31 
A perfectly cylindrical internode, 8 ¢.M. long, lengthens to 
an inflorescence which is over its whole range — more than 
2 d.M. — strongly contorted and consists of numerous small 
Sdntiadco, 
The contorted axis is moreover bent in several directions. 
The bracts which constitute one single row are linear with 
broadened base. By their cohesion they produce the effect called 
biostrepsis. To each of these bracts is opposed a branch of 
2—3 internodes. The first of these is short, flattened and grooved 
like a, petiole: 4—6 m.m., the second (cylindrical) has a length 
of 2—3 ¢.M., the third, if present, 3 c.M. Next follows a panicle 
of various dimensions. The leaves at the top of the first inter- 
node are scarcely bigger than the stipules, those of the fol- 
lowing internodes are of normal size and shape. 
The whole thing makes the impression as if the first and 
shortest internode is nothing but’a flower with median proli- 
fication, producing from its centre one or two leafbearing inter- 
nodes and a terminal inflorescence. 
I had no normal material at my disposal to confirm this: 
supposition. 
CAMPANULACEAE. 
Platycodon grandijlorus DC. 
a) Legit C. A. Backer in suo horto 1916. 
Habitat Siberia. 
One of the flowers of an inflorescence has an intercalated 
corolla alternating with the normal one. The difference with 
Eicuier’s fig. 150B') is that in our case the anther bearing 
stamens are antisepalous whereas in Eica.ér’s diagram they 
show an antipetalous position. As in both cases the fertile 
stamens oppose the cells of the ovary, neither of them can 
serve as a support of the hypothesis according to which a whorl — 
of stamens has been suppressed and accordingly the seas vo 
should have a diplostemonic character. 7 
One of the anthers replaced by a bifid stigma (see nf. 
Se en 
1) Riciinapaghenine 1, p. 204.5 
