45 



The broken seeds when tested for germination, germinated 12' 5 per 

 cent. The small seeds contained 11 -6 per cent, hard seeds, as against 

 a content of 4 • 5 per cent, hard seeds in the large grained seeds. 



The percentage of pure germinating seed, if taken according to number, 

 was : — 



Broken seeds 0-45% 



Small seeds - ... 7-60% and 1% hard. 



Large seeds - - 83-00% and 4% hard. 



91-00% and 5% hard, 

 and calculated according to weight — assuming each seed in any of the 

 three grades to have equal average weight — 



Broken seeds 0-40% 



Small seeds .... 4- 16% and 0-54% hard. 



Large seeds . . - 87-09% and 4-21% hard. 



91-29% and 4-75% hard. 



When germination was taken according to number, and broken seeds 

 according to weight, as is customary, the percentage of pure germinating 

 seed was — 



96% X (95% + 5% hard) = 91-2% + 4-8% hard seeds. 



The following account of determinations made on the sample of Eed 

 Clover circulated by Dorph-Petersen in 1923 (No. 64a!) is also pertinent. 

 Portions of this sample after purity determination (which was 

 92% ± 0-75%) were divided into brown shrivelled seeds and full round 

 seeds in a ratio of 33% ± 3% brown seeds to 67% ± 3% round seed 

 by weight, the numerical ratio for the same sample or any portion of it 

 taken for the germination test being approximately 37% ± 4% and 

 63% ± 4% respectively. The numerical ratio of the germinating capa- 

 cities of the two portions was 42% ± 0%: 80% ± 1%. The difference 

 between the weight of the same number of shrivelled seeds and of full 

 seeds was 17 per cent, of the weight of the full seeds. If it be assumed 

 that the individual seeds in either of the two portions were equal in weight, 

 then these percentages of germination may be taken as equivalent to 

 weight percentages. 



Thus, the pure germinating seed, neglecting hard seeds — the ascer- 

 tained purity being 92% ± 0-75%, — ^would be: — 



/(33% ± 3%) X (42% ± 0%)\ inoo/ I n-7S0/1 — fi20/ 4- 9<V hir 

 \(67% ± 3%) X (80% ± 1%)/ ^ '■•''^ ■° ± " '^ /°^ - «>2 /o ± ^ /o by 

 weight and, numerically, 



J (37 ^ ± 4%) X (42% ± 0%) \ /Qoo/ J. n-VfiO/ s— fift.70/ 4. o. co/ 

 \(63% ± 4%) X (80% ± 1%)/ ^ ^ '°^ ^°'~ '° ^ '"' 



Were it possible to separate this sample accurately into germinable and 

 ungerminable seed, the difference in the average weight of the germinable 

 ajid ungerminable would be found to be less than 10 per cent., and, by 

 calculation, the difference between pure germinating seed by weight and 

 pure germinating seed by number would be less than 2 ■ 5 per cent. 



II. The adoption of the proposal here suggested would lead to greater 

 uniformity in reports. 



Evidence of this is provided by reference to the results reported by 

 various Stations on samples circulated by Dorph-Petersen in 1921 and 

 1922, and by M. T. Munn in 1923. 



If certain European Stations which are obviously, frbm the nature 

 of results reported, working according to a fairly uniform routine be 

 selected and results of the three complete series compared, it will be 

 fo\ind that the variation in the determination of the percentage of p\u:e 

 germinating seed is smaller than that in the percentage of germination in 

 the case of over 60 per cent, of the samples. It, therefore, follows that 

 the latitude of variation in germination at present in use could be adopted 

 for the pure germinating seed. 



