LEPIDOSIKEN PARADOXA. 
107 
(Text-fig. 23 a, a. A.). These afferent branches are at first 
very short and each is continuous with an efferent branch 
that passes back along the gill, dorsal to the afferent vessel. 
Each efferent vessel in turn joins the dorsal aortic root (Text- 
fig. 23 a, D. a. R.), which passes from just in front of the 
mandibular arch, backwards and inwards towards the middle 
line, where ultimately it meets its fellow of the opposite side 
behind the heart in the region of the pronephros, forming the 
unpaired dorsal aorta. In the region of the mandibular and 
hyoid arches the dorsal aortic root receives a slender first and 
second efferent aortic vessel (Text-fig. 23 a, A. A. 1 and 2). 
The present observer has not been able to determine any 
connection between the lateral ventral aorta and these two 
vessels, though possibly an exceedingly fine communication 
may exist for a time. The ventral end of the first aortic 
vessel is prolonged downwards and inwards along the outer 
side of the mandibular arch over the ventral surface of the 
pericardium and terminates in the region of the cement 
organ. With the commencing atrophy of that organ, the 
little vessel breaks up and disappears about Stages 33-34. 
The second incomplete aortic arch, which is even more insig- 
nificant than the first, disappears still earlier. 
The four posterior external gills grow out rapidly into ccn- 
spicuous projecting structures on the sides of the head, and 
give off long feathery filaments through which the branchial 
vessels pass, forming a fine capillary loop in each. Subse- 
quently with the development of the opercula, the external 
gills gradually disappear, and, with the establishment of the 
mouth-cavity and the definition of the gill-clefts, each pair of 
vessels becomes continuous round its respective branchial 
arch by means of a new short-circuiting vascular channel that 
develops between the dorsal and ventral (efferent and 
afferent) vessels (Stages 34-38, Text-fig. 23 b, 5. c.). These 
new channels — where the fourth, fifth and sixth pairs of aortic 
arches are concerned — appear as little widening chinks 
between the respective vessels, but unlike what happens in 
urodeles ( 18 ), cannot be said to develop either from afferent 
