318 BOTANICAL GAZETTE [APRIL 
with thirteen bivalents. This sequence has been completely 
followed in pollen through heterotypic mitosis and the ensuing 
homotypic division where (fig. 22) thirteen univalents show. In 
the embryo sac it has only been traced with definiteness through 
the compact orientation stage (fig. 17), while fig. 21a, 6 represents 
a badly masked anaphase showing thirteen chromosomes at each 
pole. In addition, it is likely that the metaphase shown by 
Stork in his fig. 16 is heterotypic, since it agrees with the general 
aspect of such a stage as found frequently in pollen. Whether the 
second (quantitative or homotypic division) can occur in the 
embryo sac as it does in the pollen maturation, giving true reduc- 
tion, is not known. Inspection of hundreds of embryo sacs failed 
to disclose tetrad formation, and yet the large numbers of empty 
fruits in T. laevigatum may eventually be explained by occurrence 
of reduced embryo sacs, never fertilized, quite as much as by the 
occurrence of amitosis of sequence D. 
Sequence B is shown in figs. 23-300. It is ostensibly the 
sequence which results in reproduction, inasmuch as it is the only 
mechanism found which in the absence of fertilization insures 
preservation of the constant chromosome equipment characterizing 
the species. In this sequence the nuclear membrane disappears 
before synapsis and orientation are complete. Synaptic pairing 
is end to end, but takes place so slowly that spindle fibers become at- 
tached to each of the halves of each univalent instead of to the 
univalent asa whole. In consequence the pairs come to metaphase 
thirteen in number, but with components still end to end and trans- 
versely oriented. The resulting spindle (figs. 27, 28) is much broader 
than that of sequence A, and the division is quantitative instead of 
qualitative, if ordinary canons be right. The partial or delayed 
pairing here was noted by Osawa in the pollen of 7. albidum 
only, did not attract the attention of SrorK, and seems to have 
been interpreted by JuEL (12) as a splitting. Hocpen (7) has 
described similar phenomena (delayed synapses) in parthenogenetic 
animals, while the present observations are amply verified by 
numerous counts made throughout the sequence. Sequence B in 
pollen seems to result in diads of fairly uniform nature, which as a 
rule do not undergo further growth. In the embryo sac it likewise 
