THE MEASUREMENT OF TEMPERATURE. 69 



meet, and I conclude are not intended to meet, the demands 

 of exact inquirers. Very wisely they, in no case, give the 

 correction term beyond 0'Oi°C, for they do not supply any 

 of the data (as far as my knowledge goes) for making the 

 corrections rendered necessary by change in position, in 

 external pressure, in the temporary change of zero conse- 

 quent on the preceding temperature of the thermometer, 

 etc., and as the combined effect of such corrections alone 

 may considerably exceed o'Oi°C. it is evident that until the 

 department supplies further information it has carried its 

 method of correction to its limits. I am afraid that the 

 issuing of those Kew certificates, which give the second 

 decimal figure of the correction, so far from being a preven- 

 tion is a cause of inaccuracy, as, unless used with a full 

 knowledge of the variations consequent on changes in the 

 conditions, they impart a false confidence to their possessors. 

 In this section I have endeavoured to show that although 

 it is possible to accurately determine ordinary temperatures 

 by means of a mercury thermometer, the process is an intri- 

 cate and unsatisfactory one at the best, and that even under 

 the most favourable circumstances, the determination of a 

 single temperature requires considerable expenditure of time 

 and attention. The removal of the thermometer from its 

 initial position, its immediate immersion in ice, the complete 

 immersion of the stem, the measurement of the external 

 pressure, etc., make up a series of operations which the condi- 

 tions of an experimentusuallyrender difficult, if not impossible. 



SECTION II. 



A consideration of the difficulties inseparable from the use 

 of mercury thermometers would naturally lead to the conclu- 

 sion that if some more simple but accurate form ot thermo- 

 meter, free from such unavoidable causes of error, could be 

 devised, it would meet with general acceptance. The plat- 

 inum thermometer, as now constructed, is an almost ideal 

 instrument, and is especially free from those defects insepar- 

 able from mercury thermometers. It can be used under 

 circumstances which forbid the employment of a mercury 

 one, and it surpasses that instrument^both in accuracy and in 



