But, as I have explained above, I include all those in my definition of imperial public burdens, in spite of the fact that they are used for local purposes. Turning then to the items of imperial revenue, we have to consider which of them are public burdens. The official returns described the revenue raised under the first six heads as "Revenue derived from Taxes," that raised under the next three heads as "Revenue derived from other sources." This classification may be useful for historical and departmental reasons, but scientifically it is useless, if not absurd. It is, however, generally agreed that the customs, excise, stamp and home duties and the income tax are public burdens in my sense, that is, they are taxes in the broader sense of the term. The income derived from Crown lands and the interest on Suez Canal shares and other loans are certainly not public burdens. But what about the land tax, the postal receipts and the miscellaneous revenue? I will deal with these three separately.

On looking at the details of the miscellaneous revenue, we see that a great number of the items are clearly not public burdens. A few possibly are, but there is some difficulty in deciding the question. Much ingenious argument might be employed in considering how far conscience money might be considered a public burden. In order to avoid these niceties, and partly because it would be very tedious and rather futile to discuss a great quantity of small items, most of which I should finally classify as not public burdens, I omit the miscellaneous receipts entirely.

The land tax was originally a property tax, but it gradually came to be a fixed charge upon land. In 1799 Pitt rendered the land tax perpetual, subject to redemption and purchase (39 George III. cap. 60). I do not think that at the present time we can (at any rate for the purpose of this article) look upon it as a public burden. Whenever land is conveyed, enquiries are made to find out whether the tax has been redeemed, and land is conveyed subject to the tax just as it might be to any other annual rent charge. The fact that it was rendered perpetual nearly a century ago makes it certain that all the present owners of land have acquired their land, whether

by purchase or descent, subject to the tax. Those who have bought their land have taken the fact that the land tax was unredeemed into consideration when they agreed upon the purchase money. Those who have acquired their land by descent, or under the operation of a will or settlement, cannot very well now be said to be taxed; we must consider the matter as if the original owner through whom they claim had had to pay the capital value of the tax in a lump sum. I therefore do not consider the land tax to be a public burden.

I have already stated that it is a matter of some difficulty to determine whether the profits of any business undertaken by the state should be considered as a burden on the community, and have mentioned that the answer to such an inquiry must to some extent depend upon the normal view of the functions and organization of the state. Roughly speaking, in England our industrial organization is individualistic. That is to say, we think that prices are best determined under free competition. When, therefore, the Government deliberately creates a state monopoly we are inclined to think that the difference between the monopoly price charged by the Government and the free competition price is a burden, and in many countries the creation of Government monopolies is a recognized form of taxation. If I buy a box of matches in France I find on the box the words "Contributions Indirectes," together with the date of the law which established the monopoly. Now the post office is a Government monopoly. The profits from it exceed three millions a year. Is any, and if so what, part of this three millions to be considered a burden? The finance accounts consider that this is not a tax. The learned author of Whittaker's Almanack says in reference to it, "we all contribute cheerfully." The word "contribute" is here very significant. We don't talk of contributing when we pay our butcher's bills. In fact, it is certain that many people consider that high postal rates are not merely a burden, but a very objectionable burden. But if this postal profit is a burden, is all of it a burden? On the whole, I am inclined to think so. It is very difficult to draw the line at which a fair rate of profit made by a Government business becomes a burden, but I think that it is arguable that any rate

of profit is a burden. A certain portion of the community want postal services, the Government sells them these services. If it sells them at cost price, is any one injured? I do not see that any one is. I, therefore, am going to consider the whole of this profit as a public burden. I, therefore, finally take as the measure of the amount of incidence of the imperial public burdens of the United Kingdom to be £87,853,000. made up as follows:

The next step is to divide the community into two or three classes in order to determine what is the incidence of the public burdens on each of these. That is to say, we want a definition of "upper," "middle" and "working classes." I am, however, going to divide the community into only two classes, viz.: "working" and "other." There are various principles upon which we can found our definition so long as we only care about a rough, practical grouping. For many purposes, the best line of cleavage between the two parts of the community is given by considering who keep servants and who do not, but there are some difficulties in the way of adopting this for my present purpose. In 1860 Professor Levi classified the population into four groups, as follows:

