Mr. F. D. Brown's JVotes on Thermometry. 63 



and have modified accordingly the correction to be applied to it. 

 Xow, no observations of other thermometers — no curves or 

 equations representing their zero-movements — could be of any 

 assistance to me in this matter. I knew that the zero would 

 probably rise, and that the amount of the rise would not be 

 the same in my case as in that of others, and that, therefore, 

 I must obtain the index-error experimentally. 1 also knew 

 that if I boiled the thermometer I should cause irregular 

 changes in the position of the zero: and as there was no 

 necessity for the operation, I avoided boiling it. But if by 

 mischance it had fallen in boiling water, no equations repre- 

 senting the zero-movements of other thermometers would have 

 told me exactly what had happened to mine ; I should simply 

 have been obliged to observe its index-error more frequently 

 thai: before the accident happened. 



The question which seems to me to be of the greatest impor- 

 tance with regard to zero-movements is, how we can best 

 reduce the trouble which they cause us. In the case of all 

 meteorological and clinical thermometers, where the changes of 

 temperature are small, as in the above case, it is evident that all 

 we can or need do is to protect the instrument from unnecessary 

 changes of temperature. When, on the contrary, our observa- 

 tions extend over wide ranges of temperature, the difficulties 

 increase considerably. Suppose, for example, that I want to 

 use a thermometer to indicate accurately a series of tempe- 

 ratures between 70° and 90°. It is obvious that if I observe the 

 index-error beforehand, and apply the correction thus obtained 

 to my readings, I shall not be doing right ; for the very heating 

 of the thermometer to 70°-90° will have altered the index- 

 error. But if, on the other hand, I first heat the thermometer 

 to 100°, then ascertain its index-error, then make my experi- 

 ments with it, and finally observe its reading in ice a second 

 time, I shall be tolerably certain, if the index-error is the same 

 at the end as at the beginning of the experiment, that no 

 variation has occurred during the observations. 



In most laboratories, however, the frequent determination of 

 the zero-point of a thermometer involves a considerable expen- 

 diture of labour : ice has to be purchased, broken up into small 

 pieces, washed, and placed in a suitable vessel. All this 

 requires no little time, and has, moreover, to be repeated at 

 every determination, since the broken ice melts away in the 

 interval. On the other hand, the apparatus for the observation 

 of the steam-point is always in readiness ; if, therefore, no 

 greater error arises when the index-error is determined before 

 and after the experiments by means of the steam-point, a great 

 saving of time will be effected, without any corresponding loss 

 of accuracy. 



