Stout & Boas: STATISTICAL STUDIES IN CICHORIUM 355 
flower number of the terminal heads, but also by the ray-flower 
number of the later heads, for he found that some plants having 
21 ray-flowers in the main head gave for all flower heads curves 
with maxima lower than 21, often at 13 or 14. These plants were 
discarded as not belonging to the desired race and only those 
giving "partial curves" (the curves obtained from the flower 
heads on a plant) with maxima at 21 were retained. No attempts 
were made to isolate races with numbers intermediate between 
13 and 21, and no further studies were made of the very irregular 
cases of partial variability which were in evidence. 
After isolating the two races, de Vries observed variation in 
the race with 21 ray-flowers in terminal heads in respect to in- 
crease of ray-flowers. In a crop of 1,500 plants one was found 
with each of four lateral heads having 22 ray-flowers, a higher 
number by one than was previously seen in any of the terminal 
heads. Open-fertilized seed of this plant gave a progeny in 1897 
of 414 plants; the number of ray-flowers in terminal heads ranged 
from 14 to 34. From seed of the plant having 34 ray-flowers in 
1898, 241 plants were grown; for these the range Was I9 to 48; the 
modes were at 26 and 34. The average of the population was 38. 
The next year 194 plants, offspring of the plant with 48 ray- 
flowers, produced terminals with ray-flowers ranging from I9 to 
67; the modes were scarcely pronounced at 32, 37, and 45, and the 
average number of rays was 41.5. Іп this crop, it is stated, ligulate 
flowers appeared among the disk-flowers in one head of 62 ray- 
flowers. The seed of this plant gave a progeny in 1900, 31 plants 
in all, ranging in ligulate flowers of terminal heads from 33 to ror; 
there were no pronounced modes either primary or secondary; 
the average number was 53.2. One plant had some heads with 
only ligulate flowers. 
Evidently for the first few years there was an increase in the 
size of the heads and the accompanying number of ray-flowers. 
Then there came also a change of disk-flowers to ray-flowers. 
Throughout there had been great irregularity in heredity revealed 
in range of numbers observed in individual variability both for 
data of terminals alone or for terminals and laterals. Such varia- 
tion de Vries assigns to mutation and attempts to show that the 
increase in number follows the Fibonacci series. There can be 
