358 HENRY A. ROWLAND 



which for the constants used in Eecknagel's table becomes 



Error = d -- I = .000 000 33 1 29000 780^ + 3f \. 

 clt 



This amounts to nearly one per cent at 0, and thence decreases to 

 45, after which it increases again. As only 0-2 at the 40 point 

 produces this large error at 0, it follows that an error of only 0-02 

 at 40 will produce an error of y^nro at 0. At other points the errors 

 will be less. 



Hence extreme care must be taken in the comparison and the most 

 accurate apparatus must be constructed for the purpose. 



2. DESCRIPTION OF APPARATUS 

 The Air Thermometer 



In designing the apparatus, I have had in view the production of 

 a uniform temperature combined with ease of reading the thermom- 

 eters, which must be totally immersed in the water. The uniformity, 

 however, needed only to apply to the air thermometer and to the bulbs 

 of the mercurial thermometer, as a slight variation in the temperature 

 of the stems is of no consequence. A uniform temperature for the air 

 thermometer is important, because it must take time for a mass of air 

 to heat up to a given temperature within 0-01 or less. 



Fig. 1 gives a section of the apparatus. This consists of a large 

 copper vessel, nickel-plated on the outside, with double walls an inch 

 apart, and made in two parts, so that it could be put together water- 

 tight along the line a &. As seen from the dimensions, it required 

 about 28 kilogrammes of water to fill it. Inside of this was the vessel 

 mdefghkln, which could be separated along the line d Ic. In the 

 upper part of this vessel, a piston, q, worked, and could draw the water 

 from the vessel. The top was closed by a loose piece of metal, o p, 

 which fell down and acted as a valve. The bottom of this inner 

 vessel had a false bottom, c I, above which was a row of large holes ; 

 above these was a perforated diaphragm, s. The bulb of the air ther- 

 mometer was at /, with the bulbs of the mercurial thermometers almost 

 touching it. The air thermometer bulb was very much elongated, being 

 about 18 cm. long and 3 to 5 cm. in diameter. Although the bulbs of 

 the thermometers were in the inner vessel, the stems were in the 

 outer one, and the reading was accomplished through the thick glass 

 window u v. 



