MTJLTIJTJGATE TYPES. 177 



anomalies. It is now possible to consider the actual specimen 

 (fig. 64) in relation to the theoretical scheme; this particular 

 head (23 + 24) represents a range of variation not included in 

 the observations of the Bravais, and the (16 + 26+42) type is 

 fairly constant for strong terminal heads. 



It will be noted that on tracing the ramifications of the long 

 curves in the manner adopted for Helianthus, the areas leading from 

 1, 5 and 7, as also 1', 5' and 7', correspond member for member, 

 but not these leading from 3 and 9. 



This is further seen to be due to the fact that 3 overlaps 7, 

 instead of falling clear of it; so that 3 is possibly the member 

 which has gone wrong, and the fact that the divergence angle be- 

 tween 1 and 3 was only 60° would be confirmed by the subsequent 

 error of the system. N'ew paths are being opened up from 9 as 

 compensation at this point, but it appears that the construction has 

 been thrown out by this displacement of one particular pair of 

 leaves. To what extent such an effect might be ascribed to the 

 action of the frosts at the time the capitulum was commencing is of 

 course not evident from the consideration of one specimen alone.* 



* It is clear, on the other hand, that too much importance must not be attached 

 to the low divergence angle between 1 and 3, when it is borne in mind that these 

 members are also contained in an expansion system derived from the (2 + 4) of 

 the vegetative shoot. 



That the new (16 + 26) system commences at 7, 7', suggests that the (6 + 10) 

 system was only completed at 5, 5', these members adding the last two short 

 curves of the system. Allowing a new curve for each member, on the lines of 

 Helianthus, this would imply that 1, 1' added short curves, and four long 

 curves were put in with the upper two pairs of foliage leaves : thus on a 

 capitulum (fig. 66) which agreed with the postulated construction of fig. 65, the 

 divergence between 1 and 3 was 59|°, which agrees with the preceding within 

 the limit of the error of observation. 



It will also be noted that subsequent growth is not uniform : the members 

 tend to come away from full contact, and a small amount of sliding growth 

 accompanying the dorsiventrality must be allowed for. It is possible that the 

 expansion from (2 + 4) to (6 + 10), as in the succeeding phase, is more rapid than 

 the Helianthus type, since lateral capitula of Dipsacus show the inflorescence 

 commencing immediately beyond the two vegetative prophyUs. The case of the 

 (23 + 24) capitulum, granting a loss in the shorter curves, suggests that another 

 expansion had commenced and added extra long curves beyond the type 

 (16 + 26). 



