These experiments, which were carried on in various forms, demanded as a natural supplement a study of the effects of repetition in relation to size.  This was the object of a series of tests which I carried on recently in the Harvard laboratory.  I constructed the following material:  60 sheets of Bristol board in folio size were covered with advertisements which were cut from magazines the size of the “Saturday Evening Post” and the “Ladies’ Home Journal.”  We used advertisements ranging from full-page to twelfth-page in size.  Every one of the 6 full-page advertisements which we used occurred only once, each of the 12 half-page advertisements was given 2 times, each of the fourth-page size, 4 times, each of the eighth-page size, 8 times, and each of the twelfth-page size, 12 times.  The repetitions were cut from 12 copies of the magazine number.  The same advertisement never occurred on the same page; every page, unless it was covered by a full-page advertisement, offered a combination of various announcements.  It is evident that by this arrangement every single advertisement occupied the same space, as the 8 times repeated eighth-page advertisement filled a full page too.  Thus no one of the 60 announcements which we used was spatially favored above another.

Thirty persons took part in the experiment.  Each one had to devote himself to the 60 pages in such a way that every page was looked at for exactly 20 seconds.  Between each two pages was a pause of 3 seconds, sufficient to allow one sheet to be laid aside and the next to be grasped.  In 23 minutes the whole series had been gone through, and immediately after that every one had to write down what he remembered, both the names of the firms and the article announced.  In the cases where only the name or only the article was correctly remembered, the result counted 1/2.  We found great individual differences, probably not only because the memory of the different persons was different, but also because they varied in the degree of interest with which they looked at such material.  The smallest number of reproductions was 18, of which 14 were only half remembered, that is, only the name or only the article, and as we counted these half reproductions 1/2, the memory-value for this person was counted 11.  The maximum reproduction was 46, of which 6 were half remembered.

If these calculated values are added and the sum divided by the number of participants, that is, 30, and this finally by the number of the advertisements shown, that is, 60, we obtain the average memory-value of a single advertisement.  The results showed that this was 0.44.  But our real interest referred to the distribution for the advertisements of different size.  If we make the same calculation, not for the totality of the advertisements but for those of a particular size, we find that the memory-value for the full-page advertisement was 0.33, for the 2 times repeated half-page advertisement, 0.30, for the 4