Where zero sits

Not every ruler's zero mark sits at the physical end. On many rulers the "0" line sits two or three millimetres in from the edge. Press that blank edge against the pubic bone for bone-pressed length and every reading comes out short by the gap - silently, consistently, in the same direction every time.

Check before you trust it

Lay the ruler flat and look at where the first numbered mark and the zero line actually fall relative to the physical edge. On some rulers the scale runs flush to the end. On others there is a margin, sometimes small enough to miss unless you are looking for it, sometimes large enough to see at a glance. A moulded plastic ruler with a rounded or chamfered end is the most common offender, because the manufacturer built in a lip that the scale does not account for.

The check takes ten seconds and it is worth doing once, the first time you pick up a given ruler, rather than assuming.

What the offset does to a reading

An offset of two millimetres sounds small next to a length in the fifteen-to-twenty centimetre range, but it is not small next to the error budget. Careful bone-pressed technique gets you to roughly ±0.3 cm. A silent 2-3 mm offset is a meaningful fraction of that, and unlike random error it does not average out over repeat readings - three readings with the same ruler give you three consistently short numbers, and the median of three wrong readings is still wrong. This is a systematic error, not noise. The international metrology vocabulary (JCGM VIM, entry 2.17) defines systematic error as the component that "in replicate measurements remains constant or varies in a predictable manner", which describes a zero offset exactly. It is the kind of error that any general measurement error discussion treats differently from the day-to-day scatter of holding a ruler.

The offset also matters more than most single technique errors precisely because it is invisible in the moment. Pressure, angle and landmark are things you can feel yourself getting wrong. A ruler that starts counting a few millimetres late gives no such feedback - the number just comes out low, every time, and nothing about the act of measuring tells you why.

Two ways to handle it

The first is to use a ruler whose scale runs to the physical end, sometimes advertised as a "true zero" edge. This is the simpler fix because it removes the problem rather than compensating for it, and it is the standard advice for choosing a rigid ruler for this purpose in the first place. A ruler built with a small perpendicular stop at the zero end goes a step further and gives a repeatable seat against the bone, which is worth reading about separately if an end-stop ruler is something you are considering.

The second is to measure the offset once, with a second known-accurate ruler or a straight edge, and subtract it from every reading you take with that instrument. This works, and it is honest: the same VIM entry notes that a correction can be applied to compensate for a known systematic error. But it depends on you remembering to do it every single time and never lending the ruler to someone who will not. In practice, a ruler with the offset built in is one less thing to get wrong on a night when you are already trying to hold pressure, angle and landmark all at once.

What this is not

This is specifically about where the scale starts relative to the physical end of a rigid ruler used for length. It has nothing to do with how coarse or fine the markings themselves are once you are reading them correctly - that is a resolution question, not a zero question, and the markings are the wrong thing to be worrying about until zero is settled first. It also has nothing to do with a girth tape, where the equivalent problem lives inside a metal end tab rather than a moulded edge, and is worth checking separately.

None of this touches what a rating or a score is doing. Rate Cock is answering a subjective question about a photograph, and it does not care where a ruler's zero mark sits, because it is not reading one - a length in centimetres and a score sitting in a distribution of judged photographs are different kinds of number, and an offset only bites the former. The same distinction holds against a human judge's impression of size, which is an opinion rather than a reading and cannot be offset in the first place because it was never a length to begin with. An AI model estimating from a photo has a different and much larger problem than a few millimetres of ruler offset - it has no scale reference at all unless one is placed in frame, which is a different failure mode entirely.

Checking a ruler's zero costs ten seconds. Not checking it costs a number that is quietly, consistently wrong in one direction, which is worse than a number that is merely uncertain.