amount should be expected also to increase. People ordinarily think of a "large fortune" as some more or less definite absolute amount. If the words be given this meaning, then increase of large fortunes evidently does not involve concentration of wealth. We should expect the number of large fortunes to increase not merely as population increases, but also on account of the increase of per capita wealth. Since per capita wealth has been increasing in the United States, we should expect the ratio of millionaires to population to increase. The fact that this ratio has increased does not prove concentration of wealth. For the same reason, neither does the fact that our greatest fortunes to-day are larger than were those of a generation ago.

We may look at the question in a different way. We may say that a man is rich when he has 100 times the average or per capita amount of property of his countrymen, and that he is very rich when he has 1,000 times the average property. According to this definition, if per capita wealth be $500, any man who has as much as $50,000 is rich. But, if per capita wealth be $1,000, a man must have $100,000 to be rich. This, I think, is the better way to define the "large fortune." The idea should be that of a relative quantity. Men's ideas of "middle" and "upper" economic classes are certainly relative. If we assume that the growth of large fortunes and the concentration of wealth go together, we must take the term in its relative signification. Ordinary usage is of course ambiguous, not being clearly thought out. Taking the phrase thus relatively, we find it is not obvious that "large fortunes" have increased in importance in the United States in the last one or two generations. It may be so. But it is not the kind of conclusion that can well be based upon mere observation. The proof of concentration of wealth is a difficult and technical statistical problem.

An obvious and apparently the usual method of interpreting statistics of fortunes or incomes classified by size is to compare the per cents. of number in the several classes at the different

dates and also the per cents. of property possessed by the several classes. An objection to such procedure is that the uppermost class has no upper boundary, though this criticism is met in part where per cents. of amount of property as well as of numbers of owners are used. To be complete, the scheme should also include a class without property; classes should be of comparatively small range. But there still remains the fundamental objection that such classes, as they have been used, are bounded by absolute and fixed amounts, while concentration is a fact of relation. The boundaries of the classes should themselves be relative numbers, and should change with the per capita of the total amount of fortunes. If wealth classes are to be employed, the dividing lines should be at multiples of some sort of an average.*

The essence of the following attempt to interpret statistics relating to concentration of wealth is its use of thoroughly relative criteria. Another important difference from usual methods of treatment consists in keeping this question as to concentration of wealth separate from the other and larger question as to the character and tendency of the distribution of incomes, including thus incomes from labor. Special attention is given also to the peak of the pyramid of fortunes. One reason for this-in addition to the peculiar interest of that aspect of the question which relates to the growth of great fortunes-is the difficulties met in trying to obtain tolerable statistics of small properties, as distinguished from small incomes.

To test for a tendency to concentration of wealth, probate and inheritance statistics are the best available material. There is an assumption involved in the use of such material. But it is a fair assumption that any considerable influence affecting the degree of concentration of wealth among the living will soon affect in the same direction the distribution of property in the estates of deceased persons. In other words, the errors are constant and do not affect tests of tendency.

* For an extended discussion of this rather technical subject of the method of measuring concentration of wealth, the reader is referred to articles by Messrs. Lorenz, Holmes, and Watkins, Publications American Statistical Association, Nos. 70, 71, 72.

The best evidence we have for the United States bearing on the question under consideration is certain probate statistics compiled by the Massachusetts Bureau of Statistics of Labor. Following are the important primary tables:

PROBATES IN MASSACHUSETTS: NUMBER AND AMOUNTS, AND THEIR PERCENTAGE DISTRIBUTION, 1829-31.*

PROBATES IN MASSACHUSETTS: NUMBER AND AMOUNTS, AND THEIR PERCENTAGE Dis

* The absolute numbers are from the Twenty-fifth Annual Report of the Bureau of Statistics of Labor, Massachusetts, 1894, p. 265.

† Ibid., p. 267.

PROBATES IN MASSACHUSETTS: NUMBER AND AMOUNTS, AND THEIR PERCENTAGE DISTRIBUTION, 1879-81.*

PROBATES IN MASSACHUSETTS: NUMBER AND AMOUNTS, AND THEIR PERCENTAGE DIS

* The absolute numbers are from the Twenty-fifth Annual Report of the Bureau of Statistics of Labor, Massachusetts, 1894, p. 267.

+ Ibid., p. 267.

In these statistics there is exhibited a notable tendency to an increased number of the larger fortunes. There were two estates of over half a million probated in the earliest threeyear period, and 30 such probated sixty years later. Meanwhile population had increased 267 per cent. The estates above $100,000 were 11 in number in the earliest period, and 244 in the latest. The wealth of the community certainly did not increase at such a rate. The greatest growth of riches, but chiefly in numbers of moderate fortunes, appears to have occurred in the thirty years preceding the Civil War. In the latest decade covered there appears a tendency to reverse the process. For the latest period there is a decline in the relative number of half-million dollar estates probated. The tendency to concentration doubtless began to produce marked effects earlier in Massachusetts than elsewhere, owing to its leadership in manufactures and its lack of capacity for agricultural expansion. Hence the tendency may well have worked itself out by 1880.

But let us consider these figures in their proper relation to the growth of per capita wealth. According to the United States Census, the "true valuation" of real and personal property in Massachusetts for several decades past was as follows:

The per capita wealth of the State thus declined noticeably between 1880 and 1890. The two sets of figures, probate and census, are in general agreement as regards the indication that the relatively large number of estates of half a million or more in 1880 was in part but an expression of the greater per capita wealth of the State in that and in the immediately preceding years, as compared with ten years later and with twenty years