BY THE SELECTION OF SOMATIC VARIATIONS. 23 



no sectorial distribution could be traced; and (4) 5 plants with green 

 and yellow distributed irregularly. 



Aside from the fluctuations in relative amounts of green and yellow, 

 there was also much fluctuation in number and size of the blotches of 

 epidermal red. On some of the plants there was rather gradual increase 

 or decrease both in number and size of these blotches, giving such 

 differences in respect to red as are shown in figures 5, 17, 26, and 28. 

 Such plants were, however, still considered as blotched in the sum- 

 maries. Selection of typical red blotched epidermis for various types 

 has been directed to plants having the epidermal blotching as in figures 

 2, 5, and 6, rather than as in figures 23 and 28. 



53 plants produced bud variations giving loss of green 7 times, loss 

 of the yellow bordering-band 8 times, complete loss of yellow 24 times, 

 reversal of the relative positions of the green and yellow 6 times (4 cases 

 appeared in half of a leaf only, as in fig. 25), increase of red to com- 

 plete epidermal red 4 times, complete loss of epidermal red 10 times. 

 In clone 14, 4 plants gradually developed a green-yellow spotted-red 

 blotched pattern in part of the branches. On the basis of my descrip- 

 tions the change was a fluctuating variation affecting only a few 

 branches of a plant. These 4 cases are not included in the summaries 

 of bud variations. 



In their extent the bud variations gave extremes in development of 

 yellow, of green, and of epidermal red. There were cases of nearly 

 pure yellow and of absolutely pure green ; there were cases of solid red 

 epidermis and others with no red epidermis. 



Furthermore, the changes in green and yellow or in epidermal red 

 occurred entirely independently of each other. In general, the different 

 types of bud variations were quite uniformly distributed in the various 

 clones. 



The type green-yellow spotted was produced on 2 plants by a fluctua- 

 tional increase of green after the loss of epidermal red had occurred. 



In table 2, as in other tables, when the totals given for constant, 

 fluctuating, and sporting plants exceed the number of plants grown, 

 it shows that a certain number of the fluctuating plants produced also 

 sharp, clear-cut bud variations. Also, when the total of cases of bud 

 variations exceeds the plants giving them, certain plants produced 

 more than one bud variation. 



For the purpose of establishing an index of the frequency of bud 

 variation we may take the ratio of bud variations to the estimated 

 number of buds developed. Each plant produced an average of at 

 least 200 branches which made sufficient growth to reveal the pattern 

 of the leaves. On this basis the index of total bud variation for this 

 group was about 1 to 1,110. The ratio of constant plants to fluctu- 

 ating plants was almost exactly 3 to 1 , not counting the plants with bud 

 variations many of which were otherwise constant. 



