57^ Ascorbic Acid Subniitrition 



To define the degree of this correlation more accurately, a correlation co- 

 efficient was computed from the data given in table i. This was found to be: 



r = — 0.155 =t 0.040 (S.E.). 



As this coefficient is nearly four times its standard error, it is certainly statis- 

 tically significant. Therefore, a negative correlation undoubtedly exists be- 

 tween the ascorbic acid intake and the number of petechiae in Gothlin's test. 

 This correlation is, however, of fairly low degree. If compared with some 

 known correlations (Pearl"*), it is, for example, seen to be no better than the 

 correlation between the brain weight and the body weight and only slightly 

 better than that between intelligence and the horizontal head circumference. 



There are, however, factors which may have unduly lowered this coefficient. 

 First, the regression of the number of petechiae on the ascorbic acid content 

 of the diet need not necessarily be rectilinear; some other function might give 

 a better fit and, conseqtiently, a higher correlation.* In view of the irregular 

 course of the array means (table 1), it seems very doubtful, however, whether 

 any simple function could be found which would give an essentially better 

 fit with the data. In any case, the maximum influence of this factor may be 

 estimated by computing the correlation ratio, a value which provides an 

 upper limit such that no regression function giving a higher correlation can 

 be found. This was found to be: rj = 0.225. This is higher than the correlation 

 coefficient but still a relatively low value. Furthermore, the ascorbic acid values 

 may be somewhat inaccurate because individual differences may have existed 

 in the ascorbic acid intake among the different members of the families, differ- 

 ences which have been disregarded by using the familial averages of the as- 

 corbic acid intake values for the individuals. These two factors may have 

 exerted a somewhat depressing action on the value of the correlation co- 

 efficient. Their influence, however, can hardly have been great enough to alter 

 significantly the fact that the correlation between the ascorbic acid content 

 of the diet and the capillary resistance is low. 



Effect of Treatment with Ascorbic Acid on Capillary Fragility. The signifi- 

 cance of ascorbic acid as a capillary controlling factor should also manifest 

 itself in the capacity to raise an abnormally low capillary resistance, and to 

 prevent a normal resistance from falling. The earlier experiments on this point 

 have yielded discordant i-esults.*'^"'"'"'^®'^""'^ Besides, many of these experi- 

 ments have been inadequately controlled and on a relatively small scale. The 

 problem was further studied on the basis of the data next presented.-j" 



In March, 1941, the capillary resistance of a number of elementary school 

 children was measured by Gothlin's improved method," the double-sided 

 petechial index being determined. From these children two equal groups 



* See Ahlborg and Brante," who found that the relationship between the phisma ascorbic 

 acid and the capillary fragility expressed as the "petechial index" could best be represented 

 by a hyperbola. 



f This part of the work was aided by a grant from the foundation "Suomen Kulttuurira- 

 hasto." 



