'I marvel how Nature could ever find space

For so many strange contrasts in one human face.'

William Wordsworth, 'A Character' (1800)

Never mind fingerprints or DNA. What make us recognisably individual to eachother are, of course, our faces. Nature, nurture and circumstances combineuniquely in each of us to create a set of features that is the key toidentification for everyone who knows us. At one time or another, we've allbeen misled by the similar body shape or gait or hair of a stranger; but whenthey turn or come close enough for us to see their face, we know our error atonce. But death steals our faces from us. Our flesh decomposes, nature stripsus back to the bone and the skull beneath the skin means nothing to the peoplewho knew and loved us.

Thankfully, there is a small band of scientists whose work is dedicated togiving the dead their faces back. In the UK, Richard Neave established thetechnique of facial reconstruction from skeletal remains at ManchesterUniversity. He was part of a team put together in 1970 to investigate theEgyptian mummies housed at the Manchester Museum and in 1973, using plaster andclay, he rebuilt the faces of two 4,000-year-old Egyptians, Khnum-Nakht andNekht-Ankh, known as 'The Two Brothers'. 'Right from the beginning,' wroteNeave, 'I endeavoured not to rely merely upon intuition – what was irritatinglyreferred to as "artistic licence".' Instead he determined the shape of thefaces using average tissue thickness measurements taken from a collection ofcadavers in 1898 by the Swiss anatomist Julius Kollmann.

Neave developed great skill in modelling the muscles of the face and skull,which provided a latticework for the rest of the flesh and skin to sit on.Having refined his skills in the archaeological sphere, he turned to forensicwork and was involved in more than twenty cases of unidentified remains, with a75 per cent identification success rate.

One of his most challenging cases began, paradoxically, with a headless corpse.The body of a man wearing nothing but a pair of underpants was discovered in1993 under the railway arches of Manchester's Piccadilly Station. In spite ofthe best efforts of the police, his identity remained a mystery.

Three months later, a man was walking his dog across a playing field inCannock, Staffordshire, seventy-five miles from Manchester. Suddenly the dogstarted digging and continued frantically until he uncovered a severed head. Ithad been smashed into more than a hundred pieces; later, it emerged it had beenmangled by a machete. DNA tests connected it to the headless torso inManchester, but that still didn't take the police any closer to an ID. And atfirst it seemed unlikely that the face could be rebuilt. A significant amountof bone was missing, especially from the crucial middle section of the skull.Police presumed the murderer had intended to make it impossible for anyone torecognise the victim of this vicious attack. But painstakingly Richard Neaveglued together what remained of the skull and cast it in plaster, filling thegaps to the best of his ability and with the benefit of his extensive knowledgeand experience. When the Independent newspaper published a photo of Neave'sclay head, seventy-six families came forward thinking that they recognised theface.

The police collected photographs and details from those families and begancomparing the faces of their missing relatives to the skull. As they workedtheir way down the list without success, it began to look as if the murdererhad succeeded. Finally, they reached the last name. Adnan AlSane had been givensuch a low priority because there had been nothing about the body or skull tosuggest that the victim wasn't Caucasian. But the details matched. At last, thepolice knew who the victim was.

Adnan Al-Sane was a 46-year-old Kuwaiti businessman who had been living inMaida Vale, west London. He came from a wealthy family and had made a fortunerunning a bank in his native country, before retiring at only thirty-eight. Hehad last been sighted, the day before the decapitated body was found, havingdinner at the Britannia Hotel in Grosvenor Square, central London. Dentalrecords and fingerprints from Al-Sane's flat confirmed his identity. The postmortem showed that he had swallowed a tooth during the attack that killed him,but his head had been hacked off after he was dead. To this day, his murderremains unsolved, the motive a mystery. But at least his family knows his fate.

