DEVONIAN FISHES OF IOWA 171 



comes intelligible only through comparison with surviving Dip- 

 noans, and without such aid must remain an enigma sui gen- 

 eris. The pertinency of this statement will appear from the fol- 

 lowing description of the headshield. 



Cranial characters. — The arrangement of dermal roofing 

 plates in the headshield of Macropetaliohthys is shown in the 

 accompanying restoration (text-fig. 24), which may be profitably 

 compared with the diagram given on page 197 (text-fig. 29) of 

 the cranial roof of Neoceratodus. One perceives that there is a 

 general correspondence between the two genera as regards cra- 

 nial pattern, and especially noteworthy is the similar disposition 

 of the median series of plates, the more posterior of which is 

 elongated nearly to the same extent as in Homosteus. Other 

 points of agreement between the form under discussion and 

 Homosteus consist in the elongation of the external occipitals, 

 and enclosure of the orbits within the headshield. In more spe- 

 cialized forms, the preorbital and postorbital plates are merely 

 notched externally, but in Macropetaliohthys, Homosteus, and 

 presumably also in Asterosteus, these two plates are in contact 

 with each other externally so as to form the inferior border of 

 the orbits. 



A conspicuous difference between Macropetalichthys and other 

 Arthrodires, one which has proved a stumbling-block to a correct 

 understanding of the cranial osteology, lies in the fact that the 

 central elements are divided so as to form two small plates on 

 either side of the headshield back of the orbits. These more or 

 less rounded plates are placed one behind the other, the two 

 pairs being separated from contact with each other in the median 

 line by the elongated median occipital plate (MO, text-fig. 24), 

 very much as is the single pair of corresponding plates in 

 Neoceratodus (text-fig. 29). That the plates here called the cen- 

 trals are correctly determined as such is evident from the follow- 

 ing reasons: First, the two pairs together occupy the usual 

 position of the centrals with reference to the preorbital plate in 

 front, and to the postorbital and marginal plates externally ; and 

 secondly, they are proved to be such by the disposition of the sen- 

 sory canals. Imagining the suture line as obliterated between 

 the two independently ossified plates on either side, we shall have 



