Face To Face

Joelle Steele Interviewed by Valerie McCady, October 20, 2012

If anyone has ever found a unique career niche, it must be Joelle Steele. Steele is a writer, artist, photographer, and publisher, who also makes her living as a court-certified expert in the analysis and comparison of ears and facial features to authenticate identities of people in photographs. She is the author of five books on the subject.

Valerie McCady: How did you discover such an unusual kind of work?

Joelle Steele: In the early 1970s, I became interested in family photos after my mother brought back a lot of them from a visit to her relatives in Finland. At the same time, a friend of mine had just inherited four old albums and a bunch of loose photos. Many of my mother's and my friend's photos were unidentified. So, I saw it as a big puzzle to solve.

VM: What did you study to learn to do this kind of work? Did you have to study anatomy?

JS: Actually, I began by studying anthropometry - the older name for what we now call biometrics - and craniofacial anatomy under Leslie G. Farkaš, Ray Iwata, and W.R. Stark. I traveled a lot to to meet these experts and learn whatever they could teach me that applied to identifying faces. I also studied forensic anthropology under Clyde C. Snow and Simon Fuchs, which covers, among other things, the identification and analysis of bones. Bones are so important. And I took a class in biological anthropology, which also addressed bones. I learned all about ear structure and ear patterns by studying under ear identification expert Alfred V. Iannarelli – with whom I don't always agree, but he is the expert on ears as identifiers. And, by the way, I was introduced to him by the husband of my friend who had just inherited the photos.

VM: When did you first start practicing professionally?

JS: In 1980. My first client was a Nazi hunter with more than 200 photos purported to be of missing Nazi war criminals, two of which were exact matches. I have since gone on to analyze more than 100,000 photographs for measurement and comparison to 25,000 unidentified faces.

VM: And how did you solve the identification of your family and friend's photos?

JS: Very, very slowly, and at first I made a few mistakes — which I quickly corrected!

VM: How many photos are we talking about?

JS: My friend had about 400, and at that time my mother had about 1,200.

VM: That is a lot of photos.

JS: It's actually not that many when I think about some of the family photo ID projects I've worked on in recent years. I've had a few projects with more than 3,000 family photos spread throughout four to six albums, plus loose photos. These days, we're often talking about photo collections that span more than 150 years, so they can be quite large. I personally have more than 4,000 family photographs from my mother's side of the family.

VM: How many family collections do you normally analyze over the space of a year?

JS: Not many. One or two is average. The majority of my clients just have a single photograph of a family member or some historic figure or missing person that they want me to examine to see if it is who they think it is.

VM: What do you need from them to make this kind of analysis?

JS: I need to have a very high resolution scan of their photograph, at least 300 dpi, preferably 600 dpi. And, I need to have at least two or three exemplars, also high resolution scans..

VM: Can you explain what exemplars are?

JS: Exemplars are photographs used to make the comparison. They are known photographs of the person. For example, if someone brings me a photo they think might be their great-uncle Harry, then I need to see some photos that definitely are of Harry. And, if Harry has brothers, I need to have photos of them too, especially if there are strong resemblances in a family. I don't want to identify the photo as Harry if it is really his brother Fred.

VM: Don't all people who are related look alike?

JS: Surprisingly, they don't always look alike at all. It is not at all unusual for a person to not resemble their parents, but to instead resemble a grandparent or a great-aunt, or even a great-grandparent. And they may have a sibling who resembles a grandparent on the opposite side of the family. No logic to genetic inheritance, DNA recombines with every child. I see this in my own family all the time, both sides, multiple generations..

VM: How does this affect your ability to identify people?

JS: Well, it doesn't. I'm not trying to show that anyone is related. I'm just trying to accurately match up their faces to give them a name, an identity.

VM: Are there a lot of unidentified faces in family albums?

JS: Tons of them, mostly in bins in antique malls. People often don't label old photos at all, and sometimes they label them incorrectly. And some people get cute about labeling photos, writing on the back that someone is a famous person when they aren't, simply because they kind of look like that person or are in a pose that resembles some famous person. I get a lot of those from people who claim to have photos of Lincoln or Jesse James or Bonnie and Clyde.

VM: So, what exactly do you look for in a photo to distinguish, say, a photo of Lincoln from one of a farmer with a beard and stovepipe hat?

JS: Well, again, the exemplars are a necessity. I gather them up first, and then I take a quick look at them onscreen next to the questioned photo to see if there is any resemblance at all. I have a really good eye for spotting resemblances, so if one is there, I usually see it immediately. Then I have to select the exemplar that is about the same age as the person in the questioned photo or that is facing the same way. Then I look for the main things that rule out matches.

VM: And those would be ...?

JS: Ears and bone structure. Ears are as unique as fingerprints. If the ears are visible, they are the very first things I look at. If the ears of the person in the questioned photo don't match the ears in the exemplars, then it isn't the same person. If the ears are not visible in one or both photos, then my next step is to align the eyes. I reduce or enlarge the questioned photo with the best exemplar until the pupils are aligned in a straight line. Then I make the irises the same size. This is a critical first step to comparing faces that most people do not do, even other experts. So, after the eyes are aligned, I look at how things line up between the two faces. If, for example, the questioned photo has eyes that are larger in proportion to their face than those of the exemplar, it will be very obvious, because the head in the questioned photo will now appear to be much smaller than the face in the exemplar.

VM: So the first step is ruling out matches?

JS: Absolutely. You can waste a huge amount of time examining faces for surface traits that are very similar, when the ears or the bones of the face are completely different. I don't want to charge someone to measure everything – 100 or so items per face – if a preliminary dozen or so measurements can rule out entirely the possibility of a match.

VM: What if it turns out to be a possible match? Where do you go from there?

