Senator TYSON. I can not agree with you on that. I think when a man has his crops planted, he knows what he has done; and we know that on the first of November lots of crops are being planted in the month all over the country. Especially in the South, we plant wheat in November.

Mr. OLSEN. How is he going to bear in mind the crop that he has planted in 1928 on which he has to report?

Senator TYSON. He knows whether he has put in 25 or 50 acres. In three of four months his memory will not be lost.

Mr. OLSEN. Nevertheless, our experience is that the farmers will forget.

Senator TYSON. I know they do, but

Mr. OLSEN. And they forget better the further the time is removed when you ask them about the facts; the further off they get the worse their reply is going to be.

Now, we also feel that you will have real difficulty in getting an adequate census of livestock, where you take it as late as May 1. The tendency and the desire has been to get the livestock census that would give us classes of livestock and also by age groups. It is very desirable to obtain information of that kind in order to determine what the likely increase in the production is going to be in the livestock field. Taking the census in the spring is going to give us quite a difference, possibly, in the livestock population, from what we would have if we took it in the fall, and it would not be at all comparable with the figures we have been getting in the past, and which have been so helpful to us in our work.

Then, in addition to that, we really need inventories of livestock at very nearly the close or beginning of the year, in order that we can appraise the total of livestock production. We have other means by which we can estimate the production of various livestock products, and with the inventories we can determine what was the livestock production in the year. If you move the date, you are going to increase the difficulty of and complicate our problem very seriously in the livestock field.

The CHAIRMAN. Will you go a little bit more into detail on that? What brings about this complication?

Mr. OLSEN. I am going to ask one of our men who is devoting his time to our crop and livestock estimates to develop that, but those are the principal points that we would like to stress in urging that the earlier date be taken, first in the interest of comparability of the census data, and secondly in the interest of an increased accuracy. We have every reason to believe that the census is not going to be as accurate if it is taken April 1 as it will be if taken at the earlier date, and it is of so much importance to us that we do feel very strongly that the earlier date should be used.

Now, to amplify what I have said on these various points, I would like to ask Mr. Becker to develop the facts with reference to livestock, and also the difficulties we would have in connection with some of the crops.

The CHAIRMAN. First, are there any further questions anyone would like to ask Mr. Olsen?

Senator VANDENBERG. I would like to ask him if, in a general way, it is not his opinion that agriculture does very largely itself favor the November date?

Mr. OLSEN. So far as I know, all farm organizations favor that date. I know the research people in the land-grant colleges are very strong for the earlier date. Various research committees and organizations over the country have expressed themselves the same way. Some time ago we had the question up in the world agricultural census. One of our men is the Director General of the Census; and the committee that has that work in charge decided on, I think, January 1 as their date, for similar reasons comparable to those we

urge.

Furthermore, we think that is also an additional reason why our census should be taken as of dates that would make the results comparable with the results of the world agricultural census.

Senator BURTON. Is there any possibility of taking the population census at a time separate from the agricultural census?

Mr. OLSEN. In case it is found desirable to take the population census at some other time, for other reasons, I see no reason why the two should not be separated. We think there would be some advantage in taking the population census and the agricultural census at the same time. It might add somewhat to the completeness and the accuracy of an agricultural census if the population census were taken at the same time, in that additional tests would be made upon all the returns.

Senator LA FOLLETTE. It would greatly add to the expense, would it not, to take them at separate times?

Mr. OLSEN. I think it has been estimated by the Bureau of the Census that it might cost an additional $2,000,000 or somewhere in that neighborhood, if the censuses were taken at different dates.

Senator JOHNSON. It would result in duplication of the work, too, would it not?

Senator TYSON. There is bound to be some additional expense and duplication if it would cost $2,000,000 additional.

Mr. OLSEN. Necessarily, there would be some duplication.

Senator BURTON. Is it provided that the agricultural census if taken in November shall be taken entirely by the same enumerators as the population census?

Mr. OLSEN. I could not say. It is my understanding that the same organization would do the work; yes, sir.

