The Mor'phology of Ruppia Maritima. 149 



a four-celled embryo is formed from the first embryo-cell by two 

 Jongitudinal divisions occurring at right angles to each other, and 

 these quadrant-like cells are next each divided by a transverse wall. 

 So that, although the divisions here occur in different sequence, 

 the two four-celled segments nevertheless result. 



Although these first stages were not followed out by Rosenberg, 

 he presents a figure (1901, I, PI. I, fig. 20, and P. 12) of an eight- 

 celled embryo similar to that of Wille's (PI. XIII, fig. 99). It is 

 perhaps possible that later investigation will show that the first wall 

 in Zostera is transverse instead of longitvidinal, as Hofmeister has 

 described it. 



The task of tracing out the lines of demarcation of the primary 

 segments during the later stages, after the manner of recent embryo- 

 logical investigations, is very difficult, probably from two principal 

 -causes. First, the remarkable number of early longitudinal divisions 

 which have already been indicated to a certain extent obscure the 

 segment limits. Second, the form of the mature embryo differs so 

 widely from that of a typical monocot3dedon that even in the earl}^ 

 developmental stages this influence makes itself felt, and renders 

 comparison with type forms rather unsatisfactory. Wille, indeed 

 (1883, pp. 2, 3), describes his eight-celled stage as forming a 

 sixteen celled structure by transverse divisions in all of the eight 

 cells. But of the subsequent divisions he says (1. c, p. 3), " Herefter 

 ^aa Delingerne uregelmsessigt , saa man ikke laengere med Be- 

 stemthed kan f»lge de enkelte Cellers Delinger." 



Of stages older than the eight -ceUed embryo of Wille and Hof- 

 meister, I was fortunate in having a fairly large number, and in 

 the following description I shall attempt to show to what extent 

 -the segment boundaries may be traced during the embryo develop- 

 ment, and how they may be correlated to such a type as Sagit- 

 taria. 



I have already suggested that such an embryo as is represented 

 in PI. XIII, fig. 100, composed of twelve ceUs, or three four-celled 

 segments, arises from Wille's younger form by a transverse wall 

 through one of the segments, and that this is parallel to the case in 

 vSagittaria, where a three-celled row arises by a transverse division 

 in the lower embryo-cell. 



Fig. 101 is a slightly older embryo, in which the divisions are 

 somewhat irregular, but still admit of an interpretation which reveals 

 the outlines of the cells and segments in the preceding stage. It 

 is evident that a transverse division has occurred in the terminal 

 segment. 



