MAMMALS FROM WESTERN CHINA. 167 



SO ruflimentaiy nor so hair-covered as is tlie case in Marina ; 

 their conch wholly below the surface of the fur. Tail shoi't, but 

 not nearly so shoi't as in Blarina. Claws large, the anterior 

 inconspicuously larger than the posterior. 



Skull in genei'al build about as in the more delicate species of 

 tlic sul)genus Cryptotis^ though the rostrum is hardly so high. 



Teeth numbering 32, as in Sorex and true Blarina. Upper 

 unicuspids normally 5, but of such proportions and so slanted that 

 only 3 are as a rule visible in a side view, as in Cryiitoiis. First 

 two large and subequal, third about half their size, its hinder 

 edge level with the front of p'; fourth flattened from before 

 backwards, about half the size of third, and jammed close against 

 it, between it and the minute fifth, which is entirely in the inner 

 half of the tooth-i'ow, close against the large p'. Molars S(iuare, 

 practically without intei'spaces posteriorly. Lower teeth about 

 as in Blarina, the large incisors with thi-ee notches. All teeth 

 heavily pigmented, as in Blarina. 



Type, B. quadraticauda {Sorex quadraticauda M.-Edw.). 



The discovery that the Blarina group is represented in the Old 

 World is a most interesting one and very similar to that of the 

 Sze-chwan Zapus, published by M. Pousargxies in 1896. The 

 present species was described as a Sorex by Milne-Edwards, but 

 when examining the type in Pax-is some years ago I felt sure 

 it was either Blarina or neai'ly allied to it, and Mr. Anderson's 

 series now enables me to place it with certainty. 



Of the four specimens obtained thi'ee have five unicuspids, as 

 above described, but in one (No. 2505) the minute fifth is absent 

 on both sides, so that it seemed possible the type was in the same 

 condition, only four unicuspids having been described and figured. 

 A i'e-examina.tion of it, which I owe to the kindness of Prof. 

 Trouessart, shows that it presents the intermediate condition of 

 having on the right side only 4 unicuspids, the 5th minute one 

 being missing, just as in no. 2505, while on the left this tooth 

 is present. Owing, however, to the accidental loss of the 3rd and 

 4th unicuspids on this left side, the fact that the minute tooth 

 still present, lying pressed against p'', corresponded to the fifth 

 tooth that we now know to be normal to the species, had not 

 been previously observed. There is therefore no doubt whatever 

 as to the specific identity of the pi-esent series with Milne- 

 Edwards's animal. 



The four groups of short-tailed Shrews may be i-eadily dis- 

 tinguished from each other by the number and relative visibility 

 from outside of the upper unicuspids, as follows : — 



5 unicuspids, 4 visible from outside. Blarina. 



5 ,, 3 ,, ,, Blnrinella. 



4 „ 3 „ ,, Cryptotis. 



3 ,, 3 ,, ,, Notiosorex. 



So entirely are both posterior unicuspids generally hidden in 

 Blarinella that the figure of the teeth of B. {Cryptotis) parva 



