Six years later, a Canberra public servant, David Eastman, was convicted of murdering Winchester. The conviction followed a chain of circumstantial evidence which linked, first, bullet cartridges found on the ground next day to a type of gun, then to a particular gun linked to a Queanbeyan gun dealer, and some particularly dubious evidence suggesting Eastman had bought it. Gunshot residue at the scene appeared similar to gunshot residue vacuumed from the boot of Eastman's car. Eastman, who was unable to account for the gunshot residue in his car, had florid mental health problems. At trial he fought continually with his lawyers, often sacking them and hiring others, sacked in their turn. He was often unrepresented, refusing to participate, and there was little cross-examination of the forensic evidence.

What we think about the Eastman case is now affected by another murder, 10 years later, in suburban Fulham, in London.

In 1999 British TV presenter Jill Dando was about to enter her London house when a man walked up and shot her once in the head. There were no witnesses.

A year later, police searched the flat of a Barry George, a man who had been seen in the neighbourhood about four hours before the murder. He had a jacket, hanging on a kitchen nail. A vacuum cleaner put in its pockets produced, on close examination, a single speck of gun-shot residue: it was 11.5 microns in diameter - about a hundredth of a millimetre. It was ''similar to'' gunshot residue found on the cartridge case, and in Dando's hair.

This, the Crown alleged, was an important link in a chain of circumstantial evidence proving George was the murderer. The chief forensic witness, Robin Keeley, said the particle was ''consistent with'' having come from the bullet used in the killing. The judge showed he thought this was indicative of guilt, and, although he noted defence objections to its significance, was dismissive of the idea that the residue could have come from anywhere else.

The jury convicted George.

George, like David Eastman, continued to protest his innocence. Repeated appeals failed. Appeal courts were not much interested in a review, using phrases, again redolent of the Eastman case, saying ''the residue evidence was capable of supporting the prosecution case; its weight was a matter for the jury''. Yet, as continuing critics pointed out, jurors were in no great position to assess the scientific merit, or weight, of an arcane art such as gunshot residue analysis.

As it happens, one of those uneasy with the verdict was a colleague of Keeley, Dr Ian Evatt, who worked in the British Forensic Sciences Institute. His role involved helping scientific colleagues with their scientific method, and making sure that evidence put before the courts was ''logical, transparent and robust''.

His very job was a response to a criticism often made of some forensic scientists, including in Australia. Some are accused of losing detachment and becoming advocates for a side or a theory of a case (most often the prosecution side). Some, having made a tentative honest hypothesis, tend, unconsciously to look for material supporting it and to ignore evidence against it. Moreover, it is said, they tend to overstate their expertise or the significance of their conclusions, and to understate the limitations of, or caveats attached to, their conclusions.

Evatt was developing a new standard technique called Case Assessment and Interpretation. The purpose of this technique, now widely accepted around the forensic world, is to clarify (before evidence is examined or analysed) what findings would support or tend to disprove particular hypotheses.

Evatt, at the time of his initial unease, knew nothing particular about the case, apart from some general gossip about the labs, before trial, that the forensic evidence was ''neutral'' - which is to say that it neither particularly supported nor tended to disprove George's guilt.

He was thus puzzled, in reading newspaper reports of the trial, to see that the prosecutor, and the judge, in addressing the jury, was tending to rely on it as a proof of guilt.

Evatt and Keeley decided (because Keeley was at heart a scientist) to apply the CAI techniques to the gunshot residue evidence. The propositions were (1) George shot Dando or (2) he was not the man who shot Dando.

Looked at this way, Keeley estimated that the chance of not finding a particle of residue in the pocket was 99 in 100 on either proposition. The chance of finding a speck was one in 100 on either proposition. The Crown, and the judge, were at least unconsciously unbalanced, or biased, because they had simply not understood the witness. While it was unlikely that the speck had come from somewhere else, it was equally unlikely that it came from the gun that killed Dando.

The lab, and now all reputable labs, and later, the courts, changed the rules on gunshot residue evidence. These days no one attaches much significance to finding low levels of residue. Alas, that case was resolved only in 2007, 18 years after the Winchester murder, and 12 years after Eastman's conviction.

Critical evidence which helped convict Eastman would not be regarded as admissible today. This is not because of some ''legal technicality'' about the admissibility of evidence. It is because modern science now understands that conclusions they once drew confidently, and methods they once used, have been shown to be wrong, or, at the least, unsustainable. A true forensic scientist is at heart a scientist, and scientific knowledge, technique and capacity continually improves.

But, as it happens, much more is involved in the Eastman case than ''new'' knowledge. There seemed to be significant weaknesses in gunshot residue evidence from the start. The chief ''expert'' used by the AFP was Robert Barnes, who had been in the Victorian Forensic Science unit. Barnes gave different evidence at different times, often without explaining how or why he had changed his mind. He changed his mind about the type of gun used, whether a silencer was used, whether the rifle barrel (if it was a rifle) was shortened, and about what the residue tended to prove. Like detectives he has always assumed, without evidence, that cartridges found, trampled in mud near Winchester's car a day or two after the murder, were from the murder weapon.

Barnes was a confident, even dogmatic, witness. But, because of Eastman's antics, he was never cross-examined, asked about contradictions in his statements, or even about his tendency to overstate his qualifications, a one-year trade certificate in metallurgy. And, as the judge had in the George case, the trial judge, Ken Carruthers from NSW, seemed to form an early antipathy to the defendant and, if formally fair, gave him no quarter, because he thought Eastman was faking mental illness as a tactic.

Dr Jim Wallace, of Belfast, is the generally acknowledged international expert on gunshot residue - and author of the standard text on the subject. He is scathing of Barnes and his findings.

The AFP has said that international labs, in Israel and the FBI, have supported Barnes's conclusions. But these did not repeat tests; they simply read Barnes's reports and said that if tests found what they said, the conclusions would be sound.

At trial, the jury and defence were not told that Barnes had been recently sacked after Victorian judges and his own scientific colleagues had raised serious concerns about his methodology, his tendency to overstate conclusions, and the fact that, in several critical cases, his overconfident and dogmatic evidence had been shown to be wrong.

A year ago, I asked Victoria Police, under FOI for its file on the investigation. It decided to give me all but purely personal documents on him. But Barnes, as a third party with rights, has resisted disclosure, saying it would breach his privacy. He effectively gave up his resistance late last month, but the documents are not yet handed over.

A review of the safety of Eastman's conviction should not begin with a lawyer's picnic, a cascade of ''he said-you said'', or attempts to cross-examine witnesses about events 23 years ago. Long before it comes before a judge, or lawyers, there ought to be a calm and cool independent review, from the papers, transcripts and such exhibits as remain of the police investigation and the forensic evidence, including some new items which have emerged. This should lead to a report to the judge, counsel assisting and representatives of Eastman, and a hearing about what follows.

No AFP officers should be involved: from Day 1, when senior officers and detectives failed to secure the crime scene and tramped all about it, this was an inadequate and seriously compromised case. That's why so many entertain doubts about the safety of the conviction.

But I expect nothing of the sort, because that would involve some initiative by ACT Attorney-General Simon Corbell, of whose conduct Justice Shane Marshall, ordering a judicial inquiry, was scathing yesterday. On a pretence of saving money, there will be ''justice'' on the cheap, and - I should not be surprised - heroic, massive and dubious expenditure by the DPP or police to confine and narrow external review of their work. If politicians had any guts, the doubts would have been resolved, one way or another, years ago.