JAN. 19, 1923 WHITE: ELECTRIC HEATING OF CALORIMETERS 23 
quently, the two triangular areas are not independent of each other, 
and (6) does not hold. Indeed, blank heatings alone might be used 
in most cases. The preceding analysis, however, seems generally 
useful in giving an idea of the sources of error, and of the precautions 
that must be taken if there is an initial thermal head, or if the jacket 
temperature changes, matters whose purely empirical investigation 
will generally be very tedious and of inferior accuracy. 
Extension for Initial Thermal Head 
An initial thermal head, that is, a difference between calorimeter and 
jacket at the time, A, will cause a thermal leakage which will be simply 
added to the one already considered, due to the heating. This leakage 
may properly be taken as modifying the initial thermal head which 
causes it, but as having no other effect. The two thermal heads, each 
with its resultant leakage, may therefore be considered as separate. 
The measurement made at the time of H in fig. 1, however, takes them 
together. If the simple approximation of (5) is then used the rise 
due to the heating, to which the approximation is applied, is under- 
estimated, and no.account is taken of the change in ¢, as such. Greater 
accuracy would therefore be obtained by adding again the fall in 4¢,, 
7 
multiplied by a +n — L, and subtracting the (integrated) real loss of 
thermal head due to the diminution of ¢,._ The fallin ¢, is ¢,K (7+7), 
: ae 4 T+n\ 
the loss of integrated thermal head from it is ¢, K 5 ; the 
application of the method of (5) is thus easily found to give a result 
which is too low by 
wk(3—z)@4n (7) 
in addition to (6). 
Change in jacket temperature can be similarly treated. If the 
jacket is electrically heated, so as to give a nearly adiabatic method, 
its lag, as usually in adiabatic work, will be as important as the lag 
in the calorimeter. 
Initial Change in Heating 
When the current is first turned into the heater the resulting change 
of temperature of the wire will generally cause an initial change of 
resistance and the value for the first few seconds will be different from 
