FIGURES THAT MISLEAD.

45

25

spirituous liquors consumed is correctly stated at $900,000,000. In very proper contrast with this enormous waste, the cost of public instruction is set down at $85,000,000, and the cost of missions at $5,000,000. So far well! But the statistician in comparing these figures with the cost to the people of their annual consumption of meat, inadvertently assumes that if the United States census gives any of the figures of meat consumption it must give them all. In fact, however, it only gives the value of the meats packed in the packing houses, these being all which can be classed under "manufactures," and it is as a form of manufactures only that meat statistics are taken. All the meat supplies furnished by the 76,241 butchers of the United States, and all meats butchered on farms by those who raise it, and all poultry, game and small meats of every kind are omitted from the census. Hence, the unwary statistician in search of the figures of meat consumption to compare with spirituous liquors, states them at $303,000,000, being the correct amount named in the census, but only for the meats packed in packing houses. This would make the meat consumption for 51,500,000 people about $5 per capita per year, or, say ten cents per head per week. In fact, in manufacturing and urban populations, the cost of the meat bill is more nearly $1 per head per week. The consumption of bread, groceries, and vegetables combined is about half or two-thirds as much more. Assuming the cost of consumption among farmers to equal that in cities, the two classes of food, animal and vegetable, would cost about $3,300,000,000, or about four times the expenditure for spirituous liquors, which is about the proportion which a person of good judgment would assume if not misled by the figures.

An imposing pretense of statistical proof is also sometimes made by collecting a formidable array of statistics of a feature of the case which is not in dispute, and leaping over the real point in dispute without any statistics whatever that bear upon it.

Thus Mr. Springer, of Illinois, in an article in the North American Review, purports to state in statistics, by how much the cost of various articles of American production is increased to the consumer, by the duty on importations. He concedes that they are not increased in cost by the whole amount of the duty, but adopts a purely deductive, a priori, hypothetical and unsustained fraction of a third, half, or two-thirds of the duty, and says it is reasonable to assume that they must be increased by

this amount at least. The point to which his statistics were needed was that of the degree in which in any one case the duty causes an increased price. This was the knob of the question, and this he covered wholly by a guess. Having guessed at his standard, the calculations based upon it were as devoid of any element of statistical proof as if he had first guessed at a quarter of the amount sought and then multiplied it by four.

Again, economists sometimes assume that statistics show as to one class what they only show as to a class having very unlike interests. Thus Mr. Fawcett says: "There is no surer test of the prosperity of the laboring class than the low price of bread, and there are few statistical facts better substantiated than that the marriages among the laboring class increase with the fall in the price of bread."

But Mr. Fawcett does not say whether the laboring class to which he refers is that which produces or that which consumes the bread. In 1849, when the effort to make bread cheap in England by giving it free importation from abroad, regardless of the interests of the bread producers, culminated, the ejectments in Ireland among the bread producers rose to eight-fold their previous number, and the arrests for crime increased in like proportion, population diminishing in five years by three-eighths. Whatever force Mr. Fawcett's alleged statistics, if such there be, may have, it could only relate to a portion of the laboring class, viz. to those that labored in factories, and not on farms.

There are also errors in the collected statistics which could only be removed by costly investigation, and, no such investigation having been made, no removal of the error is possible. This arises especially where statistics of the same fact are taken by the officials of two governments, as where both keep an account of the shipment or receipt of a product which to the one is an import and to the other is an export.

Thus in a recent year French customs officers returned the export of silk from France to Belgium at 36,862 kilogrammes, while the Belgian officers returned the import of silks from France for the same year at 25,947 kilogrammes. In the same year the export of coals from Belgium to France figures in Belgian exports as 21,121,520 cwts., whereas the same fact figures

"The Condition of Nations," by Kolb, Brewer and Streeter, p. 895.

CONFLICTS IN STATISTICS.

