Monday, April 22, 2013

Is the American Criminal Justice System Incapable of Adopting Valid Scent ID Procedures?

Additional Note: Important research by a group of French researchers indicates that rigorous training procedures may be necessary for scent lineups to be effective forensic tools.  Marchal, S., Bregeras, O., Puaux, D., Gervais, R., and Ferry, B. (2016).  Rigorous Training of Dogs Leads to High Accuracy in Human Scent Matching-To-Sample Performance. PLoS ONE 11(2): e0146963. In a forthcoming book on canine olfaction, Tadeusz Jezierski and I both discuss deficiencies in many scent identification procedures. 

Scent identifications result from various procedures, including station identifications and scent lineups, the former of which has become a sort of odd forensic procedure in southern California and the latter of which has apparently always been structured without regard to forensic research throughout the United States.  Having recently appeared as a witness regarding scent identification in a European trial, I also begin to despair of the ability of the American judiciary to calculate the forensic value of scent identification evidence.  Two recent cases, one from California, one from Texas, illustrate why I believe there is reason to despair. 

American law enforcement has almost insisted on performing poorly structured scent identifications, with the possible exception of agencies (like the FBI) that prefer to confine scent lineups to the investigatory stages of their treatment of a crime and avoid using scent-lineup results during prosecutorial stages.   In other words, neither American law enforcement nor American courts have adopted standards to assure that scent identifications have a reasonable diagnostic ratio (which could place such procedures in an intermediate category for forensic value, as I and two scientists have discussed elsewhere).  This means that the risk of misidentification in scent identification procedures that have been described by U.S. courts is often high, and even those judges that know or suspect this only allow such procedures to remain on the record by deeming their admission “harmless error.”

J.E.R. Frijters and J. Boksem, a scientist and a lawyer, described Dutch scent lineup procedures as akin to Russian roulette (“Russische roulette”).  Yet those procedures were better designed than the procedures described in the cases here.  The history behind the exclusion of such evidence in the Netherlands was discussed extensively by Resi Gerritsen and Ruud Haak in a book previously reviewed here.  The problem is that if Dutch courts were playing Russian roulette with a pistol that had one bullet in a chamber that can hold six, U.S. courts are using a pistol with a chamber that has three or four bullets.  Even Nick, the character portrayed by Christopher Walken in The Deer Hunter, was not taking that much risk. 

Murder at a Pasadena Nightclub (California v. Reynolds)

Ebony Huel was shot and killed in a crowd outside the Underground Club in Pasadena on August 17, 2007.  Johnl Dvon Reynolds was charged with the crime.  The prosecution’s theory was that Reynolds was attempting to kill Sean Quintero, who had implicated him in a robbery that occurred in 2005.  Reynolds was on parole at the time of the murder. 

Earlier on August 17, Reynolds had showed Vanessa Calderon a handgun with a yellow or tan barrel.  At the time he was wearing a beanie and a red Chicago Bulls jersey.  Later that day, Leeban Adan drove Calderon and Reynolds to the Underground Club.  Calderon was waiting near the car later when she heard a gunshot and Reynolds and Adan ran towards her.  When they got in the car, either Reynolds or Adan tossed a gun into her lap, the same gun she had seen that morning.  Calderon threw the gun out the window.  Reynolds told her to get it back and she left the car to look for the gun but she did not find it. 

Huel died from the gunshot wound to her head.  A search of the area produced a red beanie and a yellow, four-barreled handgun.  Reynolds was wearing a Jordan Chicago Bulls jersey when he was arrested the day after the shooting. 

“Reynolds was placed with Calderon in an interview room that was electronically monitored. Detective Max Dahlstein testified he overheard Reynolds instructing Calderon to tell the police the handgun had accidentally discharged when Reynolds attempted to grab it away to prevent her from shooting ‘Shady Stupid’ (Quintero) for snitching on Reynolds. Reynolds explained to Calderon that, unlike her, he could be imprisoned for life. Calderon agreed to 'take the rap' for Reynolds. She subsequently told police that she had been playing with the handgun when it had accidentally went off. Calderon later recanted this admission.”

Adan did not testify because he was killed before Reynolds’ trial. 

