Interview Gerald Hurst

A top fire science expert and chemist, he's been doing pro bono arson defense work for more than a decade. In February 2004, days before Willingham's execution, he submitted a report to Willingham's attorney stating the house fire was not arson. This is the edited transcript of an interview conducted on Jan. 15, 2010.

• HIGHLIGHTS
• The Lime Street Fire -- the case that changed fire science
• Willingham's investigators relied on "old wives' tales"
• "95 percent of fire investigators get it wrong today"
• An arson-murder case, similar to Willingham's, but with a different outcome

[How did you get involved in arson science? What's your background?]

... I began with a Ph.D. in chemistry. My work [in] chemistry was in "high-energy" chemistry; that is, explosives, incendiaries, rocket propellants. I started out as literally a rocket scientist during the moon race. As the moon race wound down after we got to the moon, I moved into corporate warfare, working on the problems that we had in Vietnam, mainly on the methods that we had in destroying the enemy infrastructure, behind enemy lines; in other words, legal arson, if you will. I went from there. After shooting some oil wells with liquid explosives, I formed an explosives company and again worked in incendiaries and explosives. A long time ago, [I] invented a binary explosive that's still on the market. ...

And then [I] became chief scientist of the Atlas Powder Company, which at the time was the largest manufacturer of explosives and squibs and energetic stuff in the country. I started consulting on fires in about 1972 and did civil cases mainly. By civil cases I mean product liability -- finding the cause of the origin of fires. … And we would look at the evidence to see if, once again, is there any real evidence that this particular item started the fire or not? So I was looking at the cause of origin of fires, largely in an industrial setting: a lot of plant fires, a lot of explosions, ship fires -- you name it -- fires anywhere.

I had done a workman's compensation case. I had a Mexican laborer who had been burned, and we did the cause of origin on that. [I was] working for a Texas attorney, and a few years later he got saddled with an arson case. This was 1995. And he didn't know where to go with it. It's very difficult to find a defense expert locally, because people don't want to go against the police or what have you. He remembered going with me on a civil case, so he brought me the Sonia Cacy case [of alleged arson-homicide] in 1985. ...

They were having a rehearing on the case. She had already been convicted, but they were having a sentencing, a rehearing -- not looking at the original facts of the case.. ... I read the Sonia Cacy case trial, and I was appalled. It was just garbage. I couldn't believe that arson investigation could be that bad. So we went over and testified, and she got 99 years in the resentencing, which really kind of ticked me off. So I undertook to get her back out. It took two or three years, but I did.

[I] got her out of prison, got a lot of publicity early on in that case, and my phone began ringing. I found out there were people all across the country in Sonia's position. These are largely housewives, believe it or not, being accused of torching their houses or their little businesses or what have you. I began consulting with them and helping them out. Got a ton of publicity on the first case and the second case and the third case, ... and it just kept going. 1995 was the end of my career [of] making money doing fire. That grew so rapidly that I have not had time for any civil work since.

So the vast majority of my work has been pro bono arson defense. I'm kind of the go-to guy when you can't find anybody for advice. I get calls from public defenders from all over the country. I always have them send me the fire investigation report and some pictures and what have you. [I] evaluate the case, maybe spend a week on the case, and get back to them. What I try to do is tell them whether the case seems to have merit, what kind of additional expertise they need, and explain to them how the prosecution is going to handle the case and what they need to look for. If it's a paying case, I need to get rid of it, pass it on to somebody else. If it's a white-knight case, I can also get rid of them. ... And then the cases that look really bad, ... I will delve into them and see if there's anything at a deeper level, and every once in a while it turns out that there is, that a case is not what it appears to be.

[How did you get involved in the (Cameron Todd Willingham) case?]

Shortly before the execution, I was contacted by Walter Reaves, the attorney on the case [during Willingham's appeals], ... and Walter was able to get me the documentation that I needed, but we were on a very short fuse. ... I asked him for a deadline for getting a report out. He gave me a firm date by which a report had to be ready. I studied the case and almost immediately found that it was clearly a bogus arson case. Got my report out by the deadline, and Walter subsequently filed a report with the parole board and governor's office. It didn't do any good. Willingham was subsequently executed.

