In No. II. of this series of papers, measurements were given of the striking change shown by cobalt chloride when an aqueous solution was heated or was altered in concentration, and it was announced that a more accurate investigation was being made in the hope of finally determining its cause. Chemists suppose the cause to be the dehydration of the hexahydrate, but the matter is not beyond doubt.

First of all, aqueous solutions were used. Except in regard to the heating arrangements the methods and apparatus were those used in the previous research. The heater in this case was a rectangular copper casting 8 cm. long, 2·5 cm. broad, 4·5 cm. high, which at one end was attached to a tin in which water was boiled. At a distance 1·2 cm. from the other end a rectangular hole was cut down through the casting, of dimensions such that it was just filled by the cell containing the solution to be heated. At each side of the hole from the base upwards a part of the casting 1·6 cm. high and 1·1 cm. broad was cut away. This permitted the transmission of light horizontally through the lower part of the vertical cell, which was of the ordinary rectangular type, and obviated the use of mirrors. As in the previous research glasses were inserted in the cell to reduce the thickness of the layer of solution examined, and the temperature of the solution was read by a thermometer.