XIV,2 Witt: Effect of Calcium Sulphate on Cement 297 
per cent. If less than 1.80 the set cannot be depended upon, 
and if greater than 2.10 the cement will not conform to Gov- 
ernment specifications. This area is shown by the darkly shaded 
rectangle INOF (fig. 2). 
At many plants the allowable variation in sulphuric anhy- 
dride is considerably greater than 0.3 per cent, and sufficient 
accuracy may often be obtained by measuring the gypsum; that 
is, by adding a certain volume to a given weight or volume of 
the clinker. This method was used formerly, and in a day’s 
run the maximum variation of sulphuric anhydride was 0.46 
per cent. By exercising greater care and stricter supervision, 
this was lowered to 0.30 per cent. With such small limits 
allowable, this was still not sufficiently accurate and other 
changes were made. It was found necessary to weigh both 
gypsum and clinker. Two small bins were made, each con- 
taining sufficient gypsum for a two-days’ run of the mill. The 
gypsum to fill one of them is thoroughly mixed, and sampled 
as it is placed in the bin, and then analyzed. The weight of 
gypsum added to the clinker is calculated from this analysis. 
While the first bin is being used, material for the second is 
prepared. By this method the maximum variation for a day’s 
run may be kept as low as 0.17 per cent. By comparing many 
analyses of the finished cement with the theoretical sulphuric 
anhydride content calculated from the mix, it has been found 
that on account of irregularities such as the sulphuric anhydride 
in the clinker, the loss of water in the gypsum during grinding, 
the loss of gypsum from dusting, and the like, there is prac- 
tically a constant variation which can be corrected in the orig- 
inal calculation of the mix. Taking this into account, the 
sulphuric anhydride content can be kept within a few hun- 
dredths per cent of the calculated value. 
THE EFFECT OF EXPOSURE TO THE AIR, AND OF HYDRATED LIME, IN 
PRESENCE OF CALCIUM SULPHATE 
A cement may contain sufficient gypsum to render it normal 
setting at the time of manufacture, but may lose the effect of 
the retarder when exposed to the air. Quick-setting cement 
may also become normal under the same conditions. The effect 
of air exposure on a number of the samples in Table III was 
investigated. It has been shown that the effect of the atmos- 
phere extends only a few millimeters below the surface,’ but 
conditions more severe than cement is ever subjected to during 
' Philip. Journ. Sei, Sec. A (1910), 5, 398. 
164001——5 
