The blank for this test was specially prepared, as no printed form with which we were familiar entirely met our needs. In many of the printed forms available the print was not so legible as desirable. Furthermore, and what was especially important for this work, they did not provide enough material for several successive trials with the same subject to avoid the danger of some memorizing.

The subject should feel that he is adding new material; there will then be no tendency for him to try to remember distinctive combinations in certain columns and the sums for these columns. In an effort to meet our need, digits were arranged in blocks of 100. Figure 19 reproduces full size a portion of one of the blanks. The type was 12-point. The digits 1 and 0 were not employed, and an effort was made to avoid combinations which added to 10. Each digit was used with approximately the same frequency in the material taken as a whole, as is seen in figure 19, the blocks being arranged in 10 columns, each containing 10 digits. A page was composed of 8 of these blocks. Six different pages of such material were prepared.

1 Simpson, Columbia Contributions to Education, No. 53, New York, 1912, p. 122.

The printed forms were set up on a linotype machine, each horizontal line of digits being a separate linotype slug. One hundred of these slugs were made up and numbered with a punch. They could thus be readily combined into an indefinite number of blocks of addition material. This method made a substantial saving in the cost of composition.

The procedure followed in the test was to begin adding the 10 vertical columns in the upper left-hand block of material, placing the sums below. When the 10 columns had been added, the subject added the same material, but in the horizontal lines, so a block of 100 digits in the form shown in figure 19 provided 20 lines of addition. The method of adding, and of entering the sums, is shown in figure 20. The subjects were told that they had 10 minutes for the addition. No clock or watch was visible and no warning signal given until the signal to stop. The time interval was measured with a stop-watch. The men were instructed to work as swiftly and accurately as possible. Equal emphasis, so nearly as may be judged, was placed on both factors. Particular attention was called to the most frequent types of errors in addition. The attention of the subject was also frequently directed to the ease with which numbers may be transposed in writing them down. The greatest care was taken to maintain silence in the room. The men worked on the honor principle, and without exception they worked in this adding test very silently and intensely (at least so far as could be judged from facial expressions and physical attitude).

There is no precedent at hand for using horizontal lines of digits in a test of adding efficiency. The method suggested itself as an economical expedient to minimize movements, such as shifting of paper and position. Preliminary trials with other subjects did not disclose any particular difficulties in adding digits arranged horizontally when the material was spaced as it is in figures 19 and 20. It must be noted that in printing from linotype slugs the horizontal columns always remain the same, a particular horizontal column being subject to shift from one block to another and to different position in a block of such digits, but the sum of it being always the same wherever it is. No information as to the method of makeup of the blanks was given the men. They never saw the forms except at the 10-minute test periods. They did not know how many different blanks existed, as these were designated from each other by a secret mark, which was the digit in the extreme lower left-hand corner. It is hardly likely that'under these circumstances they would note that the horizontal columns were unchanging and would try to memorize them.1

Immediately at the close of the 10-minute period of addition, the blank was turned face down on the table in front of the subject, in the same position as before the signal was given to begin the test. The subject then took out his watch, laid it on the table, and made a pulse count on himself for a period of 30 seconds. He recorded this count on the back of the addition blank, under the date with his name.2