368 WILLIAM PATTEN. 



tacle itself; moreover, there would be no necessity for regard- 

 ing these terminal cells or teloblasts as any part of a gastrula. 

 Bearing in mind the above conclusions, it is evident that two 

 surface teloblasts or mesoblasts multiplying by tangential and 

 radial division would produce two long mesoblastic bands 

 lying beneath the ectoderm, as in most Annelids. Many 

 surface teloblasts dividing by radial division alone would pro- 

 duce a band of mesoderm lying its whole length on the sur- 

 face. Then, if tangential division began, a longitudinal cord 

 of cells would be either proliferated or invaginated inwards ; if 

 this median cord divided into two lateral ones, which then 

 became segmented, we should have the essential features in the 

 development of the mesoblastic bands of many insects. The 

 important point is that the segmented mesodermic bands of 

 insects have been produced in exactly the same way as those 

 of Annelids, except that the tangential division of the teloblast 

 and of its products is slightly postponed. That the median 

 furrow of insects is merely an ontogenetic adaptation is suffi- 

 ciently evident from the fact that it may be present or absent 

 in closely related forms. When it is absent the resemblance 

 of the mesoblastic bands and their growth, to those of Anne- 

 lids is more evident. 



In Doryphora, Acilius, Musca, and others, the great mass 

 of terminal proliferating cells also gives rise to two cords of 

 endoderm extending forward, one on either side of the 

 median line. In some insects the median furrow produced by 

 the invagination of mesoderm tends to close from before back- 

 wards, leaving a terminal pore, the anterior wall of which con- 

 tinues to proliferate endoderm and mesoderm : in this case the 

 mesodermic bands would appear to grow forwards from the 

 anterior median wall of the pore, while the endodermic bands 

 would appear to arise from its anterior lateral walls. 



Ordinarily the mesodermic bands become segmented while 

 the endoderm is yet only two narrow bands (Fig. 16, c), 

 but later the endodermic bands begin to spread out ventrally 

 and dorsally to enclose the yolk. Suppose the segmentation 

 of the mesoderm is retai'ded and the extension of the endo- 