In some of the articles [written about the Willingham case], it says there were only 48 hours, ... that you were in such a hurry that you had to handwrite the notes and that there were typos.

Well, there were probably. ... It went out about four days before the execution. It was a short report, a very succinct report. It was all of the important points. There was plenty of time for it to be digested. It's not as if we were trying to get Willingham out of prison with that report; we were just looking for a 30-day stay of execution to give the state time to check with their own experts, to bring someone else in for their own verification. But I was not optimistic at the time. I've dealt with this justice system many, many times, and usually a work such as I did will fall on deaf ears, and that's what happened in this case.

[... Can you just walk me through the arson investigator's initial findings and the inconsistencies and inaccuracies and scientific untruths? ...]

Let me begin by saying in 1990, there was a sort of landmark fire investigation called the Lime Street fire. There had been a fire in the duplex, and that fire left patterns on the floor that looked to the investigators as though gasoline had been poured on the floor. There was some other ignitable liquid -- a typical arson, in other words. The investigators in that case were John DeHaan and John Lentini. Both of these gentlemen thought it was an arson fire because they were seeing the typical signs of arson that the entire industry was used to looking at.

The district attorney was a bit unhappy with the case for other reasons, and they managed to find a matching duplex that was available. They bought the duplex. They furnished it with furniture identical to the original fire, and they set a fire in the same point of origin that was in the original fire, but using no accelerant, nothing on the floor. They let the fire burn until it had reached about the same level of destruction that had been reached in the original fire, and the fire department extinguished it. When they went in and looked at the fire patterns, especially those on the floor, they were, for all practical purposes, the dead match with the original scene.

This was an epiphany for DeHaan and Lentini, especially for Mr. Lentini. He realized that for years, he and other investigators had been misidentifying patterns on the floor that were a natural consequence of accidental fires being mistaken for accelerated fires with gasoline or some other ignitable liquid poured on the floor.

In 1992, John Lentini wrote a very brief paper [PDF]on the Lime Street fire, and it was published in a journal that is pretty much read by the entire industry. So by 1992, it was common knowledge among people who were current with the [state of arson investigation] … that a phenomenon known as flashover and post-flashover burning could produce patterns that were usually mistaken as pour patterns -- that is, patterns caused by pouring accelerant on the floor and igniting it. Like a lot of new information, it was not absorbed by the fire investigation community in general. ...

Taking a look at the photographs and video and testimony and investigation report[s], it became apparent we were dealing with a fire which had gone to flashover, as it had in the Lime Street fire, and [had] burned for a longer period of time than the Lime Street fire. ... Flashover had left natural patterns on the floor that all post-flashover fires tend to leave behind, and these had been misidentified as pour patterns, and thus the fire had been labeled an arson.

Define flashover.

Flashover is a phenomenon which occurs when a fire has burned intensely in a room -- a local fire, maybe a couch or a bed on fire. That fire produced a hot cloud of smoke, and because the gases are hot, they have a low density, and they bank down against the ceiling. So you get a two-layer structure: a smoky layer from the ceiling down about two-thirds of the way down, and then clear air underneath it. When that cloud reaches a temperature of about 550 degrees Celsius, the radiation from the cloud becomes so intense that it reaches a stage where it sets everything in the room on fire simultaneously. Everything that's combustible bursts into flames, and areas it would not normally burn in the self-sustaining fashion, such as a carpeted floor or a wooden floor, begin to burn. When that happens, you get a surge in temperature. The radiation associated with that surge becomes astronomical, and it very quickly begins to burn patterns into the floor that look like accelerant pours.

If that fire burns longer than even two minutes, it is highly unlikely that you would be able to see, even if gasoline had been poured on the floor, that you would be able to discern the patterns. So when you find a fire like that, the only way you're going to know that an accelerant was used is by getting a positive laboratory analysis for an accelerant from the floor or from the furniture in that room.