Scent Transfer Unit and Station Identifications

The scent identification evidence began with the preparation of scent pads using a scent transfer unit:

“Michael Wooldridge, a canine handler for the Long Beach Police Department, went to the shooting scene the night of the murder. Using a special device called a scent transfer unit, which is similar to a small, portable vacuum cleaner, Wooldridge transferred scent from a beanie and a handgun he was given to separate sterile gauze pads, making three scent pads for each item.”

The trial court held a pretrial hearing under California Evidence Code § 402 to determine the admissibility of dog-scent identification testimony because “no published decision has yet affirmed under People v. Kelly [17 Cal.3d 24 (1976)] that the scientific community has generally accepted use of a scent transfer unit.”  The trial court held that the evidence was admissible. 

The actual identification of Reynolds by a dog was what is generally called a station identification and is described by the court as follows:

“Three days after the shooting Edward (Ted) Hamm, a scent hound handler for the Los Angeles County Sheriff's Department, and his dog Bojangles, a trained scent hound, went to the police station where Reynolds was being held. Hamm had Bojangles sniff one of the handgun scent pads prepared by Officer Wooldridge. Bojangles then led Hamm through the first floor of the station to the room in which Reynolds was being held and ‘alerted’ to Reynolds by walking up to him. Officers moved Reynolds to a room on the second floor of the police station, while Hamm had Bojangles sniff one of the beanie scent pads prepared by Wooldridge. Bojangles again led officers to the room where Reynolds was being held and alerted to Reynolds.”

Hamm’s testimony at the trial is summarized by the appellate court as follows:

“Hamm initially explained his general approach to training dogs to trail or follow scents. He then described the training Bojangles had received, specifically focusing on the dog's training in different environments including police stations (station identifications). Hamm testified, at the time of this proceeding, five of Bojangles's station identifications for the Pasadena Police Department had been confirmed. Hamm then described the station identifications conducted at the Pasadena police station on August 20, 2007 in which Bojangles twice followed a trail that led to Reynolds, who had been placed in different parts of the station, each time after the dog had sniffed a different scent pad.”

The trial court also took notice of earlier cases regarding Hamm’s use of Bojangles (California v. Salcido (Super. Ct. L.A. County, 2005, No. GA052057) and California v. Elias et al. (Super. Ct. L.A. County, 2010.No. GA069722).   Defense counsel objected that the prior unpublished cases did not establish that the STU was generally accepted in the scientific community. 

“Counsel also argued the results of Bojangles's scent discrimination and trailing were ambiguous. Rather than place his paw on the person he has trailed, the dog was trained to stand by the person, which could mean the dog was simply waiting for the next command rather than signaling he had matched the scent to the person. Counsel additionally argued it had not been shown that Bojangles's initial scent identification of Reynolds had not influenced the dog's second scent identification of him. In this regard counsel asserted, ‘But it seems to me that we would have to rule out the dog's ability to remember the scent he had just done a trail on when he gets a new one and trails my client again.’”

As to the use of the scent transfer unit, following a hearing in an earlier trial, “the trial court tentatively ruled that dog-scent identification evidence obtained through the use of an STU was generally accepted by the scientific community.”   

Testimony of Forensic Psychologist

Dr. Nancy Kaser-Boyd, a clinical and forensic psychologist, testified that Reynolds had PTSD.  In an interview, Reynolds told her that Calderon was attempting to shoot Quintero for having sent Reynolds to prison.  Reynolds said he tried to stop Calderon and the gun discharged accidentally.  When Kaser-Boyd said no one saw a girl with a gun, Reynolds said “Okay, I was the one with the gun.” 


The jury convicted Reynolds of first degree murder by a person with gang affiliations.  Reynolds was sentenced to a term of 25 years to life for first degree murder, which was doubled under the Three Strikes law, with another 25 years added for a firearm-use enhancement. 

Dog-Scent Identification Challenged on Appeal

Admission of evidence that a dog trailed or tracked a suspect from a crime scene has been upheld if an adequate foundation is provided, i.e., that the dog was properly trained in tracking humans and the handler was qualified to use the dog (citing California v. Craig, 86 Cal.App.3d 905 (Ct. App. 1978); California v. Malgren, 139 Cal.App.3d 234 (Ct. App. 1983)).  (For a discussion of the different standards for admission of scientific evidence under state and federal decisions and how these standards have been applied in scent identification cases, see Police and Military Dogs, 62-63.) 

