THE PRIMARY SEGMENTATION. 121 



depend on the steps of ancestral history being repeated in a 

 complete and orderly way in any animal form, and our observa- 

 tions give us only circumstantial evidence from which a balance 

 of probabilities must be struck to determine what is ancestral 

 and what is secondarily acquired. The chain of evidence is 

 incomplete, and must always be supplemented by a certain 

 amount of inference, but it has not been fully recognized in 

 the practical study of embryological development that the 

 shortest intervals of time may be very important in keeping 

 the connection. Coherency of history must be preserved. 

 The difificulty of doing so is greatly increased by the fact that 

 the new is made to proceed out of the old, and frequently one 

 organ insidiously takes the place of an earlier formed one. I 

 am glad of the opportunity to say, in this company of investi- 

 gators and students who are preparing for independent research 

 in biology, that too great stress cannot be laid on the desir- 

 ability of having a more complete series of stages for study. 

 The traditional method has been to study one stage, and then 

 another "a little older," and fill in the gap with inferences. 

 This has proved to be inadequate and misleading. It is now 

 required that we shall have stages near enough to trace the 

 history of the transitory as well as the permanent organs ; 

 embryologists are just beginning to realize how transitory 

 some organs are. I have recently had occasion to examine a 

 set of embryonic structures in the chick, which do not appar- 

 ently last more than an hour or two in the course of develop- 

 ment, but which are, nevertheless, clearly defined for that 

 period and then fade away. In this particular case the agree- 

 ment of two observers, even as to the presence of these organs, 

 would depend on their having stages of identically, not approxi- 

 mately, the same period of development. A wider recognition 

 of the existence of such conditions would give us fewer contro- 

 versies and less biological mythology. 



Above, it was stated that the vertebrate head is primitively 

 segmented, and it is manifestly an interesting problem to deter- 

 mine the number, the nature, and the transformations of the 

 segments that have entered into the composition of the head. 

 The individual segments are called metameres, or somites, and 