Since they had no such evidence in this case, they had no evidence of arson. That was the major mistake, but in addition to that, I found that they had employed a whole series of old wives' tales in finding, they claimed, 20 indicators of arson. Quite honestly, in any fire, nobody ever finds 20 indicators of arson. Even in a real arson fire, you might find two indicators of arson, but 20 is absurd. These were things like the threshold on the front door was aluminum; it was melted. They attributed that to gasoline ... or some other accelerant being poured across the floor. But we know that even if you pour gasoline or kerosene or what have you across an aluminum threshold and ignite it, it will not melt the threshold. ... [The] two things that can melt it are the radiation from post-flashover burning, or fall down of wood from a ceiling. Embers can bank on it.

They found crazed glass, this glass that has a spiderweb pattern, finely divided glass, which they attributed to the rapid buildup of heat by a liquid accelerant, but it's actually not caused by liquid accelerants. ... It's caused by the chilling effect of the water from the fire hoses on the hot glass.

They found brown stains on the concrete, which they attributed to charcoal lighter fluid being poured in that area, but charcoal lighter fluid doesn't leave brown stains. Brown stains are usually rust stains or they are the residue of wood smoke, which can be brown. If you burn a piece of wood against concrete, it will leave a brown stain. If you burn a typical liquid accelerant on concrete, the most it will do is leave perhaps a little soot ring around it, a dead black soot ring. ...

It's necessary to eliminate accidental causes before you declare a fire to be arson. They had an obligation to eliminate electrical causes, ... but they didn't. What they did was look around, didn't see any obvious electrical cause and say, "We have eliminated electrical causes." They said there were not electrical faults in the room. That is incorrect. Although I've never seen an electrical fault in there and none show in the pictures, it is perfectly obvious there had to be electrical faults in that room, in the bedroom, and the wiring in that house was surface wiring. That is, it wasn't run in the walls; it was run on the front surface of the walls. The insulation was burned off. All of that wiring, which was highly twisted and contorted -- when you have a live electrical wire and you burn the insulation, it will short.

So somewhere in that room there was a short, even if it was caused by the fire, but since they did not find the short, it means they did not look. They simply scanned it and said they didn't find any electrical faults, but there would have been an electrical fault in there.

The fire could not have started as they said, on the front porch, where they claimed that the charcoal lighter fluid had been poured, because you can't ignite charcoal lighter fluid on a concrete floor. It will not continue to burn. It will simply back away from the match when you put it down. The concrete sucks all the heat away, so the fire did not start outside.

So the patterns of that fire are typical of exactly what you would expect if you would start a fire in the bedroom -- comes out into the hallway, goes out the front door. You would get exactly what you see in that fire: heavy charring on the floor and in the room of origin, charring down the middle of the hallway floor, melted threshold, and then charring to the outside of the building because you have a covered porch. The hot gases come out, are burning fiercely; they set the wood on fire up above; this radiates down on the front of the building, and you wind up with burning on the outside as well. It is, I would say, a classical example of that sort of fire. And one could point to numerous other post-flashover fires where you see exactly that same set of indicia.

One other quasi-indicator of arson classically has been the V pattern on walls. A poor fire investigator in the old days and even today frequently would look at a pattern in the shape of a V against a wall and say the apex points to the origin. In post-flashover burning, everything in the room is set on fire within a period of a few seconds, several seconds. Anytime you have anything that will burn in the vicinity of the wall, it [catches] on fire and burns very intensely because it is reinforced by the radiation from the hot smoke cloud, and that leaves a V pattern. Typically, in a structure such as the Willingham house, it would be surprising if you didn't find at least half a dozen and maybe 20 V patterns in there. They found, for instance, a V pattern on the doorway going out of the room of origin into the hallway, a very clear V coming off of the sides of the door. All that is, is doors are framed in wood; hot gases come out; they bank against the ceiling; the wood is on fire, and even without the wood, you get a V pattern from gases coming out into the hallway. So those V patterns are just a natural consequence.

There were three beds in the room. If you looked at a wall carefully, you would find three V patterns associated with the beds. If there was a chest or a stool or anything else there, you would find a V pattern associated with that item. The other thing is, those who are aware of post-flashover burning always assume that because it was caused by radiation from a hot gas cloud that the damage would be uniform to the floor. It is never uniform. At the very beginning of the post-flashover, yes, it's fairly uniform as things ignite, but then they burn according to their nature. And those things that are close to other fuel supplies burn more intensely because they also get radiation from the local fires that are burning in the midst of this holocaust. So you wind up with very complex patterns, and it's a lot like a Rorschach pattern.

