SERIES CORRELATED WITH GROSS EARNINGS 221
correlation is higher for ratios of total expense than for those of
net earnings; and (3) the type of correlation characterizing the
changes in all districts and all years obtains for the Boston district
 for the years 1919 to 1925, and for all districts for the
changes between 1924 and 1925. That is, in general, the relations
obtaining for all districts obtain for the Boston district, and those
which hold for all yearly changes characterize those between 1924
and 1925.
Space is not available in which to insert, for all series correlated
with gross earnings, tables similar to Table 134. The net figures
given in Table 133 reflect the prevailing relations.
Moreover, as is apparent from Charts 46 and 47, relating, respectively,
 to (1) ratios of gross earnings and of total expense
and (2) ratios of gross earnings and of net earnings, all ratios
being expressed in terms of earning assets, the positive correlation
 of both series with gross earnings, found for the twelve districts
 combined and for the Boston district alone, holds also for
the other districts. These charts are drawn according to the
method described on page 154, and should be interpreted in keeping
 with what is there said. They illustrate, more graphically
than do words, the close relationship obtaining between the respective
 pairs of ratios.
There is another basis according to which various banking
ratios may be correlated with those of gross earnings. If the
average yearly amounts for the combined districts are taken as a
base and if those for the respective districts are expressed as
percentage deviations, plus and minus, the distribution shown in
the stub (horizontal) classes of Table 135 is secured. The net
percentage deviations in different series, for the districts having
the classified types and amounts of dispersion with respect to
gross earnings, are given in the caption classes of the same table.
This arrangement indicates the manner in which different series
are related to variable gross earnings.
It is obvious that if the ratios of gross earnings in a given district
 are persistently above or below the country level, which is
frequently the case,? the position of that district in the dispersion
classification, provided for in the stub of Table 135, will tend to
remain fixed. There is less occasion for such a condition to obtain
when the dispersion-classes apply to deviations from district levels
?See Table 53, page 77, and the discussion in Chapter V.