The police service finds itself completing a full circle as far as the examination of forensic evidence is concerned. Initially forces developed their own expertise or approached experts on a case-by-case basis. This led the Home Office to set up the Forensic Science Service (FSS) which dealt with the needs of the police and other government agencies. In the 1990s the FSS was changed from a government department to a government-owned company and began to charge police forces for every examination that it carried out. This has led to forces developing their own forensic science capability in order to be able to screen exhibits so as to reduce the number which are sent for examination or to be able to carry out the examination themselves so that they can obtain the evidence without any additional cost. Anything complicated will be sent to the FSS, but forces have a growing capability for examining footprints, shoeprints and marks left by tools etc. at the scenes of crime.
Locard’s Exchange Principle
Wherever he steps, whatever he touches, whatever he leaves, even unconsciously, will serve as a silent witness against him. Not only his fingerprints or his footprints, but his hair, the fibres from his clothes, the glass he breaks, the tool mark he leaves, the paint he scratches, the blood or semen he deposits or collects. All of these and more, bear mute witness against him. This is evidence that does not forget. It is not confused by the excitement of the moment. It is not absent because human witnesses are. It is factual evidence. Physical evidence cannot be wrong, it cannot perjure itself, it cannot be wholly absent. Only human failure to find it, study and understand it, can diminish its value.
So said Professor Edmond Locard (1877-1966) whose Exchange Principle is the basis of forensic evidence. He created the first crime laboratory in Lyons in 1910 and was the father of forensic science.
Forensic literally means ‘used in court’. Forensic evidence has come to mean the physical evidence recovered from a crime scene, a victim or a perpetrator, which links the latter to the crime in some way. Its value lies in the fact that it stands by itself and is objective, so the more interpretation is needed the less useful it is. As we will see this is an important point when considering DNA and blood, and it can now be an issue with finger-prints. Forensic evidence’s objectivity means that it is essential that its integrity is sustained. This integrity applies to two elements: the scene and the exhibit. In a murder it is assumed that the last person at the scene was a killer. It is essential that the scene is isolated to minimize the probability that anyone can contaminate it. As far as the exhibit is concerned the prosecution must be able to prove two things: that the piece of evidence which they produced as an exhibit is what they say it is – human hair not animal fur, fibres that match the clothing of the victim, etc.; and that nothing has happened to it between its recovery and its examination which could cause it to be changed or contaminated in any way. As we saw in Chapter 1, to do this they must establish a clear chain of evidence which links the movement of the exhibit fundamentally undermine the integrity of the exhibit.
Forensic evidence is only useful if it either links a suspect to the crime in a way that indicates guilt or undermines their credibility. For example, a fingerprint found at the scene of the crime is of no value if the suspect had, and if they could not lawfully have been there or if they deny ever having been there. Similarly, with any object found at the scene, the prosecution must be able to show either that it could only have been left there by the perpetrator or that it is highly improbable that anyone else could have left it there. This means that fibres from a suspect’s jacket which are found on the clothing of the victim have much more evidential value than those found snagged on a nearby bush. If that bush is on or near a footpath their value is even more diminished. It will increase if the jacket fibre is an unusual one and if the suspect denies ever being anywhere near the scene. The relative value of exhibits is shown graphically in this illustration, known as a Johari Window.
[Figure 2: Relative Value of Forensic Material]
Fibres and other material from the victim found on the suspect have much more weight and value than the other way around, as common material could have been picked up by the victim anywhere, at any time. If the material is found either on the suspect or in some unusual spot, such as the boot of their car, it has more value.
Forensic evidence is best dealt with in three classifications: exchange materials, fingerprints and body fluids which provide DNA.
It is impossible to list all the possible external exchanges that can occur, but those described below are the most common.
These are the most common exchange materials in both fact and fiction for the obvious reason that most murders in Britain require contact between the victim and the killer, as the use of firearms is still uncommon. In non-domestic murders there is usually some attempt to hide the body and this again requires some form of contact. Once contact has occurred the only certain way of completely destroying the evidence is the destruction of any material involved – all clothing and any carpeting or lining in a car used to transport the victim, living or dead. Even the most thorough cleaning cannot be relied on to remove all contact material and science is continually developing ways of finding trace materials, such as the use of powders which will attach to contaminants and can then be found by using ultraviolet light to fluoresce them.
