G. UNDERWOOD 



scales mounted are the frontal, a parietal and a vertical series 

 of scales at about mid-trunk from row one to the vertebral. 



The immunological studies of Cadle (1984) suggest that 

 species of Pseudoboa are sister to Clelia; his sample included 

 the type species of each genus. He also found that Oxyrhopus 

 fitzingeri is closer to Clelia plus Pseudoboa than it is to other 

 species of Oxyrhopus. 'Oxyrhopus' fitzingeri is therefore 

 included as part of the outgroup. In the absence of an 

 analysis, the majority outgroup condition is taken to be 

 primitive for the ingroup. 



The four species of Pseudoboa recognised by Bailey share 

 the special feature of single subcaudals. No evident special 

 feature unites the species of Clelia. On further study the 

 boundary between the two genera may be redrawn, or 

 abolished so that the assignment of the species discussed here 

 may be changed. 



The loreal is rather variable. It may be about as large as the 

 preocular, in which case it meets supralabials two and three. 

 It may be smaller and may meet only supralabial two. It may 

 be absent allowing the prefrontals to meet the supralabials, or 

 the nasal to meet the preocular (as a unilateral variant, C. 

 clelia BMNH 89.4.8.2). A large loreal is assumed to be 

 primitive. 



Boulenger (1896) distinguishes between those Oxyrhopus 

 (s.l.) in which the preocular reaches the upper surface of the 

 head and those in which it does not. the distinction appears to 

 be real but it can be influenced by the condition of the 

 specimen and by the angle of view. The supraocular scale 

 meets the preocular and often, also, the prefrontal scale. The 

 lengths of the supraocular-preocular and the supraocular- 

 prefrontal sutures are compared. The prefrontal suture may 

 range from two or three times the length of the preocular 

 suture, as in Clelia rustica, to absent, as in Oxyrhopus petola 

 in which the preocular meets the frontal. 



There are one preocular and two postoculars. The tempo- 

 rals are 2:3. The upper anterior temporal always meets 

 postoculars; the lower may meet them (represented by 

 +/+2:3) or may not meet them (+/-2:3). In every species 

 represented by more than two specimens this contact is 

 variable. Where the temporal does not make contact the 

 labial scale is higher than its neighbours. However, rustica is 

 nearly constant, on only one side of one of nine specimens 

 does the lower temporal make contact (BMNH 95.9.17.21). 

 The outgroup is also variable so polarity is uncertain. In one 

 rustica there is no suture between prefrontal and preocular 

 (BMNH 86.1.19.21). 



There may be three or two anterior supralabials. As three 

 is the commonest number in the outgroup it is assumed to be 

 primitive. There are two supralabials meeting the eye and 

 three postocular supralabials. Anterior and posterior infrala- 

 bials are distinguished. The last anterior infralabial is pentag- 

 onal; from it starts the posterior row and a mesial row. The 

 anterior infralabials are usually five. Four and six occur as 

 unilateral variants; one rustica (BMNH 1909.11.2.16) has 

 four symmetrically. The posterior infralabials range from 

 four to two. As four is the commonest number in the 

 outgroup this is assumed to be primitive. 



The genials range from posterior about as large as the 

 anterior to somewhat smaller. This does not appear to be a 

 taxonomically useful feature. 



There are always two, sometimes three, scale row fusions 

 from the back of the head onto the neck. The rows on the 

 neck may be 19 or 17. As a majority of the outgroup have 19 

 this is inferred primitive. In those with 19 rows this number 



continues through mid-body and then reduces by fusion of 

 row four or five with the row below to 17 rows. The vertebral 

 or paravertebral rows were not seen to be involved in 

 scale-row changes. Those with 17 neck rows may continue 

 without change to the end of the trunk. In some specimens, 

 however, a lateral row splits on the posterior neck to give rise 

 to 19 rows through mid-body, then reducing to 17 rows. Such 

 a minimum on the neck rising to a maximum around mid- 

 body and falling towards the vent is widespread in henophid- 

 ian grade snakes (Underwood and Stimson, 1990). However, 

 this condition is found in none of the outgroup so it is here 

 assumed to be a derived, pseudoprimitive feature. The scale- 

 row pattern is scored as 19:19:17 (inferred primitive), 

 17:19:17 or 17:17:17. 



In some specimens the vertebral scale-row is undifferenti- 

 ated, but in many it is wider than the paravertebral rows. This 

 modification is a little more pronounced towards the poste- 

 rior trunk. As most of the outgroup have undifferentiated 

 vertebral scales this is assumed to be primitive. 



These snakes usually have both scale pits and tubercles 

 (Underwood, 1963). The pits are confined to the apices of the 

 scales of the trunk and tail, where they usually occur in 

 pairs. Tubercles may be numerous on the head scales. On the 

 trunk they are rather irregularly distributed around the centre 

 of each scale; numbers range from five to zero (Fig. 2). Both 

 pits and tubercles tend to be better developed on the upper 

 scale-rows. Where the vertebral scale-row is enlarged 

 they may be reduced or missing. Pits and tubercles are 

 found in the outgroup and their presence is inferred primi- 

 tive. 



The anterior viscera were examined and their positions 

 recorded in ventral scale units. This has the advantage that 

 juveniles and adults can be compared. Organ positions deter- 

 mined by measurement are subject to allometric change 

 (Thorpe, 1975). The features recorded are: tip of the hyoid, 

 tip of the ventricle, anterior end of the liver and the end of 

 the tracheal cartilages. In all the entry of the trachea into the 

 right lung is terminal. This is a derived condition which 

 Wallach (personal communication) reports to be general in 

 Xenodontine snakes. In all there is some extension of the 

 vascularisation of the lung into the roof of the trachea, but it 

 does not extend far forwards of the heart. The left lung may 

 be present but small, up to about one or two scale-units long 

 and vascular; it may be a non-vascular vestige or it may be 

 absent. Presence is inferred primitive. 



The trachea may terminate no more than three or four 

 scale-units beyond the tip of the heart. It may extend to 

 overlap the liver, in some cases extending the full length of 

 the vascular portion of the lung to reach the terminal air-sac. 

 A short trachea is inferred primitive. 



Counts are made of the teeth, and empty tooth places, of 

 the left maxilla. The solid teeth show a moderate increase in 

 size from front to rear. They are followed by an interval and 

 two obliquely placed teeth with anterior grooves (Fig. 3). 

 Polarity is not inferred. In one juvenile specimen (clelia, 

 BMNH 1933.6.24.102) no grooves could be seen, even when 

 the maxilla was removed and dried. The anterior tooth count 

 is recorded. For sample specimens counts were made of the 

 teeth on the dentary, palatine and pterygoid bones. They all 

 have a full length choanal process of the palatine bone with a 

 broad base, about half the length of the palatine, which 

 sweeps backwards into a process which overlaps the ptery- 

 goid bone by two to three teeth (Fig. 3). The maxillary 

 process of the palatine is turned backwards; it bears a 



