According to the OECD, in Finland and New Zealand the top level 'genius strand' is reached by 4 per cent; in the United Kingdom and Australia it is 3 per cent; in Germany and the Netherlands 2 per cent; in the United States and Sweden 1 per cent; and in Portugal and Italy nearer to O per cent. [...] if you read the report's technical manual (released three years after the survey), you will find hidden after 144 pages of equations and procedures the statement that, in calibrating the results (adjusting the scores before release): 'those releasing this data . . . assumed that students have been sampled from a multivariate normal distribution'.2 Given this assumption, almost regardless of how the students had 'performed', the statistical curves produced would have been bell shaped.