The appellate court held that Reynolds failed “to provide an adequate record on appeal” to question the dog-scent evidence.  This means that the appellate court could not review the evidentiary basis for the trial court’s decision” regarding the admission of the evidence derived from the use of the STU.  In any case, any error that might have occurred by admitting the dog-scent evidence was, according to the court, harmless. 

“There was overwhelming evidence Reynolds was the individual who fired the shot outside the Underground Club, killing Huel. The evidence at trial established a clear motive for the murder: Reynolds, an admitted gang member, was angry with Quintero for having informed the police about Reynolds's involvement in a robbery, which targeted Quintero for retaliation by Reynolds. Two eye witnesses, Adan and Bonwell, identified Reynolds as the man they saw extend his arm towards Quintero, followed by a gunshot. Before the shooting Calderon saw Reynolds in possession of a distinctive handgun, which Reynolds demanded she retrieve after she had thrown it out the car window following the shooting. Reynolds also made highly damaging, incriminating statements: In urging Calderon to claim responsibility for the shooting, Reynolds said he was facing a life sentence. He also admitted to Dr. Kaser–Boyd that he, not Calderon, held the gun; and he conceded he was angry with Quintero for having betrayed him to the police. Based on this largely undisputed evidence, admission of the dog-scent identification evidence could not have prejudiced Reynolds.”

The judgment of the trial court was affirmed. 


Station identifications, a term derived from the location where such procedures are generally conducted, are a flawed forensic procedure.  When they occur by accident, such as when a dog loses a trail and is brought back to a police station where, by chance, the person being trailed has arrived by other means, the dog’s alert provides useful evidence.  It is a variation of a trail being lost at one point and resumed at another. 

When station identifications are structured as was done in this case, they lack the cautions that can be built into a scent lineup.  It would be as if visual lineups of eyewitnesses to crimes were to be conducted by putting the witnesses in parks near the courthouse and having various suspects come near the bench on which the witness in random order, along with whoever else happened to pass by.  There are too many aspects that cannot be controlled. 

The identification of a real person in a scent lineup, as opposed to a scent (in a procedure sometimes just called a “scent match”) increases the possibility of cueing.  It is likely the defendant is dressed differently from most people in the station, likely that he is not working the way many other people are. Where did the tracking inside the station start?  Did the dog have to go up several floors? How many people were in the room where the dog found him?  Were any of them sufficiently like the defendant that a real choice was available to the dog?  In an experimental environment near Phoenix, it was demonstrated that dogs can trail from an IED explosion to a room with potential suspects including the “perpetrator,” yet alert to a non-perpetrator despite the presence of the individual that they have presumably been tracking from the explosion site.  Numerous questions like this have to be answered before the procedure can be deemed objective, and it is doubtful that most of the flaws in such a haphazard method can be eliminated. 

The court may be correct that any error in introducing the station identification was harmless.  The non-canine evidence was overwhelming.  Nevertheless, California courts should be disturbed that scent evidence so often has to be deemed harmless to avoid overturning convictions as there will be cases where better procedures could provide solid evidence that could withstand challenges where other evidence needs less corroboration than was the case here.  California v. Reynolds, 2013 WL 604188 (Ct. App. 2013).

Suit for Deprivation of Rights (Curtis v. Anthony)

The federal district court for the Southern District of Texas granted summary judgment on claims by three men who were arrested, charged, and incarcerated as a result of two different sets of scent lineups conducted by one of the defendants, Keith Pikett.  The lineups are described by the Fifth Circuit as follows:

“To conduct his lineups, Pikett made use of scent-discriminating bloodhounds. First, Pikett would obtain a scent sample from the suspect under investigation by wiping the suspect with a sterile gauze pad. The gauze pad, containing the suspect's ‘human scent’ and ‘skin cells,’ would be stored in a Ziploc bag until the time of the lineup.  At the time of the lineup, a second officer would arrange six cans, one containing the suspect's scent pad and the other five containing scent pads from other persons of the same gender and race. The officer would arrange the cans approximately ten feet apart and positioned perpendicular to the wind so as to minimize the crossing of scents.  Thereafter, Pikett would expose a bloodhound to a scent sample taken from the crime scene. The trained bloodhound would ‘alert’ if the scent pad from any of the six cans matched the crime scene sample. Pikett would repeat the exercise with a second bloodhound to confirm the first bloodhound's alert.” 

