2G 



mouly called, the number of rows (longitudinal)* of scales 

 is a valuable distinctive character, as it is comparatively 

 rare that individuals of the same species should have a 

 different number of rows. 



In Part II it will be seen that the number of rows of 

 scales is generally the first element in the diagnosis of species. 

 When the physiognomy of a snake does not indicate its 

 family the collector begins at once to count the number of 

 scales in a transverse series along the course of a pair of 

 ribs ; but in the tree-snakes and some others it will be 

 found more convenient to count across the ribs. 



This number is nearly always odd, the vertebral row 

 being azygos, and often of a different shape ; in onlj'^ two 

 genera (Zaocys and Peltopelor) is there a double row of 

 vertebral scales, and, consequently, an even number in the 

 transverse series. An even number of scales consequent on 

 the doubling of the vertebral may be present as an anomaly 

 in individual snakes. I have a Bipsas gohool with 22 rows 

 of scales instead of 21. The number of rows should be 

 counted at a distance from the head equal to about one 

 quarter of the length of the body, as the number on the neck 

 exceeds the normal number by two or more ; the number 

 settles down at a point varying from the tenth to the 

 thirtieth ventral shield and remains constant for at least 

 halfTway down the body ; then, sooner or later, the scales 

 begin to diminish, always in uneven number, down to the 

 root of the tail. There several rows may be observed with- 

 out corresponding ventrals, three or four being rudimentary 

 or deficient where the anal orifice is covered by its large 

 shield. The number of scales on the tail is nearly always 

 even, beginning with about ten and diminishing by pairs to 

 four or two. 



* The number of transverse series is not counted ; the number of 

 ventral shields is, practically, more constant and is stated instead, at 

 least in the snakes with ventral shields. 