Senator BURTON. For the same identical purpose?

Mr. STEUART. Yes.

Senator JOHNSON. May we say in a way that those who are charged with the administration of the law, and those who are interested in it, with substantial unanimity have agreed that November 1 is the appropriate date on which to begin the census?

Mr. OLSEN. That is certainly true as to the Department of Agriculture, and I understood the Director of the Census to say the same. Senator JOHNSON. The farm organizations and agricultural colleges alike are of the same opinion?

Mr. OLSEN. That is correct. So far as I know, the sentiment in that direction is just about unanimous. I have heard of no farm organization and of no land-grant college or agricultural economist that has taken any other position than this one.

A reference was made to the distribution of the farm population at the two dates we were discussing, from the point of view of the population census. We have made two or three surveys of persons

engaged in agriculture, to ascertain what is the distribution of the rural people month by month. We do not claim 100 per cent accuracy for these figures, but we think they may be helpful, to indicate the general status of the rural population in the various months. I have a chart here with presents the results of those surveys. This chart shows the monthly change in the number of persons engaged in agriculture, and the percentage of change as compared with January. January, in other words, is taken as the base. These lines are merely indices of your level of population on May 1-your rural population, as you will observe from the chart. Take, for example, this solid line. That represents 1927. That is 110, or a little better than 10 per cent over the January figure. If you will compare it with November 1, you will notice that the rural population on that date is slightly higher than it was in May—that is, in 1927.

The CHAIRMAN. That is, there is a greater rural population?

Mr. OLSEN. A greater rural population, according to that, November 1, than there was on May 1.

Senator JOHNSON. What is your peak?

Mr. OLSEN. The peak is October. The top line [indicating on chart] is the one I am referring to here for 1927.

In 1926 we had another survey, and there, again, the rural population is slightly higher than it was in May.

The CHAIRMAN. At what time?

Mr. OLSEN. November 1-it is slightly higher than it was in May, in 1917, which is the dotted line. November 1 the rural population is slightly lower than it was in May; but in the last, you will notice that from the point of view of numbers out in the country, November 1 shows as many if not more people than May 1.

The CHAIRMAN. That is very interesting, because it was one of the main objections I have had, coming from an agricultural State that has a large rural population. The boys want to be in the cities the 1st of November, while they would be on the farms the 1st of May.

Senator TYSON. That is exactly what has been represented to me, Mr. Chairman.

The CHAIRMAN. And yet what you get in these figures is that there are more on the farms the 1st of November than the 1st of May.

Mr. OLSEN. We have the same results in these three different surveys we have made. I can not say, offhand, whether we have additional figures to substantiate that.

Senator RANSDELL. In the cotton country we have more of the boys on the farms the 1st of November, getting the cotton out, than the 1st of May. There is not nearly as much for them to do, down in my country, on the 1st of November, as there is the 1st of May. That is the harvesting time.

Senator VANDENBERG. That is what you use the Republicans for, is it not to get the cotton crop in?

Senator RANSDELL. We use them for anything we can; but they are pretty shrewd fellows, and we can not do much with them. [Laughter.]

The CHAIRMAN. How much of the country did that cover?
Mr. OLSEN. The entire country.

STATEMENT OF JOSEPH A. BECKER

Mr. BECKER. May I correct one thing, Mr. Olsen said? You referred to this as farm population. These are persons working on farms, which does not include the children or the housewives and so on and people in the houses. These are farm workers.

The CHAIRMAN. The point is, that covers the country? Mr. BECKER. Take the 1927 figures for cotton. We asked a question on the first of each month. This is not a survey made the first of the year, for the year. This is the question we asked our crop reporters.

The CHAIRMAN. Were those crop reporters scattered all over the country, or did you take them in some particular locations?

Mr. BECKER. Scattered all over the country, and in the sum total, they ranged from 11,000 to about 13,000 farms, and the average number per farm runs around three, so that it would be three times that many, or from 33,000 to 39,000 persons, represented by this monthly sample.

The CHAIRMAN. Taken from all sections of the country?

