158 CRUGER's ORGANOGRAPIIICAL OBSERVATIONS 



organs must present themselves first, while according to various 

 observations, organs cease to grow there soonest, and the points 

 of the petals remain permanently free, it may be supposed that 

 the lower parts of the flower are developed in a state of cohesion. 

 The development of stamens, where, as is well known, the anthers 

 are complete long before the filaments begin to show themselves, 

 offers another proof for this hypothesis. Since the two lateral 

 outer petals are adherent to the three inner below, while, how- 

 ever, they must be supposed to stand in an external circle with 

 the large, intermediate outer petal, and originate at the same 

 epoch of development, it is natural that the two lateral outer 

 petals should cohere only by the borders where they touch the 

 middle inner petal, with each other and with the latter, while 

 the borders which touch the large outer petal should remain 

 free ; which is really the case, since the border of this petal is 

 interposed between them above, — of course, however, only to- 

 wards the apex of the flower, since all are uniformly coherent 

 below. 



The only difference met with in the other species of Heliconia 

 we possess here, is, that the outer lateral petals remain free 

 somewhat lower down. In other respects all is as in H. Bihai. 

 The inflorescence of these other species is the same, only the 

 flowers are less crowded, and the outermost bract standing next 

 the rachis is abortive, which confirms the above explanation of 

 the inflorescence of the first-named species. 



The following description of the flowers of Musa, as well as 

 the diagram (PL V. fig. 2), show that the Heliconics differ from 

 it not only in the inflorescence, but also in the place where the 

 sixth stamen is suppressed. This is not indicated in the syste- 

 matic works to which I have access. 



Musa. 



I have made out the following in the two species cultivated 

 here. The inflorescence is simpler than in Heliconia, i. e» be- 

 hind each bract (the apex of this bract sometimes grows out 

 into a leaf) occur two transverse rows which are developed si- 

 multaneously and have no partial bracts. The three outer and 

 three inner petals are confluent below, and the third inner, the 

 back of which is towards the rachis, is free. The suppressed 



