Roicland's Thermometers with the Paris Standard. 13 



high and 15*2 centim. in diameter. At the bottom there was 

 an opening through which the excess of water could be drawn 

 off. At a distance of 7 centim. from the top of the tank two 

 openings 2 centim. in diameter were made, one on each side 

 of the tank. To these there was fitted, in a manner that it is 

 not necessary to describe, but so that it could be readily 

 detached, a tube of hard rubber or ebonite 2 centim. in 

 diameter. This ebonite tube extended horizontally across the 

 tank from one side to the other, except that at the centre of it 

 there was a little brass piece that held a sort of drum with 

 drum-heads of pure sheet rubber. These were perforated 

 with small circular holes in the centre, and a thermometer 

 could be inserted perpendicularly through them both. When 

 in position the whole tank could be filled with ice and water, 

 and very little if any water escaped through the joints. As 

 represented in the drawing, the two rubber diaphragms, or 

 drum-heads, were 3' 7 centim. apart, and a portion of the 

 thermometer-stem of that length was therefore not exposed to 

 the ice. This portion of the stem, however, must have been 

 within 0°'l of 0° C, because when the bulb of a small ther- 

 mometer which could detect a difference of o, l was placed 

 in the air in the centre of this ebonite tube, and left there a 

 short time, it indicated a temperature of 0° C. There was 

 also no dew deposited on the stem unless there happened to 

 be a draught of air blowing through the room, as happened in 

 one or two cases. 



In using the tank the ice was broken up into very fine 

 pieces by means of a w T ooden mallet in a zinc-lined wooden 

 box and transferred by a tin scoop to the zero tank. Ordinary 

 distilled water was poured on it, and it was packed dow r n closely. 

 The thermometer was then placed in position, and the tank 

 was filled to the brim with ice. The actual zero point was 

 generally observed in the middle of the tube. The water in 

 the ice was also generally kept at about the same height. 

 The actual height of the w^ater was measured by means of a 

 vertical glass tube placed alongside the tank, and communi- 

 cating with the outlet tube at the bottom. A brass frame, 

 not represented in the drawing, was fastened to the tank to 

 support the thermometer. 



The ice used was made artificially out of filtered Baltimore 

 city water. Clear pieces were chosen, and it was found to 

 be very pure by testing the water melted from it for ammonia, 

 and for dissolved substances in general by measuring its 

 electrical conductivity. The depression of the freezing-point 

 caused by impurities in the ice must have been less than 

 0°001 0. 



