[Sess. 
178 Proceedings of the Royal Society of Edinburgh. 
between TO and 1*5 ; (3) those in which the error is less than 1. Tabulate 
alongside each of these numbers the corresponding excess of the R.B. 
reading over the air temperature. Adding and taking the average values 
for each group, we find as follows : — 
Average 
Error. 
Average 
Excess. 
No. of 
Cases. 
Group (1) 
2-8 
47 
33 
Group (2) 
1-1 
36 
7 
Group (3) 
0*4 
22 
30 
Comparing groups (1) and (3), we see that with an average increase of 
25° in the excess of the R.B. temperature above the air temperature, the 
average error of the Stevenson screen thermometer is increased sevenfold. 
From the middle of August to the end of October nearly 50 per cent, of the 
morning readings of the thermometer in the Stevenson screen were in 
error by more than 1°*5, the average being 2° *8. — C. G. K.] 
In addition to the regular morning and evening readings, many others 
were taken at different hours during the rise of temperature, and these all 
tell the same tale as to the lag in the records of the Stevenson screen 
thermometer. 
A comparison of the figures given here with the results given in 
previous papers shows that the conditions at Ardenlea were much more 
trying for the Stevenson screen than those at Darroch, where the early 
observations were made. The situation at Darroch had a much freer 
exposure ; but there are sure to be some situations with a worse exposure 
than the Ardenlea tennis lawn. It is only on the top of a hill or on 
a wide perfectly open space that the temperature within the Stevenson 
screen can be kept down to near the true air temperature. Even at the 
Ben Nevis Observatory the tests showed that the Stevenson screen was 
influenced by radiation. 
There are still two points to be settled with regard to the C screen, 
namely, the size of the upper screen and the size of the bulb of the 
thermometer. Both must be standardised if different screens are to give 
comparable and concordant readings. The screen used in the present 
investigation was small, the upper protecting screen being a circular disk 
19 inches (48 cm.) in diameter, and the lower 14 inches (36 cm.). The two 
under screens a and b were 4*6 inches (11 ’6 cm.) in diameter. With so 
small a screen the sun’s rays at early morning and late evening will fall 
on the bulbs and tend to heat them. This does not, however, happen near 
