so 



TWENTY-FIRST REPORT. 



The computatiou pointed out a fact which should have been foreseen, but 

 was not, viz. : that the order in which the mixed size of advertisements were 

 seen by the subject exerted an influence upon the memorability of the adver- 

 tisement. The dummy already mentioned had not been prepared to take 

 accoimt of this feature, so in giving the results, three curves were prepared, 

 showing the obtained results and the maximum and minimum theoretical 

 results. This graph is appended : 



COMPARISON OF MEMORY EFFICIENCIES. 



Dotted line : maximum and minimum 



Fig. 2. Solid line : experimental results 

 theoretical results. 



a = 1 full, 



5 = 1 half — 3 quarter, 

 c = 2 half — 2 quarter, 

 d = 3 half — 1 quarter, 

 6=1 full — 3 uparter, 

 / = 1 full— 3 half. 



gr = 1 full — 1 half — 2 quarter, 

 »!= 1 full — 2 half — 1 quarter, 

 i = 2 full — 1 half — 1 quarter 

 y = 2 full — 2 quarter, 

 fc = 2 full— 2 half, 

 I = 3 full — 1 quarter, 



m = 3 full — 1 half. 

 It is obvious that if the idea in mind is a condition of attention, that a 

 large advertisement appearing first should create a more vivid idea than a 

 smaller one and hence should prepare the mind better for the following adver- 

 tisements of the same commodity. In our theoretical curves, this factor was 

 given an arbitrary \alue of 10%. To determine the actual value, three dum- 

 mies were prepared, all made up on the same general plan. In each dummy, 

 four firms advertised with each of the mixed sizes which were possible by a 

 combination of full, half and quarter pages, a total of twelve combinations. 

 In eacli case, half of the firms showed their largest advertisements first, fol- 

 lowed by the smaller sizes, and in the other half the smaller advertisements 

 appeared first, followed by the larger. Alternate subjects began the reading 

 with the first page and the last page of the dummy respectively. Approxi- 

 mately 500 subjects were used. The i-esults follow in tabular form : 



