Ernest Warren 
345 
Relative Sizes of the various Castes, — Ratios. 
Soldiers of two sizes occur in Termes natalensis, Termes badius, Termes 
trinervius and Rhinotermes sp., and the ratios are given in the 2nd column of 
Table IX. There is considerable variation in the ratios, and the mean is 62'5. 
Two sizes of workers occur in live species. It should be noticed that it does 
not follow that there are two sizes of soldiers when there are two castes of workers 
and vice versa. The ratios are given in the 3rd column of Table IX. It will be 
seen that they are fairly constant, the mean (66"9) is somewhat similar to that for 
the soldiers (6 2 '5). 
The ratios of large workers to large soldiers are given in the 4th column of 
the table and they are exceedingly variable, ranging from 5 7 "6 in Termes natalensis 
to 129'5 in Termes latericius. 
Various other ratios are given in the remaining columns of the table, and 
there is no marked tendency for constancy in any of them ; although the ratio 
of workers to winged imagos is less variable than the others. 
Means of the Coefficients of Variation. 
The arithmetic mean of the coefficients for large soldiers for the 11 species 
(see Table VIII.) is 3'26 ; for small soldiers (3 species) 3'68 ; for large workers 
(11 species) 374; for small workers (5 species) 3"89 ; for male imagos (6 species) 
2'06 ; and for female imagos 1'98. 
It is evident from these results that the male and female imagos are con- 
siderably less variable than any of the asexual castes. The special manipulation, 
whether of the nature of specialised food or of any other influence, necessary for 
the production of these castes, appears to increase the variability. 
With regard to the variability of the asexual castes, it will be noticed that the 
mean results for the series of species are not in accord with those for the species 
Termes natalensis. In this species the soldiers were more variable than the workers, 
but in the mean values for all the species the reverse is the case. 
(12) Comparison of the termite measurements tvith those of Wasps. 
Miss Alexandra Wright, Dr Alice Lee and Professor Karl Pearson have recently 
completed an investigation on the variability of various parts of the wings of 
worker, drone and queen wasps (Biometrika, Vol. v. Part iv. p. 407, 1907). The 
material employed comprised the individuals from one large nest, and it will be of 
interest to compare as far as possible the results obtained with those from the 
termites. 
In the absolute dimensions the queens, drones and workers of the wasps are in 
descending order of magnitude with respect to the wing-measurements. 
In the case of the termites, the means of the head-breadths of the five species 
(T. vulgaris, T. incertus, T. parvus, T. bilobatus and T. trinervius) were for male 
imagos, 163-15 units; workers, 117-3 and soldiers, 121-3. Thus, although the 
Biometrika vi 44 
