White — Melting Point Methods at High Temper atitres. 487 



When the latent heat is small, as in most inversions, the 

 variations in the values of G, in the furnace rate, etc., are also 

 small, and the radiation method is easily made fairly accurate. 

 Two determinations of the inversion heat of wollastonite, made 

 with the simple 2'5 gram charge, agreed to 0*3 calories, and a 

 very generous estimate of the possibilities of systematic error 

 does not allow more than 1 *5 calories.* 



The experimental arrangement for quasi-calorimetric deter- 

 minations usually does not differ in any respect from that for 

 simple melting points at constant values of G. The determi- 

 nation illustrated on p. 484 was, in fact, quasi-calorimetric. For 

 the greatest accuracy the equipotential leakage shield (p. 474) 

 should be as complete as possible, and should be on the inclos- 

 ing 4 cm tube, instead of connected to the crucible itself.f 

 The heat supplied to the charge within any interval is deter- 

 mined simply by adding the values of Q% for that interval ; 

 since each value of G corresponds to one minute, this gives at 

 once the value of TG. A correction for the observed varia- 

 tion in K may also be needed. The other steps in the cal- 

 culation do not seem to call for special description. 



IV. Small thermal effects. — Some thermal effects, espe- 

 cially inversions in the solid state, are so slight that their mere 

 detection is a problem in itself. Determinations of this 

 character, like the preceding, depend on accurate heat meas- 

 urement, though here the behavior of the charge is no longer 

 a complication. The difficulty is rather to secure, in the fur- 

 nace itself, conditions so uniform and reproducible that very 

 slight variations occurring within the charge may be recog- 

 nized at once. This requires, of course, chiefly precision in 

 temperature measurement and regulation. It is also promoted 

 by increasing the size of the charge, thus increasing the tem- 

 perature difference, furnace to crucible, and diminishing the 

 effect of small temperature irregularities. 



The feeble intensity of these thermal effects is usually due 

 as much to the slowness of the heat absorption or evolution as 

 to its small magnitude. Hence a rapid furnace rate, which 

 concentrates the effect, greatly aids in its detection. The 

 wollastonite inversion, for example, with a rate of 10° per 

 minute, can be accomplished in about 10 minutes with an 

 uncertainty of less than 1*5 calories (2*5 gram charge). Under 

 like conditions two calories should therefore be capable of 

 detection. 



*The result was 10"0 ± 1*5 calories. 



f The more elaborate arrangements referred to in the footnote to page 486 

 are not properly within the scope of this paper. 



£ Really, the values of the temperature difference between center of 

 charge and furnace, which is greater than G by the amount of the gradient 

 in the charge, but proportional to it. 



