( OMPARATIVELY little is known of the a'.4 degree to which income concentration has changed in the United States, either secularly, or from prosperity to depression.2 Even less is known of how these shifts are related to variations in the proportion of income arising from property ownership or from personal exertion, or to other economic movements less directly related to income distribution. A knowledge of the mechanics of income distribution is of obvious importance for all policy-making purposes, but as yet we are unable to describe adequately the processes which have made our distribution what it has been in the past, and much less able to predict the future. Aside from the Census samples for I944, I945, and I946 and the Federal Reserve studies for I946, I947, and I948, Statistics of Income, tabulated as a by-product of income tax reporting, is the only source giving distributions of income by size for successive years or periods on a national basis. As is well-known, however, separate reporting by husbands and wives makes it difficult or impossible to interpret taxable incomes in terms of income of families. Furthermore, taxpayers' incomes lie in the upper range of the entire income distribution. Despite these limitations, Statistics of Income has been used as the source for the following analysis, which attempts to relate shifts in the magnitude or size distribution of the principal components of income to changes in the concentration of the total in several selected years. A number of studies of the income tax data have been made by various analysts for the purpose of inferring changes in the entire income distribution from that of the upper income groups. These studies have assumed that an increase in equality among the wealthy is symptomatic of a general movement toward greater equality.3 This will not be true, however, if the decrease (or increase) in concentration of incomes among the upper income groups arises from a general shift in money incomes accompanied by a change in the numbers having incomes above any given money level. An income level which in one year may indicate the lower limit for the top one-half of the income recipients may in another (lower) income year show the lower limit for the top one-third. In I933, for example, 332,000 income reclpients reported net incomes of $s,ooo or over, while the same number of recipients, cumulating from the top down, had incomes in excess of $II,ooo in I929 and in I942. A shift in the concentration of income lying above any given money level from year to year consequently carries no implication as to the direction of change in the entire income distribution. The conclusion is simply being drawn for a different segment of the distribution. The following analysis departs from usual practice by constructing distributions for groups whose income comprises a given top fraction namely, the top I2 per cent of the national income, rather than for groups whose income lies above a given level. This does not imply that the distribution for the top I2 per cent is typical for the whole range of incomes; however, the method described may be applied to the entire range when more data are available. Twelve per cent of income recipients represents the largest fraction which could be shown for the years selected (I929, I933, I940,