222 The Philippine Journal of Science 1919 
curves were plotted, but no two of these curves resembled each 
other. At the plant where this cement was manufactured 
similar irregularity was also noted, and considerable difficulty 
was experienced in controlling the addition of gypsum. Some 
samples taken from the conveyors, having a low sulphuric 
anhydride content, were normal in set, while others, with 
much higher content, had a flash set. In such cases the other 
physical tests showed no essential differences; nor were there 
sufficient differences in chemical analyses to account for the 
irregularity. It developed further that samples prepared in 
the laboratory by crushing the clinker and grinding it in a 
small ball mill did not have the same behavior as the cement 
which had been ground in a large mill at the plant.® 
Most of the work reported in the present paper was carried 
out to solve the problem of satisfactory gypsum control for 
the cement in a single plant and with no intention of publishing 
the results. To obtain the number of results necessary to ac- 
complish the desired purpose it was necessary to work with one 
cement alone, even though an investigation of several simulta- 
neously would have contributed more to the general subject. 
All physical tests have been made in accordance with United 
States Government specifications for Portland cement,® except 
that the ball method for normal consistency has been used, and 
no effort has been made to control room temperature (usually 
25° to 35° centigrade). The setting time as determined by the 
Gillmore needles is given to the nearest ten minutes, and any 
value less than five has been called zero. As a rule the final set 
has been omitted in the tables, because most of the irregularity 
has been noted with the initial set. The normal consistency is 
given only when it shows some characteristic of interest. Most 
of the samples have been analyzed individually, but a few of 
the values were calculated from the sulphuric anhydride content 
of the clinker and of the, gypsum. In Table I the samples 
shown as containing no sulphuric anhydride were not analyzed. 
No gypsum was added, but the clinker contained a trace. The 
tests were made at different times and by different operators, 
so that some slight variation may be due to personal equation. 
Samples that gave the most erratic results have purposely been 
* In this paper, “normal set” means a satisfactory initial set — usually 
two to four hours. 
“ For simplicity, samples ground from clinker in a laboratory mill 
will be called “laboratory samples” while those taken from the conveyors 
in the plant will be called “plant samples.” 
' No. 59 Cla, January 1, 1917. 
