App4tutorial

back to FRSA home
back to July 13 workshop
back to Seed Analyst Fact Sheets

AOSA Rules for Testing Seeds Appendix 4 Tutorial prepared by FRSA for July 13, 2001 workshop.

Normal type is the exact wording of the AOSA appendix 4. Bold type indicates additional explanation and examples. (ALM)

APPENDIX 4 TUTORIAL:

DETERMINATION OF WEIGHT OF WORKING SAMPLES FOR KINDS NOT LISTED IN TABLE 1

To determine the weight of the purity working sample and its corresponding noxious-weed seed working sample the Weight Determination Method will be followed.

Use Appendix 4 to figure out, in a statistically correct manner, weights for species that are to be added to Table 1 or use appendix 4 whenever you need purity weights for something not listed in Table 1.

From the working sample count out at random, by hand or with a germination counter, eight replicates, each of 100 seeds. Weigh each replicate in grams to four significant figures (see Appendix 2). With small-seeded kinds it is not necessary to weigh the replicates with greater precision than four decimal places.

Key points:
·    At least eight replicates of 100 seeds each. (The appendix is not specific about how many lots to use for establishing weights for a rule proposal but, conventionally, data from at least 8 samples are submitted.)
·    Seeds must be representative.
·    Four significant figures or four decimal places.
Examples: 10.25 1.025 .1025 .0102

Example of seed weights
Species: Dimorphotheca sinuata DC. - stick type seed

Weight of 100 seeds (8 replicates)
Lot No.    1    2    3 4    5    6    7    8
1    0.2224 0.2134 0.2130    0.2116    0.2145 0.2071 0.2116 0.2125

Calculate the variance, standard deviation and coefficient of variation as follows:

Variance = n (Ex2)- (Ex)2
   n(n-1)

Where x = weight of each replicate in grams
   n = number of replicates (8)
   E = sum of

Variance = n (Ex2)- (Ex)2
   n(n-1)
_______________________________________________________________________

Ex2 = (0.2224)2 + (0.2134)2 + (0.2130)2 + (0.2116)2 + (0.2145)2 + (0.2071)2 + (0.2116)2 + (0.2125)2
Ex2 = 0.04946 + 0.04554 + 0.04537 + 0.04477 + 0.04601 + 0.04289 + 0.04477+ 0.04516

Ex2 = 0.36398
_______________________________________________________________________

Ex = 0.2224 + 0.2134 + 0.2130 + 0.2116 + 0.2145 + 0.2071 + 0.2116 + 0.2125
Ex = 1.7061

(Ex)2 = 2.91078
_______________________________________________________________________

n (Ex2) - (Ex)2 = 8 (.36398) – (2.91078) = 2.91184 – 2.91078 = .00106 = .0000189
   n(n-1)    8 (8-1)    8(7) 56

variance = 0.0000189
______________________________________________________________________
   ________
Standard deviation(s) = √Variance

__________
Standard deviation(s) = √ 0.0000189    = 0.004296

Coefficient of variation = S x 100
   x
_
where x = mean weight of 100 seeds

_
x = 1.7061 ¸ 8 = 0.2133

Coefficient of variation = 0.004296 x 100 = 2.014
0.2133
If the coefficient of variation does not exceed 6.0 for chaffy seeds, or 4.0 for other seeds, the result of the determination can be calculated.

If the coefficient of variation exceeds whichever of these limits is appropriate, count and weigh a further eight replicates and calculate the standard deviation for the 16 replicates. Discard any replicates that diverge from the mean by more than twice the standard deviation so calculated.

For the Dimorphotheca example, here are the rest of the samples and the calculations:

Lot No. 1    2 3    4    5    6    7 8 mean    variance SD Var. Coeff.
1    0.2224    0.2134    0.2130    0.2116    0.2145    0.2071    0.2116    0.2125    0.2133    0.0000184570 0.004296    2.014492
2    0.2111    0.1918    0.2017    0.2021    0.1991    0.1950    0.1999    0.2003    0.2001    3.19336E-05    0.005651    2.823725
3    0.2229    0.2106    0.2117    0.2205    0.2223    0.2200    0.2130    0.2153    0.2170    2.45827E-05    0.004958    2.28444
4    0.2500    0.2450    0.2362    0.2446    0.2468    0.2560    0.2381    0.2549    0.2465    5.08057E-05    0.007128    2.892194
5    0.1908    0.1896    0.1925    0.1912    0.1794    0.1952    0.1896    0.1886    0.1896    2.12584E-05    0.004611    2.431634
6    0.2076    0.2113    0.2003    0.2105    0.2055    0.2053    0.2021    0.1964    0.2049    2.59107E-05    0.00509    2.484567
7    0.2334    0.2354    0.2271    0.2370    0.2210    0.2224    0.2143    0.2252    0.2270    6.18879E-05    0.007867    3.465969
9    0.2630    0.2750    0.2767    0.2615    0.2704    0.2782    0.2612    0.2696    0.2695    4.76743E-05    0.006905    2.5625
11 0.1908    0.1981    0.1907    0.1990    0.2011    0.1901    0.1894    0.1944    0.1942    2.13657E-05    0.004622    2.380178

   mean    0.2180

The coefficients of variation for these samples are all less than 4.0. If any were above 4, an additional 8 replicates should be counted out and weighed. Use the 16 replicates to calculate a new standard deviation. Follow the instructions in the appendix for deciding if any of the replicates should be discarded. Double the standard deviation. Add the doubled standard deviation to the mean. This number is the high limit for any replicate. Subtract the doubled standard deviation from the mean. This number is the low limit for any replicate. If any replicate falls outside these upper and lower limits, it can be discarded. The coefficient of variance calculations should be made on the remaining replicates.

To calculate the minimum weight for purity analysis (grams): multiply the mean weight of 100 seeds by 25 (2500 seed weight).

Mean 100 seed weight = 0.2180 grams
0.2180 x 25 = 5.45

5.45 grams is the estimated weight of 2500 seeds.
This may be rounded to 5 grams.

Weights may be rounded down to the nearest half gram if over one gram. Under a gram the weight may be rounded to the nearest tenth of a gram (not to fall below 0.2 grams).

To calculate the minimum weight for noxious weed seed or bulk examination (grams):
multiply the minimum weight for purity analysis by 10.

5 x 10 = 50 grams for the noxious weed exam.

To calculate the approximate number of seeds per gram: divide 100 (the number of seeds) by the mean weight (of the 100 seed replicates).

_100__ = 459 (approximate number of seeds per gram)
0.2180

To calculate the approximate number of seeds per ounce: multiply the approximate number of seeds per gram by 28.35.

459 x 28.35 = 13,013 (approximate number of seeds per ounce)

For species which have small and large seeded varieties, more than one morphological seed form, or have seed weights which may vary with processing, counts should be made for each type.

Example: Dimorphotheca sinuata has two types of seeds: stick and flake types. As a result the entry in table 1 is as follows:

Table 1. Weights for working samples.

   Minimum    Minimum    Approximate Approximate
Kind of seed    weight for    weight for    number of number of
   purity noxious    seeds per seeds per
   analysis weed seed or gram    ounce
   (grams)    bulk examination
(grams)
Dimorphotheca
sinuata de Candolle
African-daisy
cape-marigold, blue-eyed
stick type seed    5    50    459    13,013
flake type seed 4    40 591 16,755

Spreadsheet programs like Microsoft Excel work very well for doing these calculations. Mean, sum, variance and standard deviation are formulas available in the program for doing these calculations quickly.

back to FRSA home
back to July 13 workshop
back to Seed Analyst Fact Sheets