A fire investigator who is looking for arson will always be able to find numerous indicia of arson there because they're not taking into account the natural patterns that are left. Although the Lime Street fire occurred in 1990, and although it was published in 1992, today I still get cases coming across my desk every month in which they have used the same techniques and continue to use the same techniques that we used in Willingham. The number of these bogus arson cases has dropped in recent years, but it is still very, very high, and there are areas of the country where the information about these things simply has not been able to infiltrate the system there. ...

How did the Willingham fire start? ... Your educated guess?

I don't know. I mean, you can take your choice: an electrical fire, a kid fire. I don't know how it started. But what I do know 100 percent is that there is not a single bit of evidence that this was an incendiary fire, that it was started by human hands.

I'm curious. A question that gets begged: As much as I would say perhaps that investigators are trained to look for arson, you also are trained perhaps to look for ways that this is not arson. ...

Well, you have to look at how I look at my role. My role, if I take a case, is to see that the defendant gets a fair trial; that all the cards are put on the table. I get a lot of cases -- I have several over here -- which really are arson, and they really are murder, and then my role is to convince the attorney or the defendant to cop a plea and take life in prison without possibility of parole to avoid a trial in which he will wind up with the death penalty, or if it's just a minor arson, to cop a plea and get 10 years instead of 30. ... I always assume that a fire is accidental and then study the case and let the evidence overwhelm me. Sometimes it takes about 15 minutes for the evidence to overwhelm you. Sometimes it takes a month of studying trial transcripts. ...

Cases fall into three clear categories: those people who are clearly innocent, those people who are clearly guilty, and those cases where you just don't know and [a] jury is going to have to decide it. Those middle cases are very troublesome, so you look at a case like that and you make sure that you keep the prosecution honest, so the jury knows all the facts and isn't subject to junk science, and you turn it over to the jury because that's what juries are for. ...

We experts, our job really should be to intersect these cases up front and see that people who are clearly innocent don't get indicted in the first place and take that roll of the dice that happens once they get into court, because once a case goes to court, no matter how bad the case is, the defendant may lose it, simply because the prosecution gets an investment in time and energy, and after a couple of years, they don't want to lose that case. Their careers are involved in this. So if you can stop them early, that's the ideal way to do it.

The far end of the spectrum, and the cases I get more often than any other kind, are post-conviction cases, and these cases, you'd rather have your teeth pulled out without anesthetic than work on those cases, but they are numerous. That's the people who have been convicted by the system and stowed away in a prison, and you don't see the case until 10 or 20 years later, and you go through the case and you say, "Oh, my God, this is a totally bogus case," and you have to try and figure out a way to get the attention of the justice system. That can take years and years and years. I have cases that I have been working on for 10 years. The last time I got someone out of prison, it took me seven years -- seven years to get a hearing in front of a judge and then two weeks to get him set free.

The Willingham case is not one you can pass on. ...

Nobody would have taken that case. Nobody would have wanted that case. ... I couldn't have given that one away. I couldn't have paid anybody to take that case. … The Willingham case is one that looked really bad on the surface. In fact, I think in my first e-mail back, I wrote, "It's going to be pretty hard to get any sympathy for this guy." But then when I actually got the thing and looked at it, I said, oh, he might not be a particularly nice guy, but this is really a railroad job.

You have found cases where people are guilty of arson?

I have two of them this week. It's that simple, right. I've already contacted the lawyers, ... [and] I've told the attorneys: "You don't want me to testify in this case; you don't want me to write a report. Tell your defendant to cop a plea, because he's going down." ...

So please tell me the difference between cases that are absolutely guilty, murky or innocent. Tell me where Willingham's case falls into those three. ...

