Sunday, October 23, 2011

Station Identification as Forensic Procedure? California Court Strains Tracking Law in Drive-By Shooting Case

Tony Walker was sitting in a parked car in the afternoon on May 7, 2007, when a Dodge Intrepid came past with three bald-headed Hispanic men, at least one of which, according to a witness, had a gun. Multiple shots came from the Intrepid, killing Walker and Jamal Varcasia, who was running across the street at that moment.

The witness, Raeshel Shay, was frightened and did not immediately report what she had seen. Later she was in custody for having a fraudulent gift card when, during a monitored phone call to her sister, she mentioned the shooting. Using photographic six-packs, she identified Jose Elias as the driver, Joseph Ruiz as the rear passenger, and Eric Perez as the front passenger. At trial, she identified them again but said Elias had been in the back seat and Perez had been the driver.

Elias and Ruiz were also identified by Johann Montoya as being in a car near the scene of the May 7 shooting, though Montoya made this statement to an officer after he was beaten by some of the men in the car on or about May 11. At trial, Montoya denied making a statement about the May 7 incident.

Bullet casings were recovered from the scene of the murders. All casings were fired from the same Remington Peters 9mm Luger, and probably all the recovered bullets and fragments were fired from that weapon as well. A .32 caliber handgun was recovered later during the investigation at a different location. This was not the murder weapon but there was some evidence that a second gun had been in the Intrepid on May 7 and the police considered it possible that this gun had been in the car during the shootings.

Station Identification

A station identification was performed on May 16, 2007, by Ted Hamm, a civilian handler working for the police, and his dog, Bojangles. Detective Grant Curry had three detectives take the three defendants—Ruiz, Elias, and Perez—to the third floor of the Pasadena Police Station and put them in different rooms. The opinion states:

“Neither Detective Curry nor Hamm knew where on the third floor the defendants were. Detective Curry then gave Hamm three bullet casings recovered from the crime scene and the .32 caliber gun. Using a Scent Transfer Unit (STU), described as a modified dust buster, Hamm extracted scent from the three casings and placed it on a sterile guaze pad. Hamm did the same with the .32 caliber gun, placing its scent on a different gauze pad.”

It is not clear why Hamm felt it necessary to place scent from all three casings on the same pad since more than one person may have loaded different rounds and separate pads from each casing might verify this. In any case, the scent pad created from all the casings was used by the dog to identify two individuals. There was no discussion concerning whether this might indicate that one of the rounds was loaded prior to the incident on May 7, which could argue that one of the defendants might not have been present during the shootings.

“Hamm brought his dog, Bojangles, up the elevator to the third floor and let Bojangles smell the pad containing the scent from the casings. Bojangles immediately went down the hall, making turns, and led Hamm and the detective to Perez. After Bojangles smelled a pad with the scent from the casings again, Bojangles this time led them to Elias. But when the scent pad was reintroduced a third time, Bojangles did not move.”

Why did the dog not move after being scented a third time? Presumably we are to understand that the dog had completed its tracking, but this was not explained. Did Hamm and Bojangles begin each sniff in the same location? Once the dog alerted to a suspect, was the suspect removed? Was a replacement put in the same place as the identified suspect had been? How many choices did Bojangles actually have in terms of suspects or individuals who looked like suspects after being scented each time?

A third identification took place in the Pasedena Police Station parking lot:

“The other detectives then took Elias, Ruiz, and Perez to the parking area, and Hamm had Bojangles smell the scent pad from the .32 caliber gun. Bojangles led to Ruiz.”

This change of location should also be explained. A parking lot identification has some of the same concerns as a station identification and most of the same questions raised above should have been of concern to the police and defense counsel.

A jury convicted Elias and Ruiz of the murders in February 2010. Elias received a sentence of life without possibility of parole plus two consecutive 25 year terms. Ruiz was sentenced to two life terms, each with a 15 year minimum plus two consecutive 25 year terms.

