THE MEASUREMENT OF TEMPERATURE. 67 



necessarily the case. Occasionally all the signs are similar, 

 and the effect is considerable, amounting in some cases 

 that I have observed to nearly o'4°C. Again, the cor- 

 rections in the case of hard glass thermometers are smaller 

 than when soft glass is used, and nearly all the ther- 

 mometers used in our English laboratories are of the 

 latter material. 



It is therefore obvious that the endeavour to obtain one 

 accurate reading by means of a mercury thermometer in- 

 volves considerable labour, and presupposes a prolonged 

 and exhaustive study of the thermometer used. Many 

 observers disregard these precautions and defend themselves 

 by the following statements : ( 1 ) that certain of these correc- 

 tions are uncertain and possibly imaginary ; (2) that cases 

 requiring such extreme accuracy are rare. The opinion of 

 such an authority as Dr. Guillaume ought alone to be a 

 sufficient reply to No. 1. It is, however, an objection 

 which I confess to having at one time entertained, and I 

 find it is very prevalent. The best answer that I can make 

 is to give particulars of a case which has come under my 

 notice within the last few weeks. 



In February, 1893, I received from the International 

 Bureau of Weights and Measures the thermometer by Ton- 

 nelot, No. 11,048, and in Phil. Trans., A, vol. clxxxiv., pp. 

 427-433, I have already published a full account of the com- 

 parison of this thermometer with my own standards. In the 

 autumn of last year I received from Professor Threlfall of 

 Sydney University a request to procure for him, if possible, 

 thermometers fulfilling certain specified conditions. Three 

 thermometers were accordingly prepared by Tonnelot, 

 standardised at the International Bureau and they reached 

 my hands the first week of June, 1894. One of these was 

 for low temperature, and I have made no comparisons 

 between it and others. No. 11,402 had a range from 

 — i° to + 2i°C. No. 11,403 had a range from - i° to 

 + i° then a bulb and then from 20" to 4i°C. Each degree 

 on these thermometers was divided into twenty parts. 



On June 8th and 9th Mr. C. T. Heycock was so good 

 as to assist me in a careful comparison of these instruments 



