Shoulder mobility is the one area where the popular explanation and the anatomy actually agree, and then the popular advice ignores the agreement. Getting an arm straight overhead is not a shoulder movement. It is a shoulder movement plus a shoulder blade movement plus an upper back movement, and if the third one is missing you will arch your lower back to make up the difference and call the result overhead.
This cluster treats the region as the chain it is, and is careful about pain, because shoulder pain is one of the areas where the confident advice on the internet has been contradicted by good trials.
Shoulder mobility is three joints
Raising an arm to vertical takes roughly 180 degrees of travel, and it is not one structure that supplies it.
The shoulder itself — the ball in the socket — supplies most but not all. The shoulder blade has to rotate upward and tilt back across the ribs to keep the socket underneath the arm. The upper back has to extend, because the ribs the blade slides on are attached to it: a thoracic spine held in flexion holds the blade forward, and a blade held forward runs the shoulder into its own limit early.
That is why the person who cannot reach overhead often has a perfectly normal shoulder and a spine that has not extended since the second year of a desk job. It is also why chest and lat stretching alone so often does nothing: the constraint was one segment lower.
Thoracic spine mobility covers the segment; scapular control covers the blade; shoulder mobility exercises covers the joint itself.
Measuring the chain, test by test
Three self-tests cover the region, and each is honest about a different part of it.
The Apley scratch, or back-scratch reach. One arm over the shoulder, one behind the back, and you record the gap between the fingers. It has the virtue of being a single number that a person can take alone, and the vice of combining shoulder rotation, extension and flexion into that number. 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, which is more than most home shoulder tests can claim. It is a screen, not a diagnosis: a poor reach tells you the chain is short somewhere, not where. The Apley scratch test sets out the protocol.
The wall angel. Back against a wall, arms sliding up, and you record what leaves the wall first — the hands, the wrists, or the lower back. It exposes the substitution the scratch test hides, because the wall will not let you cheat with lumbar extension. Wall angels is the guide.
Seated thoracic rotation. How far you can turn the upper back with the pelvis fixed. This is the reading most desk workers are worst at and least aware of.
In Limber, the shoulder reach and the wall angel together make the shoulder area score — 60 per cent reach, 40 per cent wall angel — and the seated rotation is the whole thoracic score on its own. The neck has no direct test in the battery and is scored from the shoulder reach as a stated proxy. That is a compromise, and the app labels it as one rather than implying it measured your neck. The overhead mobility test explains what each reading covers.
The shoulder reach is free here: the shoulder mobility test reads both sides on the same five bands the app uses. The wall angel and the seated rotation are not built — on either surface — so the instrument case lists what is finished rather than what is planned, and the Range Score prints the shortfall instead of quietly re-weighting around it.
Shoulder pain, and honest uncertainty
This is the part of the cluster that matters most, because the last decade of shoulder research went badly for the confident version of the story.
For years, pain at the top of the shoulder on reaching was explained as impingement: a bony spur pinching a tendon, correctable by shaving the bone. Then Beard and colleagues ran the CSAW trial, published in The Lancet in 2018, comparing arthroscopic subacromial decompression against placebo arthroscopy — the same operation without the decompression — and against no treatment. Both surgical groups improved. They did not meaningfully differ from each other. Ketola and colleagues' Finnish randomised trial had already found that adding the operation to an exercise programme added nothing to the exercise programme.
Grade: strong, and it changed practice. These are randomised trials with placebo control, which is rare in surgery and hard to argue with.
What follows is a lot of honest uncertainty. Lewis, writing in Manual Therapy in 2016, retitled the whole area "rotator cuff related shoulder pain" precisely because the mechanical explanations had not held up, and set out how much remains unknown about who gets it and why. The structures blamed on a scan are frequently present in shoulders that do not hurt.
There is a related caution about shoulder blade movement. Hickey and colleagues, reviewing the prospective studies in the British Journal of Sports Medicine in 2018, found that asymptomatic athletes with scapular dyskinesis had a modestly increased risk of later shoulder pain — a real finding, with wide uncertainty and a small effect. It is a long way from there to "your shoulder hurts because your blade moves wrong", which is what the finding usually gets turned into.
And a caution about manual work: Mintken and colleagues, in the Journal of Orthopaedic & Sports Physical Therapy in 2016, randomised people with shoulder pain to exercise with or without cervicothoracic manual therapy and found the manual therapy added no benefit.
What this means for this cluster. We write about range, because range is measurable and responds to work. We do not write about how to make a painful shoulder stop hurting, because the evidence does not support anyone claiming that from a web page. Shoulder impingement and mobility is deliberately the most careful article in the cluster, and rotator cuff mobility is written the same way.
