connection the work carried out in 1903 by a special committee nominated by the Royal Statistical Society of London deserves mention. Its terms of reference were "to inquire into the statistics available as a basis for estimating the production and consumption of meat and milk in the United Kingdom."

Reference may also be made, inter alia, to the estimate made for Switzerland by the Swiss Peasant Secretariat; for Germany by Kirchner, Kuczinsky-Zuntz, and Semler; for Italy by Professor Besana; for Sweden by Haglund; for Hungary by Koerfer, etc.

5. There already exist in certain countries partial statistics on milk and dairy products or at any rate such indirect information as is contained in the data supplied by producers' organizations, cooperative dairies, establishments manufacturing butter, cheese, and the like, municipal returns, etc. Canada, the United States, and Argentina give annual returns for the output of dairy products in industrial establishments or factories, excluding the produce on the farms themselves. Sweden publishes full annual statistics of dairy work (number of dairies, amount of unskimmed milk produced, amount sold, amount of cream sold, and cheese and butter produced). Norway supplies every year the amount of milk handled in the dairies and condensed milk factories, and Finland gives figures for the butter and cheese manufactured in the dairies. Denmark on the other hand publishes figures for a large number of dairies, indicating the number of cows and the average annual milk output per cow, etc. In Holland the annual amount of butter produced under the surveillance of the control stations is ascertained. It is also possible to obtain periodical statements of the amount of milk and dairy products that are brought in to certain large urban consuming centers such as Paris, Vienna, New York, Montreal, etc.

6. Finally and on the international side, reference should be made. to the information collected every quarter by the office of prices information of the Swiss Peasants' Union, concerning the position of the output of milk and dairy products in different countries. By arrangement with several official institutions and a large number of farmers' associations in different countries this body collects, in addition to information on the forage situation and the prices of milk products, figures showing the changes in milk production as compared with the figures for the corresponding periods in the previous year. These are expressed by means of the following index figures: 0=no change, -1 to -3, slight, marked, very marked decrease, +1 to +3, slight, marked, very marked increase.

THE POSSIBILITY OF DEVELOPMENT IN THE ORGANIZATION OF MILK

STATISTICS.

1. While, as suggested by the above summary, statistics on milk production and dairy produce are available for a certain number of countries, it must be recognized that in the majority of cases the figures appear at irregular intervals or are incomplete and obtained by different methods and are in any event insufficient to serve as a basis for an international statistical service such as the institute has provided for several years for a quite considerable range of crop prod

ucts.

The special difficulties that arise in connection with the question of a regular system of statistics for milk production explain the position of general inferiority of this particular section in comparison with other branches of agricultural statistics. We may here consider the possibility of its development along uniform lines.

Granted the importance of cow's milk and its derivatives, from the point of view of world production and consumption these particular products will for the sake of brevity be mainly considered. It should, however, be understood that whatever may be said with reference to cow's milk and its derivatives will be equally relevant for sheep's milk and goat's milk and their derivatives for all countries in which these products are of any considerable importance.

2. A knowledge of the amount of milk produced in each country may be reached either directly, by obtaining from the individual dairy farmers returns of the amount of milk produced by their herds for a given period and available for human consumption, either as such or after transformation into dairy products; or indirectly, through an estimate based on the number of milk cows in the country during the given period and the average amount of milk produced per head during such period. The first system is followed by some of the few countries which make a regular census of production; these include in the list of questions sent to the farmers for the purpose of the census a heading for the amount of milk produced on their farms. This system is the one that in general gives the most satisfactory results. It must not, however, be assumed that this form of direct information presents no special difficulties. As a matter of fact an exact registration of milk production can only be made for the more highly organized farms, whereas elsewhere the producers can merely provide an estimated and approximate figure. In this connection it is interesting to note that in the United States a certain number of farmers who reported dairy cows on the census date failed to report any milk production for the preceding year so that the production of their dairy cows had to be based on the average production per dairy cow as shown by the complete reports. But, speaking generally, producers should be in a position to give a sufficiently accurate estimate of the amount produced by their own animals. It is certain that direct returns give the surest basis for a knowledge of milk production; even if there are good reasons for not obtaining annual figures, this method should in any case be adopted when a census is taken.

Unfortunately the practice of making censuses at regular intervals has hitherto only been adopted by a very limited number of countries. The institute lays great stress on the general adoption of the system of making censuses of production as far as possible at uniform dates and along uniform lines. Such censuses should include milk production among other categories.

3. Apart from the direct returns, which, on account of the amount of labor and expense involved in a census taking, can as a rule be made only at relatively infrequent dates, estimates of milk production can also be made indirectly on the basis of the number of milk cows and the average yield of milk per cow. This method, at first

For the census of 1920 the amount of production thus estimated represented 12 per cent of the total.

sight resembles closely that followed very generally for the valuation of a particular crop production in which an estimate is made on the basis of the area sown and the average yield per acre for each zone. A closer study of the parallel brings out, however, the greater difficulties in practice of making estimates of milk production.

The growth of any particular crop and the condition of such crop just before the time of harvesting (or, in other words, the forecast of the yield per acre) are matters that within the limits of a determined area can be estimated with a reasonable degree of accuracy by one or more observers. The reporter, after a survey of his district and a careful examination of the fields where the crop is ripening, is in a position to form an opinion of the probable extent of the yield for the area. The estimate has an element of security by reason of the fact that it is based on what is actually visible and can be readily translated into figures by an expert. Moreover harvesting takes place once a year only and practically all at one time for a given area; hence a simple estimate made on a single occasion covers the whole of the production for the area and crop under consideration.

