76 MORPHOLOGY OF ANGIOSPERMS 



(Marshall-Ward 8 ), Jeffersonia (Andrews 37 ), and Potamoge- 

 ton ( Holferty ' J1 ). In the ease of C ' ynomorium (duel 1 ' 1 ), the 

 two cells resulting from the first division of the mother-cell are 

 very unequal, the micropylar one being the smaller. This 

 smaller cell divides longitudinally and the larger one trans- 

 versely. Transitions to this condition in the formation of 

 oblique walls sometimes occur, as in Delphinium (Mottier 36 ). 

 The case of Fatsia japonica has been referred to above. As 

 already indicated, the completion of a tetrad is by no means 

 always attained, for there is every gradation between a row 

 of four megaspores and an undividing mother-cell that func- 

 tions directly as a megaspore. The explanation of this tend- 

 ency to shorten the megaspore series is probably connected 

 with the fact that only one megaspore of the tetrad functions. 

 The number of megaspores formed by the mother-cell has been 

 reported for very many plants, but the records are by no means 

 of equal value. The reasons for this are obvious. One is that 

 the sterile axial cells of the nucellus often simulate megaspores, 

 so that too large a number might easily be reported, and great 

 care is necessary to distinguish them ; and another is that the 

 technique of the earlier observers did not always permit cer- 

 tainty. By far the most important source of inaccuracy, how- 

 ever, is the hasty examination of a great number of forms by 

 a single investigator. Two megaspores might be reported, 

 when the same ovule collected a few hours later might have 

 shown four megaspores. As a consequence, much of the avail- 

 able data can be used only in a very general way as indicating- 

 tendencies of groups. 



Among the Monocotyledons, about one-third of those investi- 

 gated are reported as forming complete tetrads, in another 

 third the mother-cell does not divide, while the remaining forms 

 show every intergradation. Although one might expect the 

 complete tetrads to be characteristic of the more primitive 

 Monocotyledons, and the undividing mother-cell characteristic 

 of the higher families, there is as yet no such evidence, both 

 conditions occurring in all grades of Monocotyledons. 



The greatest variability is found among the Liliaceae, possi- 

 bly because more of the species have been investigated (compare 

 Fig. 28 with Figs. 35 and 36). For example, without attempt- 

 ing to include all the recorded cases, in Hemerocallis (Stras- 



