26 The Philippine Journal of Science i9i8 



shown in a series of experiments in which concentrations as 

 high as 2 per cent of glucose — a condition not met with in prac- 

 tice — were employed without appreciably affecting the end point. 

 Since volumes may be read to 0.1 cubic centimeter with a tall 

 100 cubic centimeter cylinder, the error involved here is also 

 negligible. The error due to the excess of reagent present when 

 an end point is obtained may be reduced to 1 or 2 per cent by 

 making the excess small. The error due to lack of uniformity 

 in the pellets may be placed at a like figure. 



It is, therefore, reasonable to assume an error of about 5 

 per cent in using this method, an estimate that was verified 

 in the case of several natural and artificially prepared waters. 

 This degree of accuracy is ordinarily sufficient in questions of 

 potability or of the suitability of a water for technical purposes. 



No extensive comparison was made of results obtained from 

 this method and those from gravimetric determinations. This 

 was due to the fact that the first method was discontinued in 

 favor of the second, as soon as the latter had been shown to 

 be sufficiently accurate for field work. 



The second procedure possesses three important advantages 

 over the first: 



(1) It is more rapid. The solution of the pellets in the 

 first method requires considerable time. It was found that 

 from fifteen to twenty minutes were required for a determination 

 by the first method, while ten to twelve minutes sufficed for 

 the second. (The five minutes employed in aspiration in both 

 methods may be usually subtracted from the time of the analyst 

 required, as this may be performed by an unskilled attendant.) 



(2) Less reagent is needed. As only a part of the sample 

 taken is treated with potassium palmitate, the second method 

 will ordinarily require only a third or a fourth of the number 

 of pellets required in the first. 



(3) Several determinations may be rapidly made from the 

 same sample, thus avoiding any gross error that may occur in 

 a single determination. 



The error of the second method is, like that of the first, about 

 5 per cent. It is believed that the accuracy of the method could 

 be increased by machine-molding the pellets, thus securing 

 greater uniformity, and by the selection of a material more 

 suitable for binding and filling than glucose. 



In Table II are shown the results of a number of field deter- 

 minations of total hardness by the second method, compared 

 with the calculated values of the same from gravimetric analyses 

 of the calcium and magnesium contents. 



