safford: classification of rollinia 373 



others they may be nearly horizontal when immature and at 

 length more or less decurved. Moreover, the members of a 

 group are not always botanically close to one another. Never- 

 theless the arrangement of the various species into groups accord- 

 ing to the shape of the corolla is a great aid to classification and 

 will . prevent many errors. A striking example of erroneous 

 identification is that of the flower figured by Baillon and repro- 

 duced in Engler and Prantl's Natiirlichen Pflanzenfamilien under 

 the name Rollinia mucosa. This is certainly, not the flower of 

 Jaequin's Annona mucosa, the cachiman morveux of Martinique, 

 which is the type of the species. The slender, ascending, in- 

 curved lobes place it at once in the same group with Rollinia 

 orthopetala A. DC. and R. laurifolia Schlecht. On the other 

 hand, the widely spreading lobes of Annona obtusijiora, as figured 

 by De Tussac, place that species in the same group with Rollinia 

 Pittieri and R. deliciosa, described below. In some cases two or 

 more species with similar leaves and fruits but with very distinct 

 flowers have been wrongly associated under a single name, 

 as in the case of R. sylvatica, as usually treated by botanists. 



In a systematic study of any group of plants the desirability 

 of going back to the original description of each species will at 

 once be recognized. In certain monographs more easily acces- 

 sible to the student than the various publications in which the 

 species were first described, amended descriptions are often given, 

 based not on the species itself but upon some allied species mis- 

 taken for it. In many cases the monographer has never had 

 the opportunity of examining the type material. Martius, for 

 instance, in describing Annona obtusijiora and Duguetia Marc- 

 graviana could not possibly have seen the plants on which these 

 species were based; and nothing in De Candolle's description of 

 Rollinia orthopetala indicates that the fruit of the plant he de- 

 scribed was "of the size of a child's head." The plants growing 

 in the inundated forests along the banks of the Amazon, in the 

 province of Para, which yielded the fruits described by Martius, 

 may have been specifically distinct from the type of De Candolle's 

 species, which grew near Demerara, in British Guiana, of whose 

 fruit we know nothing but of whose flower we know certainly 



