106 Mr. H. L. Callendar on the 



very small traces of either are sufficient to ruin the wire for 

 thermometric purposes. 



For accurate work at high temperatures it is necessary to 

 use platinum leads, and to inclose the coil in a tube of glazed 

 porcelain or silica. 



For insulating the coil and leads I have found nothing that 

 answers so well as mica. Biscuit porcelain is fairly good, 

 but most varieties of clay are apt to attack the wire at high 

 temperatures. If the wire is wound on a clay cylinder, the 

 relatively large mass of the clay has also the effect of mate- 

 rially reducing the sensitiveness. 



The wire is preferably doubled on itself like an ordinary 

 resistance-coil, and wound on a thin plate of mica. The 

 leads are insulated by being made to pass through a series of 

 mica wads cut to fit the tube containing the instrument. This 

 method has the advantage of giving very perfect insulation, 

 and of preventing convection-currents of air up and down the 

 tube. 



The resistance of such an instrument may conveniently be 

 measured by means of an ordinary post-office box. If the 

 resistance-coils are of German-silver wire, the temperature of 

 the box must be taken at each observation and a correction 

 applied. It is better therefore to use a box with coils of 

 copper-nickel-manganese, or one of the many other alloys 

 that do not change appreciably in resistance at ordinary 

 temperatures. 



With the above simple arrangement it is not difficult to 

 obtain results consistent to a few hundredths of a degree at 

 500° C, provided that the resistance of the leads is relatively 

 very small and fairly constant, and that the stem of the ther- 

 mometer is always immersed to nearly the same extent. There 

 are some objections, however, to the use of thick leads. They 

 are necessarily wanting in lightness and flexibility, and they 

 tend to cool the bulb of the thermometer by conduction along 

 the stem. 



For most purposes it is better to insert in the stem of the 

 thermometer a second pair of leads similar to those of the coil 

 itself, so that their resistance can be measured separately. 

 It is then possible to use leads of any convenient length and 

 flexibility, and to make the observations of temperature inde- 

 pendent of the length of stem immersed. 



In the ordinary method of measuring a resistance with a 

 post-office box, it is necessary to observe galvanometer- 

 throws in order to find the last two figures of the value of the 

 resistance. If, however, a divided bridge-wire be used in 

 conjunction with the resistance-box for determining the frac- 



