230 
EDWIN S. GOODRICH. 
tlie litnb as a whole (12). The nerves supplying the ratlinl 
muscles are branches of the spinal nerves cori-esponcling to 
the inyotoines from which the inuscle-biuls were developed ; 
and the course of these nerves in the adult necessarily indi- 
cates the source of the fin muscles they supply, while their 
size is proportional to the amount of muscle derived from each 
particular segjuent. This is the case even when neighbouring 
muscular segments fuse to compound muscles, as usually 
happens in paired limbs.^ The branches of the segmental 
spinal nerves may combine more or less completely to form a 
complex limb-plexus, but they nevertheless preserve their 
original connections with the spinal cord, and also their 
original peripheral connections with the muscle-substance 
derived from their own segments. It follows that the nerve- 
supply in the adult is a safe guide in deciding both which 
segments have been concerned in the formation of the limb, 
and in what proportion they have contributed to it. 
We must conclude from these facts that the paired limbs do 
not bear a constant relation to any given segments in the 
Vertebrates, that they appear to shift along the segmented 
trunk, that this apparent shifting has taken place indepen- 
dently and repeatedly in the different classes, and that it can 
take place either forwards or backwards. For in the 
modern Anura the pelvic limb is almost undoubtedly farther 
forwards than in the ancestral Amphibian, and in the teleos- 
tean fish the more recent and modified forms may have the 
pelvic fin far forwards, until in extreme cases it reaches a 
position immediately behind the pectoral. 
Not only do the paired limbs shift their position, but they 
also vary greatly in their extent, being sometimes formed by 
many and at other times by only a few segments. In the 
fishes this variation is extraordinary, and in ^J'etrapods it is 
less pronounced, but still considerable. 
^ In my paper on Fins (12) I maintained, against the views of Brans 
and others, that the radial muscles preserved the original segmentation. 
But this is not the case, and the error was corrected in a subsequent 
paper (15). 
