Observations made with a Black Bulb Thermometer. 85 



the period May-August the variation of cloud appears to have very- 

 little influence upon the behaviour of the black bulb, and that the 

 maximum at 50 per cent, shown by Table III. is due entirely to 

 the readings during the eight months September-April. By com- 

 paring the solar maxima with the difference of maxima between sun 

 and shade, we elicit the curious circumstance that from January to 

 April the temperature in the shade is practically unaffected by any 

 quantity of cloud below 50 per cent. ; whereas during the period 

 May-August the shade temperature shows a tendency to fall as the 

 amount of cloud becomes greater, and actually to rise somewhat 

 during September-December as the amount of cloud increases from 

 zero to 50 per cent. 



Seeing that the temperature of the dew-point, and the relative 

 humidity, both rise as the amount of cloud increases, in each of the 

 three terms of the year, the explanation is not by any means self- 

 evident. 



For the purpose of getting a better idea of the effect of moisture 

 upon the temperatures as shown by a black bulb thermometer 

 in vacuo, the observations under an absolutely clear sky have been 

 separated from the rest and considered alone. Some such process 

 seems to be wanted because for small amounts of cloud the actual 

 places of the clouds in the sky will be of the first importance. 

 According to both Stow and Park Harrison clouds near the sun 

 increase " radiation," whereas clouds near the horizon can have but 

 little influence. Thus a mere cloud percentage may be misleading. 

 A further important advantage of a consideration of clear skies only 

 is that the maxima in sun and shade fall at very nearly definite 

 hours, and are therefore much more readily comparable. Sequences 

 of temperature under clear skies have been made for dew-points and 

 humidities at noon in ascending order of magnitude, after the 

 algebraic addition of a monthly constant which raises the monthly 

 mean temperatures to the mean of the year. 



Table V. gives the temperature variations corresponding to 

 assigned dew-points. It should be explained that the dew-point 

 28° includes really all dew-points under 30° ; the dew-point 33° all 

 dew-points from 31° to 35° ; the dew-point 38° all dew-points from 

 36° to 40° ; and so on ; all over 50° being classed with the dew- 

 point 53°. 



In this Table the values corresponding to a dew-point of 53° are 

 somewhat doubtful, partly because they are obtained from a very 

 few observations ; but the other columns are better. Our impres- 

 sion from the Table is that Stow's result is correct (even if his way 

 of getting it is not quite satisfactory), and that the difference of 



