394 
44. RUBIACEJ2. 
margins. Cymes on erecto-patent stalks or short side-branches 
exactly as in G. aparine L. but 4-10-fld. and 2-3-compound, 
the divisions finely capillary and sprinkled with a few very 
minute prickles, 3-2-forked, the ultimate division with a 
single leaf or br. as long or sometimes twice as long as the 
quite smooth divaricate or often in fr. deflexed 1^-3 milliin. 
long, mostly binate pedicels. FI. reddish or purplish outside, 
very small, pet. not longer than the ov. not apiculate, scarcely 
spreading. Fr. notched, transversely reniform, f-f millim. 
long, f-1 millim. broad ; mericarps globoselv oval, mostly quite 
free from all trace of pubescence but finely granulated. 
Two opposite extreme forms or states of this plant occur 
occasionally, equally diverging from the normal type. In hot 
sunny situations or open hill-sides and mountain pastures, it 
becomes more robust, with shorter stouter st., more crowded 
whorls of 1. and close congested cymes of fl. and fr., turning 
very black in drying, and when growing on exposed flat tops of 
walls or rocks with numerous procumbent st. radiating from 
the crown of the root. On the other hand, in holes or crevices 
of walls or rocks, or when drawn up amongst other herbage 
in damp and shady spots, it assumes a very different aspect, 
becoming excessively attenuated and delicate in all its parts 
with erect often unbranched single st. of extreme tenuity not 
thicker than a hair, long internodes, finer or thinner 1. in re- 
mote whorls, and loosely divaricate cymes with more elongated 
spreading pedicels and branches ; and the whole pi. also turns 
less black or sometimes not at all so in drying. Such spec, in 
the Canaries were considered by Webb 1. c. to be identical 
witli G. divancatum Lam. Diet. ii. 580 (1)C. Ic. PI. Par. Gall, 
t. 24); which however, notwithstanding the adverse opinions 
of Smith E. Fl. 1. c., Koch (ed. 2) 1025, and Cosson and Ger- 
main Fl. Par. ii. 304, appears to be sufficiently distinct. In- 
deed Cosson and Germain’s G. anylicum erection, with st. 
(i fortement denticulees-scabrcs,” agrees thus far as exactly 
with this form of the Mad. G. parmense L. /3 as it differs from 
the true G. divarication Lam. 
I loll and Iteichenbach’s reference (Mad. List) of this deli- 
cate form of the Mad. pi. to G. minidijionon Brot. is doubtless 
incorrect; for Brotero’s pi. belongs by its “hispid fr.” rather 
to G. parmense L. a tnchocarpum Tausch (G. litiyiomm DC.), 
