EXPERIMENTAL ANALOGIES TO ATMOSPHERIC MOTIONS 
fruitful there must be a systematic and sustained effort, 
guided by theory, to obtain numerical measurements of 
as great a variety as is practically possible. The rather 
sterile, though often remarkably interesting, results of 
the large majority of these investigations can probably 
be traced to the qualitative character of the observa- 
tions more than to any other factor. Another basic 
difficulty with applications of these experiments to the 
atmosphere has been the lack of simple, theoretical, 
guiding principles which take into account the thin- 
ness of the atmosphere. Experiments need to be 
designed around principles, even if only semi-quantita- 
tively. In the experiments at the University of Chicago, 
for example, various applications of the vorticity the- 
orem due to Rossby (e.g., [51]) have provided the 
theoretical suggestions for much of the experimental 
work. It is not going to be easy, nor even possible in 
some cases, to measure adequately the field quantities 
which may appear to be most immediately required in 
comparing possible theoretical explanations. Neverthe- 
less, certain types of measurements are quite possible. 
As experimental work progresses, techniques and altered 
concepts of the quantities to be measured should de- 
velop to a far higher point than is the case at present. 
Im connection with general circulation problems, the 
following experimental steps appear to me to be the 
next ones to take. 
1. The simple types of techniques used in hydraulic 
or aerodynamic experiments (or in the studies cited 
above) to measure velocity and temperature should be 
extended to the cylindrical discs studied by Vettin, 
Thomson, and Exner. Several modifications would be 
necessary and height-diameter ratios should be investi- 
gated to as low values as are possible.* Besides the 
general possibility of investigating the motions near a 
pole for thin shells, there are indications that evidence 
could be obtained bearing on a comment by Jeffreys 
[36] that a reversed distribution of thermodynamic 
sources might reverse the sense of the average relative 
zonal motions of the atmosphere as functions of lati- 
tude (see also below). Preliminary observations on a 
rotating dishpan apparatus constructed by F. Hall 
for a similar type of study indicated that with a Bunsen 
burner under the center (instead of a cold source there) 
the more rapid outward radial currents show relative 
6. It has recently come to my attention that in 1902 C. A. 
Bjerknes carried out experiments with a rotating cylinder of 
water (12 cm high, 36 cm wide, 7 rpm) to verify qualitatively 
Ekman’s theory of the surface drift current in the ocean. A 
jet of air 10 cm broad was blown along a diameter relative to 
the rotating cylinder. Measurements of the currents with 
floating balls and a sensitive directional vane showed the 
proper deflection to the right (Northern Hemisphere rotation) 
at the surface and increasing rightward deflection in the top 
centimeter where the Ekman spiral would be expected. This 
suggests the possibility of studying wind-driven currents in 
oceanic basins of suitable shapes on an experimental basis, 
so long as variations of the Coriolis parameter are not impor- 
tant, and of utilizing paraboloidal shapes where such varia- 
tion is important. (See Ekman, V. W., ‘‘On the Influence of the 
Earth’s Rotation on Ocean Currents.’? Ark. Mat. Astr. Fys., 
Vol. 2, No. 11, pp. 51-52 (1905) and also [85] and [86].) 
1243 
westerlies instead of easterlies as should be the case to 
correspond with Vettin’s result. The particular equip- 
ment is subject to vibration so that some reservations 
must be entered, but very recent (March, 1950) ro- 
toscope observations and photographs with the dishpan 
(radius 7 in.) filled with water to a depth of from 214 
to 3 in. and heated at the outer rim show some extremely 
striking patterns. Figure 11 shows an aluminum-powder 
tive circulation at the top surface of water in a pan rotating 
clockwise (12 rpm), which is being heated at the bottom near 
the rim. The narrow current near the rim is westerly (¢.e., 
moving more rapidly than the pan) while that near the middle 
is predominantly easterly. The second current radially in 
from the counter, for example, is almost certainly easterly, 
on the basis of visual observations. (Photograph 8 min after 
heating began, exposure 14 sec, westerly current u/C, = —0.05; 
at 12 min after heating began, temperature at top center was 
a at bottom rim 31.9C taken with a mercury thermome- 
ter. 
streak photograph taken through the rotoscope with an 
exposure of 14 sec so as to show the motions relative 
to the pan. The narrow band of motion near the rim 
(moving clockwise) is a westerly current with u/Cz 
values running to about —0.06 or more, when the angu- 
lar velocity of the pan is 12 rpm. The rest of the top 
surface is occupied by a number of eddies of various 
sizes which, for example, give average values of +-0.02 
for u/Ce at a distance of 0.67) from the center (ro is 
the rim radius). In this condition the water temperature 
away from the pan surface is hotter at the rim than at 
the center by 1-2C. The burner was also moved in to 
0.579. The average temperature gradient in the water 
is then reversed with the temperature at the pole per- 
haps 1¢C warmer than at the rim because of the rapid 
heat conduction in the metal of the pan. The relative 
