1958J 



THE BOTANY OF THE 6UAYANA HIGHLAND PART III 



103 



a different habit, glabrous branches, much longer leaves, and stamen connective 

 conspicuously prolonged ventrally. The stamens of C. serpyllacea are more remi- 

 niscent of C. montana Gleason and C. coriacea Gleason although the basal con- 

 nective thickening of those species is more pronounced. The small leaves are 

 somewhat reminiscent of those of C. sertularia (DC.) Tr. but the habit, curved 

 leaf -margins, glandular-puberulent branches and lower leaf -surfaces, long stamen 

 connectives, and quadrilocular ovary of the Brazilian species make the resem- 

 blance seem superficial. The Brazilian genus Fritzschia has similar leaves and 

 stamen connective prolongation but glabrous stems, foliage, and hypanthium, 

 more abruptly truncate anthers, and 4-locular ovary. 



Vegetative variability in savanna plants is usually quite great ; dimensional 

 floral variation is also significant. The latter type of variability is certainly 

 influenced by the flowering period, with the later flowers on a single plant often 

 being much smaller; this phenomenon can be seen wherever a large stand of 

 Rhynchanthera serrulata (Rich, ex Bonpl.) DC. or Pterogastra minor Naud. is 

 found. In Tibouchina spruceana Cogn., the last few flowers on a plant often 

 are semiabortive with distorted petals and sterile stamens. 



Gleason has several times suggested the possibility of heterostyly in Andean 

 Miconia of sect. Cremanium. Such floral diversity, if finally recognized taxo- 

 nomically will lead to the synonymy of many species of Miconia. For specific 

 limits in the rare genus Pachyloma, it has seemed necessary to recognize a floral 

 latitude uncountenanced previously, with basic criteria being vegetative. 



This genus is restricted to the savannas of the upper Rio Negro and upper 

 Orinoco drainages. Although it was described in 1828, it has hitherto been known 

 by only five collections, with three described species. Additional recent col- 

 lections by Schultes and ourselves have both clarified and confused the species. 

 Upon examination of single branches, the most striking floral character is the 

 connective development differences in the two whorls of stamens. However, while 

 the dimorphic or essentially monomorphic character of the stamens seems rela- 

 tively constant for single sprigs, examination of entire collections reveals quite 

 a disturbing floral variability in the sprigs assigned to a single collection num- 

 ber, despite vegetative constancy in the collection. The heterostemonous feature 

 is especially evident in the Schultes collections from the Rio Vaupes drainage. 

 Within the recent Venezuelan collections, the stamen variability exists, although 

 less striking fashion. One other disturbing phenomenon is the presence or 

 absence of glandular hairs on style and/or filaments of some collections both 

 Colombian and Venezuelan. Since both P. coriaceum and P. huberioides are 

 spare shrubs, a collection number includes branches from several plants, at least 

 in the Venezuelan collections. 



I am indebted to the United States National Herbarium for the opportunity 

 to examine all the specimens for each Schultes & Cabrera collection number. 

 The herbaria at Munich and Paris have permitted the study of the holotypes of 

 P. coriaceum and P. huberioides. Drs Baehni and Weibel of Geneva kindly sup- 

 plied information from the de Candolle herbarium and on the relations between 

 Martins and de Candolle. 



Pachyloma DC. Prodr. 3: 122. 1828. 



Urodesmium Naud. Ann. Sci. Nat. III. 15: 338. 1851. 



Flowers 4-merous. Hypanthium campanulate to urceolate, the calyx-lobes 

 fleshy and inconspicuous. Petals oval to obovate, acute to mucronulate-obtuse, 



