372 FRANCIS B. SUMNER 



whether or not the respective characters tend to vary together. 

 Of course, the size of all the parts of the body is correlated with 

 the general size of the animal. Thus, larger animals have longer 

 tails, feet, etc. Also, in a series of mixed size, any parts which 

 are correlated with body size are necessarily correlated with one 

 another. But this is not the sort of thing which concerns us 

 here. Of chief interest are the correlations which are found to 

 exist between two characters, when the element of body size has 

 been eliminated. For this purpose, we may either parcel out 

 our animals into groups having approximately the same body 

 length, and determine our coefficients within each of these, or 

 we msiy employ material of varying size and resort to the method 

 of 'multiple correlation.' For certain reasons the former method 

 has been employed exclusively, at least for such characters as 

 vary with the size of the body.^ 



Thus proceeding, I find that there is a small, though probably 

 significant, positive correlation between the length of the tail 

 and that of the foot and skull, and a less certain one between 

 tail and ear, while the length of the foot likewise appears to be 

 positively correlated with that of the pelvis. 



The tail-to-body ratio is negatively correlated with body 

 length; i.e., longer mice have relatively slighter shorter tails. On 

 the other hand, there is probably no correlation between body 

 length and the relative width of the dorsal tail stripe (ratio of 

 arc to entire circumference of tail). 



The number of tail vertebrae is positively correlated with the 

 relative, but not with the absolute, length of the tail. In other 

 words, animals, large or small, which have relatively long tails 

 possess, on the average, a slightly greater number of vertebrae. 

 But larger animals have no more vertebrae than smaller ones, 

 despite the greater absolute length of the tail. In any case, the 

 length of the tail is chiefly determined by the size of the indi- 

 vidual vertebrae, rather than by their number. 



■» Certain other characters, such as the relative length of the tail (regarding 

 it as a percentage of body length), the relative width of the tail stripe (expressed 

 as a percentage of the circumference of the tail), and the number of vertebrae 

 are so slightly correlated with the general size of the body that correlations 

 have been computed directly in populations of mixed size. 



