946 JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 
study. Those of D. filiformis are abundantly covered externally with 
glandular hairs (fig. 105, No. 7a) which vary greatly in structure and 
length. They may be from 180-380 » long. Generally it may be said 
that all of them end in a flat-topped mass of 4-12 glandular cells, but 
the stalk supporting these may be a single cell row or two rows below 
gradually running into one or two rows above. Each stalk also consists 
of 3-6 or 7 tiers of elongated cells. In the distal hairs of each sepal 
spiral trachee may enter the base of the hair and soon end blindly, or 
may be prolonged about one-third the length of the hair. Rosette- 
shaped, four-celled, sessile glands, and toward the base of the sepals 
two-celled glands of similar appearance, are also disposed over the 
exterior. Stomata are abundant, are nearly circular in outline, and 
measure 380 X28. The internal (upper) epidermis is glabrous, and con- 
sists, like the upper epidermis, of very variable cells, alike in size and 
shape. In D. ixtermedia the outer surface of each sepal is glabrous, 
but a moderate number of two-celled glands on a slight stalk cell, as 
well as stomata, occur over its surface. In D. hybrida the outer-surface 
of each sepal bears the same type of stalked gland hair as is seen in 
D. filiformis (No. 7b), but here they are on the average only one-third 
to one-fourth the length of those on the parent species. But, like the 
parent, they vary in structure amongst themselves. Besides these are 
the four-celled sessile glands, which differ in no way as to size from the 
parent, each being 30 » across. They are more scarce, however, than in 
the parent. We thus have the interesting case shown of one type of 
parental hair inherited in greatly reduced numbers and of much smaller 
size, and another type inherited in reduced numbers, but of exactly the 
same size. Still further, the hybrid inherits bilobed glands from 
D. intermedia in moderate quantity. Three types of epidermal glandu- 
lar appendage are thus inherited by the hybrid from its parents. 
I do not propose to describe in detail at present all the points of 
floral or fruit and seed structure. The careful observations of another 
season will be needed before exact comparisons can be made. It may be 
said, however, that the pollen grains of D. jiliformis are richly granular, 
largest in size, and measure 56 across. Those of D. intermedia are 
also granular and plump: they measure 44 yp across. Most of the 
hybrid grains are more or less starved or poor in protoplasmic substance. 
They measure 48-50 1, so that development of the pollen grain walls has 
proceeded perfectly, though the enclosed fertilising substance appears to 
be imperfect. ‘The ovules and seeds of both parents matured well ; those 
of the hybrid remained small and in most instances developed as empty 
or nearly empty shells. Cultivation and future study will demonstrate 
how far this may be a constant character. 
The naked eye and histological details described above emphasise the 
position first fully established by the writer * that a hybrid is, as a rule, 
down to its minutest details, a blended reproduction of both parents in 
which the morphological and physiological details of each are reduced by 
half. In no ease has it been possible to detect the entire loss in the 
hybrid of some parental condition, and this cannot be too strongly in- 
sisted on in view of the loose theoretical reasonings often indulged in 
* Trans. Roy. Soc. Edin. vol. xxxvii. p. 203. 
