worth while to study more carefully this decomposition under 

 atmospheric pressure, and for this purpose tlie following ai)})aratus 

 and methods were devised: 



EXPERIMENTAL METHOD. 



The apparatus finally adopted for the experiments is sketched 

 in fig. 1. For the thermostat a mixture of sodium and potassium 

 nitrates was used, contained in a cylindrical vessel of enameled 

 iron ("agate ware"), about 30 centimeters in diameter and 25 

 centimeters in height. (A in fig. 1.) This vessel rested on a 

 tripod and was heated from below by two Bun sen burners. The 

 heating efficiency and the constancy of the thermostat were sur- 

 prisingly increased by a sheath of asbestos (B) surrounding the 

 bath and projecting 10 or 15 centimeters below it. This sheath 

 was separated about one-half centimeter from the side of the 

 vessel l:)y a number of corrugated brass strips not shown in the 

 figure. An annular space was thus left between the asbestos and 

 the bath through which the hot gases from the burners could pass 

 freely and thus heat the thermostat from the sides as well as from 

 below. The careful adjustment of this sheath seemed to be 

 almost essential to the satisfactory operation of the thermostat. 



The stirrer was made very simply Avithout the use of solder, 

 combining the ordinary method of stirring with the centrifugal, 

 and it was so effective that the different parts of the bath never 

 differed by more than one or two hundredths of a degree. 



The temperature regulator was simply a large mercury ther- 

 mometer made of a spiral of thin glass sealed to a vertical 

 capillary (C). Into this capillary two fine platinum wires were 

 sealed so that when the expansion of the mercury brought it 

 into contact Avith the upper wire an electric current from a storage 

 battery was made, Avhich, passing through a telegraph instrument, 

 shut off a thin rubl:)er tube supplying gas to one of the Ijurners. 



This regulator was in fact the only part of the apparatus that 

 caused the slightest inconvenience. The spiral being made of thin 

 glass and holding nearly a kilogram of mercury was very fragile, 

 and, being weakened further by the continued high teniperature, 

 broke several times and had to be replaced. If a regulator of the 

 same type could l)e made of some suitable metal, which would 

 also have the advantage, because of its high conductivity, of 

 responding more rapidly to changes in the temperature, a very 



