MULTIJUGATE TYPES. 189 



variation capacity on the part of the plant, and entirely indepen- 

 dent of circTimstances of nutrition: experimental evidence may 

 throw light on the point. 



The (10 + 16) type was found to be constant to a remarkable 

 extent for lateral capitula of Bipsacus pilosus ; such capitula are 

 easily cut in early stages, and owing to the relative length of the 

 spiny bracts, the whole of the system may be obtained in one 

 section. As in other examples, growth is extremely uniform, and 

 although the members lose their lateral contact except at their 

 bases, they maintain their relative positions with great accuracy. 

 A section of such a capitulum, taken near the base, shows 

 unmistakably, however, that the contact edges of the rhomboid 

 members lie along the paths (16 + 26) (fig. 24, 2), and that the 

 appearance of (10 + 16) is therefore secondary, and due to the 

 fact that in the adult head the curves are counted from the con- 

 tact lines of florets rather than of the bracts. When these florets, 

 which tend to be more constant in volume on capitula of different 

 sizes, subtend a greater angle than the original member in whose 

 axil they arose, it is clear that new contact lines will be empha- 

 sised and the system apparently altered. A similar result occurs 

 in the elongated fruit-heads of Scabiosa atropurpv/rea, these in the 

 flowering condition show most usually terminal heads (10 + 16), 

 and laterals (8 + 13), as contact-lines for the florets which diminish 

 in size towards the centre ; in the fruit-head, owing to the greater 

 development of the involucels, the more obvious curves reduce 

 to (6 + 10) and (5 + 8), while the fact that the fruit must be all 

 equal in bulk is correlated with an elongation of the axis and the 

 tendency for the conversion of the curves into intersecting 

 helices on a cylindrical surface. 



A section of a similar capitulum of D. pilosus, taken at the 

 insertion of the terminal members, is of further interest in that 

 the fall of the bijugate system is shown to be absolutely regular, 

 and the last two sterile members are diametrically opposed. The 

 system, that is to say, remains bijugate to the end ; this may be 

 more strikingly demonstrated by numbering the members back- 

 wards ; the contact paths will be seen" to change from differences 

 by two to six, and by four to ten, as perfectly as in the number- 



