508 Agnes Arber. — On the Structure of the Androecium in 
a fascicle. In reply to this argument it may be urged that, while all 
evidence from abnormalities requires cautious handling, this is more than 
ever the case when the abnormality in question is of rare occurrence. When 
the same deviations from the normal occur in a large percentage of the 
flowers borne by any species, they may perhaps be regarded as the outcome 
of some definite tendency inherent in that species. For instance, the fact 
that five carpels instead of four are very frequently found in Parnassia 
palustris 2 and P. Kotzebuei 1 may well have some significance, especially 
since, in these pentamerous forms, the carpels alternate in normal fashion 
with the inner whorl of the androecium. But it is dangerous to draw any 
phylogenetic conclusions from the structure of two isolated, abnormal 
flowers. It must also be remembered that, many years ago, Buchenau 3 
described an abnormal flower of Parnassia palustris , in which one of the 
nectaries was partially transformed into a carpellary leaf, bearing both 
ovules and typical stalked glands. If we are prepared to attach phylo- 
genetic significance to such abnormalites, and to conclude from Wettstein’s 
work that the nectary is equivalent to a single stamen, we are also 
logically compelled to conclude from Buchenau’s observations that the 
nectary is equivalent to a carpel — a result which appears to be a reductio ad 
absurdum. 
When we leave the subject of the staminodes of Parnassia and turn to 
the comparison of the fertile stamens of this genus with those of the 
Hypericineae, we find that the anthers differ in a point upon which stress 
has sometimes been laid by systematists — namely, the dehiscence. In the 
Hypericineae this is introrse, while in Parnassia it is commonly described 
as extrorse. However, the distinction has lost some of its force since Gris 4 
showed that the anthers of Parnassia palustris are introrse in the young 
flower and only become secondarily extrorse at a later stage of development. 
Another and much more important difference between the fertile members 
of the androecium in Parnassia and the St. John’s Worts is the fact that we 
have to deal with individual stamens on the one hand, and fascicles of 
stamens on the other. At first sight it may seem that, even if the homology 
of the nectaries of Parnassia with the stamen phalanges of Hypericum 
were accepted, it would be impossible to imagine that the fertile stamens of 
Parnassia could come into the same category. Externally, these stamens 
are perfectly simple structures (Text-fig. 2 , A and D), each consisting of an 
undivided filament and a single anther. But, as we have shown in Sec- 
tion II (ii) of this paper, the vascular system of the filament and connective 
1 Seemann, B. : The Botany of the Voyage of H.M.S. Herald , London, 1852-7, p. 25. 
2 Pace, L. : 1 . c., p. 306. 
3 Buchenau, F. : Einige Beobachtungen aus clem Gebiete der Pflanzen-Teratologie. Bot. 
Zeit., xx, p. 307, 1862. 
4 Gris, A. : 1 . c., p. 915. 