If the proportions are about the same at the present time, this would make the upper and middle classes about 34% of the population. Mr. Dudley Baxter (after making certain allowances on account of domestic servants) makes the proportions in 1867 much the same. In 1884 Professor Levi took the working classes to include 70% of the population, leaving 30%

for the upper and middle classes. In an investigation as rough as this, or any other similar one, must necessarily be, there is no need for any pretence at an accurate or very scientific division. For our purpose it may be convenient to define the working classes as the 70% of the population who have the smallest incomes; for another, as that class that does not pay income tax; for another, as that portion of the population that does not employ domestic servants, and so on. These classes given by these different methods are, of course, not identical, and in a delicate investigation it would be stupid to consider them to be so. But so much of the work that follows is, of necessity, so coarse and rough and so many of the estimates hereafter made are only guesses, that any labour spent in striving after an accurate, sensible or even clear definition would be wasted. I shall, therefore, define "working" class as the 70% of the community with lowest incomes, and "other" as the remaining 30% of the community. Taking the estimated population, this would give 27,387,000 to the working and 11,737,000 to the other class.

The next stages are the most dubious and difficult of all. We know that nearly 88 millions of pounds were raised by the imperial authorities in the year 1894-95; we assume the lower 70% of the population may be called the working classes; we want to know what was the proportion of this 88 millions paid by the working classes. We might be able to answer the question, "what was the actual amount paid by the working classes?" if we had sufficiently accurate budgets of family expenditure for the whole kingdom. But even if we had this information (which is not the case), it would not give us a complete answer to our actual problem unless a good deal more were known about the shifting and incidence of taxation than is at present. Still the budgets would be of enormous value. In default of a complete set of budgets, we must fall back upon typical budgets. But is any budget fairly typical? The ordinary housekeeper in the working classes does not keep a detailed account of expenditure. If she does, she and the family she is at the head of are almost sure to be exceptional. Most, if not all, of the working class budgets that I have seen represent families of almost superhuman prudence and sobriety.

For certain items, such as tea and sugar, these budgets may be of some use; for estimating the average consumption of tobacco and alcohol, they are practically worthless. However, I have been compelled to make use of budgets as best I can.

But the real question "what is the actual incidence of any public burden?" is of the very greatest difficulty. Theoretically, there is no insuperable difficulty in the investigation of the incidence of taxation, but the practical difficulties, which arise from insufficient data, are enormous. Even if we had (based on statistical data) supply and demand schedules for all those articles of consumption which are subject to taxation, we should not have enough. The element of time is very important in all questions of taxation, and the amount and direction of economic friction has much to do with the ultimate incidence of a public burden. Unfortunately very little has been written and almost nothing is known about economic friction. All we know is that the incidence of an old tax may be quite different from an exactly similar new tax.

one.

With all these difficulties in method and all this imperfection of data, it may well appear that the present investigation is doomed to be worthless. But there is one cause for hope. It is this. We can make a lot of small estimates instead of one big For instance, instead of considering en bloc how much of the 20 millions of customs duties is paid by the working classes, we divide our 20 millions into some 20 distinct items and make a separate estimate for each of these. Then the sum of all our errors in these 20 items is not likely to be 20 times as big as the average individual error, but it is much more likely to be only V20 times as big as the average error in each. My predecessors in this task were doubtless well acquainted with this fact and derived some consolation from it, but they have not explicitly said so. It is very difficult to estimate the error of any particular estimate, and therefore it is difficult to estimate the total error; it is perhaps absurd to try to state it in figures. Yet this does not diminish the importance of the fact that the percentage error in our final result is likely to be much smaller than that in each of our minor results. It is quite possible to obtain roughly accurate results from only very roughly accurate data.