JS: From the very start, I offer a client two options. One is just a yes/no as to whether it is a match or not. The second is a brief explanation as to why it is or is not a match. It is almost always about either the ears or the bone structure and not the fleeting idiosyncratic traits such as moles or scars.

VM: Fleeting?

JS: Yes, surface idiosyncracies are fleeting. In other words, they may not appear in every photo due to the photographic process itself, or they may appear darker and more pronounced in some photos due to the photographic process, or they may appear at certain times in a person's life and may either worsen with age or fade away entirely over time.

VM: How can you be sure the measurements will ever match when photographs are different sizes?

JS: You can't. That's why you don't ever compare the measurements. They have to first be converted into proportions (ratios, percentages) of the face, and those proportions are what are compared. So, for example, if one nose length represents 23% of the total face height, and the other nose represents 31% of the face height, the faces are not a match.

VM: What about problems with the way the person is facing in the photo? Doesn't that affect the measurements?

JS: That can be problematic at times, but it really depends on what I'm analyzing, whether it's a photo from a family collection, or a single random photograph compared to a couple of known exemplars. With the latter, some angles could make analysis impossible, or they could show enough data to measure to rule out a match, but not enough to verify an exact match, and the latter is essentially the same as a non-match.

VM: Is it harder to identify people in families or in individual photos?

JS: The analysis of the face is the same, but with family albums it takes much longer because I have to sift through so many photographs. With a single face analysis, I already know who the client thinks the face in their photo might be, and they have sent up to three exemplars for comparison. That means I'm comparing only one face to another face, and not trying to find possible matches from a collection of 100 or more photos.

VM: I guess that would be similar to law enforcement officials trying to find a suspect in their criminal databases.

JS: Yes and no. With law enforcement, they are looking at a computerized database of faces that includes other identifying features such as height, weight, hair color, age, etc., that can be used to narrow the search. With family collections – especially ones from the 19th century – you don't get all those other details. And, if there are strong family resemblances, you frequently find lookalikes that may be hard to distinguish from each other.

VM: Does that mean you can identify people who might be related by looking at their photos?

JS: Actually, no. Not with any level of accuracy. Family members can resemble each other – or not. They can resemble an ancestor instead of a parent, they may not resemble their siblings at all, etc. Even identical twins (and conjoined twins) often have differences in ears and bone structure. But what is more interesting to me is that people who are not even distantly related can look uncannily alike and share bone structure similarities as well. There are lots of photos of unrelated people who look alike all over the Web. There's even a Canadian photographer named François Brunelle who takes photos of these unrelated lookalikes.

VM: I guess that means DNA is the only true proof of a biological relationship.

JS: As far as I know, that is the case.

VM: What about age progression? Is that accurate at all?

JS: I don't practice age progression, but I know that to do it accurately it's important to have access to family photos to see how people develop and age within a particular family. I know from my own experience with family photo collections that all people age differently from one family to the next. I have heard pros and cons as to how accurate age progression is, but from what I have seen, it seems to be close enough to make a good facsimile of an adult based on a photo of a missing child or young person. I think that's the main purpose.

VM: What other types of facial features analysis or photo analysis do you do?

JS: I examine faces in snapshots, surveillance photos, or security camera footage, usually for private investigators or criminal attorneys. I've had quite a few post-mortem photos come my way as well. Sometimes I'm asked to look at faces and other aspects of photos and digital photos to determine whether or not they have been altered in some way. It's not always easy to tell because people who do these alterations can be very skillful. I have also examined faces in fine art portraiture, usually to match unidentified faces in paintings with known faces in photographs. And, I do get the occasional "is this a ghost in my photo" questions from people who have reflective oddities appearing in their images.

VM: Do you have experience analyzing photos of people who have undergone plastic or reconstructive surgery?

JS: Yes, a little. It doesn't come up very often. If a person hasn't had their ears modified too severely or at all, those are always a good place to start. The bone structure is usually not changed much, if at all, unless it's a major reconstructive surgery. But my experience has been that most people who have plastic surgery do not go to the extremes of a Michael Jackson. Most of them just have their eyelids lifted, noses narrowed, ears pinned, etc. So the bones are unchanged and there is usually still a lot of recognizable face under all the improvements and enhancements.

VM: Are there any criminal issues that come up for you in this type of work?

JS: Not often. I work mostly on civil cases. I am often asked to prepare a report of findings, and/or a declaration regarding my analysis and comparison of the facial features of two people, one of whom is a defendant or plaintiff in a civil case. And I have had to testify in court a few times, which is how I became court-certified as an expert in face and ear comparisons. Criminal cases involve the police, and they generally have their own experts, usually forensic anthropologists and medical examiners.

VM: Do your findings ever vary from those of other experts?

JS: Not usually, but our face measurement methods of getting to the truth can vary. For example, a forensic anthropologist can analyze faces in photos, but they normally work in the 3-D, real world with live humans or their remains. The 2-D world of photography is quite different, so how we approach an analysis is not the same.

VM: Does this mean that your training is also different than theirs?

JS: Absolutely. I understand what they do, just as they understand what I do. But I would be lost in the 3-D real world, as they are often lost in the 2-D world of photography. They train with real human bodies, and all I've got to work with are pictures. Huge difference.

VM: Do you advocate for one party over the other in a lawsuit?

JS: Never. That is the attorney's job. My job is to remain neutral, look at the faces, look for the truth in or behind them, and present my findings to the court. I learned many, many years ago to make this perfectly clear whenever I am hired by an attorney. I do not allow myself to be put in a position where I would have to commit perjury — either by an outright lie or by an omission or alteration of the facts.

VM: So you really do have an unusual niche carved out for yourself?

JS: Yes, it is unusual. And it's always interesting and challenging.