PRETEND GENES 41other (as if they were random combinations of Mendelian genes) and just
add together. Th en we can take it that individual diff erences are just dif-
ferences in the sums of such genes in those individuals. Th at is, the over-
all ge ne tic eff ect on variation— the heritability— can be treated as if it
were random variations of such sums, so long as the size of the ge ne tic
eff ect can be separated from the environmental eff ect. And that can be
achieved because it will produce certain observable patterns of resem-
blance among known relatives. If the size of the eff ect is large, then iden-
tical twins should be closely alike, nonidentical twins or ordinary siblings
less so, cousins still less so, and random individuals least of all.
So the heritability could be calculated aft er all. (Th e paper was actu-
ally titled “On the Correlation Between Relatives on the Supposition of
Mendelian Inheritance”). Or so it was thought. As for humans, Fisher
reached the remarkable conclusion that “an examination of the best avail-
able fi gures for human mea sure ments shows that there is little or no
indication of non- genetic causes,” and that “the hypothesis of cumula-
tive Mendelian factors seems to fi t the facts very accurately.”^2
Fisher, however, noted in the paper that “throughout this work it has
been necessary not to introduce any avoidable complications.” One of the
complications he mentioned was the possibility that factors might not just
add together like in de pen dent weights or forces; that they might actually
have eff ects on one another (called gene- gene interaction). Also, their
eff ects may be diff er ent in diff er ent environments— what is called gene-
environment interaction. Either complication screws up the formula.
Another complication is that environmental eff ects can be controlled
in animal experiments. In practice, this usually means attempting to ran-
domize environments, so that every individual genotype has an equal
chance of experiencing every environment. Th at may be reasonable for
cows in a fi eld, or hens in a chicken shed, but it is clearly not pos si ble in
humans. Th e equality of environmental eff ects across diff er ent kinds of
relatives just has to be assumed, and I return to that matter below. Such
simplifying assumptions are common and oft en useful in science. In later
life, however, Fisher seemed to realize he had devised a spherical horse
solution when he admitted that “[heritability] is one of those unfortunate
short- cuts which have emerged from biometry for lack of a more thor-
ough analy sis of the data.”^3
This content downloaded from 139.184.14.159 on Tue, 17 Oct 2017 13:51:26 UTC