PRESERVATION OF CROPS FROM INJURY BY FROST. 335 
an equally rapid in-moving high. There results boisterous west to 
north winds, subsiding as night approaches. Then comes a period of 
stillness and there is no mixing or churning of the air. With the 
cessation of convectional currents there is also diminishing amount of 
vapour and haze or dust. The first law of frost-forecasting is that 
conditions must favour the surface inversion of temperature character- 
istic of still, clear nights with dust-free atmosphere and low humidity. 
Air can flow in any direction, and the forecaster will follow closely 
the great displacements in horizontal directions ; but in dealing with 
frosts of the late autumn and early winter he must be keenly alive to 
the action of slow, vertical currents, chiefly descensional. These slow- 
moving surface currents, which might be called creeping currents, com- 
plete the fall in temperature which the larger horizontal movement 
began. Marked inversions take place after the cyclonic winds have 
calmed down. Then begins what might be called local air drainage, 
and the formation of ponds of stagnant dry cold air. On windy nights 
there is little likelihood of frost, and the temperature is as a rule higher, 
because there is thorough mixing. It should not, however, be for- 
gotten that wind favours evaporation, and, unless from a water surface, 
tends to dry and cool vegetation. Again, temperature is largely 
controlled by humidity, and if there is much moisture present the 
amplitude or range will be small. More important still, the rate of 
cooling will be a minimum. Therefore, when high humidities are 
reported the probabilities are that there will be no frost. On nights 
when frost is anticipated, data showing temperature gradients in a 
vertical direction will help the forecaster. If special readings can be 
made of two similar thermometers not in shelters, one 100 millimetres 
above the soil and the other a metre higher, these will help. On 
frosty nights the minimum temperature is found to occur not on the 
soil itself, but about 10 millimetres above. With regard to data from 
wet-bulb thermometers there is some uncertainty, because unless a 
sufficiently low dew-point be obtained the readings may mislead. The 
trouble is that, as generally obtained, the wet-bulb reading is not low 
enough, because proper provision has not been made for determining 
the amount of air passing over the wet bulb in a given time. While 
sling psychrometers and whirled psychrometers improve matters 
somewhat, still there is no way of knowing definitely how much air 
has passed over the wet bulb. It is assumed that the distance travelled 
by the wet bulb is practically equivalent to an air flow of three or four 
metres per second, and that this is adequate ventilation. But, of 
course, there should be a definite time and a definite radius of rotation, 
or else each observer will report a depression of the wet bulb, depending 
upon his own idea of what constitutes a sufficient ventilation. With 
the whirled psychrometer it is possible at small expense to improve 
matters by adding a small toothed disk and a cheap spring bell. Thus 
an instrument which the writer uses sounds the bell for every 100 
revolutions, and with watch in hand the speed can be easily regulated 
to 100 revolutions a minute. To get the true depression the whirling 
