tion in alcohol. The first segment is not enlarged to conceal the head, 

 nor are the anterior segments larger than the others. The general ap- 

 pearance is strikingly different from that of other Diplopoda, the resem- 

 blance being rather to certain lepidopterous larvse. 

 Familly Comodesmid^e, new. 



The type of this family is a small, reddish-brown, subcylindrical 

 form, very rare, and also inhabiting the denser parts of the forest. The 

 pore-formula is unique : 5, 7, 9, 12, 15, 17, 18. The pores are located 

 in the front part of the posterior subsegments. The dorsal surface is 

 beset with conic piliferous granules, giving a wooly appearance. The 

 last segment is scarcely produced beyond the anal valves, but is 

 rounded off at apex as in many Iulidse. The head is not concealed by 

 the first segment, which is narrower than the second and somewhat in- 

 cluded between the carinse of the latter, much as in Scytonotus gran- 

 ulatus (Say). Two other allied genera, also granular and hairy, are 

 found in similar situations in Liberia, but both have the normal pore- 

 formula as in Polydesmus. Thelydesmus is nearly black, larger and 

 much more abundant than Comodesmus. The generic name alludes to 

 the fact that the females are in a large preponderance. Although 

 about a hundred females were taken, careful and extended collecting 

 resulted in only four males. The remaining genus is minute and very 

 rare, cylindrical, and without carinse. The posterior subsegments are 

 abruptly thicker than the anterior, giving the appearance of a series of 

 of discs laid together, whence the generic name, Discodesmus. In the 

 Berlin Museum is another form evidently allied to Thelydesmus, but 

 with broader carinse and more resemblance to the Pterodesmidae, to be 

 noted later on. It was collected in the German Colony of Togo by Dr. 

 K. Buttner, and may be known as Xyodesmus planus. 



In addition to the above genera there may be referred to this family 

 Cylindrodesmus Pocock, from Christmas Island. It is even more evi- 

 dently allied to Comodesmus than the other genera mentioned. There 

 is also in my collection a new generic type from the mountains of Java, 

 not closely related to the other genera, but evidently belonging to the 

 same family group. 



Family Prepodesmid.e, new. 



Under this name it is proposed to arrange West African forms 

 hitherto referred either to Paradesmus or to Oxydesmus, from the 

 latter of which they differ in having the apex of the last segment nar- 

 row and bituberculate. The affinities of the group seem to be with the 

 Oxydesmidse, although no connecting links have yet turned up. In a 