Canine Evidence at Trial

In a pre-trial hearing, the court determined that a scent transfer unit (STU) is, in the words of the appellate court, “generally accepted in the scientific community as a collection device and that it can pick up scents from items that have gone through volatile events or scents.” The trial court also determined that a foundation had been laid for the handler and the scent dog.

The appellate court found the assignment of error of the defendants regarding the canine evidence “imprecise,” but concluded that they were objecting to the STU on scientific grounds, but to the dog scent identification as lacking a proper foundation. The defense argued at trial that it had not been established that an STU could pick up multiple scents. It is not clear where the defense was going with this argument. Did someone on the defense team think that an STU selectively picked up only certain scents? In any case, the defense seems to have dropped this line, but then suggested that a dog could not differentiate multiple scents. The appellate court called this a foundational issue. In any case, canine detection is based on the knowledge that a dog can be scented to one odor and thereafter distinguish this odor from others that might be present at the same time. This was established scientifically by Otto Kalischer more than a century ago. Dogs can also be taught to remember multiple scents (up to 17 separate individuals were distinguishable by dogs in one experiment) and will continue to distinguish them for much longer than was required for the testing discussed here.

The appellate court concluded that the trial court did not abuse its discretion in finding that the prosecution established a foundation to admit the dog scent identifications. The court analyzed the history of dog tracking and scent lineups in California, describing cases reviewed extensively in Police and Military Dogs, including People v. Willis, 115 Cal.App.4th 379, 9 Cal.Rptr. 235 (2004), a station identification in which a California appellate court determined the admission of the identification using an STU was clear error, though harmless because of the other evidence. Despite the rejection of canine evidence in Willis, the appellate court here found that the testimony of the experts had provided an adequate foundation for admission of the scent identification evidence:

“We find that the trial court did not abuse its discretion in finding that the People laid an adequate foundation to admit the dog-scent identification evidence through the testimony of its expert witnesses: Dr. Kenneth Furton, a professor of Chemistry and Biochemistry; Rex Stockham, an FBI supervisory special agent who is the forensic canine program manager and research program manager for the evidence response team unit; and Ted Hamm, a civilian contract canine handler primarily employed by the Los Angeles County Sheriff's Department and Bojangles's handler and trainer.”

Hamm and Bojangles

The defense objected that there were no negative scent pads used, i.e., negative controls or zero trials in which the dog was scented on a pad unconnected with the case to verify that the dog would not start tracking if there was no scent related to the pad to be followed. One of the experts described his use of a negative scent pad to ensure his dogs are working properly, before the actual trailing is done. The court rejected this argument as follows:

“Using such ‘controls,’ however, is not common practice, and the only law enforcement agency that uses controls is the FBI. Hamm's failure to use a negative scent pad did not therefore establish that the procedure he used was a bad practice.”

Hamm is familiar with negative controls. In a 2005 case, the Los Angeles Superior Court cited several experts (including Stockham) regarding the value of negative controls and said that Hamm, a handler in that case as well, “uses negative checks and Knight responds reliably.” People v. Salcido, Docket No. GA052057 (Los Angeles Superior Court 2005). The authors are aware of law enforcement agencies other than the FBI using negative controls, but perhaps the court is correct as to California police practice.

The defense also argued that Hamm’s procedures were faulty because he only used one dog, not two. The court also rejected this:

“Stockham, however, merely testified that while he prefers to use more than one dog in his investigations, he uses single dogs too. Stockham did not testify that the only generally accepted practice is to use two dogs. In fact, Stockham praised Hamm as ‘very competent’ and in the ‘top echelon of experienced handlers in the U.S.’ He considers Hamm to be a leading expert in scent trailing.”

In a properly conducted scent lineup, it is our opinion that several dogs increase the reliability of an identification. (See draft paper of Ensminger, Jezierski, and McCulloch, pp. 58-63, describing how requiring three or more dogs for a positive identification significantly reduces the number of identifications, but also reduces the chance of a misidentification almost to zero).

