4.28 M. Berthelot on a New Thermometer for measuring 



taken two by two*. It suffices, therefore, to note exactly the 

 length of these interspaces, once and for all, in the first experi- 

 ment ; so that, by taking one of the four fixed points and rede- 

 termining its position immediately before using the instrument, 

 the position of the remaining points may be ascertained. When 

 the instrument is about to be used, one of the points, the zero, 

 for example, or, better still, the 100°-point, is redetermined, and 

 by relaxing the screws pp, the scale is moved along the stem 

 ////until the fixed point marked on the scale exactly corre- 

 sponds to the position of the mercurial column. The tube is 

 then once more fixed to the scale by means of the screws. 



3. Graduation of the Scale. 



The most exact method of graduating the instrument consists 

 in jotting upon curve-paper the temperatures taken for fixed 

 points, the length of the spaces being expressed in millimetres ; 

 these points are then connected by a curve after the well-known 

 method. The curve at one end is prolonged to +500°, and at 

 the other to —100°. The abscissae will then represent the 

 spaces on the scale expressed in millimetres, and the ordinates 

 the temperatures. The latter are transferred to the graduated 

 scale and engraved to the right of the tube ////, parallel to the 

 millimetre-divisions. 



When the capillary tube is of uniform calibre, the length of 

 each degree diminishes very slowly with the increment of tem- 

 perature. The variation becomes less when the bore of the ca- 

 pillary tube changes. This may even accidentally produce such 

 a compensatory effect that the length of each degree remains 

 constant; in fact I possess such an instrument. 



But the above method of graduation may be considerably sim- 

 plified when the amount of error to which the instrument is ne- 

 cessarily subject (reaching 2° or 3° at high temperatures) and 

 the fact that it is specially adapted for temperatures ranging be- 

 tween 300° and 500° are taken into account. In fact it is suf- 

 ficient to divide the space between 350° and 440° into ninety 

 equal parts ; each of these divisions represents consequently a 

 degree of temperature. This graduation is extended on the one 

 hand to 300°, and on the other to 500°. Between these limits 

 the graduation obtained in this manner agrees sufficiently well 

 with that resulting from the curve deduced from the fixed points ; 

 the differences do not exceed the experimental errors. 



When the thermometer is graduated according to this method, 

 it is sufficient to mark the 0° and 100° points without dividing 

 the intermediate space into degrees ; these points are intended to 



* Provided, of course, that the capillarity of the tube be perfectly uniform. 



