PORTLAND CEMENT TESTING. 165 



Fig. 12. 



The foiirteen blows illustrated b}' fig. 11 are repeated and the final two 

 struck directlj' across the middle as shown by fig. 12. 



Feat briquettes made in this way always break across the center in 

 the Fairbanks roller clips, and seldom vaiy more than 10 per cent from 

 the highest (5 per cent from the mean). At times, batch after batch 

 will break within a few pounds. Again, at rarer intervals, an occasional 

 break occurs which is 20 per cent or more away from the normal. This 

 variation depends to a great extent upon the nature of the cement and 

 the consistency which the per cent of water used produces. Quick- 

 setting cements give the greatest variation in results. 



Sand briquettes still continue to differ considerably, as is true with all 

 other methods. The variation in the size and shape of the sand particles 

 and the corresponding voids and excesses of cement are such that it does 

 not seem piossible to contrive any method to eliminate the differences 

 in the tensile strength. The chief value of our system in this re- 

 sjDect is that it insures good breaks and hence gives more data to 

 report from. For the purposes of investigation and for work which 

 is under dispute, the question of variation in the force of the blows and 

 their application is eliminated by our method. It is also true that the 

 mechanical tamper renders it possible to depend upon the labor of 

 assistants. The variations in tamping having been eliminated, a smaller 

 number of breaks will suffice. We find that four briquettes from two 

 batches of mortar will almost invariably cover the entire range of pos- 

 sibilities, and show any undue variation in the quality of a series of 

 cement samples, this is illustrated Ijy Table XVIII which shows the 

 uniformity of the cement very plainly, despite the small number of breaks. 



71978 — -5 



