854 
tain themseives with a reasonable value for at least 
three days, the analogues are, on the average, synop- 
tically and kinematically sound, (2) for the purpose of 
making long-range forecasts up to ten days’ duration 
there usually exists one analogue which gives a pre- 
cipitation forecast as good as, if not better than, the 
present 24-hr forecast (this, of course, is after the fact, 
and at the present time no method is available for 
detecting such an analogue in advance), (8) there are 
indications that certain previous temperature distribu- 
tions aid materially in picking that analogue, and (4) 
maps which show the same general movement of the 
essential features are usually good analogues. 
In order to determine why it is that the major centers 
of action have such variable dynamics from year to 
year, it may be desirable to examine more carefully 
variations in sea-surface temperature. Preliminary 
studies have indicated that in certain regions the sea 
surface has undergone rather marked changes in tem- 
peratures and that the variation of the difference 
between the air temperature and the water temperature 
shows considerable change from year to year. Since the 
water temperature is a very persistent phenomenon, 
these long swings of above and below normal over a 
period of many months may have considerable influ- 
ence on the behavior of the weather. Also, since about 
three-quarters of the world is ocean, perhaps too little 
thought has been given to these variations as potential 
sources of the fluctuations in the dynamics of the 
weather system. 
There are, of course, many variables which should 
be considered in attempting to get a complete statistical 
model, particularly in light of the fact that we are 
dealing with a servomechanism. The low-pressure area 
produces the clouds, the clouds change the energy 
input, the change of the energy input causes a change 
in the energy of the system, and that in turn affects 
the movement of the low in its future course. Sunspot 
activity, variations in total pressure over the poles 
and over the equatorial regions—all these must con- 
tribute their part. It therefore seems advisable that 
investigations utilizing the statistical approach be 
greatly increased so that more and more information 
will be available, thus clarifying the nature of the 
phenomenon with which we are dealing. As this statis- 
tical information establishes the variables which have 
dynamic properties, it should be possible to set up 
mathematical models on the bases of reasonable 
hypotheses inferred from the observational facts, and 
in this way we may hope that the final solution of the 
weather forecasting problem will be reached. On the 
other hand, it might develop that meteorology is faced 
with a situation analogous to that treated in the kinetic 
theory of gases, wherein the behavior of an individual 
molecule is impossible to predict, and only certain 
average features of all the molecules can be computed. 
Nevertheless, if the problem has to be reduced to this 
type of conclusion, it would at least be possible to 
estimate the respective probabilities for the occurrence 
of each possible state. 
WHATHER FORECASTING 
Thus it appears that if the general forecasting prob- 
lem of meteorology is to progress as rapidly as possible 
toward its solution, meteorologists and statisticians 
should combine their efforts in a more united fashion. 
It is clear from the type of problem with which we are 
dealing that it is only through the statistician’s under- 
standing the meteorological point of view and his being 
able to understand and utilize the experience of the men 
who have been watching the progress of weather situa- 
tions for a long period of time that he can be effective. 
Reciprocally, the meteorologist cannot take advantage 
of statistical techniques which would be of great ma- 
terial aid until he himself understands the basic prin- 
ciples upon which the techniques are based and also 
understands what the statistician wishes to accomplish. 
REFERENCES 
In the following publications the reader will find an account 
of applications of statistics to weather forecasting. 
1. Baur, F., Hinftihrung in die Grosswetterkunde. Wiesbaden, 
Dieterich, 1948. 
2. —— ‘Der gegenwirtige Stand der meteorologischen Kor- 
relationsforschung.”’ Meteor. Z., 47:42-52 (1930). 
3. —— “Die Stérungen der allgemeinen atmospharischen 
Zirkulation in der gemiassigten Zone.’’ Meteor. Z., 54:437- 
444 (1937). 
4. Daruine, D.A., Report on the Accomplishments of the Statis- 
tical Project. Calif. Inst. Tech—Army Air Force Res. 
Unit, 1942. 
5. —— Relationship between the Analogue Forecast and Other 
Methods of Forecasting. Calif. Inst. Tech—Army Air 
Force Res. Unit, 1944. 
6. Exuiorr, R. D., Hxtended Weather Forecasting by Weather 
Type Methods. U. 8. Navy Dept., Long Range Weather 
Forecasting Unit, Washington, D. C., 1944. 
7. Haurwitz, B., ‘Final Report on the Use of Symmetry 
Points in the Pressure Curves for Long-Range Fore- 
easts.”” U. S. Air Force, Air Wea. Serv. Tech. Rep. No. 
105-7 (1944). 
8. Namras, J., Hatended Forecasting by Mean Circulation 
Methods. U. S. Weather Bureau, Washington, D. C., 
1947. 
9. Scout, I. I., ‘‘Dynamic Persistence and Its Applications 
to Long-Range Foreshadowing.” Harv. meteor. Stud., No. 
8 (1947). 
10. Scumauss, A., ‘‘SSynoptische Singularitaten.’’ Meteor. Z., 
55:385-403 (1938). : 
11. Scoumann, T. E. W., Statistical Weather Forecasting. 
Meteor. Res. Bur., Pretoria, Un. of South Africa, 1947. 
12. Srarr, V. P., A Physical Characterization of the General 
Circulation. Dept. Meteor., Mass. Inst. Tech., Rep. No. 
1, General Circulation Project No. AF 19-122-153, Geo- 
phys. Res. Lab., Cambridge, Mass., 1949. 
13. Wapsworts, G. P., “Short Range and Extended Forecast- 
ing by Statistical Methods.” U. S. Air Force, Azr Wea. 
Serv. Tech. Rep. No. 105-38, 1948. 
14. —— Analogue Techniques in the Forecasting of Rainfall. 
Geophysical Research Directorate Reps. No. 1, 2, and 
3, 1948; Dynamics of the Semi-permanent Pressure Cells, 
Rep. No. 4, 1949; Further Analysis of Dynamics of Major 
Pressure Cells, Rep. No. 5, 1949; Position of Major Pres- 
