Races of Rattus norvegicus labelled A and B give characteristic curves relating angle of orientation θ during geotropic progression to the inclination (α) of the surface. The orientation-angles for B are higher at every slope, and the threshold slope for orientation is lower in B. When these races are cross-bred, the F1 progeny show a θ-α curve in general corresponding to that for the B parent, as regards both threshold slope and magnitudes of θ.
Differences between the curve for F1 and B have to do (1) with a slight but significant distortion of the curve, such that from α = 15° to about (α = 35° the mean curve is slightly above that for B, whereas above this slope of surface, θ is consistently below that for B, and (2) below α = 35° the horizontal latitude of variation in the curve θ vs. log sin α is much greater than above α = 35°. The first distortion is interpreted as due to the fact that, as a manifestation of heterosis, developmental processes which upon the one hand lead to growth in bulk in the posterior region of the body, and on the other which would lead to a proportionate development of tension receptors in the legs, fail to keep pace harmoniously. In ordinary development of individuals in a pure line, harmonious relationships between these two aspects of growth are maintained. The variation of response (θ) is not altered in F1, if attention is paid to the total observed variation; but the unmodifiable or uncontrollable portion of the total variation of θ, not affected as a function of sin α, is increased. This fact is more pronounced in female offspring from the matings A ♂ x B ♀. This effect in F1 is interpreted as due to the introduction of modifying genetic influences, which affect variation of orientation. The percentage of modifiable variation is reduced from 74 per cent (mean) in B to about 59 per cent in F1.
These interpretations are checked by the behavior of offspring produced in the backcross F1 x B. The disharmony of developmental processes thought to be signalized by the distortion of the form of the θ-α curve disappears completely in these backcross individuals. It can be computed that an equivalent distortion of the θ-α curve for B would be produced by the posterior attachment of a mass of about 0.5 gm., unaccompanied by its proportion of receptors in the legs; in backcrossing to the B line, this should be reduced in the average to about "0.25 gm.," which is at about the threshold for any detectable effect of a mass added posteriorly.
The variability of θ in (F1 x B) is not different from that for B, or for F1 (total variation); but, as must be expected if the unmodified variation has been increased in F1 by the action of modifying genetic influences not directly connected with the genes determining the numbers of tension-receptors, we find that in (F1 x B) the percentage of modifiable variation is returned half-way (65 per cent) toward that for the B line.
The bearing of certain considerations derived from the treatment of this case is discussed with reference to the desirability that other phenomena of "hybrid vigor" and heterosis be similarly analyzed.