58 Salisbury . — Variation in Er ant his hy emails , 
instance one member of the outer whorl had become transformed into 
a trilobed bract (Fig. 5, C) and the other two members of the whorl were 
undivided but exhibited a yellow central region and a green margin 
(Fig. 5, d). One or more members of the perianth may be trilobed and 
only partially virescent (Fig. 5, a). More rarely the trilobed structures are 
uniformly yellow (Fig. 5 , B). Very rarely, too, one or more members of the 
involucre of bracts may exhibit partial petaloidy (Fig. 5, H). Such variations 
are too frequent to be regarded as mere abnormalities, for not only have 
they been observed in every locality from which material was obtained by 
the present writer, but they were recorded by Irmisch in i860 as of frequent 
occurrence in the following passage: ‘Nicht selten wird ein Hiillblatt 
mehr oder weniger vollstandig in ein gelbes Kelchblatt verwandelt, oder es 
ist ein Kelchblatt zur Halfte grim, zur anderen Halfte gelb, oder einzelne 
Kelchblatter erscheinen, ohne die gelbe Farbe aufzugeben, nach Art der 
Hiillblatter mit tiefen Einschnitten, 
lauter Beweise fur die innige Be- 
ziehung dieser Blattkreise zu ein- 
ander’ ( 1880 , p. 225). 
There have thus been observed 
almost every possible transition 
between the involucral bracts and 
the perianth members. Moreover 
the latter may be replaced by the 
former. That such phenomena 
repeatedly exhibited by healthy 
plants have some significance can 
scarcely be denied, and one can 
therefore only conclude that either 
the perianth members are petaloid bracts and that both had their origin in 
foliar structures, or that the bracts are derivatives of the perianth members 
and that both had their origin from sporophylls. 
That the former view is the correct one is indicated on general grounds 
that will be considered later, but in this particular species we may note that 
the replacement of perianth members by bracts is not infrequent, whilst 
petaloid bracts are rare, and, as is well known, atavistic variations are usually 
much commoner than those of a progressive character. 
Moreover, the vascular supply of the perianth member corresponds 
closely with that of the bract, whilst even petaloid nectaries when well 
developed do not show the same vascular organization. It is, however, true 
that the supernumerary petals do help to bridge this gap. 
The most important argument is undoubtedly that in no case have we 
found a perianth member replaced by a nectary or showing any signs of 
transition to one. Thus if we were to accept the other alternative we 
