336 JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 
should be continued fully five minutes. It is necessary to emphasize 
this point, because one reason why the dew-point fails to indicate the 
minimum temperature of the coming night is faulty manipulation. 
The theory that the succeeding minimum will not be below the dew- 
point presupposes that both general and local conditions will remain 
constant, and while in general at times of frost conditions do remain 
unchanged, there are exceptions. If the humidity is high when the 
observation is taken near sunset, the chances are that the dew-point 
will be two or more degrees centigrade above the night minimum. If, 
however, the station is in the lowlands and the instruments not far 
above the soil, the readings will not fairly represent the tempera- 
tures of the lower air mass. It would be better to get the dew- 
point at higher levels. 
The forecaster should have before him a rather detailed topographi- 
cal map with contours for every ten metres. He must appreciate the 
trend of valleys and differentiate between valleys more or less walled 
in and those which are open, particularly those open to the sea. The 
inclination of ridges to the prevailing winds is of some importance. A 
comparatively slow flow of air will prevent frost unless the air in its 
passage over higher land has been dynamically dried. Such air in its 
descent may gain in heat, but not enough to offset the cooling due to 
contact with the soil cooling rapidly by radiation. The forecaster, 
then, carries a mental picture of basins filling with dry cold air from 
rather shallow streams. These ponds are only a few metres in depth, 
and are capped by a layer of much warmer air widely spread. And 
this is what we mean by inversions ; the cold air is down and cannot 
get up ; the warm air is up and cannot get down, and there is marked 
stratification within a few metres from the ground. 
Besides the topographical map the forecaster must know something 
of the nature of the soil, for this is important in determining what may 
be called the effective radiation. Dark soils and certain covers * will 
radiate more rapidly, and hence show lower temperatures. The general 
forecaster can hardly be expected to be an expert for all localities, 
hence it is a good plan to have in selected districts a local forecaster, 
preferably an agriculturist who knows something of soils, and also the 
relative susceptibility of different plants to frosts. He will know 
where the land has been ploughed and where it has been left unbroken, 
the crop cover, the presence of wind-breaks, and the minor slopes. 
For crops on hillsides the factor of safety is large. Protection 
comes from both mixing the air and the warmth derived from the 
great reservoir of warm air which we have described as lying above 
the cool air. During the night hours the air moves toward the hill- 
sides, and whatever convectional changes take place, the resultant 
* The streakiness of frost even on a nearly level field may be due, nrst> 
to differences in thermal capacity of soil and cover, and second, to the minor 
circulations thus established. For example, over a dark soil the cold air will 
settle, forcing warmer air elsewhere. And as even a slow flow will prevent 
frost, we find frost-free areas not far from frosted areas. 
