I9I3] 



G. 0. Higley 



401 



The products XY are now collected, the negative in column 7 

 and the positive in column 8, and the totals determined. We have, 

 then, 2(XF) =2,749.4—1,174.8 = 1,574.6. 



From this the coefficient of correlation (f ) is obtained : 





^xy) 



1,574.6 



Na.a^- 



= + 0.236 



28 X7.65 X 31. 1 



The positive sign of this coefficient indicates, of course, that the 

 relationship between barometric change and carbon dioxide excre- 

 tion in this case is direct or that the two variables change in the 

 same sense. Since a coefficient of correlation of i indicates perfect 

 correlation, the result obtained in the series of experiments repre- 

 sented in Table i indicates a slight degree of correlation. The 

 probable error of a correlation coefficient of this value for a series 

 of 25 observations is at least 0.13 so that the value of r is 

 0.16 ±0.13. 



The results of the whole series of experiments are summed up in 

 Table 4. 



TABLE 4 



General Summary 



Conclusions. There were indications in this work of an in- 

 fluence of barometric change on carbon dioxide excretion in the 

 case of one subject, C, since there were three positive coefficients of 

 correlation having the value of 0.316, 0.39, and 0.248, for morning, 

 noon, and evening experiments (perfect correlation would be in- 

 dicated by a coefficient of i ) ; a slight direct influence is also indi- 

 cated in the case of A, whose coefficients were 0.12, 0.236 and 

 — 0.12. In the case of B, whose values of carbon dioxide excre- 



