Tongue Rolling: Why the Textbook Traits Are Wrong

If you learned genetics at school, there is a fair chance the lesson involved rolling your tongue. Rolling was dominant, non-rolling recessive, and the class counted hands to prove it. Earlobes, widow’s peak, hitchhiker’s thumb and which thumb lands on top when you clasp your hands got the same treatment. These traits were chosen because they were easy to see in a classroom, not because the evidence for them was good. Most fell apart within a decade or two of being proposed, and the geneticist who introduced tongue rolling spent the rest of his career trying to take it back.

The tongue-rolling story

Alfred Sturtevant, who had drawn the first genetic map of a chromosome as a student in Thomas Hunt Morgan’s fly room, published “A New Inherited Character in Man” in 1940. His family data suggested that the ability to roll the tongue into a tube was dominant, with some exceptions he noted but set aside. The trait went straight into textbooks.

In 1952 Philip Matlock tested 33 pairs of identical twins. In seven of them, one twin could roll and the other could not. Identical genomes, different tongues: whatever determines rolling, it is not simply genes. A year earlier Taku Komai had followed Japanese schoolchildren and found that the proportion who could roll rose from 54 percent at six or seven years old to 76 percent at twelve. A fifth of the population, in other words, learns. In Komai’s families, a third of the children of two non-rollers could roll, which under Sturtevant’s model should never happen.

Sturtevant conceded. In his 1965 A History of Genetics he wrote that he was embarrassed to see the trait “listed in some current works as an established Mendelian case”. Textbooks kept it anyway.

The rest of the list

• Earlobes. Two 1937 papers reached opposite conclusions about which form was dominant, and one of them pointed out that lobes vary continuously rather than falling into two types. Family studies in the 1960s found free-lobed children born to two attached-lobed parents. In 2017 a genome-wide study of nearly 75,000 people found 49 loci associated with lobe attachment, including EDAR, a gene better known for its effects on hair and teeth. Earlobe shape is heritable in the way height is heritable: many small effects, no single switch.
• Hitchhiker’s thumb. The 1953 study that proposed a recessive allele included families in which neither parent had the bendy thumb and a child did. Thumb angle is a continuous measurement with an arbitrary cutoff.
• Widow’s peak. No properly conducted family or twin study supports a single-gene model. Hairlines vary continuously and change with age.
• Hand clasping. Across 18 family studies, about a third of the children of two right-thumb-on-top parents put the left thumb on top, which a dominant-recessive model forbids. Among identical twins, more pairs disagree than agree on the left-on-top form, and fraternal twins look almost the same. The genetic contribution, if any, is small.
• Cleft chin, dimples and hair whorl direction. Anecdote, no replicated model.

Two classroom traits survive partially. Eye colour is close to a single-gene story for the blue-versus-brown split, but two blue-eyed parents can have a brown-eyed child, as our eye colour explainer describes. Bitter tasting of PTC really is mostly one gene, TAS2R38, though it has three linked variants and a middle group that the two-box version leaves out.

Why the myths stuck

A dominant-recessive model is hard to falsify with casual observation. It fits any family in which at least one parent has the trait, which is most families. Only the exceptions - two non-rollers with a rolling child - count against it, and those were easy to wave away as misclassification or incomplete penetrance. Add the fact that the visible, harmless, genuinely single-gene human traits are rare, and teachers reached for what was available.

The real single-gene traits you can see are a short list, and it is telling that few of them are classroom staples: wet or dry earwax, red hair for the most part, the alcohol flush, and adult lactose tolerance. Each traces to one gene, each is visible in a raw file, and none involves rolling anything.

What this means for your raw file

There is no rsID for tongue rolling, earlobe attachment or widow’s peak, and any tool that reports one is extrapolating from evidence that does not exist. Genespiral does not list these traits for that reason. A polygenic predictor for earlobes could in principle be built from the 2017 results, and it would still explain only a small share of the variation between people.

The wider lesson is the same one that runs through modern genetics. The human traits that a single letter can meaningfully predict are a small and specific set - the ones above, plus some drug responses and a few serious diseases. Almost everything you see in the mirror comes from many variants, an environment, and sometimes plain practice. Our articles on genotype and phenotype and polygenic scores pick up where the classroom left off.

If you want to replicate the science yourself, ask a sibling to roll their tongue. If one of you can and the other cannot, you have reproduced Matlock’s result, with a smaller sample.

This article is educational only.

References

• Sturtevant AH. A new inherited character in man. PNAS. 1940. PubMed 16588317
• Komai T. Notes on lingual gymnastics: frequency of tongue rollers and pedigrees of tied tongues in Japan. Journal of Heredity. 1951;42:293-297.
• Matlock P. Identical twins discordant in tongue-rolling. Journal of Heredity. 1952;43:24.
• Sturtevant AH. A History of Genetics. Harper and Row, 1965.
• Shaffer JR, et al. Multiethnic GWAS reveals polygenic architecture of earlobe attachment. American Journal of Human Genetics. 2017. PubMed 29198719
• Reiss M. The genetics of hand-clasping: a review and a familial study. Annals of Human Biology. 1999.
• McDonald JH. Myths of Human Genetics. Sparky House Publishing, 2011. Online edition, University of Delaware.