Any recovered fibres can be linked in two ways. First they can be analysed and found to be the same material; obviously the rarer the material the more weight it has as evidence. A strand of wool from a Marks and Spencer’s cardigan will not have quite the same impact as one from a cashmere sweater sold only through Harrods. Secondly, if the material was cut or broken off in some way it may be possible to make a physical match of the cut or broken ends.
This is the next most common exchange product. Human beings cast hair all the time and there will always be some sort of exchange in a violent attack. Hair can be matched in the same way as fibres. Analysis can determine whether or not it is human hair or animal fur. If it was cut then a physical match may be possible. DNA analysis is only possible, however, if the root is still attached.
Because it is so fine and difficult to see, hair presents the killer with the same problems as fibre. Rapists have shown an increasing awareness of this in that a significant number of them now meticulously comb and clean the pubic area of the victim, adding significantly to the horror of the attack.
This is a growing field for the police, helped by the fact that most young men (who commit most crimes) now wear trainers rather than shoes, and trainer manufacturers continually change their sole patterns as part of their strategy to make their shoes distinctive. The FSS are slowly building up a database, as are a number of forces. Most forces now routinely take the footprints of suspects whom they believe to be active criminals, especially suspected burglars.
The usefulness of a footprint is determined by:
- the rarity of the sole pattern
- the size and clarity of the recovered footprint (it can be very clear if it has moved through fluid and onto a flat surface such as a kitchen floor)
- the age of the shoe – the older the shoe the more likely it is that it has acquired unique marks
- the time between the offence and the arrest of the suspect – the shorter this is the more likely it is that the patterns of damage or wear will match.
In Bedfordshire in 1996 a murder occurred where a footprint was the critical piece of evidence. The victim and her husband were renovating their large, isolated house and during the work were living in a caravan parked in the garden. The woman was attacked when on her own one afternoon and after the assault the killer left her body in the caravan together with a load of straw which he then set fire to. All the caravan windows were closed. When he closed the door on leaving, he deprived the fire of oxygen, causing it to smoulder rather than burn. Some time later a passer-by noticed smoke coming from a caravan and went to investigate. When he opened the door the fire immediately took hold. He managed to drag the woman’s body out of the caravan but the van itself was quickly devoured by the fire. This meant that the only source of forensic evidence was what was left of the victim’s body and clothing.
The victim had been killed by being stabbed in the throat with a screwdriver. During the assault the killer had at some point stamped on the victim’s face several times with considerable force, leaving it marked. The SIO, Detective Superintendent Dave Tomlinson, succeeded in finding a pathologist who could recover the mark as a useable print and it was sent to the FSS but no match was made. When it was returned to Bedford a sharp-eyed SOCO thought it looked similar to one recently taken from an active burglar and the match was made.
Like trainers most makes of tyre have distinctive tread patterns which can be helpful in confirming identification once a suspect is brought into the frame. They are usually only useful in helping to find a suspect in the first place if they are very unusual, thus limiting the number of people that the inquiry team need to eliminate. Once a suspect is arrested they can provide strong circumstantial evidence.
Tools and Weapons
The exchange that takes place here is not materials (other than blood, flesh and hair with weapons) but marks. When tools or weapons are used they will leave the mark of any peculiar wear or damage they have on the object or person they were used on. These marks and those on the tools or weapons can then be matched. The more usual the wear or damage pattern the easier it is to make a match which a forensic scientist can say is exact.
The usual source of paint in inquiries is cars, usually involved in hit-and-run cases. It can be used in two ways:
- the analysis can identify the make and model of the car
- samples recovered can be physically matched to scratches or patches on the vehicle itself.
Paint is usually used as part of a screening process; the investigator tries to find everyone who owns the make and model identified and then eliminate them from the inquiry. This is an expensive process and will only be used in a very serious crime, usually murder. In difficult inquiries where there is little to go on this has often created enormous workloads for the inquiry teams. In the Yorkshire Ripper case 5.4 million car numbers were taken and checked; in the case of Robert Black (see Chapter 4) a blue Cortina came into the frame and as a result 20,000 owners of blue Cortinas were traced and interviewed. It can lead to spectacular success as in the case of ‘The Fox’, Malcolm Fairley. In the summer of 1984 he carried out a series of rapes on the Bedfordshire / Buckinghamshire / Berkshire borders. Over 200 police officers from different forces were deployed on patrol to catch or at least deter him, without success in either. Eventually tiny fragments of ‘harvest yellow’ paint from a British Leyland car were found where the rapist had reversed his car into some bushes. When the detectives carrying out the elimination inquires came to see him, Fairley was actually washing his car. they noticed the scratches and when they searched his other car they found more evidence connecting him with the rapes.
