the Elements Carbon, Boron, and Silicon. 171 



were as accurately (and much more easily) made as would have 

 been the case had a telescope been employed. Throughout the 

 entire research a small oscillation in the mercury thread of the 

 calorimeter was noticed (owing to the difference of melting-point 

 of the outer and inner ice of the instrument) , varying between 

 00 millim. to 020 millim. per minute in winter while the ca- 

 lorimeter was surrounded with snow, and O'O millim. to 0*45 

 millim. per minute in summer when the calorimeter was sur- 

 rounded with pure lake-ice. The correction necessary on account 

 of this oscillation in an experiment lasting twenty or thirty mi- 

 nutes would be considerable ; I therefore endeavoured to bring 

 it about that the time which elapsed until the heat had been 

 entirely spent in melting ice should be as short as possible. 

 This time was reduced to about ten or fifteen minutes by allow- 

 ing not more than 4 or 5 grms. of ice to be melted at once, and 

 by causing the formation of a new quantity of ice by means of 

 the introduction of a fragment of ice or snow, and the addition 

 of a drop or two of alcohol. 



In order to obviate a small loss of heat by the temperature in 

 the receiving- vessel of the calorimeter rising above 4°, instru- 

 ments of three sizes were used. For measuring quantities of 

 heat varying from to 50, from 50 to 100, and from 100 to 200 

 heat-units, the receiving- vessel of the instrument employed was 

 capable of containing 8, 16, and 30 grms. of water respectively. 

 When the quantities of heat taken up by substances cooled to 

 — T in rising to 0° had to be determined by means of the for- 

 motion of a corresponding amount of ice, alcohol was used instead 

 of water in the receiving-vessel, care being taken that the proxi- 

 mity of these cold bodies did not bring the temperature of the 

 snow or ice-film in the calorimeter beneath its melting-point. 



For the measurement of such high temperatures as 500°-1000°, 

 as I could not readily obtain a suitable air-thermometer, I made 

 use of the device to be described. This method may not yield 

 such altogether accurate results as those of the air-thermometer ; 

 yet it is suited for solving, in a measure at all events, all the 

 problems connected with the specific heat of elements at high 

 temperatures. Supposing that the amount of heat necessary to 

 raise a unit weight of platinum from the temperature T to the 

 high temperature T (measured by the air-thermometer) has been 

 determined by carefully conducted experiments, and is expressed 

 by the function II(T — T ), then it is possible to calculate the 



temperature T by determining the amount of heat, W T , given 

 up by the unit weight of platinum in cooling, in the calorimeter, 

 from the unknown temperature T to the known final tempera- 



