366 Mr. 0. Thomas on 



compared to many of our large series of P. ielradactylus. 

 On the other hand, I may note, as a further character in 

 P. sultan, that the anterior incisors are longer and more 

 dominant as compared to the teeth behind them than they 

 are in other species. But even this character is not true 

 of P. schioanm'j and is equalled in P. robustus. 



Then, as to P. rovumcB and the forms allied to it, I find 

 among the six skulls I refer to the group a variation in the 

 condition of the palate extending over the whole gamut 

 shown by typical Petrodromus on the one hand and Cerco- 

 ctenus on the other, the original specimens from the Rovuma 

 having the palate as imperfect as in the former, while one, 

 which I believe to represent P. nigriseta, Neum., from 

 Mandera, E. Africa, has it as complete as in the latter. 



With regard to the caudal bristles, some specimens of 

 " Cercoctenus" sultan have them, or, at least, some of them, 

 hardly more knobbed than in P. romance, and, on the other 

 hand, some of those of P. rovumce are scarcely more thick- 

 ened than in the less hairy-tailed forms of true Petrodromus. 



In view, however, of the way the species overlap geogra- 

 phically, and of the apparently superspecific value of the 

 bristle structure, I would suggest that three subgenera should 

 be recognized — the typical Petrodromus (genotype P. tetra- 

 dactylus), characterized by its perfectly normal tail-hairs, a 

 new subgenus Mesoderms (genotype P. rovumce), with thick- 

 ened and partially differentiated bristles, and Cercoctenus 

 (genotype P. sultan), with its bristles knobbed and fully 

 differentiated from the other hairs of the tail *. 



* It is difficult to enter into the mind of an author (A. Roberts, Ann. 

 Transv. Mus. iv. p. 69) who in 1913 stated that the knobs on the caudal 

 bristles of P. schtcanni were "probably" due to singeing in grass tires. 

 Of course, forty years ago, when the first specimen of P. sulta?i came, 

 this idea presented itself and was considered, but was even then disprov- 

 able by the different lengths, inter se, of the bristles, of which fresh ones 

 could be seen pushing up full}- formed, among the bases of the longer 

 ones. But now, when dozens of examples of Petrodromus with knobbed 

 bristles have been recorded, one can only wonder how Mr. Roberts sup- 

 poses that every individual singes its tail in exactly the same part and to 

 the same extent. 



Mr. Roberts has also given new names to Beira and N.W. Rhodesian 

 forms of Petrodromus, of both of which the British Museum possesses 

 topotypes. In the case of the former it seems true that Beira and Goron- 

 goza specimens may be recognized as distinct from those of the Zambezi 

 by their larger size, darker feet, and buffy thighs. But, if this be so, it is 

 certain that the Chirinda series, put in with tetradactylus by Mr. Roberts 

 in defiance of the geography of the case, should not be included in that 

 species, but be either beira or distinct. Personally I believe them to be 

 the latter, and have described them below. 



With regard to occidentcdis, it does not appear that the reputed diffe- 

 rences in size hold good, and the name should apparently be synonymized 



