462 THE SAX.IX LISTS IN THE ' LONDON CATALOGUE.' 



than either Andersson or Sir J. D. Hooker. This latter step has 

 produced a confusing result, since under S. phylicifolia Linn, stand 

 lettered as three varieties — 



a. phylicifolia Linn. auct. 



b. phylicifolia— nigricans Wimm. 



c. nigricans Bm. 



The second of these names is used, not in Wimmer's sense, but 

 " as a convenient designation for the numerous forms which cannot 

 be referred to either of the extremes " ; while under each of the 

 vars. a. and c. are ranged a number of hybrids, an innovation 

 which has a clumsy appearance and is out of keeping not ouly 

 with the rest of the list, but with the whole Catalo(jne. It may 

 readily be admitted that the numerous forms of 8. nif/ricaiis and S. 

 ph 1/1 ici folia run into one another, and on this ground, in any other 

 genus almost, this would be a strong argument for combining the 

 two species in one. But in a genus in which hybridization not un- 

 frequently occurs, and definite lines of demarcation are obliterated 

 by recrossings of the species with one of its hybrid offspring, the 

 many intermediates between type S. niffricans and type ^S'. phijli- 

 cifulia are easily accounted for. There is a similar case in the 

 regular series of gradations that occur between S. alba L. and 

 S./ragilis L. As Dr. White has observed, in describing <5. viridis 

 Fr., " the hybrid, as met with, more frequently shows a departure " 

 from typical viridis " towards either t'ra;/il is or alba, till finally it is 

 almost impossible to separate it from one or other of these species " 

 [Journ. Linn. Soc. xxvii. pp. 372, 37b). There are two alternative 

 views to choose between, either of which would account for the 

 present state of things : one is to regard S. alba and S. fiai/ilis as 

 old-established species which by hybridization and innumerable 

 crossings and recrossings between the types and the hybrids have 

 given us in the present day every variation ; or to regard these two 

 as one original species, which has gradually diverged in two direc- 

 tions, and presents to us now two extreme forms, each of them 

 very common, well-marked, and usually constant iu character, 

 accompanied by some variations mostly lying between these two 

 forms, but as a rule much scarcer than either of them. Even 

 if this latter theory should be the more acceptable alternative, the 

 time would surely have come w^hen the two extreme forms, in this 

 case S. alba and *S'. /ragilis, would deserve recognition as distinct 

 types and the specific rank which has always been accorded them. 

 In the same way, on either theory, it is more convenient, if not 

 more true, to regard S. ni(/)icans and S. phi/licifolia as specific types. 

 In Dr. White's list, that is, in Andersson's specific order, 

 S. viminalis L. has place between S. Arbuscula L. and S. lanata L., 

 and the S. purpurea group brings up the rear. Searching for a reason 

 why the lowland S. viminalis should be inserted between these two 

 alpine species, I perceived that, though it has little in common with 

 ^'. lanata beyond the long style, and with S. Arbuscula besides the 

 short pedicel and long narrow nectary, S. viminalis is so closely 

 allied to S. Lapponum L. in every detail of the catkin, and also in 

 the foliage, that the latter may be regarded as the alpine correlative 



