248 ANNUAL REPORT SMITflSONIAN INSTITUTION, 1919. 
into account. The problem is whether the readier devitrification 
of a glass in the flame can be ascribed solely to the surface of the 
glass being exposed to a very high temperature, and so for a thin laj'^er 
reaching the right state of fluidity for crystallization or whether 
chemical action also plays a part, i. e., whether the hot gases of the 
flame act on the glass and assist the segregation of parts of it by dis- 
turbing the chemical equilibrium of the bases and acids of the glass. 
Attempts to get an exact reproduction of the behavior of a glass in a 
flame by exposure of thin pieces of the same glass to intense heat by 
radiation, have given results showing close similarity in some glasses 
and great differences in others. It would be somewhat out of place, 
and certainly tedious in a lecture such as this, to go into the details 
which seem to justify the statement that a survey of the relative be- 
haviors of a very large number of different types of glasses exposed to 
flames or to heat by radiation, and chemical examination of the prod- 
ucts, leads to the conclusion that water, and to a smaller extent carbon 
dioxide, do act chemically when many glasses are heated in flames, and 
that this action plays an important part in the initial stages at least 
of devitrification. As the most simple example, which I can choose, 
of marked difference in composition of a glassy body heated by radia- 
tion, or in a blowpipe flame, ordinary borax, Na20.2B203, may be 
taken. Heated in a muffle up to about 1,450° C. until much of it has 
volatilized, the residue may, according to the time of heating and the 
temperature, have a composition represented by anything between 
NaoO.SBoOg to Na20.15B203; but it has not been found possible, 
under any conditions, in a blowpipe flame, to get a residue from borax 
with the proportion of boric anhydride greater than is represented 
by about Na20.3B203. It is difficult to ascribe this difference solely 
to the effect of different temperatures. With some glasses, however, 
there is visible evidence of the disturbing influence of the hot gases 
of the flame. A glass containing barium oxide, which was heated 
and reheated many times by radiation of varying intensity, and which 
was most reluctant to show any signs of crystallization, became, in 
the blowpipe flame, or in a hydrogen flame, gray at once over its 
surface, and soon afterwards signs of crystallization were readily 
noticeable. The initial gray effect was seen under the microscope to 
be due to numerous very minute bubbles caused apparently by 
the rapid absorption and subsequent evolution of gases. As the 
question considered here is the influence of the hot gases of a 
flame to hasten devitrification, there is no occasion to discuss the 
well-known effect of an ordinary blowpipe flame to blacken glasses 
containing lead or similarly reducible metals, except to say that 
experiments show that in many instances the process of alternate 
reduction and oxidation which sometimes occurs when such jjlasses 
