262 MORPHOLOGY OF GYMXOSPERMS 



(92) of the first wall-formation in Sequoia sempervirens is probably 

 true for the first walls of endosperm in general. At the last division 

 of the free nuclei (in this case at the two extremities of the sac), adja- 

 cent nuclei remain connected by the spindle fibers, which increase 

 in number, and upon them cell plates are formed. 



The details of endosperm-formation in Cryptomeria, however, 

 seem to be the most peculiar (93). The sac is filled with tissue by 

 centripetal growth as usual, but the formation of the permanent 

 endosperm is the remarkable feature. The "primary" endosperm 

 cells, that is, those open toward the center of the sac, elongate cen- 

 tripetally and by free nuclear division beome multinucleate (fig. 285). 

 There is then a stage when "hundreds" of these nuclei divide about 

 simultaneously, with no formation of a cell plate; but the kinoplasmic 

 fibrils extending between each pair of daughter nuclei increase in 

 number and curve outward on all sides, until both nuclei are com- 

 pletely surrounded by a sheath of fibrils, which fuse to form an 

 investing membrane. This method of free cell-formation goes on 

 throughout the whole endosperm except in the region of the archego- 

 nium initials; the cells become crowded, thus resembling ordinary 

 tissue composed of binucleate cells; and after this nuclear division 

 accompanied by wall plates proceeds in the usual way (figs. 285-288). 



In considering these cases of Sequoia and Cryptomeria, the second- 

 ary formation of cells by centripetally growing primary ones (" alveoli" 

 of Arnoldi) is common to both, whether the details of secondary 

 cell-formation are the same or not. There is also suggested the 

 possibility of a permanent secondary tissue which replaces the primary 

 tissue, the development of the latter usually having closed previous 

 investigations in this direction. Random reports of endosperm cells 

 that are uninucleate, or binucleate, or multinucleate may find their 

 explanation in stages of the development of permanent endosperm 

 tissue. 



More recently Lawson (156) has studied Pseudotsuga, and finds 

 that in that form also free nuclear division occurs in the primary 

 endosperm cells before cross- walls appear to form the uninucleate 

 cells of the permanent tissue. Saxton (159, 176) has observed the 

 same free nuclear division in the primary endosperm cells of Widdring- 

 tonia, but in this case the cells usually become only binucleate (occa- 



