i io Salisbury .— Variation in Anemone apennina , Z., and Clematis 
variation of the perianth, but there are good reasons for regarding congenital 
fission, of the original segment rudiments, as the chief cause. The arrange- 
ment of the supernumerary parts is one ground for this view, but much more 
significant is the relation which the perianth number bears to the stamen 
number. From the following table it will be seen that the average number 
of stamens associated with a given number of perianth segments increases 
instead of decreasing when the latter is large. 
Number of 
Average Nu??iber 
Perianth Segments. 
of Stamens . 
9 
54 
IO 
5 r 
ii 
0 
12 * 
63 
13 
68 
I 4 
72 
15 
72 
16 
77 
x 7 
82 
18 
82 
19 
86 
20 
96 
21 
90 
That a negative correlation obtains between perianth number and 
stamen number when the former is augmented by transformation of the 
latter has been shown in the case of Chelidonimn majus by Karl Sax (1918), 
and conversely Bonnier noted (1889) that* petal-less flowers of Atragene 
alpina always had more numerous stamens than normal petalled flowers. 
It is of course patent from the work of Goebel (1905) and others (e. g. 
Saunders, 1917) that doubleness in its extreme form may involve trans- 
formation of stamens, which even in the same species (cf. Saunders, 1917) 
may or may not be accompanied by multiplication of parts. It is neverthe- 
less important to guard our minds against the assumption that doubleness, 
in the popular use of the term, signifies a homogeneous group of phenomena, 
and still more, from the unsupported conclusion that the normal polymerous 
perianth is a phenomena of precisely the same nature as that exhibited in 
the extreme ‘ double * flowers of horticulture. In reference to the non- 
homogerieity of the phenomenon of doubleness it may be pointed out that 
in two closely allied species of Diant hus Miss Saunders has shown that 
‘ doubleness ’ is dominant in the one and recessive in the other— a fact 
difficult to reconcile with the presence and absence hypothesis except on 
the assumption that the ‘ doubleness ’ is of a different character in the two 
cases. We may also note that in the somewhat parallel phenomenon of 
fasciation White (1917) considers this morphological feature to be due to 
very varied causes (p. 49 °)- 
