If the relative risks of mortality in Figure 2.4 are standardized on the French crude death rate of c. 1785, one obtains the time series of crude death rates (per thousand) shown in Table A1.

Table IV in Waaler (1984) contains relative mortality rates for Norwegian males aged 50 to 64, averaged over height intervals of 10 cm and weight intervals of 10 kg. Attempts to fit iso-risk curves to these averages produced unsatisfactory results. Because at that time we did not have access to the data from which these averages were derived, it seemed reasonable to “fill in” the table by polynomial interpolation and use the generated data to estimate the risk–height–weight relationship.

The interpolation was done in two steps. The first step consisted of taking the relative mortality rates given in the table and using standard least-squares regressions to fit polynomials in weight to each column of risk values (corresponding to given levels of height) and polynomials in height to each row (corresponding to given levels of weight). These polynomials, which were each of the maximum order permitted by the number of entries in the corresponding row or column, were then used to generate values across each row and down each column of the table in intervals of whole centimeters and kilograms. This step generated 700 data points.

To further increase the available data, the rows and columns generated in the first step were used as the data for the second round of polynomial interpolations.