546 On a Practical Thermometry Standard. 



difference due to b more than a hundred times as great 

 in proportion), refer to a similar quality of porcelain. The 

 experiments of Holborn and Deville refer to the mean co- 

 efficient of expansion of porcelain between 0° and 1000° C. 

 It will be observed that Bedford's formula agrees very fairly 

 with their results at 1000°, but that the formula found by 

 Chappuis would make the correction much larger. This is 

 another indication that his value of b is too high. If we 

 were to take instead the value found by Bedford, the S. B.P. 

 would be lowered by 0°'64. It seems possible from the 

 above considerations that the uncertainty due to expansion 

 correction alone may be sufficient of itself to account for the 

 whole of the difference between our results, and that the lower 

 value may be the more accurate, so far as this particular correc- 

 tion is concerned, because we cannot regard the extrapolated 

 results with the same confidence as those deduced from ob- 

 servations made at the S. B.P. itself. A similar uncertainty 

 affects the small correction for the difference between the 

 constant-volume and constant-pressure scales, which is at 

 present deduced from an empirical formula founded on obser- 

 vations between 0° and 10j° 0. There does not appear to be 

 any prospect that these difficult questions will be absolutely 

 and finally settled in the near future, but in the meantime it 

 is highly desirable, in the interests of science, to adopt a 

 definite standard scale to which such questions may be re- 

 ferred. As a matter of practical convenience it would cer- 

 tainly be preferable to adopt without change the old value 

 444°\53 C. for the S. B.P., which has now been in use for 

 nearly ten years for graduating platinum-thermometers. By 

 adopting this course a great deal of confusion would certainly 

 be avoided, and we should secure a scale which appears to 

 give the most probable rpsnfrs at high temperatures. In pro- 

 posing this course I do not wish to appear to attach an ex- 

 cessive value to old results obtained with comparatively crude 

 apparatus. I am fully conscious that these results might be 

 improved. But the old value is probably at least as accurate 

 as the parabolic difference-formula itself, the need for a 

 practical standard is immediate, and it is not worth while to 

 risk delay and confusion in making an uncertain correction, 

 which might lead to less satisfactory results at higher points. 

 The annexed table is an illustration of the close agreement 

 between the proposed B.A. scale and the results of all 

 the best observations. Some higher points are given in the 

 last two lines of Table II. It would be possible to state many 

 of these points more definitely if a particular sample of platinum 

 had been selected for the standard thermometers. 



