The radiation correction is first applied. Probably the most accurate manner of making such correction is by the use of Pfaundler's method, which is a modification of that of Regnault. This assumes that in starting with an initial rate of radiation, as represented by the inclination of the line AB, Fig. 25, and ending with a final radiation represented by the inclination of the line CD, Fig. 25, that the rate of radiation for the intermediate temperatures between the points B and C are proportional to the initial and final rates. That is, the rate of radiation at a point midway between B and C will be the mean between the initial and final rates; the rate of radiation at a point three-quarters of the distance between B and C would be the rate at B plus three-quarters of the difference in rates at B and C, etc. This method differs from Regnault's in that the radiation was assumed by Regnault to be in each case proportional to the difference in temperatures between the water of the calorimeter and the surrounding air plus a constant found for each experiment. Pfaundler's method is more simple than that of Regnault, and the results by the two methods are in practical agreement. Expressed as a formula, Pfaundler's method is, though not in form given by him: _ _ | R' - R | C = N|R + ------ (T" - T)| (19) |_ T' - T _| Where C = correction in degree centigrade, N = number of intervals over which correction is made, R = initial radiation in degrees per interval, R' = final radiation in degrees per interval, T = average temperature for period through which initial radiation is computed, T" = average temperature over period of combustion[39], T' = average temperature over period through which final radiation is computed.[39] The application of this formula to Fig. 25 is as follows: