Reports of the pro-democracy protests in Tiananmen Square and their violent end have had a huge impact on how the outside world sees China. James Miles - who was the BBC's Beijing correspondent at the time - reflects on the difficulties of covering the story.
The first draft of history can be crude. Even if the thrust of a story is well described by journalists on the scene, some of its details might need refinement, and sometimes even correction.
Such was the case with the massacre in Beijing on 3 and 4 June, 1989. I was one of the foreign journalists who witnessed the events that night.
We got the story generally right, but on one detail I and others conveyed the wrong impression. There was no massacre on Tiananmen Square.
Looking back at the radio despatches I filed for the BBC, the following words stand out from a draft script that I sat down to write back in the office at 0230 on 4 June: "Around two in the morning, troops using armoured personnel carriers crashed through barricades set up by residents on the outskirts of the square.
"Eyewitnesses said thousands of troops later poured into the square, firing as they advanced. Tens of thousands of students and workers crouched in the centre of the square."
Scripts are often revised by hand at the last minute in the studio, but the actual words as broadcast probably did not differ considerably from these.
Towards midday on 4 June, amid reports of widespread casualties, I wrote in another draft that "many of the deaths occurred at Tiananmen Square, not only from gunshots, but also from being crushed by tanks, which ploughed relentlessly through any obstacle in their way."
Reports by other foreign journalists conveyed a similar impression. "Death in Tiananmen; Witnesses Describe the Devastating Assault" said a Washington Post headline on 5 June.
Evidence of a massacre having occurred in Beijing was incontrovertible.
Numerous foreign journalists saw it from widely scattered vantage points.
On the morning of 4 June, reporters in the Beijing Hotel close to the square saw troops open fire indiscriminately at unarmed citizens on Chang'an Boulevard who were too far away from the soldiers to pose any real threat.
Thirty or 40 bodies lay, apparently lifeless, on the road afterwards.
That scene outside the Beijing Hotel alone justified the use of the word massacre. But the students who had told me and other journalists of a bloodbath on the square proved mistaken.
Protesters who were still in the square when the army reached it were allowed to leave after negotiations with martial law troops (Only a handful of journalists were on hand to witness this moment - I, like most others at the time, had spent the night in various different parts of the city monitoring the army's bloody advance).
A few of the students were crushed by armoured vehicles some distance from the square after the retreat.
There were credible reports of several citizens being shot dead during the night on the outer perimeter of the square, but in places which strictly speaking could be said to be outside the square itself.
But we are far less certain of killings on Tiananmen proper. There were probably few, if any.
We are also still unsure how many people were killed that night. The government said 200 citizens died (from stray bullets and shootings by thugs), in addition to dozens of troops. The likely toll is almost certainly higher.
Nicholas Kristof of the New York Times, who did some admirable detective work in Beijing hospitals in the weeks after the massacre, said in a report published on 21 June 1989 that "it seems plausible that about a dozen soldiers and policemen were killed, along with 400 to 800 civilians".
The standard line now used by foreign journalists is that "hundreds, possibly thousands" died.
The Chinese government was quick to exploit the weaknesses in our reporting.
By focusing on what happened in the square itself, it began sowing seeds of doubt about the general accuracy of Western reports among Chinese who did not witness what happened.
At first this made little difference, since most Beijing residents at least had friends of friends who had seen for themselves that there had been a massacre, even if not in the square.
But as the years passed, a new generation emerged with few eyewitness accounts to cling to.
Public discussion of Tiananmen was taboo, and those who had lived through its horrors became increasingly disinclined to dwell on them.
Occasionally I meet young Chinese who work out from my dates of residence in Beijing that I might have been around in June 1989.
Bolder ones venture to ask what happened, and whether there was really a massacre.
Last year in the western province of Gansu, in a town on the edge of the Gobi desert, I was astonished when a young taxi driver reeled off the names of several of the prominent student activists and intellectual leaders of that time.
Their names now, if known at all to younger Chinese, are very rarely heard in public or private.
China's rapid economic growth, and the far less stellar progress of the ex-communist states of Europe, has helped give credence (among many Chinese) to the government's claim that China would have descended into chaos were it not for the crackdown.
Beijing was peaceful in the days leading up to the massacre and many students were beginning to grow weary of the protests.
But it is not uncommon to find Chinese who believe the Communist Party's fiction that there was a riot in Beijing on 3 June that warranted intervention.
Rioting did occur, but involving angry residents outraged by the army's brutal entry into the city.
I believe that eventually, as part of a process of political change in China, the government will revise its official account of what happened.
We journalists have long since revised ours, but misleading terms persist.
These terms can be faulted on points of detail. But their failing could also be said to be that they understate the magnitude of what happened.
There was no Tiananmen Square massacre, but there was a Beijing massacre.
The shorthand we often use of the "Tiananmen Square protests" of 1989 gives the impression that this was just a Beijing issue. It was not.
Protests occurred in almost every city in China (even in a town on the edge of the Gobi desert).
What happened in 1989 was by far the most widespread pro-democracy upheaval in communist China's history. It was also by far the bloodiest suppression of peaceful dissent.
James Miles is now the Beijing correspondent of The Economist, and author of The Legacy of Tiananmen: China in Disarray (University of Michigan Press, 1996).