120 
BULLETIN OF THE BUREAU OF FISHERIES. 
the earlier work we made some mistakes because of poor judgment in these matters. 
The correct estimation of the importance of particular scale characters can only come 
with experience, and if through lack of it some unsound opinions are expressed in this 
paper, the fact should not be used as an argument against the study of scales. 
I have adopted, for the purposes of the present paper, the classification given by 
Dr. D. S. Jordan in his recent (1907) work “Fishes,” pages 757-771. It is only necessary 
to compare the classification in recent standard works by Jordan, Boulenger, and Good- 
rich, together with the essays of Regan, to perceive that there is still latitude for much 
difference of opinion in regard to fish taxonomy. Neither the authors cited nor anyone 
else would pretend to be able to present a classification which is nearly perfect^ although 
many important matters are generally regarded as settled. In the slow approximation 
toward a system based on real relationship, lepidologists venture to think that they may 
have a part, and the present essay is written largely with this end in view. 
The cost of the photographs used to illustrate this paper has been defrayed by a 
grant from the American Association for the Advancement of Science. 
Two corrections which I have had to make in this paper since it was written may 
be worth citing, as illustrating certain dangers of error, and at the same time bringing 
out clearly the general reliability of scales for identification. 
Under Boleosoma nigrum I had written that the scales from Osterville, Mass, (from 
the collection at Wood’s Hole), were so unlike those from Indiana that I supposed them 
to be wrongly determined. Dr. Sumner has since very kindly looked up the speci- 
mens, and reports that they are really Boleichthys fusiformis. In the case of Carpiodes 
velifer, I used scales from the fishes forming the basis of Juday’s record for Boulder 
County, Colorado, without making any study of the fishes themselves. I noted with 
surprise that they corresponded “very well with the scales of the common goldfish.” 
Dr. Max Ellis, going over the collection, has discovered that the specimens are in fact 
wild, dark-colored examples of the goldfish, Carassius auratus. 
Class TELEOSTOMI. The true fishes. 
Subclass CROSSOPTERYGll. 
Order ACTINISTIA. 
CtELACANTHID^ (Fossil). 
Dr. A. S. Woodward (Catalogue of Fossil Fishes, pt. 2, pi. xiv) has figured the scales of Ccelacan- 
thus. The figures are excellent, but unfortunately show only the apical (exposed) portion, which is 
ornamented Asdth broad grooves running obliquely toward the middle line. Tlirough the kindness of 
Dr. L. Hussakof, I have obtained the loan of several scales of Caelacanthus robustus Newberry, from the 
Carboniferous of Mazon Creek, HI. These show that the scale is elongate in form, with a large basal 
region wholly free from grooves, and consisting entirelj’^ of very fine longitudinal fibrillae, exactly as in 
Ariiia. The scales are about 12 mm. long and 8 broad. 
There is a striking resemblance between the apical (exposed) area of Ccalacanthus, with its grooved 
lines, and the same area in the South American characinid Lebiasina bimactilata Cuvier & Valenciennes. 
