428 Sir James Dewar [Jan. 28, 



from below of this small improvised observatory is given in Fig. 11. 

 The batten supporting the manometer was pivoted and adjustable, 

 as in the portable form, in order to control the position of the liquid 

 index : but no equalising flask or other similar attachment was 

 necessary in this protected position. The actual opal scale was sup- 

 ported in a small rack adjustable by a pinion for easy and accurate 

 setting, with a low-power lens arranged above. The space available 

 was quite sufficient to allow re-filling and all necessary tests and 

 attention without any disturbance. 



The instrument was tested by exposure to a light metal vessel 

 above, filled with alcohol cooled to known temperatures to - 80 C, 

 and with water up to + 100° C. When it was properly adjusted 

 the results were found to be within about 5 percent, of the theoretical 

 values deduced from Stefan's law, the greatest deviations being 

 with the higher temperatures. The metal vessel for containing the 

 cooled alcohol, etc., was conical, and the bottom concave and 

 blackened. It fitted over the open top of P. 



During the Discourse, to demonstrate the working of the thermo- 

 scope, small churn-shaped vessels, covering the open neck of the 

 flask, were rested on the base plate D. They also were concave 

 below (to avoid direct contact with the flask neck). The response 

 obtained by exposure to such a vessel containing water at the tem- 

 perature of the Lecture Theatre was 2 * 5 cm. of the projected scale in 

 fifteen seconds. With the water at 95° to 98°, the displacement was 

 6*8 cm. in the same time. With the "churn " filled with boiling 

 carbonic acid paste, 30 seconds' exposure was necessary for a dis- 

 placement of l'O cm. When liquid oxygen replaced the carbonic 

 acid the index was stationary. New liquid air, being about ">', 

 lower, caused a small but persistent negative movement. When a 

 current of hydrogen was passed through a capillary nozzle in this 

 liquid air, the negative movement rapidly increased, us the lowering 

 of temperature caused the exposed thermoscope charcoal to absorb 

 more air, as contrasted with the previous expulsion of air when 

 exposed to higher temperatures. 



With the regular use of the instrument, the readings did not 

 vary appreciably, even though the level of the liquid oxygen in the 

 flask was allowed to fall several centimetres. Only when the fall 

 was sufficient to allow a rise in temperature of the diaphragms H 

 and the shoulder of the flask did the value begin to increase at all. 

 The stray radiation correction thus introduced can of course be 

 determined at any time by making a blank observation with a small 

 plain vessel of liquid oxygen over the flask neck. The amount so 

 determined is then deducted from all measurements to get the 

 correct Z/E ratio, which would otherwise be too high, and as a result 

 give deduced zenith temperatures that would not be sufficiently low. 

 An average working condition would be with the flask from two- 

 thirds to three-fourths full, and with up to 5 or cm. depth of liquid 



