CEMENT. 476 more efficiently accomplished with the Portlands than with the natural cements. With the latter class there was a greater tendency in the liner particles to adhere to the coarser grains and to be carried down with them. The relative proportions of grains of different sizes found upon sifting were as follows: Brand. Class. Size of grain. >.0058 . 0050 .0034 . 0027 <.0027 Per cent. 11.2 12.9 19.3 2.8 14.1 33.6 12.5 Per cent. 3.8 4.7 5.7 Per cent. 9.1 10.4 ״ 7.9 Per cent. 6.6 11.4 8.2 9.1 12.5 23.8 17.5 Per cent. 69.3 60.6 58.9 88.1 65.1 42.6 70 Slag 1.9 6.4 Chemical analyses were made on sifted and winnowed material. Some earlier analyses showed differences in the composition of the grains of different sizes from one brand of cement then examined. Subsequently there were 67 complete analyses made on different sized grains from seven brands, the results of which showed substantially the same composition which was found in the material taken from the barrels. In this series of determinations 5 Portland cements were used. 1 slag cement, and 1 natural cement. SPECIFIC GRAVITY. The specific gravity determinations of loose material were made with a Schumann volumeter, in which the displacement of a quantiti^ of liquid shows the volume of the cement, benzine being the liquid used. The specific gravity of hydrated material in the cake was determined by means of a chemical balance, weighing the material in air and in water. Account was taken of the water absorbed when the cement was immersed, thus making the necessary correction for voids. The results on the loose material taken from the barrels were as follows: Brand. Class. Specific gravity. Remarks. 3.11 3.09 3.10 3.12 With plaster, 38 days after grinding. Without plaster, 38 days after grinding. 3.13 3.07 3.13 14 days after grinding. 3.08 3.11 3.04 Slag 2.74 2.83 2.87 composite. stone composite. 3.15 2.85 3.06 2.93 3.06 12 days after grinding. 3.03 3.12 2.94