Evans: HEpATICAE OF PuERTO RICO 207 
walls are more delicate. Two of the marginal cells, during the 
development of the gemma, lengthen out into unbranched rhizoid- 
like structures, which ultimately become a little longer than its 
radius. These cells lie, one-on each side, about midway between 
the apical cell and the base of the gemma, and their outgrowths 
are more or less closely appressed to its convex lower surface 
(FIGURE 21). Apparently none of the marginal cells are modified 
into the curious organs of attachment, which Goebel describes 
and figures for Cololejeunea Goebelit. 
When a gemma becomes detached it must fasten itself firmly 
to the substratum before further development can take place. 
This is done by means of radiately branched rhizoids, which spring 
from the marginal cells and secrete an abundant supply of slime. 
Apparently all of the marginal cells are able to develop rhizoids 
of this character, but many of them fail to do so. The number 
and position of the rhizoids varies accordingly on different gemmae 
(FIGURES 22, 23). In all cases observed, where a gemma had es- 
tablished itself firmly, it had first of all been turned over and had 
thus been enabled to present its concave surface to the substratum ; 
but whether or not this is an absolutely necessary proceeding 
could only be proved by experiment. In any case the gemma 
would be in a most unstable position if it lay with its convex sur- 
face turned toward the substratum, and this instability would be 
increased rather than diminished by the two long rhizoid-like out- 
growths above described. By turning the concave surface down- 
ward, on the other hand, the gemma is able to present its entire 
circumference to the substratum and at the same time to enclose 
a capillary space where water can be temporarily retained. Through 
the inversion of the gemma the surface which was originally ven- 
tral becomes dorsal, and apparently the two outgrowths, which 
are now turned away from the substratum, undergo no further 
development. These structures, therefore, although without doubt 
morphologically rhizoids, do not assume the function of holding 
€ gemma in place. 
The established gemma does not give rise at once to a leafy 
axis but develops instead a protonemal structure in the form of a 
flat oblong or ligulate thallus, closely appressed to the sub- 
Stratum (FIGURES 22, 23). Its behavior, therefore, is very much 
