May 12, 1911] 



SCIENCE 



14:1 



through its hibernating mycelium in the 

 potato tubers, just as the potato itself has 

 largely lost its power to reproduce sexually 

 through the formation of seeds. This loss of 

 sexual power is shown in different degrees by 

 the different strains of the fungus in artificial 

 cultures. The fungus seems to lose first its 

 power of producing antheridia and then of 

 producing oogonia. Under favorable condi- 

 tions attempts to form oogonia first appear, 

 and under still more favorable conditions the 

 antheridia are produced, and with the forma- 

 tion of these the oospores also appear in more 

 or less perfect form. 



A further discussion of this subject, with 

 photomicrographs of the sexual stages as we 

 have gradually developed them, will appear 

 in the next report of the Connecticut Agri- 

 cultural Experiment Station. 



G. P. Clinton 



New Haven, Conn., 

 December 20, 1910 



A POSSIBLE LINE OF DESCENT OF THE GOBIOID 

 FISHES 



Indicating the doubt existing as to the rela- 

 tionship of the gobies are the several different 

 positions assigned to them in the schemes of 

 classification suggested from time to time by 

 different authors. Without attempting any- 

 thing like an exhaustive survey of the disposi- 

 tion of the group by different authorities its 

 treatment by a few of them may form an 

 introduction to the suggestions of relationship 

 in the following lines. 



Dr. Gill, in his " Arrangement of the Fam- 

 ilies of Fishes," ' places the superf amilies 

 Gobioidea and Cottoidea in adjoining groups. 

 But in his later arrangement" he has several 

 families interposed between the Gobiidse and 

 Cottidae, as the Batrachidse, the Uranoscopidas, 

 the Trachinidse, the Malacanthidse and others. 



Dr. Jordan, in his " Guide to the Study of 

 Pishes," " has placed the gobies near the cot- 

 toid fishes with the following remark : " The 

 great family of Gobiidffi, having no near rela- 



»" Smith. Miss. Col.," 1872. 



'Mem. Nat. Acad. Sci., Vol. VI., pp. 127-138. 



'Henry Holt and Co., 1905. 



tions among the spiny-rayed fishes, may be 

 here treated as forming a distinct suborder." 



Dr. Boulenger, in the Cambridge Natural 

 History,* places the Gobiidse between the Kur- 

 tidse and Echeneididas, and expresses the opin- 

 ion that the gobies " are not very remote from 

 the Perciformes, and may have evolved out of 

 a type not very different from the Percidffi." 



Mr. Began, in his classification of the teleos- 

 tean fishes,^ has placed the suborder Gobioidea 

 between the Blennoidea and the Kurtoidei. 



Recently while examining the skeleton of 

 Dormitator maculatus, a large goby from the 

 warm waters of the American Pacific and 

 Atlantic, I was impressed with the similarity 

 of its shoulder girdle with that of the family 

 Cottidae and certain other cottoid or mail- 

 cheeked fishes. In light of the fact that there 

 is otherwise very little in the anatomy of the 

 gobies that might show their line of descent, I 

 wondered that the line from some ancestor of 

 the Cottidas had not been long ago suggested, 

 more especially as there seems to be little 

 reason why such relationship should not exist. 



The similarity of the shoulder girdles of 

 these families has long been known. As early 

 as 1865 Dr. Gagenbaur published a picture of 

 the shoulder girdles of a gobioid and a cottoid 

 fish side by side in the second part of his 

 " Untersuchungen zur Vergleichenden Anat- 

 omie der Wirbelthiere." ' 



The condition of the shoulder girdle in the 

 CottidaB and Gobiidas is as follows : The cora- 

 coid elements and the actinosts are arranged in 

 a continuous row on the posterior edge of the 

 clavicle; the hypercoracoid above, next the 

 actinosts, and ending below with the hypocora- 

 coid — the actinosts attached directly with 

 the clavicle, and separating the coracoid ele- 

 ments widely from each other. In the typical 

 condition — the condition in the great majority 

 of fishes — the coracoid elements are broadly 

 attached to each other, and the actinosts are 

 attached to their posterior edges remote from 

 the clavicle. 



* Macmillan and Co., 1904. 

 'Ann. Mag. Nat. Hist., Ser. 8, Vol. III., 1909. 

 ' E emitripterus acadianus and Gohius guttatus, 

 Taf. VII., figs. 8 and 9. 



