[STUART] THE CORROSION OF METAI.S BY WATER 185 



suspended by threads, completely immersed in 400 cc. of the water 

 to be examined. The bottles were of equal size and corked, with 

 equal air spaces at the top. One set of bottles stood seven days at 

 room temperature (about 22° C.) and the other set for a similar period 

 at 60° C. At the end of that time the wires were removed, adhering 

 rust and products of corrosion rubbed off with a cloth and then 

 weighed again. Loss in weight was recorded as a measure of corrosion. 

 The agreement ol3tained from duplicates was surprising. 



In order to ascertain the effect of corrosion of unlike metals in 

 contact several sets were taken, each comprising two metals of greatest 

 difference in potential. The wires were wound around each other, 

 then coiled and suspended as before. These constituted the "couples" 

 mentioned hereafter. 



First, a series of tests were made to find the necessary amount 

 of alum required to effectively decolorize the Ottawa River water. 

 Results showed at least 50 p. p.m. to be necessary, which by the way, 

 on a daily pumping supply of 20,000,000 gallons would amount to 

 5 tons a day. In the treated water no precipitate of alumina was 

 obtained with the usual reagents, indicating that the natural alkalin- 

 ity of this water is sufficient to entirely decompose all the sulphate of 

 aluminium used. In addition to water prepared in this way, samples 

 of raw and treated water were used which had been obtained from a 

 city waterworks in the United States employing this method of 

 purification. 



Conclusions. 



Corrosion of Iron in Raw and Treated Water. 



In the first experiment no appreciable differences were found 

 between the treated and the untreated waters as regards their action 

 on iron, in other words the loss in weight of iron wires, alone or in 

 couples, was essentially the same for both waters, under similar con- 

 ditions of time and temperature. The second series of experiments, 

 using Ottawa River water, treated and raw, gave results leading to 

 the same conclusion. Singularly the index of corrosivity was found 

 to be the same in all four samples. Although these figures are prac- 

 tically identical, there are extreme and important differences in the 

 character of these waters. Thus, it was found that raw waters although 

 corroding iron to the same extent, do not coat the iron with as much 

 flocculent material nor do they allow it to settle out, as is the case 

 with treated waters, but retain a great proportion in solution, pei- 

 haps in a colloidal state. Treated waters at ordinary temperatures 

 became very murky and opaque and, when hot, large amounts of 



Sec. Ill, 1913—12 



