The nomogram figures

Veale et al. (2015) published four nomograms, one each for flaccid length, stretched length, erect length and erect circumference. Each is the same idea drawn four times: a mean, a spread around it, and a curve that lets you place a single measurement on that spread without doing arithmetic.

What a nomogram is doing here

Each chart plots a measurement value on one axis against a percentile on the other, with a curved line connecting them. The general idea of a nomogram - a chart that substitutes for a calculation - is not specific to this subject; the paper applies it to four measurements built from the pooled clinician-measured sample. Find your figure on the measurement axis, follow the curve, read the percentile.

The four charts

Flaccid length. Built from the largest subset of the pooled data, because flaccid measurement was the most commonly reported state across the included studies.

Stretched length. Built from studies that reported a stretched-flaccid figure specifically, taken to a defined resistance point rather than a guess at "as far as it goes".

Erect length. Built from a smaller subset than flaccid, because far fewer studies measured men in a full erection under clinical conditions - erect data are the thinnest part of the pool.

Erect circumference. The girth nomogram, built from studies reporting circumference at a defined site, again a smaller subset than the length charts.

Reading one, schematically

Every nomogram in the paper has the same shape: a roughly S-curve running from low measurement and low percentile at one end to high measurement and high percentile at the other, steepest through the middle where most men sit and flattest at both ends where few do.

That shape is exactly what a normal distribution's cumulative curve looks like, because that is what the nomogram is - the mean and standard deviation of a normal curve, redrawn as a lookup rather than left as two numbers. The steep middle is why a small measurement error near the mean swings your percentile far more than the same error out in the tails.

Why the four charts are not interchangeable

A percentile read from the flaccid nomogram and a percentile read from the erect nomogram describe positions on two different curves, built from two different subsets of the pooled sample with their own mean and spread. A man can sit at a noticeably different percentile on each of the four charts - flaccid length correlates only loosely with erect length at the individual level, so a high flaccid percentile does not predict a matching erect one. Reading the wrong chart, or reading one figure against another measurement's curve, produces a percentile that does not mean what it looks like it means; each chart is only valid for the exact measurement, state and site it was built to represent. Treat each of the four as a standalone reference rather than four views of the same underlying number, and a man's four percentiles as four separate, only loosely related facts about him rather than one figure repeated four times.

What the chart does not tell you

A nomogram assumes the underlying distribution is close enough to normal for the curve to be trustworthy, an assumption the paper checked rather than asserted. It also inherits everything about who was in the pooled sample - the chart cannot know whether you resemble the men who were measured, only place your figure against the ones who were.

Using a number instead of a curve

If you have a measurement and want the arithmetic behind the curve rather than the picture, the mean, the standard deviation, and the three-step calculation that gets you from one to the other does the same job with numbers. Either way, what comes out is a percentile against a specific measured population, which is a different kind of statement from a rating. A rating like the one Rate Cock produces from a photograph is a judgement placed on a scale, not a length placed on a curve built from a ruler; an AI model scoring a photo is doing pattern recognition, not measurement, and a numeric score built for comparison between photos answers a different question than a nomogram does, as does a human judge's read of a photo, which carries no percentile at all.