The Production of New and Improved Varieties of Timothy 367 



our new types, and in this experiment we should select possibly the six 

 or eight highest yielders for further trials on different soils. 



These trials in broadcast plats, together with the tests of the same types 

 in clonal rows and in rows from self -fertilized seed, give us a good basis 



Fig. 94. — Row on right No igi6, grown from selfed se^d of a select type; row on left, 



No. 1917, grown from good commercial seed 



for a preliminary decision as to the best sorts among those under trial. 

 If a type has stood high in the yield of a good quality of hay in all these 

 three methods of testing, it is believed that it m.ay with reasonable safety 

 be judged as a good race, at least for local use on the same type of soil 

 as that on which the selections have been made. On this basis of judg- 

 ment the half dozen best races, which in our experiments have been found 



