o4. 
DUPLICATE GERMINATIONS., 
In order to test the value of a single germination, or in other 
words the variation in germination between presumably equally good 
seed, a series of duplicate germinations was made with some Chris- 
tiana melon seed of known age, and of careful selection. ‘The seed 
was grown at Framingham, Mass., in 1879, and had been carefully 
winnowed, and represented the product from about 1200 melons 
carefully selected as on ile and of superior quality. The test was 
Sept., 1883, when four years of age. The germinations were dupli-: 
cates In every respect, the oe occupying the same apparatus, ex- 
posed to the same moisture and the same temperature. ‘The first 
germination in the twenty-three samples of fifty seed each was in all 
cases on the second day, half had gerimnated in twenty-two of the 
trialson the third day, and the last seed germinated in from four to 
eight days. The average germination was 78 per cent., the extremes 
of the trial were 66 and 86 per cent. 
We thus see that in this case, in trials with fifty seed, a variation 
of twenty per cent. in the expression of single germination results 
had httle meaning. We may assume that a larger number of seeds 
at a trial are required in order to obtain close duplication. Our re- 
sults can be calculated, by massing the adjoining seeds, into those 
with one hundred, two hundred, etc., seed, as below :— 
Per cent. germinated in trials of 
Ill T 
iF HE . A Ve 
100 seed. 200 seed. 300 seed. 400 seed, 500 seed. 
70 76 75 74 73 
82 73 74 76 79 
73 74 79 
73 79 
66 75 
83 
82 
76 
79 
78 
86 
With 50 seed used, the variation was 20 per cent.; with 100 seed it 
was ae per cent.; with 200 seed it was 6 per cent.; with 300 seed it 
was 5 per cent.; with 400 seed it was 2 per cent. ; and with 500 seed 
it was 6 per cent. 
The conclusion is evident, that in germinating single samples of 
melon seed from two different sources, in numbers of one hundred, 
we are not justified in pronouncing one of better quality than anoth- 
er, if the variation is not over twenty per cent., and such a differ- 
ence could have no further meaning than to suggest further trial be- 
fore decision. 
A second trial of the value of duplicates was with seedsmen’s 
seeds, each packet taken from a different bag of the same stock. 
Danvers Yellow onion seed, growth of 1882, tested 50 seeds at a 
trial in June, 1883, and fifty-one tests. The average germination 
was 89 per cent., the extremes 74 and 98 per cent., or 24 per cent. 
difference. 