Richard Neave helped to demonstrate the scientific basis for facialreconstruction, shaking off the notion that it was more an art than a rigorousscientific discipline. He spent his career working and lecturing at ManchesterUniversity, where he passed on his knowledge to the next generation, among themCaroline Wilkinson, now Professor of Craniofacial Reconstruction at theUniversity of Dundee.

One of Caroline's own landmark cases started almost as improbably as theAl-Sane case. One August day in 2001 a sunbather came across part of a girl'sbody on a beach at Lake Nulde in the Netherlands. Over the next few days otherbody parts were found at different locations along the Dutch coast. Then afisherman discovered a skull near a wharf, eighty miles from Nulde. The facehad been mutilated beyond recognition. Investigators were baffled. Theycontacted Caroline, hopeful that she would agree to remake the face.

But when Dutch police told her they estimated the victim was aged between fiveand seven years old, she found herself uneasy about taking the case. Part ofher reluctance stemmed from the fact that her own daughter was also only fiveat the time. But much more significant than her own emotional response wasprofessional caution.

Back then, anatomists doubted whether it was possible to reconstruct children'sfaces with anything like the same accuracy as those of adults, because juvenilefaces are undeveloped and lack clarity. But Wilkinson had specialised injuvenile facial reconstruction while working towards her PhD. She believed shecould bring something useful to the investigation. She locked away her qualmsand examined the damaged skull the Dutch police had sent over. As she studiedthe bones, Caroline realised the dead child had some unusual features: a large,wide nose – unlike the little upturned noses that most 5-year-olds have – and abig gap between her front teeth. Already she could see that this was adistinctive face.

In general, it's more uncommon for missing children to be recognised fromphotos than adults, despite the greater media coverage that they get, becausetheir unformed faces are more similar to each other. Only one in six missing childrenis found because someone calls the authorities after seeing that child'spicture, according to the National Center for Missing and Exploited Children,who release thousands of images of missing children every week in the US.

But Caroline was hopeful that this girl would be one of the identified ones.She applied all her skill to making a clay model of the Nulde girl's face.Photographs of the result were widely displayed in newspapers and on televisionaround Europe. Within a week the girl had been identified as Rowena Rikkersfrom Dordrecht, aged five and a half.

In the wake of the identification, a horrific story emerged. During the lastfive months of her short and tragic life Rowena had been physically abused byher mother's boyfriend, with her mother's knowledge. She had spent the last twomonths of her life locked up in a dog cage. After her death, her body had beencut up and scattered throughout the Netherlands by the two people above all whoshould have cared for and protected her. They were eventually tracked down toSpain and convicted of their crimes. It was the first time a facialreconstruction had been used to solve a crime in the Netherlands – and withoutCaroline's work, Rowena's death may never have been acknowledged or avenged.

The idea of rebuilding faces is not new, and nor is it always about murder. Itsprung from a desire to connect with lost people by visualising them, andpeople have been doing it for a very long time. In 1953 archaeologist KathleenKenyon discovered skulls in Jericho from around 7000 BC which had claycarefully worked on to them, with shells set into their sockets to imitateeyes. She was struck by their beauty: 'Each head has a most individualcharacter, and one cannot escape the impression that one is looking at realportraits.' The ancient Middle Eastern artists had used clay to model thephysical distillation of their ancestors' identities – their faces – so thatthey might conquer death.

The face has always been imbued with significance. The eighteenthcentury artistWilliam Hogarth called the face 'the index of the mind'. And there's no denyingthat faces betray our emotions and responses – they laugh, cry, scare, soothe,entertain. The tiniest movement of our facial muscles can reveal aggression, oraffection: you only have to think of the subtle difference between a confusedfrown and an angry frown to realise this. Our brains are highly skilled inrecognising minute differences in other people's faces and we can identifyhundreds of them as a result. At only five weeks old, babies can distinguishtheir mothers' faces. And 2.5 per cent of people grow up to be 'superrecognisers', capable of identifying nearly every face they have ever seen. Wecan read in a face certain key elements of our humanity – gender, age, generalhealth, for example. But just because you can see someone's face, it doesn'tmake you a mind-reader; as Shakespeare pointed out, 'there's no art/To find themind's construction in the face'. One thing we definitely can't tell from a faceis whether someone is 'constructed' like a criminal.