I don't usually use the word "innocent," but Todd Willingham's case falls into that category where there is not one iota of evidence that the fire was arson, not one iota, and much of what they considered to be evidence is not evidence at all. This case started with the fire investigation report and the misunderstanding of all these indicia that you look for in determining arson -- bad interpretation of the evidence. But then it went to the second stage, which is typical in all cases, demonization of the defendant, to make the jury not like him or her and then a full-court press to make the jury hate him during trial, and by introducing evidence of the behavior of the witness during the fire and after the fire. ... They literally expect people whose houses are burning down and whose children are dying, they think that there is a proper way to behave at that time. And of all the cases that I've ever had, I have never had a single defendant behave the way he should have behaved during and after a fire. ... Every one of them has misbehaved. Never fails.

In Todd Willingham's case, he was a heavy metal fan, and they used that against him. ... Heavy metal fan became a Satanist. The prosecutor in this case literally believed that the burn patterns on the floor were in the shape of a pentagram, like some satanic ritual. When you actually look at the burn pattern that they drew and then you look at where the windows are, windows furnish ventilation to the fire, and all they were looking at is what we call ventilation patterns.

As the fresh air pours in the five windows around there, it alters the burn pattern there, because things can't burn without fresh oxygen coming in. So you get an irregular pattern. This pattern looked like a pentagram to the man who wanted to see the work of a Satanist. It looked like an "X" to the snitch and to the fire investigators as if it had been poured in the form of an "X."

The ATF [Bureau of Alcohol, Tobacco, Firearms and Explosives] now, in their training, they take pyre investigators in -- some of them are novice fire investigators, and some of them are very experienced fire investigators -- and they offer a course, and during this course, they start test fires in a simple room with simple furnishings in it. They start a fire in one quadrant of the room. They let it burn into post-flashover, and after it has burned in full involvement mode -- that's post-flashover for two minutes -- they extinguish the fire. Then they parade these inexperienced and experienced fire investigators in, and they have to look around. ... It's a very simple scene, so it doesn't take long. And [they] read the patterns, and they are asked to identify the quadrant in the room in which the fire started. ... Ninety-five percent of fire investigators get it wrong today. ... Simple furnishings and everything that was in the room, it's all there to look at. Imagine today when they walk into what is more or less a black hole, a really burned-out room, these same 100 fire investigators go into a fire scene like that, they're going to get it wrong 95 percent of the time, more than 95 percent, because real fires burn longer than two minutes, and the longer it burns, the harder it gets. That's how bad it still is.

And why is that?

Because they don't want to know. ... This is a money-driven system. If you look at the history of fire investigation, it doesn't come out of the same roots that the other forensic sciences come from. Medical jurisprudence was the basis of modern forensic science. When you try to trace the history of fire investigation back, the closest things you can find to fire investigation if you go back more than 60, 70 years, is witch hunts, because witch hunts used old wives' tales; fire investigators use old wives' tales. They use folklore that has been handed down. Since about 1992, thereabouts, ... that was the first time science was ever really introduced into the mainstream of fire investigation.

The International Association of Arson Investigators spent the next seven or eight years trying to find ways to get around this landmark publication. ... Everybody belonged to this organization -- fire investigators did -- [and] their thesis was that they shouldn't be subject to court rules like the Daubert rule [from Daubert v. Merrell Dow Pharmaceuticals in 1993] that came out at about the same time, the rule that says if you testify in court, you have to have a reasonable, logical, rational basis for your testimony. See, in the old days, you got on the witness stand and said: "Look at me. I've got 20 years of experience, so here's my opinion." When Daubert came in and an NFPA 921 [Guide for Fire and Explosion Investigations] came in at about the same time, the courts and the NFPA [National Fire Protection Association] said no, that's not good enough; you need a basis -- call it a scientific basis, a rational basis. In other words, you need a better reason than your own experience for declaring a fire to be arson. They hated this rule. They were training investigators how to get around Daubert, how to get around NFPA 921.

They had filed an amicus brief saying that fire investigators shouldn't be subject to these rules that other forensic sciences are subject to because fire investigation was an art, not a science, and therefore their experience was good enough. That's what artists have; they have experience. ... This continued until a landmark case, [Kumho Tire Co. v.] Carmichael, [in] 1999. ...

The minds of these fire investigators became so caudated with this empty-science mode that you've got people with 30-year careers, and we still have those people out there, and they're still passing their old folklore down. ...

