The bill already introduced in Congress for the taking of the thirteenth census is the first visible result. The bill has received a more careful study by the best statistical experts than was ever before given to a census law.

I believe it has but one serious defect: that is the provision for the appointment of the temporary clerical force after noncompetitive examinations rather than through the usual civil service method. The Director ought to be wholly relieved, during the progress of the tremendously difficult work of a decennial census, of all pressure for patronage.

We hope to see it enacted at the present session of Congress. This will be a year earlier than the usual date of census legislation heretofore. This additional year for preparation, by the complete organization of the working machinery, kept alive at its highest efficiency, means a good job, more deliberately and carefully done than was ever possible before.

There has been much speculation as to the margin of error in past censuses. It is customary to reassure the doubters by the statement that one error tends to offset and balance another, giving a net result sufficiently near the truth for all practical purposes. I have always been sceptical as to the soundness of this reasoning. At the next census we may make some discoveries that will be startling. At any rate, we are for the first time in a position where we can intelligently check one census with another.

The permanent Census Office has created a training school for government statisticians. Many of you will recall the remark of General Walker in 1896, that "the government which has spent millions and tens of millions in the collection, compilation, and publication of statistics, had never spent anything in training and preparing the men who should conduct the statistical work of the country." We have an army and a military academy to train men in military science, a navy and a naval academy to prepare men to conduct the service of the navy in war and in peace; but for the development of the science of statistics, the science whose light guides and directs the action of legislature in the shaping of policies that

are to determine the future of the nation, we have simply taken our chances, with a resultant waste which is appalling and an impairment of the validity and accuracy of official statistical data even more appalling.

The permanent Census Office is destined to be our statistical West Point. Already more than twoscore of the best census clerks have gone into other bureaus and departments of the government to engage in various statistical investigations in connection with administrative work,-to the Bureaus of Corporations, of Manufactures, of Forestry, of Labor, and various bureaus of the Department of Agriculture. These statisticians have received a severe and exacting training. Every table or statistical presentation prepared in the office undergoes a merciless criticism and analysis before it reaches the printer. It must be defended, proved, and justified, or be sent back for reconstruction. No university subjects a thesis to more drastic criticism. Most of the clerks who go out after such a training can be trusted to make the most of it in their new fields. I am proud of the record that censustrained clerks are already making in other branches of the government service.

Others have gone into important private posts, still others to State statistical bureaus, and others still to the colleges and universities.

One of the most notable recent developments in higher education is the introduction of the study of statistics as an adjunct to the courses in Political Economy. I know of half a dozen institutions where well-organized statistical courses exist, and no doubt there are more. Many of these university teachers of statistics maintain intimate relations with the Census Office. It is a source of pride to us that census reports are in regular use as text-books.

The relations of the university and the Census Office should be even more intimate than at present. What we now most need is the friendly criticism and suggestion of the trained men, who, through constant use of census reports, come to know their defects. The producers and the consumers of

statistics should be in constant sympathetic co-operation. Helpful suggestions have already come from the consumers. But there should be more of them. There should exist the feeling among the teachers of statistics and of statistical methods that the federal Census Office is not only intended for their use, but is in a large degree dependent upon them for the development of its work along right lines.

Especially should this relationship exist between the Census Office and the American Statistical Association, the one organization in this country whose members are vitally concerned in the work set for us to do. Statistics are your tools. Help us, by friendly criticism and constant suggestion, to keep your tools keen-edged and well tempered, to establish a true standard gauge for the accurate measurement and comparison of the wonderful and multifold conditions of American development.

This is the more important because the Census Office is rapidly coming to be recognized as the general information bureau of the government. The correspondence of the bureau, involving inquiry of one kind and another, is enormous. Fully one-half of it is referred to the census from some other bureau or department. This fact illustrates the confusion which exists in the public mind as to where to apply for statistical information from the government. The statistical work has heretofore been so divided up and the names of the bureaus have been so misleading that the public is utterly at a loss. Letters are sent hit or miss, and the red tape of circumlocution is appalling. For instance, we have a Bureau of Manufactures in the Department of Commerce and Labor and at the same time a Division of Manufactures in the Census Office. The latter takes a complete census of manufacturing industries every five years, and alone can answer the bulk of the inquiries which pour in upon the Bureau of Manufactures. We have a Bureau of Statistics in the Department of Commerce and Labor, which deals only with the statistics of foreign and domestic commerce, and gets thousands of letters which must be referred to the Census. To make confusion more confounded, we have another Bureau

of Statistics in the Department of Agriculture. This bureau is of course concerned only with agricultural investigations, and even in that restricted field it deals not with actual statistics, which only the Census Office compiles, but with yearly crop estimates, which, it is needless to add, are totally different from statistics.

There thus exists a veritable babel in the designations of the bureaus that handle government statistics. This confusion is the outcome of a gradual but disjointed and haphazard development, which was the natural consequence of the fact that prior to the establishment of the permanent Census Office there existed no general statistical bureau to which the accretions of statistical work could appropriately be attached. It is a situation almost grotesque, but one which will gradually right itself. I am glad to be able to inform you that the present Secretary of Commerce and Labor has the problem under consideration, and is about to take an important step for the simplification and clarification of the existing confusion. The law which established the department of Commerce and Labor recognized the fact that this department was to become the chief statistical department of the government. With that end in view it conferred upon the Secretary unusual powers for the consolidation and rearrangement of statistical work. I believe this power is about to be exercised by Secretary Straus in a wise and effective manner.

It is of the utmost importance, in my judgment, that a definite apportionment of the responsibility for the government official statistics should be made. I cannot overstate the need for it, as it is revealed in the daily routine of the Census Office. The daily correspondence confirms another remark of General Walker; i.e., "The American people are intensely and passionately devoted to statistics." They make more frequent use of them, perhaps, than any other people; and, of course, they subject them to every possible misuse. It is their too frequent habit to accept any figures presented in tabular form as "statistics," and to jump at the obvious conclusion. I regard it as no minor function of the permanent Census Office to act as a

check upon this vicious habit. It is immensely important that there shall be an official "hall mark" upon statistical publications: far more important than in the purchase of articles of gold or silver, the value of which is of interest only to a few individuals. It is a part of the duty of the Census Office to furnish the "hall mark" wherever it can and to decline to furnish it wherever it must. There should be some national criterion for all statistics which are labelled "official." This is a function of the permanent Census Office which, so far as I know, has not hitherto been suggested.

By the wise, impartial, and conscientious exercise of this function the standard of statistical accuracy will be materially advanced and the indiscriminate or perverted use of statistics, or of figures purporting to be statistics, greatly restricted. The trust thus imposed upon the Census Office is a grave responsibility. It would work an incalculable injury to the cause of statistical science if anything should happen to impair public confidence in the integrity and reliability of the census; and it is one of the best traditions of this office that its reports should point out and emphasize the limitations and sources of error in the statistics which it compiles, and thus guard against their misinterpretation.

Now a word as to the practical utility of the Census Office during the intercensal years. It makes for more exact statistics to concentrate the statistical work of the government, so far as practicable and wherever separated from administrative functions, in one central bureau, under one general supervision, and with a general unity of method. There must always be exceptions, as in the case of financial statistics, which the Treasury must compile. But what we may call the general, non-administrative statistics of the government-statistics collected primarily for the sake of the knowledge they give of general sociological and economic conditions-can always be handled to the best advantage in one office, whose business it is to do nothing else, which has no executive functions, which is under constant spur to do this one thing better and better, which can measure and test the results secured in one branch of statistical