GEOPHYSICAL LABORATORY. 137 



3. Lags have a law of their own; they usually vary with K, are independent 

 of T and 4>, and can usually be made to cause little or no error. 



4. The stirring introduces a possible error from irregular heat production, 

 due to irregular speed. Since the heat varies as the curve of the speed, the 

 error tends to vary as the cube of the efficiency of stirring, and thus may 

 become very important in delicate measurements. A remedy can usually be 

 found by controlling the speed, or, more effectively, by diminishing Ki as 

 already noted. The type of stirrer, and the design of the calorimeter with 

 reference to stirring, may be very important. Evaporation is also touched 

 upon, and a discussion is given of the error from heat-flow along wires or rods. 



(28) Calorimetric methods and devices. Walter P. White. J. Am. Chem. Soc, 40, 1887- 



1898 (1918). 



In this paper various forms of jacket covers (and of stirrer mountings) are 

 described and compared. 



Vacuum-jacketed vessels are mainly useful either in relatively crude or in 

 very exacting work. They are convenient and simple in crude work, but 

 cause difficulty and complication if the work to be done demands more elabor- 

 ate arrangements. The adiabatic method does not diminish either of the 

 two main sources of thermal leakage error, but has advantages in a dozen 

 other waj^s. The Pfaundler twin-calorimeter method still enables high pre- 

 cision to be attained with relatively simple apparatus, but is most successful 

 when applied to liquids and to the comparing of two things which are nearly 

 alike. Aneroid calorimeters work quicker, and within limits also more ac- 

 curately, the smaller they are. The measured shield method, where a con- 

 vection shield supplied with thermocouples is used, diminishes very greatly 

 the difficulties of thermal head measurement and stirring, which are among 

 the greatest in work of high precision, and is more convenient than the use 

 of glass vessels. 



(29) Calorimetric lag. Walter P. WTiite. J. Am. Chem. Soc, 40, 1858-1872 (1918). 



The lag effects of bodies external to calorimeters, although more compli- 

 cated in expression, are found to follow the same general laws as the simpler 

 lag effects already described. More carefully stated than previously, these 

 effects prove to be three: (1) equivalent to the heat capacity of the calorim- 

 eter; this can be eliminated if a calorimeter is directly calibrated; (2) much 

 smaller, depending on the amount of thermal leakage; this can be avoided, 

 if necessary, by using the adiabatic method; (3) dependent on the jacket 

 temperature, which disappears for constant jacket temperature; this one 

 causes the calorimeter to have a different effective heat capacity when used 

 adiabatically. 



Of two errors that are peculiar to the aneroid calorimeter, one comes 

 from improper distribution of the thermo-j unctions (or equivalent devices) 

 which measure the surface temperatures. This error becomes zero if the sur- 

 face temperature distribution does not change from one experiment to an- 

 other. The other special error is inconstancy of final temperature due to 

 inconstancy of final temperature of the jacket. 



On account of the lag effect the effective heat capacity of a shield midway 

 between calorimeter and jacket is only one-fourth the actual capacity, and 

 with due regard to the possibility of change, such shields may often with great 

 profit be used to reduce the thermal leakage. Used as a cover, such a shield 

 has a specially small error, and offers a particularly easy method of dealing 

 with evaporation. Ordinary non-metallic covers, on account both of their 

 lag and their heat capacity, are very undesirable. 



