44 BULLETIN 1300, U. S. DEPARTMENT OE AGRICULTURE 
before has an effect explainable on the same basis as with summer 
slaughter. 
The corn crop has relatively little influence on winter slaughter. 
In the whiter following an exceptional crop the correlation is nega- 
tive so far as it goes ( — 0.08). In the second winter it is only 
slightly positive I —0.21) in spite of the heavy fall breeding determined 
by a big crop or the light fall breeding by a small crop. As we have just 
seen, heavy breeding due directly to a big crop tends to swell the 
following summer slaughter at the expense of the second as well as 
following winter slaughter. It is not until the third winter after an 
exceptional crop that there is an appreciable correlation with winter- 
slaughter (+0.42). A large crop stimulates heavy breeding for at 
least a year and the second large pig crop which results is raised in a 
season of scarce rather than abundant corn, and hence fattening and 
slaughter must be delayed into the winter season. There is IS per 
cent determination by the third preceding corn crop and only 27 per 
cent by all three preceding corn crops combined. 
In contrast is the 61 per cent determination by the second preced- 
ing summer weight, due to the correlation of —0.78 referred to above. 
The combination of this with the second and third preceding corn 
crops increases the coefficient of determination to only 64 per cent. 
The highest coefficient which has been found is from trie combination 
of the second preceding summer weight with the two preceding 
winter weights (67 per cent). 
The negative correlation (—0.36) with the winter slaughter two 
years before or later, has already been noted as a manifestation of 
the cycle in hog production discussed under summer weight. 
Summing up. winter slaughter is determined largely by the amount 
of breeding in the second preceding fall. Breeding before this time 
has an appreciable effect. The immediate effect of the corn crop is 
slightly negative, since an abundance of corn tends to concentrate 
slaughter in the summer at the expense of the winter season. 
THE SYSTEM OF HOG AND CORN VARIABLES AS A WHOLE 
From the foregoing discussion it will be seen that most of the 
correlations among the hog and corn variables can readily be 
accounted for even though they may not agree in all cases with what 
might seem most probable beforehand. In many cases, indeed, 
there seems to be a superfluity of reasonable explanations, Only too 
often, however, it turns out that what seems to be the obvious expla- 
nation in the case of one correlation is wholly incompatible with what 
seems the equally obvious explanation in another case. 
In the complex network oi interrelations, satisfactory analysis of 
the system requires that the consequences of each hypothetical 
relation of cause and effect be worked out with respect to every one 
of the 510 observed correlations. Such an analysis must be as 
rigidly quantitative as the data permit. 
This may appear to be a hopeless task, in such a system as the 
present, in which the number of possible actions and reactions among 
the variables would seem almost infinite. Nevertheless, by testing 
the consequences of different reasonable hypotheses as to causal 
relations by the method of path coefficients 6 and choosing that which 
'■ Wright, Sewall, 1921. Correlation and Causation. In Jour. Agr. Research, v. 20, No. 7, pp. 557-585. 
