at a relatively early date, and clerks from States having a deficiency of appointments were assigned to work which would continue longer. In this way the force has been gradually, although not completely, readjusted so as to secure a distribution more nearly in accordance with the law of apportionment.

TABULATING MACHINES.

The tabulating machinery of the Bureau of the Census has been employed to its maximum capacity during the fiscal year and has proved eminently satisfactory. In the population division, as stated in my last annual report, the work of tabulation is almost wholly performed by means of punched cards, each individual inhabitant having a card punched with the facts concerning his characteristics. The punching of the population cards was completed in December, 1910. About two-thirds of the cards were punched on the new electric keyboard machines and about one-third on the old-style pantograph machines. Toward the close of the work, when the operators became expert, an average of about 2,000 cards per day per operator was reached on the electric machines and about 1,200 on the other machines. The system of piece-price payment upon these machines worked most satisfactorily.

The cards, after being punched, are sorted by electric sorting machines and tabulated by electric machines, part of which are fed by hand and designated as semi-automatic, and others fed automatically. The semi-automatic machines, of which the Bureau possesses 100, have been used for the greater part of the work, although increasing use has been made of the automatic machines. The semi-automatic machines proved even more rapid in operation than was anticipated. The speed which can be made depends upon the frequency with which readings of the results have to be made, and it is therefore greater for those runs of the cards in which the units of area or the groups into which the cards are divided are large. In the first run of the cards for regular tabulation purposes-a prior run having been made for purposes of verification only-the operators working on a piece-price basis averaged, for all classes of cards combined, 17,646 cards per full working day of seven hours. On the same run the piece-price operators handling the cards of native white persons of native parentagethe class in which the groups are usually largest-averaged 22,274 cards per day. The speed of these machines was therefore much more than double the speed obtained in 1900. This was partly due to the improvement of the machines, including the use of an automatic printing mechanism, and in part to the introduction of the piece-price method of payment. The piece-price rates differ for different classes of cards. For the native white cards of native parents on the first run

*

the rate of pay was 15 cents per thousand cards. On this basis operators were able to earn in the neighborhood of $1,000 per year on the average. The system of piece-price payment has proved highly satisfactory.

On the third run of the cards, which, like the first, includes the entire population of the United States, a rate of speed has been maintained substantially the same as on the first run. On the second and fourth runs, which deal with the foreign-born population exclusively, and in which readings have to be taken for relatively smaller numbers of cards, the speed has been naturally somewhat less.

Some

The Bureau of the Census now possesses seven automatic tabulating machines. These machines were designed and built in the Bureau. Mechanical difficulties encountered at the outset have now been largely remedied. Some difficulty still occurs from the fact that the automatic machines require cards of almost absolutely perfect dimensions and with holes punched in almost mathematically exact positions. defects in the manufacture of part of the cards and in the adjustments of part of the punching machines have interfered with the greatest success in the use of the automatic tabulating machines. If these difficulties can be remedied in the future, there is reason to believe that the automatic machines can advantageously be used exclusively, or almost exclusively. The possibilities of these machines are evidenced by the fact that a single machine has in actual operation turned out as many as 150,000 cards in a day, and on certain days the average for all of the machines in operation has run as high as 80,000 cards. These machines require about three operators for each two machines.

The sorting machines used by the Bureau of the Census for the sorting of population cards have also been highly satisfactory. The suit of the Tabulating Machine Co. against the Director of the Census, on the ground that the alteration of these machines was practically equivalent to the construction of new machines and an infringement of its patents, is still pending.

In the agricultural division of the Bureau of the Census the results have demonstrated the wisdom of the decision to use adding and listing machines instead of punched cards and electric tabulating machines. The adding and listing machines have proven in every way satisfactory, and the work of tabulating the agricultural census is proceeding with much less cost than at the census of 1900. On account of the wide variety of the tasks performed, the piece-price method of payment is impracticable in the agricultural work, but by comparisons between different clerks doing the same work adequate pressure has been brought to bear to maintain a proper standard of output. The same circumstance makes it impossible to present any condensed statement with regard to the average output per operator.

CENSUS FRAUDS.

At the time of my last annual report investigations were still going on with respect to frauds in the population census, and it was therefore deemed inexpedient to make any detailed reference to them. These investigations, so far as they directly affect the census statistics, have now been completed, although the Department of Justice is still pursuing inquiries for the purpose of prosecution.

The number of instances of fraud in the census of 1910 disclosed by the investigations made by the Bureau of the Census is greater than that disclosed at any previous census, but there is reason to believe that the actual amount of fraud was not greater, as the investigations were more thorough at this census than before. It was found that the population had been overcounted in 23 cities and towns, the number of enumeration districts affected being about 200. In some cases these overcounts were due less to deliberate fraud than to errors of judgment, but in many cases the fraud was of a serious character. The cities in which overcounts were found were as follows: Fort Smith, Ark.; San Francisco, Cal.; Boise, Idaho; Duluth and Minneapolis, Minn.; Billings, Great Falls, Missoula, and Havre, Mont.; Portland, Oreg.; Ogden, Utah; Aberdeen, Bellingham, Centralia, Everett, Hoquiam, Montesano, Seattle, Spokane, Tacoma, and Walla Walla, Wash.; Superior, Wis.; and Rawlins, Wyo.

