214 BOTANICAL GAZETTE [SEPTEMBER 
short, the transition from stem to root taking place in an almost hori- 
zontal plane (jig. 30). In the absence of cauline bundles, the root 
poles are inserted upon the cotyledonary strands (A, B, C, D, jigs. 
34, 35). There is no rotation of the protoxylem, but the phloem and 
metaxylem separate opposite the protoxylem groups (Bm, Cn, jig. 
36), the cambium proliferating to fill the breach in the metaxylem, 
and the cortex invading the space left vacant -by the phloem, The 
resulting halves swing, the one to the right, the other to the left, giving 
the appearance of a double fan of phloem and metaxylem connected 
by a single group of protoxylem. The right half of the phloem of 
one pole joins the left half of that of the next, and the cambium layer 
is thus curved inward (cb, figs. 37, 38). One result of this compli- 
cated process is that the entire xylem system of the root is bordered 
peripherally by cambium. 
The medulla throughout the root is extremely meristematic, and 
its activity sometimes results in a displacement of one or more of the 
protoxylem groups, giving an unsymmetrical appearance in cross- 
section, This activity is indicated by the thin walls (0, fig. 3%) 
showing recent division. a 
In all the plants I have investigated, the number of poles remains 
constant throughout any given root. This number may be three or 
four, depending seemingly upon the degree of development attained 
by the median bundle of the aborted cotyledon. Figs. 26 and 34 
will serve to illustrate this point. They represent the same level 
in two different plants. In the former the median bundle (D) bis 
very weak, and there were but three root poles; in the latter this 
bundle was well developed and the root was tetrarch. 
Fig. 39 represents the condition of the root tip in the re 
the differentiation of protoxylem. The connective tissue is cle: y 
defined, but there is no xylem plate connecting the poles. ere 
tion of longitudinal sections of the root tip furnishes nothing that co 
be added to the description of RemnKe (ro). One initial group P*™ 
duces plerome, another periblem. There is no calyptroge? - = 
matogen, but the outermost layers of the periblem become Joosent 
at the tip and form the root cap (k, fig. 42). : ak 
There is no distinct endodermis, and the pericycle 1 ae 
layered, making it impossible to distinguish with absolute 
ion of 
