Cheilanthes and Pellaea. 
68 1 
derive Ch . Fendleri from the normal type and the other Cheilanthes petioles 
from Ch. Fendleri . In Ch. lanuginosa the series is essentially that of 
Ch. Fendleri , except that the primitive endarch condition of the median 
protoxylem is not represented, and that this protoxylem lasts only over 
a limited distance at the base of the petiole. In Ch. gracillima there is 
omission of the early stages ; there is no sign of the median splitting of the 
trace, of the central protoxylem there is only an indication, and even that 
in an exarch position. Taken alone it would be difficult to fit this Fern 
into the Sinnott type, but in conjunction with Ch. lanuginosa and Ch. 
Fendleri the connexion becomes clear. 
The derivation of Ch. persica was mentioned when dealing with that 
species. In the case of Ch. persica especially, as also in that of the two first- 
mentioned species, one sees that Sinnott rather underestimated the plasticity 
of the Fern petiole when he spoke of the ‘striking constancy ’ of the endarchy 
of the filicinean petiolar bundle. While acquiescing in his general funda- 
mental classification and derivation of Fern leaf-traces, one sees in studying 
these five xerophilous Ferns that the anatomy of the leaf-base, as of the 
stem, may furnish a sound guide to relationships, but that it is not to single 
species but to groups of allied species that the test should be applied. 
Within quite narrow circles of affinity certain forms may slip ahead of the 
general evolutionary march. This holds good for the petiole of Cheilanthes 
gracillima or Ch. persica, which have diverged from the closely related Ch. 
Fendleri , just as for a solenostelic species occurring in a genus otherwise 
haplostelic (e. g. Gleichenia pectinata). 
Another point of interest is that petiolar structure may show a secondary 
simplicity. Thus in Cheilanthes Fendleri and Ch. lanuginosa the ordinary 
transverse section of the petiole shows a single compact mass of xylem. 
Nearer the base we find a stage at which, but for the protoxylem forming 
a connexion, the xylem is split medianly into two masses. This may give 
place at a still lower level to a single C-shaped strand. We have seen in 
deriving Ch. lanuginosa and Ch. gracillima from Ch. Fendleri that the earlier 
stages may be dropped out at the base of the petiole. Such a dropping out 
of early stages at the base, together with a secondary simplicity at a higher 
level, gives the state of affairs found in many Ferns (e. g. Onoclea sensibilis , 
Peranema , and Diacalpe ), where a trace, double below, unites to give 
a single strand at a higher level. This renders advisable a relaxation 
of the inflexibility of Sinnott’s assertions as to the constancy of the 
characters of the leaf-base, but such a relaxation really adds to the strength 
of the theory, and disposes of the objections raised by Davie (1), who 
claimed that his petioles in Peranema and Diacalpe were incapable 
of plausible explanation on Sinnott’s theory. The accumulation of 
this additional evidence for the theory of the primitiveness of structures 
found at the leaf-base, entails no necessity that one should accept all 
