HEEPETOLOGY OF JAPAN. 395 



forms enuinerated [])y him] might l)e regarded as distinct species but 

 for the absence of any sharp demarcation-hnes between them," this 

 lack of correlation might be explained. Nevertheless, even if we 

 regard them as species, their geographical distribution becomes dis- 

 connected, confused, and unintelligible. 



It seems that a better subdivision might have been obtained by 

 taking the scale formula as a primary basis and color only as a second- 

 ary character. 



It might then be jDossible to separate out a form Naja 7iaja oxiana 

 (Eichwald), from Transcaspia, Afghanistan, and Gilgit, characterized 

 by a large number (averaging 268) of ventrals and caudals together, 

 the sum of these giving more definite results than each separately, 

 owing to the variable location of the vent. 



The s])ecimens from India proper, Ceylon, and possibl^^ Burma 

 and the Malay peninsula seem to be fairly well characterized by tlie 

 large number of scales around the neck, 29 to 35, rarely 27, and 

 should probably stand as Naja naja. 



The cobras from the Malay Archipelago have about the same average 

 number of ventrals plus caudals as the Indian cobras (235 in the 

 former as against 238 in the latter), but the number of scales around 

 the neck is smaller, viz, only 21 to 25. These may be Boie's Naja 

 naja sputatrix, or they may fall into several more subspecies. 



Finally, the Chinese and Siamese specimens seem to hold an inter- 

 mediary position so far as neck scales are concerned, viz, 25 to 27, but 

 with a minimum of ventrals plus caudals (average of 9 specimens 216.) 

 These are also fairly uniform in color and appear entitled to the 

 subspecific appellation Naja. naja atra (Cantor) . 



The demarcation-lines here drawn up are not very sharp, but they 

 seem better than those l)ased primarily on color and have the advan- 

 tage of a more rational geographic distribution than the others. More- 

 over, like Boulenger's scheme, this is only a provisional arrangement 

 which needs to be tested with a very large material. In tliis con- 

 nection I would earnestly warn against using specimens, to the origin 

 of which there attaches the least suspicion. Only b}^ utilizing 

 material which in every detail is authentic and beyond dispute will 

 it be possible to reach a satisfactory solution. 



Description. — Young; British Museum, No. 99.4.24.102; South 

 Cape, Formosa; J. D. La Touche, collector. Rostral very broad, 

 the height only two-thirds of the width; internasals somewhat longer 

 than prefrontals; frontal pentagonal, broader than supraoculars; 

 parietals as long as frontal and prefrontals together; nostril separating 

 the two large nasals; no loreal; one prefrontal in contact ^\■itll 

 supraocular, prefrontal, internasal, posterior nasal, and third suj^ra- 

 labial; diameter of eye as large as its distance from anterior rim of 



