The tank man photo is the most famous image of the 1989 Tiananmen Square democracy movement in Beijing. Indeed it is considered one of the most famous photos of the 20th century.   It has become a symbol of human resistance, a lone individual stopping a whole column of tanks. 

What the photo claims to symbolize, however, is only very partially true.  It is not a photo of Tiananmen Square, but of a boulevard nearby.  It was not taken during the crackdown on demonstrators, which took place on June 3 and the following night, but on the quiet morning of June 5, after Tiananmen Square had been cleared and government control had been reestablished in Beijing. And it was not a successful protest. The tanks stopped briefly; two men came into the street and took the protestor away.

The photo was taken by an American newsman,   from a hotel balcony 800 meters distant--about half a mile. It was shot through a telephoto lens, like so many news photos of recent decades. This is one of the marvels of modern technology, and a hidden one: how seldom do we stop to think of how the photographer got so close, so near the action where history is made? Compare the equally famous, infinitely shocking photo from South Vietnam 1972 of children running from napalm—where did we think the photographer was standing? 

Telephoto lenses allow us to intrude closely into events that the participants would probably like to keep hidden. It is one of the sharpest differences between our images of the world before about 1960 and the present. The Vietnam War was the first war in history where we could see what it actually looked like. Before then, we had to be content with what officials allowed for patriotic publication, plus (as of World War II) candid shots of soldiers, generally far behind the front lines. And not just for violence in war, but violence in all its peacetime forms, telephoto lenses have brought us first-hand records of how violence really looks. And other forms of conflict, too—the expressions on faces and bodies that give us clues to how conflict plays out, and enable us to cut through the rhetoric and the mythology that have obscured it since humans first began to tell lies about violence.

I would go so far as to say that the telephoto lens, even more than the advent of television, has changed our access to reality. Even more than the camcorder which in 1991 first showed the police beating Rodney King; even more than the ubiquitous mobile phone cameras that now flood the Internet-connected world with images. The reason I make this exorbitant claim is that all the other devices depend on being up close; the telephoto lens zooms in from a great distance. It can go where it is too dangerous or too private for other devices to go. Unlike TV, it gives us photos that are not posed, since no one knows there is a camera to pose for. And it can give photos of great detail—the emotional expressions on faces, the exact postures of bodies, that are so important for a micro-sociologist’s explanation.  The purpose of my writing, however, is not to pick a fight as to which visual technology is best; they all work together to make our times the golden age of visual sociology. 

Having extolled telephoto images, I want to raise a caveat about their limits. Taken out of context, they carry the danger of modern myth-making. To see what is distorted and what can be salvaged, let us examine the tank man photo in greater depth.

The Surrounding Context of the Tank Man Photo

The Beijing democracy demonstrations began on April 17, 1989, and went on for 50 days until they were crushed. The tank man photo was taken on day 51. Here I will summarize only the very last days. (More detail on the entire sequence is given in my post, Tipping Point Revolutions and State Breakdown Revolutions: Why Revolutions Succeed or Fail, The Sociological Eye, June 2013.) /sociological-eye/2013/06/tipping-point-revolutions-and-state.html

Over the 50 days, the size of the crowds at Tiananmen Square rose and fell. After most of the initial enthusiasm had fallen off, on day 28 (May 13), the remaining few hundred militants launched a hunger strike, which recaptured public attention, and brought hundreds of thousands of supporters to Tiananmen. On day 34 (May 19), the Communist elite purged its dissidents and declared martial law, and began to bring troops into Beijing.

The next four days were a showdown in the streets; crowds of residents blocked the army convoys; soldiers rode in open trucks, unarmed-- the regime still trying to use as little force as possible, and also distrustful of giving out ammunition-- and often were overwhelmed by residents. Crowds used a mixture of persuasion and food offerings, and sometimes force, stoning and beating isolated soldiers. On May 24 (day 39), the regime pulled back the troops to bases outside the city. The most reliable army units were moved to the front, some tasked with watching for defections among less reliable units. In another week strong forces had been assembled in the center of Beijing. 

Momentum was swinging back the other way. Student protestors in the Square increasingly divided between moderates and militants; by the time the order to clear the Square was given for June 3 (day 49), the number occupying was down to 4000. There was one last surge of violence-- not in Tiananmen Square itself, although the name became so famous that most outsiders think there was a massacre there-- but in the neighborhoods as residents attempted to block the army's movement once again. Crowds fought using stones and gasoline bombs, burning army vehicles and, by some reports, the soldiers inside. In this emotional atmosphere, as both sides spread stories of the other’s atrocities,  something on the order of 50 soldiers and police were killed, and 400-800 civilians (estimates varying widely). Some soldiers took revenge for prior attacks by firing at fleeing opponents and beating those they caught. In Tiananmen Square, the early morning of June 4, the dwindling militants were allowed to march out through the encircling troops.

