138 REIBLING AND REYES. 
area fractured in tension. Under similar conditions, Portland 
cements which are undeniably weak in tension always attain a 
much lower strength in compression than those strong in tension. 
It would seem that discrepancies between the two values would 
not occur unless caused by contending influences which affected 
the conditions imposed by the tests in an unlike manner. It may 
be observed that such forces are at work when the values of the 
tension and compression tests fail to rise or fall coincidently. 
There are many who advocate the substitution of compres- 
sive tests for our standard requirements in tensile strength, but 
nothing would be gained by such a change. Tension tests are 
more useful in revealing the true character of metals than any 
other kind of mechanical test, and they should be equally valuable 
to cement users. However, no test becomes of value until its 
significance is thoroughly understood, and failures properly to 
interpret the results obtained from tension tests of cement have 
had more to do with the general lack of knowledge on the 
subject under discussion than errors and imperfections in the 
mechanics of the test itself. 
THE TYPICAL CURVE OF TENSILE STRENGTH. 
Consider, for instance, the typical curve of tensile strength 
of Portland cement which, as described by W. P. Taylor,® rises 
sharply up to about twenty-eight days, then falls slightly until 
from two to six months, then increases up to one or two years 
and then falls off again as brittleness develops. Taylor believes 
that the first drop in strength is— 
probably due to the different rates of hardening of the different con- 
stituents of the cement. The sulphates and aluminates harden very rapidly 
for a short period, and then soon fall to a very low value. The silicates, 
on the contrary, harden much more slowly, but their action progresses 
regularly with age. 
This idea he illustrates in a general manner by a diagram 
which is reproduced in fig. 28. 
Taylor failed to prove that the sulphates and aluminates in 
indurated Portland cement act as stated. The indurated alu- 
minates prepared by Schott (Table XXXVIII) after being sub- 
merged in water for three months retained great strength. 
Our attempts to associate quantitatively this drop in strength 
with either the alumina, iron oxide and alumina, or sulphate in 
Portland cement, failed. We also found that it was possible to 
% 4m. Soc. Testing Mat. (1908), 3, 413. 
