New York Agricultural Experiment Station. 523 



perfect or nearly perfect clusters and also some that were quite 

 imperfect. It appears therefore that while tests with one or two 

 clusters may generally give a good indication of the condition of 

 a variety they are not extended enough to be received as conclusive. 

 Indeed some kinds of grapes show much variability in the charac- 

 ter of their self-fertilized clusters on the same vine, although the 

 average results of the variety with different vines and in different 

 tests are quite similar. In addition to the instances of the Rogers 

 hybrids which have just been cited, detailed results in two other 

 cases of this kind will be given as illustrating this 'feature in cer- 

 tain variable grapes. In 1897 9 clusters of Antoinette were cov- 

 ered during the blooming season. Three produced perfect clusters 

 of fruit, 3 were almost perfect, 2 were rather loose and 1 formed* 

 no fruit. Had only the last-named cluster been tested Antoinette 

 might have been listed as a self -sterile variety. Twenty-two clus- 

 ters of Yergennes were tested at Penn Yan in 1897. Two clusters 

 formed no fruit. The other twenty averaged about half filled, 

 varying from the very loose and imperfect clusters shown in Plate 

 XLIX, Figure 2, to the nearly perfect cluster which is shown in 

 Figure 1. 



Seasons and localities. — From a consideration of these results 

 and others which will be presented on following pages it is clear 

 that not only should more than one vine be included in tests of 

 self-fertility, but the tests should be made in more than one sea- 

 son and under diverse conditions of soil and climate. This has 

 not yet been done with all of the varieties which are included in 

 these experiments. Of the 169 cultivated varieties under investi- 

 gation but 78 have been tested by the writer in more than one 

 season and only 24 in more than one locality. 



methods of recording observations. 



When the fruit of uncovered clusters was nearly or quite full 

 grown each covered cluster was examined. Whenever any fruit 

 was found a record was made of the proportion of blossoms which 



