Atkinson and Sandmo (1980) evaluated the taxation of saving by embedding the static optimal-tax framework within the two-period overlapping-generations model. Using this model, they collected a series of important results for optimal taxes on labour and capital income under a variety of assumptions regarding the instruments and objectives of the government, in particular the availability of national debt to spread burdens among different generations of individuals. With clear intuition and presentation, the study remains an important source for understanding the welfare implications of the taxation of saving.The 1960s and 1970s saw rapid development in the field of optimal tax analysis, based on and extending a handful of much earlier and fundamental contributions such as Ramsey (1927) and Corlett and Hague (1953). This literature focused primarily on the minimisation of deadweight loss arising from consumption and labour supply distortions in models of linear taxation, although an important extension was the consideration (and general rejection) of production distortions in the work of Diamond and Mirrlees (1971).A separate line of research, on optimal non-linear labour income taxation, commenced with , and the two strands of the literature converged in a heavily cited article by , which developed conditions under which consumption distortions were no longer helpful in improving social welfare in the presence of an optimal non-linear labour income tax. Although many contributions over many years followed, the research of this period largely defined the field of 'static' optimal taxation -static in the sense that time did not play an explicit role in the models used.Of course, one can introduce time into static models in the Arrow-Debreu sense of treating consumption that occurs at different dates as distinct commodities and this approach allowed an interpretation of existing optimal tax results as applying to the taxation of savings. For example, Feldstein (1978) argued in favour of expenditure taxation over income taxation based on the results of . But this is a very limited approach to dynamic analysis, for it ignores important elements of taxation over time, perhaps most notably the existence of different generations who may lack trading opportunities and who may be affected differently by a particular change in tax policy. The article by Atkinson and Sandmo (1980), 'Welfare Implications of the Taxation of Savings', addresses this limitation by embedding the static optimal tax analysis in the classic Diamond (1965) model of overlapping generations.