On the other hand, in the case of milk the question is one of a continuous product collected from day to day. Not only is the production distributed over a wider period, but also more diversified than is the case with crops. It is impossible to have a knowledge of the actual number of milk cows and still less of their average annual output on the same basis as, for example, for an area sown with wheat and for its average yield for a given section, by means of the direct observation of an expert. It is necessary (in the case of milk), in order to get really accurate knowledge, to make special inquiries of a far more complicated and difficult order.

4. It is necessary that every country should possess annual figures for the number of milk cows. As stated above, the institute has already considered this question and brought it to the notice of the governments.

The general assembly, in 1913, approved unanimously the following recommendations which were forwarded to all the adherent governments:

The general assembly is of opinion that both in the national and international interest it is absolutely necessary that each State should possess statistical returns for livestock. It hopes that in countries where such statistics either do not as yet exist or in which they are not organized in such a way as to be able to satisfy the requirements of a service of international information. the necessary steps may be taken as soon as possible to establish such statistics in accordance with the recommendations made by the institute.

The general assembly is of opinion that statistical returns for livestock should show as far as possible for each year the number of head of each kind and for each of the categories set out in the subjoined recommendation.

These recommendations were confirmed at the general assembly of 1920, which in addition expressly introduced into the classification of the various types of cattle, a special heading for milk cows, justifying the insertion by "the great importance attaching to a knowledge of the figures as an indispensable element in an estimate of milk production."

Acts de la Cinquième Assemblée Générale, Tome II. p. 292.

5. At the present time the countries which have annual statistics for cattle are the following (names of countries in which these statistics are not made with absolute regularity each year are printed in italics and a certain number of countries and colonies of quite minor importance are omitted):

Argentina, Algeria, Austria, Australia, Belgium, Bulgaria, Canada, Chile, China, Cuba, Denmark, Egypt, Esthonia, Finland, France, Germany, Great Britain and Ireland, Greece, India, Japan, Korea, Latvia, Luxemburg, Madagascar, Morocco, Netherlands, New Zealand, Norway, Philippines, Rumania, Siam, Southern Rhodesia, Spain, Sweden, Switzerland, Tunis, Uganda, Union of South Africa, United States.

The number of these countries is considerable, and among them are to be found some which are specially important as regards milk production and yield in dairy products. The number of countries which have adopted a special heading for milk cows in their classification is a good deal smaller," while others merely give a general figure for cows or an indication of the total head of the various kinds of cattle. The following table shows the type of classification employed in the countries referred to above.

Although a certain number of important countries already possess annual statistics of milk cows, there are still many in which this fundamental element for estimating milk production is lacking. It may be noted, however, that in recent years some progress has been made in this respect. Spain and Rumania have lately introduced a heading for milk cows into their annual statistics. Certain countries recently constituted, such as Czechoslovakia and Poland, have already taken into account the recommendations made by the institute, in connection with the classification adopted in their latest censuses of livestock, and it is to be hoped also in their subsequent annual statistics. The continual pressure brought to bear by the

It should be noted that the term "milk cow" should have in the statistics of all countries a uniform definition and significance so as to include all the individuals which for the year to which the figures refer either are actually giving, have given, or are likely to give milk either for direct consumption or for conversion into derivatives, excluding those whose milk is used solely for the nourishment of calves.

institute is tending to bring about a similar acceptance by all the governments.

6. As regards an estimate of the average annual production of milk by the milk cows of any particular country, the available material is still more scanty.

The average milk production per head of cattle in any given country is the result of a great variety of factors, some of which show relatively little variation from one year to another (e. g., breeds, management, etc.), while others (e. g., alimentation, health conditions, etc.), vary considerable. These factors bring about considerable differences in the average annual production for each individual cow in the same country and for the same individual from one year to another.

Ideally, in order to arrive at a sufficiently accurate annual estimate of average milk production for the total head of milk cattle in any country, it would be necessary:

(a) To ascertain approximately, in addition to the actual numerical distribution of the milk cows in the various districts of the country what is the proportion for each district of the different groups classified by breed, age, etc. Since these proportions are not likely, as a rule, to show any great changes from year to year, it would suffice that they should be established as a result of information obtained on the occasion of the census taking or by means of a special inquiry at relatively long intervals.

(b) To make a reckoning for each year for each district by means of estimates or partial returns of the average milk yield for the year in question for each group.

With this material available it would be possible to calculate the average production per head for each district and for the country as a whole.

7. Subject to the above, it must be recognized that at the present time the preliminary material required for annual statistics for milk production per head is insufficient in the case of almost all countries or is altogether absent. Among the very few examples of classification by breeds, mention may be made of an inquiry into the distribution of cattle according to breed made in the United States in 1920 by the Bureau of Markets and Crop Estimates, and the information obtained in connection with the inquiry into the dairy industry which was carried out in France in 1920. Inquiries into average annual yield per head are made in certain countries, where the dairy industry is most advanced, by the control societies, but although these societies have been greatly developed in recent years and their work may be of value for purposes of statistics, the actual number of cattle which come within the range of their observation is still relatively small. In addition, the fact that the controlled cattle are almost always selected and above the average from the point of view of production makes it dangerous to draw any general conclusions from the results obtained from the controlled cows as to the facts for the total number of milk cows in a particular country.

8. However, by making use of the actually available material and the facts already ascertained and completing them by means of an

Leaving, of course, out of account the amount of milk used for feeding calves.