306 Royal Society. 



31°"6, and that of the dew-point 3°'2. The fraction of saturation 



19 

 on this occasion was r^. and the factor derived from this observa- 

 tion was 3"36 ; this being much the largest deviation from the 

 adopted mean 2-53. 



The cause of this discrepancy is doubtless owing to the heat that 

 the wet-bulb thermometer derives from the radiation of surrounding 

 objects ; and were observations sufficiently numerous, it might con- 

 duce to accuracy were the factors calculated for every degree of 

 difference in the value of {t — t'). 



We purpose instituting a comparison between two wet-bulb ther- 

 mometers placed in similar boxes, the one box coated with lamp- 

 black, the other with a polished metallic surface. 



Below 32° our results do not appear to coincide with the factors 

 deduced from the Greenwich observations ; and the causes of these 

 discrepancies I must leave to time. 



As, however, we have had considerable experience at these tempe- 

 ratures, I may perhaps be doing service to observers in bringing 

 before their notice two causes of error, to which we have found our- 

 selves particularly liable when the thermometer is near 32°. 



1st. If the air is a little above, and has been below 32°, there will 

 frequently be a small button of ice at the foot of the wet-bulb ther- 

 mometer, which is not easily perceived, and which will keep it at 

 32° when the temperature of evaporation is really above that point. 



2ndly. It is well known that under certain circumstances water 

 may be cooled below 32° without freezing ; and an example will 

 perhaps best show the error which this fact may occasion. 



Let us suppose that the temperature of the air is 27°, and that 

 when the thermometer is wetted it sinks to 26°, and then rises. 

 Should it rise very slowly, or not at all, the probability is that 26° 

 is the true temperature of evaporation, but if rapidly, the rise may 

 be due to the conversion of the water into ice ; and it will be prudent 

 to observe whether or not the thermometer again commences to sink. 



We have frequently observed this phenomenon, and I am quite at 

 a loss to what to ascribe its uncertainty. 



It has occurred both in a high wind and a calm (the thermometers 

 are protected from the full force of the' wind), and it also appeared to 

 be quite uncertain at what temperature the water might freeze. 



I am obliged to admit that the limits of certainty of the factors 

 below zero are not quite so close as could be desired. This is partly 

 attributable to our being obliged to reject many observations made 

 with a thermometer which was broken before its index- errors were 

 fully ascertained ; but Mr. Campbell and I must claim the indul- 

 gence of those who know the difficulty of taking observations 

 requiring so much time and accuracy at such temperatures, and fre- 

 quently at six o'clock in the morning. 



Believe me, &c., 



W. Noble, 

 Lt. R.N. 



C. R. Weld, Esq., Assist. Sec. Royal Soc. 



