Decreasing quality of semen

produce a measure of relative differences in each country. The figure combines the data for neonatal and postneonatal mortality and shows that England and Wales have a steeper and more consistent gradient of mortality across social classes than Sweden. The width of the "steps" is proportional to each class's share of total live births. The differences in total mortality and its socioeconomic distribution in the two countries are both consistent with the suggestion that Sweden's more egalitarian income distribution might be influential.' Mortality in each social class will be affected by the scale of relative deprivation in each. Social classes are heterogeneous with respect to income: there are unemployed and low paid workers in social class I occupations just as there are occasional well paid unskilled workers in social class V. Therefore, while having its greatest impact lower down the social scale, a more egalitarian income distribution may be expected to produce a small reduction in the burden of relative deprivation on mortality even in the upper classes. Though Leon and colleagues seem to assume that smaller differences in income should lead to a relative rather than an absolute diminution in differentials in mortality, if relative deprivation is regarded as a risk factor like any other, an absolute effect may be more plausible. This would also be consistent with its impact on total mortality.