BANKING STANDARDS 
loans and discounts? If time deposits are high relative to 
appropriate average levels, are operating expenses high or low? 
In general, what are the types and amounts of correlation between 
related series of banking data, and can these be expressed in 
terms of norms and tendencies? Answers to this general ques- 
tion, as well as to questions relating to particular series, are 
supplied in the following chapters of Part III. Before discussing 
these relations, however, it is necessary to explain the methods 
by which they are determined. 
At various places in the discussion under Part III, charts such 
as Chart 38 are introduced. Each of them, drawn on a ratio? 
basis, has the following characteristics: (1) the vertical posi- 
tions of the respective districts on the charts are without 
significance; (2) for the districts separately and combined, the 
seven-year average ratios, to which each of the lines refer, are 
equated to a common base and indicated by horizontal solid 
lines; (3) in each district the distances of the two types of lines 
above or below the solid horizontal line indicate the percentage 
amounts of dispersion of the amounts for such lines from the 
seven-year average; (4) the slopes of the lines indicate from 
year to year, and over the whole period, the rates of change of 
the ratios to which they apply. 
So much for the graphic form used to illustrate the similarity 
of the differences of the ratios from their respective averages 
year by year, and of the nature and rates of change of the ratios 
for two related series from year to year. 
The tabular form of analysis is somewhat different. As has 
already been indicated, three types of comparisons are used 
in detecting and in measuring the relations between different 
series. The first has to do with the paired deviations of the dis- 
trict ratios in two series from their respective seven-year average 
levels; the second, with paired rates of change from year to 
year; and the third, with paired deviations of yearly district 
averages from yearly averages for the country as a whole. 
Table 96, while applying to the first type of comparison, will 
serve to illustrate the others. Its preparation involves the fol- 
lowing steps: (1) selecting series which seemed to be or might 
be related; (2) calculating, for each district, the percentage 
1 Gee discussion, page 21. 
[54