OBSERVATIONAL STUDIES OF GENERAL CIRCULATION PATTERNS 
10,000 ft in the 20° latitude band of greatest northward 
rate of change. 
With these indices (and many more) Willett pro- 
ceeded to look for contemporary and lag correlations, 
after removing the seasonal trend. Ordinary linear 
(Pearsonian) coefficients of correlation were computed 
from fifty-two overlapping five-day mean periods for 
each six-month period (October—March) during each 
of the seven years and for the seven years as a whole. 
The terms “significant” and “highly significant’? were 
applied to those correlations which had less than five 
per cent and less than one per cent probability, respec- 
tively, of occurring by chance. 
Of the contemporary correlations one of the most 
significant was the negative one (—0.54) found with 
fair year-to-year consistency between the average pres- 
sure along latitudes 70°N and 35°N. This negative 
correlation is, on a global scale, the same phenomenon 
that Sir Gilbert Walker stressed in his North Atlantic 
and North Pacific oscillations, when he pomted out 
the opposing reactions of Aleutian and Icelandic lows 
to North Pacific and Azores subtropical anticyclones, 
respectively. These results stress the important fact 
that the polar-cap anticyclone grows at the expense of 
the subtropical highs, the shifts in mass being asso- 
ciated with an expanded circumpolar vortex of low 
index described above on page 561. 
Largely as a result of this negative correlation the sea- 
level polar easterlies correlate insignificantly with sea- 
level zonal westerlies in spite of the fact that these two 
indices possess the latitude 55°N im common. But the 
sea-level polar easterlies correlate significantly and con- 
sistently from year to year in a positive sense with the 
zonal westerlies at the 10,000-ft level, indicating that 
when the sea-level polar easterlies are strong, the pole- 
ward temperature gradient in the 10,000-ft layer above 
the surface is also strong. This is also borne out by sig- 
nificant negative correlation between solenoid index and 
zonal westerlies and positive correlation between sole- 
noid index and 10,000-ft westerlies. The correlation 
between the strength of the maximum 3-km westerlies 
and their latitude, while not significant, is consistently 
negative from year to year, suggesting the observed 
fact that the jet stream at times intensifies as it expands 
equatorward. 
The meridional index computed for 45°N shows a 
tendency (though not quite significant) for strong merid- 
ional circulations to be associated with weak zonal 
components. The meridional exchange at 45°N also 
appears to be related to the latitude of the maximum 
3-km westerlies, since a significant negative correla- 
tion exists between these two quantities. Again this 
reflects the meridional (e.g., cellular) character of the 
low-index pattern as compared with the high-index 
pattern. 
From these correlations Willett [57] concludes that 
. . the low index circulation pattern, in contrast to the high 
index pattern, is characterized by: 
1. A relatively strong poleward temperature gradient, at 
least between sea-level and the 3-km level. 
2. Relatively weak zonal westerlies at sea-level which in- 
565 
crease to relatively strong aloft, with a tendency to be dis- 
placed equatorward, that is, an intensified and expanded 
circumpolar vortex in the upper troposphere. 
3. Strong polar easterlies as a result of a relatively strong 
sea-level polar anticyclone, which in turn is produced pri- 
marily from a weakening of the sub-tropical high pressure 
belt. 
4. In middle latitudes a relatively strong meridional circu- 
lation at sea-level which tends to weaken with height as the 
zonal westerlies become relatively stronger. 
The step from contemporaneous to lag correlations 
involving features of the general circulation appears 
to be of a high order of difficulty. In earlier decades, 
particularly the 1920’s, a popular form of research 
aimed at long-range forecasting consisted of correlat- 
ing, with different lags, weather or circulation elements 
at various points over one or both hemispheres. One 
such ambitious program, undertaken at the U. S. 
Weather Bureau under Weightman [53], involved cor- 
relations between pressure, temperature, and rainfall 
for lags of three, six, and nine months. It was hoped 
that these supposedly fact-finding and purely statistical 
studies would lead to a better understanding of the 
general circulation and provide a basis for long-range 
weather forecasting. But the technique was apparently 
too erude to bring to light any very revealing secrets 
of the behavior and evolution of atmospheric circula- 
tion and concomitant weather. Similarly, m their vast 
undertaking Willett and his associates, working with 
data described previously, were unable to detect any 
reliable predictors of the general circulation in its zonal 
or meridional branches or in regional divisions, aside 
from a small degree of persistence. From these negative 
results Willett [57] concluded that the general circu- 
lation possessed no internal mechanism of operation 
and that there was apparently no one way that a sub- 
sequent state could dynamically and thermodynami- 
cally evolve from an initial state. From this conclusion 
it was apparently not a difficult step for Willett to join 
that group of meteorologists who maintain that controls 
of nonseasonal variations in the general circulation lie 
in the sun. One can, of course, question the fundamental 
premise upon which this latter conclusion was based. 
Indeed, Willett himself is quite aware of the possibility 
that the correlation technique, applied as he has ap- 
plied it, is too crude to unveil the secrets of a highly 
complex organism like the atmosphere. The restrictions 
placed upon correlation techniques by fixed time and 
space intervals, in the opinion of the present authors, 
are much too severe to fasten upon an evasive atmos- 
phere notoriously resistant to strait jackets. 
In this semiphilosophical question seems to lie one 
of the principal deficiencies of research in the problem 
of general circulation. Admittedly, all meteorologists 
aim for objectivity. But in the quest for objectivity it is 
sometimes forgotten that adequate statistical tools may 
not yet be developed to treat many complex problems. 
The problems of time series, So common yet so annoy- 
ing to meteorology, serve as an example. It thus be- 
hooves the meteorologist to encourage the development 
of statistical tools which are by definition objective 
