154 
ARCHIVOS DO MUSIJU NACIONAL — VOL. XXVI 
b) The proportion is not constant. 
There are hovewer two groups or types. In the first the male 
flowers predominate to a very great extent over the others, which 
are at times redúced to 3 or 4, or entirely absent. This types qf inflo¬ 
rescence is peculiar to some of the individual plants óbserved. 
In the second and most common type the male flowers still form 
the májority but the proportion ranges between 4,5 8,10 or more male to 
1 androgynous onei • 1 
In a very small number of specimens proportions were more or less 
equal or even a larger number of hermaphrodite flowers than of 
males was present. . ! 
. The trees which bear an apparantly higher proportion of herma¬ 
phrodite flowers, present a larger number of forms intermediate betweèh 
male flowers and perfect hermaphrodites, which is equivalent a physio- 
logic reduction in the number of the latter. ’ . 
. The hermaphrodite flowers ;do not always occupy the same position 
on the stem. They have a tendency however to appear towards the top 
of the principal and the upper secondary axes of the inflorescence and 
are not generally seen in the basal secondary axes of the inflorescence. 
This does not apply to some specimens, which have hermaphrodite 
flowers from the base úpwards. *. 
The tendency to group the hermaphrodite flowers toward .. the 
upper extíemites is more marked in the specimens that produce móre 
male than hermaphrodite flowers. 
At times we found secondary branches of an inflorescence entirely 
made up of male flowers; on trees furnishing these some infloresceh- 
ces were entireley deprived of hermaphrodite flowers. In no tree did wP 
find male flowers only. . 
Three or four individual occasionally show neutral flowers. 
The number of fruits produced is very inferior to that of flowers. 
On inflorescences which had produced several hundred flowers there 
were but 1, 2, 3, 4, at most 8 or 12, seldom up to twenty fruits to be 
seen. ; .' 
This must be attributed firstly to the existence of many flowers unfit 
for fuit production due to total (male flowers) or partial abortion of the 
pisti] (intermediate degres between male and perfect flowers). This fa- 
jtor already reduces the possibility of fruits in the best hypothesis by a 
—entieth, in the worst by much more. 
Other causes might be the singulary redueed amount of pollen and 
jerr.aps its lack of efficacity as already suggested for other families, 
f :he Umbellifera & Anacardiaceae by Engler and the Sapindacea 
7:t reduction of fruit formation is agravated further by the very 
freç~~r: fali of the small fruits formed. In Brazil it is generally attri- 