A collection of 'criminal faces' compiled by Cesare Lombroso: this plate showsmurderers. Lombroso believed criminality could be predicted by the physiologyof an individual

The nineteenth-century criminologist Cesare Lombroso thought he knew better,however. Lombroso measured the faces of 383 lawbreakers and published a book,L'Uomo Delinquente (Criminal Man), in 1878 which ascribed to criminals'enormous jaws, high cheekbones, prominent brow ridges, solitary lines inpalms, extreme size of eye sockets and handleshaped ears'. Later studies ofLombroso's own measurements have shown his conclusions to be nonsense. Theevidence didn't back up the theory; it was based only on Lombroso's ownprejudices and unfounded opinions.

But 'Lombrosia', as it came to be called, was a seductive concept, and itscreator was often asked to testify at trials, with mixed success. He wasoutraged when one jury ignored his recommendation to convict a man of murderdespite the absence of hard evidence. Although Lombroso had identified 'aphysiognomy approaching the criminal type in every way', including 'outstandingears, premature wrinkles and [a] sinister look', all of which should have beenenough to convict him 'in a country less tender towards criminals', the jurywas not convinced. He was also criticised by some contemporary scientists, but,despite these setbacks, his ideas were nevertheless influential. Peoplelistened to him because they instinctively seek meaning in faces.

Lombroso went about it in completely the wrong way. But, in a way, he was onthe right track. In order to solve crimes and unearth the secrets of the past,scientists and investigators do need to pay very careful attention to humanphysiology. In Caroline Wilkinson's view, 'Any facial reconstruction producedwithout an understanding of facial anatomy and anthropology would be at bestnaive and at worst grossly inaccurate.' Painters and sculptors have long knownthat understanding how the facial muscles attach and move can improve theaccuracy of their work, leading to a profound interest in dissection andanatomy. Leonardo da Vinci dissected thirty unrefrigerated cadavers in hislifetime, overcoming 'the fear of living in the company of these dead men,dismembered and flayed and terrible to behold'. His dissections gave rise to aseries of astonishing anatomical drawings, including of a skull incross-section, that gave Leonardo's later painterly depictions of human faces adeeper realism.

The brilliant seventeenth-century Sicilian sculptor Giulio Zumbo never sawLeonardo's unpublished skull drawings, but managed to improve the understandingof how individual faces relate to their skulls in a different way. Togetherwith a French surgeon, he worked wax on to a real skull, leaving the 'skin'peeled back to reveal the facial muscles. The resulting full-colour model of ahalf-decomposed face, replete with maggots coming out of nostrils, lookeduncannily like a real person.

In the nineteenth century, as we came to a greater understanding of theworkings of the human body, facial reconstruction became more rigorouslyscientific. Early practitioners had lacked established anatomical principles towork within, so they began to generate them. German and Swiss anatomists andsculptors collaborated to interpret the relationship between the face and theskull.

In 1894 in Leipzig, archaeologists exhumed a skeleton that they thoughtbelonged to Johann Sebastian Bach. They asked the anatomist Wilhelm His toprove it. He went about it in an original way, by taking possession oftwenty-four male and four female cadavers, and laying patches of rubber atlandmark points on their faces. He pushed an oiled needle through each rubber –which represented the level of the skin – and down through the face until ithit bone. Then he pulled the needle back out and measured the distance from theneedle tip to the rubber. These were the world's first soft tissue thicknessmeasurements. He averaged out these measurements and then, with the help of asculptor, began building up clay over the skull to match them. The resultingmodel looked remarkably similar to contemporary representations of Bach.

