these poorly sampled areas. Table | illustrates quite vividly the poor data 
distribution and density throughout the world and, therefore, the difficulty 
a cartographer faces in accurately depicting the geomagnetic field. 
In general, the difference between charted and observed data is a 
fair guide to the reliability of the world chart. Accordingly, the mean and 
median for each World Aeronautical Chart was computed fo investigate these 
differences. In each case both magnitude and direction are indicated. The 
sign associated with each value is positive whenever the charted data exceeds 
the observed data. While both values are indices of central tendency, a rela- 
tive advantage of median over mean does exist, i.e., the mean is greatly 
influenced by extreme values, while the median is not. For this reason and 
because the large volume of data used was not edited, the median appears to 
be the more reliable value. The importance of both mean and median in this 
analysis is that they provide a measure of how closely the charted isogonic 
lines and observed data agree. 
In considering the median, it seemed desirable to establish a level of 
confidence. Therefore, an 80% confidence interval based on the computed 
median was derived. This interval represents the range within which the true 
median would lie 80 out of 100 times. A broad interval then, would indicate 
that a "good" estimate is not possible. In this analysis, the median and 80% 
confidence interval are not valid whenever 50% or more of the data lie in the 
"open-ended" intervals of the histogram. This consistency occurs in charts 
1773, 1774, 1815, 1818, 1821, and appears to result from lack of data and/or 
being located in high latitudes. 
The agreement of charted and observed data in the northern hemisphere 
is somewhat better than that of the southern hemisphere. Figure 2 illustrates 
this, depicting those areas where the median exceeds 0.5°. For the northern 
hemisphere, the median in 65 or about 7% of the 904 charts containing data 
exceeds 0.5°. In the southern hemisphere, the median exceeds 0.5° in 221 or 
about 27% of the 824 charts containing data. When considered overall, only 
286 or about 17% of the 1,728 charts containing data have medians exceeding 
0.5°. In light of the sparseness of observed data and the limited detail the 
world chart is designed to show, the mathematical model of the Coast and 
Geodetic Survey would then appear to fit the observed data remarkably well . 
The root mean square, RMS, for each chart was computed to indicate 
the dispersion or scattering of the residuals about their mean. With a large 
sample and a distribution curve which approaches normality, 68% of the resi- 
duals will fall within an interval given by X + RMS, and 95% within 
