Data
Two percentiles, one person
Length and girth are only moderately correlated, so most men sit at different percentiles on the two; the same percentile on both is the exception.
A man who measures at, say, the 70th percentile for length is not thereby at the 70th percentile for girth. The two figures come from the same body but behave, statistically, almost like separate measurements. Expecting them to line up is the mistake worth correcting before either number gets read at all.
What "moderately correlated" means here
If length and girth were perfectly correlated, one figure would predict the other exactly, and a single percentile would describe both. If they were uncorrelated, knowing one would tell you nothing about the other. The published data sit between these: a longer shaft has some tendency toward a larger girth, but the relationship is loose enough that it explains only part of the variation, and a great deal of room remains for a man to be well above average on one measure and unremarkable on the other. That looseness is the whole reason two percentiles exist rather than one - what a percentile itself is stated relative to matters before either number gets interpreted.
Why a moderate correlation still surprises people
The intuition that a "bigger" man should be bigger everywhere comes from comparing whole bodies, where height, weight and limb length do correlate fairly strongly with each other. Length and girth are a poor analogy to that. They are two dimensions of one structure, not two independent structures, but the correlation between them is weaker than most people assume by default, and a moderate correlation still leaves most individuals scattered rather than lined up. The practical result: pick two men at the same length percentile, and their girth percentiles will typically differ by a noticeable margin, not by a rounding error.
What a moderate correlation looks like on paper
A correlation of zero and a correlation of one are both easy to picture: a flat cloud of points with no pattern, or a single straight line. A moderate positive correlation sits between them - a diagonal trend is visible if you plotted every man's length against his girth, but the cloud of points is wide enough around that trend that any single point could sit well off it in either direction. That is the shape the published data take: a real, directional relationship, and a lot of individual variation the relationship does not capture. It is the same shape as the relationship between height and shoe size, or any two bodily measurements that share some common cause but are also each influenced by plenty of factors the other does not share - useful to know the direction of, not useful for predicting one from the other with any precision.
Reading your own two numbers
If your length and girth percentiles come out close together, that is not a sign either measurement is correct and the other is wrong - it just means you happen to sit near the diagonal in a scatter that mostly does not. If they come out far apart, that is the more common outcome, not an anomaly to explain away. Neither figure corrects the other, and averaging them into a single number throws away the information that they are, substantially, independent measurements.
This is also a case where the measurement itself has to be trustworthy before the percentile means anything - a length reading taken along the underside rather than the standard dorsal line, or a girth reading taken at the wrong site, will misplace a percentile before correlation ever enters the picture.
What this is not
This post is about the strength of the relationship between the two figures, not about the arithmetic of dividing one by the other, and not about how the same man's flaccid and erect readings relate, which is its own question with its own answer. It is also not a subjective judgement of proportion. A photograph-based score, the kind Rate Cock or a similar rating service produces, is answering a different question entirely - how something looks in an image - and does not resolve to either percentile, because it was never measuring a distance in the first place. The same goes for a human judge's impression of proportion, and for whatever an AI model infers from a photograph: both are estimates built without a scale reference, not a second opinion on either percentile.
If you have not yet taken both figures by the standard method, start with length and girth measured to the clinical protocol before treating either percentile as settled.