| 
7 of water, and as regards intermediate points Regnault’s determinations 
of the temperature and pressure of aqueous vapour were adopted. The 
precautions observed in the construction of the apparatus, and in the 
method of observation, are described. The thermometer was tested by 
_ comparison with a number of thermometers standardised at Kew. The 
: curves, showing the result of these determinations, are in remarkably 
. close agreement, and when the observations were sufficiently numerous 
: 
ON STANDARDS FOR USE IN ELECTRICAL MEASUREMENTS. 131 
it appeared possible to calibrate the bore as accurately as by the usual 
more laborious process. The further advantage of this method is that 
thermometers can be compared under the conditions in which they are 
to be used. 
_ Ina communication to the Royal Society read on June 19, 1890, 
{ described a method of constructing and graduating platinum ther- 
mometers, and gave a table of boiling and freezing points for various 
substances lying between 100° and 500°, determined by means of these 
instruments. 
Subsequent observations indicate that a slight change appears to be 
taking place in the readings of these thermometers. I attribute this (1) 
to alterations in the glass, (2) to presence of moisture in the tube—the 
asbestos roll on which the spiral was wound being highly hygroscopic. I 
therefore decided to construct a thermometer in which there should be 
mo contact between the glass and the platinum, and which should be 
thoroughly dry and hermetically sealed. 
I was unable to discover any suitable non-conductor capable of resist- 
ing high temperatures ; but in anthracene (melting-point 213°) I found a 
substance suitable in every respect for use at low temperatures. I sub- 
jected a sample to severe tests, and, up to a temperature of about 130°, 
found it to be a better insulator than paraffin. 
_ The leads to the coil were constructed of silver, the inner one a rod 
and the outer a tube. The resistance of these leads was about ‘001 ohm, 
and therefore any change in the external resistance, caused by change of 
temperature, might be disregarded. The silver leads approached to within 
about 1 inch of the spiral, and were connected to it by moderately thick 
platinum wires ; thus a flow of heat from the spiral to the silver was 
diminished. The wire forming the coil was about 56 inches in length, 
and had a diameter of -005 inch. The spiral was about 2 inches long, 
having a resistance of about 13:5 ohms at 0° C., and the external diameter 
at the covering tube was about ‘3 inch. The ends of the asbestos roll 
were made of greater diameter than the portion on which the spiral was 
vound, and thus there was no glass contact. The tube and contents were 
veated up to a temperature of several hundred degrees, and dried air 
“passed through for some hours. It was then exhausted and the open 
end placed under the surface of melted anthracene, which was allowed to 
rise until nearly in contact with the coil. When cool, the whole of the 
thermometer from the spiral to the upper end (about 13 inches) was a 
Solid mass, while the spiral and asbestos roll were perfectly dry and in an 
almost vacuous space. I have taken nearly 600 observations with tbis 
thermometer and cannot detect any signs of change. When the lower 
‘part was undergoing rapid chan ges in temperature, thermo-electric effects 
Showed themselves, but by reversing the battery and galvanometer 
‘connectiong.during each reading these effects were eliminated. A low- 
resistance galvanometer was used, and the current which passed through 
ithe thermometer when determining its resistance did not exceed one 