27

in French imports as 19,655,869 cwts. In 1853 the export of wool from Belgium to France amounts in the Belgian returns of exports to 371,260 kilogrammes, while the French customs returns of arrivals of wool in France from Belgium show that 3,301,500 kilogrammes were received. In one year the value of wool imported into Great Britain from France is stated in the French books to be 11,750,000, francs or £470,000, while in the English books it appears as of the value of £1,280,280. In one year French returns show an export to England of cotton to £345,760, while English returns show an import of cotton from France of only £180,000. The next year France sends to Great Britain, according to French returns, 952,035 kilogrammes of silk goods, while Great Britain receives from France, according to English returns, only 233,739 kilogrammes. So by British returns Great Britain. sends into France 245,925 kilogrammes of silk, while according to French accounts France receives from England more than twice as many, viz., 597,354. The English return their export of wool to France in one year at 2,207,741 kilogrammes, while the French return their import of wool from England at 3,940,496 kilogrammes. In corn 87,716 hectoliters on one side appear as 312,768 hectoliters on the other. Coals start from England as 7,292,411, and arrive in France as 5,631,829 cwts. Comparing English customs returns with Belgian, 1,015,173 kilogrammes of coffee dispatched in one year from Great Britain to Belgium, are made to arrive in Belgium as 572,613 kilogrammes. Again from England to Belgium 4,036,049 kilogrammes of wool are exported, and only 1,643,766 kilogrammes of wool are entered in Belgium in the same year as received from England. 1,676,701 kilogrammes of hops are sent from Belgium to England, and only 861,466 appear in English returns as received from Belgium. Belgium sends glass goods to the British Isles, in quantity 3,776,544 kilogrammes, while England acknowledges the receipt of glass goods from Belgium in 5,645,826 kilogrammes. It is probable in the cases cited that the returns on both sides are accurate in the mode in which they are taken, but that the quantities which are covered by the same terms are really different quantities. For instance, in dealing with wool, the customs officers on one side may have but one designation for wool, or hair, shoddy and fleeces, while the other may have two or three, viz., wool and mixed, or wool, shoddy, hair and mixed, in which case what one sends as wool will not be received by the other as wool. This will make the returns differ in the

statistics, while each will be truthful according to its own system and law, but comparisons made on such data must be made with care, i. e., figures when compared must be known to have been arrived at by the same process of collection. Again it may be that the French system of receiving goods at the Belgian frontier does not distinguish between English goods coming through Belgium into France and Belgian goods, while the Belgian may call one a transit of English goods through Belgium and not an export of Belgian goods. It is certain that if the two countries classified goods in the same way, and both made correct returns, there could be no discrepancy.

When one sees very startling facts presented as statistics, therefore, it is proper to call for the mode, or system, or authority by which the statistics were collected. If they were collected in different countries, or at far distant periods of time in the same country, and the contrast they present with what the common judgment would infer is very great, it is reasonable to withhold assent until it is known that the general process by which the contrasting data were collected is the same.

We are told that in France, previous to the existence of railways, there was one passenger killed to every 335,000 carried, and out of every 30,000 carried one was wounded; whereas out of 1,781,403,678 passengers who traveled on the railways in France between September, 1835, and December, 1875, only one to every 5,178,890 was killed and only one to every 580,450 was wounded. This makes railway traveling seventeen times safer than traveling in a diligence or "stage" behind horses. On French statistics it has been estimated that if a person could lose his life only in railway traveling, the chances would be that he would only die at the age of 960 years. On Belgian railways the proportion killed of those who travel on railways is only one in 20,000,000, or one-fourth that of France. In the United States, on the other hand, in 1875 there was one wounded by railway travel in every 172,000 carried, and one killed in every 810,000 persons carried. Statistics also show that in England the dangers of railway travel exceed those on the continent and even those in the United States, being in 1876, 1,245 killed out of 538,287,295 carried, or one in about 430,000, which is a rate essentially equivalent to the old stage coach rate in France. Now either the railroad travel is

"Condition of Nations," Kolb, Brewer and Streeter, 899.

SAFETY IN TRAVEL.

29

much safer in France and Belgium than in England and the United States, or accidents are less fully reported.*

It happens that in France and Belgium, railways are under government control, while in England and America they are under private control. If now it should also be true that, under state control, the railway operators who have accidents are dismissed, while under private control they are not, this fact might cause more accidents to be concealed or not reported, and might also cause fewer to be committed. On the other hand, in France, under the old diligence system, the rate of accident was about as great as the rate now is in England under the railway system.

As the public mind has a fixed impression, or common-sense judgment, that railway travel is as dangerous as stage coach travel, it would need fuller statistics to make it plain whether the pretended increase of safety of railway travel over stage travel in France does not result from a less trustworthy mode of reporting the deaths under the system of state control than under that of private control. If not, then a strong case is made out for state control over private control, as respects safety in travel, as well as for railways over coaches drawn by horses.

Among the most doubtful statistics in general use are those entertained in Europe and America concerning the population of China. Very eminent publicists discredit them. In explanation of certain discrepancies in these statistics it is said that when the Chinese government called for returns of population with the view to distributing taxes, the population returned was small, but when returns were called for as the basis of distributing imperial aid or charity among the provinces the population rose to "myriads of millions."

So in parts of the United States the desire of counties to escape taxation has led to assessments of values of both land and personal property at from one-sixth to one-twentieth of their true

*In 1884 the average number of miles a passenger could travel by rail without being killed was :*