The Apley scratch test — one arm reaching over the shoulder and down the back, the other coming up from below, and the gap between your hands as the reading — is the shoulder measurement in Limber's battery. Two minutes, no equipment, a separate number for each side.
It also has an arithmetic problem, and this page states it before anything else, because it decides what your own result is allowed to mean.
The short answer
- The side is the arm that goes over the top. Name it that way every time or your two readings swap places between sessions.
- Five bands and nothing between them: 10 · 30 · 55 · 75 · 95.
- One band is worth 20 or 25 points. Limber calls a gap of 12 points or more an imbalance. So the smallest gap this instrument can express already exceeds our own threshold, and any difference at all between your sides is flagged.
- A one-band difference means re-measure, not finding. Take it again on another day before you believe it.
- It feeds 60 per cent of the shoulders area and all of the neck area — the neck as a stated proxy, because there is no neck test in the battery.
How to take the Apley scratch test
Four steps, and the app prints them the same way.
- Reach the arm on this side over your shoulder and down your back.
- Bring the other arm up behind your back and reach the hands toward each other. No forcing. Stop at a stretch, never at sharp pain.
- Hold there. Note the gap between your hands.
- Come out slowly, swap arms, and read the other side.
Measure the gap in your own hand-widths — the distance across your knuckles. It is a rough unit and that is deliberate: you have it with you, it scales with your own body, and it repeats without a helper. A ruler held behind your back by somebody else is a more precise reading of a differently performed test.
The naming matters more than it sounds. The side is the arm that goes over the top. Left arm over the top is the left reading, whatever the hand underneath is doing. Get that backwards once and your two columns trade places, and a swap looks exactly like a change.
Same conditions, too. A warm shoulder at six in the evening reaches further than a cold one at seven in the morning, and neither of those is progress.
What each band scores
| Band | Where the hands stop | Score |
|---|---|---|
| 1 | More than two hand-widths apart | 10 |
| 2 | One to two hand-widths apart | 30 |
| 3 | Less than one hand-width apart | 55 |
| 4 | Fingertips touch | 75 |
| 5 | Fingers overlap | 95 |
Those are the app's own words and the app's own anchors, and the shoulder mobility test scores a reading taken here with the same five numbers. A reading taken on the web and a reading taken in the app are the same reading, not two scales that happen to agree.
Nothing interpolates. There is no honest value between "fingertips touch" and "fingers overlap", and a hand-width is not a unit you can subdivide behind your own back. The anchors are spaced so the distance between two of your readings means something. They are not percentiles. No population has been measured against this scale, so the number says nothing about how you compare with strangers.
The top band is a ceiling as well as a score. Everything past an overlap reads 95. If both sides read 95, this instrument has stopped measuring you and the informative readings are elsewhere in the battery.
Why any side gap gets flagged
Here is the arithmetic, printed on our own page rather than left for somebody else to find.
The anchors are 10, 30, 55, 75 and 95. The steps between them are 20, 25, 20 and 20. One band on this test is worth 20 or 25 points, and Limber calls a gap of 12 points or more between your sides an imbalance. So the smallest gap this instrument can express already exceeds our own threshold, and any difference at all between your sides is flagged.
There is no such thing as a small gap on a five-band test. The instrument cannot express one.
The rule that follows is that a one-band difference means re-measure, not finding. Take it again on another day, same hour, same warm-up, before you believe it. If it repeats, it is a reading. If it does not, it was the day, or which hand you happened to concentrate on.
Set that against the knee-to-wall, which is the one per-side test in the battery fine enough to show a gap under the threshold. It is measured in centimetres and interpolated between anchors, so one centimetre is worth 5 to 10 points — ten through the middle of the scale, five at the top. The fineness cuts both ways. Its smallest expressible gap is 5 points, well under the threshold, where no band test can express anything under 20 — and a centimetre of careless foot placement manufactures most of an imbalance out of nothing. Coarse instruments flag everything; fine ones punish sloppiness. Reading a left–right gap honestly works through both failures.
What Limber does with a gap it believes: one extra move for the side that reads lower, each session, until the gap falls under six. One position added to an otherwise unchanged plan, not a separate programme.
What the reading feeds
This one reading decides a whole fifth of the composite by itself, and it is worth knowing exactly how.
Shoulders area — 60 per cent. The wall angel is designed to supply the other 40 per cent, and it is not built, in the app either. The weights of tests nobody took drop out rather than counting as zero, so today this reading carries the shoulders area on its own, and the page names the test that is missing instead of presenting the area as fully read.
