Gradual , Change , and by Guppy s Method of Differentiation. 623 
vividly red at one end and as vividly black at the other, the division 
following a perfectly straight line across the seed. Other species have simple 
bright red seeds. Incidentally, this is as good a case of f mimicry* as those 
described in animals, and was at one time seriously advanced as such. No 
use-value can be suggested, and no intermediates are possible, nor is it con¬ 
ceivable that Natural Selection would work to such minuteness of detail as to 
divide the two colours by an exact straight line in all cases. There is nothing 
for it but to admit that it must have been acquired at one step. And the 
same step must have been taken by several different plants, for the four 
genera do not belong to the same division of Leguminosae, and one of them 
does not even belong to the Papilionatae. 
We may proceed by giving one instance in some detail of the way in 
which a character that in one place is of family rank may in another become 
merely generic or specific. Whether the endosperm is or is not ruminate is 
the only general character of distinction between Anonaceae and Magno- 
liaceae, yet in Alyxia (Apocynaceae), in Aralidium and other genera 
(Araliaceae), in Polysphaeria (Rubiaceae), in Tinospora and others 
(Menispermaceae) rumination appears as merely a generic character, whilst 
in the palms, to give full details, it appears in the left-hand genus of each of 
the following pairs of genera that are placed side by side in the classification 
(numbers from Drude’s classification in * Die natlirlichen Pflanzenfamilien *, 
3 889): 
Ruminate. 
2. Chamaerops (sub-fam. I. 2) 
20. Copernicia (I. 2) 
23. Medemia (II. 1) 
30. Raphia (III. 2) 
42. Caryota (IV. 1 a) 
58. Catoblastus (IV. ic) 
69. Reinhardiia[ IV. id) 
73. Prestoea (IV. ie) 
78. Iguanura (do.) 
82. Heterospathe (do.) 
87. Phoenicophorium (do.) 
90. Oncosperma (do.) 
93. Ptychandra (do.) 
]oi. Ptychococcus (do.) 
109. Nenga , p.p. (do.) 
Not ruminate. 
3. Trachycarpus 
19. Screnaea 
24. Hyphaene 
31. Oncocalaimts 
43. Arenga 
57. Iriartea 
68. Synechanthus 
72. Hyospathe 
77. Linospadix 
81. Clinostigma 
88. Deckenia 
89. Acanthophoenix 
94. Cyphokentia 
100. Cyrtostachys 
no. Cyphophoen ix 
It is impossible to escape the conclusion that all or most of these 
fifteen genera on the left must have acquired rumination independently. 
They belong to nearly every group of palms, and agree closely in other 
characters, each with the one opposite to it. If we suppose them to have 
descended from a pair of ancestors that agreed to separate on the question 
of endosperm, we must explain why they also disagree in other characters of 
T 
t 
