In the case of mixtures the specific sera only reacted with extracts of the flesh of the two animals in question. Thus a rabbit treated with an extract from a mixture of the flesh of a hare, cow, deer, and pig, yielded a serum giving a precipitate with the extracts of the flesh of each of those animals, but not with that from any other animal. It was not long before the possibility of using the method to distinguish between the blood of different kinds of animals suggested itself, and it was shown by Dr. de Nobel in 1902, that by treating a mouse or rabbit with any fluid, such as blood serum or saliva from a human body, it eventually produced a serum that would give a precipitate with human blood, but not with the blood of different species of animals. Reactions were also obtained with old human blood. Thus stains on linen from several days to two months in age, when treated with dilute solutions of common salt gave a solution which yielded a precipitate with the prepared rabbit's serum. No reaction was obtained, however, with the preparation from a blood-stain nine years old or with that from blood which had been dried in a high temperature. It was also found that the specific sera could be evaporated in a vacuum without losing their activity, and that the dried residues could be preserved in sealed tubes in the dark, and mixed with water when required for use. Other investigators showed that it was possible to separate the active agent by adding magnesium sulphate to the serum, and that the precipitate could be dried and kept for a long period. By dissolving it in water at any time a liquid with the specific properties of the original serum could then be obtained. Later work has shown that this serum test is not quite so absolute as was at first believed. Thus, if the blood serum to be tested be used in too concentrated a form it may give a reaction with a serum that is not specific to it, though even in that case the precipitate will only appear slowly and its amount will be insignificant in comparison with that obtained when the two liquids correspond. The error is obviated by using extremely dilute solutions for the test, and when proper precautions are taken a solution of normal blood serum containing one part in 1,000 invariably gives a reliable reaction with its corresponding prepared serum. In more concentrated solutions there is an abundant deposit at the bottom of the tube within thirty minutes, whereas in the case of sera, which are not specific to the prepared serum, the formation of precipitate does not begin until the tube has stood for an hour or more. [Illustration: FIBRES OF CHINESE SILKS, SHOWING CROSS SECTION] [Illustration: KANGAROO'S HAIR] [Illustration: HUMAN HAIR] [Illustration: A. Hair of a Cat B. Hair of a Dog _By kind permission of Messrs. Scott Greenwood & Co._]