214 
PACIFIC SCIENCE, Vol. VI, July, 1952 
detailed and illustrated account. They placed 
it in the tribe Paypayroleae which had un- 
armed connectives and subequal petals, and 
put it next to the genus Paypayrola. 
The most recent tribal placement of Iso- 
dendrion is by Melchior in the second edition 
of the "Pflanzenfamilien” (1925: 347, 355- 
356). As was natural, he also relied heavily on 
Gray’s original description of the genus and 
two folio illustrations, but Melchior also saw 
and dissected a specimen of /. laurifolium 
Gray and printed the first figure of its stamen 
showing a terminally apiculate connective. 
His generic description is largely a German 
translation of Gray’s, but he adds, correctly, 
that the three placentae bear each 2-4 ovules 
(1925: 355). He also adds, incorrectly, 
"Stam. [ina] mit freien oder ganz am Grunde 
zu einem sehr niedrigen Ring verwachsen, 
flachen Filamenten . . . Nektarorgane auf dem 
Riicken der Filamente fehlend oder in Form 
von Schuppen in ihrer ganzen Lange den 
Filamenteri angewachsen. . . . Bliiten . . . 
griinlich-weiss oder rotlich . . After dis- 
secting the flowers of two of the three gen- 
uine species known to Melchior (the type of 
L longifolium having already vanished) and of 
the 11 additional species here described, no 
factual basis is seen for his statements that 
the stamens sometimes arise from a low, 
joined filament ring; that the staminal nec- 
taries may be dorsal scales growing on the 
full length of the filaments; or that the 
flowers are ever reddish. Leveille stated no 
flower color. Gray had surmised that the 
petals of /. pyrifolium and 1. longifolium were 
probably white; and Hillebrand stated for his 
composite /. pyrifolium that the petals were 
greenish white. Yet MacCaughey (1918: 10) 
stated in his compilation that I. longifolium 
had the flowers ’’purplish- white.” He left no 
specimens of this or any other Isodendrion, 
and no confirmation of his statement is 
known. Melchior’s statement that some 
flowers were reddish may have been a free 
translation of MacCaughey ’s "purplish- 
white.” He places Isodendrion in the sub- 
family Violoideae, the tribe Rinoreeae, and 
as the single genus in the subtribe Isoden- 
driinae. The other members of the tribe 
(Rinorea, pantropic; Allexis, African; and 
Gleospermum, tropical American) all differ 
fundamentally in having produced, elaborate 
connectives and the stamens mostly connate 
into a filament tube. 
During detailed dissection and illustration 
of the flowers of all the species, several correc- 
tions have been made in the generic characters 
of Isodendrion. The flowers are always zygo- 
morphic and single. The sepals are distinct 
and unequal. The petals unequal, the lower 
being longer and larger than the lateral, and 
the lateral longer and larger than the upper 
pair. The distinct stamens (never in any way 
united) are unequal in size, and all the sta- 
mens, or the two distal ones, are apically acu- 
leate or at least apiculate (in I. pyrifolium) by 
a prolongation of the connective. Several of 
the species have a swollen gland borne dor- 
sally on the filaments. The plants are shrubs, 
not trees. 
Though the flowers of Isodendrion are def- 
initely irregular and have a violet-like appear- 
ance, the lower petal lacks any spur or saccate 
enlargement, hence the genus cannot be 
placed in the tribe Violeae and must remain 
in the tribe Rinoreeae. Because of the signi- 
ficance of the staminal characters, it does not 
show close relationship to and should be kept 
distinct from the subtribes Rinoreinae and 
Paypayrolinae. It is much closer to the Hy- 
menantherinae, which differs from the Iso- 
dendriinae by having the fruit a berry, the 
sepals connate at base, the petals sessile, and 
the stipules deciduous. It has strong resem- 
blances in this subtribe to the genus Melkytus, 
containing four species of shrubs and trees 
from New Zealand, the Kermadec, Norfolk, 
and Fiji Islands. In Melicytus, however, the 
flowers are dioecious; the stamens almost ses- 
sile; ovary with 3-5 many-ovuled placentae; 
stigma 3-5-lobed; and stipules deciduous. 
These characters distinguish it clearly. How- 
ever, it is close to Isodendrion in having the 
