72 JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 



and varied, that a new arrangement, founded on evolutionary or 

 biological principles, is now desirable — a system which, begin- 

 ning with the simple, or most primitive forms of any one genus, 

 shall gradually trace out their development or evolutionary 

 growth, instead of placing them, as now, by the rule of thumb, 

 or the rule of a, b, c. In my forthcoming " History of the 

 Narcissus," I hope to arrange the species of Narcissus and their 

 neighbours and relations on this natural plan. Since I wrote 

 the above, I find this view well expressed in the Gardener's 

 Chronicle of April 13, 1889, p. 467, in the following words : — 



" An analytical table or key, being framed wholly on grounds of 

 convenience, may, of course, be purely artificial or arbitrary in its 

 arrangement. A natural system, the aim of which is to secure an 

 arrangement which shall most closely represent the real or assumed 

 affinities or degrees of filiation and descent, should, could it be 

 perfected, have nothing more arbitrary about it than a genealogical 

 pedigree ; each plant should be in its proper place, and its exact 

 relationship to all other plants should be precisely defined. Unfor- 

 tunately this ideal has not been reached, and many circumstances 

 concur to render it impossible that it ever will be realised. Neverthe- 

 less, it is capable of constant improvement, and every new fact is a 

 step to perfection." 



Of course those who know most of Botany will best know 

 how difficult this ideal system will be to work out ; and yet, if 

 every one of us works out a special group or genus, I think the 

 task a possible one, and certainly one worth attempting. 



So early as 1837 we find Herbert referring to Narcissus 

 (Hermione) deficiens (Bot. Reg. t. 22, fig. 1) (=N. serotinus, 

 var.) as being in his opinion " the nearest existing plant to the 

 first Narcissus " (Journ. Hort. Soc, vol. ii. p. 27), and he further 

 says, " I should take Zephyr anthes minima, and Z. gracilis, and 

 the genera Carpolyza, Hessea, and Acis to be nearest to the 

 created type of Amaryllidaceffi " (Journal R. H. S., vol. ii., pp. 

 27 and 28). 



When I suggest that the flowers of the Narcissi we see around 

 us to-day are developments of something else — some other plant 

 less highly specialised — you will naturally ask me one or two 

 awkward questions. Firstly, from what plants the Daffodils 

 and Narcissi have been developed ; and secondly, how or by what 

 means I can prove what I say. Well, I shall not try to prove 

 anything I do not assert. I merely make a few general sugges- 



