1915] OSTERHOUT—ADDITIVE EFFECTS 231 
of the figure that the growth in 40 A is 47, while that in 40 B is 
28.5 (this is equivalent to the growth in 71 A). The additive 
effect of (60 A+ 40 B) is therefore equal to the effect of (60+71=) 
131 A, which gives (as read from the curve in fig. 1) the additive 
effect 6. 
If we calculate the additive effect of the same mixture in terms 
of B, we find that the effect of 60 A is 39, which is equal to the effect 
of 24 B. Hence the additive effect of (40 B+60 A) equals the 
effect of (40 B+24 B=) 64 B, which gives as the additive effect 
19: 5 
TABLE I 
ADDITIVE EFFECT WHEN THE EFFECT OF M =THE 
EFFECT OF 2N 
Mixture _— — —e cc. to make} aviaitive effect 
TOO C0 Mes ee 1.0 
ae ; MM 90 M Rot cae aia kes 
in : 1 20.4 i ee ae 3-3 
Cae 5-0 
ve ; r | srelcoatans a es cans be 7-5 
TOO CN 60 Mee a II.o 
We have in this case, therefore, two values for the additive 
effect, namely 6 and 19.5. One is undoubtedly too high, the other 
too low. Instead of taking the mean (or the weighted mean) of 
these two values, it seems desirable to avoid this complication 
altogether by calculating A and B in terms of a third curve, C. 
This may be obtained by taking points midway between the two 
curves A and B (the distance being measured vertically) and 
drawing a line through them, giving the dotted line C. The curve 
C could be drawn in any convenient manner (it might, for example, 
be a parabola or a hyperbola), but it should have two points in com- 
mon with each of the other curves. This might be arranged by 
multiplying or dividing the ordinates or abscissas so as to make 
these curves coincide at the origin and at the half-way point with 
the arbitrary standard curve. 