Neck area — all of it, as a stated proxy. There is no neck test in the battery. Nothing measures your neck, and the app labels the inference rather than dressing it up as a reading.
Those two areas make the Shoulders group, which is one of the five groups the Range Score averages. The other four are the posterior chain, the ankles, the hips and the spine. With the wall angel missing, this test is the only input to both areas, so it settles that group alone — one of just two tests in the battery that decides a whole group by itself. The knee-to-wall is the other.
The spine group is fed only by a seated thoracic rotation test that is not built either, so four of five stations is the ceiling of this battery today — and the upper back is the segment that most often limits an arm going overhead. A station nobody measured is drawn as untested and left out of the mean. It is never guessed from its neighbours.
What the evidence says
The back-scratch reach is better documented than most home shoulder tests, which is a low bar that it clears.
Rikli and Jones built a version of it into the Senior Fitness Test in the Journal of Aging and Physical Activity in 1999, with normative data by age and sex. Grade: reasonably solid as a repeatable field measure. Their norms belong to their protocol and their scale rather than ours, so treat them as evidence that this shape of test survives repetition, not as a table to score yourself against.
What the test does not do is predict. Moran and colleagues' 2017 systematic review in the British Journal of Sports Medicine found that movement-screen composites predict injury poorly, and that applies to this reading as much as to any other. It tells you where a shoulder is today. It does not tell you what will happen to it. Grade: solid, and it runs against the way tests like this are usually sold.
A gap with an obvious cause deserves curiosity rather than alarm. Wilk and colleagues, in the American Journal of Sports Medicine in 2011, found that in pitchers total rotational motion — the sum of internal and external rotation — was more informative than the internal-rotation deficit taken alone. A throwing arm, a racket arm, a carrying arm and the side you sleep on all read differently from their partner in people with no complaints at all. Grade: moderate, and drawn from throwing athletes rather than from desks.
A pinch at the front or the top of the shoulder as the arm travels is not a stretch. Come out of the position rather than pushing for the next band. Pain that persists, wakes you at night, radiates into a limb, comes with numbness or weakness, or followed a fall or an injury is a matter for a clinician who can examine you. Nothing here diagnoses anything and nothing here is treatment.
Take it, then take it again
The shoulder mobility test runs the whole thing in a browser: the four steps, five bands, both sides, and a 0–100 score off the same anchors the app uses. Nothing is stored and nothing is sent — the survey sits in the address, so the link is the record.
Then leave it alone. Twenty-one days is the app's default gap between readings, and a reading older than that is marked stale rather than quietly treated as current. Same protocol, same hour, both sides written down, the date beside them.
Between readings, the work. Shoulder mobility exercises sorts the catalogue by which link of the chain each move loads, because the commonest reason chest stretching does nothing is that the constraint was one segment lower. The overhead mobility test explains why passing this test and failing an overhead reach is an ordinary result rather than a contradiction. The rest of the cluster is shoulders and upper back.
Questions
What does the back-scratch reach actually measure?
Three joints at once, and it cannot tell you which one is short. The top arm is doing shoulder flexion and external rotation, the bottom arm is doing extension and internal rotation, and the shoulder blade has to slide on the rib cage for either of them to travel. A restriction anywhere in that chain closes the gap between your hands, and so does a stiff upper back.
Which arm names the side?
The arm that goes over the top. If your left arm reaches over your left shoulder, that is the left reading, whatever your right hand is doing underneath. It is the single most common way to corrupt a set of readings, because a swap between sessions looks exactly like a change in both sides at once.
Is a one-band difference between my sides a problem?
Treat it as a prompt to measure again rather than as a result. One band on this test is 20 to 25 points, the app flags 12, so the smallest difference the instrument can record is already over the line. If the same gap shows up on a second day in the same conditions, it is a reading worth acting on; if it does not, it was noise.
How often should I take it?
Every three weeks is the app's default, adjustable to two or four. Shoulder range moves slowly, day-to-day variation is large, and taking a band test weekly mostly measures how warm you were. A reading over 21 days old is treated as stale rather than current.
Does a good score here mean I can reach overhead?
Not reliably. This test reads a mixture of rotation, extension and flexion behind the back, and an overhead reach is a different position with a different limit, usually the upper back. People pass one and fail the other in both directions, which is why the two are written up separately rather than treated as one number.
Can I train the gap between my sides closed?
Sometimes, and nobody has shown it matters. There is no trial demonstrating that reducing a left–right range difference reduces injuries or improves anything else, and this site will not imply there is. Limber adds one extra move on the lower side until the gap falls under six points, which is a modest, checkable rule rather than a claim about outcomes.