Ernest Warren 
337 
Thus it will be seen that the workers are more than double the size of the 
soldiers of the similarly sized class. 
The vai-iability of the soldiers, as measured by the coefficient of variation, is 
greater than that of the workers, the ratios being : 
Coefficient of Large Workers 2'461 „ 
Coefficient of Large boidiers 3 123 
Coefficient of Small Workers 2-647 ^_ ^ 
— — ,. , — X 100 = ^ X 100 = 8r6, 
Coefficient of Small Soldiers 3'022 
that is the workers are about 0'8 times as variable as the soldiers. 
(6) General Variahility of the sexual and asexual Castes. 
The average coefficient of variation for the four castes is 2'8 : it is thus seen 
that the general variability of these termite-castes is distinctly small. The nymphs 
of the winged sexual forms appear to exhibit extraordinarily little variation. In 
the case of 100 male and 100 female nymphs (Table VIIL nest " 653 ") the co- 
efficients were only 1'.57 and 1'60 respectively. Unfortunately there is no available 
material for ascertaining the variability of the adult sexual form before leaving 
the nest in the species natalensis ; but, as Avill be seen later, small series of winged 
imagos have been measured in some of the other species, and there is a general 
tendency towards a low variability. 
It has been remarked that there appears to be good evidence for supposing 
that the young which hatch from the eggs are all alike, and that by special 
feeding, or by some manipulation by the workers, any one of the five castes 
may be produced. From this it would be expected that the variability exhibited 
by all the castes would be about the same ; but it has been shown above that the 
workers are less variable than the soldiers, and that the asexual castes generally 
are much less variable than the sexual castes. In this connexion, it should be 
remembered that the individuals in a nest are all produced from one queen and 
king. The difference in the variability of the castes must therefore be referred to 
differences in food or other environmental conditions, since they all have a common 
parentage, and are presumably all alike on hatching. 
Thus, the tendency to vary is induced or modified by the special food or 
manipulation received, and the influences necessary for the production of a soldier 
or worker lead to greater variability than those for the formation of a sexual 
imago. It appears that either sex may be converted into either soldiers or 
workers, the potential sex having no effect on the ultimate destiny of the in- 
dividual. It is clear that the environment has an overwhelming effect on these 
organisms, it influences the variability and decides on the bodily shape and 
structure of the developing insect. 
(7) (Joviparison of the Constants of a Family with those of the Population. 
Of small soldiers, random samples from 30 nests were measured. The sizes 
of the samples were most disproportionate, ranging from 100 to 1000. On adding 
Biometrika vi 43 
