456 BERTRAM G. SMITH 



this paper I am greatly indebted to Prof. Bashford Dean, under 

 whose guidance it has been brought to completion. 



During the past seven years the collection of an abundance of 

 material has enabled me to preserve an ample supply in every 

 stage of development. At least fifteen thousand embryos have 

 been secured from nests, and nearly as many more have been 

 obtained by artificial fertilization. 



For convenience in description, the embryonic and larval his- 

 tory has been divided into stages, based chiefly on external char- 

 acters. In making the division, the usual difficulty has been 

 encountered, that the rate of development of each structure 

 varies more or less in different embryos. In detei'mining what 

 shall constitute the interval between stages, the guiding principle 

 has been to establish stages only so far apart that individual 

 variations in the rate of development of the most important 

 characteristics selected as criteria for classification shall not 

 overlap. For purposes of more intensive study, each stage may 

 be divided into tenths; this device is useful in following any single 

 character or set of characters. Since development is a continu- 

 ous process, the importance of studying a close series cannot be 

 too strongly emphasized. 



As an aid to obtaining the exact sequence of events, stress 

 has been laid on the study of a series of stages preserved at short 

 intervals, from a single lot of eggs fertilized at the same time. 

 Every period of the embryonic development has been covered 

 repeatedly in this way, an entire spawning of eggs being some- 

 times used in the study of three or four stages as distinguished 

 in this account. Thus not only a close series, but a large number 

 of embryos in each stage, representing several different spawnings, 

 have been studied, so that the typical course of development 

 could readily be distinguished from variations or abnormalities. 



Moreover, the entire embryonic and larval history has been 

 carefully followed in living material, repeatedly and for the most 

 part with embryos freshly collected. The study of living mate- 

 rial is of especial importance in the late cleavage and gastrula 

 stages of Cryptobranchus ; for in these stages the translucent 

 condition of parts of the unpigmented embryo enables one to 



