Centrography

Centrography is “the trend in thought…directed towards the establishment of laws of the distribution of phenomena based on the relationships and migrations of their centers of gravity” (Poulsen, 1959). Measures of center are geographical techniques for locating the position of a point (or area) that represents the average location of the entire population sampled. This population is applicable to a host of variables including (but not limited to) species populations, manufacturing, natural resources, epidemics, agriculture, education, etc. According to Sviatlovsky (1937), to accurately compute these centers “it is necessary to divide a region into sections small enough to constitute a basis for a relatively precise evaluation of the features involved”. In the case of finding the mean center of area in the U.S., counties (or smaller divisions) would be an appropriate unit of area for use in measures of center calculations.
Mean center calculations are very useful, however they are not all there is to measures of center. Other measures include the median point calculation, quartilides, decilides and centrilides. The median point method involves finding “the point of intersection of two orthogonal lines each of which divides the population into two equal groups” (Sviatlovsky, 1937). One major downside to the median point method is that its results may vary radically with small changes to populations because it does not have what Sviatlovsky calls a “sensitive center”. This lack of a sensitive center causes the median point to shift erratically with certain changes to the population, whereas, the mean center will not. This occurs because the median center is calculated based on the position of orthogonal lines, not a statistical operation like the mean center calculation. When populations move only slight distances (over an orthogonal), they can cause major shifts in the median point location. The opposite is true when a large population migrates but does not pass over an orthogonal. Quartilides, decilides and centrilides are similar to the median point method in that they involve lines that divide the population into equal portions, however, the number of dividing lines is higher. This produces divisions of populations in quarters, tenths and hundredths. According to Sviatlovsky, this method is advantageous in regional analysis because it allows for “a greater degree of refinement…than ordinarily profitable”.
Other methods of centrographic analysis (which were not covered) include;
Differntial Centers
Density Profiles
Ellipses of Inertia
Centers of National Economy