580 JAMES F. GEMMILL. 



question are incompletely separated from one another, partial doublinfr 

 or twin formation may result. 



Usually in starfish ova, doubling of this kind is associated with dis- 

 turbances so profound that differentiation ceases in the gastrula stage. 

 In my own experience great numbers of partially double blastulae have 

 appeared in different cultures of Asterias ruhens, A. glacialis, and Porania 

 pulvillus, but none of these was observed to reach even the early bipin- 

 narial stages, and so far as I know such stages have not been figured or 

 described. Possibly the ova of Luidia possess unusually great potenti- 

 alities of duplex development, but we may, perhaps, conjecture that the 

 long-continued shaking which the cultures would suffer during their 

 journey (the Thermos flasks were left only three-quarters full for reasons 

 connected with aeration) effected a physiological separation of masses of 

 cells during the formation of the blastula, and at the same time diminished 

 their vitahty less than do the more abrupt experimental methods 

 commonly employed in laboratory work. 



Mode of Formation. Among Fishes the first noticeable feature in the 

 genesis of double monstrosities is that two centres of gastrulation arise 

 on the margin of the blastoderm. Next, the resulting embryonic axes 

 are either brought together so as to unite posteriorly, producing the 

 anadidymous type, or else remaining separate they give rise to anakata- 

 didymous union of the embryos by means of the yolk-sac. The kata- 

 didymous condition is extremely rare, and, indeed, probably never 

 occurs in perfect form. In the birds and mammals the larger proportion 

 of double monsters arises in connection with two centres of embryo 

 formation, but Katadidymus is not uncommon, being caused in most 

 cases by fission of the posterior end of a developing embryonic axis. 

 In fishes, birds, and mammals, since growth of the axis takes place 

 almost entirely from before backwards, true anterior fission either 

 does not occur or is extremely limited in extent. On the whole, 

 we see that throughout the vertebrates the important feature in the 

 production of double monstrosities is the presence of two foci of embryo 

 formation, and that in the simplest group, the fishes, these foci are, to 

 begin with, centres of gastrulation. As regards the Asteroids, a glance at 

 the series of illustrations to this paper will show that here also the 

 formation of two centres of gastrulation precedes bipinnarial 

 twinning. Two more or less separate archentera are produced, and 

 various other structures are pai-tially or completely doubled. In the 

 end the two archentera may remain separate from one another (Figs. 1-4, 

 gastrulse ; Figs. 13-15, bipinnarige), but if the foci of gastrulation are 



