1902] UNDESCRIBED PLANTS FROM CENTRAL AMERICA 261 
scentia, petiolis 14"? longis basi dilatata r2"" latis. Spica strobiliformis 
virescens bracteis 3-4 foliaceis 13-25*" longis fulta, bracteis floralibus extus 
pubescentibus intus vinosis 5*" longis 2.5*" latis apice rotundatis, bracteola 
lineari 28?" longa pubescente, floribus generis adhuc maximis circa 12" 
longis. Calyx pubescens 10-12?" al]tus. Corolla glabra 6.5*" longa, tubo 
2.5" longo, segmentis obovato-ellipticis. Labelli lobi circa 20—25"" longi, 
postici inaequales 22-25?" lati margine erosi, laciniae anticae lineares vel 
oblanceolate 2-5" latae, Stamen 3^" longum r3"? latum ultra loculos 127" 
longos 177"? productum. Stigma cano-pubescens semiorbiculare 4?" latum, 
appendice vix breviore ovali bifido. Ovarium dense pilosum late cylindricum 
I1"" Jongum.— Foliorum florumque magnitudine, et horum fabrica insignis. 
Guatemala, alt. 350", Jul. 19ot, voz TuwerckAhezm, n. 6015 Pl. Guat. &c., 
qu. ed. Donn. Sm. 
DONNELLIA C. B. Clarke, nov. gen. Commelinacearum.— Cap- 
sula trivalvis, loculis monospermis; caetera Tradescantiae. 
Donnellia grandiflora C. B. Clarke (sp. unica).— Ca/Zisia 
grandiflora Donn. Sm. in Bor. Gaz. 31:125. I9or. 
This species cannot remain in CaZZsza, to which genus Donnell Smith 
has from the capsule referred it; as he observes, the stamens and the habit 
are those of Zradescantia Sect. Descantaria. We may either arrange the 
plant as a new subgenus onze//ia of 7radescantia, or call it DONNELLIA 
nov. gen., as here proposed. On the suggestion of Captain Donnell Smith, 
I may here give reasons for this course. 
In my monograph of the Commelinaceae (DC. Monogr., Phan. III) no 
attempt is made to deal with the genera ''logically;" the same characters, 
which in the American genera ( 7radescaztia and its allies) constitute genera, 
only constitute subgenera in CozezeZiza and A7ezleza. The main reason for 
this course (7. e., following my predecessors), was to avoid the introduction of 
new names, and in particular of a disputed synonymy; and Bentham has 
here followed me; convenience has been preferred to any logical system. 
A concrete instance may make this more quickly clear. The two sub- 
genera of Commelina are absolutely distinguished (as cf. Bentham), and I 
might have taken up 44ozoon and JDZdymeon as genera (in order to make 
the arrangement consistent with that of the American group); in this case 
Commelina nudiflora L. would have become JDzdymeon nudiflorum C. B. 
Clarke, and Commelina Virginica L. would have become J7ozoon Virgznzcum 
C. B. Clarke. We should then have had two names current for each species ; 
for many botanists would prefer to keep only one genus. But this would 
have been only a trifling inconvenience. 4A school of modern botanists, who 
believe in **laws," hold that the Linnean name Co»tzi/ina must be retained 
for something, must not be sunk altogether. Then would arise the insoluble 
