134 REPORT— 1890. 
within less than ‘02° of the true path at all points. It agrees closely 
with the curve obtained by Mr. Callendar from the parabola 
ons t 
xb ino) = r00 | 
by measuring one-tenth of the ordinate along the abscissa.’ 
The following equation, however, represents its path more accurately. 
y=018795t— 00019914? + 000000111523. The curve itself is shown 
in Chart A, fig. 12. 
fees 
} 
| 
2A SSN Be ce ea Te St NRO Ng 
16 . 20 24 28 a2 | 36 40.44. 48 52. 56. 60 68 72. 76 80. 84.88. 92° "96° 100 
We proceeded to test our conclusions by comparison with thermo- 
meters standardised at Kew; for this purpose a rotating annular ring, 
through the centre of which the platinum thermometer passed, was 
inserted in the lid of the tank, in such a manner that the mercury ther- 
mometers, fixed in holes bored near its circumference, could successively 
be brought into the field of view of the kathetometer without any re- 
adjustment of the telescope; the thermometers were then read by one 
observer, whilst the platinum resistances were taken by the other. The 
freezing-points were not, however, determined by this method, but by 
direct immersion in powdered ice, adopting the precautions recommended 
by Guillaume in his ‘ Thermométrie de Précision.’ 
The following curves were then drawn, which indicate the result of 
the comparison of our platinum thermometer with those standardised at 
Kew. 
Curve Thermometer, Kew No. | Standardised 
B 75148 October 1888. 
C 4 75149 October 1888. 
D | 43762 May 1885. 
E | 8394 December 1880, January 1882, April 1888. 
‘ It must be remembered that Callendar’s difference curve gives the connection 
between platinum and air thermometer temperatures, whilst Regnault used a mercury 
thermometer (M.A.S. XXI.), and thus curve A gives the relation between platinum 
and mercury thermometer temperature. 