Pikett was a deputy with the Fort Bend Sheriff’s Department but he volunteered his services to the Houston Police Department.  At the time of the lineups in 2007 to 2009, the Fifth Circuit notes that “Texas courts uniformly had accepted Pikett as an expert on dog-scent lineups.”

Texas Courts and Dog-Scent Lineups

The Fifth Circuit reviewed Texas decisions regarding scent lineups, but noted a trend towards increasing skepticism finally resulting in its own decision in Winfrey v. San Jacinto County, 481 Fed.Appx. 969 (5th Cir. 2012).  In that case, the Fifth Circuit itself reversed the summary judgment of a district court dismissing a suit against Pikett, “citing a factual dispute over whether a videotape of the dog-scent lineup at issue demonstrated that Pikett had manipulated his bloodhounds to cue false alerts during the lineups.”

The Fifth Circuit then reviewed the encounters of the three plaintiffs who were now suing Pikett and others.

Scent Lineup Case 1: Burglaries of T-Mobile Stores

In June 2007, Ronald Curtis was found near the location of a T-Mobile store that had been burglarized. He and a passenger in his car provided conflicting accounts to Houston Police Department officers, and various items that might be used in burglaries were found in the car, including a crowbar which had markings matching markings found on the back door of the store.  Mud was found at the crime scene, and Curtis and his passenger had muddy shoes when they were detained.  Both Curtis and the passenger were released by a magistrate who found insufficient probable cause to hold them. 

The investigation continued.  A still photo from a surveillance video at another T-Mobile store that was burglarized was thought by an investigator to match Curtis’s driver’s license photo.  The investigating detective approached Pikett to conduct a scent lineup.  Curtis refused to provide a voluntary scent sample but one was obtained by subpoena.  Concerning the scent lineups in this investigation, the court stated:

“Pikett used his dogs to compare Curtis's scent with scent samples taken from the three burglarized stores. Stivers obtained these scent samples weeks and, in some cases, months after the times of the burglaries. Accordingly, the burglarized stores had experienced routine cleanings and customer traffic between the times of the burglaries and the times that Stivers obtained the scent samples. Nevertheless, the dogs alerted to a match between each store's scent sample and Curtis's scent.”

Curtis was indicted and arrested, but a string of burglaries continued while he was in jail.  After eight months, Curtis was released and the charges were dropped without a trial.

Scent Lineup Case 2: Gang Style Murders in Houston

Three people were murdered by being shot in the head in a house in Houston in November 2007 and the perpetrators set the home on fire.  Houston Police Department officers found charred remnants of a gasoline can, a cigarette lighter, and two guns, as well as the remains of the victims.  The gas can was traced to a service station where the surveillance video recorded the purchasers.  One of them was identified as Cedric Johnson. 

Johnson provided a voluntary scent sample which was used as a comparison for four scent samples taken from items at the crime scene.  The dogs matched each item.  Johnson was arrested and charged with the murders, and a grand jury indicted him.  Johnson’s cell mate contacted the district attorney to say that Johnson had confessed to the shootings.  In speaking to the investigators, the cell mate relayed information known only to investigators. 

Johnson identified Curvis Bickham as his accomplice.  The investigators asked Pikett to conduct additional scent testing.  The scent samples from the murder scene were by then more than a year old.  Nevertheless, “the bloodhounds alerted to a match for three of the four items.”  Bickham was arrested, charged, and indicted for two of the murders. 

Then Samuels recanted his statements and confessed to perpetrating the murders himself.  In May 2009, Johnson and Bickham were released and the charges against them dropped.   Later, the Houston Police Department determined that Samuels could not have perpetrated the murders since he was in Boston at the time of the shootings.  A further dog-scent lineup b Pikett indicated Samuels’ scent did not match any of the items from the murder scene.  Samuels ultimately admitted that he had falsely confessed to the murders to stop threats and attacks made by gang members affiliated with Johnson. 

Section 1983 Action

Reynolds, Johnson, and Bickham sued, under 42 U.S.C. 1983, for deprivation of their rights. They alleged that Pikett had manipulated and misrepresented the results of his dog-scent lineups to manufacture fraudulent inculpatory evidence, that investigators had failed to intervene to prevent these frauds, or alternatively that the investigators had conspired with Pikett to manufacture fraudulent evidence, that Pikett’s supervisor had failed to train and supervise Pikett, and that the City of Houston and Fort Bend County had failed to establish policies to monitor use of dog-scent lineups for unreliability and fraud.

