VARIATION AND DIFFERENTIATION. 21 



in the normal ontogeny, then clearly the second assumption made at 

 the beginning of this paragraph is incorrect, and the suggested explana- 

 tion of our present results would fail. It is impossible to reach any con- 

 clusion regarding the point until biometrical studies on regeneration have 

 been made. 



One further point in this connection is of interest. There is no 

 reasonable doubt that the differentiated, specialized condition of the leg 

 bearing the great chela is phylogenetically a relatively late acquisition. 

 In other words, it is not a primitive morphological condition. But we 

 find that this part which has been modified most recently phylogeneti- 

 cally is also the most variable of the three appendages studied. In so far 

 we have direct statistical confirmation in a single case of the dictum that 

 phylogenetically young structures tend to be more variable than those 

 older and more primitive. In our opinion, however, no particular stress 

 is to be laid on this fact. The real cause of the greater variability of the 

 cheliped appears to us to lie most probably in ontogenetic (in particular, 

 growth) rather than phylogenetic factors. 



So far we have discussed the variation in the appendage as a whole. 

 We may now consider very briefly the different segments of the leg 

 separately with reference to their comparative variability. The data are 

 given in convenient form in tables 5 and 7. It has already been noted 

 that the most variable single joint of all those on which we have data is 

 the propodite of leg I. Leaving this out of account as a specialized organ, 

 and comparing the other segments, it is seen from table 5 that in both 

 legs II and in the carpopodite is relatively the most variable. Also in 

 the case of leg I the carpopodite is more variable than the meripodite. A 

 relatively high degree of variability of the carpopodite, as compared with 

 meripodite and propodite, is shown in Yerkes' (1901) figures for Oela- 

 simus pugilator. Our results thus stand in agreement with his on this 

 point. In legs n and in the meripodite and propodite are substantially 

 equally variable. The differences in the coefficients are almost certainly 

 insignificant in comparison with their probable errors. These are the 

 results when the variation is measured from the absolute dimensions. If 

 we consider the variation in the proportions of the different segments as 

 given in table 7 the results are not quite so regular. Leaving out the 

 great chela as before, we see that in legs n and in the meripodite is more 

 variable than the carpopodite. The difference is probably not significant 

 in the case of leg I, but it certainly is in leg n. In leg in the carpopo- 

 dite is again the most variable joint. It might be thought that the 



