14 



bushels per acre as compared with an average of 492 bushels per acre for the 

 low-yielding strain, and a yield of 607 bushels per acre secured from seed tubers 

 selected without reference to individual hills. The high-yielding strain produced 

 an average of 262 bushels per acre, or about 35 per cent more than the low- 

 yielding strain, and 167 bushels per acre, or about 22 per cent, more than the 

 check as the result of three years of bud selection work. His studies showed 

 clearly that with few exceptions the low-yielding plants remained unproductive 

 and that the high-yielding hills for the most part remained productive, as was also 

 shown to be the case in the Cornell experiments referred to above. 



In citrus varieties, the writer and his associates have recorded the behavior 

 of 14 strains of the Washington Navel orange (64) [typical fruits of some of 

 these strains are shown in Plates 17, 18, and 19] ; 12 strains of the Valencia 

 orange (65) ; 7 strains of the Marsh grapefruit (66) [typical fruits of two of 

 these strains are shown in Plates 20 and 21] ; 8 strains of the Eureka lemon (67) 

 [typical fruits of three of the strains are shown in Plates 22, 23, and 24] ; and 6 

 strains of the Lisbon lemon (68) [typical fruits of several strains are shown in 

 Plates 25 and 26] . These strains originating as bud mutations have, for the most 

 part, been unintentionally propagated. They have been found to be capable of 

 perpetuation through bud propagation. The isolation of the valuable strains and 

 the elimination of the undesirable ones through systematic bud selection has been 

 demonstrated to be possible and commercially practicable. 



In timothy, Webber (78) found great differences experimentally and com- 

 mercially in the yield of high- and low-producing strains isolated through bud 

 selection. Experimentally, he shows that the high-yielding strains produced 

 an average of more than six times the yield of the low-yielding strains. 



In prunes, the writer has begun the study of the behavior of seven strains 

 originating as bud mutations of the French prune variety in California ; an 

 example of one of these mutations is shown in Plate 9. The data are incomplete 

 as yet, but enough information has been secured to warrant the statement that 

 these strains are very different in fruit and foliage characteristics and that these 

 differences are of great commercial importance to the prune industry. 



In sugar cane, the writer and associates in the Hawaiian Sugar Planters' 

 Experiment Station during 1920 found in fields of the H 109, the Yellow Cale- 

 donia, and the D 1135 varieties many plants resembling distinct strains. Their 

 appearance and occurrence indicated that they originated as bud mutations. The 

 plant differences found in these strains included the number of stalks of each 

 stool and their habits of growth, weight, length, circumference, shape, and color 

 of stalks, amount of juice in the stalks and its sucrose content, and other impor- 

 tant characteristics. These data are not complete as yet, nor available for pub- 

 lication, but it is likely that they will be announced when these investigations 

 have proceeded far enough to warrant this action. 



Most of the work in the isolation of strains has not been reported because 

 such records have usually been withheld until the strains have been tested and 

 introduced as varieties. In strawberries (5, 32), apples (41, 42, 43, 44), and 



