148 A. H. Graves, 



monocotyledonous embryo (Coulter and Chamberlain, 1903, pp. 188, 

 190 ff.). 



Instead of the increase of this row of cells from three to four, 

 or, in other words, to a row of three embryo-cells, as is the case 

 in the above-mentioned genera, Wille, as already stated, finds that 

 two plates of four cells each are formed from the two embryo-cells. 

 These two resulting four-celled segments are, nevertheless, still the 

 representatives of the two small embryo - cells, which are formed 

 first, in Ruppia as well as in Sagittaria and Alisma. 



That this condition is the usual one in Ruppia is shown not only 

 by Wille's (1883) observations, which may be correlated so well 

 with my own, but also by the figures and description of Hofmeister 

 (1852, p. 143 and figures 41-46, and 1861, figures 1-7, pi. \lf. 



My twelve -celled embryo (PL XIII, fig. 100), composed of three 

 four -celled plates, has obviously arisen by a transverse wall through 

 one of these four-celled segments — which one, it is not possible to 

 state. In the typical embryo of Sagittaria, indeed, it is the lower 

 of the ^two segments that undergoes a transverse division (Schafiher, 

 1897. II, p. 262 and PI. XXIV, figures 46, 48, 49), and this may be 

 the case here. 



What has occurred then is simply the formation of three seg- 

 ments, one above the other, comparable to the three upper cells in 

 Sagittaria (Schaffner, 1897, II, PI. XXIV, figures 48, 49), with the 

 difference that in Ruppia longitudinal divisions precede the trans- 

 verse ones. 



A comparison of this stage with those embryos of related genera 

 which have been worked out, brings to light the following points. 

 The embryos of Zannichellia (Campbell, 1897, p. 48 and cf. fig. 63 

 PL III; Hofmeister, 1861, PL I, fig. 18) and Potamogeton (Wiegand, 

 1900, pp. 37, 38, and PL VII, figures 25, &c.) are essentially like 

 that of the typical Sagittaria, consisting of a row of three cells 

 above the suspensor-cell, the terminal one being the first to undergo 

 longitudinal divisions. 



As to the embryo of Zostera, it was investigated at a very early 

 period by Hofmeister (1852), and quite recently by Rosenberg 

 (1901, I). Their results are not complete in the early stages, but 

 enough has been shown to indicate that Zostera is more like Ruppia 

 in the early development of its embryo than any of the Potamo- 

 getonaceae so far investigated. Hofmeister (1852, p. 139) states that 



' Practically the same figures are presented in , each of these articles : 

 in the former, the species is given as rostellata^ in the latter, q.s maritima. 



