168 C. J. WHITE. 
did not bear out these surmises, in fact the nearly colour- 
less glass (B) was found to have relatively slightly less 
manganese than either A or C. Still it is quite possible 
for one and the same quantity of iron to require different 
quantities of manganese to decolourise it (according to the 
state of oxidation of the two in the raw materials and the 
conditions of fusing) since the action of pyrolusite depends 
on two factors—(1) its oxidising action, (2) its power of 
forming a silicate complementary in colour to that of iron 
silicate. Also specimen B was of much finer quality than 
Cand hence it is probable that pains would be taken to 
secure such conditions in fusing as would be most con- 
ducive in producing a colourless glass. In commoner 
varieties, where colour is no object, it is not likely that 
such trouble would be taken, and it is possible for a 
specimen of glass to contain quite a large percentage of 
manganese and yet retain a green colour due to ferrous 
silicate if a reducing atmosphere is maintained throughout 
the fusing process. 
Also it is rather hard to see how manganese could be 
present as a mere impurity to the extent of ‘5 to 1°5*. 
According to Pliny the Romans employed sand and Hgyptian 
soda (from sea weeds) in glass making; the latter would 
account for the presence of iron and calcium as wellas the 
preponderance of soda over potash, but not for the man- 
ganese, and one would scarcely expect to find 1% of man- 
ganese in sand. However, there is not sufficient informa- 
tion available to definitely prove or disprove that manganese 
was purposely added, the question therefore remains an 
open one. 
Complete analyses of B and C were not made, iron and 
manganese only were estimated, these were obtained in 
solution by treating with hydrofluoric acid and then 
evaporating down with sulphuric acid. 