The district court was affirmed by the Fifth Circuit in its grant of summary judgment based on qualified immunity.  The Fifth Circuit cited its earlier language in Winfrey v. San Jacinto County, 2012 WL 3062159 (5th Cir. 2012), that the defendants had not been “objectively unreasonable in seeking Pikett’s assistance, and then using the resulting information as part of their investigation,” given that, at the time, “Pikett enjoyed a solid reputation.” Also, there was other evidence corroborating the results of the scent lineups, which the Fifth Circuit was sufficient to satisfy “the standard for probable cause even without the evidence from Pikett’s lineups.” 

In the Fifth Circuit’s Winfrey case (which was not published and therefore not binding precedent in the Circuit), the panel concluded that the plaintiff had raised a sufficient factual issue as to whether Pikett had cued his dogs during the scent lineup.  Although Pikett had denied this, an expert witness had described certain behaviors, such as jerking on leashes and strategically stopping as he and the dog paced a row of cans, that the expert described as consistent with an attempt to induce alerting behavior.  Thus, the court in that case could appropriately reverse the granting of summary judgment.  Here, however, there was “independent and untainted evidence … corroborating the results of Pikett’s lineups.”  Therefore, there was no “alternative reason to reverse summary judgment” as to the defendants. 


The plaintiffs here, all defendants in criminal cases, tried to push their luck too far in filing suits against almost everyone involved in their prior indictments and incarcerations.  Sufficiently few facts are given concerning the scent lineups conducted by Pikett that a detailed criticism is not possible.  It appears that, as with prior lineups that have been discussed here (see, e.g., blogs for April 22, 2010, and March 29, 2011), Pikett used old samples, may not have properly isolated scents, and conducted lineups in a manner where cueing was possible.   Curtis v. Anthony, 2013 WL 823428 (5th Cir. 2013)

The Unfortunate Belief That Scent Procedures Can Ignore Science

Courts have admitted dog tracking evidence over challenges based on arguments that such evidence does not pass as scientific, saying for instance that a scientific inquiry was “not required because it would be a superfluous confirmation of that which is already known.”  Connecticut v. Kelly, 2009 WL 323481 (2009).  This logic has been inappropriately extended to scent lineups.  See Michigan v. Giles, 2008 WL 2436529 (Ct. App. 2008) (“We are not persuaded that the evidence offered was of such a ‘scientific’ nature as to necessitate application of Daubert to the scent line-up and defendant has not presented a case which would persuade us to do so.”), 769 N.W.2d 683 (2009) (order denying leave to appeal). For additional citations, see discussion and notes in Police and Military Dogs at 62-3. 

The fact that canine behavior and procedures that involve such behavior cannot be fully subjected to scientific analysis because of the variables in a dog’s behavior seems to have led to a belief that even research on how those procedures can be made more reliable can be ignored. This amounts to saying that traditional handler practices need take no notice of science. There is a considerable and growing literature on the use of scent lineups, some of which could make the evidence of significant forensic value, as has been discussed elsewhere.

Whether the American criminal justice system would accept the cost of scientifically conducted scent-lineup procedures is a separate issue, but failing to conduct scent identifications and lineups in optimal formats will continue to mean that scent identification evidence should be viewed as very low on the evidentiary ladder and probably not worthy of being on the ladder at all. 


There has been at least one wrongful conviction based largely on a scent lineup. (Dedge v. Florida, 442 So.2d 429 (Ct. App. 1983; Dedge was exonerated by DNA evidence in 2004; see also Merran, A.H. Anatomy of a Wrongful Conviction: State v Dedge and What It Tells Us about Our Flawed Criminal Justice System. University of Pennsylvania Journal of Social Change, 13, 137). California v. Reynolds and Curtis v. Anthony do not involve wrongful convictions, but they do put scent identification in a bad light.  

The Innocence Project has labeled scent lineups as junk science.  This is a harsh criticism but is actually correct as to those scent lineups that have been described in any detail by American courts.  The criminal justice system should be making a better effort to assure that this label does not stick or, as happened in the Netherlands, our game of Russian roulette with scent identification procedures will result in more wrongful convictions. There is a fix to this problem, but law enforcement administrators and trial and appellate judges have to first open their eyes. 

This blog was written by John Ensminger and L.E. Papet.

No comments:

Post a Comment