The jury had retired to a hotel for the night. A ouija board was used to consult with the deceased - and to ask him who the killer was.
Contact was supposedly made, and the victim named the accused as the murderer. When the guilty verdict was challenged, the British Court of Appeal decided it could investigate only what happened at the hotel - but the jury room was out of bounds.
Peeking into the minds of "12 good men and true" has long been considered taboo; the sanctity of the jury room is to be protected at all costs.
But two recent NSW cases have persuaded the judges of the Court of Criminal Appeal that juries need more explicit instructions on how to go about their work.
In December, the court quashed the conviction of a twice-accused murderer because jurors had gone to the internet to learn about his previous trial.
And an "experiment" by two jurors during the trial of the notorious gang rape ringleader Bilal Skaf could end up leading to a new trial for Skaf and his brother Mohammed.
The court is clearly sick of this amateur sleuthing and wants the power to declare such jurors in contempt and to send them to jail for up to two years.
But how far can it go in investigating jury verdicts? And when does an irregularity become so material that appeal judges will order a retrial?
The long-standing rule is that a court can consider the nature of the information or experiments, but not whether those experiments affected the deliberations of the jury.
If it believes the sleuthing amounted to gathering evidence, the next question is whether it can "be satisfied that the irregularity has not affected the verdicts, and that the jury would have returned the same verdicts if the irregularity had not occurred".
In the internet case - R v K - Justice Wood said the only way to answer the question with any certainty was to kick down the jury door and the court would not do that. It would have to presume the "potentially prejudicial" searches affected the verdict.
Interestingly, a juror in that case also visited the alleged crime scene - conduct which Justice Wood said was "clearly wrongful". But he said that without knowing if he had done anything more than pass by the house, the visit was not enough for the court to intervene.
In the Skaf case the two jury members did much more.
Justice Wood was unimpressed by the Crown's argument that their visit to Gosling Park, Greenacre, was an inspection rather than an experiment. "Why not? Two people go to different parts of the park and ask, 'Can you see me'? Surely that is an experiment," the judge said.
If the experiment had happened during the trial "this clearly would have been part of the evidence collating exercise". That would have allowed the defence to argue - as it did yesterday - that the foreman and the other juror had not conducted their experiment in the same conditions as existed on the night of the rape.
Only scant details of the sheriff's report into the Skafs' case have been made public. However in R v K, the sheriff looked at not only what searches each juror had made, but whether they had shared that information with other jurors and whether it affected their own deliberations.
That information may have been helpful in determining whether their view of the accused had been prejudiced, but Justice Wood said it could not be considered.
"This is consistent not only with important public policy considerations, but also with the long-established rule that the court will not hear evidence of the deliberation of the jury."
That rule was reaffirmed this year in a case the court said could help guide it on jury reform - Regina v Mirza in the House of Lords.
A Pakistani man was found to have sexually abused his stepdaughter. He failed to win a retrial despite one juror telling Mirza's lawyers that most of the panel considered his use of an interpreter to be a devious ploy.
Lord Steyn was alone in arguing that the jury "reached their verdict on perverse grounds which included a pronounced racial element".
The rest of the law lords relied on a 1785 ruling that a court cannot hear evidence about jury deliberations. They said that to unlock the door to the jury room and find out what happened would only undermine confidence in the jury system.
But there are a number of cases which make it clear that when matters extrinsic to the jury's deliberations are brought to the notice of the court that it will intervene and set aside a verdict.
This happened in 1992 when some handbags were tendered into evidence at a NSW murder trial. It was later found that notes which the Crown said should have been removed had alleged the accused was paranoid, delusional and obsessive and had "damn near succeeded" in an earlier attempt to murder her.
The Court of Appeal said the appeal should be allowed because of the extremely prejudicial potential of the material. As in R v K it said it could not be satisfied that the irregularity had not affected the verdict. It is clear from yesterday's appeal that Justice Wood also regards the "experiment" at Gosling Park as potentially prejudicial.
However, the deputy director of public prosecutions, Greg Smith, argued there was plenty of other evidence which showed the complainant had clearly identified Bilal Skaf.
Skaf's barrister, Peter Zahra, SC, from the Public Defender's Office, argued that the experiment may have resolved the issue of identification. The foreman had asked the judge if they could go home early because there was "a little bit of frustration" among jurors.
The foreman and one other jury member went to Gosling Park that night. The next day deliberations continued for less than two hours before the guilty verdicts.
Last week's decision to hear appeals by the Skaf brothers is already causing confusion on the bench.
At the start of a murder trial on Monday, Justice Bruce James said he would seek advice on what directions should be given to his jury in light of yesterday's appeal.
Yesterday the state Attorney-General, Bob Debus, yesterday said he was "consulting on these issues and will also wait to see what, if any, recommendations the court makes in relation to jury reform".
It is understood he has invited comment on Justice Wood's view in R v K that there was much to like about a provision in Queensland's Jury Act that dealt with inquiries.
He said it should be inserted in the NSW act and widened to include "any matter relevant to the trial, including the history of the offence, and its investigation, and the proceedings themselves".
And he said juries should be warned not to undertake any independent research concerning the proceedings and that it would be an offence for them to make any such inquiry.
The court expanded on these suggestions when the visit to Gosling Park came to light on March 4 and even asked for submissions from the Director of Public Prosecutions and Solicitor- General on possible jury reform.
Justice Mason even suggested that jurors should be required to dob in those who do their own thing during a trial.
During the past decade there have been two systematic jury surveys in NSW. One, led by Professor Michael Chesterman, of the University of NSW, studied the influence of media publicity on juries. It found juries were capable of ignoring pre-trial reports and were appalled at the standard of reporting of their trials.
The other, led by Professor Mark Findlay, for the Australian Institute of Judicial Administration, surveyed jurors on all aspects of their experience. He said they were are a lot more intelligent but not as well informed as the court system gives them credit for.
After serving on a District Court jury two years ago, Herald writer Malcolm Knox decided to write a book, Secrets of the Jury Room.
He said he agreed with the calls for an overhaul of the jury system. "Not to abolish unanimous verdicts or to decrease jury sizes or, heaven forbid, reduce the use of juries, but to supply ordinary people with a clear guide to the criminal jury trial, to make the thing fairer and to let the community profit from the experience of those who undertake this pivotal civic duty."
A leading British judge, Lord Justice Auld, has already said the British Court of Appeal should be allowed to investigate jury misbehaviour - whether it occurred during deliberations or not.
However, in Mirza the law lords said it was such a fundamental departure from existing law that Parliament - and not judges - should be the ones to make the change.
Lord Steyn said this would mean there would be no review even if jury members tossed a coin to arrive at a verdict or some urged a conviction because the accused was a black immigrant.
"It is to the effect that in the interests of maintaining the efficiency of the jury system the risk of occasional miscarriages of justice may acceptably be tolerated. In other words, one must accept some dubious verdicts . . . as to the cost for protecting the jury system."