EXTENDED-RANGE FORECASTING 
The indefatigable work of Baur on the subject of 
types must also be mentioned, for he has probably done 
more work of this sort as it applies to Huropean weather 
than anyone else [1]. For his types he has tabulated the 
seasonal frequency of occurrence, associated weather, 
and other statistics which are generally worked up by 
various typing schools. 
Since both Baur and Elliott have articles describing 
their methods in this Compendium, we shall make only 
a few remarks on this general subject. 
In the first place, the question of uniqueness of classi- 
fication arises. Will two equally trained meteorologists 
give the same type designation to an observed sequence 
of maps, and furthermore, will they agree as to the dates 
of its beginning and termination? For the most prom- 
inent regional types it appears that there can be fairly 
good general agreement. However, there seem to be 
numerous “‘borderline”’ or transitional cases where there 
is disagreement. The dates of inception and termination 
of a type seem to be difficult to define. In the minds of 
some meteorologists, these difficulties have raised the 
question as to whether discrete types, with the neces- 
sary discontinuities between types, exist, or whether 
there is a more or less gradual transition between circu- 
lations. The final answer to this problem cannot yet be 
given. On the one hand the supporters of the type ap- 
proach emphasize those periods in which a very rapid 
change of the circulation takes place and completely 
alters the weather regime in a short period of time, 
while its detractors point to many periods where the 
evolution of the general regional circulation is of a 
gradual nature for many weeks. Unquestionably the 
truth is composed of both these viewpoints. 
When we come to the question of prognosis the lines 
become more sharply drawn. The exponents of weather 
types maintain that there exist methods of predicting 
the subsequent type. Statistical studies of the probabil- 
ity of one type following another and the relationship 
of subsequent types to the position and orientation of 
some center of action have been carried on for the pur- 
pose of prediction. Perhaps the safest method of predic- 
tion, though limited, is that used by the Multanovsky 
school—to wait until the first one or two days of a type 
have elapsed. By that time the type has been identified 
and one issues a forecast only for the remainder of the 
natural period during which the type exists. In the 
opinion of the present author, the weakest phase of the 
entire weather-type approach is the lack of any reliable 
methods of predicting what will be the coming type. 
This statement does not involve denial of the utility 
of weather types in diagnosis, or particularly in peda- 
gogy, but it questions their utility for purposes of prog- 
nosis. This question was raised in the introduction to 
this article where the inadequacies of a regional ap- 
proach, contrasted to a hemispherical approach, were 
cited. 
The analogue method appears to crop up more or less 
periodically in the history of synoptic meteorology, 
showing perhaps a tendency to reach maximum popu- 
larity at the times of greatest stress (wartime). Wads- 
worth [22] was one of the principal exponents of this 
807 
approach during World War II. The basic theory under- 
lying it is entirely logical: If two maps or weather situ- 
ations are found which are identical, then so are their 
physical properties, and therefore the subsequent evo- 
lution will be identical. Assuming no discontinuities in 
extraterrestrial influences, no one can take serious issue 
with this concept. The problem then resolves into one 
of finding the proper analogue in the archives of weather 
maps. 
The proponents of the analogue method cite several 
advantages of its usage: 
1. It is objective and does not rely upon complex and 
subtle reasoning inherent in physical methods which 
are far from perfected. 
2. It is relatively fast in operation and forecasts can 
be turned out rapidly G@ndeed, analogues can now be 
selected by punch-card machinery). 
3. Analogues automatically contain the climatologi- 
cal peculiarities of every location for which a forecast 
is required. 
These three arguments are indeed potent and un- 
questionably have a strong appeal, particularly to ad- 
ministrative officials. To counter these arguments the 
opponents of the analogue technique point out the 
following: 
For the analogue to be successful for a period of a 
few days or more the agreement between current and 
historical maps must be three-fold: (1) spatially similar 
for large areas covering a large portion of the hemi- 
sphere, (2) similar in the third dimension, and (8) 
similar in evolution. The necessity of considering large 
areas is a direct consequence of the rapid dispersion of 
planetary atmospheric waves indicated earlier. The 
other two considerations are obvious. 
When these restrictions are placed upon the selection 
of an analogue it becomes virtually impossible to find a 
good one in the relatively limited archives of available 
weather maps. The very necessity of considering the 
upper-air flow patterns greatly shortens the available 
historical record. To this objection the proponents of 
the analogue method offer the rebuttal that it is pos- 
sible to estimate with sufficient accuracy the appear- 
ance of the upper-air flow patterns from surface maps. 
They also maintain that while no analogies are ever 
complete, the discrepancies can be evaluated subjec- 
tively and allowance can be made for differences. 
From the long-range view perhaps the most damaging 
blow to the analogue technique is that it is at best a 
substitute for understanding—and without understand- 
ing it is unlikely that meteorology can elevate itself 
above the low plateau of skill obtainable by analogue 
methods. Moreover, there is no evidence that this 
plateau is any higher than that of other more physical 
methods. 
Physical Methods in the Circulation Prognosis. With 
the increase in upper-air data brought about largely by 
the progress in aviation there has come a better under- 
standing of the general circulation. Of course the com- 
plete solution of the general circulation problem con- 
tains the key to long-range weather forecasting. Prog- 
ress along these lines is described elsewhere in this 
