Trends in the world distribution of income are much discussed and philosophised about. The issue of increasing economic inequality among nonsocialist countries in the post-war period is central to the debate about the International Economic Order. In I969 the Pearson Commission Report was already pointing to the 'widening gap between the developed and developing countries' as a 'central issue of our time', and was proposing the reduction of that gap as the main objective of international co-operation.' According to that report, GNP per capita had increased between I960 and I967 at annual rates of 2-5 % and 3-6 % in developing and developed countries respectively. Morawetz (I978) found that the gap between the two sets of countries rose from i 2 6 fold in I950 to I3-2 fold in I975. Other authors see a more complicated picture. Atkinson (I975), for instance, has argued that although the share of poor countries in world income probably declined between I950 and I970, that of the countries at the middle of the world distribution probably increased.2 What may be a mere assessment of relative growth performances among nonsocialist countries becomes a basic theoretical element in the radical stream of economic literature. For authors like Amin (I974), Emmanuel (I972), and Frank (I978), the capitalist mode of development necessarily leads to the rich countries of the 'centre' becoming richer and poor countries of the 'periphery' becoming poorer, in a degree depending on their involvement in international trade. For them, a widening gap between developed and developing countries is simply the statistical confirmation of an unavoidable consequence of the national and international capitalist system. In the latter perspective, the issue at stake is of such importance that the contrast between the abundance of theoretical literature and the lack of empirical work is striking. The few existing studies of changes in world distribution provide an inadequate picture. Intra-country income distributions are not taken into account, observation periods do not exceed I5 years (8 for Andic and Peacock (I96I) and for Beckerman and Bacon (I970), I5 for Kirman and Tomasini (I969)), socialist as well as some non-socialist countries are excluded, and national incomes are converted to U.S. dollars at official exchange rates.3 Finally, none of these studies covers the I970s, although several shocks which might have affected world distribution occurred during that decade. For the non-socialist countries to which they refer, the Andic/Peacock and Beckerman/Bacon studies found the Gini coefficient almost constant over