The Tank Man Photo and What It Shows

The Tank Man photo was taken the following morning. The revolutionary crowds had been beaten. Massive arrests were being made, especially of workers, whom the government regarded as far more dangerous than students. Hundreds of thousands of security agents were beginning to spread across the country, picking off suspects one by one, ultimately arresting tens of thousands in the following months. The tipping point had passed, and the regime had clearly won.

What then was the point of the tank man protest?  By his white shirt and dark trousers, we can surmise that he was a government bureaucrat, a class of people whose sympathies were strongly on the side of the protestors. But it is also a category of persons, numerous in all demonstrations, who offer support but do not take part in the actual confrontations with authority.  In virtually all photos of demonstrations and riots everywhere in the world, a small portion of crowd is at the front doing the violence, while most stand at a distance and watch. Very likely tank man had seen or heard about the previous days’ violence, and came forward in the quiet atmosphere to do something to demonstrate his own commitment.

As we can see in the photo, the streets are virtually empty. He has no visible supporters, although a small audience gathered on the sidewalk to watch from a distance. On the other hand, the tank troops too are anonymous, hidden inside their armored stations. The tanks are moving slowly, making a show of force, not an actual military operation.  – One can know this, because the tanks are in column, a parade-like movement; deployed into combat they would go into line. I would surmise that the soldiers are calm; their action has been over for 24 hours or more.

Thus it is a symbolic confrontation: the lone man, respectably dressed in the garb of the urban apparatchik, stepping in front of the column of slow-moving tanks. In that atmosphere, there is little danger of being run over. The lead tank swerved to avoid him, but he kept in its path until it stopped. Very likely the troops had returned to the orders that prevailed during days 34-38, when unarmed troops were sent to assemble in the city as quietly as possible, and had given no resistance when crowds forced them back. On the whole, the regime had used a mixture of appeasing the crowds, waiting for them to dwindle away, and sporadic application of military force. On day 51, they were back into the mode of calm normality. The government machinery was operating again; bureaucratically organized investigations and individual arrests were the regime’s weapon now. The rebellious crowd has its best chance when it is assembled in huge numbers, in an atmosphere of emotional support that flows outward, dangerously lapping at the solidarity of the government apparatus. Now the crowd has dispersed; and it is in this configuration that bureaucratic authority can exercise its unrelenting and comparatively unemotional control.

And that is what happens. Tank man steps in front of the tank column; the lead driver stops; the tank drivers behind him stop because the tank in front stops. Two men in dark suits come and take tank man away.  The column grinds slowly on.

When Does Local Resistance Succeed?

It would have been an unknown incident except for the newsmen in the hotel with the telephoto lenses. Pictures of violence on the previous days were just making their way into Western newspaper and television, so little attention was paid to the exact sequence when things happened. A famous photo showed bicycles crushed in a street where fighting had taken place—and in the absence of photos of actual bodies, these were taken as emblems of how the revolution had been crushed. It was easy to conjure up a scenario of tanks rolling over a crowd of demonstrators. And then, in the midst of this—the heroic image of the man who stopped the tank column. All was not lost: the human individual still prevails.

We are living in the realm of symbolism here, not in the realm of history. Never mind that no one stopped the tanks, or more likely the trucks that rolled over the bicycles and carried troops into the streets where fighting had taken place 36 hours earlier.  It carries a nice message, although only through the more careful retrospect of micro-sociology do we actually see what it is:  the violent confrontations between crowds and army on June 3 and the early hours of June 4—confrontations in which violence was used on both sides—did not stop the army. But at the right moment, approached with the tools of non-violence, the army was stopped. 

Micro-sociology, above all, attempts to be realistic.  Violence cannot be stopped everywhere. Sometimes force rolls on and crushes everything in its path. But—sometimes violence is stopped. It happens locally, and by persons acting in local conditions.  This is something to build on. What are those conditions?

