SERPENTES. 137 



N. n. atra from the Siamese and Chinese regions. 



To the above may be added : — 



N. n. paucisquamis ; this is known from a single specimen taken in Borneo. 

 Peracca, after correspondence with Boulenger, proposes (Rev. Suisse zool., 1899, 

 7, p. 328) the name for an example with 19 rows of scales on the neck, and 

 15 on the body. He mentions 46 subcaudals, and coloring similar to N. n. 

 mtolepis. 



Since the preceding pages were written, Bryant's Javan cobras have come ■ 

 to hand, and bear out what has been said as to the ventral counts of Javan and 

 Sumatran specimens just meeting but not overlapping in numbers. From the 

 nine specimens which Bryant preserved at Buitenzorg and Depok we get the 

 following counts: — Ventrals 166-180, subcaudals 45-53. These counts are 

 almost the same as those taken from my specimens. Bryant's show a greater 

 variability in numbers of rows on both neck and body. Instead of these being 

 expressed by the formula which I gave before, ig::2i> we must now write igz:23' 

 Only the ventral counts, then, distinguish Javan from Sumatran cobras. 



Doliophis intestinalis (Laurenti). 

 Plate 5, fig. 16; Plate 6, fig. 17. 

 Latjbenti, Syn. Rept., 1768, p. 106. Boulenger, Cat. snakes Brit, mus., 1896, 3, p. 401-404. 



Type locality: — "Africa." 



An excessively variable species. Studies of material from Java and Penang 

 in the collection of the M. C. Z., and data made available by Boulenger, lead me 

 to conclude that this form is not divisible into subspecies. Color varieties which 

 embrace examples from widely separated localities express no relationship, and 

 hence do not merit a name. The averages obtained by adding the number of 

 ventrals and subcaudal scales of specimens from the same locality teach us little. 

 Thus, seven specimens each from Java, Sumatra, and Nias, and from the Straits 

 Settlements, give averages respectively as follows: — 267.7, 261, and 260.4. 

 Twelve specimens from Borneo and Labuan give an average of 258.3. And 

 finally a specimen from Menado in northeastern Celebes has scales which total 

 262, The Javan average is raised by the counts of two specimens which have 

 extremely high scale-counts, — 294 and 287. This is paralleled by a single 

 specimen from Singapore with 288, and one from Borneo with 283. If we had 

 fifty specimens from each of these localities, we might be able to separate satis- 

 factorily a Bornean form; for these do seem to have a smaller average than 

 those from the other islands. With small series, and from scattered parts of the 

 various regions, it seems unwise to try to separate races, especially since color 



^^^ 