There will be an exchange of materials in any contact. Science now helps us trace and then analyse microscopic amounts of material. As the sample gets smaller it is important to bear two issues in mind. First, the smaller the sample the more scrupulous the SOCO must be in preventing contamination. The rejection of the DNA evidence by the judge in the Omagh bombing case powerfully demonstrates what happens when you get this wrong. This means that the scene of a murder will only be visited by SOCOs in the first instance. Some forces now have a special camera that can take a 360-degree photograph of the scene so that the SIO need never actually enter it. Since it can take two or three days before SOCOs can allow access, this has become more and more important. The second issue is that analysis of the sample requires its physical destruction so the smaller sample the more confident the analyst must be of getting a result as there may not be enough for a second attempt.
Fingerprints were first used for identification in Persia and India in the fourteenth and fifteenth centuries, usually in place of a signature. They became the object of scientific study and development in the late nineteenth century by a number of academics, notably Dr Henley Faulds, Francis Galton and Sir William Herschel. Herschel was the first to use fingerprints in a formal and systematic way to identify criminals. Faulds was the first to propose that they be used to identify and eliminate suspects. The first fingerprint bureau was set up in Calcutta under Sir Edward Richard Henry, who is credited with the development of the classification system used in most English-speaking countries, although there is no doubt that much of the credit should also go two the two Indian fingerprint experts, Azizul Hacque and Hem Chandra Bose, who worked with him at the time. Scotland Yard set up the first fingerprint bureau in Britain in 1901, using the Henry classification system.
By the 1980s the fingerprint database had become unmanageable in terms of searching marks (prints found at the scene of a crime). The volume of material was such that manual systems could not cope. A mark would only be searched against the general database in very serious cases, where there seemed no other way are of taking the investigation forward. Those marks found at lower-level crime scenes such as burglaries were only searched against a force’s target criminals or anyone that the investigator could put up as a suspect. At the time the service was confident that the identity of criminals was not a problem, but since automation a number of criminals have been found to have used more than one identity. In the early 1990s the process was computerized and the service got its first automatic fingerprint recognition (AFR) system. This meant that for the first time in decades the police were able to do what everyone thought they had always been able to do, i.e. check marks against the whole database. This development has played a significant part in improving police detection rates.
The palmar surfaces of the hand (and feet) are covered with ridges which are formed randomly in the foetus, developing unique patterns which do not change over a lifetime. No two fingerprints are the same, not even those of identical twins. The valleys between the ridges secrete on to all the surfaces with which they come into contact, leaving the ridge pattern. The secretions are a mixture of water, inorganic salts and organic material such as urea and amino acids. For a long time the prints on the surface at crime scenes, called latent prints or marks, could only be recovered in very limited circumstances, such as if they were recently deposited on a clean, nonporous surface. Now they can be recovered from most smooth surfaces using a variety of approaches ranging from the standard fingerprint powders, through reagents which can be fluoresced by an ultraviolet light and then photographed, to smoking superglue over the surface. It is normally necessary to remove the item or the part containing the fingerprint to a laboratory to do this.
The standard approach by most English forces is to use aluminium powder or flake, applying it with a fine bristle brush so as to minimize the danger of rubbing and thus damaging the print. This is adequate for most scenes of crime. The grey powder is visible against most other surfaces. Other colours are available if this is seen to be a problem. Ideally the SOCO wants the powder to adhere to the latent prints but not to the surface, so as to show a clear powder but it may be too sensitive and, by clinging to the surface, cover (or paint) the latent (print) so as to obscure it. Which to use is a matter of professional judgement and SOCOs use their years of experience to decide where they are most likely to find prints and what techniques could and should be used to develop those that are there.
