EXPERIMENTAL CLOUD FORMATION 
against any great rise of temperature at the top of the 
chamber. In a series of experiments carried out by 
Chandra [4] at the Imperial College of Science and 
Technology, South Kensington, London, a chamber so 
constructed was used, and the top of the chamber 
was supported on a series of brass cylinders, sets of 
such cylinders being made of lengths from 2 to 16 mm. 
The sides of the chamber were filled in by layers of 
felt, sufficient to prevent a rapid leakage of air into or 
out of the chamber. 
In Chandra’s experiments the temperature of the 
air was measured by means of platinum resistance 
thermometers, fixed as near as possible to the top and 
bottom plates respectively, while a third was fixed 
halfway between the plates. Observations of the tem- 
perature distribution within the chamber could thus 
be made at any time. The motion within the chamber 
was made visible by means of cigarette smoke, which 
was injected by means of a two-way pump. The motion 
could be watched and photographed through the top 
plate. 
As might be inferred from Rayleigh’s formula, no 
motion was observed until the excess of temperature 
at the base over that at the top exceeded a finite 
limit. Thus, with a chamber of depth 10 mm, the 
critical temperature difference was 11.4C. Observations 
were made with depths from 2 to 16 mm. 
Figure 3 shows the structure found in a chamber 
of depth 8 mm, with a temperature difference of 28C 
i : : 0 |. 203 455 0M 
bel {mS 
SCALE Wie 8 
ns ae eo coo 
Fra. 3.—Convection pattern in a chamber of depth 8 mm; 
pomperatute difference between top and bottom of chamber, 
28C. 
between top and bottom of the chamber. This shows a 
series of polygonal cells, in each of which the motion 
was downward at the centre, upward at the outer 
margin, and inward at the top. In addition, there are 
some long rolls which fill the chamber when the smoke 
is first injected, and which gradually break up into 
separate polygonal cells. This feature was particularly 
noticeable in all chamber depths of 8 mm or more, 
but eventually the long rolls were replaced by poly- 
gonal cells. 
1257 
Figure 4 shows the structure found in a chamber 
of depth 7 mm, the whole chamber being filled with 
polygonal cells almost instantaneously after the in- 
: pO i ai 
Fie. 4.—Convection pattern in a chamber of depth 7 mm. 
jection of the cigarette smoke. The rapidity with which 
the initial long rolls or strips of smoke would break up 
into the cellular pattern in a chamber of depth 7 mm 
was very striking. In Fig. 4 the diagonal of the hex- 
agonal polygons was approximately three times the 
depth of the chamber, while in Fig. 3 the diagonals of 
the polygons were usually little greater than twice 
the depth, as the steady state had not been attained. 
With a chamber of depth 6 mm or less, polygons 
could not be formed with any vertical distribution of 
temperature. The change in the nature of the phenom- 
ena in going from 7 mm to 6 mm is best illustrated by 
Fig. 5, in which the base plate of the chamber was 
Fie. 5.—Convection pattern in a chamber of depth 6 mm. 
Where polygonal cells appear, a dent in the base plate yielded 
a depth of 7 mm. 
dented downwards over the area covered by polygonal 
cells in the diagram. Over this area the depth of the 
chamber was about 7 mm, and above the remainder of 
the plate the depth was 6 mm, when the photograph 
was taken. When the smoke is first injected into the 
chamber it forms long rolls such as still appear in the 
photograph at one side. Hach roll has a central white 
