Push Hands (Tui Shou)

It is also the part of the curriculum most often done badly, in two opposite ways. This page explains what push hands is for, how it progresses, and how to recognise both of the standard failures — because a school that has fallen into either one will not tell you.

What it is for

Solo form practice can only take you so far. You can build structure, learn the movements and develop the qualities the classics describe, but you cannot find out whether any of it survives contact with another person. Push hands is where that is tested.

Specifically it trains three things. It develops the ability to feel what another person's body is doing through the point of contact. It builds the habit of responding to force without either resisting it or losing connection to it. And it exposes structural faults immediately, because a partner will find every place where your organisation breaks down.

The traditional formulation describes a progression: first you learn to listen, then you come to understand force, and eventually understanding becomes immediate. The first two stages are ordinary training. The third is what people spend decades on.

Listening

The skill called listening is the foundation, and the name is misleading in English because nothing auditory is involved. It means detecting, through touch, the direction, magnitude and timing of a partner's force, and detecting changes in it before they become visible.

There is nothing mysterious about this. Human beings have dense mechanoreceptors in the skin and highly developed proprioception, and the information available at a point of contact is genuinely rich. What training does is teach you to attend to it, and to interpret it fast enough to be useful.

The main obstacle is muscular tension. A tense arm is a poor sensor: it registers only large forces, and its own activity masks what it is trying to detect. This is the direct practical reason relaxation matters so much in this art, and it is a better reason than most of the ones usually given.

Listening improves faster than almost anything else in taijiquan. Most people notice a real difference within months, which makes push hands unusually rewarding early on.

Neither disconnecting nor resisting

The governing rule of push hands is stated as a pair of prohibitions: do not lose contact, and do not push back. Everything else follows from it.

Losing contact means your arm comes away from your partner's, at which point you can no longer feel what they are doing and you are working blind. Pushing back means meeting their force directly, at which point the exchange becomes a contest of strength and the whole method is abandoned.

Between those two failures is a narrow path: maintaining light constant contact while continuously yielding along the line of the incoming force and redirecting it. The associated instructions — adhere, connect, stick, follow — all describe aspects of staying on that path.

This is much harder than it sounds, because both failures are reflexive. When pressure arrives, an untrained person either braces or pulls away, and both happen faster than thought. Retraining that reflex is most of what push hands practice actually consists of.

The progression

Standard training moves through recognisable stages, and a school that skips them is usually skipping the difficult parts.

It begins with fixed-step single-hand practice: feet stationary, one arm in contact, a simple repeated circle. This is where listening is first developed, and it is far less trivial than it appears.

Then fixed-step double-hand practice, and the four-cardinal pattern, which cycles through the four principal energies — warding off, rolling back, pressing and pushing — in a set sequence. Learning the pattern takes weeks; doing it well takes years.

Then the four corner methods, which introduce plucking, splitting, elbow and shoulder, and are rougher. Then moving-step practice, which adds footwork and makes everything harder because your base is no longer fixed. Then large-scale rolling-back patterns covering more ground. And finally free push hands, in which the pattern is abandoned and both partners do as they like within agreed limits.

Most practitioners never leave the fixed-step patterns, and skill developed there transfers only partly to the free version.

The first failure: collusion

This is the commonest problem and the harder of the two to see, because everyone involved is having a pleasant time.

Over months of practice with the same partners, people learn each other. They learn when a push is coming, what response is expected, and — crucially — that the session goes more smoothly if they move when moved. Nobody decides to do this. It develops because cooperative practice is more comfortable than the alternative and because students want to please their teacher.

The result is a group in which everyone appears skilled, the teacher can move anyone effortlessly, and none of it works on an outsider. It is not fraud. It is a training artefact, and it is extremely common.

The test is simple: bring in someone who has never trained with the group. If the effortless throws stop working, you have found collusion. Good teachers arrange this test themselves rather than avoiding it.

The second failure: shoving

The opposite error. Push hands becomes a contest, both partners commit strength, and the exercise turns into a slow wrestling match won by whoever is heavier and more determined.

This is what competitive push hands rules tend to produce, because a rule set that awards points for displacement rewards the thing that most reliably causes displacement, which is strength. Watch footage of most push hands competitions and you will see very little that resembles the classics.

The error here is not that resistance is bad. It is that undifferentiated maximum resistance removes the conditions under which the skill can develop. If both partners are braced, nobody can listen, nobody can yield, and the exercise trains something other than what it is for.

Graded resistance

The solution both failures point to is the same: resistance should be real, deliberate and adjustable, and both partners should know what level they are working at.

At a low level, the partner offers steady light pressure and moves honestly when their balance is genuinely taken, but does not fight. This is where technique is learned. At a middle level, the partner actively seeks openings and does not concede anything unearned. At a high level, both partners work at full commitment within agreed safety limits.

The important thing is that the level is chosen and stated rather than drifting. Most schools drift towards the low level and stay there. A useful question to ask when visiting is whether anyone can describe how they vary resistance, because a school that has thought about it will have an answer ready.

What push hands is not

It is not fighting, and confusing the two causes real problems.

Push hands begins from contact, which a fight does not. It excludes strikes, kicks, grappling on the ground, weapons, multiple opponents and the entire problem of managing distance before contact is made. Its goal — displacing the partner — is not the goal of a fight. It is a drill isolating one component of a much larger skill.

People who are extremely good at push hands and have never trained anything else are sometimes surprised by this. The exercise is valuable precisely because it isolates something, and the cost of isolation is that everything else has been removed.

How it connects to the form

Push hands and form practice are supposed to feed each other, and in schools where they do not, both suffer.

The form supplies the movement vocabulary and builds the structure. Push hands tells you which parts of your form are real. A posture that feels fine in solo practice and collapses the moment someone leans on it has a fault you would never have found alone, and once you have found it you take it back into the form and work on it there.

Running the other way, push hands without form practice tends to produce practitioners who are good at the game and have no particular quality of movement. The form is where the fine detail gets built, because it is the only place slow enough to build it.

A useful sign of a well-run school is that corrections cross over: the teacher stops a push hands session to fix something in your stance, or points out during form practice that a particular transition is where you keep getting caught.

Common errors

Using arm strength, which is the default and takes years to unlearn. Leaning forward to reach the partner, which surrenders your own balance to gain a small advantage. Chasing the hands rather than attending to the partner's centre, which is how you are led into empty space. Tensing at the moment pressure arrives, which is the reflex the whole exercise exists to retrain. And treating push hands as a competition to be won every time, which stops you from experimenting and therefore stops you from improving.

The last is worth emphasising. Someone who never loses in push hands is playing safe, and playing safe means practising what they can already do.

Key points

• Push hands tests whether solo practice survives contact with another person, and it exposes structural faults immediately.
• Listening is tactile detection of force and change, not anything mysterious; tension is the main obstacle, which is the clearest practical reason relaxation matters.
• The governing rule is neither to lose contact nor to push back, and both failures are reflexes that take years to retrain.
• The first standard failure is collusion, in which long-term partners unconsciously cooperate; the test is to bring in a stranger.
• The second is shoving, which competition rules encourage because points for displacement reward strength.
• Push hands is a drill that excludes strikes, distance and everything before contact, so skill in it is not the same as fighting ability.