90 



SCIENCE. 



[Vol. vm, No. 183 



tives were so adjusted that the pupils of the eyes 

 in each case fell upon the same points of the 

 sensitive plate, while in fig. 1 the distance from 

 the line of the eyes to the mouth was made con- 

 stant. 



The result of these different modes of adjustment 

 is apparent in the multiple mouth which disfigures 

 fig. 2, and in the less clear definition of the eyes 

 in fig. 1, in which the component eyes fell upon 

 slightly differing points in the same horizontal 

 line. 



The question at once arises, which of these 

 faces, if either, in its general outline and expres- 

 sion, is the true average of the group V In seek- 

 . ing the typical features should we choose fig. 1, 

 and correct the dimness of the eyes, or take fig. 

 .2, and substitute a single mouth in the mid- 

 dle of the blur? As far as I can learn, this 

 question of adjustment and its results has not 

 before been raised. It is, liowever, a question of 

 importance to all who are interested in composite 

 photography ; for only those composites which 

 are made according to the same method of adjust- 

 ment can be properly compared as types. 



In any group of persons not chosen with 

 ■special reference to facial symmetry, the ratio of 

 the distance between the pupils of the eyes to 

 that between the line of the eyes and the mouth 

 is a variable one ; and adjustment to either dis- 

 tance as a constant for the group will give its 

 corresponding and differing composite. Mr. Gal- 

 ton makes the distance from eyes to mouth con- 

 stant (' Inquiries into human faculty,' p. 359). The 

 portraits of American men of science {Science, v. 

 No. 118) seem (from the tendency to multiple 

 mouths and noses, especially noticeable in fig. 1) 

 to have been made, as fig. 2 was, by matching the 

 eyes, though in these cases the beard prevents the 

 prominence of the disfigurement which this adjust- 

 ment gives in the case of smooth faces. 



If a fixed distance between eyes and mouth be 

 taken for adjustment, the composite will have a 

 single distinct mouth, but will differ in form ac- 

 cording to the distance chosen; if it be that of 

 the shortest or of the longest face in the group, 

 the composite face will be correspondingly short 

 or long, and the indistinctness of the eyes at a 

 maximum. But if, on the other hand, a com- 

 ponent face of average length (i.e., one in which 

 the ratio of the distance between the pupils of the 

 eyes to that between the line of the eyes and the 

 mouth is a mean one) be chosen, the i-esultant 

 portrait will show a minimum indistinctness of 

 eyes, and give what we may fairly call the jjic- 

 torial average of the group. The average ratio 

 which must serve for fixing the fiducial lines can 

 be obtained from dii-ect measurements on the 



negatives. This will not be a formidable task, if, 

 as is usual, the negatives are taken so that the 

 distance between the pupils is the same in all ; 

 since in this case it is only necessary to measure 

 the distance from eyes to mouth in each, and 

 take the mean. 



This point is one which should be carefully 

 attended to in making composites, for it would 

 seem to be the only normal method of adjust- 

 ment ; all other adjustments giving more or less 

 pronounced variants from the type. 



Composites made in this way lose something 

 of the deep-eyed, earnest expression, which is 

 the result of superposing all the eyes of the 

 components on exactly the same points. This 

 loss, however, is a real gain in the truthfulness 

 of the composite portrait, for the deep 

 dark eyes do not represent the average, but 

 rather a summation, and hence exaggeration of 

 earnest expression. The face in fig. 1 is, I be- 

 lieve, a fairly normal composite of the group of 

 forty-nine from which it was made ; fig. 4 is 

 from a group selected for facial symmetry, i.e., 

 constancy of the ratio indicated, and is a type of 

 this group with the exaggeration which comes 

 from superposition of the eyes. Questions as to 

 the possible dependence of the result on the order 

 in which the components are taken, and on the 

 time given to each exposure, occur to every one 

 who interests himself in composite photography. 

 In Mr. Galton's earliest paper on the subject, he 

 speaks of six composites made from the same 

 three components taken in their six possible com- 

 binations, and says, " It will be observed that 

 four at least of the six composites are closely 

 alike, ... the last of the three components 

 was always allowed a longer exposure than the 

 second, and the second than the first, but it is 

 found better to allow an equal time to all of 

 them. In a later experiment, composites were 

 made of four differently colored disks, whose 

 images were superposed in four different orders, 

 while the times of the successive exposures were 

 equal. The result was four composite disks ' of 

 precisely uniform tint.' The inference from this 

 is, of course, that the order of exposure makes 

 no difference ^^•hen the times of exposui-e are 

 equal (equal illumination of the image is as- 

 sumed). The experiments which I have made on 

 this point by taking composite portraits from the 

 same components in different orders (with equal 

 times of exposure) have shown that the order of 

 exposure does affect the result. I have also re- 

 peated Galton's other experiment in several modi- 

 fied forms, both with disks of colored paper and 

 with colored glasses (by transmitted light), and 

 obtained results which, especially in the case of the 



