578 JAMES F, GEMMILL. 



Monstrosities are conveniently divided up into Anadidymi, Katadidymi, 

 Anakatadidymi, Mesodidymi (5 ; 14 ; 6, p. 3). To these a small group 

 falls to be added containing the few recorded examples of simple longi- 

 tudinal or parallel union (6, pp. 4, 29), which I venture to suggest may 

 suitably be termed Paradidymi. The Anadidymi are, of course, forms 

 with the anterior end more or less double, and the posterior end single ; in 

 the Katadidymi these conditions are reversed. The Anakatadidymi 

 show anterior and posterior doubling, but are single in their middle 

 regions, while the converse relations are characteristic of the Mesodidymi. 

 In the Paradidymi doubling occurs in an equal or sub-ecj[ual degree 

 throughout the whole of the longitudinal axis of the twins. 



In fishes and other vertebrates, the notochord, the vertebral column, 

 the central nervous system and the alimentary canal, serve as our prin- 

 cipal guides in judging to which group a particular double monster should 

 be assigned. In double bipinnarise, on the other hand, we have to 

 depend entirely on the alimentary canal, inasmuch as the only other 

 easily recognisable longitudinal structure, namely, the posterior ciliated 

 band, owing to its superficial position, in most cases shows a greatly 

 lessened amount of doubling throuoh the w^orking of " regulation "' 

 processes. Nevertheless, if the alimentary canal be taken as a guide, it 

 is remarkable how readily the various types of duplex bipinnarise fall 

 into the same kinds of groups as double-monster fishes. Thus Figs. 13 

 and 14 illustrate longitudinal or parallel union and are therefore Para- 

 didymi ; Figs. 16 and 17 belong to the Anadidymi ; Figs. 18 and 19 to 

 the Katadidymi ; Fig. 20 is Anakatadidymous, and Fig. 21 Mesodidy- 

 mous in type. Probably, further search among the abnormal bipinnarise 

 would have revealed a still fuller and more representative series. There 

 remains to make mention of Fig. 15, which illustrates what may be 

 called tangential union, and would no doubt have included the bipin- 

 naria from the larva shown in Fig. 7 had survival been allowed. The 

 twin embryonic axes, as represented here by the alimentary canals, 

 are independent of, and widely divergent from, one another, but there 

 is superficial union of the lateral or frontal body-walls. Among monster 

 fishes we have no exact counterpart of this type, since, owing to the 

 manner in which the twin embryonic axes develop, practically the only 

 alternative to axial union is an Anakatadidymus effected through the 

 intermediary of the yolk-sac. However, in the amniotic vertebrates, 

 and particularly in the mammals, numerous instances occur in which, 

 without axial union being present, the twin organisms are united to 

 one another by paraxial or superficial structures. 