Despite the scientific value of the Bach reconstruction being compromised byWilhelm His's familiarity with contemporary portraits of the composer, hisneedle and rubber technique was of lasting value; the measurements he took haveremained remarkably consistent, and are still used today, although facialreconstructionists think that in recent years faces in the Western world havebecome fatter. In 1899 Kollmann and the sculptor Büchy used the technique torebuild the face of a Neolithic woman who had lived by a lake in Auvernier,Switzerland. The woman is regarded as the first properly scientific facialreconstruction, because Kollmann had based his model on so many soft tissuemeasurements, taken from forty-six male and ninety-nine female cadavers fromthe local area – the same tissue measurements Richard Neave would use in the1970s to reconstruct the faces of the Two Brothers.

As the twentieth century moved forward, so did the techniques of facialreconstruction. Anthropologist Mikhail Gerasimov developed what is now known asthe 'Russian Method', which pays great attention to muscle structure and lessto tissue thickness measurements. He modelled muscles on to the skull one byone, and then covered them with a thin layer of clay to represent skin. Hereconstructed over 200 archaeological faces – including that of Ivan theTerrible – and was involved in 150 forensic cases. In 1950 he founded theLaboratory for Plastic Reconstruction at the USSR Academy of Sciences inMoscow. It still exists, and makes an important contribution to the field.

Developments in medical technology have generated important developments in thefield of facial reconstruction. X-rays and CT scans of living people have beenremarkable sources of data. Until the 1980s all measurements had been takenfrom cadavers, which led inevitably to some inaccuracies. The walls of ourcells start to break down immediately after we die, which causes fluid to drainto the back of our heads and our faces to lose their plumpness. Also, asAmerican facial reconstructionist Betty Gatliff noted, 'When people die, theydon't die sitting up, they die lying down. The soft tissue shifts.'Three-dimensional models of living faces and their skulls were always the holygrail for reconstructionists and CT scanning has provided more widely acceptedthickness measurements. As a result, facial reconstruction is now more accurate– and consequently more trusted – than ever before.

Investigators call forensic artists when they've found a skull that can't beidentified after crime scene clues, missing person files and forensic evidencesuch as DNA and dental records have led them nowhere. If investigators don'tknow who they are looking at, the last best hope is that a member of the publicmight. So it was with Rowena Rikkers and Adnan AlSane. A reconstructed face isa recognition tool, a memory jogger. It's not strictly 'forensic' because thereconstruction itself has no weight in a courtroom. It's only after familieshave contacted the police that the forensic procedure of identification begins.

But why does a face look like it does? How did it develop into this means ofidentification? We tend to think of the face as a social tool, which is whywhen we want to dismiss people disrespectfully, we tell them to 'Talk to thehand 'cos the face ain't listening' or turn our head away. In fact, our facesevolved as they did primarily for utility. Having a pair of eyes at the frontof our head gives us overlapping fields of vision and thus depth perception.Our lips and jaws are perfectly evolved to chew, swallow, breathe and talk.Having one ear on either side of our head helps us to pinpoint the origin of asound. But there are other elements, too. Familial resemblances reinforcedtribal loyalties in early communities, as well as in later dynasties such asthe Hapsburgs, famous for their hereditary malformation of the lower jaw.

The shape of the face depends on the twenty-two bones of the skull. The complexshape of these bones, and to a lesser extent the muscles that are attached tothem, explains the variation between individual faces. Understanding the myriadvariations these bones and muscles can produce is the starting point of facialreconstruction.

To deduce the shape and prominence of somebody's eyes, forensic artists look atthe depth of the eye orbit and the shape of the brow. The shape of the lips andhow they meet is taken from the size and position of the teeth. Ears and nosespresent a challenge, because cartilage decomposes after death. All we can knowabout the ears is where they were and whether they had earlobes; although, inlife, every pair of ears is as unique as a fingerprint. It's hard to know if anose was a button, Roman or tipped up like a pig's. But the 'bony nose' cantell anatomists a surprising amount about the 'soft nose' which sits on top ofit. For example, the pointed piece of bone – the nasal spine – at the bottom ofthe bony nose usually has one point to it. If it has two points that makes thenose split slightly at its tip.

