872 
tudes, we become convinced that there are no horizontal 
temperature gradients in latitudes below 20°. 
The new model, then, is open to the same empirical 
objections as the old. But there are also serious objec- 
tions to the theoretical explanations which have been 
advanced, inasmuch as the principle of the conservation 
of the vertical component of the absolute vorticity is 
appealed to. Without entering into the details of the 
new theory, which would take us beyond the limits of 
this article, we may sum up these objections by saying 
that it has yet to be shown that the mean motion of the 
tropical atmosphere is autobarotropic, horizontal, fric- 
tionless, and nondivergent to a high degree of approxi- 
mation. Until this has been shown, we must suspend 
judgement as to the applicability of Rossby’s concepts 
in low latitudes. Even if we admit them, it will be only 
in certain longitudes, and not as a general explanation 
of the tropical part of the planetary circulation. 
It is clear now that the great success of the per- 
turbation school has been in the realm of descriptive 
synoptic meteorology; the models of the perturbations 
which they have promulgated have been applied in the 
field with far greater success than was ever achieved by 
the air-mass school. In dynamical meteorology and in 
their treatment of the general circulation, members of 
this school have been less successful. There have been 
complaints, moreover, that the model of the easterly 
wave published by Riehl is far too rigid. In particular, 
the slope of the actual systems is frequently different 
from what might be expected from the published ac- 
counts and, what is more important, the weather dis- 
tribution about the wave axes departs widely from the 
model [26]. 
But these, as we shall see, are not serious objections; 
they can be removed by a more careful study of the 
perturbations m low latitudes and by avoiding the 
tendency to apply models to synoptic maps as if they 
were rubber stamps. I ought perhaps to pomt out that 
my sympathies lie with the perturbation school and 
that if the criticisms here seem to be harsh, it is because 
I believe that future advances in tropical meteorology 
will come through the application of the concepts and 
methods of this school, provided they are freed from 
fallacious generalizations. 
THE PRESENT AND THE FUTURE 
General Observations. Each of the methods of ap- 
proach to the tropical problems which were described in 
the preceding sections has something to recommend it; 
each has contributed something to our empirical knowl- 
edge and our understanding. As far as we can see at 
present, progress will come by pursuing to further 
lengths, with new and more detailed data, the lines of 
investigation that have already proved profitable. It is 
unlikely, that is, that we shall entertain any radically 
new theory of tropical meteorology in the near future. 
As a guide to the assessment of the present state and 
possible future development of the science we ought 
therefore to extract from the confusion of present con- 
flicts those concepts upon which the majority of trop- 
ical meteorologists would agree. Before we do this we 
TROPICAL METEOROLOGY 
ought, however, to make a few general observations on 
method, deriving our principles from the study, not 
only of the achievements of the three schools, but of 
the errors they have committed. 
1. Before attempting causal analyses of tropical 
phenomena, or dynamic explanations, we ought to be 
sure that our empirical descriptions are correct. This 
principle was violated by the climatological school in 
describing the doldrums, particularly the doldrums in 
the Pacific; by the air-mass school in describing fronts, 
since they neglected to describe accurately the tempera- 
ture and wind changes observed at the supposed dis- 
continuities; and by the perturbation school in omitting 
all mention of the convergence lines of the frontal school 
in their description of the model of the easterly wave. 
2. The fallacy of generalization from a given longi- 
tude should be avoided as far as possible. We ought 
always to be on guard against it, particularly when 
dealing with the mean circulation in the tropics. But 
faulty generalization also enters in other contexts, 
especially when we have theoretical preconceptions. 
Thus the frontal school, as soon as it found evidence 
of empirical ‘fronts” in the neighborhood of the equa- 
torial low-pressure trough, assumed that the occlusion 
process would also occur. No evidence was, so far as I 
know, ever produced to back this up. We ought then to 
remember at all times that, owing to the fact that we 
never have as much data as we really need, the habit of 
hasty generalization is an occupational disease of mete- 
orologists. All the geophysical sciences suffer from this 
handicap; we ought to be at least as careful as the pure 
physicists, and probably more careful. 
3. We ought to guard against the statistical fallacy 
in meteorology. This consists in assuming that a mean 
motion of the atmosphere, arrived at by the standard 
methods of climatology, represents a physically pos- 
sible steady-state motion. 
Achievements and Future of the Climatological 
School. There is no doubt that the departures from the 
mean values of the meteorological elements that are 
attributable to the diurnal cycle of radiation and to 
orographical effects can be of great magnitude in the 
tropics. It is the virtue of the climatological school that 
they have always emphasized these variations. How- 
ever, in recent years little research work that might 
lead to a better understanding of the phenomena has 
been carried out. It is encouraging to notice that Leo- 
pold [41] has entered this field with papers on the 
so-called sea-breeze fronts in the Hawaiian Islands. 
Wartime synoptic experience everywhere in the tropics 
has confirmed the existence of organized lines of cumu- 
lus or cumulonimbus, accompanied by wind shifts, 
which simulate the cold fronts of high latitudes but 
which are demonstrably due to diurnal and orographical 
causes. The dynamical problems raised by these 
“fronts” are among the most interesting problems of 
tropical meteorology. Their properties, however, can 
be investigated only on location; it is to be hoped that 
young meteorologists resident in tropical areas of high 
relief will, in the future, undertake the investigation. 
Apart from the theoretical interest of such “fronts,” 
