NOTES ON SCOTTISH WILLOWS 67 



structure of the stigmas and the length of the pedicel in the 

 ? flowers. From leaf-specimens it might readily be considered 

 to be 5. ludificaus, since the leaves combine the characters of 

 S.phylicifolia and .S. aitrita, their glabrosity, texture, and in 

 some degree their pubescence showing the former species, and 

 their shape, veining (especially of the young leaves), and 

 pubescence the latter. The catkins show some affinity also 

 to that species, but the evident style points to phylicifolia. 



It is to be hoped that plants bearing unisexual catkins 

 will yet be discovered. Although I have, in the description 

 given above, mentioned the hermaphrodite nature of the 

 catkins, that is of course no characteristic of the hybrid, 

 being merely peculiar to the specimens seen by me. 



In connection with this willow I should like to say a few 

 words about the nomenclature of hybrids, since there is much 

 divergence of opinion on the subject amongst botanists. 



In the " Revision " I expressed my belief that " whilst 

 there is a decided advantage in employing a compound 

 name — since it conveys distinct information — such can be 

 used in those cases only where no earlier name exists, and 

 where there is no doubt about the parentage." 



This opinion I am now inclined to modify. If only 

 binary hybrids existed the utility of adopting a name composed 

 of the designations of the parent species (when these are 

 certain) would admit of no doubt. Such a name conveys 

 information and is not too unwieldy. But when we have 

 to 'deal with hybrids into whose parentage three or more 

 species enter, it is doubtful whether the inordinate length of 

 the compound name would not altogether outweigh any 

 advantage it possesses. Names are not essential parts of 

 organisms, but are merely convenient symbols for indicating 

 the particular plant or animal under discussion. If such a 

 symbol is cumbrous — as it would be if the binomial system 

 of nomenclature were replaced by (say) a trinomial system 

 — its convenience would disappear. 



It seems to me therefore that for ternary hybrids we 

 must adopt a single and not a compound name, and that 

 whilst there is no imperative necessity — it would be 

 advantageous to do the same in the case of binary hybrids 

 also. 



