36 The Philippine Journal of Science i9i8 



under the same conditions, the time and expense would be hardly 

 justified. 



2. Details which can be easily controlled by the investigator. 

 These include the 'kdoption of some definite plan for the work, 

 so that results obtained with different substances may be directly 

 comparable in so far as the concentration and purity of the solu- 

 tions employed are concerned; analysis and physical tests of all 

 cements as received ; the choice of a sufficient number of cements 

 so that results may be considered as averages. 



Chemical and physical tests of all cements investigated should 

 be made because the results may point out the constituent of 

 the cement that is most concerned in the observed results. The 

 conditions under which work is carried on, including the average 

 temperature, humidity, and laboratory methods, should be given 

 because they are of recognized importance. As a rule, solutions 

 are made on the percentage basis and are not afterward analyzed. 

 The results obtained are not directly comparable in some re- 

 spects. For instance, it usually happens that no account is taken 

 of atomic weights and valencies. An investigator may make 

 solutions of aluminium chloride and sodium chloride and oom- 

 pare results obtained by the same percentage concentrations 

 of each, and yet aluminium chloride contains a much higher 

 percentage of chlorine than does sodium chloride. Likewise 

 in comparing the relative effects of calcium chloride and calcium 

 sulphate the percentage basis is not the proper one. 



It is apparent that much progi'ess cannot be made in this 

 field of investigation until the results of various workers can 

 be connected by certain generalizations. The points of basic 

 importance are: 



1. Does the effect of an electrolyte depend on the negative ion, 

 the positive ion, or on both? 



2. Is the degree of solubility of the corresponding calcium 

 salt important? 



3. Can the effect of an untried electrolyte be predicted from 

 results obtained with others ? 



4. Can the observed effect be traced to any particular constit- 

 uent of the cement ? 



When 1 and 2 have been determined for a large number of 

 electrolytes, it should be possible to predict the effect of an un- 

 tried electrolyte — at least qualitatively — on many cements. 

 Number 4 is of great practical importance. Such data could 

 be used as a factor in determining the cement best suited for 

 construction that is to be exposed to certain conditions. This 

 has been already taken advantage of in the study of the effect 



