MORE SIDE-LIGHTS ON THE STRUCTURE OF COMPOSITES. 41 



the somewhat hooded tips of the petals ; so that, at first sight, the 

 stamens appear to be superposed to the petals as they are in 

 Primroses, but a httle further examination shows that the filaments 

 originate from the tube of the corolla, alternate with its lobes in 

 the ordinary manner. 



In the Dahlia above described, it was the hgulate condition of 

 the corolla which was exaggerated. In such flower-heads a much 

 larger number of the florets than usual become ligulate, and 

 physiologically emasculate. I have now to refer to some specimens 

 of Gaillardia jncta, in which precisely the opposite condition of 

 things was met with. This plant belongs to the same great group 

 of Composites, with strap-shaped imperfect ray-florets and tubular 

 4- or 5-toothed disc-florets, with stamens and pistils structurally 

 perfect. In the specimens to which I now more particularly refer 

 (see figure), the ligulate florets were completely absent; all the florets 

 had a 4- or 5-lobed tubular corolla, but the lobes were much longer 

 than usual ; the four or five constituent lobes of the corolla were 

 in fact nearly, if not quite, as large as those of the ligulate corolla. 



/N>^S7Vt%^ 



Taking the two cases in conjunction with what we know of the 

 mode of development and structure of the order generally, it is 

 reasonable to suppose, without drawing too much on the imagina- 

 tion, that the Composites are a highly specialised offshoot from 

 that large assemblage of plants which Linnaeus would have 

 grouped under Pentandria Dir/ynia, itself probably derived from an 

 earlier or possibly concurrent type, which would have been 

 included in former days under Pe^itandria Pentayynia. What was 

 the exact line of descent is at present so much a matter of mere 

 surmise that it is hardly profitable to enter upon it here. 



