1891.] ^OL [Frazer. 



ber of close correspondences with the oreodonts and especially in those 

 particulars in which that group differs from other artiodactyl families. On 

 the other hand, there are significant deviations from the oreodonts, which 

 are to be found more particularly in the structures correlated with the 

 curious change in foot structure. It seems on the whole highly probable 

 that the two families are not distantly related, especially if the somewhat 

 intermediate character of Protoreodon be considered. 



The conclusion to which the available evidence leads is, then, that 

 Agriochcei'us is the last term in a succession of species which form a 

 curiously specialized offshoot of the OreodontidcB, its divergences from 

 that family being principally the results of a change in the functions and 

 uses of the feet. The separation of the two series was probably already 

 established in the Uinta Eocene, for, in spite of its somewhat intermediate 

 character, Protoreodon can be a forerunner only of the oreodonts. The 

 Bridger beds may be expected to yield the common ancestor of the two 

 series, and this animal will probably turn out to be a pentadactyl form, 

 with buno-selenodont dentition and quinquetuberculate upper molars, the 

 unpaired lobe in the anterior half of the crown. As I have elsewhere 

 suggested, this hypothetical form may have been already found in the im- 

 perfectly known Helohijus. 



The likeness of the AgriocJio&rus molars to those of Eyopotamus has 

 often been noticed and the inference drawn that these two genera were in 

 some manner more or less closely related. Mr. Hatcher writes me that 

 he has lately found feet of Eyopotamus which suggest the same affinities. 

 Until this material has been carefully studied, it will be the part of pru- 

 dence not to prejudge the question. 



Three New Methods for the Detection of Forgery. 

 By Dr. Persifor Frazer. 

 {Read before the American Philosophical Society, May IS, IS94.) 



I wish to put on record three new methods which I have applied suc- 

 cessfully for the purpose of detecting frauds in written documents. 



The first enables one to determine with comparative ease which of two 

 crossing ink lines was made first, and consists in observing the cross- 

 ing by a lens of low power (four or five diameters) at a very oblique 

 angle. If a light ink line be made over a darker one the appearance to 

 the eye when viewing the crossing perpendicularly to the plane of the 

 paper will be that the darker line is superposed. The reason of this is 

 that ink lines are quite transparent and the darker line is seen through 

 the lighter one and seems to make one continuous line with its two limbs 



