696 



TEANSACTIONS OF SECTION T. 



3. By means of ' partial correlation ' the influence of the hypothetical general 

 factor may be eliminated, and the partial or specific intercorrelations calculated. 



The arrangement of these coefficients suggests a scheme ■which_ is not hier- 

 archical, and has not hitherto been described. The positive coefficients form a 

 diagonal line running from the top left-hand corner to the bottom right-hand 

 corner (see Table II.), with outlying groups at the top right and bottom left 

 corners : the negative coefficients run in lines parallel to the diagonal from the 

 centre of the top row to the centre of the last column and from the centre 

 of the first column to the centre of the bottom row. 



This arrangement suggests that the specific tendencies may form a circular 

 series, each member being most closely related to its immediate neighbour, less 

 closely related to remoter members, and antagonistically related to members 

 half-way round the cycle. 



To illustrate the Chief Specific Correlations between the Chief Emotional Tendencies. 



(Distances of arc represent inversely the chief positive correlatioas ; chorda represent 

 the average direction of the chief negative correlations.) 



Sex 



Sociality 



Elation 



Joy 



Wonder 



Anger Wonder 

 '+■34- 



Tenderness 



Anger D/sffusC 

 = +■23 



Disgust 



Subjection 



Fear 



Sorrow 



This phenomenon — the cyclic overlap of specific factors — has also been 

 observed in partial coefficients obtained by eliminating 'general intelligence ' from 

 the intercorrelations of complex mental tests. It may, therefore, have a wide 

 significance for the study of mental organisation. 



The theory of a general factor is commonly associated with the views that 

 specific factors are negligible and that the general factor is simple. The 

 problem arises as to whether the above specific correlations invalidate the 

 hypothesis of a general factor, and, in particular, whether the peculiar relations 

 between the specific factors may not of themselves produce the appearance of a 

 general factor. Independent evidence, however, appears to confirm the exist- 

 ence of both general and specific factors underlying emotional reactions and to 

 indicate that both are highly complex. 