The only way to guarantee that the print is not disturbed is to ensure that it is examined in situ, only moving it as much as the examination itself requires. It is not safe to pick it up, no matter how it is done. Let us take a gun, for example. Picking it up with a handkerchief will ensure that you do not leave any prints on it, but it will almost certainly guarantee at the same time that any prints that are on it are at best smudged and damaged, and at worst completed rubbed off. The next favourite method on TV and film is to put a pen or pencil through the finger guard. This is great, unless of course there is a partial print on the trigger – not at all unlikely since that is where the finger is supposed to go. Finally, there is the pen put into the barrel. Again this is great as far as the outside of the weapon is concerned but if it was fired close enough to the victim the inside of the barrel may have minute traces of the victim’s blood or clothing sticking to it, which would be invaluable in proving that it was the murder weapon; the pen will knock them off, so that they may not be recovered. The old rule for a crime scene of keeping your hands in your pockets at all times still holds good. Don’t touch anything until the SOCO tells you he or she is finished with it.
The features of the fingerprint are called characteristics. They consist of three basic ridge patterns: the arch, the loop and the whorl. When these are present in a latent they give the fingerprint expert a good starter. The ridges themselves are made up of three basic minutiae: ridge endings, bifurcations and short ridges (or dots). These last are much shorter than the average ridge and are often contained by adjoining ridges, like a little island.
[Figure 3: Finger print Characteristics]
Until quite recently a fingerprint match could only be made if an expert could match sixteen characteristics in the mark if an expert could match sixteen characteristics in the mark found at the scene with the suspect’s fingerprint. This was the highest number required by any agency worldwide; many were content with nine, and eleven was the standard in the USA. The UK has now moved to a qualitative rather than quantitative measure. This has been helped by the fact that there is now an approved certification of expert status in the individual must have at least five years’ experience and be qualified in the Professional Investigation Process. An expert can now make a positive identification based on fewer than sixteen characteristics if he or she can say that, taking the print as a whole, they are the same. This does introduce that element of interpretation already discussed which can undermine the objectivity of the evidence but it also allows for marginal identifications, for example with twelve characteristics, which would have previously been arbitrarily excluded regardless of the certainty of the expert that they were the same.
From a crime writer’s point of view it should be borne in mind that the fewer characteristics the more the finding is a matter of opinion and not fact. This means that mistakes can be made. A very good recent example of this involved a DS in Scotland in 1997. DC McKie, a successful police officer in the Strathclyde Police, was involved in a murder inquiry in Kilmarnock. Her thumb print was allegedly found by the Scottish Crime Record Office (SCRO) experts in the kitchen of the murder house. She denied ever having been in the house. Since she was not involved in that element of the inquiry it would have been unprofessional of her to visit the scene at all, never mind being careless enough to leave a print. In the face of the denial she was arrested and charged with perjury. It took an enormous effort on her part and her supporters to have the Scottish Office acknowledge the error and agree to compensation. The case exemplifies the difficulty that can occur when expert evidence is taken as face and not opinion.
Difficulties in Fingerprint Identification
The major difficulty is that the SOCO rarely gets a complete fingerprint at a scene. It is usually only a part of the print (called a partial) and it may contain the least useful elements in terms of identification. In addition, the expert has the following issues to deal with:
- since skin is flexible no two copies of the same fingerprint will be exactly the same; there will always be some differences, no matter how slight
- it is not always easy to tell which finger the mark comes from, although its position on the object, the nature of the object itself, etc. will probably give some clues
- on partials the orientation may not be clear – which bits are the top, the bottom or the side. This is critical in working out the relative position of the characteristics
- the original police record against which the match is being made may not be a good one. The clarity of this print is critical. When they were taken using printer’s ink on paper the quality was very variable and since the prisoner had inevitably been released before the prints reached the department it was difficult to get them redone. Today most, if not all, forces have Livescan, which allows an optical reader to capture the print so that its quality can be assessed immediately
- the suspect may have accumulated scars since the last time his or her fingerprints were taken. The scars will not be shown on the previous prints and they may cut through or distort key characteristics.
The police can take the fingerprints of anyone arrested for an offence. If the person is subsequently acquitted the prints must be destroyed, even if the suspect has a long record and the most recent set of prints is better than those currently on record. The law for retaining prints was set in more liberal times (despite the absence of the Human Rights Act), when Parliament believed that the privacy of the subject had some meaning.
When investigating offences police will often ask for ‘elimination prints’. These are most often sought when it is necessary to eliminate all the people who had lawful access to the scene. They are also taken where there are a large number of possible suspects and a mark has been found at the scene. It is important to remember that the fact that the mark is not proven to be the suspect’s does not eliminate him or her. Until the owner of the mark is identified it is impossible to know its worth or relevance and there may be other strong reasons for retaining the individual as a suspect. Elimination prints must be destroyed at the end of the inquiry and cannot be entered into the national database.
