The men marry later than the women. Many of the men marrying over forty marry women much younger than themselves. As the husbands are older than the wives, the expectation of life for the husband is, of course, less than for the wives, and the number of widows at any age is far in excess of the number of widowers, on this account alone. Industrial accidents, war, etc., also tend to make a high death rate among the men and cause more widows than widowers. In Fig. 145 the curve for men has been labeled "widowed" to follow the Census Office practice in Fig. 144.
With Fig. 144 some of the age classes are for an interval of only five years while other age classes have an interval of ten years, yet the different lengths of interval are shown by the same distance on the scale. For Fig. 145 the horizontal scale has purposely been made such that the ten-year age intervals are set off by twice the distance used for the five-year intervals. As there are very few marriages under fifteen, the space for "under fifteen" has been made three times the space for the five-year interval. Taking the standard life as four-score years, the space for "over sixty-five" has been made three times the distance used for the five-year intervals. This selection of horizontal distances makes the curves into much more nearly their correct shape than is possible on the Census Office chart, where five-year and ten-year class intervals are shown by equal scale distances.
Perhaps the greatest gain made in clearness in Fig. 145 is due to the placing of the curves for male and female on the same ruled field, so that they can be compared instantly and correctly without need for any right-hand and left-hand measurements such as were necessary in Fig. 144. No claim is made that Fig. 145 is suitable for untrained readers. Since, however, it is doubtful whether many unskilled readers ever refer to the Census Atlas, it would seem desirable to use the general scheme of Fig. 145 for clearness and convenience.
In Fig. 146, also taken from the Census Atlas for the 1900 Census, a right-and-left measurement must be made to compare death rates in two different years, ten years apart. The chart was drawn to bring out the data clearly and, if clearness is not attained, the data might just as well be expressed in columns of figures. Here again the variables have been reversed and the independent variable improperly made the vertical scale.
In Fig. 147 the data of Fig. 146 are redrawn into two curves by which the number of deaths occurring at different ages can be readily
