502 
While the coverage shown in these figures is reason- 
ably adequate in heavily populated land areas and in 
the much-traveled North Atlantic, there are still many 
important gaps, notably in the southern portions of the 
North Atlantic and North Pacific Oceans and in many 
parts of Asia. This situation is due in part to commu- 
nication difficulties and to the fact that it is more ex- 
pensive and difficult (or even dangerous) to establish 
stations in unfrequented and less habitable areas. Nev- 
ertheless, political and economic forces are also partly 
responsible for the tendency of the density of observa- 
tions to be proportional to the number of inhabitants. 
Since there is mounting evidence (some of which will 
be presented in this paper) that weather in any given 
area of the globe may be importantly influenced by 
conditions in other areas, no matter how remote, it 
would appear preferable to reduce the amount of data 
in populated areas and use the money thus saved to 
increase the number of reports, particularly upper-air 
reports, from remote areas. Certainly it is a mistake to 
reduce the amount of hemispheric data on the grounds 
that it is no longer necessary for military purposes. 
For the past two or three years there has been a strong 
tendency in this direction. The International Meteoro- 
logical Organization, the International Civil Aviation 
Organization, the Pacific Science Association, and simi- 
lar organizations should be commended for and en- 
couraged in their present efforts to improve the data 
coverage of the world. 
A brief list of published hemisphere charts is given 
in the references [1, 9, 17, 28, 29, 30, 46, 50, 51]. The 
Air Weather Service and U.S. Weather Bureau North- 
ern Hemisphere sea-level and 500-mb analyses contain, 
in addition to analyzed charts, a valuable listing of all 
data received. To date the Air Weather Service charts 
have been published only from October 1945 to Sep- 
tember 1946. The U. S. Weather Bureau commenced 
analysis and publication of this series beginning with 
January 1949. Charts for January through November 
1949 are currently available. Efforts should be made to 
close the gap in the historical map series from 1939 to 
1945. This includes the years 1944 and 1945 when hemi- 
spheric data, particularly in the South Pacific, were 
more complete than at any other time. 
Many of the investigations to be discussed in this 
article were made using unpublished operational analy- 
ses. These include twice-daily sea-level and 10,000-ft 
(or 700-mb) charts over the Northern Hemisphere from 
1940 to date as analyzed at the Extended Forecast 
Section of the U. S. Weather Bureau. Not until mid- 
1943 do these charts cover half a hemisphere or more. 
From them, twice-weekly five-day mean maps and 
twice-monthly thirty-day charts have been constructed. 
Five-day mean twice-weekly sea-level and 3-km charts 
for the cold season October through March 1933 to 1940 
have been prepared by the Department of Meteorology 
at M.1.T. The 40-yr daily historical map series [28] 
has led to the construction of monthly mean sea-level 
charts. Corresponding 10,000-ft monthly mean charts 
THE GENERAL CIRCULATION 
have been constructed partly by extrapolation from sea 
level for the period 1932 to 1939.1 
Reasons for the Use of Mean Charts. Mean charts, 
showing the average state of the general circulation for 
specified periods of time, have been used extensively by 
many investigators, including the authors, for studying 
the observed behavior of the general circulation. Here 
are some of the advantages of mean charts over daily 
synoptic charts: 
Because of the unreliable nature of portions of hemi- 
sphere-wide analysis, the use of means leads to a 
smoothing which tends to eliminate experimental errors. 
Furthermore, such smoothing also eliminates irregulari- 
ties in the flow pattern, such as small eddies or minor 
atmospheric waves. These irregularities, while real 
enough, tend to confuse the large-scale features of the 
circulation which are of most importance in long-range 
forecasting. To a lesser degree this second type of 
smoothing may be attained by the use of synoptic charts 
from higher tropospheric layers or by mathematical 
devices, such as Charney’s geostrophic wind approxi- 
mation [10]. 
Perhaps the most important justification for the use 
of means is that they reveal large-scale features having 
a long period. Indeed the speed with which a given 
pattern changes appears to vary inversely as the num- 
ber of days over which the averaging is done. The 
authors feel that slow evolution of this character holds 
out the hope for successful long-range weather fore- 
casting. 
Granting the correctness of the arguments given 
above, the question has often been raised as to how 
mean charts are to be interpreted from a physical point 
of view. In particular, it has been argued that the equa- 
tions of motion cannot be applied to a mean flow pat- 
tern, for, since these equations are nonlmear, the 
average value of acceleration will not equal the sum 
of the corresponding average values of the other terms. 
The answer to this question cannot yet be given. How- 
ever, this problem seems to be closely connected with 
the statistical theory of turbulence [3]. Briefly, this 
theory states that if a given fluid consists of an average 
flow upon which are superimposed random eddies, then 
the equations of motion for laminar flow can be applied, 
provided certain frictional stress terms are added which 
express the cross-flow eddy transfer of momentum. 
There seems to be no obvious reason why it should be 
necessary to place any arbitrary time limit to the aver- 
age flow, so that the equations of motion ought to be 
just as applicable to mean charts of a month’s duration 
as they are to mean charts of a few minutes (synoptic 
charts)—provided the proper frictional terms are added 
[20]. This type of reasoning has been used by Defant 
[16], Lettau [32], and more recently by Elliott and 
Smith [21] to measure the frictional stress exerted by 
traveling cyclones and anticyclones. The omission of 
frictional terms in applying the equations of motion to 
1. All of these charts have been microfilmed and copies may 
be purchased by writing to the Librarian, U. 8. Weather 
Bureau, Washington, D. C. 