Mr. BECKER. Yes.

Senator TYSON. And from every State in the Union?

Mr. CALLANDER. One from every township.

The CHAIRMAN. I hope that gets in the record; one from every township.

If there is anything else on that, Mr. Becker, will you tell us about that stock proposition which has been referred to?

Mr. BECKER. If you do not mind, Mr. Chairman, I would like to use this chart, as long as it is up in connection with this question of memory bias.

The CHAIRMAN. Certainly.

Mr. BECKER. The particular purpoes of this survey was to point out changes in the people working on the farms from month to month.

The CHAIRMAN. That is the chart already referred to?

Mr. BECKER. When we started these surveys back in October, 1925, I believe it was, we started by asking these individual farmers the number of persons on their farms this month and last month, to get the change from one month to another. The memory bias was so large in even that short time of one month that if we took and linked those together at the end of the period year you would run off of this chart. In other words, they would forget the people they had a month ago. The result was that they always showed more people when asked at a particular time than they would report when asked to report a month later for the same time.

This is constructed not on the basis of a figure that includes the memory bias, but on the reports made each month and then worked back to the average number per farm.

Senator JOHNSON. How do you know that they ever gave it correctly if there is such a discrepancy in the statements they make? How do you know there is accuracy in any statement?

Mr. BECKER. If it comes out accurately at the end of the year with the beginning of the year. But if it comes out twice as many

Senator JOHNSON. But at the end of the year and the beginning of the year they would still have a memory bias, would they not?

How can you depend upon such a thing; if there is such a tremendous difference as you have indicated in their recollection during the elapsing of only 30 days? You know, I am not criticizing you: I am asking you for information. I can not see that.

Mr. BECKER. In one case we ask, "How many are on the farm now?" Then we ask him again a month from now.

Senator JOHNSON. And their statements would vary?

Mr. BECKER. No, sir; those figures give us this chart

Senator JOHNSON. Wait a moment. You say you ask them to-day, "How many people are there on your farm?" They say "seven. Mr. BECKER. That is right.

Senator JOHNSON. You ask them next month, and they say "17." Is that correct?

Mr. BECKER. No, sir.

Senator JOHNSON. You asked them how many there were a month ago, and they tell you a different story; is that correct?

Mr. BECKER. If we ask them on the 1st of April, "How many persons are on your farm?" and they say that the number averages exactly three persons, then we come along to the 1st of May and ask them again, and their average for this month we will say is 3.1 per cent. That is an increase of one-tenth of 1 per cent in a month. But if on May 1 we asked him also how many they had in April, that figure will drop to 2.8, and the difference between 2.8, which is a memory figure, and the 3.1, which relates to the immediate present, varies about 10 per cent from the first figure.

Senator JOHNSON. Now, if you cumulated that

Mr. BECKER. I would be rather skeptical of the individual person's memory, even if he was telling you how many were in his house at the time.

The CHAIRMAN. I understood you asked them in April and they would say there were so many on their farm; but when you asked them in May how many were on the farm in April, they would give you a different figure.

Mr. BECKER. Yes; they would have forgotten some of them.

The CHAIRMAN. And when you asked them a month later, relying on memory they made a different statement then from what they had made at the time, relying on what the fact was?

Mr. BECKER. Yes. I wanted just to amplify what Mr. Olsen said with respect to the confusion which would be caused in a census taken on May 1 because a new crop would then be in course of being planted. The farmer will be asked the number of acres of winter wheat, and while it is true that in the schedule we will ask for the acreage of winter wheat planted in 1929 and harvested in 1929, there is always the danger that he is confusing it with the last crop. There is danger that he will confuse it, and there is more tendency there for confusion, particularly as the enumerator will be somewhat hurried, and will shorten his inquiries as much as possible.

Of course, in a State like Texas, on the 1st of May the new cropin other words, the 1930 crop-of winter wheat will be practically ready to cut, and we will be asking him at that time for statistics of a wheat crop which he harvested nearly a year ago. I think that answers the question the Senator has asked.