2. The merits of the different sets of crop statistics now issued by the government.

3. The history of the controversy which has been raging between partisans of the two different departments during the

year 1902.

4. The condition of the statistical service in agricultural matters and the demand for reorganization.

First and most important, it will be desirable to consider the agricultural information published by the Department of Agriculture and by the Census Bureau, and to study the relations of the two offices.

I.

At the present time the Department of Agriculture publishes crop reports through two distinct agencies- the Weather Bureau and the Division of Statistics. The report of the Weather Bureau is published during certain parts of the year monthly, and during other portions weekly, and is designed to show in a general way the effect of weather conditions upon the state of crops in the various sections where the process of growth is in progress. These reports are, of course, never specific, and can be merely the result of the guesses or impressions of a large number of men as to the probable extent of benefit or damage inflicted by good or bad weather conditions as the case may be. As an example of the kind of work done by the Weather Bureau the following may be cited:

September 23. This week was unseasonably cool in nearly all districts east of the Rocky Mountains, with light to heavy frost, more or less damaging, throughout the central valleys, middle Atlantic states and northern portions of the central Gulf states. Excessively heavy and damaging rains.

occurred in the east Gulf and south Atlantic states. On the Pacific coast the weather conditions were very favorable, except in northern and portions of central California, where rain caused damage to grain, hay, and grapes. Late corn was damaged to some extent by heavy frosts in North Dakota and portions of Nebraska, Kansas, Missouri, and Iowa.

From such extracts as this it will be readily seen that no scientific value whatever can be accorded to the crop reports of the Weather Bureau. They are at best no more than useful

hints to men engaged in trading corn, wheat, and other farm products. As to how far the returns of the bureau are comparable for one section as against another may be gathered from the following (supposititious but representative) examples:

Alabama.-Cotton is showing the bad effects of the drought and there is a hot wind blowing.

Texas. Signs are multiplying that boll weevil has injured more cotton than was supposed.

Tennessee. The bad conditions prevailing during the past few weeks are now intensified, and the outlook for a good cotton crop is discouraging.

Evidently, if one desired to obtain an accurate knowledge of crop conditions from such returns as these, he could do so only by possessing, in the first place, general knowledge of the conditions governing the particular crop in which he was interested throughout the different sections where that crop might be growing. He would also be compelled to go through some mental process whereby these conditions were translated into a percentage estimate, however vague, concerning the extent of damage done by unfavorable weather.

Turning now to the Division of Statistics in the Department of Agriculture, it appears that in general two kinds of returns may be recognized among its figures.

1. In certain crops (hops, flaxseed, sugar, and rice) the Division of Statistics lacks a basis in acreage and so, being unable to furnish figures by its own methods, takes them from the best accessible publications. In the same way, also, are made up all tables of commercial prices as well as tables of "visible supply" of grains and those of freight movements. The tables furnished by railways afford the source from which transportation rates are drawn, while the summaries of the Treasury Bureau of Statistics furnish the material for the figures concerning imports and exports of produce. Tables for the world's production of various staples are taken from the official (or, in default of these, from the commercial) estimates of the countries to which they relate.

2. The only results for which the Division of Statistics is

primarily responsible are those which are based on the returns of its correspondents, and cover cereals, live stock, cotton, etc. Formerly there were some two or three thousand of these, one for each county, but their number has largely increased within recent years, and all told is now reckoned by some at two hundred and fifty thousand.

Each of these correspondents supposedly looks over the area to which he has been assigned and forms a mental estimate of the relations existings between a hypothetical normal crop and the forthcoming crop, the former being regarded as the base. The forthcoming crop is then expressed in terms of the normal by a percentage method. These correspondents number, as has been said, perhaps two hundred and fifty thousand. They are, at the present time, divided into three classes—one for states, one for counties, and one for smaller districts subordinate to counties. These three classes of correspondents are entirely independent one of the other, and report in every case directly to the Division of Statistics in the Department of Agriculture. Thus the division has at the end of each report period three sets of data upon any one of which it may base opinions regarding. the condition of crops throughout the country. It is proper to add that of these correspondents, only those representing states are paid by the department.

