404 



BOTANICAL GAZETTE 



[DECEMBER 



dividing by several successive periclinal walls, as most of the segments 

 promptly do, the superficial parts remain for some time conspicu- 

 ously undivided (p, p' , fig. /), and so are distinguishable by their 

 depth. In particular it is the final periclinal divisions that fail. 

 This will be more evident by following the usual segmentation of 

 the mother cell of an air chamber. In fig. 2, the young air chamber / 

 originated as usual in a cell which underwent one periclinal division, 



Fig. 5. — Elongation of a single gem- 

 miparous cell (p); w, rim cells, the 

 posterior oldest and most advanced in 

 division. 



Fig. 6. — A somewhat later stage than 



forming the wall /, 1; then an anticlinal one (2, 2) ; these two surface 

 cells each divided periclinally (3, 3), and cleavage occurred at the 

 junction of walls 2 and 3. (The curved anticlinal walls, 4, 4, suc- 

 ceeded the cleavage.) It is the divisions 1,1, and 3, 3, or only the 

 latter, which do not appear in the gemmiparous areas (p, p'). 



The relative extent of the area which these undivided cells cover, 

 and the occurrence of somewhat different cells (w) between them, 

 make it not unlikely that in figs. 1 and 2 the primordia of two cupules 

 are laid down in close succession; but of this we cannot be sure. 



The failure of the gemmiparous cells to divide allows their neigh- 



