SERPENTES. 125 



higher average; viz. V. 225, subc. 92.8. The range is also different: — V. 220- 

 227, subc. 81-100. Should we add to this series Cantor's specimen from Penang 

 (V. 223, subc. 106), the average would be materially raised; but without a series 

 from Penang, this example, differing so widely from all the rest, can not be 

 considered as necessarily included in this race. In a general way, however, this 

 condition of affairs is paralleled by that found in Matrix suhminiata (q. v.), in 

 which species there is a Javan form with fewer scales than in the mainland 

 representative. In N. suhminiata also the single specimen from Hong Kong 

 (Bowring's) shows a condition more like the Javan than the Burmese. Bow- 

 ring's specimen in the British museum has V. 203, C. 80, which is very near the 

 average for Javan specimens, and widely different from the counts of specimens 

 geographically much nearer. 



I wish again to thank Dr. Stejneger for having a verbatim copy of Boie's 

 description taken for me from the Isis in the Library of Congress at Washington. 



This snake is known from southern China, Indo-China, Burma, Malay 

 Peninsula, Sumatra, Java, and Celebes. 



Boiga dendrophila (Boie). 

 BoiE, Tsis, 1827, p. 549. Boulengee, Cat. snakes Brit, mus., 1896, 3, p. 70-71. 



Type locality: — Java. 



Boulenger (loc. cit.) gives in his synonymy of this species Coluber peruvianus 

 Shaw (Gen. zooL, 1802, 3, p. 483, pi. 122). If this species were B. dendrophila, 

 then of course Shaw's name, however inappropriate, would have to stand. I 

 heartily agree with Dr. Stejneger, who has expressed his opinion in a letter, that 

 Shaw's figure and description can not be identified with this species, and proba- 

 bly not with any other. In this case, of course, it drops from literature, so far 

 as this form at any rate is concerned. 



In Java this snake was seen on several occasions, although no specimens were 

 obtained. A specimen from Borneo in this Museum seems to connect varieties 

 B and C of Boulenger; it is over five feet long. Another, smaller, example 

 from the Philippine Islands has 76 cross-bands on the body, almost all complete, 

 and 26 around the tail; this represents var. E, or better, perhaps, B. d. multi- 

 cincta (Blgr.). Bryant had seven from Depok and one from Buitenzorg, all 

 var. A of Boulenger. This seems to be really confined to Java. The scales 

 range V. 218-226, C. 96-108. 



Judging by Boulenger's account, there seem to be some fairly definite 

 groupings into geographic races. There is no necessity to recapitulate these 

 conclusions here, as the material at hand is unimportant. It seems worth 



i^^i 



