268 DR J. M. MACFARLANE ON THE 



rows of hairs, and also median ones, as in C. purpureus ; but both sets are less numerous 

 than in the parent. The pollen-cells are of a pale reddish-brown colour, and each is 23 

 to 25 fi. 



It has repeatedly been pointed out as a rather peculiar fact in connection with the 

 high sterility of the hybrid parts, that the pollen-cells of it are mostly well formed. On 

 Plate VIII. figs. 14 a, 14 b, and 14 c, illustrations are given of the three, and though two 

 bad cells are figured from the hybrid, such are decidedly rare. 



Pistil, — In C purpureas the receptacular stalk and ovarian wall are quite glabrous. 

 The circumstigmatic hairs when fully grown are very densely set, and the longest are 100 

 to 120 fi. In C. Laburnum the receptacular stalk is glabrous, but the ovarian wall is 

 densely covered with spindle-shaped hairs. The circumstigmatic hairs are rather loosely 

 set, and are 160 to 180 ^ long. In C. Adami the stalk and ovarian wall are glabrous. 

 The circumstigmatic hairs are more closely set than in the last, and the longest measure 

 120 to 140 fji. 



The above results prove C. Adami to be even more unique in its minute anatomy 

 than in its naked-eye characters, remarkable though these are. The promiscuous mixing 

 up of tissue masses in the vegetative organs, such as stem, petiole, and lamina, and the 

 union of these, so as to give a tolerably intermediate physiological result in cork formation, 

 strengthening stays, sap conduction, and transpiration, is as curious as is the distribution 

 of the secretions of a tannin nature throughout the composite organism, or of histological 

 details in the floral organs. But the very striking resemblance which the epidermis of 

 the hybrid portion has to that of C. purpureus, not only in the general structure of the 

 cells, but in the size and structure of the cell nucleus, the distribution of the stomata, and 

 specially of hairs, would seem at first sight to prove that the hybrid portion was wrapped 

 round, so to speak, by an epidermis of C. purpureus. Other considerations, however, 

 show that the effect of the Laburnum parent has been to swamp or reduce by half as 

 exactly as we can estimate many of the " purpureus " peculiarities. Thus the number of 

 stomata over one side of the leaf; the reduction, as a rule, by half in the number of the 

 hairs over the floral parts ; also the reduction in size of the cells that form the floral parts, 

 all give countenance to this. It is, nevertheless, remarkable that where hairs grow out 

 from any epidermal surface in C. Laburnum, these should never be inherited by C. Adami, 

 and conversely where hairs grow out from C. purpureus, these are always inherited by C. 

 Adami, though reduced in number by about half. Still in Bryanthus erectus we have 

 shown that an approach to this condition is observable. 



If, however, we select either stem or leaf, and go over seriatim the structural resem- 

 blances or differences as compared with the parents, it will be found that, with the exception 

 of the epidermis, the tissues have greatest affinity with C. Laburnum. But some seed 

 hybrids, which we have described, may share in a similar one-sidedness of growth, though 

 not to so exaggerated an extent, so that the gap separating graft from seed hybrids is 

 not so wide as some have supposed. So far as the floral parts of C. Adami are concerned, 

 these might quite pass for products of a well-balanced seed hybrid, and it is only in the 



, 



