The most common interpolation procedure in the top income literature has been toassume that incomes in the top are Pareto distributed. This goes back to Pareto (1897),who was the first to make systematic observations of the size distribution of income.Given the nature of data, his observations were confined to the upper tail, and eventhough he initially thought that the Pareto function was a correct description of thewhole distribution (with a bound at a “physiological minimum”>0), he eventually recognizedthat the distribution function over the whole population was probably humpshapedand not Pareto distributed.27