The case of Robert Kennedy

August 31, 2011

Arthur Eathorne, found murdered in his flat in 1990

In a recent Australian TV magazine feature on the Meredith Kercher case, the British journalist Nick Pisa said something I found a little odd. In his view, it seems, the case would never have come to court had the murder taken place in the UK, because the evidence is just insufficient.

Now, there are worse things in the world than being a little chauvinistic about standards of justice. But I don’t think that’s exactly what’s going on here. Many people who seek to defend Amanda Knox and Raffaele Sollecito do so, in my opinion, on the basis of a wholly unrealistic view as to how much evidence ought to be enough to convict someone in a court of law.

There’s a lot of evidence in this case. You only need to be able to to count to know that. Of course, for each item of evidence there’s a defence argument, with greater or lesser plausibility, that it doesn’t say what it might appear to say. But the whole purpose of taking the matter to court is to have those arguments adjudged. No prosecutor has ever dropped a case on the grounds that footprints found near a body which tests say were probably made in blood might actually have been made in fruit juice. Or that falsely accusing someone of a murder is a relatively normal thing to do if you are suffering from stress.

I thought I’d take a look at a British case which, in my view, was prosecuted on evidence which was far thinner than in the Kercher case.

Robert Kennedy is a former prostitute, convicted in 1991 of murdering a client. He has always maintained his innocence and has unsuccessfully appealed his conviction twice, the last time being in 2007.

The only physical evidence linking Kennedy to the scene of the crime, the victim’s home, came in the form of two cigarette butts. These were believed by police to have been left by the culprit because a taxi driver had called at the house a short time before the murder, and he testified that the ashtray in which the butts were found was empty at that time. The butts were not initially examined for DNA, but saliva on them was found to be of a type matching Kennedy and also matching about 25% of the population. This evidence was admitted at trial. Years later, the saliva was tested for DNA and it turned out not to be Kennedy’s, but that of an unidentified person.

The main evidence against Kennedy comes from two witness statements relating to a conversation that took place at Kennedy’s shared house. The first witness said that she walked into a room in the middle of a conversation between Kennedy and four other people. They were, according to her, discussing a murder that had happened locally. Since news of the murder had not yet broken in the media, it is unclear how they knew about it. The witness also recalled Kennedy mentioning something about an ashtray and then being told to shut up by someone else present. This is significant, because a large glass ashtray (not the one with the cigarette butts in) was missing from the scene of the crime and may have been the murder weapon. Witness number one also seemed unsure about whether the conversation had taken place on the day of the murder or the day before, but that does not seem prevented her testimony being accepted by the trial jury.

Witness number two was party to the same conversation, but did not hear the word “murder” used, only talk of some incident or other involving two men and a woman. She said that Kennedy had been told to “shut up”, but that this was not anything to do with him mentioning an ashtray. She couldn’t remember what night the conversation had taken place on. She also recalled seeing a pullover with blood-stains on it, but wasn’t completely sure whether that was on the same night.

According to the police, Kennedy had lied to them on two occasions. Firstly, he had at first denied, then later admitted, having had sex with the victim. I think it seems obvious that even an innocent person might have told that lie. Secondly, he told police that he had read in the newspaper that a large glass ashtray was missing from the scene (a fact that had indeed been reported in the media) and offered a description of it from his recollection, saying that it was oval and had a label saying “Hofmeister”. But enquiries with the makers of Hofmeister lager revealed that they had never produced an ashtray fitting the description given by Kennedy.

On remand, Kennedy had also been alleged to have confessed to three fellow inmates. One of these was applying for parole and had agreed with the police that he would try to get information for them by talking to Kennedy. A second was also applying for parole and told the police he had information about Kennedy but would not tell them anything further until he found out whether or not he was going to be released. A third had no parole hearing in process, but had given details about Kennedy’s alleged account of the crime which describing things which were known not to have happened. Kennedy’s story was that he had spoken to all three of the witnesses about his case, which is how they were able to concoct accounts for the police. But he denied having confessed to any of them.

The defence produced evidence that the victim had been wearing a wind-up watch which had stopped a certain time after the murder. Unless this was almost fully-wound when the crime occurred, Kennedy could not have been the murder, because he had a solid alibi.

