All that has preceded concerning hot tests refers to their use for testing Portland cements. Very little is known concerning the value of hot tests for natural cements. There are comparatively few natural cements that are absolutely bad, but to distinguish between the first and second quality of this variety of products is much more difficult than to make a similar distinction with Portlands. One point is certain, natural cements must not be expected to withstand boiling water. Mr. de Smedt experimented with fifteen brands of natural cement, and found that thirteen of them went to pieces in boiling water in two hours, although none of them was thought to contain caustic lime. Prof. Tetmajer has stated that for Roman cements, boiling water, or even 75° C, is not at all conclusive, and recommends 50° C. for trial, but our natural cements are not strictly comparable with Roman cements.

138. The author has experimented with three temperatures, namely, 50°, 60°, and 80° C, and is inclined to consider that 80° C. is likely to give the most useful information for sand mortar briquets but not for neat cement pastes. Table 41, § 227, gives the results of hot briquet tests on six brands of natural cement. It is seen that, with two parts sand, brands Jn, Hn, and Bn, give very low results at 80° C, and these brands are really inferior cements as shown by the two-year cold tests. Brand Jn is the only one that gave a lower result at seven days than at five days when tested at 80° C, and this brand failed entirely at two years, though it gives normal results in cold water up to six months. Neat cement pats of this brand, after being stored in cold water for nearly one year, were found to be cracked, although they had been perfect after one month in cold water. It was also found that neat cement pats of this brand warped and cracked in two days when placed in water of 60° C. when set.

1 " Tests Hydr. Materials," by H. LeChatelier. 2 Trans. Amer. Soc. C. E., Vol. xxvii, p. 438.

139. Conclusions

It may be said that although the limits within which the hot tests are reliable have not been well established, and although a strict adherence to them may at times reject a usable product, yet it is believed that sufficient experiments have been made to indicate that they are of much value, and should be made in all cases where the quality of the cement is of high importance.

The present indications seem to be that Portland cements may well be tested in the form of neat cement pats and sand mortar briquets at a temperature of about 80° C. Natural cements in the form of neat paste should not be called upon to resist a temperature above 60° C, but 80° C. will probably give the most useful information with sand mortars. In either case, the mortar should be allowed to set in moist air of ordinary temperature, then transferred to the vapor, to remain two or three hours before immersion in the hot water. It is not recommended that these hot tests should replace the ordinary cold tests, but simply that in cases where the extra work involved is not prohibitive, the hot tests should be made in connection with the cold tests.