HAZEN: TRIMORPHISM AND INSECT VISITORS OF PONTEDERIA 467 
containing a blood-red tannin; later it transpired that Zopf had 
confounded true anthocyanin receptacles with “зас-сейз” occur- 
ring throughout the vegetative structures of Fumariaceae, which 
he finally considered to be alkaloid-receptacles, though Heinricher, 
who gave the name “‘sac-cells,’”’ states that their contents аге a 
mixture of substances, including a fatty oil. 
Тһе rôle of these peculiar perianth cells in Pontederia (and in 
Piaropus) can only be surmised at present, whatever the nature 
of their contents. In fixed and stained sections they often behave 
much like mucilaginous or gummy substances, and if of such a 
nature might possibly function in protecting the perianth from 
danger of desiccation until after anthesis, when the upper part 
promptly rolls up and soon dries, though the tube persists as an 
increasingly fleshy envelope around the ovary until the seed is 
mature. Even in the open flower, some of these large hypodermal 
cells are often found with their thin protoplasmic layer collapsed, 
and the contents apparently discharged; half of such a cell is 
shown at the bottom of FIG. 2. It is interesting to note that the 
stamen filaments in both Pontederia and Piaropus, which have 
blue anthocyanin in their epidermal cells, are well supplied with 
these long, mostly subepidermal idioblasts; but in the case of 
Pontederia montevideensis, though they are conspicuous іп fila- 
ments of flowers grown out of doors in September, they appear 
to be entirely absent in filaments of mid-winter, conservatory- 
grown flowers, while still persisting in the perianth of the latter. 
In April, after two or three weeks of sunshine, the flowers of 
the same conservatory plant have the cells sparingly developed 
in the filaments and showing a pink anthocyanin color. That 
a temporary suppression of such structures should occur in con- 
sequence of lack of need for them is rather incredible; the sugges- 
tion, rather, presents itself, that the contents of these idioblasts, 
as perhaps also the numerous raphide-sacs which are early found 
in a similar position in perianth, pistils, filaments, and most 
abundantly in the anthers, are after all only in the nature of 
by-products of metabolism. 
The outer surface of the perianth is clothed with spreading 
glandular hairs (FIG. 2); their elongated terminal cell, rich in 
protoplasmic contents and sometimes binucleate, is not infre- 
