942 DE- W. 6, EIDEWOOD OS THE [NoV. 28, 



and the consequent difficulty of dissection, or to the bad state of 

 preservation of the parts, it was not possible to trace out the 

 vessels so delineated, but that, from analogy with other forms, 

 there is reason to believe that the vessels occupy the positions 

 indicated. There appears to be no reason to suppose that the 

 circulus cephalicus is ever incomplete in front. The lumen of the 

 transverse commissure between the roots of the anterior carotid 

 arteries may possibly be closed in some cases, but the failure of the 

 iiiiection-mass to pass into the commissure is not necessarily a 

 proof of the fact, since the pressure during the process of injection 

 is equal on the two sides of the circulus. The transverse vessel is 

 usually of small size, and its traversing the parasphenoid bone 

 makes it difficult to dissect out with any degree of neatness. 



In the selection of characters by which to classify the various 

 types of arterial disposition, it has been assumed that the condition 

 found in Clupea (PI. LXIII. fig. 2) and Engraulis (fig. 1), where the 

 circulus cephalicus is small, and does not involve the second, third, 

 and fourth efferent branchial vessels, is the most simple and 

 primitive, and that the connection of all four efferent branchial 

 arteries with the circulus, such as occurs in Gadus (PI. LXV. fig. 34), 

 is the most specialized. This assumption is based partly upon the 

 fact that the Gadoids are highly specialized in numerous other 

 respects, whereas the Clupeoids are generally recognized as among 

 the lowest of the Teleostean series ; partly upon the fact that in 

 Amia, an admittedly primitive Granoid with Clupeoid affinities, the 

 last three efferent branchial vessels ai-e unconnected with the 

 circulus cephalicus ^ ; partly also upon the researches of Ayers (4) 

 upon Elasmobranch fishes, which go to prove that the right and 

 left sides of the circulus cephalicus are not the primitive paired 

 aortfe such as occur in Amphioxus and in embryos of the true 

 Vertebrata. but that the true dorsal aorta may persist as a median 

 vestigial vessel traversing the circulus cephalicus, in the same 

 n)anner as, according to Miiller (16), it does in the Cyclostomi. 



Pursuing this line of argument, we may legitimately conclude 

 that where, as in the Salmon (fig. 7), Mackerel (fig. 6), and 

 Carp (fig. 13), the circulus cephalicus, receiving the first and 

 second efferent branchial vessels, is separated from the point 

 of entry of the third and fourth by a length of the median 

 aorta, the condition is more primitive than that in which the tliird 

 and fourth vessels open at the posterior exti-emity of the circulus 

 cephalicus, as in the Bass (PL LXIV. fig. 17). And further, the 

 separation of the third and fourth vessels in the Anchovy (fig. 1) 

 by a portion of the aorta indicates a more lowly condition than 

 that seen in the Herring (fig. 2), where the two vessels open close 

 together. There are thus two lines upon which we may regard 

 specialization as proceeding : firstly, by the circulus cephalicus 



* If such exists. The circulus appears to be suggested in Allis's figure 

 (3. pi. xxxvi.), but its existence is denied by Eamsay Wright (25. p. 495). The 

 arrangement in Lepidosteus is somewhat similar to that oi Amia. See Hyrtl, 8. 

 p. 235, and Miiller, 17. pi. T. fig. (5. 