Comparisons:  Pockets of Successful Non-violent Peace-making in Riots

Turn now to the work of Dr. Anne Nassauer, of the Free University, Berlin. Using videos posted on-line from mobile phone cameras, plus GPS maps of streets, charting time-lines from police radio traffic, in short with the whole array of tools now available, she reconstructs protest demonstrations in Germany and the US.  With this cutting-edge media high-tech, she is able to reconstruct the micro-history of protests and to pin-point just when and where a demonstration will turn violent.  On the whole, I should mention, Nassauer finds that most demos stay peaceful, and their peacefulness can prevail even when militant protestors announce in advance that they will use force; or indeed, when police announce a tough crack-down-on-everything policy.  That is to say, whether a protest turns violent or not depends on local and emergent conditions; violence-threatening events can end up peaceful, and peaceful demos can turn violent.

I will not try to summarize here Nassauer’s findings of the several pathways that lead to violence. Let us concentrate on a single point:  if violence has already broken out, nevertheless all is not lost. It is not too late to stop the violence—not everywhere, but locally, at the place where human individuals use the right techniques.  What are those techniques?

When the police surge forward and the crowd starts running and ducking, people are likely to be beaten.  Photos often show clusters of police or soldiers, attacking anyone in their path—swinging clubs at women, old people, news reporters, anyone. 

Minsk demonstration 2006

Katmandu 2006

This is an emotional rush by the police, that I have called Forward Panic.  It is like the crowd contagion of running away, except in this case forces that have been pent up by confrontational tension, run forward into the vacuum left by a sudden weakness on the other side. It is an adrenaline surge that has been kept in suspense, suddenly released into action. That is why the police go out of control, swinging at anything in their path. It is important to see that this is a reciprocal emotion—the crowd running away is the counterpart of the police running forward, the display of emotional weakness feeding the surge of dominance of the attackers. In Nassauer’s data, she often finds that when one cop at the front swings at a target—it may be a person who has stumbled and fallen to the ground—the cops just behind will also swing at the same target. One policeman’s attack leads others to repeat the attack.

But—and here is the good news—Nassauer also finds that these attacks can be stopped locally.  When an individual stands still, directly facing the police, and calls out in a strong, clear voice:  “We are peaceful. What about you?” – or words to that effect, the attack almost always stops. 

This does not mean that the riot as a whole can be stopped in this way.  There can be hundreds or thousands of persons spread out over a considerable space. Violence in a riot is not like one huge rugby scrum, not like huge battle-lines of ancient phalanxes, but a series of little clusters of violence here and there.

Each one of these clusters may be checked, could be rendered no longer violent, by the right local action.

To repeat: the details are important. The peace-making person must stand still, no longer moving. When almost everyone’s back is turned, he or she stands in direct eye contact with the on-coming forces. And one’s voice must be clear and steady, neither threatening nor fearful.

Especially important is not to scream.  Someone in the crowd, in the fear or rage of being attacked, can cry out the identical words:  “WE ARE PEACEFUL!! WHAT ABOUT YOU!!” but in this case it will not work.  The police perceive and feel the crowd as being out of control. To scream at the police does not correct this impression, but reinforces it. Screaming is an expression of being out of control; and that is precisely the problem with the interactional situation. Tension and fear pervades everything, and the violence is coming out of the situation of one-sided emotional dominance by the police. The victim who screams does nothing to change the emotional field. It is the strong, calm tone that changes it, back towards local equilibrium, where the violence stops.

A similar technique can work when it is not a confrontation of police (or soldiers) versus a protesting crowd, but a violent attack by one crowd upon another. David Sorge, in research at University of Pennsylvania (2014), shows that in an incident of communal violence in India, the technique was used that stopped violence in a specific location.  The individual under attack was a peace-maker, a citizen who had stood up in a town meeting the day before, to urge the Hindu populace not to pay attention to rumours and not to attack the local Muslims.  As often happens in the early phases of communal violence, the peace-maker became targeted as a traitor. A crowd gathered in front of his house and pelted it with stones, the usual preliminary to an attack. But the peace-maker came out of the front of his house carrying a chair. Before anyone could attack him—there is usually a time-lag of shouting before someone starts the personal assault—he stood up on the chair and started to make a speech in a loud voice. The crowd quieted down and eventually dispersed.

Notice the details. He stood up above the crowd, where he could be seen. He met them face to face. For the members of a violent crowd, usually the target is someone anonymous up there behind all the surging bodies; for the few in the front with clearer visibility, someone cringing, showing weakness and fear, usually cowering, hiding their face, or knocked to the ground where all we can see is their side and back. Standing up in a prominent position, in this instance the peace-maker remained a human individual. He spoke in a loud, strong voice, not in anger, but resolutely.  He spoke to them as individuals, and took apart the collective emotion of the crowd, where each relies on the others to carry out acts of violence that ordinarily would outrage our moral sensibilities.

