In this way alone, M. Bertillon claims that 100,000 persons can be classified into groups of ten each, the portraits in which would offer no difficulty in examination. M. Bertillon undoubtedly puts the position too favourably here, in assuming division into equal groups; for out of his hypothetical 100,000 individuals, seventy-five per cent. might conceivably be tall men, and seventy-five per cent. of these, again have long heads, so that the final groups would in some cases have no representatives, while in the other groups there might be 1,000 individuals. In recording the colour of the eyes a special table is used, the scale of which is based upon the intensity of the pigment of the iris. A number corresponding to one of the following groups is then assigned:--(1) Iris, azure blue, with areola pale but free from yellow pigment; (2) Iris blue or slate, with light yellow areola; (3) Same shade, with larger areola approaching orange; (4) Iris, greenish reflection; hazel areola; (5) Same shade with dark hazel areola; (6) Hazel distributed over surface of iris; (7) Eye entirely hazel. When first the system was introduced into Paris it was a common practice for the old offenders to change their names and try to escape identification, but, according to M. Bertillon, after a few years this was only done by those who had been away from Paris for a long period, or had some very special reason for attempting to slip through the examination unrecognised. A similar method is employed in the United States Army for recognising deserters. Each man on joining is measured, and an outline figure card showing the measurements of the front and back surfaces, which are divided into areas by means of dotted lines, is filed in the Medical Department of the War Office. When a man deserts or is dismissed his card is placed in a separate file, and the new cards of recruits are compared with those in this particular file. A special register, ruled into columns corresponding to the areas on the cards, and giving the measurements and any peculiarities such as scars, tattoo marks, etc., is used to facilitate the search, and when, on reference to this, there appears to be a probability of a recruit being identical with a deserter, the original card is used for the comparison. During the first five months after the system was instituted (1891) sixty-two men were suspected of concealing their identity, and in sixty-one of these cases the suspicion was justified and the identity acknowledged. A drawback of the Bertillon system of identification is that much depends upon the accuracy of the person who takes the measurements, and that, therefore, a permissible error must be admitted. In the United States Army an error of one inch in either direction is allowed, for the recorded height. In addition to this, some degree of natural variation will take place in the course of years, and due allowance must also be made for this influence upon the measurements.