The methods of fraud were numerous. The names improperly enumerated included names of purely fictitious persons; names of persons who were residents of the city, but who did not reside in the district in which fictitiously reported, and had already been enumerated at their true residence; names of former but not present residents of the city; names of persons who had no permanent residence in the city, but who were temporarily present and not entitled to enumeration; and names of persons who had never had a permanent residence and were not present at the time of enumeration, but who had been temporarily present at some previous time. In many cases the names of persons thus fraudulently enumerated were assigned as additional residents, usually under the guise of boarders and lodgers, of houses which had already been properly enumerated. In other cases they were assigned as residents of vacant lots or fictitious houses.

In a considerable proportion of the cities in which overcounting took place it was due largely to the unauthorized activity of private individuals in collecting names and turning them over to the enumerators. In a number of cities so-called census committees were constituted. The nominal purpose in all cases was merely to make sure that the enumeration was complete, and in some instances the activities of these committees were entirely legitimate and helpful to the census.

Even where the purposes of the committees were legitimate, however, the result was sometimes an overcount of the population through the failure of the enumerators to verify the information turned over to them by these private organizations. There is reason to believe that in some cities, at any rate, there was a deliberate design on the part of certain citizens or organizations, in conjunction with enumerators, and possibly in conjunction with supervisors and special agents, to pad the returns.

This Department has turned over to the Department of Justice the facts with regard to census frauds, and that Department, with the assistance of special agents of the Bureau of the Census, has been conducting careful investigations to develop the responsibility for the frauds and to secure prosecution of those guilty. A large number of enumerators have already been indicted. In a majority of these cases which have come up for trial the enumerators have pleaded guilty and been punished with fines and imprisonment; in several other cases convictions have been secured on trial. Only two enumerators indicted have been acquitted. Still other cases remain to be tried. One census supervisor and several special agents who acted as assistants to supervisors have also been indicted. Two private individuals were also indicted for conspiring with Government employees to secure fraudulent returns and, having pleaded nolo contendere, were heavily fined. I consider it of importance that all cases of census frauds should be prosecuted. The abuse is of a serious character, and it can not be wholly stopped except through a wholesome fear of the consequences. It is believed, however, that the publicity which has been given by the Bureau of the Census to the facts with regard to the padding of the population returns will itself have a powerfully deterrent effect upon attempts to commit similar frauds in the future.

APPROPRIATIONS.

In my annual report for 1910 I estimated that the total expense of the Bureau of the Census during the Thirteenth Census period, including the census work proper and the annual work of the Bureau, would amount to $14,500,000, and requested an appropriation of $2,500,000 in addition to the $12,000,000 theretofore appropriated. This appropriation was duly granted by Congress. It is now found, however, that the estimate of expense was too low. On account of the extraordinary changes made with respect to the questions in the schedules, the character of the machines employed, the methods of checking the accuracy of the returns, and the character and form of the tables prepared, it was quite impossible, from the experience of the census of 1900, to make any accurate estimate of the cost of the office work of the census of 1910. The estimate contained in my last

!

annual report was necessarily only approximate. It now appears that the total cost of the work during the Thirteenth Census period will be approximately $15,500,000, and I have therefore submitted to Congress an estimate for a further appropriation of $1,000,000 to complete the work up to June 30, 1912.

Should this appropriation be granted and the amount expended, the total cost of the Thirteenth Census work proper during the census period will be in the neighborhood of $13,800,000, the remaining $1,700,000 representing the cost of the current annual work of the Bureau. This estimated expense of $13,800,000 is all that can be compared with the expense of about $11,770,000 incurred by the Census Office during the Twelfth Census period, ended June 30, 1902. The increase in the total cost of the Thirteenth Census as compared with the Twelfth will therefore be about 17 per cent, whereas the population of the country has increased 21 per cent. So far as office work alone is concerned, the actual cost of the Thirteenth Census will probably be less than that of the Twelfth Census, notwithstanding the increase in population. The increase in the aggregate cost is wholly in the field work.

The approximate cost of the field work of the Twelfth Census was $4,925,000, and that of the Thirteenth Census $7,059,000, an increase of a little over 40 per cent. This increase was due to three principal causes-(1) the great increase in population and number of farms, and in magnitude of manufacturing establishments to be canvassed; (2) the necessity of paying higher rates to enumerators and supervisors by reason of the general increase of wages throughout the country, a necessity which was recognized in the census act itself, which fixed higher salaries for supervisors and a higher minimum rate of compensation for enumerators for collecting farm schedules; (3) the increase in the number of inquiries upon the schedules, in part due to positive enactments of law, the increase in territory required to be covered, and the addition by law of certain special investigations not covered by the census of 1900. The present census covered Hawaii and Porto Rico, which were not covered in 1900; it also included special investigations of irrigation; of institutions for the dependent, defective, and delinquent classes; and of slaughtering establishments, the latter for the purpose of ascertaining the number of animals slaughtered for food or for hides.

The actual cost of the office work of the Twelfth Census, including salaries, tabulating machinery, printing, and miscellaneous expenses, was about $6,845,000, while the estimated cost of the same items for the Thirteenth Census is slightly less, about $6,730,000. As a large part of the office expenditures of the present census have already been incurred, this estimate is a fairly close one, and it indicates that, despite the great increase in population and in the magnitude of the