TROPICAL METEOROLOGY 
the doldrums. But even the doldrum origin of tropical 
storms was not quite general. 
At least some of the hurricanes that afflict the West 
Indies and the southern coast of the United States 
originate in the Caribbean or in the Gulf of Mexico, 
and when this happens they appear almost without 
warning in the trade current, not in the doldrums [25]. 
The second circumstance that is fatal to the extreme 
climatological view is that the really steady trades 
occupy only a small fraction of the total oceanic areas 
in the tropics. Further, the antitrades are similarly 
found only in comparatively restricted areas. The typi- 
cal trades, as described in models of the general circula- 
tion, are found only in the eastern parts of the equatorial 
branches of the great subtropical anticyclones. The 
regions of high persistence are these same areas. In the 
western fractions of the tropical oceans the trade wind 
is variable, not as variable as the belt of the westerlies, 
it is true, but still subject to perturbations of the wind 
field, so clearly independent of thermal causes as to 
belie the classical description. In the western oceans 
also, the meridional component of the wind may not 
be directed toward the equator, even in the mean. At 
Pukapuka (11°S, 166°W), for example, the mean wind 
at the surface is directed during the summer away from 
the equator, though this station always lies south of 
the doldrums [24]. Similarly the antitrade at higher 
levels is best developed and most easily discovered 
above the eastern borders of the subtropical highs. To 
the west, the westerlies are found at higher and higher 
levels, and in the lower latitudes and near the western 
borders of the anticyclones they may not be present 
below the tropopause. The deathblow to the notion of 
universal antitrades in tropical latitudes was given by 
J. Bjerknes in his discussion of the subtropical anti- 
cyclones [6]. It is clear from that paper that the oceanic 
anticyclones are major perturbations of the atmosphere 
with features independent of any direct thermal cause, 
and that any explanation that leaves them out of 
account by dismissing the trades as part of a simple 
“direct cell” symmetrical on all longitudes is artificial. 
While such explanations may suffice in a discussion 
that is directed mainly toward clarifying the vagaries 
_ of the westerlies, they are too unrealistic to be useful 
in tropical meteorology. 
A more serious difficulty arises when, having recog- 
nized this oversimplification of the general circulation, 
we attempt to discuss the details of the observed 
tropical circulations, daily, seasonal, or annual. While 
we have an excellent series of observations over the 
oceans (excellent, that is, from the climatologists’ point 
of view), we have very few observations and those for 
comparatively short periods over most of the land 
masses. India, it is true, has long provided observations 
from a dense network, but we still lack long and reliable 
series from Southeast Asia. The data are even more 
unreliable and sparse from equatorial Africa and from 
Brazil. As we shall see later, the wartime synoptic data 
from the latter areas strongly suggest that the simple 
picture of the monsoons in those areas is probably a 
863 
climatological abstraction as misleading as that of the 
trades. What good data we have, moreover, consists of 
surface observations. Upper-air observations are still 
so widely spaced, even at the present day, that most 
high-latitude analysts would refuse to commit them- 
selves to a space analysis of either the synoptic data or 
the means. It must be remembered, also, that except for 
some Indian stations, and a few in the Caribbean and 
Panama, practically no radiosonde or meteorograph 
data antedates World War II. Further, the upper winds 
in India and the East Indies, upon which much empha- 
sis has been placed in some papers [23], are derived from 
the observations of pilot balloons. Dr. Mintz of the 
Meteorology Department of the University of Cali- 
fornia at Los Angeles has pointed out to me how 
misleading mean winds derived from such observations 
in the tropics can be. In “The General Circulation of 
the Atmosphere over India and Its Neighbourhood”’ 
[51], for example, are published data giving the mean 
wind at 4 and 8 km according to the pilot balloons. In 
the same volume are also given the mean directions of 
the clouds at and about those levels. The mean circula- 
tions for January derived from these two independent 
sources are directly opposite to one another, that indi- 
cated by the balloons corresponding to the east end of 
an anticyclone situated at the appropriate level, that 
from the clouds corresponding to the west end. It is 
clear that the pilot balloons are selective and that mean 
winds derived from them are probably typical of clear 
weather; the balloons are very easily lost from view in 
tropical regions even with small amounts of convective 
cloud, since the cloud pillars are so much higher in low 
than in high latitudes. On the other hand, winds derived 
from middle and upper clouds are selective in the oppo- 
site direction, since such clouds are, in low latitudes, 
most commonly the result of convection, being formed 
from the middle and upper parts of cumulonimbus. 
Such considerations as the foregoing ought to make 
us hesitate to generalize when discussing the upper 
parts of the tropical circulations, either in the trade or 
in the monsoon regions, especially when our point of 
view leads us to place the greatest emphasis on seasonal 
and annual means. There are literally not sufficient 
data to make a single reliable statistical generalization 
applicable to the upper levels. Our best hope is to allow 
the lapse of time and the gradual extension of the tropi- 
cal observing networks to accumulate data that might 
later give us an insight into the upper circulation in low 
latitudes. In the meantime, the chief problems seem to 
be synoptic and dynamic, particularly the dynamic 
problems growing out of synoptic experience. 
THE AIR-MASS METHOD 
When, after World War I, meteorologists of the 
Bergen group began to extend their application of 
frontal and air-mass analysis to regions outside Scandi- 
navia, they found the climatological concepts of trop- 
ical meteorology already adapted to their purpose. We 
have already referred to Bergeron’s approval of the 
notions, trade and monsoon, as modified to include not 