Facial reconstructions based on skulls have to work without the importantdifferentiators of hair and eye colour, at least for now. Geneticists haverecently learnt how to pin down nineteen different eye colours from DNA. Butthis information is expensive to extract – far beyond the budget allocation forreconstruction, even in a murder inquiry. DNA can also reveal hair colour but,even if that had a negligible cost, it would be of limited value to artists.Caroline Wilkinson explains: 'I took photographs of all my students this year.Only two of them have their natural hair colour. I'm forty-eight and I reckon mostof my friends have no idea what my real hair colour is. I'm not sure even Iknow anymore.' So most artists sidestep the problem. They subtly blur out thehair (and the unpredictable ears) of their models. And yet the overall resultscan still be uncanny, often because of the accuracy of the soft tissuethicknesses that CT scanning has provided. The closer a model resembles a realface, the greater the chance of someone recognising their loved one. Theeffectiveness of an acute resemblance was proved in an extraordinary case inEdinburgh in 2013.

On 24 April Philomena Dunleavy arrived in Edinburgh from her home in Dublin. Aslightly built, shy woman of sixty-six, she had come to visit her eldest son,Seamus. At his flat in Balgreen Road, they started to catch up. Seamus talkedabout his recent work labouring on Edinburgh's tram network. In return,Philomena tried to fill him in on the news of his four siblings. But Seamus wasbehaving oddly, first distracted and then agitated.

Philomena was alarmed. She told her son she was going to have a look aroundEdinburgh, but instead she went to Portobello Police Station. She asked anofficer there where she could get a cheap room. She said, 'I don't want tospend the night with my son whilst he's having an episode.' A few days later,Seamus called his father in Dublin to say his mother was on her way home. Shenever arrived.

On 6 June a 24-year-old ski instructor went for a bike ride in Edinburgh'sCorstorphine Hill nature reserve. The weather was hot and he decided to stopriding and find a place to sit in the sun for a while. He was pushing his bikealong a narrow path when he saw a set of brilliant white teeth gleaming at himfrom the dirt. The teeth were set in the remains of a severed head. Most of theflesh had rotted away, but the carrion-loving flies were still there.

In the shallow grave that the shining teeth had revealed, forensicanthropologist Jennifer Miller unearthed two severed legs and a human trunk,which she ascribed to a woman of about sixty. She noted that the brilliantteeth were the result of expensive cosmetic dentistry. One of the rings sheremoved from the cadaver was a traditional Irish Claddagh ring. Equipped withthis limited information, the police spent weeks searching missing personslists.

Finally they asked Caroline Wilkinson to make a facial reconstruction, whichshe did using 3D scans of the skull, then filling in the soft tissue digitally.The resulting image was circulated to police forces throughout Europe, andshown on BBC Crimewatch. The presenter of Crimewatch also mentioned theCladdagh ring, which made a family member in Dublin feel doubly sure that shewas looking at Philomena. The likeness of Wilkinson's image was uncannilyaccurate. The identity of the body was put beyond doubt by dental records.

A few days later Seamus was arrested and charged with his mother's murder,which he denied.

The jury didn't believe him. Instead, they accepted the prosecution's case thatPhilomena had gone back to Seamus's flat sometime after she'd spoken to thepolice. There, she had died. The pathologist noted damage to the small bones inher neck (which often signifies strangulation), injuries to her head andsmashed ribs. Seamus had cut off her head and legs with a saw. But it was impossibleto tell whether these injuries had been sustained before or after death. Ajournalist for Herald Scotland reported on the more disturbing possibility:'Philomena Dunleavy may still have been alive, but unconscious, when her sonbegan to hack off her legs.' The exact circumstances of her death will never beknown.

