CONSTANT PHYLLOTAXIS. 97 



leaf by leaf.* If then in the same construction the genetic spiral 

 remains constant, the direction of the parastichy curves must be 

 reversed ; although it is possible that the plant-apex might stop and 

 reverse to complementary lines of equal distribution, such an 

 arrangement would undoubtedly appear in the form of a sharp line 

 of demarcation between the cycles, and would be recognised, if the 

 system were continued to any distance, as a reversal of parastichies.f 



There does not appear to be, then, any practicable method of 

 passing from one cycle to the next without distortion being appar- 

 ent. Thus, in the inflorescence of Bipsacus fuUonum, a new paras- 

 tichy curve may be initiated in the middle of the capitulum 

 (fig. 38a), or curves may be dropped out one at a time (fig. 386). 



The distortion is very obvious, and appears, in the former case, 

 to be initiated by the development of two primordia in the place 

 of one when the error of construction, due to the greater increase 

 of the axis bearing a constant lateral member, becomes sufficiently 

 large. An identical phenomenon, in which a new ridge is 

 added, or dropped out, probably in correlation with conditions of 

 nutrition, is afforded by the stems of Cacti, the distortion produced 

 being again considerable | (fig. 39a, b). 



From the standpoint of uniform growth, no transition from one 

 phyllotaxis series to another is possible without distortion. Any 

 such changes, therefore, when they occur, must be secondary, and 

 the compensatory allowances must render the distortion more or 

 less obvious. 



Thus, given a certain ratio between the diameter of the primordium 

 and that of the axis producing it, the growing apex works out the 

 system, until, hy constant repetition, certain members fall into series 

 vjhich give the subjective appearance of log. spiral parastichies. 



Any secondary change in the system must result in the disturbance 

 of these parastichy curves, and either appear as a " distortion " of the 

 series, or completely break up the system. 



In other words, so long, and so long only, as the ratio between 

 axis and lateral primordium remains constant, and a given axis 



* Gf. Bellis (fig. 47). 



t Gf. Saxifraga umbrosa (fig. 52). 



J Gf. Vbclitiiig, Brings. Jahrb., xxvi. 438. 



