The usual target height for a child is the parents' average height plus 6.5 cm for a boy, or minus 6.5 cm for a girl, with a range of 8.5 cm either side. We applied that formula to the 934 adult children in Galton's family records whose heights were written down as numbers, and 87.9% of them finished inside the range. The 113 misses leaned in opposite directions at the two ends: on average, children of the shortest couples finished above their target and children of the tallest couples below it. The records describe 205 English families of the 1880s, with heights reported by a family member by post, and include no Asian families.
The formula casts a ±8.5 cm net around a sex-adjusted parental average
As a 2025 review in BMC Pediatrics by Ciancia, Ribes Cajas and Cools gives it, the target for a boy is (father + mother + 13) ÷ 2 and for a girl (father + mother − 13) ÷ 2, in centimetres. The review credits it to Tanner, Goldstein and Whitehouse in 1970. Tanner used 13 cm for the gap between adult men and women. The review notes that the 1990 UK growth charts show a gap of 13.2 cm and WHO's charts 13.3 cm. Some populations run wider, an Indian one at 15.8 cm. Among Galton's adult children it was 13.0 cm, on our arithmetic.
The review says the range of ±8.5 cm "reflects the within-family variance". A 2024 study in Children by Zeevi and colleagues describes it as Tanner's 3rd-to-97th centile band, meant to leave out about 3 children in 100 at each end, and says Tanner arrived at it through "theoretical considerations" rather than a count of grown children. We could not open Tanner's 1970 paper, so its wording reaches us only through these two accounts.
Tanner's band netted 88% of the Victorian sample
The GaltonFamilies dataset, transcribed from Galton's notebook, lists 963 adult children in 205 families; 29 have heights recorded as words and are left out, which may slightly under-represent the extremes. We converted the heights from inches and counted how many children finished within each width of their own target.
| Range around the target | Share of Galton's 934 adult children inside |
|---|---|
| ±5 cm | 64.0% |
| ±8.5 cm (Tanner) | 87.9% |
| ±10 cm | 92.2% |
| ±10.2 cm (AAP teaching material) | 92.9% |
| ±11.1 cm | 95.0% |
Sons fell inside ±8.5 cm less often than daughters, 86.5% against 89.4%. Resampling whole families 2,000 times, since siblings are not independent, kept the share inside ±8.5 cm between 85.6% and 90.1%. Children scattered around their targets with a standard deviation of 5.6 cm, so ±8.5 cm spans about 1.5 of them here. MSD Manuals' calculator describes the same ±8.5 cm as ±2 standard deviations, and the American Academy of Pediatrics' EQIPP teaching page treats ±10.2 cm as 2 SD instead. Zeevi's team reports that Tanner later widened his own band to ±9 cm for girls and ±10 cm for boys.
Until 14 October 2026 our mid-parental height calculator stated that a recent study had found a gap larger than ±8.5 cm in only about 3% of children. That 3% was Zeevi's probability for one worked example, a girl of nine, which the 2025 review repeated without that context. The revision of 14 October removes it, gives Galton's one in eight as our count, names the wider bands in use, and says the calculator still applies Tanner's uncorrected formula.
Short-parent families slipped through the net on the tall side
We ranked Galton's 205 couples by their average height. The shortest 20 couples had 83 adult children, who finished on average 2.6 cm above the formula's target; 15 were above the range and one was below it. The tallest 20 couples had 66 children, on average 2.3 cm below target, with 9 under the range and 2 over it. This is regression to the mean, the pattern Galton described in 1886: children of unusually tall or short parents tend to finish nearer the population average than their parents did.
In a Swedish population study (n = 2,402), Luo, Albertsson-Wikland and Karlberg (1998) reported that "the Tanner method may lead to a 6-cm error in underestimating target height" where the parents' average falls below 163 cm. Wright and Cheetham (1999), measuring 419 children aged 8–9 rather than adults, found 90% within 1.5 SD scores of their reported mid-parental height, close to our 88% within a similar width.
Our arithmetic suggests where the misses come from. A 3rd-to-97th centile band of ±8.5 cm implies a spread of about 4.5 cm, which matches the sibling spread Zeevi measured within families (4.5 cm for sons, 4.2 cm for daughters) and the 4.7 cm we found averaged over Galton's larger sibling groups. The band seems to be about the right width around a family's own centre. The extra misses come from the parental average placing that centre too far from the population mean.
Swedish regression constants snag on Victorian families
Corrected formulas shrink the parental average toward the population before adding a range. The Hermanussen and Cole method of 2003, as Ciancia's review gives it, multiplies the parents' averaged SD score by 0.72. A 2011 letter by van Dommelen, Schönbeck and van Buuren describes the Dutch version: a multiplier of 0.7519 from the 2009 growth study, built into the country's height-for-age charts. Galton's families give 0.69 for sons and 0.73 for daughters on the same scale, on our arithmetic.
Each correction carries its own reference population. Luo's equations add a fixed constant for each sex to about three-quarters of the parents' average, and that constant holds the Swedish mean. Applied to Galton's Victorian families, they came out about 2 cm too tall for sons and daughters alike (our arithmetic; the paper's full text, which we could not open, defines its input).
Taller generations escape a net set at the parents' height
A formula built on parents' heights cannot see a generation growing taller than the last, a secular trend. The Dutch letter records that the 1997 growth study added 4.5 cm to Tanner's target for this. Ciancia's review reports that Japanese researchers added 2 cm in 1990, then, under twenty years later, returned to Tanner's original, and that in Taiwanese children Tanner's formula underestimated adult height by 2.3 cm (Su and colleagues, 2007).
An apparent gap between generations can also be the parents shrinking. Zeevi's team, with 303 adult children in 23 large families, found the children 2.7 cm taller than the target on average; after correcting the parents' heights for their ages, around 62, the difference dropped to 0.06 cm. Galton's grown children matched their same-sex parents to within a quarter of a centimetre, a figure that cannot separate the two effects.
Child health records nearer home haul in no target height
Singapore's Health Booklet, in the edition dated 2014, plots children on local survey charts from 2000 and 2002 and contains no target-height method. A Malaysian Ministry of Health guideline dated February 2025 names WHO's 2006 charts, and its text, as far as we could search it, gives no such formula. Our guide to baby growth charts and percentiles covers which charts each place uses.
View author profile → · Editorial policy
