4 PROCEEDINGS OF THE NATIONAL MUSEUM vol. los 



Fossils presumed to be remains of early vertebrates have been 

 recorded on two occasions from North American deposits that are 

 older than the Harding Sandstone of Colorado and the other verte- 

 brate-bearing beds of Wyoming and South Dakota referred to above, 

 viz.: (a) small, tuberculated plates from the Middle Cambrian St. 

 Albans Shale of Vermont, described by Bryant (1927) and Howell 

 (1937, pp. 1200-1202, p. 2, figs. 7, 8) under the name of Eoichthys 

 howelli Bryant, and (b) Archeognathus primus Cullison, a peculiar 

 V-shaped element with tooth-like cusps along one margin, which comes 

 from the Low^er Ordovician Dutchtown formation of Missouri (Culli- 

 son, 1938, p. 227, pi. 29, fig. 16a,b; Miller, Cullison, and Youngquist, 

 1947, pi. 1). It is by no means certain, however, if any of these fossils 

 really are of vertebrate origin (0rvig, 1951, footnote 1, on p. 381). 

 The Eoichthys plates, at any rate, which I had the opportunity to 

 study in 1953 during my visit to the department of geology of Prince- 

 ton University, Princeton, N. J., cannot possibly have belonged to 

 any true chordate. According to Rhodes and Wingard (1957, p. 

 453), Archeognathus primus is, in several ways, reminiscent of the 

 Neurodontiformes (the "fibrous conodonts" of previous w^riters). 

 Whatever its affinities are, tliis fossil cannot possibly be a detached 

 cornu of an early representative of the Osteostraci (as tentatively 

 suggested by G. M. Robertson, 1954, p. 733). 



Outside North America, pre-Silurian vertebrates are only known 

 with certainty from the lower Ordovician Glauconitic Sand of Es- 

 thonia. From these beds Rohon (1889) has described small tooth-like 

 fossiles belonging to two genera, Palaeodus and Archodus, which, by 

 Russian paleoichthyologists, are nowadays referred to the Thelo- 

 dontida (Obrutchev, 1948, p. 285; Bystrow, 1955, p. 473; Berg, 1955, 

 p. 33; according to Denison, 1956, p. 367, they are indeterminable as 

 to group). As will be shown in another connection (0rvig, in MS., a), 

 however, it is fully evident that Palaeodus represents broken-off 

 tubercles from the exoskeleton of a form belonging to the Astraspida. 

 That the Astraspida are met with not only in various places in the 

 Cordilleran Region of the United States (Colorado, Wyoming, South 

 Dakota, and Montana) but also in Northern Europe is a circumstance 

 of some interest as it indicates that this ancient group of the Hetero- 

 straci may have had a worldwide distribution in Ordovician times. 



Methods and materials: All of the specimens figured in this paper 

 except two (figs. 1-3 of pi. 2 and figs. 1-4 of pi. 3) are contained in 

 beds of Upper Ordovician age from the Rock Creek section on the 

 eastern slope of the Bighorn Mountains. As far as can be judged 

 from the limited material at my disposal, these beds are similar to the 

 other vertebrate-bearing rocks from the Ordovician of North America 

 in that they contain a thanatocoenotic assemblage of detached exo- 



