172 Proceedings of the Royal Society of Edinburgh. [Sess. 
XIX. — Thermometer Screens. By the late Dr John Aitken, F.R.S. 
(The paper was finished shortly before his death.) Communi- 
cated by Dr C. G. Knott, F.R.S. 
(MS. received September 10, 1920. Read January 10, 1921.) 
It is now thirty-two years since I last brought this subject before the 
Royal Society of Edinburgh. Since then nothing has been done to remedy 
the imperfections which were then shown to characterise the thermometer 
screens used throughout the country. In some respects these imperfections 
may seem to be of small importance. Yet in these days when accuracy of 
measurement is being more and more insisted on, and particularly now 
when reconstruction is everywhere in vogue, it does seem desirable that 
apparatus largely used by thousands of observers should be as perfect as it 
is possible to make it. 
In my previous papers * it was shown that in sunshine the Stevenson 
screen always gave too high readings. The trouble taken to get Kew 
corrections for thermometers whose errors in that respect rarely exceed 
0 o, 2 or 0°*3, and the application of the corrections to readings obtained 
under conditions in which they may be easily 2° or 3° in error, do not 
commend themselves as worthy examples of scientific method — it is very 
like a case of straining out a gnat and swallowing a camel. 
Having changed to other surroundings since the former series of 
observations were carried out, I determined to repeat the work in the 
altered conditions ; but it was not till after the middle of August of this 
year (1919) that I was able to instal all the necessary instruments and 
make comparisons. Soon after I came to Ardenlea, the Stevenson screen 
had been fitted up on a lawn-tennis green more or less surrounded by 
trees, but subject nevertheless to a fairly free circulation of wind. On this 
green the new tests and comparisons were made. 
Taking the temperature of the air seems to be, at first sight, a very simple 
matter. It is, however, very far from being simple ; and, as a matter of 
fact, no one has ever attempted to define what we are to measure. 
The fine-bulb f thermometer used in previous tests showed that, when 
* Proc. Roy. Soc. Edin ., vol. xii, 1882-84 ; vol. xiii, 1884-86 ; vol. xiv, 1886-87. 
t The bulb is cylindrical, with a length of 25 mm. and a diameter of a little over 1 mm. 
It is provided with a sheath of pure silver which fits it closely. The readings of this instru- 
ment when simply placed under a sunshade were found to agree very closely with those 
given by shaded thermometers in a steady draught of air. It may be regarded as giving 
very nearly the true temperature of the air. See Proc. R.S.E., vol. xii, p. 688, 1882-84. 
