the specific heats of certain solids, 273 



the same time possess the requisite subdivision of degrees the bulb 

 must be large, or the stem very long. Could we employ a cylindric- 

 al metallic containing vessel, fitted up with an apparatus to measure 

 its own longitudinal expansions with perfect accuracy, it would per- 

 haps be the best kind of thermometer for such experiments. The 

 specific heat of mercury, at least within the range where a thermom- 

 eter for our present purpose would be used, is, according to the four 

 independent detemiinatlons of Lavoisier, Kirwan, Crawford and Du- 

 long, .0327. The specific heat of glass given by six different phi- 

 losophers is at a mean .18511, that of Irvine being .2000, and that 

 of Kirwan .1740 at the extremes. By three trials on flint glass in a 

 method hereafter to be referred to, I obtained a mean of .17854, 

 which is less than the above mean result by .00657 and more than 

 that of Dulongand Petit by .00154. 



If the scale be of brass we have its specific heat by the mean re- 

 sult of Wilke, Crawford and Dalton's determinations .11276, but as 

 the conducting power of that metal is high as well as its rate of ex- 

 pansion it ought if possible to be avoided as a part of the immersed 

 thermometer. 



The thermometer which measures the heat of the solid before 

 immersion, should be faithfully compared with that which is used in 

 the water. Thermometers of extensive range are often found inac- 

 curate from containing minute portions of air. It would for this rea- 

 son be desirable to compare their indications with the fusing points 

 of tin and lead, as well as the boiling points of water and mercury. 

 To be sure of at least two points in the temperature of the hot body 

 it will be well to place it in an iron vessel containing mercury, im- 

 mersed in boiling water, for that point, and in a bath of melted tin 

 immersed in boiling mercury to get the utmost range of temperature 

 measurable by that liquid. By forming a suitable covering for the 

 bath of mercury, and providing for the exit and condensation of its 

 iumes we may operate with perfect convenience in the method just 

 described. 



6. I have already mentioned the necessity of confining the range 

 of temperature taken by the water during these experiments. If we 

 terminate the experiment but one or two degrees above the actual 

 temperature of the room the loss by radiation and conduction on one 

 side will in general be so nearly counteracted by the gain on the oth- 

 er, as to influence very little, the actual result. But if we employ 

 too small a vessel the high temperature of our solid may give too 



Vol. XXVII.— No. 2. 35 