The defense then argued that a finding of adequate training and reliability could not be made because Bojangles was not certified. The court rejected any certification requirement, and concluded that one of the experts in the case, Furton, though mentioning certification, had not meant that it was required:

“Defendants incorrectly assert that Dr. Furton testified that a dog must go ‘through certain blind tests and reach a set percentile of correct identifications’ to be certified in California. What Dr. Furton actually said was a California organization he belongs to requires a training log, blind tests, and a certain percentile of correct identification to certify a dog; he did not testify that California has a certification standard or process. Rather, there is no national standard for certification or training of scent dogs, and standards vary from state to state. Neither California nor the Los Angeles County Sheriff's Department have requirements for certifying trailing dogs. The Scientific Working Groups on Dogs and Orthogonal Detection Guidelines (SWGDOG), an international working group funded by the FBI, is developing best practice certification and assessment guidelines for police-related canine disciplines.”

Hamm is presumably not opposed to certification. In Salcido, he used a dog named Knight that had been certified in 1998, apparently with the California Bloodhound Handlers Coalition, though Hamm did not purchase the dog until 2000. The court then summarized the position of another expert, Rex Stockham of the FBI, regarding training and certification:

“Although there are no national or specific state standards for training or certifying a scent dog, Stockham described his training process and how he determines a dog's reliability. Stockham keeps detailed training records to compare and contrast the milestones a dog should reach. To determine a dog's reliability, he relies on frequent observation of teams and blind testing, which the FBI routinely conducts. Before Stockham considers deploying a dog for a job, he subjects it to a series of tests and controls. The ‘best practice’ is for someone not routinely involved in the dog's training to assess its readiness for fieldwork. But a dog's handler also has the ability to determine whether the dog is positively or negatively trailing a scent. Once a dog is ready for fieldwork, the dog still requires maintenance training, and the industry standard is a scent dog should have 16 hours of maintenance training, although some people in the field believe that once a month is sufficient.”

Although some of Stockham’s comments would seem to provide a model different from what Hamm actually used, there was no further discussion of the disparity.

The defense also pointed out that Hamm had lost training logs dated before 2008 in a computer crash and argued that this meant that “Bojangles’s training and abilities were merely anecdotal.” The court responded:

“We disagree. Hamm testified extensively and specifically about his and Bojangles's background. Hamm has trained bloodhounds for 21 years and participated in over 2,000 investigations. He begins training a dog when it is a puppy. The dog first watches a person run away and then follows. Then the person drops an object that has his or her scent on it. Eventually, the person goes out of the dog's sight, and the dog must transition from visually following the person to using its nose to find the person. The length, age, and complexity of the trail is slowly increased. Hamm also works blind, meaning he doesn't know where the person is or what the result should be. The training process takes about two years, although a dog continues to do maintenance training even after it is ready to work. He tries to do maintenance training at least once a week and more often if he is able.”

Hamm is no doubt busy. In a 2008 opinion, the court cites the trial court’s description of Hamm as having been involved in 1,200 cases. People v. Alonzo, 2008 WL 2248628 (Ct. Appp. 2008). This suggests that Hamm works on at least 200 investigations a year, a significant number for a contract worker.

The absence of training records is not made trivial by a restatement of the training procedures preferred by the handler of the dog. Hamm has testified in many cases, and it would seem likely that prior records might have been introduced in other cases where he testified, but this possibility was not discussed. The court continued regarding Hamm’s training regimen:

“Hamm trained Bojangles in this manner, on trails fresher than 48 hours, to discriminate between human scents and to identify only the scent he's given. Bojangles has also been trained in different environments and on different surfaces. He's been trained with the STU and on a variety of scent articles, including spent cartridge casings. Bojangles frequently participates in a weekly maintenance training session. Hamm has been doing casework with Bojangles for three or four years with no history of falsely identifying targets in either blind testing or casework. Bojangles has confirmed cases, which are cases in which the dog's conclusion is confirmed by other sources, for example, witness statements, confessions, and DNA. To Hamm's knowledge, Bojangles has no confirmed negatives, where the dog found a scent and trailed to a subject who did not match the scent.”

