228 
Proceedings of the Royal Society of Edinburgh. [Sess. 
bunsen burners were used in these tests — one an old form, which does not 
give a very high temperature ; the other the recent form, which gives 
much higher temperatures. 
The first thing to be done was to heat the tube very highly to cleanse 
it, filtered air being drawn through it to carry away the impurities. This 
air always gave very dense condensation ; the test-flask looked as if it 
were packed with cotton-wool. After the high-temperature flame had 
been applied for some time, moving it over some length of the tube, the 
low-temperature flame was used to see if it gave any nuclei at the lower 
temperature, as at the higher temperature all the tubes gave great quanti- 
ties of nuclei. At a low red heat some of the tubes were inactive, but all of 
them became active after the temperature was raised a little higher. It soon 
became evident that, if we simply observe the effect of the flame on the 
tube, we can get the information wanted without testing the air for nuclei. 
So long as the flame is unaltered by the presence of the tube, no nuclei are 
produced ; but when the temperature is such that the flame above the tube 
is different in appearance from what it is below, it is found that some 
action has taken place inside the tube which produces nuclei. Take, for 
instance, the glass tube : it is inactive till the temperature is high enough 
to show the sodium colouring in the flame ; whenever that appears, dense 
fogs are formed in the test-flask. In most cases the flame above the tube, 
when examined with a pocket spectroscope, showed an increase in the 
brightness of the D lines. 
There is, however, an interesting exception to this rule. The trans- 
parent silica tube produced very little change in the flame, and yet at a 
certain temperature it was an active nucleus-producer. Watching this tube 
as the temperature was raised, it was noticed that it had the appearance of 
cracking into pieces. Bright, shining facets appeared, which looked as if 
they were reflecting light, which, however, could not be the case, as there 
was none to reflect, the room being dark except for the pale light of the 
bunsen flame. These shining facets were self-luminous, and shone with the 
brightness of any small piece of opaque substance which happened to be 
in the tube for testing. It was noticed that so long as these bright cracks 
were absent the tube did not produce nuclei, but whenever they appeared 
nuclei also appeared in the test-flask, and the density of the condensation 
was roughly proportional to the extent of the facets. It may be mentioned 
that these crack-like facets, or at least some of them, appeared at the same 
places on successive heatings. One irregular crack seemed to pass nearly 
round the tube, which looked as if it would fall in pieces, but no change 
was noticed when the tube was cold. These bright cracks seem to be the 