Those are the facts of the case. It might seem to you like an enormous miscarriage of justice. After all, prisoners, particularly those who might be trying to bargain for a release, are not always reliable witnesses. The witness who heard Kennedy talking about a murder and an ashtray is clearly contradicted by another witness. And surely it is possible that Kennedy just made an honest mistake in describing the ashtray to police. Add this to the dispositive evidence in terms of the DNA and the thing about the watch, and there must be reasonable doubt. Mustn’t there?

I’m not saying any of that is wrong. But perhaps it makes incorrect assumptions about the normal standard of proof in criminal cases. Presumably, the jury heard and evaluated the witnesses, took the evidence before them into account and felt able to make a decision. “Reasonable doubt” is not the same as “no possible objections”.

In the Kercher case, as well as the DNA evidence which is currently the subject of debate, we have all the evidence of a staged break-in, footrprints made in blood (or, possibly, pear juice), Knox’s DNA found mixed with the victim’s blood, her false accusation that someone else committed the murder, two witnesses who saw Knox and Sollecito near the murder scene at times when they claim to have been elsewhere. Maybe the court will eventually decide that all the evidence was misleading. But the process is more-or-less the same as it would be for a British jury. How much do you trust the witnesses? How certain do you feel you need to be about the science? And, for prosecutors, the decision about whether to go to trial is mainly based on making a guess as to what the chances of conviction are, on that basis. I think the idea that all that evidence was never enough to even put before a court just doesn’t show common sense.

By the way, apologies to anyone who Googled in here expecting to find information about a link between Amanda Knox and the other Robert Kennedy.

R v Kennedy [2007] EWCA Crim 3132

Knox/Sollecito: How to spot a fake burglary.

August 8, 2011

In the ongoing appeal of Amanda Knox and Raffaele Sollecito against their convictions for the murder of Meredith Kercher in 2007, attention has recently been focussed on two pieces of evidence, a clasp from the victim’s bra and a knife, said by prosecutors to be incontrovertible proof linking the DNA of the two suspects to the crime. There will be more courtroom discussion about that in September but, as things stands, it does not seem easy to guess what the view of the court with regards to these two items will eventually be.

If the DNA from the knife and clasp proves resilient to assaults on its value, then it would seem likely that the convictions will be upheld. But what if that is not the case? Supporters of Knox and Sollecito will point out the central position of this DNA evidence within the prosecution case. Take it away, and what is left? Those who believe that Knox and Sollecito are guilty will argue in return that there is plenty left. This was never a case that hung by a single thread.

In a way, both sides may be right. Clearly, if you take away two pieces of evidence that, essentially, purport to show that Knox stabbed Kercher and that Sollecito removed her bra after she died, then the case against them can only be weakened. But by how much? Can the rest of the evidence stand up on its own, or has it always relied on the knife and clasp to hoist it out of the mists of uncertainty through corroboration?

Prosecution lawyers have said that, even if the current battle is lost, there is still DNA evidence, in the form of mixed samples from Kercher and Knox found at the scene, which they believe is equally compelling. But that is one of a number of possible views on the case. In the event that the knife and clasp go MIA, I wonder whether the best part of the prosecution case to deputise might be the question of whether Knox and Sollecito faked a burglary at the crime scene in an attempt to cover up their crime.

The day after the murder, police officers arrived at the flat where it had happened because two mobile phones used by Kercher had been found abandoned. When they arrived, Knox and Sollecito told them that they were concerned because their appeared to have been a break-in and there were spots of blood in the flat. A short while later, Kercher’s bedroom door was broken down and her body was discovered.

Investigators soon arrived at a belief that the apparent break-in was in fact simulated. This was, perhaps, the earliest indication that Knox and Sollecito were involved in the murder. Apart from Kercher, Knox had been the only person in town with a key to the flat. What’s more, it seemed unlikely that anyone else could have a motivation to interfere with the crime scene in the manner that had taken place. A burglary would draw attention away from the occupants of the flat, but from who else? And since the other occupants of the flat were not in town…

The potential power of a staged crime scene hypothesis is shown by a study conducted in 2000 by the forensic scientist Brent Turvey. This looked at twenty years’ worth of US criminal cases where a crime scene had been staged and attempted to identify commonalities between them. Although this turns out to be only 25 usable cases (all domestic homicides), what is striking is that in 11 of them, convictions appear to have been secured purely on the basis of police and expert testimony as to the staging. And in all the other cases, the additional support to this testimony came in the form of some sort of confession or self-implicating statement by an offender (which is relevant, because Knox made a statement accusing her boss of the murder and placing herself at the crime scene, which she later retracted).