The court apparently accepted Hamm’s testimony that Bojangles had “no history of falsely identifying targets,” and that “[t]o Hamm’s knowledge, Bojangles has no confirmed negatives.” Not many handlers could say this about the entire history of any dog. A footnote explains that “Hamm has been able to confirm approximately 40 of Bojangles’s cases out of 300.” This meant that the defense apparently had access to about 13% of the dog’s field work.

The defense also argued that there was no evidence that a bluetick coonhound had particular scent abilities. Here, the court correctly observed that “to the extent a dog’s breed is related to its ability to discriminate scents, there has been an insufficient foundation….” Coonhound mixes have been effective trackers since the nineteenth century.

Thus, the defense had no access to a significant part of the dog’s training history or to the vast majority of its field work. Consequently, neither did the experts. A good part of the prosecution’s case, therefore, rests on the testimony of the handler, to which the court was quite receptive.

A Florida court recently excluded canine evidence where adequate field records could not be supplied (though the records there were more substantial than those available here), stating that if “an officer fails to keep records of his or her dog's performance in the field, the officer is lacking knowledge important to his or her belief that the dog is a reliable indicator of drugs.” We suggested in that context, which could have involved the exclusion of evidence in the future from a large number of canine teams that had no advance notice that additional recordkeeping would be required, that the Florida court should perhaps have fashioned some interim means of qualifying a dog. The solution in the case here would have been to allow independent testing of Hamm and Bojangles prior to the admission of the canine evidence. We suggest that, if such an approach were accepted, the testing should consist of an independently monitored scent lineup, since duplicating conditions on the third floor of the Pasadena Police Station would be impossible with any scientific objectivity.

Human Scent

The court then discussed the uniqueness of human scent, the ability of dogs to pick up scent from an object, and the ability of dogs to differentiate multiple scents on an item. Relying primarily on the experts, the court correctly determined that these elements were supported by scientific findings as accepted by prior California cases.

Instruction to the Jury

California provides for standardized instruction in tracking cases, and the CALCRIM No. 374 was used for this purpose:

“You have received evidence about the use of a tracking dog. You may not conclude that the defendant is the person who committed the crime based only on the fact that a dog indicated the defendant. Before you rely on dog-tracking evidence, there must be, one, evidence of the dog's general reliability as a tracker, and, two, other evidence that the dog accurately followed a trail that led to the person who committed the crime. This other evidence does not need to independently link the defendant to the crime. [¶] In deciding the meaning and importance of the dog-tracking evidence consider the training, skill and experience, if any, of the dog, [its] trainer or [its] handler together with everything else that you learned about the dog's work on this case.”

The defense pointed out that the case at hand was not a typical tracking case in that the dog did not follow a scent from the crime scent to a location where the defendant had been or was found, yet the instructions were issued without objection. The court acknowledged that the situation was not typical:

“Although CALCRIM No. 374 might seem more appropriate to [tracking situations], it is not inapplicable to a station identification. The instruction simply says that there must be ‘other evidence that the dog accurately followed a trail that led to the person who committed the crime.’ (Italics added.) Bojangles smelled scent from the .32 caliber gun and trailed to Ruiz. Although there was no physical evidence that a .32 caliber gun was used during the crimes (only nine-millimeter bullets and casings were recovered from the crime scene and from the victims' bodies), Raeshal Shay testified that she might have seen two guns; hence, it was the prosecution's theory that Elias and Ruiz committed the crimes together and that there was a second gun, possibly the .32 caliber gun found at Ruiz's girlfriend's house. The dog-scent evidence therefore merely connected Ruiz to the .32 caliber gun; there had to be other corroborating evidence that Ruiz was the person who committed the crime.”

The court found the “other corroborating evidence” in other testimony regarding the actions of the defendants, and after reviewing gang-related issues, affirmed the conviction.

A Hybrid Procedure?

