170 REIBLING AND REYES. 
As the free lime in the commercial cements could not be elimi- 
nated entirely without greatly reducing the available strength, 
exact measurements of the loss of strength due to free lime were 
impossible. However, contrary to the general belief, our results 
indicated that a normal Portland cement which contained no free 
lime would contract in either air or water. For example, sample 
2, Table L, had been aérated as ground cement until it con- 
tained no free lime which would react with the phenolate solution. 
It hardened with considerable rapidity in water, the 7-day bri- 
quettes showed 482 and 208, and the 28-day, 612 and 329 pounds 
for neat and 1:3 sand mortars, respectively. During the same 
period of time and for at least a month longer both the neat and 
sand mortars continued gradually to contract. Therefore, it 
seems evident that as long as the free lime remained inert the 
cement continued both to harden and to contract. 
The fact that sound cements even when fresh expand in water 
offers no contradiction to this belief, because the soundness of 
the cement does not prove that free lime is absent. On the other 
hand all results show that the more perfect the burning of the 
raw materials into compounds of lime, silica, alumina, and iron, 
the less will be the expansion of the nonseasoned cement in water. 
Also, this belief is not contradicted by the established fact that 
Portland cement mortars and concretes exposed to alternating 
wet and dry conditions show corresponding expansion and con- 
traction, because the more rapid slaking of the free lime when the 
material is wet accounts for the changes in volume. 
Effect of changes in volume on the strength.—From the action 
of the nonseasoned samples in Table L and fig. 26, it is evident 
that hard-burned Portland cements may show a constant gain 
in strength in spite of considerable expansion. However, if this 
expansion is due to the latent slaking of free lime and this action 
contributes nothing to the strength of the cement, it must tend 
to weaken the indurated material. It seems logical to assume 
that within the limits of perfect elasticity the internal stress due 
to the slaking of free lime should manifest itself more in the 
determinations of tensile than of compressive strength. 
To test this assumption sample 4, Table L, was aérated as 
ground cement and tested after three and seven days’ exposure. 
Table LI records the results obtained. 