When there are several returns from correspondents for a county, the usage is to convert the figures into an adopted county return by taking a simple mean; but the county figures are combined into state averages by applying the proper weights. These are furnished in all cases by the last decennial census; thus, if the census figures show three counties to have produced 7,000, 3,000, and 15,000 bushels of wheat, respectively, and each furnishes a figure for condition of the growing wheat crop, or for price per bushel of the harvested grain, or for percentage of the last harvest remaining at a certain date in the farmers' hands, the three figures have weights in determining the state average condition, or price, or percentage, relatively as 7, 3, and 15. Similarly, census reports of area are used in weighting county figures for percentage of area as compared with last year, and

for yield per acre all through the ensuing decade. State averages are built up into general averages for the county by the use of weights, also; but in this case it is not necessary to depend upon a census which may be seven or nine years out of date, because the recent figures of the division itself are for this purpose available. When the present crop is unknown, as in calculation of condition of growing grain, the reported crop of the preceding year is taken as a basis.

The three quantities on which the report of the crop depends are: (1) the area in acres; (2) the average yield per acre; (3) the average price per bushel. Correspondents are also asked to estimate the total product as compared with that of the preceding year, as a check; but little practical use is made of this. Areas can be determined comparatively only, in the absence of a regular farm-to-farm visit by a paid agent. The basis employed is the acreage of the preceding year, of which the present acreage is expressed as a percentage. An average percentage having been found for a state, by census county areas as just explained, that percentage is applied to the state acreage reported for the preceding year, and the present year's acreage thus inferred. For example, if the state shows a weighted mean percentage of 103, this year's area is found by adding 3 per cent. to last year's. This year's area, thus determined, becomes in like manner the basis for calculating next year's, and so on, while the new census furnishes an improved basis. Any error that enters into the deduction of any acreage, from that of the year preceding, must affect, not only that year's figures, but the next, and all thereafter till the end of the decade. There is no tendency in such errors to cancel one another out, as where like quantities are added and accidental differences are neutralized by the law of averages; and unfortunately the defects of an estimate of area by percentage are shown by frequent experience to be of an accumulating nature, so that a divergence from the fact tends to become aggravated rather than corrected. Comparison of the last census figures of acreage and product of wheat, for some of the far southern as well as New England states, with those of the Agricultural Department for the same crops, will show several instances of this.

Besides the three quantities named, several others are ascertained by the Division of Statistics. Amount of grain in farmers' hands on the 1st of March, and again at the end of the crop season, reported as a percentage of the total crop harvested; proportions consumed in the county where grown, and proportion shipped out, reported also in percentage; general quality, and weight per bushel; all these figures are brought out in returns from correspondents, made at suitable times. The crops on which annual reports of area, product, and value are made are the six cereals (corn, wheat, oats, barley, rye, and buckwheat); also potatoes and hay. Figures for condition, during the growing season, cover many other crops, for which final estimates are, for reasons already sufficiently given, impossible. An attempt was formerly made, year after year, to calculate the tobacco crop; but the tendency of its acreage figures to run progressively too short was too strong for the department to counteract, and the results became so unsatisfactory that the statistician decided to publish no more of them. The results for annual production of tobacco are therefore not given by states in the Year-Book, but appear, for the whole country, in a table based on figures from the Bureau of Internal Revenue.

For determining the acreage, yield, total product, farm price, and total value of cotton, the same methods used for other crops are adopted; and are regularly used by the Department of Agriculture. But, in addition to these methods, others are made possible by certain facts peculiar to that crop; that practically none of it is consumed on the farm where grown; that it must all go into definite and comparatively few channels, where its amounts may be readily and exactly ascertained; that it does not decay and is bulky. It is possible, on these accounts, to ascertain the total amounts at points of shipment, by water and by rail; and then, allowing for duplication, for amounts that cross state boundaries, for amounts taken by mills within the state, and for amounts still held by the growers, to arrive at a more exact result for the production of the several cotton states without depending necessarily on the acreage. Besides the main facts with regard to this crop, the correspondents report others;