OBSEBV^ATIONS ON SOME SOUTH AFRICAN TERMITES. 423 



applying- his method to the antennee of termites some very 

 satisfactory results promise to follow, as is indicated in the 

 provisional antennal charts now submitted for T. natal- 

 ensis, T. badius, and Eutermes trinervius (PI. XXXIII). 

 I do not anticipate equally satisfactory results from its 

 application to the genus Hodotermes, although I do for 

 most genera. 



Mr. Brain has been kind enough to permit me to include 

 here a note by him on his use of antennal charts. 



"The most useful arrangement of antennal data seems to 

 be ari'ived at by giving the range of variation in measurements 

 of the different segments with the addition perhaps of the 

 mode of each. After working over a large series of slides 

 one is impressed with the characteristic appearance of different 

 coccid antennse ; but this diiference is difficult to express. 

 The nearest approach is obtained by charting a polygon 

 showing both the mode and the range of variation in the 

 segment lengths. This supplies a most useful aid for the 

 preliminary location of coccids from slide specimens. 



" Whenever possible, the range of measurement should 

 represent at least ten measurements, twenty to thirty being- 

 made where material is sufficiently plentiful to allow such to 

 be done. The measurements are then plotted in the centre 

 of each antennal column and the polygon completed. Thus 

 it appears that where any segment is constant the charted 

 area approaches a simple line, while a wide band represents 

 a wide range of variation. 



" Similarity of antennal chart, whilst indicating- similarity 

 of antennal formula, does not of necessity indicate identity 

 of coccid species; but it does give a clue to work upon, and 

 possibly at times indicates relationship." 



2. Systematic Account. 



hodotebmes. 

 I have ventured to split into four species a certain group 

 of specimens in the collection which, in part, has been identified 



