412 SEMI-CENTENNIAL OF TORREY BOTANICAL CLUB 
In the course of the studies a few plants were found which in 
the performance for a season exhibited an actual increase of average 
flower number per head. This is quite the reverse of the usual 
performance. Data for one of the most marked cases of such 
increase are given in the following table (TABLE 33). This plant 
was one of the Е, generation grown in 1916 and the data as col- 
lected distinguished between terminals and laterals for the dif- 
ferent clusters. The average or mean number for all flower heads 
is 17 and the value for the first date of bloom (a) is 16.4. Neither 
of these is an especially low value. It should be noted that this 
plant was one of a series of sister plants which were quite uni- 
formly sparsely branched and had few solitary terminal heads: 
the heads were in clusters, quite as shown in C and D of PLATE I3. 
Such clusters evidently represent a shortened and compacted 
system involving terminals of different relative ranks with their 
respective laterals. The group is so compact that only the first 
head to bloom can be definitely regarded as terminal. If the 
group had not been compacted (but expanded as in A and B of 
PLATE I3), many of these heads would have been terminal for 
their respective clusters. The prevailing high numbers in the 
so-called lateral heads of these groups as seen in TABLE 33 may indi- 
cate that in these compacted groups of heads the laterals of lower 
ranks are crowded out and fail to develop as they may when the 
branching is more profuse. However, it must be recognized that 
many plants with the grouped-head habit, quite identical with the 
one under consideration, showed the seasonal decrease most char- 
acteristic of the species. 
5. VARIATION IN PARTIAL VARIABILITY WITH THE AGE OF A PLANT 
The data collected from plants in successive years of growth 
may now be presented with respect to the very important question 
of variation in seasonal performance of a plant from year to year. 
In the case of all perennials and especially of herbaceous peren- 
nials this point needs careful analysis before adequate comparisons 
between individuals can be made. All of the plants studied, with 
the exception of the variety red-leaved Treviso, grew as perennials, 
and although plants of each series winter-killed, the roots and 
basal portions of such plants were apparently fully alive at the end 
