(JOO REPORT— 1898 



of them, or 12 units in all : then I made another composite of A, B, and 

 the four others, allowing 4 units to A, 4 units to B, and 1 unit to each of 

 the four others, forming a total as before of 12 units. So while the com- 

 posite which I will call A 6, B 6, illustrates the combined features of the 

 two parents, that ofA4, B4, C1,D1,E1,F1 illustrates those of two 

 parents and four grandparents in the proportions laid down by the ances- 

 tral law. I proceeded similarly with C, D and with C, D and the other 

 four, and again with E, F and with E, F and the other four ; I submit 

 these six composites. Of course the process could be extended indefinitely, 

 working backwards to include as many previous generations of ancestors as 

 desired, and it might be equally well applied to portraits of other animals 

 than horses, including men and women, whose features combine unex- 

 pectedly well in composites, though one sex be bearded and the other not. 

 A composite may be made of any separate part of an animal, but hardly of 

 the whole animal at once, because each separate joint is liable to be flexed 

 differently in the different portraits. The ears of the horses in the illus- 

 tration indicate what would then occur. This is not the place to enter 

 further into the details of composite making, which I have now reduced 

 to a very simple process whose accuracy is evidenced by the identity of the 

 composites that have been re-made at different times from the same com- 

 ponents. The specimens I submit would have been better if they had 

 been made from the original photographs and not from photo-process 

 copies of them, still they will serve to gauge the amount of information 

 which composites are likely to give to the breeder. They should be care- 

 fully scrutinised and compared, when more differences and points of 

 interest will be found than are apparent at a first glance. 



Measurement of Photographs. — A photograph considered merely as a 

 portrait tells about as much of an animal as can be gathered from a 

 single view of it ; it defines the contour, the slope of the shoulders, the 

 set of the head, the forms and the positions of the limbs, but this is by no 

 means all that is obtainable from a photograph. It may be so taken that 

 measurements made upon tlie photograph, after certain corrections have 

 been applied to them, will be nearly as good as those made on the animal 

 itself. Now, measurements are of the highest importance to the theoretical 

 study of heredity, for science is based on numerical data, and the science 

 of heredity is no exception to the general rule. Its progress depends 

 primarily upon the power of procuring large collections of measurements of 

 the same parts, which admit of being combined in any proportions by simple 

 arithmetic. It matters little what limb, or bodily part, or faculty is 

 the subject of measurement, because laws which are true for one particular 

 quality, and for one particular race of animals or plants, will presumably 

 apply with small modifications to any other quality and race. Therefore 

 it would be no unworthy occupation for a scientific man to devote years 

 of labour to carefully measuring each of many parts in the photographs 

 of offspring and their ancestry, and to discuss the results by the elaborate 

 methods of the higher statistics. 



The photographs of which I speak are assumed to have been taken 

 under the following conditions. They would represent side views of 

 the animals and therefore be comparable on equal terms so far as 

 position is concerned. The animals would have been photographed at a 

 distance of not less than thirty feet from the camera, in order to avoid 

 sensible distortion of the portrait. They should be taken while standing 



