340 
PACIFIC SCIENCE, Vol. IV, October, 1950 
genera Lagenophora , T etramolopium, Ar gy- 
ro xiphium, and other genera with the "Du- 
bautia type pollen.” So, at least, the pollen 
structure provides no significant differences 
between Dubautia and Railliardia. Selling 
twice accompanied Skottsberg on field trips 
to the Hawaiian Islands and Selling’s rejec- 
tion of Keck’s work seems in harmony with 
that of his elder companion, Skottsberg. 
Skottsberg (1944: 510) discounted Keck’s 
revision. He discussed Keck’s evaluation of 
the characters and consequent fusing of the 
two genera under Dubautia, and said, "I can- 
not find, however, that he has brought to 
light any new circumstances, not known to 
Hillebrand, Sherff, Degener and other writers 
on this subject, and they have considered it 
better to keep them separate. Nothing is 
gained by uniting them.” Then Skottsberg de- 
scribed a new species of Raillardia {— Rail- 
liardia}. To this it may be replied that the 
gain attained by Keck is the delimiting of a 
genus with diagnostic characters that can be 
stated in words by one botanist and applied 
with the same understanding by others. The 
reduction of Railliardia to Dubautia is here 
accepted. The only recent authors accepting 
the view of Keck seem to be Hartt and Neal 
(1940: 264), Fosberg (1943: 395-397 and 
1948: 115), and Neal (1948: 743). Fos- 
berg, in his "Summary of the Hawaiian Seed 
Plants,” places Railliardia as a synonym of 
Dubautia and indicates that it is a descendant 
of the original immigrant Ar gyro xiphium. 
Of the several characters previously used 
for generic separation, the best and most sig- 
nificant are those of the freedom or union of 
the involucral bracts and the degree of cilia- 
tion of the pappus. Both of these characters 
are significant and have always been used in 
the classification of these species into larger 
groups. However, the two pairs of characters 
are not always correlated. Six species have 
the involucre of separate bracts, and the pap- 
pus awns short ciliate lacerate or short ciliate, 
these representing Dubautia of Gaudichaud. 
Twenty species have the involucral bracts 
united into a campanulate or funnel-form in- 
volucre, and the pappus awns long plumose, 
these representing Railliardia of Gaudichaud. 
There remain three species, all from Kauai, 
which have the involucre campanulate, united 
for from 14 to ^4 of its length, and the pap- 
pus awns short ciliate. These three perfectly 
recombine the characters of the two older 
groups. As they cannot be logically fitted into 
either of these groups, a new subgenus, Mixta, 
is made to receive them; the other two groups 
are described as subgenera. 
Keck removes the Hawaiian genera Du- 
bautia and Railliardia from the Madieae and, 
quoting Bentham’s opinion, refers to their be- 
ing likened to Robinsonia and Rhetinoden- 
dron of Juan Fernandez. The similarity is in 
their habit of growth as small rosette trees, 
but in characters of flowers and fruit there 
are fundamental differences from these ge- 
nera of the Senecioneae. Keck inclines towards 
the views of Skottsberg that close relatives 
are Bedfordia of Australia and Brachiono- 
stylum of New Guinea. These are both of 
the Senecioneae, and, on analysis of their 
characters, are markedly different. Bedfordia 
is a genus of two species of Tasmania and 
Victoria in Australia — shrubs or trees with 
stellate tomentose leaves, heads axillary or 
in axillary panicles, and the denticulate capil- 
lary pappi very numerous and caducous. 
Brachionostylum is a monotypic genus from 
the high mountains of southwest New Guinea 
— a shrub with the heads unisexual, hetero- 
gamous, anthers free, and the pappus bristles 
slender, capillary, roughened, and early cadu- 
cous. On comparison, neither of these genera 
has the habit, or similarity of flower and fruit 
structure, to suggest that it is a close relative 
of Dubautia. 
Keck agrees with the Engler and Prantl 
system in placing the Hawaiian Dubautia and 
its relatives in the Heliantheae-Galinsoginae; 
that seems the best placement at present, 
though by its structure and habit it is not a 
