( 446 ) 
of the region of the fourth ventricle the cleft between the top of 
the chorda and the anterior row of cells widens. This anterior row 
of cells is then to be seen to extend to the infundibulum, and is at 
both sides connected with the head-mesoderm. 
At first solid, it becomes hollow, is shortened and lays itself closely 
at the back of the infundibulum; finally it disappears altogether. 
It seems to be the homologon of the connection-piece of the head- 
mesoderm described for selachians by BaLFouR and MARSHALL, closely 
studied by vAN Wine, the ,Sclerotomcommissur’” of KILLIAN. 
In now examining cross-sections through the head, we see that 
the mesoderm on both sides of the chorda does not break up (as is 
the case with the other Teleosts) into mesenchym, without being 
segmented, but that the somites of the trunk are continued without 
break as far as the auditory and farther on as far as the optic 
vesicles. These somites are, it is true, smaller and not as regular 
as the somites of the trunk, but everywhere myotomes and side- 
plates are to be distinguished very sharply. There is no trace of 
mesenchym until a late period. 
There where the chorda is rounded off as a distinct rod, the 
myotomes on both sides are separated from the chorda and from 
each other, more in front behind (and in young stages beneath) the 
infundibulum they are united by the connection-piece mentioned 
above. The somites are to be traced up to the foremost part of 
the head, where the optic vesicles are formed. They do not seem 
to be connected with the anterior mesoderm mass. 
In later stages of development the posterior somites of the head 
become hollow. The lumina fuse with each other, and the somites 
become small epithelial vesicles, which enlarge and form the head 
cavities; by the counection-piece they are connected with each other. 
The anterior walls of the head cavities and the solid somites lying 
in front of the head cavities form the eye muscles, but at present 
I cannot tell from which somites the different eye-muscles are formed. 
For the same reason I cannot fix the exact number of the head- 
somites, for, although the segmentation of the head-mesoderm into 
somites is to be seen in longitudinal sections as clearly as the 
peculiar form of the somites (as real myotomes) in cross-sections, Ì 
could not until now follow the different somites in their entire 
development. To make the necessary plastic reconstructions of the 
section-series time failed me and I am not sure if it will be possible 
at all, for especially the foremost somites break up after a time into 
mesenchym and are to such a degree compressed by the hind wall 
of the optic vesicles, that it is difficult to recognise them as different 
