THE SHAPES OF VERTEBRATE SKELETONS. 213 



caused by muscular contractions on the concave side, will be 

 made clear by the rude illustration which a bow supplies. 

 A bow may be bent by a thrust against its middle (the two 

 ends being held back), or it may be bent by contracting 

 a string that unites its ends; but the distributions of me- 

 chanical forces within the wood of the bow, though not quite 

 alike in the two cases, will be very similar. Now while the 

 muscular action on the concave side of a fish differs from that 

 represented by the tightened string of a bow, the difference 

 is not such as to destroy the applicability of the illustration: 

 the parallel holds so far as this, that within that portion of 

 the fish's body which is passively bent by the contracting 

 muscles, there must be, as in a strung bow, a part in com- 

 pression, a part in tension, and an intermediate part which 

 is neutral. 



After thus seeing that even in the developed fish with 

 its complex locomotive apparatus, this law of the transverse 

 strain holds in a qualified way, we shall understand how 

 much more it must hold in any form that may be supposed 

 to initiate the vertebrate type — a form devoid of that 

 segmentation by which the vertebrate type is 1 more or less 

 characterized. We shall see that assuming a rudimentary 

 animal, still simpler than the Amphioxus, to have a feeble 

 power of moving itself through the water by the undulations 

 of its body, or some part of its body, there will necessarily 

 come into play certain reactions which must affect the median 

 portion of the undulating mass in a way unlike that in 

 which they affect its lateral portions. And if there exists in 

 this median portion a tissue which keeps its place with any 

 constancy, we may expect that the differential conditions 

 produced in it by the transverse strain, will initiate a dif- 

 ferentiation. It is true that the distribution of the viscera 

 in the Amphioxus, Fig. 191, and in the type from which we 

 may suppose it to have -arisen, is such as to interfere with this 

 process. It is also true that the actions and reactions de- 

 scribed would not of themselves give to the median portion 



