290 CALIFORNIA ACADEMY OF SCIENCES [Proc. 4th See. 



southeastern Albemarle many of the female tortoises have these 

 knobs as highly developed as in any male.-^ 



The females are broader than the males. In them also the 

 height to marginals usually is greater. The males attain a much 

 larger size than the females. 



3. Variation with distribution. — We have already stated 

 that the tortoises from any island differ in shape from those 

 from any other. When we consider how close these islands 

 are one to another, it is not strange that these differences should 

 be slight ; nor are they the less interesting and worthy of study, 

 on this account. When we compare Testudo abingdoni with 

 the tortoise of Indefatigable, or that of Narborough with that 

 of Chatham, the differences are, indeed, great. Had we speci- 

 mens from these islands only, we should, without hesitation, 

 regard them as very distinct species. But when we have before 

 us a large number of tortoises from many islands, we find that 

 the matter of their separation becomes most difficult. When 

 we chart our measurements, we see at once that, while certain 

 forms are very dissimilar, others are much less so, and that 

 when the entire group of races is considered the change is so 

 gradual that no sharp lines of distinction can be drawn. It is 

 evident that there are two main groups : the saddle-backed and 

 non-saddle-backed races. But the differences between even 

 these are to a great extent bridged by such forms as the James 

 (T. darwini), the Tagus Cove {T. microphyes), and the Chat- 

 ham Island (T. chathaniensis) tortoises. Nevertheless, the dif- 

 ferences are real, and appear in the table of averages. The 

 extremes of individual variation in races so closely related 

 must overlap and prevent clear diagnosis, unless this variation 

 can in some way be hidden. Now, extremes of variation in any 

 one tortoise rarely affect more than a few measurements. It 

 therefore is possible, by selecting the measurements which best 

 bring out the racial differences, and by combining them in 

 various ways, to bury, as it were, the extremes of individual 

 variation by a process of summation of characters. It is only 

 in this way that we can hope to make a key for the separation 

 of the various races. Even when thus constructed the key 

 must be inadequate for the separation of some specimens. All 



iSince this seems to be true also in the type of Giinther's T. microphyes I was at 

 first inclined to associate the latter name with this race, but it seems better to follow 

 Gunther's use of the term. See remarks p. 253 ante. 



