PROPAGATION OF PLANTS 53 



If one sample of seed contains 90 per cent of live seed 

 and costs $9, and another sample contains 80 per cent 

 of viable seed and costs $8, they would appear to be 

 equally cheap. But the former sample is to be preferred, 

 because, if a sample germinates poorly, we may expect 



FIG. 34. Poor clover seed contain- FIG. 35. Good pure clover seed 



ing many weeds 



that the same causes that killed many of the seeds weak- 

 ened all the others. 



There is one case in which the cheaper seed might be 

 best, and that is, if the other contained seed of some very 

 serious weed that was not present in the cheaper kind. 



52. Size and Weight of Seeds. Many experiments 

 have been tried with large and small seeds, and with seeds 

 of high and low specific gravity. In the majority of trials, 

 the larger and heavier seeds have proved best. Usually 

 the small seeds are lighter for their size, or have a less 

 specific gravity, than the large seeds. 



Heavy cotton seed separated by an air-blast was grown 

 in comparison with unseparated seed at Lamar and Harts- 

 ville, S. C., in 1906. 1 At each place, equal areas of about 

 an acre were planted with each kind of seed. The average 

 yields of cotton were: 



Heavy seed 1,106 pounds 



Unseparated seed 1,010 pounds 



1 Farmers' Bulletin No. 285 



