256 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1958 
The factors affecting the average air temperature are interrelated ; 
there are in fact what electronic engineers call “feedback” relation- 
ships among them. ‘These feedback linkages are both positive and 
negative. For example, an increase of air temperature from whatever 
cause would result in a melting of part of the snow and ice cover of 
the earth and a corresponding reduction in the reflection. Conse- 
quently, the amount of absorbed radiation would increase and the 
temperature would rise still further. This is a positive feedback. 
Similarly, an increase of average air temperature would increase the 
evaporation from the oceans; hence the water-vapor content of the air 
and absorption of infrared radiation. The temperature would not 
increase without limit, however, because an increase in evaporation 
must eventually result in an increase of cloudiness as the water vapor 
condenses; hence an increase in the proportion of refiected sunlight. 
This is a negative feedback. Such complex feedback linkages tend to 
hunt or oscillate, with time constants determined by the speed of the 
different processes involved. 
Thus far we have been discussing comparatively small changes in 
average air temperature over the earth. Such changes may be of 
great significance—it is generally estimated by meteorologists that 
a 4-degree drop in average air temperature would be sufficient to bring 
on anew ice age. But with present meteorological observing facilities 
they would be almost impossible to measure. What is observed are 
local changes of much greater magnitude. These must be brought 
about chiefly by variations in atmospheric and perhaps oceanic cir- 
culation, specifically in the locations of north-south transport of heat 
and matter. 
For example, the January mean temperature at Spitsbergen in- 
creased by 24 degrees from 1913 to 1937, whereas during almost the 
same period there was a 3 to 5 degree drop in January mean tempera- 
ture in the Great Basin of the western United States. 
Because of the complex relationship between the amounts of inso- 
lation and infrared absorption on the one hand and the circulation 
patterns of north-south transport on the other, it is by no means cer- 
tain whether an increase of insolation or absorption would bring on 
a colder or a warmer climate. 
PAST CHANGES 
The circulation patterns are profoundly affected by the distribution 
of continents and oceans, and therefore it is of great importance to 
make comparative studies of past climatic changes in the Northern 
and Southern Hemispheres, because of their markedly different pat- 
terns of sea and land. Since changes in the intensity of the north- 
south circulation should have different effects in high and low latitudes, 
