THE ATTIC : What's In A Face?

Date Added
MAY 2004

The motorist could hardly believe his eyes.  It had looked like a bundle of rags at the side of the road, but something about it seemed odd..  He stopped to have a closer look, and to his horror discovered the almost unrecognisable remains of a corpse.


Police called to the scene just outside Manchester quickly established that the body was that of a man and that he had been murdered.  The skull had been fractured by a blow and the victim had also been garotted with a length of wire.  The body had been set alight after death.


But who was he?  Burning the body had made the face unrecognisable, and no one with his general height and build had been reported missing in the Greater Manchester area.


After a period of getting no further with the case, the investigators decided to reconstruct the features of the victim.  They sent the skull to Richard Neave of the University of Manchester, one of the world's leading exponents of the art of facial reconstruction.


Neave was told that the dead man had probably been Chinese or Malaysian, and was most likely in his mid-20s.  With this information he began to model the victim's features in clay on to a cast of the skull.  Some of the victim's longish hair had survived, and it was included in the model.


The completed bust was not much help at first, since nobody could suggest an identity.  Perhaps the victim was from another area?  If so, he could have been murdered hundreds of miles away before being dumped by the side of the road.  As a result, the reconstruction was shown nationwide on the BBC's Crimewatch programme.  Within hours of its broadcast a man from Leicester called to say that it looked like his uncle, Sabbir Kassam Kilu, who had disappeared just before the body was found.


Investigations showed that the body was indeed Kilu.  He had fallen foul of his wife's family and been killed by them.  His wife, brother-in-law and sister-in-law were later convicted of his murder.

FAMILIAR FEATURES


Sabbir Kassam Kilu was in fact Asian and in his 40s, so the likeness created by Neave had the wrong ethnic group and wrong age.  Nevertheless, people are extremely good at recognising familiar features.  Evidently there had been enough similarity in Neave's model to strike a spark of recognition in Kilu's nephew, and his identification enable the police to solve a particularly nasty crime.


Identification seems a simple process.  After all, we do it every day.  We see our friends and our neighbours, and have no problems in putting a name to a face. 'Someone you know well goes into hospital,' says Richard Neave, 'and you don't see them for six weeks.  They come out looking thin and haggard, but even though their appearance may have changed dramatically, you can still recognise them, maybe comment on how poorly they look.  You don't say, Who are you?'


Neave works by moulding clay on to a cast of a skull.  He first builds up the muscle groups and then adds a layer of clay to represent the skin.

In 1971, a mild-mannered accountant, John List murdered his wife, mother and three children and then seemed to just disappear into thin air.  The Police in New Jersey were at a dead loss.  As the years passed, the List killings became just another one of the many unsolved crimes.

Then in 1989, the TV programme America's Most Wanted featured the List case and called on the specialist skills of artist Frank Bender to produce a facial reconstruction.  Working from an old photograph, Bender produced a bust of John List, but aged by 17 years.

The TV station received hundreds of calls, many identifying a man called Robert Clark who lived quietly in Richmond, Virginia.

The FBI called on Clark, who protested his innocence.  But although he could change his name and his address, he could not change his fingerprints.  Seventeen years late, John List, alias Robert Clark, was returned to New Jersey to face prosecution on five counts of murder.

LANDMARK SITES


The depth of flesh on a human face is fairly standard and well known at key 'landmark' sites on the skull, and the depths elsewhere are averaged between these known points.  There are difficulties with individual features - but an experienced modeller will be able to make a reasonable estimate as to what type of nose fits a certain face.  For instance, a long, thin face will generally have a long, thin nose, while broader features will be reflected by a broader nose.


'There was a case in 1983,' Neave recalls.  'A skeleton was found in the grounds of the Duke of Marlborough's estate at Blenheim in Oxford.  There was nothing by which the remains could be identified, and eventually the skull was brought to me.  It became clear as I worked on the head that this was an absolutely classic north European Scandinavian face, so that's how I completed it.


'The police continued to search the area, looking for anything with a Scandinavian link, and also checked for missing Swedes and Finns.  Eventually they found a passport and some clothes belonging to a 23 year old Finnish nurse named Eila Karjamailen, who had disappeared while backpacking through Europe some time before.  The reconstruction looked like the passport photo, so the police checked with Helsinki.  Dental records from Finland showed that this was indeed Eila.  They never did find out who killer her, but at least they found out who she was.'

The high-tech method of facial reconstruction starts with accurate measurements of the skull by using the reflections of a laser beam.  As the skull is rotated the colour laser scans the whole surface of the skull.  The resulting measurements, accurate to fractions of millimetres, are then interpreted by modelling software to build up a three-dimensional image.

Modern digital film and computer technology enable investigators to quickly compare an unidentified skull with a photograph of the person it is suspected to have come from.  The idea is to overlay the photograph with an image of the skull.  If the angle view and magnification of the images are the same, and if the skull is indeed that of the person in the photo, the images should coincide with key areas.  Two important comparisons are the nasal aperture and the nose, and the orbits of the eyes.

