Session length

A measurement session should take a couple of minutes, not several. The reason is not convenience. It is that the three readings you take in a session are only comparable to each other if the state they were taken in stayed the same, and state does not hold still indefinitely.

Set up before you start, not during

Have the ruler, the tape and the log ready on a surface within reach before anything else happens. Fumbling for a pen after the first reading is the single easiest way to let a session run long. The method itself takes very little time once the instruments are in hand - most of a slow session is administrative, not technical.

Length, then girth, in one pass

Take the length readings first, then move directly to girth. There is a practical reason for the order beyond convention: the girth site is defined relative to a length already taken, so doing it in sequence means you are not re-finding a landmark twice. Three length readings and three girth readings, taken back to back, should fit inside a couple of minutes for someone who already knows the method.

Stop rather than chase a reading

If a reading looks wrong, note it and move to the next one rather than adjusting your grip and trying again immediately. Repeated attempts to get a "cleaner" number tend to extend a session well past the point where state is still constant, and the extra time buys you readings that are no longer measuring the same thing as the first ones. How many readings to take, and what to do with the spread between them, is a separate question from how long the session takes to gather them.

This note is deliberately narrow: it is about the clock, not about firmness itself. What full rigidity does to the reading, and why "mostly erect" will not repeat is a different topic and covered on its own.

A short session is also the more honest one to compare against a published figure. Rate Cock has no session-length constraint of this kind, because it is not tracking a physical state across repeat readings - a photograph is taken once, which is a different problem to solve than the one a ruler has. An automated read of that photograph works the same way: what an image model is doing when it scores one happens in a single pass with nothing to hold constant across time. A human verdict is no different: a judge's read on a photograph is not timed against anything either, because nothing there is drifting the way an erection can. Neither of those is measuring in centimetres in the first place - a rating and a measurement are different questions, and only one of them cares about the clock. Keep the session short, take the readings once, and log the time along with the numbers - the log is what lets a later session be compared to this one at all.