1903.] Specific Heats, especially at Low Temperatures. 179 



result of the comparison it appeared that the relation between the 

 Beckmann scale and the corrected scale of the secondary might be 

 taken as linear throughout. In obtaining the value of a degree on the 

 Beckmann in terms of a degree on the secondary standard, it was 

 found necessary to exercise considerable care, as the observed values 

 for the ice-point of the secondary varied between 0'61 and -0'04 

 according to its previous state. The temperature t of the secondary 

 was accordingly calculated from the reading T in accordance with the 

 equation t = 100 (T t - T<f)/(T m - T 100 ).* One degree on the Beckmann 

 (kept, as regards the amount of mercury in the bulb, in a constant 

 state during all except the earliest experiments) was found to be equal 

 to r-009 on the secondary standard. 



The value of the Beckmann degree was however more satisfactorily 

 ascertained by comparison with a Baudin thermometer, No. 12772, the 

 corrections for which are very accurately known from the work of 

 Schuster and Gannon, t The comparison was made at two points, 

 distant each about half a degree from one of the extremities of the 

 Beckmann scale. A considerable number of readings were taken in 

 the neighbourhood of each of the two points ; all were rejected except 

 those which indicated in the case of each thermometer a very slowly 

 rising temperature. The corrections of the Baudin for calibration and 

 unequal division, though small, were relevant. The only other correc- 

 tion requiring consideration was that for stem temperature. Since in 

 the specific heat determinations no correction for the exposed part of 

 the Beckmann (scale-length 5) is applied, the uncorrected Beckmann 

 should be compared with the corrected Baudin. The Baudin was 

 exposed from division 12, and the temperatures of comparison were 

 approximately 15 and 19J. On the range of 4|, the correction for 

 exposed stem might possibly approach 1 part in 1000, but would 

 probably be much less than this. Disregarding the stem correction, 

 the following values were obtained for the Beckmann degree in terms 

 of the Baudin degree, 1-0101, 1-0117, 1-0110, 1-0086, 1-0093, 1-0103, 

 1-0113, 1-0109, 1-0111, 1-0110, mean 1-0105. Making an allowance 

 for the stem correction, this has been called 1-011. To reduce an 

 interval 2 B on the Baudin (used vertically) to the corresponding interval 

 t on the French hard-glass scale, the following relation has been given 

 by Schuster and Gannon, t = t B (1 - -00135). The Beckmann degree 

 is therefore equivalent to 1-010 on the French hard-glass scale of the 

 mercury thermometer. The results which follow are expressed in the 

 first instance in terms of the French hard-glass scale. An interval on 

 this scale lying between 14 and 20 is reduced to the scale of the 



* ' Phil. Trans.,' A, 1895, p. 428. 



t 'Phil. Trans.,' A, 1895, p. 432, et seq. t also 'Owens College Book of 

 Standards.' 



VOL. LXXII. O 



