A census of all housing values was made in the United States in 1798 which provides the basis for obtaining an estimate of the distribution of income in that year. A 7% sample of the 576,800 dwelling units was adjusted for the number of families in a house and various assumptions were made concerning the elasticity of housing with respect to income. The relative inequality of income with a Gini coefficient of between 0.63 and 0.71 was much greater than it is today. Similar results were found for housing distributions then and now. How much income inequality existed in the United States two centuries ago? Was the Gini coefficient, G, less than now? These are questions that can be answered by studying the 1798 census of housing. This data set provides an excellent opportunity for the study of income groups in America, including a detailed quantification of the extremes, the affluent groups and their mansions and the many who lived in humble circumstances. This data set shows such surprising inequality that adjustments for crowding (the number of famlies in each house) and for possible income-rent elasticities have little effect. Analysis of the data supports the hypothesis that income inequality in the United States at the end of the 18th century was quite large, and that it is more tenable than is a hypothesis of mild

Ahstrrct-The distribution of wealth in real estate am?ong people in the United States at the end of the eighteenth centuLry has hcen estimated frornt a samplc of rolls of the censuLs of real estatc in 1798. Mean real wcalth was $1,433 and the (lini coeficieint of incqualitv was 0588 with half the aduLlt male population owning property. A comparison with the distribultion for rcal estate in 1860 shows that wealth grew 1.9% a year pcr person. Relative inequality was a little larger than in 1798 buLt the diflcrcnce can be explained by crrors in mncasurcmcnt.

JN working with various data sets for the United States in the 19th century, it has come to my attention that pronounced patterns emerge which depict the relationship between increases in wealth of individuals as they become older. When we abstract from long-run growth, by examining age-specific wealth averages in a given year such as 1850, 1860, or 1870 or even a century later, this wealth-age gradient appears to rise about 4% a year. The increase is usually attributed to savings and capital gains of individuals rather than to transfers from inheritance. If such is the case, the reason for taxing wealth is less compelling since wealth is, in part, a reward for current effort and can not easily be identified as the cumulative effort of past generations. John Brittain has recently publicized and criticized this particular way of measuring and factoring the influence of individual endowment, that is, the effort of activity of the present generation as contrasted to that which is the product of past generations. He has suggested that specialized studies be made of inheritance in an endeavor to better determine whether the wealth of one's parents might explain, in some statistical sense, 1/10, or 1/3, or even 2/3 of the wealth of the present generation. ' It is the purpose of this paper to study some of the relationships between wealth, number of children, and age of fathers, as related to the inheritances and ages of sons, by using data from the 1870 U.S. census of wealth. Statistics will be presented which describe the possible amounts of wealth that are passed on from generation to generation in any given year during the life cycle of the son. I shall argue that a 4% wealth-age gradient can appear within one generation which duplicates any original wealth-age configuration; incentive implications of any current patterns are less cogent in this sense even though only 1/3 of all current aggregate wealth may be considered a transfer from the past. The patterns to be described are of significance not only for the general cultural historian, but also for anyone trying to understand an economic model of inheritance since the extent of parental domination, the influence of family size, and, more generally, the degree of wealth inequality imposed from one generation to the next all contribute some sense of the quantitative impact of inheritance factors even if the model is very elementary.

T HE recent emphasis on the plight of lowincome groups necessitates an examination of how these groups have fared in economic progress compared to higher income groups. The pattern of inequality within higher income classes is relevant to the relative poverty of lowincome groups. A detailed examination of inequality within specified groups is made in this paper. The inconsistency of the pattern facing those in the lower portion of the distribution in the postwar period is emphasized.

Wisconsin state individual income tax data present an excellent opportunity for a detailed trend study of the changes in the dispersion of taxpayers' income over a long period. The available concept, net taxable income, which is income after deductions except federal income taxes but before exemptions, remained essentially constant in definition from I929 to I949. The minimum filing requirement, rate, and exemption schedules remained the same in this state which has a long heritage of income tax payments. These data are more nearly comparable over a long period and had a much lower filing requirement in 1929 than federal data. This situation makes it plausible to work within the intricate detail of small class intervals for incomes above $2,000 in comparing the income distribution for the year 1929 with that of I 949. The purpose of this study is to examine quantitatively how people in different income ranges lost or gained relatively in comparison with the average percentage increase in the twenty-year period and to state, with qualifications, how much more or less people would have made before federal taxes in 1949 than they did, had the distribution of income remained constant from I929 to I949 for different income ranges. Comparisions of distributions. Both the increase in population and per capita income are considered in accounting for the upward shift in the distribution during the period. The I949 population was I 15 per cent and the Department of Commerce I949 state per capita personal income was 200 per cent of that in I929. If the distribution of income had not changed, the I949 frequencies with incomes approximately double those in I929 should the frequencies of the I929 distribution increased by I5 per cent. The data for such an analysis are given in the accompanying table for individuals making over $i,8oo net taxable income in I929 and $3,600 in I949. It can be seen from the data that only in the income classes from $2,IOO-2,200 in I929 and $4,200-4,400 in I949 are there an equivalent number of returns' The disparity between the two distributions becomes relatively greater for rich income class groups. In an attempt to quantify the change in the concentration ratio or shape of the distribution from I929 to I949 for different segments of the distribution, the question arises how much income rose for different classes in the I929 distribution in the twentyyear period if it did not double. The simple expedient of determining where in the I949 distribution there are as many returns in a class whose class limits and class interval are a certain percentage of those of a I929 class with the same number of returns is used. An example is the class from $2,8002,900 in I929 which has a frequency equal to that from $4,860 to $5,040 in 1949. The best match class limit percentages for the various I929 classes is presented in the last column of the table, with three distinct groups appearing,