— 76 — 



from the branch to organ 4 after the main branch had entered the main infraorbital canal itself, 

 and then continued its course inside tliat canal. This all certainly indicates tliat organs 4 and 5 lie 

 so close together at the time that they become enclosed in the canal that no primary tube can be 

 developed between them, this tube thus never being formed. In Menidia also these two organs lie 

 close together (Herrick, '99, p. 198), but there is a primary tube between them as there is in Cottus. 

 In Gadus (Cole '98) there is, as in Scorpaena, no primary tube between the 4th. and 5th. organs 

 of the supraorbital canal, but Gadus differs from Scorpaena in that there is no 6th. supraorbital 

 organ and related primary tube. 



Organ 6 of Scorpaena difEers from the other organs of the line in being much smaller than 

 any of them. It is innervated by the first branch of the ophthalmicus lateralis, this branch. perforating 

 the alisphenoid from its lateral surface and then running upward inside the cranial cavity to per- 

 forate the frontal immediately beneath the organ it supplies. 



The preoperculo-mandibular canal begins near the Symphysis of the mandible, and, running 

 posteriorly, traverses the dentary, articular and preopercular, and then anastomoses with the main 

 infraorbital canal at the hind end of the squamosal. The canal lodges eleven sense organs, four lying 

 in the dentary, one in the articular and six in the preopercular, all of them innervated by branches 

 of the ramus mandibularis externus facialis. A primary tube arises from the canal between each 

 two consecutive organs, this making, with the two terminal tubes, twelve tubes in all. The eleventh 

 tube anastomoses secondarily, as already stated, with the seventh tube of the main infraorbital canal, 

 the twelfth tube anastomosing primarily with the tenth infraorbital tube and secondarily with the 

 eighth aml eleventh tubes of the same line. 



6. NERVES. 



The investigation of the nervous system of the several fishes of the group has been mainly 

 limited to the adult of Scorpaena scrofa, and to serial sections of small specimens of that fish and of 

 Lepidotrigla aspera; but certain features of the Innervation in the adults of Cottus octodecimospinosus 

 and Trigla hirundo, in small specimens of Dactylopterus volitans, and in embryos of Cottus scorpius 

 have been also somewhat carefully examined. The Intention at first was to simply determine the 

 relations of the roots, ganglia and proximal portions of the cranial nerves to the skeletal Clements, 

 the study of the skeleton being the principal object of the research; but as certain of the series of 

 sections examined permitted the tracing, with comparative accuracy, of the components of the 

 several nerves, these results, so far as obtained, are given. There was however no attempt whatever 

 to make these results complete. So far as given they are intended to be correct. 



In recent English and American works on the cranial nerves of the lower vertebrates there 

 is a marked tendency to consider the central origin of a given cranial nerve of much more import- 

 ance for the determination of its segmental position than the course of the nerve and its general 

 relations to the skeletal elements. Underlying this manner of considering the subject, is the implied 

 acceptance of the neurone theory of the nervous system, according to which theory all nerve fibers 

 grow eitlier centrifugally or centripetally between two primarily disconnected points, choosing always 

 the path of least resistance. Directly opposed to this nianner of considering the subject is the earlier 

 conception of the nervous system, recently re-presented by Gaskel ('05), according to which the peri- 

 pheral and central cells are from the very beginning, and as soon as they begin their separate exis- 



