ase such as lime, I am disposed to ascribe the above-mentioned
phenomenon to chemical rather than to purely physical causes.[23] It is
certain, however, that silica apparatus must never be too strongly
heated in contact with basic substances. Silica is easily attacked by
alkalis and by lime, less readily by copper oxide, and still less by
iron oxide.

[23] In a recent communication Professor Callendar tells me that the
devitrification commences at the outside and is hastened by particles of
foreign matter.

The rate of expansion of vitreous silica has been studied by H. le
Chatelier, and more recently by Callendar. The former found its mean
coefficient of expansion to be 0.0000007 between 0 deg. and 10000 deg.,[24] and
that it contracted when heated above 700 deg..

[24] The silica blocks used were prepared by fusion in an electric
furnace; it is therefore probable that they were not quite pure.

Professor Callendar used rods of silica prepared by the author from
"Brazil crystal"; these were drawn in the oxy-gas flame and had never
been heated in contact with solid foreign matter, so that they
consisted, presumably, of very pure silica. His results differ in some
respects from those obtained by Le Chatelier, for he finds the mean
coefficient of expansion to be only 0.00000059, _i.e._ about one
seventeenth as great as that of platinum. Callendar found the rods of
silica expanded very regularly up to 1000 deg. but less regularly above that
temperature. Above 1200 deg. they contracted when heated.

The behaviour of vitreous silica under sudden changes of temperature is
most remarkable. Large masses of it may be plunged suddenly when cold
into the oxy-gas flame, and tubes or rods at a white heat may be thrust
into cold water, or even into liquid air, with impunity. As a
consequence of this, it is in one respect much more easily worked in the
flame than any form of glass. Difficult joints can be thrust suddenly
into the flame, or removed from it, at any stage, and they may be heated
unequally in different parts with impunity. It is safe to say that
joints, etc., in silica never crack whilst one is making them nor during
the subsequent cooling. They may be set aside in an unfinished state and
taken up again without any precautions. Therefore it is possible for an
amateur to construct apparatus in silica which he would be quite unable
to produce from glass.

The behaviour of vitreous silica with solvents has not yet been