— 151 — 
to traverse the white substance and to extend to the envelope 
of the spinal cord. In a great many cases, I have been able to trace 
them to the external layers of the white substance (vide fig. 90, pp ; 
g2,pp). In some cases I even beheve to have observed cells which send 
protoplasmic processes to the envelope on both sides of the spinal 
cord; fig. 90, gc represents siich a case; though there is not quite 
complete continuation of the process in this section, it seems, how- 
ever, to be the same process which crosses the central canal, sub- 
dividing and sending its branches just to the external envelope on 
the other side. The relations of the protoplasmic processes seem, 
thus, to be the same in Amphioxus as are found in the other animals, 
examined, where such processes were present; viz. they are peripheric- 
ally directed (here traversing the white substance). If we assume that 
they have a nutritive function this is easily understood, because as 
there are no blood vessels present within the nerve-cord, the pro- 
cesses must penetrate to the external layers to absorb nutrition. 
I have not yet had sufficient opportunity to investigate the 
course of the nervous processes, and will therefore reserve them 
entirely for the future memoir. 
The protoplasm of the ganglion cells has, in sections, a 
reticular appearance, produced, as I suppose, principally by a trans- 
section of primitive tubes, composing the chief constituents of the 
protoplasm (vide fig. 92, gc). These tubes penetrate into both the 
protoplasmic and the nervous processes and form their contents, 
whilst giving them a longitudinal striation. The protoplasmic pro- 
cesses have, however, a more granular appearance than the nervous 
processes. 
The cylindrical cells of the ep ithelium investing the 
central groove (or canal) have, in their external extremities, pro- 
cesses which traverse the white substance, and unite with the enve- 
lope of the nerve-cord, as already mentioned. These fibres are partly 
united into bundles (vide fig. 90), which apparently divide the longi- 
tudinal nerve-tubes into various columns. This division into columns 
is, however, only partial as the bundles do not form any continuous 
septa; they are only seen in some sections, whilst in other sections 
bundles of similar fibres may occur on other places. There are, 
however, some definite places where such bundles of fibres especi- 
ally occur; as some of the most prominent of these, two places, 
one on each side of the ventral colossal nerve-tube, may be men- 
tioned (fig. 90, ns, 7is). 
I have often observed very strong fibres or even bundles of 
