542 MR. J. PARKIN ON THE 
and the shallow nature of the corolla suggest primitive characters. On the 
other hand, the plant is somewhat of an alpine; and the possibility of 
reduction having taken place in the inflorescenes, often correlated with an 
inerease in size of the individual flower, is not to be wholly disregarded. 
(2) Law panicles—the terminal flower first to bloom. 
The species coming under this heading differ from the foregoing in that 
the main axis branches more and to a higher degree, forming dichasially 
branched pleiochasia. 
The common Bluebell of Britain ( C. rotundifolia) may be taken as the type 
here (fig. 5, A). .The terminal flower is the first to bloom and the lateral 
flowers, which may number as many as eight, then open in the ascending 
order for the most part. Bach lateral produces from the axils of its 
bracteoles a pair of tertiary flowers, so the inflorescence becomes a dichasial 
pleiochasium. 
C. patula may be mentioned as another example under this heading, like- 
wise C. isophylla. Also to some extent C. barbata, C. Portenschlagiana, 
C. primulefolia, C. rhomboidalis, and C. Scheuchzeri—the last two are without 
tertiary flowers. 
(3) Spicate forms—the terminal flower no longer blooming first. 
In Campanula rapunculoides, C. latifolia, and C. spicata the “spike” ends 
in a terminal flower, which, however, does not expand first (fig. 5, B). The 
flowers open in the ascending order for half to two-thirds of the way up the 
"spike," then the terminal flower expands and afterwards the upper (the 
remainder) bloom. The suecession is by no means fixed. In long inflor- 
escences the terminal flower-bud may be the last to open. Tertiary floral 
shoots are for the most part suppressed, though the paired bracteoles are 
present. There is a tendency, noticed in several examples, for the upper- 
most flower-buds to become arrested in their development, so that they never 
expand. In some specimens of C. rapunculoides (fig. 5, C) observed in 
Switzerland this had gone so far that no terminal flower-bud was evident ; 
in faet, through the uppermost part of the inflorescence becoming arrested 
in its growth, a true racemose spike was produeed, all the flower-buds 
expanding in acropetal succession. Further, in these Swiss examples the 
bracteoles were noticed to be absent (evidently aborted) in the upper flowers, 
though present in the lower ones. This species (C. rapunculoides) would 
repay a more extended investigation in different localities, from the point of 
view of the change from a cymose to a truly racemose tvpe of inflorescence. 
Besides the tracing of the evolution of the simple racemose spike from the 
eymose panicle, two other lines in the evolution of flower-clusters in this 
genus, Campanula, have been noticed. 
