102 BELL SYSTEM TECHNICAL JOURNAL 



service is not very well defined. Cities with a very high residence 

 development have low values for the index and Southern cities with 

 poor residence development have high values for Q, but the inter- 

 mediate scattering of data is quite wide. It might be supposed 

 that cities with high values for Q, which indicate a wide spread of 

 social strata, would have a relatively large number of business firms, 

 either total or retail, to meet the widely divergent needs of the popula- 

 tion. As a matteg- of fact, no such relationship is apparent. There is, 

 however, positive correlation between Q and the proportion of insti- 

 tutions to population. This may be due in part to the fact that high 

 values of Q are found in Southern cities which have separate churches 

 and schools for whites and negroes. 



Although no special significance has been found for the particular 

 degree of rent dispersion found in any city, some interest attaches to 

 the fact that this index remains practically constant in a given city, 

 regardless of changes in the level of prices. The diagrams illustrating 

 this point have already been discussed. If the type of distribution 

 is not found constant in a particular city, it would seem probable 

 that a change in character of the population is taking place, but a 

 change in the average economic grade might occur without any 

 change in the type of distribution. When two distributions, each of 

 which agrees with a logarithmic skew curve, are added together, 

 the new combined distribution may be represented by another loga- 

 rithmic skew curve only in case both the medians and coefficients of 

 dispersion for the two original curves are identical. It follows that 

 if the index of rent dispersion in a city is found to be the same in suc- 

 cessive surveys and if it may be assumed to have remained constant 

 during the interval, then the new families which have come into a city 

 at any time comprise a group having substantially the same coefficient 

 of dispersion and median rent as the families which made up the original 

 population. The apparent permanence of the type of rent distri- 

 bution in a city may be considered, along with the telephone habit, as 

 a reasonable explanation of the rather high degree of stability of 

 station distribution by classes of service among residence subscribers. 



In commercial survey work a city is divided into market areas, 

 known also as homogeneous sections, which are so laid out that in any 

 one section the families at any stated rent are similar telephone 

 prospects. A study of rent distributions in market areas was carried 

 out in a number of cities, considering only those market areas in each 

 city which had fairly large populations. There appears to be no 

 relationship between the index of rent dispersion and either the median 

 rent, the per cent of families in private residences, or the per cent 