What we do know is that Seamus then put his dismembered mother's remains in asuitcase, and took her up Corstorphine Hill. He hacked out a shallow grave witha spade, and dumped his mother in it. As forensic experts often tell us, murderis easy compared with the difficulty of disposing of a body effectively. It wasonly two months later when her body resurfaced and, with it, the vital cluesthat would lead to his conviction. The prosecutor called it 'a case in whichpieces of evidence came together like strands in a cable'. In January 2014,Seamus Dunleavy was convicted of murder, in no small part because of CarolineWilkinson's work.

Such a swift identification of a victim is not guaranteed. On 18 November

1987, a cigarette butt ignited some rubbish under a wooden escalator atLondon's busiest train station, King's Cross. The fire grew in intensity untilthe escalator vented a fireball burning at 600°C, which hurtled up theescalator and into the underground ticket hall above it.

Hundreds of people were trapped in the complex of tunnels connecting King'sCross's six Tube lines. Some took the escalator up to escape the black smokeunderground, and were burned alive. Others thumped on doors to try and get ontrains that didn't stop. When firefighters finally fought back the blaze, theydiscovered thirty-one dead bodies.

Over the following days and weeks the police managed to identify thirty of thedead. But one middle-aged man eluded them. Richard Neave was asked toreconstruct the man's face, which had been terribly burnt by the fireball. Hefound some pieces of tissue around the nose and mouth which helped him predictthe shape of that part of the face. And he was given an extensive dossier outliningthe victim's height, age and state of health.

Interpol were approached for help and enquiries were made as far away as Chinaand Australia. Richard Neave's reconstruction was shown in all the majornewspapers in the UK and hundreds of people phoned up, believing it to besomeone missing from their circle. But no definite matches were possible.Meanwhile the body was buried in a grave in north London marked 'AN UNKNOWNMAN'.

In 1997 Mary Leishman, a middle-aged Scot, made enquiries about her missingfather, Alexander Fallon. When his wife had died in 1974, Fallon's life hadfallen apart. He had been unable to cope with everyday life. He had sold hishouse and ended up sleeping rough on the streets of London, among thousands ofother virtually anonymous homeless people. Mary and her sister had begun towonder whether the unknown victim of the King's Cross fire might be her father,but she wasn't hopeful. At the time of the fire her father had beenseventy-three years old and five foot six, whereas the post mortem had put thedead man at between forty and sixty years old and five foot two. Yet the corpsehad smoked heavily, as had Alexander Fallon and, like him, it had a metal clipinside the skull as a consequence of brain surgery. At the time of Mary Leishman'senquiry the police thought they had a match with another missing man, HubertRose, so they didn't follow up her query. Then, in 2002, a servicecommemorating the fifteenth anniversary of the victims of the fire was held innorth London. This nudged Mary Leishman to raise her concerns with the policeagain.

In 2004 Richard Neave was shown photographs of Mary Leishman's father. Herifled back through his records to find photographs of the mystery victim'sskull, and his own clay model. He compared frontal and profile photos and sawthe similarities immediately – both had prominent cheek bones, thin lips,similar spacing between the eyes, the same laughter lines running from thecorner of the mouth to the chin, although the man in the photograph had a muchmore bulbous nose than his model. With the further corroboration of dentalrecords and of the neurosurgeon who had inserted the metal clip, the finalvictim of the King's Cross disaster was finally identified as Alexander Fallon– sixteen years after his death.

A photograph of Alexander Fallon, a victim of the King's Cross fire, comparedwith the facial reconstruction created using his remains

Richard Neave's model of Alexander Fallon prompted his daughter Mary'senquiries. That was all it was designed to do. A string of other factors,including documentary evidence, supported the identification and made atraumatic exhumation unnecessary. And, as Mary Leishman said: 'One thing thatmakes us feel certain now that my father was a victim of the fire is that wehave, with the help of the police, established that no benefits were upliftedin his name after the date of the fire. If my father was alive, he would havebeen first in the queue whenever there was money to be had.'

