Warning: Totally speculative
This is an attempt at a coherent explanation for why left-handedness (and right-handedness) exist in the distributions that they do.
Handedness is a rather exceptional human trait. Most animals don’t have a dominant hand (or foot.) Horses have no dominant hooves; anteaters dig equally well with both paws; dolphins don’t favor one flipper over the other; monkeys don’t fall out of trees if they try to grab a branch with their left hands. Only humans have a really distinct tendency to use one side of their bodies over the other.
And about 90% of us use our right hands, and about 10% of us use our left hands, (Wikipedia claims 10%, but The Lopsided Ape reports 12%.) an observation that appears to hold pretty consistently throughout both time and culture, so long as we aren’t dealing with a culture where lefties are forced to write with their right hands.
A simple Mendel-square two-gene explanation for handedness–a dominant allele for right-handedness and a recessive one for left-handedness, with equal proportions of alleles in society, would result in a 75% righties to 25% lefties. Even if the proportions weren’t equal, the offspring of two lefties ought to be 100% left-handed. This is not, however, what we see. The children of two lefties have only a 25% chance or so of being left-handed themselves.
So let’s try a more complicated model.
Let’s assume that there are two alleles that code for right-handedness. (Hereafter “R”) You get one from your mom and one from your dad.
Each of these alleles is accompanied by a second allele that codes for either nothing (hereafter “O”) or potentially switches the expression of your handedness (hereafter “S”)
Everybody in the world gets two identical R alleles, one from mom and one from dad.
Everyone also gets two S or O alleles, one from mom and one from dad. One of these S or O alleles affects one of your Rs, and the other affects the other R.
Your potential pairs, then, are:
RO/RO, RO/RS, RS/RO, or RS/RS
RO=right handed allele.
RS=50% chance of expressing for right or left dominance; RS/RS thus => 25% chance of both alleles coming out lefty.
So RO/RO, RO/RS, and RS/RO = righties, (but the RO/ROs may have especially dominant right hands; half of the RO/RS guys may have weakly dominant right hands.)
Only RS/RS produces lefties, and of those, only 25% defeat the dominance odds.
This gets us our observed correlation of only 25% of children of left-handed couples being left-handed themselves.
(Please note that this is still a very simplified model; Wikipedia claims that there may be more than 40 alleles involved.)
What of the general population as a whole?
Assuming random mating in a population with equal quantities of RO/RO, RO/RS, RS/RO and RS/RS, we’d end up with 25% of children RS/RS. But if only 25% of RS/RS turn out lefties, only 6.25% of children would be lefties. We’re still missing 4-6% of the population.
This implies that either: A. Wikipedia has the wrong #s for % of children of lefties who are left-handed; B. about half of lefties are RO/RS (about 1/8th of the RO/RS population); C. RS is found in twice the proportion as RO in the population; or D. my model is wrong.
According to Anything Left-Handed:
Dr Chris McManus reported in his book Right Hand, Left Hand on a study he had done based on a review of scientific literature which showed parent handedness for 70,000 children. On average, the chances of two right-handed parents having a left-handed child were around 9% left-handed children, two left-handed parents around 26% and one left and one right-handed parent around 19%. …
More than 50% of left-handers do not know of any other left-hander anywhere in their living family.
This implies B, that about half of lefties are RO/RS. Having one RS combination gives you a 12.5% chance of being left-handed; having two RS combinations gives you a 25% chance.
And that… I think that works. And it means we can refine our theory–we don’t need two R alleles; we only need one. (Obviously it is more likely a whole bunch of alleles that code for a whole system, but since they act together, we can model them as one.) The R allele is then modified by a pair of alleles that comes in either O (do nothing,) or S (switch.)
One S allele gives you a 12.5% chance of being a lefty; two doubles your chances to 25%.
Interestingly, this model suggests that not only does no gene for “left handedness” exist, but that “left handedness” might not even be the allele’s goal. Despite the rarity of lefties, the S allele is found in 75% of the population (an equal % as the O allele.) My suspicion is that the S allele is doing something else valuable, like making sure we don’t become too lopsided in our abilities or try to shunt all of our mental functions to one side of our brain.