There was far too little discussion of the station and parking lot identifications, and too little consideration of how they were conducted. A floor plan of the third floor of the Pasadena Police Station would have been helpful in determining what sort of tracking was really involved. How many rooms were on that floor? How many suspects were on the floor? If there were only a few suspects on the floor, then the possibility of cueing is not negligible. Did any officers on the floor, either with the defendants or otherwise, know why the dogs were working there? If so, the possibility of third-party cueing was present, a factor that scientifically conducted scent lineups remove by insisting that no one within sight of the dog know the correct station in a lineup.

If the dog was not trained with negative controls, it may have performed its work in expectation of a reward. How many individuals not in police uniforms and dressed as gang members were actually on the third floor that the dog could choose from? In a scent lineup, the possibility of randomly correct identification increases as the number of stations in the lineup decrease. Was this issue considered?

The tracking aspect of the procedure also needs to be explained in more detail. Did all the defendants on the floor use the same hallway and follow the same path except when entering separate rooms? Did any of the suspects enter any rooms where other suspects were placed, even if only temporarily? Did the dog use the same hallways and passages as the defendants? Did any defendant enter a room through a passage where the dog could not sniff? If so, this begins to look like a modified scent lineup rather than a modified tracking or trailing.

Was the procedure videotaped? Although scent lineups should, in our opinion, be videotaped, it is seldom practical to videotape a tracking or trailing. Here, however, given the limited space and the advance placement of the suspect, it would have been advisable to use video cameras. If this is to continue to be the Pasadena Police Department’s method of scent identification, stationery video cameras should be installed in the third floor in places that will capture the movements of the suspects as well as those of handlers and dogs.

The presence of other individuals than those directly involved in the identification should have been recorded. Certainly this would be required in a scent lineup. So who was present in the hallways? What did they know? Were doors to the rooms with the defendants open? Were all other doors on the floor open as well? Could individuals in the hallways see the dogs when they entered the rooms and presumably alerted to the suspects? If only the doors to the rooms with the defendants were open, there was little possibility for randomness in the dog’s indications.

There were other differences from tracking. In most tracking situations, the dog has the possibility of following a trail that ends with no suspect in the vicinity. The suspect may cross streams, enter cars, disappear in countless ways. There is a good chance that the dog will abandon the trail. The possibilities of what can happen in an outdoor tracking are often infinite. Here it appears the handler knew there was a correct choice to be made in a finite space, much as if the handler was guaranteed that one station in a lineup would be a positive match. This is another reason why some sort of negative control should have been performed, one where Hamm would not know that there was no match to be found on the floor, but where individuals who looked like suspects were present in some of the rooms.

Some research has indicated that a longer trail reinforces the dog’s ability to identify a suspect as the individual being tracked. (Police and Military Dogs, p. 62) The trails here were presumably very short. One study found that dogs often follow the correct trail but fail to alert to the correct party at the end. In one segment of that study, dogs followed the correct trail 100% of the time, yet alerted to the correct “suspect” only 73% of the time. Did Bojangles ever alert to anyone besides a suspect? Apparently not, yet the fact the defendants were accompanied by escorting detectives raises the possibility that the dog was not always tracking a defendant. Without a videotape the only witness whose testimony could be elicited on this question was apparently Hamm. Did defense counsel obtain detailed accounts from the defendants of what happened on the third floor of the Pasadena Police Station? Even if so, would they have understood that they were seeing an alert to someone besides themselves?

Hamm and the Pasadena Police Department have begun to use station identification as a procedure. This formalized station identification fits neither the parameters that are applied by SWGDOG or any certifying organization to a scent lineup or most tracking environments. The court rejected that the argument that the procedure used by Hamm had to be vetted as a novel scientific approach. That may be correct because it was an amalgam of two generally accepted procedures. Yet because it was an amalgam it must be analyzed for what it is, not assumed to pass as a tracking with insignificant variations.

Station identifications have often occurred accidentally. The trail goes cold, the dog stops tracking and is returned to the station, where a suspect has been brought in on other evidence, the dog picks up the scent inside the station and resumes tracking, then alerts to the suspect. Here this result was not accidental but expected.