Taking fingerprints used to be a bit of a nightmare. A block had to be prepared, usually a long brass plate mounted on a wooden base. This was done by rolling out a thin layer of printer’s ink. The prisoner then had each digit rolled in the ink and onto a fingerprint form. If the ink had been applied too liberally all you got was a black smudge; if the digit was rolled too lightly, or the subject had not been made to clean his or her hands first, part of the print would be missing or unreadable. Both thumbs were then taken simultaneously then a set of all four fingers of each hand. This was essential as the officer taking the prints may have put them on the form in the wrong order, or got the hands mixed up, something that happens surprisingly often.
Today fingerprints are all taken using optical readers and can be checked immediately against the national database to confirm identity, but the process is the same. The prints are also run against the force and national database of marks recovered from scenes of crime. Despite the difficulties, this makes a large number of hits. Since the whole process is now computerized it is now possible to take a laptop into the field loaded with a force’s database so that checks can be done in the course of an operation with a suspect still present. If a secure encrypted link can be provided the laptop can be connected to the national database.
With the arrival of DNA, fingerprints have tended to slip into the background and their value has become less and less appreciated, with the spotlight being placed on their more glamorous competitor. However fingerprints are likely to continue as a mainstay of forensic detection for the following reasons:
- they have a track record of over 100 years in which no two fingerprints have ever been found to be identical and they are the basis for criminal identification in every police agency
- since every police agency uses them they are a worldwide tool and allow the certain exchange of information between agencies
- not only is there a vast existing database but it continues to expand daily
- in light of the current terrorist threat it is the only certain way of identifying suspects if their origins are in developing countries
- forensic practitioners continue to develop methods for recovering prints where it was previously not possible – e.g. from materials damaged in a fire by the use of a play-dough-like material which removes the soot without damaging the print, or from metallic surfaces using a Kelvin probe
- it’s cheap. In the UK every DNA analysis costs money whereas fingerprints have no additional costs
- even where the identity is not certain enough to support a conviction it can be very useful intelligence.
Dr Alex Jeffries of Leicester University was the first in the scientific community to notice that people’s DNA has different patterns which can be used in identification. It was first used for forensic detection in 1987 to identify the rapist and murderer Colin Pitchfork. In 1983 a 15-year-old girl had been found raped and murdered on a lonely footpath outside the town of Narborough. Semen was recovered from the body but the killer was not identified and the case was left open. In 1986 another 15-year-old girl was found raped and murdered on another lonely footpath just outside the town. Both had been strangled and police believed that both were killed by the same man, partly due to the modus operandi and partly because semen samples found on both bodies came from a man with type A blood with an enzyme profile that was only shared by 10 per cent of the population. A 17-year-old youth, Richard Buckley, who had discovered the second body, was arrested for that killing. He confessed to the second murder but denied any involvement in the first.
Dr Jeffries offered to assist the investigating officer by testing the semen for its DNA profile. Everyone was surprised when it showed that both girls were raped by the same man but that that man wasn’t Buckley. Five thousand local men were asked to give blood or saliva samples but no match was found. Later a man named Ian Kelly was heard bragging in a pub that he had stood in for his friend Colin Pitchfork. When Pitchfork was arrested and tested his sample was found to match the killer. DNA analysis very quickly became a major investigative tool.
DNA, deoxyribonucleic acid, is often described as the blueprint for the body’s structure, but it is much more than that. A blueprint is essentially passive. It is read by somebody who then puts it into action. DNA not only has the pattern but it also issues instructions on how a cell will be built and when. It is a 1-foot-long strand in the form of a double helix. It fits into a cube one-millionth of an inch on each side and is present in nearly every body cell – not, strangely enough, in red blood cells, but only in white ones.
Every strand is made up of four bases, usually referred to by their initials of A, T, G and C, shorthand for their proper scientific names. Strings in DNA are described in terms of these letters, e.g. TTTGAAAACTTTAAATGATGA. Some DNA sequence encode information for the cell (coding DNA) and some don’t appear to code anything that we know about at the present (non-coding or junk DNA). DNA is passed on through the sperm and egg cells, each containing about 3 billion bases that follow a well-defined sequence. The sequence of bases, coding and non-coding, varies from person to person and can be used to distinguish one person from another. The standard approach used by the FSS produces a result which enables their experts to state that the chances of getting two the same are one billion to one.