Things are getting better; there's no doubt about it. If you look at the statistics here in Texas, arson cases, we have the same number of fires that we had 15 years ago. We have 60 percent fewer arson cases. But we've still got a lot of bogus arson cases, and this is not a Texas phenomenon. This is virtually universal all across the United States. ...

The timeline and how it all went down for him [Willingham], ... can you lay that out for me?

Well, my understanding is, of course it was out of my hands when I transferred it to [appellate lawyer] Walter Reaves. But Walter Reaves sent it to, I think, the Texas Board of Pardons and Paroles -- Texas parole board -- and it was subsequently forwarded to the governor's office with their recommendation. I think a judge also looked at it. But you have to understand, the chances of anything like that working are usually more optimistic than I was. The report is solid. You could show it to any number of the top fire analysts in the country, and they all would just bob their heads -- which they have, by the way. John Lentini et al's report is a much longer version of mine because of the same key elements, and the Beyler report [PDF] [by arson expert Craig Beyler] is even longer, but it has the same key elements. We've all seen the same thing in it. ... And fundamentally, this was a classic accidental fire. If Willingham had poured gasoline in there or lighter fluid in there, they would have found it. In all probability they would have found it in the lab. There's nothing in there. …

Look at it as they looked at the Willingham fire, and you will find the same evidence that they found. You have no reason to believe it was arson unless you have to say Willingham was a really bad guy, and because he was a bad guy he must be guilty.

Can you briefly compare the case of [Ernest Ray] Willis to Willingham?

It's very instructive to compare the Willingham case with the Willis case. Both of these cases were arson-murder cases. The fires occurred in roughly the same time frame; i.e., 1988 to 1990. They both occurred in Texas. They were both fires that went into post-flashover burning, prolonged in duration. They both had almost exactly the same elements of evidence -- that is, burn patterns on the floor -- similar flooring, by the way; damage to the threshold; stains on the porch; some indication of an accelerant on the porch. They were both brought to me in roughly the same time frame, within nine months of each other. The difference between the two cases was that in the case of Willis, the prosecutor was not the same man who had originally prosecuted the case, and he was unhappy with the case because Willis had been drugged during the trial, had been overdosed with his own medications, and he wanted to know whether he should retry the case or not.

When he brought me the case, he also brought me the help of the attorney general's office. So, with their cooperation, I was able to get all of the information I needed for a thorough case study. I was able to spend a month on the case, and I found that the Willingham and Willis cases were twins. They were eerily similar -- had the same elements of evidence. I was able to write a very thorough and long report, and because I had the cooperation of the district attorney, we were able to get Willis off of death row and out of prison, and he's off somewhere driving a truck now. In the case of Cameron Todd Willingham, we had the same elements of evidence, but we had an ex-prosecutor and a governor and a parole board and people who didn't want to hear it, and they executed him.

It was like a roulette wheel. It was not an iota of difference between the two cases. Nothing. No important difference between the cases. One man is set free, and one man is executed. It shows you the randomness that our judicial system sometimes has built into it.

Did Texas execute an innocent man?

Yes, and I say that without hesitation. I say it without hesitation because our rule is really very simple in American jury prisons, isn't it? You're innocent until you're proven guilty. Now, when you put a bunch of junk in front of a jury, junk science, and they find a person guilty primarily on that basis that you have told them it was an arson fire, and you know it was arson because you have all of this experience and you can read all those fire patterns, and that evidence, you haven't proved that he's guilty. What you've done is conned [the] jury into believing he's guilty. So yeah, from that standpoint, yes, he was innocent, because you're innocent until you're proven guilty. ...

Cameron Todd Willingham was not proved guilty. And if innocent until proven guilty has real meaning in our society, on that basis he was innocent, because nobody ever proved he was guilty. ... Yes, Texas executed an innocent man.

Give me some examples of your other clients who have exhibited, ... like your story of the schoolteacher. ... What did she do after the fire that people thought: "Oh, that's bad behavior; that's not correct; ... that's not how one would act if she were truly grieving"? ...

