hardly hold good for the whole community. The abounding and growing expenditure which we see on all sides by those whose lot is easy, suggests at least a doubt whether the budget method, as applied in this investigation, supplies us with a key to all the difficulties in giving to different commodities their due weight in getting a general expression of the rise or fall in the value of money.

So much as to the questions which may arise on the soundness in principle of the budget method of weighting; from which we may proceed to a consideration of the success with which it was applied in detail. The difficulties in the way of the other method of weighting, that by total expenditure, are obvious. It is impossible to ascertain the total expenditure by the community on more than a few articles, and these largely raw materials, like cotton, wool, iron, wood, which enter into actual consumption in the most diverse and complicated ways. Hence the total expenditure method, as applied for example by Mr. Palgrave to the Economist's figures, can not be said to promise more satisfactory results than the simple method of unweighted arithmetical average. The question arises whether the budget method is susceptible of more accurate application in detail, and so promises better practical results.

Here we find that Professor Falkner's industry and discrimination, and the large resources put at his disposal by the Bureau of Labor, have yielded results solid in character and as little in need of correction and allowance as the inherent difficulties of the problem made possible. As has already been said, the expenditure by the selected families (232 in number) was followed into its details for individual items,for beef, eggs, flour, potatoes, clothing, light, fuel, and so on. The price of each commodity was known year by year, and was given weight, in making up the general index number. for each year, according to its importance in the family budget. This could be done with ample completeness and accuracy as to food. With other items of expenditure, the method was more difficult of application. The budget stated what proportion of the family expenses went for the clothing of husband, wife, children. But the list of prices showed

only what certain woollen goods, cotton goods, silks, linens, leather, and shoes, had cost. It was necessary to combine and group the quotations of prices in such manner as to fit the items of expenditure. Thus the prices of suitings were used in connection with the preparation of expenditure on men's clothing; of shoes and leather combined, for shoes; of dress goods, for women's clothes; of blankets, flannels, cotton textiles, and linen goods, for all other items of clothing. These combinations were made with skill and judgment; yet they inevitably introduced an undesirable artificial element into the process of weighting.

Clearly, some items of expenditure could not be fitted at all, or only by arbitrary supposition, into the figures of prices. Rent is a large item in expenditure; but how much of this goes for bricks, wood, glass, it is impossible to say. A considerable expenditure among the selected families had not been itemized at all, but simply set down in the budget as "miscellaneous." This was taken into account, nevertheless, in making up the weighted average, by assuming that one half of this miscellaneous expenditure was for the direct purchase of commodities; and by assuming further that these commodities, already assumed to be directly purchased, consisted of all the articles in the list which had not already found a place in the specifically itemized articles of the budget. The prices of all these articles, not traceable in the budget statements, were yet given an importance in forming the general average, determined by the proportion which one-half of the miscellaneous expenditure had in the total expenditure of the family. These other articles, it may be noted, included all the metals and implements whose prices were quoted, all the drugs and chemicals, all the lumber and building materials. Here again we have an artificial element of considerable importance,-a supposition and not a fact in the distribution of expenditure and the consequent weighting of commodities. This description of itself suffices to show how difficult it is to carry out into practice the budget principle of weighting commodities according to their importance.

In one way or another, partly by direct and specific assignment of items of expenditure to prices of individual

articles, partly by adjustments and suppositions such as have been referred to, 68.6 per cent. of expenditure was accounted for, and a weighted average of prices was calculated, resting on the importance of the various commodities in this part of the expenditure in the families of moderate means. Two sets of general index numbers were made out on this weighted method. In one, the index numbers were calculated by applying the figures of prices only to this 68 per cent. of specifically apportioned family expenditure. In the other, the same apportionment or weighting was used, but it was assumed that the rent and other items of expenditure, whose prices were not ascertainable, remained equal in amount. This second set of index numbers seems to be of very doubtful value; and in the comparison of the results of the weighted and the simple average, it will be neglected, and attention will be confined to the first set, in which the greater part of the family expenditure is itemized and is used to give importance to the different commodities.

If these two methods, then, of simple arithmetical average on the one hand, and average weighted by family budget importance on the other hand, yielded greatly different results, we might be perplexed which to use as significant of the general course of prices. The arithmetical average is simple and straightforward, and brings little liability of un. noticed error. It is obviously faulty in principle, since it gives all commodities equal weight; but, by taking a very large number of commodities, the mistakes of emphasis may be expected to balance each other. The budget-weighted method is sounder in principle; but the budgets may not be fairly representative, while the ascertainment of proportional expenditure is difficult, and necessitates artificial allowances and assumptions. Fortunately, in the present investigation, the two methods lead to results so surprisingly in agreeBement that we may be sure neither is greatly in error. tween the two, we have an indication of the general movement of prices as accurate as is likely to be secured by any method.

A glance at chart I will show that from 1861 to 1891 the lines indicating the movement of prices, one drawn by

the simple arithmetical average, the other by the budgetweighted average, run together with remarkable closeness.'

During the years from 1840 to 1860, it is true, they diverge, the weighted average being lower than the simple average. But this difference is probably due largely to the fact that a much smaller number of commodities was taken in account in forming the averages before 1860; and further to the greater difficulty of securing quotations of prices before 1860, which it was possible to fit to the items of family expenditure. There is, indeed, another not improbable explanation of the divergence. Prices of food of various sorts were lower before 1860 than since; food is given a large weight in the budgets; hence a low price of food before 1860 would perhaps lower the general index number as ascertained by

1 This chart also has a line indicating the course of the premium on gold from January to January of each year, to which reference will be made at the close of the paper. For the present, the lines indicating the movement of prices, as ascertained by the two methods, alone need attention.

the budget-weighted method, more than by the simple average method. It is unfortunate that the less complete material at our disposal before 1860 makes it uncertain whether we have here a case of really important differences in result, due to difference in method.

But, to repeat, in the thirty years since 1860, for which we have ample material, the results are in remarkably close correspondence. During the civil war, the lines move together with an evenness which is extraordinary, in view of the anomalous conditions of that period of inflation and confusion. The prices from which the averages are reduced, it must be remembered, are January prices. During 1863, 1864, and 1865, January prices might differ widely from December or February prices; and either the weighted or the simple average might be seriously affected by the accident of a month's variations. Bearing this in mind, we must be surprised to find that both methods agree as closely as they do in their evidence on the rise of prices due to the great paper issues. They agree, too, and even more closely, in showing the gradual decline in prices after the high water mark had been reached in 1865. A check to the downward movement came in the speculative period which preceded the crisis of 1873. The rapid decline was resumed after 1873, and reached its end with the period when the resumption of specie payments was effected, in 1879. It is clear that the effects of the paper money issues of the war did not exhaust themselves until that year, when prices were finally at the point from which they had started before the inflation began. The wrongs, the injustice, the chaos in the relations of debtor and creditor, the instability of all industrial enterprise, the occasions for unearned fortunes and undeserved calamities, which must result from sudden changes. in the volume of the currency, could not be more vividly illustrated than by the soaring line of 1862-65 and its sinking successor in 1865-79.

That the fall in prices to the specie point coincided with the date fixed for the resumption of specie payments, can not be regarded otherwise than as a lucky accident. The fall was due in but slight degree to legislative action in contract