By cooling a solution of table salt saturated at the ordinary temperature to -15°, I obtained first of all well-formed tabular (six-sided) crystals, which when warmed to the ordinary temperature disintegrated (with the separation of anhydrous sodium chloride), and then prismatic needles up to 20 mm. long were formed from the same solution. I have not yet investigated the reason of the difference in crystalline form. It is known (Mitscherlich) that NaI,2H2O also crystallises either in plates or prisms. Sodium bromide also crystallises with 2H2O at the ordinary temperature.
Notwithstanding the great simplicity (Chapter I., Note 96) of the observations on the formation of ice from solution, still even for sodium chloride they cannot yet be considered as sufficiently harmonious. According to Blagden and Raoult, the temperature of the formation of ice from a solution containing c grams of salt per 100 grams of water = -0·6c to c = 10, according to Rosetti = -0·649c to c = 8·7, according to De Coppet (to c = 10) = -0·55c - 0·006c2, according to Karsten (to c = 10) - 0·762c + 0·0084c2, and according to Guthrie a much lower figure. By taking Rosetti's figure and applying the rule given in Chapter I., Note 96 we obtain— i = 0·649 × 58·5 / 18·5 = 2·05. Pickering (1893) gives for c = 1 - 0·603, for c = 2 - 1·220; that is (c up to 2·7) about - (0·600 + 0·005c)c. The data for strong solutions are not less contradictory. Thus with 20 p.c. of salt, ice is formed at -14·4° according to Karsten, -17° according to Guthrie, -17·6° according to De Coppet.