Though Fig. 149 might not attract deep interest on the part of a newspaper reader, it would not be likely to be misinterpreted. Fig. 148 might serve to attract attention, but what is the use in attracting attention unless a correct impression is given after attention has been attracted.'*

Frequency curves thus far considered have permitted reading from the vertical scale only the actual number or percentage observed corresponding to any point which may be selected on the horizontal scale. Thus, in Fig. 143 (see page 167) we can read from the vertical scale only the per-

Bquitable Life AssvTance Society

Fig. 148. Change Since 1880 in the Death Rates of Americans at Various Ages

The use of the separate triangles here is confusing to the reader. One is apt to interpret the chart by the contour of the shaded areas rather than by the points at the tips of separate triangles. Compare this illustration with Fig. 149

centage of marriages which corresponds to any selected age on the horizontal scale. In Fig. 150, however, we have the same data of Fig. 143 plotted in the form of a cumulative frequency curve. With a cumulative frequency curve the vertical scale shows not the actual number for any point of the horizontal scale, but the number cumulatively up to any point which may be selected on the horizontal scale. In Fig. 143 the percentage who married at each age is given in figures at the top of the chart. By observing the figures at the top of Fig. 143 and the figures at the top of Fig. 150, the method for plotting a cumulative frequency curve will be apparent. Beginning with the percentages for the later ages in Fig. 143, the figures for the various years are added cumulatively to give the figures seen at the top of Fig. 150.