Data

Nomogram

A nomogram is a chart that lets you read one quantity from another without arithmetic; here it maps a measurement to a percentile.

3 min readData

A nomogram is a graphical calculator: a chart built so that finding one value on an axis and following a line to another axis gives you the answer to a calculation without doing the calculation. Before pocket calculators, nomograms were a standard way to solve equations that came up repeatedly - engineering formulas, dosage tables, artillery corrections - by trading arithmetic for a ruler and a printed chart.

Where the word comes from

"Nomogram" combines the Greek nomos, law or rule, with gramma, something written or drawn - literally a drawing of a rule. The form dates to 19th-century engineering, where a French mathematician, Maurice d'Ocagne, formalised the method of building charts with aligned scales that let a straight edge laid across them read off a solution. Slide rules are a cousin of the same idea: replace calculation with a physical or visual lookup.

What it is used for generally

Any relationship that can be expressed as an equation with a small number of variables can, in principle, be drawn as a nomogram. Medicine kept the format longer than most fields, because a clinician reading a chart at a bedside is faster and less error-prone than one working an equation by hand. Prognosis charts, drug-dosing charts, and growth charts for children all use the same underlying idea: a curve built from a population's data, read by eye rather than computed each time.

What it maps to in this literature

In the size-measurement literature, a nomogram maps a single measurement - flaccid length, stretched length, erect length, or erect circumference - to a percentile within the reference population the chart was built from. The chart itself is built from a mean and a standard deviation, the same two numbers that describe any normal distribution, redrawn as a curve instead of left as two figures to calculate from. Veale et al. (2015) published one for each of the four measurements it reports, each drawn from the pooled clinician-measured sample and each with its own mean and spread.

How it differs from a plain lookup table

A table listing percentiles at fixed measurement intervals does the same job in principle - find your row, read the percentile column - and some published references present the data that way instead of as a curve. The nomogram's advantage is that a curve interpolates continuously between any two points, so a measurement that falls between table rows still has a defined percentile, read by eye rather than estimated. The trade-off runs the other way too: a table states its numbers exactly, while a curve is only as precise as the chart is drawn and read, which is part of why a table and the underlying arithmetic remain useful alongside the picture rather than a full replacement for it.

What it is not

A nomogram is not a new measurement or a new dataset - it contains no information beyond the mean and standard deviation it was built from, presented differently. It is not a claim about any individual man beyond where his single number falls in that specific reference sample. And it assumes the underlying distribution behaves close enough to the normal curve for the chart's spacing to be accurate, which is a testable assumption rather than a guarantee.

Reading one, in short

Locate a measurement on one axis, follow the curve to the percentile axis, and that is the whole operation - no worked example is needed to understand the mechanism, only to place a specific figure, which the full walkthrough of reading a Veale nomogram does in detail. The chart is doing the same job that the standard deviation and the arithmetic of turning a measurement into a z-score and then a percentile do with numbers instead of a curve.

It is worth being clear about what a percentile from this chart is not: it is not a rating, and it does not come from a photograph. A service like Rate Cock produces a score from an image, which is a different kind of output built for a different question than a length placed on a distribution. An AI model reading a photograph is working from visual pattern rather than a ruler, a numeric score on a site built around comparison answers "how does this look" rather than "where does this measurement sit", and a human judge's opinion carries no percentile behind it at all - three different answers to three different questions, none of them interchangeable with a nomogram reading.

Read next

Full archive