530 THE AMERICAN NATURALIST [Vol. LIII 



much as 9.4. On 15 branches the primary terminal is 

 actually the largest head on the branch. In 3 cases the 

 primary terminal is smaller than the average, and in 10 

 cases it is exceeded in si'ze by one or more of the lateral 

 heads. It Avill be noticed that these conditions occur only 

 on branches with numerous heads, where the terminal 

 heads of certain individual cymes tend to raise the aver- 

 age. In fact, on those branches which bear a total of 

 less than ten heads, and in which there are accordingly 

 fewer chances for large secondary terminal heads, the 

 average sizes of the two classes are 55 and 50.8 and with 

 two exceptions (branches nos. 24 and 27) the primary 

 terminal is actually the largest head on the branch. On 

 branches with a total of 10-20 heads the averages are 55 

 and 51.9 and the primary terminal is actually the largest 

 in only two fifths of the branches. In the two cases with 

 over 20 heads the averages are 52 and 53.2 and the pri- 

 mary terminals are conspicuously exceeded in size by 

 some of the other heads. Since the heads of each cyme 

 differ but little in age, the variation in their size may 

 possibly be due to difference in the amount of food-stuff 

 6r water available, by which the terminal heads at the 

 end of a continuous axis are favored. 



2. It has already been stated that the solitary heads 

 and the floriferous branches appear in basipetal order 

 and that those from the lowest nodes may not be suf- 

 ficiently developed to bloom before frost stops all further 

 growth. Table II shows the variation in flower-number 

 correlated with the position of the branch. 



The table indicates a steady increase in the number of 

 abortive heads from the older branches at the summit 

 to tlio yonii,a:or ones at tlie l>a.se. The greatest number of 

 \w;uU nvo foniid ii.'nr tlic middle of the series on the 

 ]u„uv^t l.Mt.T.-il hiniK-hrs. wliich rise from the middle 

 iiitcnnxlfs to a lici.uiit (Miual to or surpassing the summit 

 of ihr stem. I>ut the average number of flowers is re- 

 markably constant throughout, varying only from 50.7 

 to 52.9 for each set and, in general, reaching the maxi- 



