34 



To see how differences in cranial measures (i.e. "y") depend on skull size (i.e. "x"), I 

 decided to calculate an artificial volume quantity called SIZE. In order to exclude 

 influence of rostral length on the chosen reference quantity, SIZE (=x) is determmed by 

 neurocranial measures only (cf. fig. 3). 



Fig. 3: SIZE = (Condylocanine Length minus Upper Tooth Row) ■ Height of Braincase * Width of 

 Braincase 



Calculations 



Univariate Analysis 



For each skull measures in each species, the mean value and standard deviation was 

 calculated. If possible, males and females were compared to each other as seperate random 

 samples (F-test, t-test). 



Allometrics 



In individuals of different size, allometric calculations allow to distinguish proportional 

 shifts exclusively based on size increase from differences in proportion caused by other 

 reasons. Often, such differences in proportion reveal deviating construction principles 

 wliich can also be evaluated m a taxonomical respect. 



Mathematically, allometrics are recorded by means of the allometric formula: 

 y = b * x' 



log y = log b + a * log x 



Tills causes comparision of measures representing different dimensions: units of length 

 against volimie measures. Thus, for allometric analysis instead of the usual regression 

 lines the reduced elhptic major axis of distribution (Rempe 1962) was referred to. 



All necessary calculations were performed on a IBM compatible personal computer by 

 means of a regression calculating software written in BASIC (D. Vinyard, after 

 Müller/Kick 1983, see appendix 9.4) as well as DIVA (Ver.2.0), a statistics software by 

 D. Plogmaim. All calculations were based on the works of Rempe (1962). 



