58 Scientific Proceedings, Royal Dublin Society. 



convenient to allow 75 sees, to elapse before making a reading. During 

 this time the stirring of the test-tubes is actively kept up ; for it is surprising 

 how quickly the ice rising in the salt solution allows the lower layers round 

 the junction to become supercooled. In the other test-tube the same does not 

 occur, as the ice soon forms a lining lying against the wall of tlie test-tube, 

 and the junction is supported in water surrounded by ice. When the first 

 reading is made, the clip is disconnected and the galvanometer mirror swings 

 free. Keversed connection is made when the spot of light is at the limit of 

 its swing on the side on which the first deflection was recorded. In this way 

 the suspension of the galvanometer is kept from any sudden strain which 

 might be produced by suddenly checking its movement. After 75 sees., 

 during which tlie same active stirring is kept up, a second reading is made. 

 This first observation after putting in tlie solution should be regarded merely 

 as a preliminary one ; but still if too much ice has not been present, it will give 

 tlie freezing-point within a couple of hundredths of a degree. 



The test-tubes are now raised from the freezing- chamber, and the one 

 containing the solution momentarily touched by the finger to give it a little 

 heat. When the upper cork is readjusted and stirring recommenced, it will 

 be noticed, that the spot of light retires towards zero. If all the ice is not 

 melted, it will quickly recover its former position; and the test-tube should 

 again be touched. When it is certain that all the ice is melted in the solution, 

 it is left in the freezing-chamber and allowed to cool. Meanwhile connection 

 is broken by removing the clip from the support in the petroleum key. 



When it is judged that radiation has cooled the solution nearly to its 

 freezing-point, connection is again made by the clip, and stirring is recom- 

 menced. The spot of light then travels to near its previous resting-place, or 

 possibly beyond it. Supercooling may proceed, and the spot of light will 

 slowly travel indefinitely beyond its previous position, or crystallization may 

 supervene, and the spot will return somewhat on its path and tend to take up 

 a steady position. In the latter case connection is broken at the clip, and the 

 mirror allowed to swing free. Connection is again made, and, after 75 sees., 

 during which vigorous stirring is kept up, a reading is made. The 

 cmTent is then reversed, and, at the end of 75 sees., another reading is made 

 on the other side of the zero point. If, however, supercooling proceeds, and 

 crystallization does not automatically occur, it is necessary to inoculate the 

 solution with a little hoar-frost. The inoculation should be carried out when 

 the spot of light has definitely passed the limit of the first deflection. If it 

 is allowed to cool too much, too much ice will separate, and the concentration 

 of the solution left over will cause too large a depression ; if, on the other hand, 

 it is inoculated just at its freezing-point, so little ice separates that the solution 



