SECRETIVE TISSUES. 149 



invariably the case ; so that one cannot but infer that a 

 common cause has brought about their correlated positions. 

 This close correlation is, of course, especially observable in 

 the more highly differentiated flowers. In regular flowers, 

 accessible on all sides, the glands are placed symmetrically 

 round the flower — whether on the sepals, as in Lime ; on the 

 petals, as in a Buttercup; or on the receptacle, as in Geranium 

 pratense, — or else there is formed a disk, as in so many "disci- 

 floral " plants. As soon, however, as a flower begins to show 

 some tendency to irregularity, or the flow^er is visited in one 

 way only, the honey-secreting organ at once becomes more 

 restricted in localization; as in the Wallflow^er, where it forms 

 two cushions, out of the middle of which the shorter stamens 

 arise, while the petals form two pseudo-tubes leading down 

 to those two glands. Again, in the Lahiatce, so markedly 

 zygomorphic, the honey-gland is often restricted to the 

 anterior side, on which the proboscis is inserted. Similarly 

 in Antirrliinum majus, "the honey is secreted by the smooth 

 green fleshy base of the ovary, whose upper part is paler in 

 colour and covered with fine hairs ; ... it remains adherent 

 to the nectary and to the base of the anterior stamens. The 

 short wide spur permits the insect's proboscis to reach the 

 honey from below ; above and in front it is protected by a 

 thick fringe of stiff knobbed hairs on the angles of the 

 anterior stamens."* 



It is hardly worth while giving other cases to prove the 

 universal rule, that the position of the honey and its gland is 

 always where it is most accessible ; and the position of the 

 anthers is, at the same time, just where they w^ill be most 

 conveniently struck by the insect; while the style and stigma 

 supply a third correlation, so that the latter organ invariably 

 hits the insect where the pollen has been previously placed. 

 * Miiller, Fertilisation, etc., p. 433. 



