Gear
String and ruler
Wrap a string, mark the overlap, lay it on a ruler: it works, and its two errors are string stretch and the width of the mark you make.
A length of string, wrapped around the shaft and pinched at the overlap, then laid flat against a ruler, gives a girth reading close to what a tailor's tape gives. It carries two specific errors a tape mostly avoids, and both are correctable once you know they are there.
Why it works at all
String traces a circle the same way a tape does: wrap, mark the overlap, straighten, read the distance against a ruler. The paper strip covered in Tape, Ruler, or a Strip of Paper is the same idea with a stiffer material, and stiffness is exactly what removes string's first error.
Error one: stretch
Ordinary string and thread stretch under tension by an amount that depends on the fibre, its age, and how hard you pull. A stretched string reads long, because you have straightened out some of its own slack along with wrapping the shaft. The direction is consistent - stretch only ever adds length - which at least tells you which way to distrust a given reading. Cotton kitchen string stretches noticeably under a firm wrap; unwaxed dental floss and fine synthetic cord stretch far less, and either beats string sold for tying parcels.
Error two: the mark
Pinching the overlap point with a fingernail or a pen leaves a mark with width, not a point, and that width is itself a source of error on a small circumference. Read from the centre of the mark rather than either edge, and use a fine pen rather than a fingernail, which tends to compress the string before you can pin the spot. A mark half a millimetre wide sounds trivial until you remember a typical girth reading is only a handful of centimetres around: half a millimetre is a real fraction of it, not noise.
Check the string before you trust it
Do not assume a given piece of string is low-stretch just because it feels stiff in the hand. Lay it along a rigid ruler, hold one end at the zero mark, and pull with roughly the tension you would use for a wrap; if the far end creeps past where it sat with no tension by more than a fraction of a millimetre over ten or twenty centimetres, that string is stretching enough to matter. Testing a Tape for Stretch describes the same check applied to a tailor's tape, and the logic transfers directly: any flexible material used as a length reference should be checked against something rigid before you rely on it, not assumed to behave. A piece of string that fails this check is not disqualified - just downgrade how much you trust the reading, or switch to dental floss or a fine synthetic cord and test again.
A worked comparison
Suppose a wrap with a stretch-prone cotton string reads a touch higher than the same wrap taken moments later with a low-stretch cord at the same site. That gap is the stretch error made visible, and it is the reason to prefer the cord as a default rather than treating the two as equivalent: the direction of the discrepancy will always point the same way, with the stretchier material reading higher, which is a useful check even without a precise figure attached to it. If a tailor's tape is available at all, use it and skip this comparison entirely - string is worth reaching for only when nothing better is at hand.
Minimising both
Use a stiff, low-stretch cord rather than kitchen string, and mark it with a pen rather than a nail. Wrap once, mark immediately, and lay the string against the ruler without re-pulling it - re-tensioning after the wrap reintroduces the stretch you were trying to avoid. Take three wraps and use the median, the same as any other girth reading, described in full for the standard method.
Where this sits against the alternatives
String and ruler is a legitimate fallback, not a first choice. A tailor's tape or a paper strip removes the transcription step entirely, and its behaviour is more predictable across repeated uses, as the direct comparison of instruments covers. Where measurement error comes from puts instrument choice near the bottom of what actually moves a reading, and string's extra errors are a small addition to a total already dominated by technique rather than equipment.
What none of this gives you
A girth figure, however carefully taken, is a number in centimetres with an uncertainty attached to it, not a verdict on anything. That distinction matters because judgement-based numbers exist too, taken by an entirely different method: a photo scored by something like Rate Cock rates an image, which is a different kind of number from a distance along a string. A rating out of ten sits on a distribution of photographs rather than a tape measure, and a human judge's opinion on Rate Penis is an impression rather than a figure, however confident it sounds. An AI model asked to size something from a photo is closer to recognition than measurement, as AI Penis explains about its own instrument - which is a separate subject from a string and a ruler, and worth knowing you are not accidentally comparing the two.