When to stop reading and get examined. Pain after a fall or a dislocation; an arm you cannot lift at all; sudden weakness; night pain that wakes you consistently; numbness or pins and needles down the arm; or shoulder pain with fever or feeling unwell. Any of those is a reason to see a clinician who can examine you. Nothing here diagnoses anything and nothing here is treatment.
What actually moves the region
The unglamorous answer, and it is the same answer as everywhere else on this site.
Extension of the upper back, worked often, over a support or against a wall. This is the constraint most people have and the one that changes most visibly.
Range worked under load, not only stretched. As set out in mobility, measured, Afonso and colleagues' 2021 meta-analysis in Healthcare found strength training and stretching produced comparable range gains, on a small and varied set of trials. For the shoulder specifically the strength argument is stronger than average, because overhead is a position you have to hold as well as reach — shoulder flexibility for the overhead press works through that.
Time, honestly counted. Chest openers are pleasant and do not, on their own, move a thoracic spine that has not extended in years. Chest opener stretches and upper back tightness both take that on.
The neck
The neck sits in this cluster because it is rarely a neck problem alone: it is the top of the same chain, and it is where a screen day lands. The Cochrane review of exercise for mechanical neck disorders found some benefit for specific strengthening work, on low-quality evidence — which is roughly where the whole neck literature sits.
The posture explanation for neck pain is weaker than it is sold as, and that argument is made in full in desk and posture, including Richards and colleagues' finding of no association between sitting neck posture clusters and neck pain in over 1,000 adolescents. Neck stretches and text neck stretches are written on that footing: gentle range work is reasonable and low risk, and it is not a correction of anything.
Sleeping position and shoulders is the cluster's thinnest-evidence article and says so at the top.
How you would know it is working
Reach is easier to fool yourself about than almost any other measurement, because a warm shoulder at 6pm and a cold one at 7am produce visibly different results, and because the substitutions are invisible from the inside. So the rules from mobility, measured matter more here, not less: same protocol, same conditions, both sides recorded separately, the date written down, and weeks between readings rather than days.
Both sides is the one people skip and it is the one that pays. Shoulders are commonly asymmetric — a throwing arm, a side you sleep on, an old injury — and an average of the two hides exactly the finding worth having. Limber records left and right separately for every per-side test and flags a gap of 12 points or more, because a difference that size is bigger than the measurement's own noise.
Questions
Why can I not get my arms straight overhead?
Most often because the upper back is not extending, not because the shoulder is short. Overhead reach needs the thoracic spine to extend so the shoulder blade can tilt back and rotate up underneath the arm. If the spine stays flexed, the shoulder runs out of room early and the lower back arches to finish the movement, which looks like overhead and is not. A wall angel exposes this quickly because the wall will not let the lower back help.
What is a good shoulder mobility score?
On Limber's scale, fingertips touching behind the back reads 75 and fingers overlapping reads 95, with hands more than two hand-widths apart at 10. Those anchors are the app's own, designed so the distance between two of your own readings means something. The Senior Fitness Test's back-scratch has published norms by age and sex if you want a population comparison, but a comparison to strangers is a much weaker guide than a comparison to yourself six weeks ago.
Does stretching help shoulder impingement?
That question has become harder to answer honestly than it used to be. Randomised trials — including CSAW in The Lancet in 2018, which compared the decompression operation against placebo surgery — have undermined the mechanical pinching story that the term describes. Exercise-based care is what most guidelines now recommend for rotator cuff related shoulder pain, but range work alone is not established as the reason people improve. If your shoulder hurts, particularly with weakness, night pain, or after an injury, see a clinician who can examine it.
How often should I work on thoracic extension?
Often and briefly beats rarely and heavily, because the upper back's problem is usually time spent in one shape rather than a short structure. Daily short exposures are a sensible default. The evidence base here is thinner than for hamstrings, so treat that as reasoning from the mechanism rather than as a tested prescription.
Is a rounded upper back something I need to correct?
It is a shape, and shapes vary a lot among people with no symptoms. What is worth knowing is whether the segment can move — whether you can extend and rotate it when you ask it to — rather than what it looks like at rest. Measure the rotation and the wall angel, work them, and re-measure. That is a claim about range, which is testable, rather than a claim about posture, which mostly is not.
Can I test my shoulders at home without equipment?
Yes. The back-scratch reach needs nothing, and the wall angel needs a wall. Record each side separately, use the same setup every time, and note the date and roughly the hour, because a warm shoulder reads further than a cold one. The shoulder mobility test scores the back-scratch reach here for nothing, on both sides. The wall angel has no instrument, on either surface.