Chap, in.] the Dogma of Constancy of Species. 133 



of the leaf-stalks of Acacia heterophylla, which do not develop 

 their laminae, he refers also to this correlation of growth. He 

 finds the most remarkable example of the kind in the doubling 

 of flowers, where according to his view the disappearance of 

 the anthers is a condition of the corolline expansion of the 

 filaments; in the same way sometimes the carpel is changed 

 into a petal through the disappearance of the stigma. Though 

 in many of these cases it is quite possible to conceive of 

 the relations of cause and effect in the reverse way, yet 

 De Candolle's principle of correlation will be equally ap- 

 plicable. 



The second cause by which the symmetry may be obliter- 

 ated, namely degeneration, asserts itself in the formation of 

 thorns, of threadlike prolongations of membranous expansions, 

 and in the production of fleshy parts, or of parts with dry 

 membranes. 



The third kind of departure from the symmetrical plan is 

 the adherence of parts, the theory of which he grounds first 

 and chiefly on the phenomena of grafting, and then passes 

 to more difficult cases. The close packing of the ovaries in 

 some species of honeysuckle, is, he says, the primary cause of 

 their adherence. This therefore does not depend on the plan 

 of symmetry, but upon an accident, which however is constant 

 in its appearance, owing to the specific constitution of such 

 plants. In connection with the phenomena of adherence he 

 next considers the question whether a structure composed of 

 several parts, as for instance a compound ovary, should be 

 considered as originally simple and afterwards divided into 

 parts, or whether the converse is the true account, and he 

 says that we must examine each particular case and decide 

 which is the correct .conception. Thus it may be shown that 

 the perfoliate leaves of honeysuckles, as well as the involucres 

 of many Umbelliferae, and monosepalous calyces and mono- 

 petalous corollas are due to adherence, and he proceeds to 

 prove that ovaries with several loculaments and several parts 



