164 Eeport of the Department of Botany of the 



the percentage of violet-colored seed, and while the range of 

 difference is fairly constant in samples from one season, in 

 samples from different seasons the percentage of violet-colored 

 seed would be increased or decreased ; and with this change there 

 would he a variation in the number of seeds per unit weight. 



In grass seeds, especially redtop and blue grass, the number of 

 seeds per unit weight varied greatly and followed closely the 

 grade of the seed. In two samples of redtop seed the number of 

 seeds varied from 8,500 to 14,500 per gram. Extra clean red- 

 top seed gave a much higher number of seeds per unit weight than 

 did the poorer grades. The number of seeds per gram in timothy 

 seed depended upon the extent to which it was hulled. Seed that 

 was nearly hull-less gave a count of 3,010 seeds per gram, while 

 seed with hulls still adhering gave a count of 2,600 seeds per gram. 

 It seems certain, then, that standards per unit weight would be 

 necessary for each grade of seed, and it would be desirable to deter- 

 mine into which grade of seed the sample of grass seed under con- 

 sideration belongs in order to arrive at a fairly accurate purity 

 percentage. 



A comparison of the two methods, by weight and by count, 

 as employed in the determination of purity percentage, was of 

 much interest since it showed the possible cases in which the two 

 methods gave widely different results. The seed impurities from 

 samples were both weighed and counted and results showed that 

 in every case where the seed was well cleaned and the seed 

 impurities were about the same size as the crop seed, even if they 

 represented several species, the two methods agreed quite closely. 



It was found that the lower the grade of the seed and the less 

 the amount of cleaning it had received, the smaller in size were 

 the seed impurities and consequently the higher was the percentage 

 of impurity by count as compared with the percentage of im- 

 purity by weight. The smaller and lighter the foreign seeds are 

 with respect to the size and weight of the crop seed in which they 

 are found the greater is the variation in the two methods, the per- 

 centage of impurity by count increasing more rapidly than the 

 percentage by weight as the relative size of the impurities de- 