If the King's Cross fire happened today, Alexander Fallon's face would berebuilt by computer. Digital modelling has not replaced clay modelling – whichCaroline Wilkinson still teaches to her students in Dundee – but nowadays 80per cent of forensic facial reconstruction is computer-based.

First, Caroline will scan the skull in three dimensions, usually with a CTscanner, and then import the resulting model to an image-editing program. Thenshe chooses one of a number of basic muscle templates and overlays it on to theskull. Now Caroline tweaks the muscles manually – click, drag, click, drag –based on the same standard thicknesses that she uses when working in clay.Computer modelling is quicker than clay modelling because having the templatesmeans Caroline doesn't have to start from scratch every time. But not by much.It takes a long time to add skin, eyes and hair, and to texture them properly.

But there are advantages to the computer method other than speed. Caroline canvary elements such as skin tone and hair colour, then print out a dozenpossible images for investigators to look at. Three-dimensional scanning allowsthe reconstructor to see injuries to the skull, such as a hammer mark, moreclearly than plaster casting. With precise modelling of the wound and theweapon it's possible to make a model of the event as well as the face, whichcan be shown in a courtroom further down the line. If someone recognises areconstruction and sends in a photo of their missing loved one, artists canscan it and superimpose it on to the skull. This is the digital version of thetechnique that was used for the first time to incriminate Doctor Buck Ruxton inthe Jigsaw Murders of 1935 (see pp.48– 51).

Craniofacial modellers don't only use computers to create a face as it oncewas, but also as it might now be, especially in the case of missing persons.The process of 'age progression' can be automated to a significant degree. Ourears grow longer as we grow older, at a more or less predictable rate, andalgorithms exist to plot the basic sagging and puffing of an aging face. Butage-progressed images are largely down to the instincts and experience of theartist, who looks at sequences of photographs of people as they get older andidentifies general trends. The artist uses photos of older siblings forguidance, adapts the image to reflect the kind of life a subject may have led,and adds distinctive clothing or facial hair. Fine details like liver spots canbe added manually, too. For Caroline Wilkinson, 'The most difficult things towork out are skin colour, eye colour, how fat or thin they are, and whetherthey have wrinkles.'

The 'Butcher of Bosnia'. From left: former Bosnian Serb leader Radovan Karadžićin 1994; as he appeared while evading capture after his indictment for warcrimes; and at the International Criminal Tribunal for the Former Yugoslavia inthe Hague in July 2008. He was charged with 11 charges of genocide, war crimesand crimes against humanity

The hunt for missing people can also be hampered by changes to their appearancethat have nothing to do with aging, and which can be effected by techniques assimple as growing facial hair. Radovan Karadžić is a former Bosnian Serbpolitician who was indicted for war crimes in 1995 by the InternationalCriminal Tribunal for the Former Yugoslavia. Among other atrocities, Karadžićwas charged with ordering the 1995 Srebrenica Massacre in which 8,000 Bosnianswere murdered. After his indictment, the 'Butcher of Bosnia' disappeared,shaved his hair, grew a beard, donned a priest's robe and lived an itinerantlife, wandering from monastery to monastery.

Caroline Wilkinson was asked to make an age-progressed image of Karadžić. Shegot the shape of his face spot on, but underestimated his beard. He had movedto Belgrade and started wearing his long hair in a ponytail, donning largesquare glasses and hiding behind an enormous white beard. Calling himself'Dabić the Spiritual Explorer', he masqueraded as an expert in human quantumenergy, worked in an alternative medicine clinic and gave public lectures. Butthe age-progressed images gave a new impetus to the hunt for Karadžić. In 2008,a year after Caroline had sent them her image, he was arrested by Serbiansecurity forces, and extradited to The Hague to stand trial. The trial is stillgoing on.

