How the tip meets the ruler

Most of the attention in a length reading goes to the base end, because that is where bone-pressed pressure is applied and where most of the technique lives. The tip end gets less attention and still has a way to go wrong.

Two opposite mistakes at the same end

The ruler should rest alongside the shaft, its scale reading wherever the tip of the glans reaches - not pressed against the tip, and not held a visible gap away from it.

Push the ruler into the tip and you compress the glans slightly, which shortens the reading by a small but real amount, the same soft-tissue compression that makes bone-pressed pressure at the base such a large source of error in the first place, here happening in miniature at the other end. Hold the ruler off the tip - a habit some people develop to avoid the first mistake - and you read a gap as if it were part of the tip's position, which adds length the shaft does not have.

What to do instead

Rest the ruler against the side of the shaft without pushing toward the tip, then read wherever the glans's furthest point actually falls against the scale, sighted square on rather than from an angle. The ruler is a straightedge to read against, not a tool to press the tip flat with.

If you are unsure whether you are compressing, ease off pressure at the tip end specifically while keeping the base-end pressure unchanged, and see whether the reading moves. A reading that shortens when you press the tip and lengthens when you ease off is telling you the compression was real.

Scope

This is only about contact at the tip once you already know where length ends. Where length ends in the first place - the glans's furthest point along a straight line from the base, not the meatus and not the far side of a curve - is a separate and prior question, worth reading first if that definition itself is unclear.

A millimetre or two here is small next to the base-end pressure error that dominates most self-measurements, but it is a millimetre or two you do not have to give away, and it costs nothing to get right once you know to look for it.

None of this has an analogue where a photograph rather than a ruler is doing the work. Rate Cock reads a static image rather than resting anything against tissue, the comparison an image model makes happens entirely in pixels with nothing to compress. A numeric score or a human judge's opinion of that same photograph has its own sources of disagreement, just not this one.