vi. PREFACE 
hybrids it is felt that the plan is not satisfactory. Primary hybrids usually 
combine the characters of their parents in sucha way that they can easily 
be recognised, and the variations assumed by different individuals from the 
same seed pod or the same cross are seldom great enough to prevent 
them from being recognised, and had the matter gone no further 
most of the difficulties could have been got over with very little trouble. 
But we now have hybrids of almost every degree of complexity—primary 
hybrids recrossed with their own parents, or crossed with other species or 
hybrids; hybrids derived from two species in which the parents are 
combined in equal and in unequal proportions; hybrids derived from three 
species, from four, and one even from five species ; and while some of these 
complex hybrids vary enormously between themselves they also in some 
cases resemble others that are known to have been derived from different 
crosses. In short there are hybrids whose parentage cannot be fixed with 
any degree of certainty by an analysis of their characters, because of the 
amount of reversion that has taken place. Then there is that curious 
complication that certain crosses which from their parentage appear to be 
distinct, yet on analysis prove identical. To put the case differently, the 
same hybrid may be obtained in two different ways. For example, 
Odontoglossum nobile crossed with O. X spectabile would appear to be a 
distinct hybrid from O. X Rolfez crossed with O. xX armainvillierense, but 
an analysis of parentage shows that both are composed of half O. nobile, 
a quarter O. crispum, and a quarter O. Harryanum. The two have been left 
under their respective names of O. X percultum and O. X Ossultoni, but 
it is at least a question whether they ought not to have been regarded as 
forms of one. And this is only a type of a series. Owing to these 
perplexing facts it becomes a question whether secondary and more 
complex hybrids may not in the future have to be treated in a different way 
from primary ones, or at all events have to be classified separately. The 
species in certain genera are now becoming so completely linked up by 
chains of hybrids that we may have to treat these complex forms purely as 
florists’ flowers, selecting and naming only such as show distinct improve- 
ments on their predecessors and ignoring the remainder. Some such 
selective method seems almost inevitable in the near future. 
With these remarks we commend the work to our readers, in the hope 
that their friendly criticism and kindly help will lead to its improvement 
in the event of another edition being called for. 
To Messrs. R. T., G. B. and F. W. Rolfe our thanks are due for help in 
reading the proof and in some other details. 
R. A. Roire, 
C. C. Hurst. 
September, 1908. 
