PORTLAND CEMENT TESTING. 139 



Such a conclusion nia}^ he satisfactor}' to the engineer, but should the 

 tests be close to the margin specified for acceptance, the selling agent 

 is sure to protest and order a retest of the material. He may allow the 

 original tester to retest the cement, or he may send the samples to one 

 of the many commercial laboratories whose reputation for high results in 

 cement testing is well estal^lished. The retesting may produce satisfac- 

 tory results in either case owing to the weakness of all cement specifica- 

 tions. ''It is not to be inferred however that the highest results are neces- 

 sai'ily the outcome of the greatest skill. As a rule the most expert 

 and reliable operators get only moderate strength for the best material." " 



Such a condition of cement testing is very deplorable. Unless speci- 

 fications guarantee an acciiracy within 10 per cent, the greatest'efficiency 

 of a cement laboratory is also lost, as the mere mechanical routine testing' 

 of various brands of cement should be the least important part of its 

 work and satisfactorily to accomplish the more important object, namely 

 a systematic study of the peculiar effects of climatic conditions upon 

 them, a variation factor of not more than 10 per cent is essential. 



Sahin states that "the chief object of testing cement is to arrange the 

 various products in their true order of merit. Cement is at present used 

 in a very crude way and it is only in exceptional cases that poor quality 

 of material may be detected in the completed structure. This is suf- 

 ficient reason why so few failures can be found in cement work which 

 may be attributed to the poor qualit}^ of the cement. But in the more 

 economical manner in which the material is, even now, being used, it 

 is absolutely essential to know what its future behavior will be." " 



We believe that the inefficiency of all American specifications lies 

 in the fact that they do not outline sufbeiently in detail the minor con- 

 siderations and operations, and that to these minor details, owing to the 

 peculiar and sensitive character of cement, is readily attributed a 

 jDossiljle variation in the results of testing of 30 to 40 per cent. There 

 are certain qualities in cement manipulation that can not be controlled, 

 such as the size, shape and intermingling of crystals, nonhomogeneous 

 voids in sand briquettes, unequal action of the water upon the hardening 

 of lu'iquettes, etc., but we believe that by far the greater variation is 

 caused by the different manner in which different laboratories interpret 

 the minor details of manipulation and treatment; and we also believe 

 that if specifications were more explicit in this respect it would be safe 

 to predict that different laboratories would agree within 10 per cent. 

 This assumption is supported by the well-known fact that the system of 

 the individual laboratory usually produces fairly uniform results, but 

 a com|)arison between different laboratories which differ only in those 

 details not explicitly treated in the specifications, often shows the most 



"Spalding, Frederick C: Hydraulic Cement. New York (1904), IGO. 

 'Cement and Concrete. New York (1!)05), 82. 