Hamm and Pasadena police officials have used the third floor of the Pasadena Police Station before. See California v. Smith, 2011 WL 1350762 (Ct. App. 2011). Were the defendants put in the same rooms as those in prior cases? If so, this also raises the possibility that Hamm may have expected the defendants to be in certain rooms on the floor. Were these rooms ever sterilized? How often was the procedure on the floor being used? Could anyone be certain that odor from prior suspects was not still present?

There are reasons why scent lineups must be so rigorously conducted in order to be fair, and why their use in some countries in Europe (the Netherlands, Poland) has been so considerably restricted. Hamm’s station identification procedures were not required to meet any scent lineup standards because the court accepted that this was a tracking situation. Yet the variations from tracking were sufficiently great that it must be asked why scent lineups were not used, since these could involve parameters that would assure a high level of reliability.


The case is disturbing at a number of levels. The court affirmed the reliability of Bojangles based on the handler’s description of the dog’s history, without either adequate training or field records, or any certification of the handler and dog as a team. Although the court was correct that certification can vary from state to state and organization to organization, the presence of records of a certifying organization could have been of considerable assistance in a case where the handler could offer only fragmentary records.

Station identifications have a legitimate place in American legal history, and have resulted in properly admitted identifications of suspects since at least 1918. Nevertheless, the Pasadena Police Department, with the help of California courts, has inappropriately elevated this procedure to a forensic technique. Significant differences of this formalized station identification from a typical tracking situation were ignored in this case, and have given the green light to further use of a procedure that fails to take into account the controls that can assure that a scent lineup is objective. California has here accepted an approach that may far too easily result in misidentifications.

It must be acknowledged that canine evidence is often important in drive-by shootings where witnesses can disappear for many reasons, and California is right to want to preserve the possibility of using this evidence. The solution is to adopt skillfully managed scent lineups, not to use a slapdash cross between tracking and scent lineups.

A larger concern is that if American law enforcement authorities continue to employ poorly designed scent identification techniques, the label of junk science that has been given by the Innocence Project and others to such techniques will stick after a few more reversals and there will be considerable pressure on U.S. courts to stop allowing scent identification altogether. Unfortunately, the junk science label is correct as to most--perhaps all--scent lineups described in any detail in American judicial opinions. It would be a shame to have such a wholesale rejection of scent identification since scent lineups can be made much more rigorous than has been true in the United States so far and could be a valuable forensic technique for law enforcement.

People v. Elias, B224372, 2011 WL 3949808 (Ct. App. 2011). It has been pointed out to us that we should note that this opinion has not been certified for publication and, under California Rules of Court, Rule 8.1115, may not be cited or relied on by a court or party except under limited exceptions. This rule has been the subject of considerable criticism. One commentator noted that "[t]here is no valid reason why litigants should not be able to let a trial judge or appellate justice know how their colleagues decided a case involving similar facts even if that opinion is unpublished." Keshavarzi, M. (May 7, 2010), Daily Journal. In any case, our criticisms are not part of a legal proceeding.

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


  1. Mr Ensminger from this blog I have started reading your book Police and Military Dog. I find it a very interesting read and so I have a question I am hoping you will answer.

    In your research did you find where the Defense used a situation where the dog in a scent lineup ID someone other then the person of interest. As the case went forward the Defense tried to say my client did not do this and the dog shows this as he did not id my client in the lineup? Was there any thing that showed how a lineup gone bad by the dog not performing to expectation?

    Thank you for any input.

    1. Look at page 82 of my book on exonerating identifications. Also read People v. Melara (Cal. App. 2006), summarized in "Scent Identifications and Lineups in US Courts, blog of May 9, 2010 where identification of someone not indicted is mentioned. Unfortunately I have a significant backlog of scent ID cases to write up for a revision, but I suspect a Westlaw or Lexis-Nexis search for scent lineups might produce some additional results.