The Sunday school teacher [Sheila Bryan], ... she was ... a sweet little old church lady, about 50 years old. She was out for a drive with her mother. ... Something went wrong; she became distracted and ran off into a bar ditch. She got out of the car, with her mother still sitting in the car, and found that she couldn't open the door to get back in, climbed up the embankment, flagged down a passerby. She walked over a bridge, flagged down a passerby, and when they looked back, the car was on fire. Her mother died in that fire, but she had no carbon monoxide in her. In other words, the woman had already passed away before the fire began.

They demonized this woman unbelievably. [Said] she didn't take care of her mother. One day the old woman wandered out to the road and fell and was lying there and a passerby came by, picked her up and put her back in the house, and that became evidence that she didn't care about her mother. The motive in the case was that the old woman was getting old, so she had become incontinent, and the daughter was tired of changing her diapers. This just went on and on and on, including publishing a photograph of her that made her look like an ax murderess.

… When you first looked at the case, it looked like this horrible woman, until I went around and talked to all her relatives, everybody involved in the church group and so on and so forth, figuring out what kind of relationship she had with her mother. And what I found, contrary to all this demonization, is that they were buddies, right? These people had taken exquisite care of Grandma. When she was getting too old to take care of herself, they literally picked her house up, moved it onto their property, built a path to the main house between the two so that the grandchildren could go up to have tea with the grandmother and picked up every one of her plants and transplanted them into the original configuration around her house. This was a loving family relationship.

The demonization was so intense that the town became polarized, people who didn't know her between people who knew her, which was basically her church group ... and people who read the newspapers and press releases from the prosecution. So by the time it went to trial, … they found her guilty, and they put her in prison. She was there 11 months. Then I got in on the case, and I began calling Georgia and asking lawyers, "Who is the best defense lawyer in south Georgia?" And [I] mounted a campaign to get this guy to take the case. Took him six months, and the church was running bazaars, and they were collecting money for the defense. ...

We got a second trial, and I managed to get 48 Hours to come in and put their cameras in the courthouse, because there had been a lot of technical lying that went on in the first trial. And I reckon that if we could point the cameras at the prosecution experts, they would sing a different tune. And it worked. ... We went in there, and they got their clocks cleaned the second time, and she was set free. ...

When the police interview people, people want to please the police. So, with a little encouragement from the cops who are giving them "attaboys" for telling them what's going on, the stories become a little bit altered, and pretty soon you see that the person totally misbehaved after the fire.

... I've never had a witness that behaved right. They all looked like they were having a good time during or after the fire. They all looked like they could[n't] care less what happened. But think about it: If you were an arsonist, you deliberately set a fire to kill people, you would behave normally then. You would express the right amount of grief, because if you don't have grief, it's much easier to regulate yourself. But when you're in a state of shock because your family's dying, nobody knows how they'll behave under those circumstances. ...

... What has proved for you that certain clients are absolutely guilty versus murky, versus no forensic evidence? ...

Well, obviously the key is the forensic evidence. I mean, that is the rock-solid stay. But the other thing you look at is the nature of the defendant and the relationship of that defendant with the deceased. Is there really a motive here? There's always a concocted motive.

I'll give you an example. David Mann from The Texas Observer was here interviewing me about one of my cases, and I told him, "If you'll bring me three arson cases, three people who are in prison, I'll show you at least one bogus arson case." So he went down to Houston to look at a case. He only brought me one case. ... He came back with that stack of documents over there. He said: "I got one. This guy was a gang member. ... They torched the place, killed some people," and so on and so forth. So I began looking into photographs and the write-up and so on and so forth. ... In this case, demonization is really easy because he is a "gang member."

It turns out when I looked into the case a little deeper, he wasn't really a gang member. He was a Hispanic kid with an IQ of around 70 or 75 who got hooked on drugs while he was in Germany with the military, and he became a peripheral guy around the gangs because that's where he got his fixes, so he was an errand boy for them, not a gang member. And his involvement in this fire was that the same house had been burglarized the day before, and he was the guy that had given the burglars a ride to the pawn shop to pawn what they had swiped out of the house for $100. ...

... [Mann] ended up writing that case up and it's in the hands of the Innocence Project now. ...

