384 Transactions of the Boyal Society of South Africa. 
Upon the rare occasions when a single observation was missed during 
the evening, values were interpolated for that hour. If more than one 
observation was missed during an evening, that evening's observations 
were rejected altogether. Of the whole number of evenings in which a 
complete set of hourly observations was obtained, the seasonal distribution 
was as follows : — 
Summer. 
Winter. 
Total. 
Very clouded nights ... 
41 
52 
93 evenings 
42 
43 
85 „ 
Clear nights 
46 
144 
190 
From which it is manifest that a preponderance of winter observations 
has come in for the clear night results. 
For our immediate purpose it may be sufficient to assume that the 
temperatures of the two thermometers are governed first by radiation of 
heat from the lower thermometer to space, and second by the conduction 
— evidently slow — of heat from the level of the upper thermometer to that 
of the lower. The actual circumstances are certainly much more complex, 
but they may be regarded as reducing to that, as a first approximation. 
On this assumption the greater the intensity of radiation the greater will 
be the fall of temperature of the lower thermometer below what it would 
otherwise have shown ; but at the same time the fall of temperature of 
the upper thermometer will also be the greater since the temperature 
gradient is greater. A steady difference of temperature between the two 
thermometers will be attained when the flow of heat along the temperature 
gradient in the space between them is sufficient to counterbalance the 
radiation. It is the purpose of this investigation to discuss the meteoro- 
logical conditions that determine the said steady difference. 
The velocity of the wind cited in the last three Tables was measured 
by a Eobinson Anemometer whose cups are 45 feet above the ground. On 
the ground itself the air movement will have been much less. 
The given relative humidity is for the standard height of 4 feet, and 
clearly, therefore, if the aqueous vapour content of the lower air is fairly 
uniform, the relative humidity must be much greater near the cold surface 
of the ground than it is at a height of 4 feet. This is demonstrated by the 
fact that dew will often begin to form on the ground when the standard 
relative humidity is only 70 per cent. 
In Table 5 is tabulated the number of times dew (or hoar-frost) was 
noted in the period, of 15 months or so, under review, corresponding to 
the state of the sky and the relative humidity. 
