XIV, 2 Witt: Effect of Calcium Sulphate on Cement 225 
raised to bring the setting time within the required limits. A 
number of laboratory samples low in sulphuric anhydride had 
been studied, but no conclusions concerning the regular product 
could be drawn. It often has been found that when an electro- 
lyte is mixed with a cement there is a maximum point in the 
setting-time curve. When a given amount of the electrolyte is 
present, the cement will have the slowest set—either less or 
more of the substance causing a quicker set. The peculiar be- 
havior of the plant samples in requiring so much calcium sul- 
phate to produce a normal set and the results obtained with 
laboratory samples containing little or none of the retarder 
suggested the possibility that the optimum sulphuric anhydride 
content was at some point below 1.50 per cent. In that event, 
the difficulty in the manufacture was that the sulphuric anhy- 
dride was always above the optimum point. 
Arrangements were made by which the output of a tube mill 
_ for a number of hours was taken for the experiment. A quan- 
tity of clinker recently made and known to be very uniform 
was selected. Sufficient gypsum was mixed, sampled, and ana- 
lyzed. At 6 o’clock in the morning the mill was started; it was 
run twelve hours in the usual manner, keeping the sulphuric 
anhydride at approximately 2 per cent. The cement was test- 
ed repeatedly during the day and was found to be normal in 
every respect. At 6 in the afternoon all additions of gypsum 
were stopped and clinker alone was fed to the mill until 9 o’clock, 
after which gypsum was added as before. From 6 p. m. until 
1 a. m. samples of 4 or 5 kilograms each were taken every half 
hour, and at times every fifteen minutes. Physical tests were 
started at once, and the following day each sample was ana- 
lyzed for sulphuric anhydride. The results are shown in Table 
III. Samples 1 to 4 and 15 to 18 gave satisfactory tests for 
’ soundness. The others could not be tested because they were 
quick setting. The normal consistency of all samples tested 
was the same. . The fineness did not vary greatly during the run. 
The lowest point reached in sulphuric anhydride content was 
0.62 per cent. It would have been difficult to obtain a lower 
result than this unless the tests had continued a long time, 
because’ the cement was run through an air separator, and 
the cement returning to the mill always carried considerable 
sulphate. The test demonstrated that satisfactory cement could 
not be made with a small amount of the retarder, because only 
one sample having less than 1.50 per cent sulphuric anhydride: 
had a normal set. 