Computers often help forensic artists identify less heinous criminals, too.They analyse CCTV footage and compare the images to a suspect. When offendersdo not simply break down and confess at the sight of their blurred selves onvideo – which they often do – it is difficult to prove conclusively that it wasthem. Even when footage is high quality, identifying an unfamiliar face bysight is not the most reliable of procedures. Computerised facial imagecomparison can offer a more reliable alternative. One method is to superimposea still from the video on to a photo of the suspect, although that can beawkward when criminals haven't been looking straight at the camera, which theytend not to do. Another technique, which has been used in UK courts for thelast fifteen years, is called photoanthropometry. This involves comparing theproportional distances and angles between landmarks on two facial images. Butthe technique is not perfect. Even when the suspect is asked to pose for theirphoto in the same alignment as the person in the video, there is a complexbundle of variables to accommodate, such as distance from the camera, cameraangle and head orientation.

We have seen how forensic artists identify the dead from their skulls, missingpeople from photographs, and wanted people from video footage. One othersignificant aspect of their work is to depict wanted people based on eyewitnessaccounts. Historically, this was the job of a sketch artist who would translatean often shaky witness recollection into a drawing of a suspect. But in the1980s researchers at the University of Kent helped develop an alternativemethod known as E-FIT (Electronic Facial Identification Technique). Policeforces around the world now use E-FIT and it makes regular appearances in themedia. To make an E-FIT, an eyewitness looks at a swatch of computer-generatedfaces and clicks on the one that looks most like the person they saw. They arethen presented with another, more narrowly defined set of faces. In this waythe image is refined, until it is a relatively close representation of theperson the witness remembers.

Facial reconstruction began as a way of bringing us face to face with ourhistory – and we are still using it for that purpose. In 2012, a set of boneswas found underneath a car park in Leicester. They were suspected to belong toRichard III, the last Plantagenet king of England, who died at the nearbyBattle of Bosworth Field in 1485 and had been buried in a local church.

The Richard III Society assembled a team to investigate the remains. Scientistsbegan analysing DNA samples, and scanning the skull in three dimensions. Theysent the digital skull to Caroline Wilkinson, who set to work on making theking's face, avoiding looking at existing portraits of him so as not tocontaminate the scientific process. She and her team modelled the muscles andskin using stereolithography, a computer process whereby a moving laser beambuilds up a structure, layer by layer, from a liquid polymer that hardens oncontact with laser light.

When the DNA results came back, and matched with a descendant of the king,Caroline finally compared her model to portraits. It was strikingly similar,with its arched nose and prominent chin. 'It doesn't look like the face of atyrant,' said Philippa Langley of the Richard III Society. 'I'm sorry but itdoesn't. He's very handsome. It's like you could just talk to him, have aconversation with him right now.'

Caroline is proud of her work on Richard III. 'Our facial reconstructionmethods have been blind tested many times using living subjects and we knowthat approximately 70 per cent of the facial surface has less than 2 mm oferror,' she reveals. To reach that level of accuracy Caroline stands on theshoulders of all the face makers that have come before her, from Giulio Zumboto Wilhelm His and Richard Neave. But it's her own artistic obsession withobservation that has helped her to do it so effectively. She describes herselfas 'really annoying to go out with because if I'm watching a film I spend allmy time going, "Oh, look at his ears, look at his nose, what a great nose," andeveryone's going, "Shut up! Just watch the film." When I'm on the train I quiteoften get my phone out and take sneaky pictures. I get my iPad out, pretend I'mreading something and take a picture, I'm terrible.

'I also collect photographic portrait books wherever I travel overseas, which Ido mainly for my archaeological work. The places I visit have books of photosthat you can't get on the internet. So if I go to Egypt I'll try to buy a bookof pictures of Egyptian faces, and so on. So now I've got this great databaseof faces that we can use to inform what we do.'

And it's that access to such a vast range of faces from all over the world thatmakes our present-day forensic artists more useful artist-anatomists thanLeonardo da Vinci could ever be. It's the application of science to the worldof artistic representation that makes it possible for the dead to tell yetanother chapter of their story to us.