— 66 — 



51. Cabbage, Marblehead Ma7ninoth. — Department of Agricul- 

 ture. 



50 seeds in each of ten tests in both the Geneva tester,* bearing folds of 

 cotton flannel, and in potting soil in forcing-house. 

 Series I. — Tests in Geneva tester. 

 Series II. — Tests in earth. 



SERIES I— GENEVA TESTER. 



Average 

 per cent. 



of 

 Sprouting 



88 



Actual 

 per cents. 



of 

 Sprouting 



Per cent. 



of 



variation 



from the 



mean. 



Average 

 per cent. 



of 

 variation. 



72 

 80 

 82 

 86 

 90 

 92 

 94 

 94 

 94 

 96 



16 

 8 

 6 

 2 

 2 



4 

 6 

 6 

 6 



8 



6.4 



SERIES II— EARTH. 



Aveiage j Actual 

 per cent, per cents. 



of I of 

 Sprouting, Sprouting 



Per cent. 



of 



variation 



from the 



mean. 



Average 

 per cent. 



of 

 variation. 



77.6 



6.7 



CONCLUSIONS. 



1 . One test cannot be accepted as a true measure of any sample 

 of seeds, 



2. Variation in duplicate tests is likely to be greater when .seeds 

 are planted in soil than when tested in some sprouting apparatus 

 like the Geneva tester. (Cf. introduction to § III.) 



X. COMPARISONS OF RESULTS OF SEED- 

 TESTS WITH RESULTS OF ACTUAL 

 SOWING IN THE FIELD. 



It has been said recently that the ideal test of seeds is actual 

 sowing in the field, inasmuch as the ultimate value of the seed is 

 its capability to produce crop. This notion of seed tests is obvi- 

 ously fallacious, although the statement upon which it is based is 

 true. In other words, actual planting rarely gives a true measure 

 of the capabilities of all the seeds of any sample, because of the 

 impossibility to control conditions and methods in the field. The 



* This apparatus holds the seeds in pockets of cloth which hang over a 

 pan of water. For a full description, see Second Rep. N. Y. Exp. Sta. 67. 



