194 MISSOURI BOTANICAL GARDEN. 



which two or three stout branches usually develop by the 

 side of the original apex, which now has ended its growth. 

 When these have reached a length of two or three feet 

 each forms a terminal inflorescence, and branches in its 

 turn, in this way giving rise to a repeated forking or 

 tripartition. On an overturned trunk, however, several of 

 the stronger branches usually become erect and grow to a 

 height equal to that of young trees, before blooming and 

 branching (PI. 8), but so far as I have seen they do this 

 without forming roots of their own, their supply of food 

 and moisture coming through the persistent roots of the 

 main trunk.* On the large stems, and even on some of the 

 larger branches of old trees, the reilexcd leaves gradually 

 fall away, and expose to view a very thick gray bark, deeply 

 fissured into quadrangles measuring about 1X2 inches. 

 As the trunks increase in height they also become much 

 thicker, the loosely fibrous, water-soaked wood being marked 

 in concentric rings, resembling those of Dicotyledons and 

 Conifers.j At the base, these older trees dilate quite 

 abruptly, from the development of a circle of thick con- 

 fluent roots, which correspond to those so commonly seen in 

 the form of more or less marked prolongations of ridges 

 and buttresses on Dicotyledons, and constitute the root- 

 bearing disk mentioned above (PI. 9). These large roots 

 possess a structure superficially similar to that of the trunk. 

 When preparing to flower, this is one of the most 

 attractive of all the Yuccas, the ovoid inflorescence buds, 

 each as large as an ostrich egg, being closely invested by 

 large thick white bracts, after the manner of a banana 

 inflorescence; but the bracts soon become dry and crum- 

 bling, and the expanded cluster, though of very compact 



* An excellent wood cut showing this habit of growth occurs in the 

 Gray Herbarium, evidently clipped from one of the English horticultural 

 journals, but I have been unable to obtain a more exact reference to it. 



t The structure and mode of thickening in arborescent Liliacere is very 

 fully treated by Roseler in Pringsheim's Jahrb. f. wiss. Bot. 1889, xx. 

 292-348, with several plates. 



