247 



ment these were sown. The result, greatly to my surprise, 

 was the appearance among a number of variants, normal 

 or irregularly inheriting the comparatively simple parental 

 form, of nine plants, all alike, in which the flat strap-like 

 specific frond had disappeared, being replaced by three 

 ball-like crests agglomerated together to form one ball-like 

 mass at the top of a long, bare stalk. This evidently 

 resulted from the fingered lobatum terminal of the parent 

 and the fanned tips of the basal lobes having been so 

 modified as to form three heavy crests on grandiceps lines, 

 while these absorbed so much of the energy of each frond 

 that the leafy or strap-like portion became suppressed, the 

 three crests thus being merged into one, though on 

 examination the triple character is clear enough. I 

 therefore named it as above. It is, so far as I know, an 

 absolutely unique form, and has, moreover, on several 

 occasions developed bulbil plants on the frond surfaces. 



By way of further experiment I sowed spores, which are 

 sparsely produced, some two years ago. The result has 

 been a considerable batch of plants, every one of which is 

 producing exactly the parental character without any 

 reversion at all to the grandparental type, so that the new 

 character seems to be entirely fixed. 



The original batch of nine plants arose, it will be 

 observed, from a sowing of spores from the wild form 

 before it had been under cultivation, taken, in short, from 

 a wild frond, and it is clear that had these spores been 

 scattered the new form might have appeared as an inde- 

 pendent find, which nobody would have been inclined to 

 attribute to the actual parent. As a rule, when discovering 

 varieties, it is impossible to find forms in the locality which 

 indicate the probability of step-by-step variation instead 

 of the sudden assumption of a new type in its full develop- 

 ment. Obviously, however, in this case, the possibility of 

 such gradings being masked by the extent of the change is 

 not excluded. 



C. T. D, 



