Flowers of Stachys sylvatica, Linn. 421 
carpels of Clematis Vitalba Goebel (18) mentions that in the Caryophyl- 
laoeae the first flowers to open are often hexamerous, while the later ones 
are pentamerous, and that in Rut a graveolens the terminal flowers of the 
cyme are pentamerous, the others tetramerous. A survey of similar cases 
can be found in Vernon’s ‘Variations in Animals and Plants’ (40) and in 
de Vries’s ‘ M utation Theory ’ (43). The position of the flower together with 
a reduced number of parts in Stachys sylvatica suggest that bad nutrition is 
the causal factor, and it is probable that some cases of increased numbers of 
parts are due to increased nutrition, though some of these cases are probably 
the results of fusions between two or more flowers, and an explanation must 
be sought for these by considering the causes of fusions in Stachys flowers. 
The cases where five stamens develop, instead of the usual four, cannot 
be regarded as infringements on the ‘ law of loss ’ recently put forward by 
Dr. Agnes Arber (1), for in the allied Stachys recta Payer (29) described 
five rudiments of stamens, four of which develop. In the cases mentioned 
above it is possible that the posterior rudiment develops instead of aborting. 
The decrease in number of members in the corolla seems to take place 
independently of the time of year. An increased number of petals seems 
commoner in summer, and the change in the number of members in the 
androecial whorl, if not accompanied by any change in any other part of the 
flower, seems to be more dependent on the time of year than on the position 
of the flower in the inflorescence. 
As regards the fused flowers, it is difficult to assign any particular 
reason for the phenomenon. Worsdell (49) truly says that it is not 
surprising that such an occurrence is common amongst the Labiatae 
because the flowers are so very near to each other. The actual number of 
flowers is not of much importance, for in such species as Ballota nigra , where 
the number of flowers in a dichasium is very much greater than in Stachys 
sylvatica , fusions are rare — partly, no doubt, owing to the longer flower 
peduncles — and in Stachys itself the fusions do not seem to be found in 
greater numbers in portions of the plant where most flowers are found in 
a verticillaster ; and not only so, but where a fusion is found there is some- 
times no fusion in the opposite half-verticillaster, even if the latter contains 
more flowers than the former. The fusion, indeed, seems to be mostly con- 
ditioned by the non-development of the flower peduncle. Some check to its 
formation must happen just before the floral organs are about to be laid 
down. It has been noticed that in depauperate forms not only are the 
numbers in a whorl smaller than in the normal species, but that they are 
often fused. In very small scapes of Ophrys apifera I have found small 
flowers without the two posterior petals and with the posterior sepal fused 
with one of the side sepals, and similar cases have often been described 
elsewhere. Also in Fraxinus excelsior vegetative buds, that would other- 
wise only develop next spring, if at all, often open, in late summer or the 
