178 



Kleinere Mitteilungen. 



manner in which in every case the maxima of the number of complexes 

 are distributed over the correlation table. If both the external properties 

 are essentially functions of one and the same factor these two kinds of 

 properties are constantly well matched in pair after pair. If the one in- 

 creases then so does the other also in a special manner, or when the corre- 

 lation is negative the decrease is in a specific manner. The result manifests 

 itself in the correlation table as a pronounced diagonal decline from the 

 maxima. 



This diagonal decline from the maxima is the only criterion which 

 I assume in the following tables. Thereby I have been in the position to 

 isolate some noteworthy correlation-tables from out of my coffee material. 



Let us now consider from the beginning, first table I and table 11. 



table I 



Here in a very remarkable way the maxima lie arranged according 

 two parallels, of which however the most striking point is that both run in 

 opposite directions (the maxima are connected by dotted lines). Each of 

 the diagonals displays a pure correlation. Yet in the very same 

 table the one correlation is positive and the other negative. The corre- 

 lation between the number of blooms per head and the number of heads 

 per axil can in the very same plant be equally well negative or positive, or 

 indeed as in our case both alike! Yet it thereby naturally follows that 

 both diagonals do not belong to one and the same factor — there are here 



