AGAVE IN THE WEST INDIES—TRELEASE. 15 
marginal arming of their leaves is not even indicative of the condition that will prevail on the 
mature foliage, the essentially smooth-edged and the most prickly forms of Agave sisalana 
agreeing in having the bulbil leaves finely prickly on the margin. 
TAXONOMIC APPLICATION. 
The actual use of such differences as have been noted is more difficult than would at first 
sight seem probable. In trunk, leaves, inflorescence, flowers, fruit, seeds, and bulbils the exotic 
Sisalanae are easily differentiable from all of the agaves native to the West Indies, even the 
rigid-leaved Inaguenses; the latter, too, differ in habit, foliage, and flowers, and apparently 
in fruit, from the other endemic groups. Beyond this, except that the known Antillares are 
small-flowered and small-fruited, and, so far as known, bear their flowers rather densely clus- 
tered at the ends of the panicle-branches, even the principal groups into which the primal West 
Indian stock has become differentiated must be limited rather by various associations of 
characters than by modifications of any one fundamental character. Within each group the 
difficulties of classification are equally great because of the corresponding impossibility of 
selecting a basic series of successively subordinated characters. The capsules of Agave papy- 
rocarpa and A. tubulata are much alike, but their flowers are very different (pls. 97, 99, and 
106), and the same is true of A. antillarum and A. Underwoodwi (pls. 43 and 71); on the other hand 
the capsules of A. montserratensis and A. Dussiana differ markedly, though their flowers are 
nearly alike (pls. 19 and 29). The rather close agreement between the flowers of A. barbaden- 
sis and A. unguiculata is attended by a totally different leaf-spine (pls. 35, 38, and 39); but 
nearly identical spines accompany very different flowers in A. montserratensis, A. grenadina, 
and A. barbadensis (pls. 18 and 19, 21, 35, and 38). 
Though it is hard to separate the dissociated spines of A. vivipara and <A. Brittoniana 
(pl. B), species which belong to different groups and could not otherwise be confused; and though 
the spine of A. portoricensis is as suggestive of one of the Caribaeae as that of A. ventum-versa is 
of one of the Antillanae, groups in which the spines are reciprocally shaped; the arming of the 
leaf appears, on the whole, to have changed less, in the differentiation of these plants, than 
have either flower or fruit, due allowance being made for discrepancies between spines of inde- 
terminate development and the prickle-bases of varyingly repand species, as well as for the occa- 
sional partial or all but entire abortion of the prickles which occurs with considerable fixity in 
A. sisalana and perhaps also in A. anomala among the Antillean species. 
It is obviously impossible to arrange the species in a single series corresponding to their 
relationships. Since no existing group can be considered as the parent stock of the others, 
the groups are treated here in the sequence which, on the whole, affords the easiest transition 
from one to the other. In each group the species are so arranged as to bring together those 
which show the largest number of common characters, though this convenient sequence does 
not necessarily bring at the beginning or end of a given group the species which most closely 
resembles that of the adjoining group. 
The specific descriptions here given might have been shortened by omission of charac- 
ters common to groups of species, so as to save much space and with attendant facility in 
comparing the few differential characters of related forms; but experience has shown that such 
abbreviated diagnoses are so likely to be regarded as complete that they are scarcely permis- 
sible except in revisions of entire genera. When prepared for a partial or local revision like 
the present, they may become absolutely misleading if applied in the study of other forms, or, 
apart from rather full synoptical characters, even of the groups they are prepared for. In the 
following account, therefore, each species is as fully described in all essentials as the material 
at hand has made possible. 
The Antillean species of Agave are so largely endemic that the references given under each 
of them are essentially complete, even for notices of it as a garden plant. If many familiar 
uses of such names as A. vivipara be missed from the citations given, it is because they pertain 
to other species and are of later application than the first use of these names as here employed. 
For the few species not limited to the islands, reference is made merely to the place of original 
publication, and to mention of it as Antillean. 