Again, we must recognize, it was a local solution only.  The riot as a whole was not stopped. The crowd moved elsewhere, where emotional dominance was easier to establish. Nevertheless, this is a hopeful sign. The whole pattern of a riot consists in all its local parts; and the more of these parts that can be stopped, the less damage it does.
Practical Advice in Violent Crowd Situations

--Don't turn your back.

--In a situation of violent threat, don't hide your face.

--Don't run away in panic.

--Above all, don't fall down.

That is to say: your eyes and your face are your strongest weapon of defense.

--Keep up a clear confrontation with a potential attacker. But don't raise the level of tension; don't scream; don't make further threats; just keep it steady as you can.

--Don't get isolated as a single individual surrounded by a cluster of about half a dozen attackers. This is the configuration in photos where persons are badly beaten. Try to stay with at least a small cluster of your own side, but not in the panicky flight mode.

I should add that this advice is for non-violent participants. It is unclear that they will work if you are throwing rocks, fire-bombs, or engaging in other kinds of violence.

This advice is drawn from research on the micro-sociology of riots. Does it work in other kinds of threatening situations, both more organized or macro-structured violence such as massacres and war, and in more individualized confrontations like street fights?

Our field of research has much more to do in examining all these types. But so far, the results are optimistic. The desk clerk in the Atlanta school on August 20, 2013 [http://edition.cnn.com/2013/08/21/us/georgia-school-gunshots/] who calmed down an armed man threatening a rampage shooting shows that even the most dangerous situations may be defused. Research colleagues have told me they have walked safely through a violent riot in Tehran, by keeping in mind what emotional tone they were projecting in their body language, playing neither attacker nor victim.  Stefan Klusemann's research (2010, 2012), on the tipping-points to genocidal ethnic cleansing in Bosnia, and in Rwanda, shows that even in the midst of a mass-murder campaign, there are micro-situational stumbling blocks, and threatened victims sometimes escape by a timely show of emotional resoluteness.

Lowering the Tension: Putting the Situation Back in Emotional Equilibrium

When the confrontation is one-on-one, the prospects are especially optimistic that violence can be avoided.  We are accumulating a significant amount of data on such situations, and two patterns stand out.

First:  The audience has an important effect.  Public fights rarely get very far without audience support.  Most angry arguments stay at the level of bluster and insult, unless the audience shows that it wants them to fight. The audience that cheers and urges them on will almost always get a prolonged fight. A neutral or uneasy audience, standing at a distance, usually results in a brief fight without much damage. And when an audience (or part of it) tries to intervene, it is almost always successful in stopping a fight. This pattern is shown in my comparison of fight incidents with different audience reactions (Collins 2008); in British research using CCTV videos of fights in pubs (Levine et al. 2011); and studies of what network relationships result in successful third-party interventions (Phillips and Cooney 2005).

There are limits to this pattern. It applies to arguments and fights in public, but not to domestic violence, which often takes place without much of an audience. (But there is ongoing research here, too, investigating the effects of indoor audiences that may be present.)  Since domestic violence is a considerable portion of small-scale violence, that is a serious limitation on our optimistic news. On the other hand, the following point applies both to domestic and public violence:

Second:  Small-scale conflict and violence peters out when the emotional field is in equilibrium.  That means:  when both sides are showing the same amount of emotional energy, the same degree of bodily agitation, the same emotional intensity.  This equilibrating effect can take place at any level of intensity, as long as both sides are evenly matched.  Research on mobile phone videos of street fights (Jackson-Jacobs) shows that even in the case of fights that have already started (and where the audience is distant and neutral) tend to wind down after both sides have thrown a few blows. On the whole, evenly matched fights do not do very much damage; the high level of adrenaline arousal makes fighters sloppy and incompetent, and Jackson-Jacobs’s videos show the fighters who have thrown a few wild punches tend to let their swings carry themselves out of range, where the fight devolves into threats, and eventually into mutual disengagement.  After all, in an honor fight, it is showing one’s willingness to fight that counts, not the result.

Let me conclude with a favorite photograph.  It was taken in Jerusalem during the height of the second Intifada, and it shows an Israeli soldier and a Palestinian political leader locked in angry conflict.

Jerusalem stalemate 2000

The news story tells there was no violence at this flash point of sacred territories that day. The angry confrontation wound down and ended. How?