COMPUTERS REPLACE CLAY


Clay modelling is the oldest technique used in reconstruction.  The pioneering work was done in the 1920s by the Russian palaeontologist Mikhail Gerasimov, and all the work carried out since then is a refinement of his techniques.  But it is being supplemented by - and may ultimately be replaced by - modern high-technology.  Lasers and computers can be used to create a graphic image in the same way that the medical artist uses clay.  The skull is scanned in all dimensions by a laser, and the resulting measurements are fed into a computer, which can then produce a highly accurate three dimensional image of the skull called a wireframe.  Data on standard skin and muscle thickness are kept in a data bank, and using this information the image of the skull is 'fleshed' out.  Common facial and racial characteristics can also be added to the program.


'There are problems of course,' says Dr Peter Vanezis, formerly of the Charing Cross and Westminister Medical School.  'At the end of the day the computer cannot produce as good an image as the medical artists, at least using current technology.  And then there is the problem of over  finishing the image.  If you put too many details in, then it acts against recognition.  The hair might be different, and someone will say that although it looks a bit like their cousin X, who is missing, the hair is wrong so it can't be them.


'Funnily enough, you have a better chance of identification with a generalised image because people recognise shapes.  They will not be able to tell you, most of them, how they recognise somebody - it is very intuitive.  Psychological studies show that what they recognise is the forehead, the eyes and the mouth and much more rarely the nose.'

In 1948 a skeleton was found in a trench in the Ukraine.  The skull had been severely damaged and the police thought it might be the body of a woman who had been reported missing less than a year before.  As there was no way of confirming her identity the skull was sent to Professor Mikhail Gerasimov at the Laboratory for Plastic Reconstruction in Moscow.  Gerasimov was a palaeo-anthrapologist whose field of study was ancient human remains.  In the early 1920s he learned of the work carried out by pathologist A.D. Grigoriev in Moscow, which correlated the dimensions of the skull and the appearance of the face.  Gerasimov went to work for Grigoriev, and built up enough data on skin and flesh thicknesses to attempt facial reconstructions.


During the next 20 years he became the world's first expert on forensic reconstruction and was called upon to help out in more than 140 criminal cases.

Gerasimov thought that the Ukraine skull had been damaged by shell splinters, which indicated that it was probably a wartime casualty.  The police were sceptical of this new technique, and they remained certain that the remains were those of a murder victim.


Gerasimov's reconstruction revealed a young woman's face with pronounced Slavonic features that did not look like the missing woman.  The investigators circulated photographs of the bust and within weeks the head was identified as being that of Valentina Kosova, who had disappeared in early 1941 as the Germans fought their way through the Soviet Union.  Gerasimov had been right all along - the skeleton was a victim, not of murder, but of war.

DEAD ACCURATE DATA?


Computer modelled facial reconstructions are the future, however, and at Sheffield University the technique continues to be developed and refund with funding from the Home Office.  One of the limitations of early computer programs was the quality of data.  For many years forensic scientists relied on skin-depth data collected from cadavers in the early part of the twentieth century.  These measurements have always been considered suspect for a number of reasons:  the sample size was small, tissues alter size after death, and the data did not take into account variations due to age, sex or ethnic origin.   Computed tomography, known as CT or CAT scans, can provide a wealth of tissue data that will eventually provide a far more accurate database for the modelling software to draw upon.  Refining the data into 'datasets' applicable to individuals of a particular age, sex or ethnic group will further enhance the forensic scientist's ability to recreate a recognizable face from a fleshless skull.



GENETIC VARIATION


However sophisticated the software becomes and however fast the computers, creating an exact replica of a person's face from the bare bone of their skull will, in the foreseeable future at least, remain a Holy Grail.  This is because a number of our soft-tissue features, such as ears, noses and lips exhibit a wide variety of forms determined to a large degree by genetics.  Until we can 'read' a person's DNA with sufficient proficiency to predict the size and shape of the nose, forensic science will have to content itself with a 'reasonable likeness' based on the best available data and a dash of artist invention.


But reconstruction is not the one way in which to identify the skull.  The Charing Cross and Westminister Medical School has also used computer and camera technology to make identifications using the superimposition of images.


First made use of in the 1930s and 1940s in notable investigations such as the Ruxton and Dobkin murder cases, image superimposition is essentially a technique for confirming identity.  It involves comparing a photograph of a skull with a photograph of the suspected victim.

Computer technology is the key to modern facial reconstructions and comparison.  Medical artists require a few days to complete a reconstruction, while a computer can complete the process in a matter of seconds.  It is also easier to change a computer model.  If more information on the age or the race of the victim becomes available, a sculptor may have to start all over again, but the computer operator can simply add the new information to the existing image and modify it all at the touch of a button.

A TIME CONSUMING PROCESS


'In those days, you had to compare the photograph with a traced overlay,' says Dr Vanezis.  'Your image of the skull had to be in exactly the right position, angle and magnification as the photography you were comparing it with.  What you were looking for was a match in the brow ridges, the orbits of the eyes, the jaw line and the outline of the skull.  It sometimes took a lot of photography and a long time to get it right.


'Using video was a big step forward.  We have one video camera on the skull and a video image of the photograph.  We can combine the two images using a mixer.  Having a zoom lens means that you can get the magnification right very quickly, and you can just adjust the angle and attitude of your subject skull as required.  You can do various wipes - horizontally, vertically, diagonally - and you can see almost at once whether you've got a match.'


'It's not conclusive proof of identity,' Dr Vanezis warns.  'Given the state of the art, what we're doing is providing corroborative evidence.  But we're getting better all the time.'

BACK TO TOP