1 66 CONVERGENCE IN MINUTE STRUCTURE 



generation. At the same time it would seem 

 proper to insist here that in the face of an 

 assertive logical system which can only be com- 

 bated at the expense of infinite weariness and 

 vexation, the best and only course is to hold 

 fast to what one adopts intuitively as first 

 principles ; and, with regard to the special case 

 before us, to regard the gill-cleft as an autonomous 

 morphon and the Limuloid gill-book, with its 

 derivative the lung-book of Arachnida, as another 

 autonomous morphon, having nothing in common 

 except their function. 



The histological resemblance between the 

 glandular part of the thyroid of Ammocoetes 

 and the glandular part of the uterus of the 

 scorpion may be regarded as a case of glandular 

 convergence, and as such it possesses a peculiar 

 interest of its own. 



Now if we reject the Limuloid Theory of 

 Vertebrate descent, why should we accept as 

 normal the theory of the Arachnoid affinities 

 of Limulus? The answer in brief is that the 

 normal morphology of Arachnoids and of Verte- 

 brates respects phyletic boundaries; and that 

 the convergent morphology of these groups tran- 

 scends those limits. Both are good so long as 

 they are kept apart. 



It would, of course, be possible to multiply 

 examples of histological parallelism to an almost 

 unlimited extent by comparing distantly related 



