The squamous covering of these three regions, the upper 

 parts, the lower parts and the head, aiTord together such a 

 large proportion of the characters used in classification that 

 they require attentive study. 



We have ah-eady seen that each pair of ribs supports and 

 moves a ventral shield ; to each also appertains a corres- 

 ponding transverse row of scales. The ribs not being fixed 

 at aright angle to the vertebral column, but raking more 

 or less backwards, the transverse row of scales corresponding 

 to each pair is inclined backwards in a similar manner. If 

 this incline is at an angle of 45°, the rows of scales will be 

 crossed by lines at an equal angle in the opposite direction ; 

 the scales will be of a rhombic or lozenge shape, and the 

 rows capable of being counted in two cross directions at 

 about equal angles of inclination. But if the ribs be inclined 

 at a slight angle to the spine, then the scales will be more 

 nearly square ; whilst an excessive incline causes them to 

 be rhomboidal or elliptic, and the I'ows to be more or less 

 longitudinally inclined. In the neck of the cobra, for 

 instance, the ribs lie down like the ribs of an umbrella, the 

 scales are consequently arranged in such acutely inclined 

 rows as to become quite linear and imbricate ; when 

 the snake raises the ribs, expanding the skin of the 

 neck into what custom calls the hood, ,the scales are 

 seen dotted like long grains of linseed on the stretched 

 surface. 



The number of scales in each transverse series is variable 

 but very regular. The extreme range is from 12 to 100 or 

 thereabouts, but 13 to 25 is the range in the great majority 

 of snakes. A number above 31 is only found in the ErycidcB, 

 Pythonidce, Acrochordidce some of the Homcdopsidce and 

 Hydrophidce. In conjunction with other characters, the 

 number of scales in each transverse series, or, as it is com- 



4 