This doesn’t provide a solid guide to the chances of the current appeal being upheld purely on the basis of the burglary having been staged, because Turvey is not able to tell us about the number of cases where there was an acquittal despite an allegation of crime scene staging. But it does give us an indication that such an allegation can be enough to sustain a conviction, provided a court buys into it. Neither DNA evidence, nor eyewitnesses, nor bloody footprints, nor phone intercepts, nor analysis of computer hard-drives are necessarily required.

In Knox and Sollecito’s 2009 trial, the court did exactly that. The appeal is hearing no new evidence on this question, but the judges are entitled to review the evidence from the trial and come to a different conclusion if they so wish. It may be that they have to do just that if the appeal is to be successful. But is that likely?


There is no single handbook that gives a definitive explanation as to how the staging of a crime scene can be established. But there does seem to be broad consensus among experts on a number of guiding principles.

Firstly, the question is, primarily and ultimately, concerned with physical evidence in the form of the state of affairs found at the crime scene. Experts differ as to how relevant non-physical evidence might be in certain cases, but it is clear that, with the obvious exception of a confession with regards to the staging, such evidence can only ever be vaguely indicative one way or the other.

The victimololgy of the case (essentially, does anyone benefit from the death of victim?) may be a consideration. But that does not seem to be relevant here. Another factor might be the existence of a potential suspect who has made attempts, apart from the supposed staging, to steer the attention of investigators away from himself (or herself). An example of this might be where a suspect has made a false accusation against someone unconnected to the crime…

But the final proof or disproof of staging is not in such details. It is in the physical arrangement of the crime scene.

There is a long-established view that the investigation into staging should focus not on attempting to prove it, but on attempting to disprove the competing hypothesis (in our case, that there was ever a genuine burglary). This may sound like a pedantic distinction and perhaps it is. But it is held to be important because it provides clarity of method. First you search for evidence which is inconsistent with the competing hypothesis. You then test these inconsistencies against all the known facts of the case. Can they reasonably be given any explanation other than staging? If any inconsistencies then remain, your conclusion must be that the scene was staged.

Perhaps I am labouring this point. But I’d remark that much Internet theorising in this area with regards to the Knox/Sollecito case focuses on explaining how a genuine burglar might have committe the crime, drawing attention to evidence that is consistent with that account: the position of a shard of glass or the way a rock appears to have landed in relation to the window.

I can understand the urge to do this – obviously is is useful to have an alternative to the account offered by the prosecution. But at the same time, it ought to be obvious why such approaches are problematic from a crime scene investigation point-of-view. People who stage crime scenes typically do everything they can think of to ensure that the staging looks as realistic as they can make it look. Therefore, any evidence that is consistent with a genuine break-in is also consistent with staging. You can only disprove the staging by focusing on the alleged inconsistencies. That is a one-sided approach, admittedly. But it is, apparently, the correct approach.

The inconsistencies

This book offer a three-point checklist as a starting point in identifying staged break-ins: were inappropriate items taken?; did the point of entry make sense?; did the hypothetical perpetrator have a high-risk of being seen?

The items removed from the flat, apart from the murder weapon or weapons, were: two mobile phones, later abandoned; Kercher’s bank cards; probably cash; Kercher’s house keys. The cash aside, it is not clear what value these items would have to a burglar. The phones, cards and keys of someone recently murdered are obviously very risky things to take from the scene. Perhaps it might be argued that our burglar didn’t realise that at first. But why the keys? Indeed, why was the door to the room in which the murder took place locked?

Point of entry is discussed below, but it suffices here to say that our burglar has chosen a particularly difficult, perhaps impossible, way of gaining access to the flat. The risk of being seen was also reasonably high. Although it was dark, the point of entry is on the part of the building most visible to passing pedestrians and traffic.

So, we seem to have a case which fits well with these recommended indicators.

This book provides a useful guide to the types of inconsistencies that might be looked for.

Point of entry

The hypothetical point of entry is the green-shuttered window at the top of this photo.

First we consider the point of entry. Specifically, we should be looking for anything that might have made the point of entry impassable (was it barricaded or bolted, for example?); what transfer evidence is apparent (e.g. blood, fingerprints, footprints etc); whether the entry could have been possible in the manner required to commit the crime (this is, we are told, “often the most dispositive feature” in establishing crime scene staging).

