No. 3.] AMPHIOXUS AND THE MOSAIC THEORY. 597 



Part III. 

 A. Cleavage and Germ-layers in Annelids and Chordates. 

 A comparison of the cleavage of Amp/iioxus with that of anne- 

 lids shows that they differ widely, not only in form but also in 

 the mode of differentiation of the germ-layers. I was led to 

 examine the cleavage of Aniphioxns primarily in order to 

 determine the origin of the mesoblastic pole-cells described by 

 Hatschek, and thus to find a definite basis for comparison with 

 the annelids, where they are known to exist in a large series of 

 forms, always arising in the same way, having the same relation 

 to the blastopore, and agreeing exactly with those of mollusks 

 (35). I was hardly prepared for the result, although it was of 

 extreme simplicity and beauty ; the pole-cells of Amphioxns are 

 a myth. 



Not unmindful of the proverbial difficulty of proving a nega- 

 tive, I nevertheless speak unreservedly on this point, after a 

 painstaking search, with complete conviction that the same 

 result will be reached by any observer who will take the trouble 

 to make a close study of the actual embryos. At no period 

 during the entire cleavage and gastrulation can the pole-cells 

 be distinguished from the other entoblast cells, either in living 

 embryos or in perfectly fixed and stained preparations (picro- 

 sulphuric, Flemming's fluid, osmic-carmine, sublimate-acetic, 

 etc.). Neither do they exist in later stages. In these the pos- 

 terior region of the larva is rapidly growing and numerous 

 mitoses may be observed in all the cells in the region of the 

 neurenteric canal. The most careful study of this region in 

 sections and total preparations, in various stages, fails to show 

 any cells that can be identified with the pole-cells. Hatschek 

 was, I believe, misled in the early stages by observing entoblast 

 cells at the lip of the blastopore in the rounded form assumed 

 during division. I have occasionally seen a pair of such cells 

 in the position described by Hatschek, but exactly similar pairs 

 of cells may be seen at the same time at other portions of the 

 blastopore-lip and in various parts of the entoblast-plate. In 

 later stages the error seems to have arisen by mistaking the 

 lateral walls of the neurenteric canal as seen in optical section 



