential interaction of seeds within the lot, e.g., stratiication within containers. Heavier
seeds, for example, tend to settle in the core and the bottom of a storage bin or bag, while lighter seeds are
displaced toward the top and the sides. Such stratiication often occurs during the illing of bags or bulk
containers and may continue during storage.
Every seed lot should be considered potentially heterogeneous. Thus, sampling technique, sampling
design, and sampling intensity are important because they directly affect the accuracy and precision of
the test results. Sampling technique refers to the mechanics of extracting the sample from a lot for testing,
including the type of probe used. Differences among probes can contribute to experimental error (variabil-
ity). Therefore, sampling design and intensity should be selected to represent the conditions of the seed lot
without bias and to provide suficient information to estimate its characteristics reliably. Two tests for seed
lot heterogeneity are given in in the second section of this chapter.
Sampling Technique and Intensity
The manner of selecting individual bags or containers within the seed lot to be sampled should be carefully
chosen. A uniform sampling design should be used for every bag sampled to ensure that a representative
sample is obtained from each. Seed in storage may be stratiied into layers, because heavier seeds tend to
settle toward the bottom of the bags (or containers). Since mixing prior to random sampling is impracti-
cal, each bag should be probed in a way that will represent the entire container. The proportion of the bags
selected for sampling depends on the total number of containers in the lot and is known as primary sampling
intensity. For example, AOSA rules specify that, for lots of ive bags or less, every container should be
sampled. For larger lot sizes, a minimum of ive bags plus 10% of the remaining bags, but no more than
30 bags need to be sampled, regardless of lot size. Subsamples from individual containers are composited,
then subdivided to obtain a submitted sample which meets the minimum size of sample speciied by AOSA
rules.
Subsampling, the division of a sample into smaller units, can also introduce additional variability.
Heterogeneity among subsamples thus can affect the precision of the test results. However, seed testing
rules, procedures, and equipment (dividers) have been developed to minimize subsampling variability.
The Association of American Seed Control Oficials published a “Handbook on Seed Sampling” in 2006.
It includes sampling procedures, guidelines, equipment, recommended, minimum sample size of various
crops and maximum lot size, as well as recommended references and on-line resources. In addition, the
International Seed Testing Association has published a “Seed Sampling Instruction Handbook” which con-
tains similar seed sampling subjects.
APPLICATIon of ToLErAnCES In SEEd TESTInG
A tolerance is the greatest non-signiicant difference between two values (or test results) such as a labeled
value (or irst test result) and a result of a subsequent test. It is used to compare two test results such as
purity or germination and determine whether the difference between them is signiicant or not, i. e., within
tolerance or out of tolerance. Theoretically, tolerances for comparing test results of two sub-samples from
the same submitted sample tested in the same laboratory should be computed allowing for random sampling
variation only. However, practically variation in test results among laboratories, and even within the same
lab is usually greater than that due to sampling variation. Therefore, other sources of variation (or errors)
that are collectively known as “experimental error” should be considered when calculating tolerance tables.
Search WWH ::




Custom Search