Our hypothetical burglar gained entry to the flat through a second-level bedroom window. It was not barricaded or bolted, but the outside shutters were firmly closed – the wood was swollen, so they were wedged in place. They might have been openable from the outside, if with a little difficulty. But this, it has been argued, would have required the burglar to make the ascent twice: once to open the shutters, then back down to the ground to hurl a rock at the window, then finally back up. The first climb would have been speculative – had the shutters been latched on the inside, that would have been the end of the attempt.

The window would also be a strange choice for a burglar on account of its position. It is in full view of the road that passes by the building and, as you can see from the photo, it doesn’t offer an easy climb. To the rear of the building is a balcony, which certainly looks more accessible and is relatively secluded. What can have been in our burglar’s head?

What’s more, no transfer material was found. No dirt or bits of leaves on the wall, no footprints on the bedroom carpet, no nothing. The burglar managed, so it seems, to perform the break-in without leaving a trace of himself.

The trial court was convinced that the window could not be considered a realistic target for a burglar. But, all the same, might the appeal court not see some room for doubt? It looks to all the world like an unlikely point of entry. But, surely, you can’t send two people to prison if there is even a slim chance that the break-in might not have been staged.

Man climbing up wall

Does this photo help the defence case or the prosecution case?

What the textbooks recommend is to perform a reconstruction experiment in order to give a definitive answer to the question. In other words, you get someone (maybe these guys) to attempt the climb. That’s something that investigators never did. Perhaps they felt that the impossibility of the exercise was obvious, or perhaps they didn’t want to risk a success. The defence, though, do seem to have conducted some sort of experiment, as shown by the photo to the right. But why were the results of this never discussed in court? Does the photo show someone succeeding in accessing the window, or failing?

The next relevant question, working through the list, is whether a weapon has been removed from the crime scene and, if so, why? At least one knife must have been used in the murder, but it was not found with the body. How consistent would this be with the actions of a burglar? I don’t think we can be entirely sure that it is inconsistent (after all, for all we know the knife may have been monogrammed), but maybe you have a different view.

Then there is a question of whether the body was moved after the crime was committed, or whether it was dressed or undressed to any degree. Kercher’s bra was found on the floor near the body. The cups were stained with blood on the front side, suggesting that she was wearing it at the time she was stabbed. The bra also bore tiny spots of blood, similar to spots found on Kercher’s chest, suggesting that she exhaled blood onto herself while she was still wearing the bra.

The body was found covered by a duvet, but blood from the body does not appear to have transferred to any great extent to the fabric of the duvet. This suggests that the blood had substantially dried before the body was covered. Our burglar must have hung around quite a while.

The defence argument is that Kercher must have been sexually assaulted after she was stabbed. That would make the motive for the burglary an especially deranged sex attack. Such things are not unheard of, but why would such a burglar waste time rifling through clothes in the neighbouring bedroom (which is what appears to have taken place)? And how does it explain the duvet?

It is also necessary to consider the ransacking. Our burglar seems to have rifled through clothes in the room of one of the occupants of the flat, paid some interest in her laptop, but not actually taken anything of value. Why was he interested in the clothes? What was he hoping to find? Why did he not, instead, go quickly from room to room picking up items of obvious value? Does the existence of an apparent ransacking fit easily with the defence supposition of a particularly unusual sex attack? If this was the motive for the burglary, why did the burglar not make straight for his intended victim? Or, if he decided to look for valuables after the attack, why did he take nothing?

Some glass on a window ledge

The glass on the window ledge

What would seem to be the most convincing evidence of a staging is, however, none of the above. It is all to do with glass. Broken glass was found on the window ledge in a pattern that was consistent with the shutters having been closed when the window was broken. That is, glass was found distributed on the ledge, but no further forward that would have been permitted by the shutters in their closed position.

Both the occupant of the room that had been “burgled” and one of the first police officers on the scene testified in court that they had found shards of glass from the window on top of the belongings found strewn across the room. How could this be possible unless the ransacking had occurred before the window was broken.

So there’s quite a pile of evidence to suggest that the crime scene was staged. How much of it can be explained away? I don’t know. But the difficulty for the defence is the logic of the methodology. If there’s one detail that can’t be reconciled with a genuine burglary, then there must have been a staging. That could be the glass on top of the ransacked possessions. If the appeal judges can’t think of a reason to doubt the witness testimony (I can’t, although that’s not necessarily a guarantee that they won’t), then there must have been a staging. And if there was a staging, Amanda Knox must have been involved in it.