The photo shows the two men exactly mirroring each other. Reading the facial expression of emotions using Ekman’s methods, we see them displaying anger, and in an identical manner: both have the hard, staring eyes, the clenched eyebrows with the vertical line between them, the square, shouting mouth. As is characteristic of angry talk, both are vocalizing at the same time, not listening to what the other has to say. Their faces, like their bodies, are tensed like muscles about to strike. But they do not strike.

They are in equilibrium at a high level of intensity.  From similar incidents observed over a few moments of time, we can surmise that they eventually become tired of the situation. No one else in the crowd is taking up their level of intensity; they are doing all the audience’s work for it. It is boring to say the same thing over and over again, getting no intelligible response. They will deescalate, going down the scale of emotional intensity simultaneously, keeping in equilibrium step by step.

They will become bored. And in situations of conflict, boredom is the pathway to peace.

The Tank and the Human Face

Micro-sociology delivers some good news. Some kinds of violence we are able to mitigate. This is on the micro-level, face-to-face with a potential attacker.

Another level is harder to handle, or at least it will take another approach. This is violence at the level of the organization or bureaucracy. 

If we take the column of tanks as a symbol of the hundreds of military vehicles and thousands of soldiers in Beijing, we are seeing the public face of an organizational network stretching far off into the distance. Orders to advance are given somewhere else, by a face we never see, a voice we never hear. Techniques of human face-to-face confrontation will not work here.


This is not to claim that the distant strategists are purely rational and coolly calculating. Their decisions are made in an atmosphere of emotions pervading the network of organized power, in counterpoint to the waves of emotions among the crowds who come into the streets over a period of weeks. It may be a long-distance chess game, but one played in shifting moods of anger or fear, confidence or deflation, righteousness and revenge-- and occasionally magnanimity.  The macro-level pattern is one of counter-escalation and de-escalation, and it has its time-dynamics that spread over weeks and months (Collins 2012).

There may be grounds for optimism about what sorts of processes can head off violence on the macro level too, although in some phases there is a kind of steam-roller momentum that is extremely dangerous once it gets rolling. We are learning about these kinds of time-dynamics, hopefully adding more tactics to the toolbox for peace.

In the meantime, as individuals in threatening situations, we can do our bit.

Appendix: How to regain calm when your heart is pounding

It's all very well to say, turn and face your attacker, call out in a firm strong voice, don't run and don't panic. But how do you manage to do this if your adrenaline is pumping and your heart is going 160 beats per minute?

There is a technique that will bring your heartbeat down, and with it, the panicky effects of adrenaline and the inability to control your voice. A useful version is described by Dave Grossman, a psychologist formerly with the US Army:

Repeat the following sequence of breathing:
-- breathe in slowly, counting 4 seconds (one-alligator, two alligator, three-alligator, four-alligator)
-- hold your breath for 4 seconds (counting...)
-- breathe out slowly, counting 4 seconds
-- hold your breath out (lungs empty), counting 4 seconds

do it again:
-- breathe in slowly, 4 seconds
-- etc.
as many times as you need until your get your breathing and heart rate under control.

Remember the details. This is not the simple cliché, take a deep breath. It is the rhythm you are after, the timing of how long each breath and holding period is. Your goal is to change your body rhythm. And after you accomplish that, to change the rhythm of the person confronting you.


Randall Collins. 2008. Violence: A Micro-sociological Theory.
Randall Collins. 2012. “C-Escalation and D-escalation: A Theory of the Time-Dynamics of Conflict.” American Sociological Review  77: 1-20.
Paul Ekman,  1985. Telling Lies: Clues to Deceit in the Marketplace, Politics, and Marriage.
Paul Ekman and Wallace V. Friesen,  1975. Unmasking the Face.  
Dave Grossman. 2004.  On Combat. The psychology and physiology of deadly combat in war and peace.
Curtis Jackson-Jacobs. 2011. "Social Organization in Violence." Research in progress, UCLA.
Stefan Klusemann. 2010. “Micro-situational antecedents of violent atrocity.”  Sociological Forum 25:272-295.
Stefan Klusemann. 2012. "Massacres as process:  A micro-sociological theory of internal patterns of mass atrocities." European Journal of Criminology 9: 438-480.
Mark Levine, Paul J. Taylor, and Rachel Best. 2011.Third Parties, Violence, and Conflict Resolution” Psychological Science 22 (3) 406–412.
Anne Nassauer.  2013.  Violence in demonstrations.  PhD dissertation, Berlin Graduate School of Social Sciences.
Scott Phillips and Mark Cooney. 2005.  "Aiding Peace, Abetting Violence: Third Parties and the Management of Conflict." American Sociological Review 70: 334-354.