GEOGRAPHIC DISTRIBUTION OF MULTIPLE SCLEROSIS 127 



prevalence is plotted against geographic and geomagnetic latitude in Fig. 2, 

 and the points suggest a sigmoid curve. A normal sigmoid curve (i.e., the 

 integral curve for a normal distribution) has been included in each plot, and 

 it is evident that it is not possible to conclude from these data drawn from a 

 limited range of longitude whether geomagnetic latitude is a better param- 

 eter than geographic latitude. The general trend shown by the points 

 between approximately 40° and 50° has been confirmed from several other 

 types of data for the United States (Barlow, 1960; Acheson, 1959) , although 

 the presence of a "knee" at about 50° geomagnetic latitude is not clear from 

 these latter data. Evidence that a rapid rise in the frequency of the disease 

 between geomagnetic latitudes of approximately 40° and 50° appears for 

 other areas of the world in both the northern and southern hemispheres has 

 previously been presented (Barlow, 1960). 



For further examination of the latitude distribution, prevalence data from 

 recently conducted surveys presented at the Geomedical Conference in 

 Copenhagen in June, 1959 (Hyllested, 1960) are listed in Table II. It 

 also includes the data from Table I and results of other prevalence surveys. 

 As in Table I, the mean prevalence ratio for locations of 50° or greater geo- 

 magnetic latitude was determined, and prevalences relative to this standard 

 (48 cases per 100,000) are shown. These data are plotted against geographic 

 and geomagnetic latitude in Fig. 3. Since a wide range of longitudes is repre- 

 sented in these plots, the sigmoid curve for the geomagnetic plot of Fig. 2 is 

 reproduced in both the geographic and the geomagnetic plot of Fig. 3, and 

 it will be reproduced in subsequent plots for comparative purposes. 



It is apparent from inspection of the two plots that at any given latitude the 

 scatter of the p>oints is greater for the geographic plot than for the geo- 

 magnetic plot, at least for latitudes of less than 50°. The scatter of the points 

 above 50° is such that a "knee" or leveling off is not as clearly suggested as 

 in Fig. 2. There is some indication, however, that the rapid increase of 

 prevalence between 40° and 50° geomagnetic latitude does not continue 

 upward in the same manner beyond 50°. 



Several independent surveys are represented in Fig. 3; therefore it is not 

 possible to state how much of the scatter of points above 50° geomagnetic 

 latitude is due to differences in survey procedures and how much is due to 

 real differences in prevalence among the population groups. It is probably 

 unlikely that the scatter is entirely due to differences in survey procedures. 

 Even if uniform survey procedures were used, such a scatter of points could 

 conceivably occur if the prevalence ratios among different population groups 

 formed a normal distribution about a mean value, and the scatter might 

 further be accentuated by differences in the size of the population groups. 

 Particularly if the population is very small, chance variations in the observed 

 prevalence ratios may be pronounced for occasional communities (Deacon 

 et al, 1959). 