Reduced to its basics, DNA analysis is very straightforward and now follows a thoroughly tested, reliable process. Every human’s DNA, taken overall, is very similar. What makes it different is the pattern of bases in any given position on the DNA strand. If enough of these locations are tested it is possible to declare that their match is extremely unlikely to have happened by chance. The analysis is done as follows:
- Targeted locations on the DNA are separated and amplified – caused to reproduce themselves.
- The resulting fragments are separated and detected using electropherosis. The fragments have a slight negative charge and they are put into a gel which has a slight positive charge at one end. This causes the fragments to move towards the positive charge. The fragments are all of different lengths and the shorter ones move more quickly through the gel, causing the sample to stretch out in separated bands.
- Fluorescent dyes which had been used in the targeting process then produce an image showing the bands clearly.
- These bands are then compared with the standardized ladder and with the sample recovered from the crime scene.
The larger and fresher the sample the less danger there is of contamination and deterioration, thus more sequences can be tested, producing a more reliable result.
[Figure 4: DNA Testing (LCN)]
Low Copy Number DNA Testing (LCN)
The initial approach to DNA analysis required quite large samples and took some time to produce results (2-4 weeks). A quicker and more sensitive process, which could work with small sample sizes, was soon developed. This is a combination of focusing the analysis on a limited number of locations on the DNA, technically called short tandem repeats (STR), and amplifying the sequence (causing them to reproduce themselves) through a process called polymerase chain reaction (PCR).
This approach can work on very small samples and can be carried out quite quickly – between forty-eight hours and a week depending on how much you are willing to pay. The method was further refined and developed so that minute samples of five to ten cells could be analysed successfully. Samples of this size can be obtained from a cup that has been drunk from a pen used to write with. From an investigator’s viewpoint it looks too good to be true – and in some respects it is.
The process causes the sequence of DNA selected to reproduce over and over until there are enough copies to carry out an analysis. It is an exponential process. The original sequence is not copied again and again like a photostat, it produces copies and those copies are copied and so on. This has the advantage that a sample large enough for analysis is produced quite quickly but it also means that any rouge DNA is also copied and since it may have been present in the same strength as the targeted DNA it has the ability to produce false and thus misleading results.
Another problem is that in specimens that have been degraded, there is a possibility that some of the sequences that have been targeted will be missing, or that some of the bases will be missing from the targeted sequence. If this is so the resulting profile will only show a partial picture. To paraphrase Donald Rumsfeld, ‘We don’t know what we don’t know’, so there is always a danger that a completed profile will be different in some way from the original.
Both of these factors are important as they significantly affect the probability of error. The experts in this field give very large odds against the sequence occurring more than once, which appears to give the process a great deal of certainty. The problem is that the samples collected may not be truly random. For example if the murder occurs in a small, remote village, many of the inhabitants may be related, two brothers will have very similar profiles and cousins may also have a higher than average similarity of profile. This means that it is necessary not only to calculate the odds of a sample randomly reappearing but to calculate the odds of one reappearing in the context of the investigation.
Where the crime scene is undisturbed the SOCOs can be reasonable certain that the minute samples they recover are uncontaminated by any subsequent contact. They cannot tell whether the same cup was drunk from by two people or whether two people used the same pen (unless it was a man and woman, when the sex chromosome may be present and may assist). Practices have been developed for recovering samples so that they will be safely transferred from the scene to the laboratory. At the laboratory there are a number of processes to ensure that any contamination is identified and, according to the CPS, ‘the statistical interpretation of the results allows for the possibility that some of the DNA may be due to contamination or other effects caused by working with such low-level samples’.
All this means that the profiles produced by LCN from specimens currently recovered from crime scenes are probably reliable. The same cannot be said for those recovered in the past where cross contamination was not a consideration when they were being recovered or handled. In these cases the safest way to approach low copy number analysis is to see it as being persuasive rather than conclusive evidence, i.e. safer and of more weight if there is other evidence to corroborate it.
I have tried to discuss this issue in a non-technical way. If you want to take it further I recommend the following websites:
The language in this entry is quite technical but it does provide access to outside links and is an easy way of keeping up to date. It also describes the difficulties in statistical interpretation that can produce the prosecutor’s and the defender’s fallacies.
This is a scientific description of the process for non-scientists which worked for me.
PBS (Public Broadcasting Service – USA)
This is aimed at older children but the ‘hands on’ feel of the virtual lab helped me understand how the process worked.