1236 
by the theory from which these experiments developed, 
this mechanism, if substantiated by further checks, 
would principally be ‘frictional’? (turbulent) mixing 
processes that are similar in the two cases. 
The obvious step beyond the purely kinematic check 
of average velocity fields is to proceed to more refined 
measurements of important dynamical quantities over 
a more extensive range of parameter values. Once an 
accurate picture is obtained of how far qualitative and 
quantitative similarity extends, many avenues of rea- 
soning and experiment will be opened for attempts to 
interpret both the agreements and the disagreements 
in terms of unifying principles. 
An important step in this process will be to attempt 
to produce experimental arrangements which have close 
resemblances in one or two selected I-parameters at 
the expense, in general, of similarity in others. Perhaps 
the most important example of this possibility is the 
case of convection in thin fluid shells on a rotating 
paraboloid as compared with the spherical shell experi- 
ments mentioned above. If the angular velocity of the 
paraboloid is such that the resultant of the gravitational 
and centrifugal forces is exactly perpendicular to the 
surface at every point, the solid surface will have the 
planetary property of being an equipotential surface. 
More realistic thermodynamic actions should be pos- 
sible in this arrangement than in the above-cited 
experiments with a rotating spherical shell. A free upper 
surface can be attaimed and the Coriolis parameter 
would still be variable, but all these advantages are 
obtained, of course, at the expense of abandoning any 
approach to geometrical similarity to planetary shapes 
and of tolerating rather considerable horizontal varia- 
tions of the apparent gravity acceleration. Various other 
practical advantages, such as the ability to produce 
vigorous convectional mixing with practicable tempera- 
ture differences within the fluid, may possibly also be 
lost. By appropriate modifications of this or other types 
it should be possible soto vary the contributions of 
the three major factors in these experiments that each 
one can be more or less isolated for study. 
In developing an experimental program, a certain 
backlog of experience will have to be built up on the 
purely practical side. Since the requirements of im- 
portance to meteorological questions involve somewhat 
different emphases than, for example, in aerodynamic 
experimentation, trials of satisfactory techniques in 
themselves will occupy a considerable period. However, 
the present development of experimental technique, 
both in general and in specifie types of model experi- 
mentation, appears very favorable. Many alternatives 
are available which will require only slight further 
development to become adaptable to such a program. 
In matters both of interpretation and of technique, 
any experimental problems which can be analyzed 
theoretically with relative completeness will be of great 
value in bringing difficulties out mto the open. For 
example, the work concerned with the stability of a 
polar vortex, which will be referred to later, has been 
particularly illuminating in this direction [23, 42]. In 
this work, a relatively complete theoretical analysis 
could be carried out for at least the basic motion and 
LABORATORY INVESTIGATIONS 
compared directly with the experimental results. By 
quantitative comparisons, it was found that surface 
tension forces were making an appreciable dynamical 
contribution in the phenomenon being studied. Such a 
contribution, having no counterpart in the major at- 
mospheric problem in view, can be eliminated either 
experimentally or analytically before drawing conclu- 
sions concerning the prototype from the model. It is 
quite doubtful that this effect would have been noticed 
if a quantitative theory had not been available. On 
the other hand, with this experience in mind, it will 
be easier to make qualitative estimates of the magnitude 
of such effects in any subsequent experiments of a 
similar nature. 
Experiments on Rotatory Phenomena—Cyclones and 
Tornadoes 
Almost from the beginning of modern meteorological 
thought, and after the discovery of the existence of 
barometric storms, attempts have been made to con- 
struct experimental models of cyclones. In fact, many 
of the cyclone theories of the early nineteenth century 
seem to have drawn their ideas from such experiments 
and from comparison with tornadoes. These experi- 
ments have usually shown a more distinct resemblance 
to tornadoes than to present-day ideas of cyclones. An 
early piece of work of this kind, which is in fact very 
similar to many later experiments, is that of Wilcke 
[838, 71] in 1780 at Stockholm. In all probability there 
were many others of the same general nature during 
the eighteenth century. 
In Wilcke’s experiments (Fig. 1), a thick steel wire 
3 Hi i : es 4 aes 
E oud ; a 
Fig. 1.—Illustration from Wilcke [71] showing a spiral vortex 
generated in water by mechanical rotation at the top of the 
cylinder. The cylinder is approximately 14 in. high and 7 in. 
in diameter. The rates of rotation are not specified but presum- 
ably are of the order of a hundred or so revolutions per minute. 
Wilcke also drove the vortex from the bottom, investigated an 
aleohol-oil system, and worked with vortices generated by out- 
flow from a hole in the bottom of the cylinder. 