Sounds difficult to use forensic evidence to actually prove arson.

It is difficult, but not difficult to get convictions. That's the irony. Yes, the actual proof of arson is sometimes very difficult; sometimes it's not. These cases I've got here, for instance, you go in, you find gasoline around the edges of a burn pattern, and there's no gasoline anywhere else. You take blank samples around it. You know gasoline was poured there if you've got a burn pattern and gasoline. Sometimes you have eyewitness testimony. Sometimes you really eliminate all possible causes. You can really do it, but it takes a lot of effort. Yes, it's very difficult to prove arson -- difficult to prove arson, easy to get convictions. ...

There's only one standard in the end that you can use: You actually have evidence of arson, or you don't have evidence of arson. Without evidence, then you're basing it on the character of the person, or their looks, or the ambitions of a district attorney. See, arson is a win-win for a prosecutor. Fire marshals get "attaboy" and promotions. The DA … gets a death penalty for this horrible person. The insurance company doesn't want to pay off, so the husband is accused of killing the wife, or the wife is accused of torching the place and killing the husband. If it is a torch job, they don't have to pay off. ... The insurance company's investigators find typically anywhere from 25 to 60 percent arson cases. National stats are like 14 percent of fires are arson. We don't really know. ... But we can always find two, three, four times as many arson cases as you can expect from whatever the statistics are. …

To tell you the truth, playing devil's advocate, I could take almost any fire and -- if I were so inclined -- convince a jury that it was arson. It's frighteningly simple, frighteningly easy. So what I try to do is make the prosecution prove the case, keep them honest. ...

How could you live with yourself if you let a guilty murderer free?

You can use the 10-1 principle, right? Better to let 10 guilty people go than convict one innocent person. But, having said that, I defy anybody to find a case where I've gotten somebody off where you can show me some evidence of arson, or any really good reason to believe it was. ... My reports are published somewhere; you can find them on the Internet somewhere. I have made them public, and what I find interesting is all the years I've been doing this, ... I bet you can't find anybody who thinks any of my people were guilty among fire investigators now, although the DAs that convicted them all think that they are guilty now, of course. ... I've never had anybody write me and say: "You can't do this. This is wrong in your report."

[Is there anything else that you would like to add?]

... I recognize that it's very difficult when you listen to me talk about evidence, about forensic evidence. But that is actually a rule. If you cannot establish that arson was perpetrated, the fire investigator has no right to declare it to be arson. Now, don't misunderstand me here. You can take a fire case in which you have no evidence of arson, and you can put it in front of a jury as an arson case. You can say, "We believe that he murdered his family," .... and the jury can decide that he's guilty. The rule that I'm talking about not breaking is to have an expert get up and tell a jury that he knows that it's arson because he has found the forensic evidence. That man should get on the stand and say, "Well, I went to the fire scene, but it was a black hole, so the cause of the fire is undetermined." That would be the truth for the jury. Then they can present whatever peripheral evidence they have. Maybe threats were made. Maybe there was a motive they want to tell the jury about. Maybe the jury doesn't like his face. Juries have a lot of power.

But the thing that drives these cases is that initial fire investigation report and that fire investigator either getting on the witness stand and saying, "I know; I can tell because I've got 20 years' experience in this." When he can't tell, when he doesn't know, and he doesn't have any evidence, you might as well put the wizard of Oz on the stand and have him tell some carny blah-blah on the jury. And there the system is really broken, and it's got to be fixed. And the best way to fix it, by the way, is to carry on looking at the Willingham case. ...

They need to let the Texas state fire marshals answer that in public so that all the other fire investigators can see what they're saying about this fire. And we need an official stamp of approval on the fact that there was no evidence of arson in the Willingham case. When that happens, then you will have a chance to go review 200 to 300 cases in Texas alone of people who are sitting in prison for similar fires. That's a big undertaking, and nobody wants to see this happen in government. ...

I can guarantee you that we've got at least a couple hundred people in prison in this state alone for accidental fires, and we need to get them out. And the only way we can do it is to start looking at their cases. But it won't do any good to look at their cases unless we have an ear in the government that we can speak into.