The Dog’s Nose and Scent

1 Sense of Smell

This chapter is concerned with the internal and external aerodynamics of the nose involved in the sense of smell (olfaction) as they relate to training and searching with SDs. The sense of smell involves detection and perception of chemicals (odorants or scent molecules) inhaled by the dog. These scent molecules in the environment enter the dog’s nose in the gas phase, in the solid phase as particulates from a source, and attached to particulate matter such as dust or skin fakes. In the warm and humid environment of the nose, scent molecules in the gas phase and those detached from particulates con­tact sensors (receptors) that generate electrical signals which are sent to the brain. Signal processing and learning by the brain result in the “perception” of an odor or scent.

1.1 Physiology and Function

Olfaction links the dog’s brain to their external environment. Figure 1 is a schematic diagram of the “wiring” system for olfaction in mammals, includ­ing dogs.

There is a cavity (olfactory recess) at the rear of the nose behind and below the eyes with a lining of tissue (epithelial layer) that contains the receptor cells (neurons) (Figure 1). The ends of the receptor cells have 10 to 30 cilia, tiny hairlike structures immersed in mucous, that are in contact with inhaled air containing odorants (Marples 1969). The binding sites for odorants in the air are located on the cilia. Each receptor binds to a single type of odorant. When this occurs, the reaction produces a tiny electrical signal that is transmitted to a kind of junction box (glomerulus) in the olfactory bulb of the brain. Several thousand receptors for a specific odorant are randomly distributed in the nasal lining and are connected to the same glomerulus. Mitral cells in the glomerulus send the signal to the olfactory cortex of the brain where infor­mation from several types of scent receptors is combined into a pattern (an odor object) that characterizes the perception of a scent. The brain can do this because each odorant activates a unique combination of receptors.

Figure 1 A schematic diagram of the olfactory system in mammals which shows how dogs detect scent molecules. (From Conover 2007. Used with permis­sion of the Nobel Committee for Physiology and Medicine.)

An odorant also possesses properties that can contribute to scent dis­crimination (Schoenfeld and Cleland 2005). These include volatility and water solubility that influence movement of molecules through the nose dur­ing inhalation. The random distribution of receptor cells makes it possible for dogs to modify scent patterns by influencing the number of molecules reaching different receptors during a sniff. Dogs can then modify and ana­lyze scent patterns by regulating their sniffing.

There is another olfactory system for detecting scent that uses the vom­eronasal organ in the roof of a dog’s mouth. It is thought that this system is used primarily for detecting pheromones relating to sexual activity. However, retrievers used in duck hunting, trailing dogs following a trail through water, and water search dogs have been observed tasting the water. The process is different from drinking in that the mouth is partially opened and water flows through it and out the sides of the mouth. This suggests that the dogs may be using the vomeronasal organ to detect scent in or on the water or it may be possible for the dogs to “taste” scent.

11 The Dog’s Nose and Scent The foregoing description of the olfactory system of mammals suggests that the physical and chemical characteristics of odorants that allow them to bind to receptors form the basis for the dog’s perception of scent. While bind­ing to receptors is an initial and necessary step in olfaction, it appears that memory and perceptual learning play a fundamental role in scent discrimi­nation and perception (Wilson and Stevenson 2003; Lovitz et al. 2012). The process of perceptual learning involves continued experience (training) with a scent which leads to improved detection, recognition, discrimination, and perception of that scent. It is important to train regularly with all the scent variations that SDs will be tasked to find (Goldblatt et al. 2009).

Most scents are mixtures of many odorants. A mixture of only two odor-ants can result in a new scent that is weaker or stronger than either odorant or one odorant can mask the other partially or completely. This makes it impossible to predict how a mixture will be perceived by SDs. The results of studies of mixtures of four or more types of scent molecules (Wilson and Stevenson 2003; Lovitz et al. 2012; Goldblatt et al. 2009) indicate that people, dogs, and other animals do not perceive all the individual scent molecules in a source but perceive the mixture as a whole; it is synthesized. The idea that dogs can smell every component in a stew is probably not correct.

Synthetic processing based on experience uses memory and pattern rec­ognition to form an odor object for the mixture. The process is similar to the sense of sight in that the brain does not perceive each pixel of a tree but forms a visual object that it perceives as a tree. Formation of odor objects by the brain is a relatively simple, efficient, and flexible way of perceiving odors in the environment that gives an almost unlimited ability to smell new odors. Tus, the search object that SDs seek is likely an odor object (Gazit et al. 2005a).

The above description of the sense of smell is primarily concerned with the details of sensing and perception that are internal processes. However, considerable practical information can be obtained by considering the fluid dynamics (air and scent movement) interior and exterior to the dog’s nose. Dogs smell by sniffing and Figure 2 is an accurate, three-dimensional model of the left canine nasal airway, reconstructed from high-resolution MRI scans (Craven 2008; Craven et al. 2010).

When inhaling, air is drawn into the nostril from a distance up to about 4 inches. Dogs tend to reduce this distance to zero (Settles and Kester 2001), which provides them with the highest scent concentration, allows them to sample the source independently with each nostril, and to discern the spatial distribution of the source. Flow paths during exhalation bypass the olfactory recess, leaving scent-laden air there for more time which enhances absorp­tion of scent molecules. Sniffing frequency for free air sniffing is about 4 to 7 times per second in short bouts. Sniffing can be observed in the movement of the thin skin on the side of my Beagle’s nose but is difficult to see on the side of my Lab’s nose due to the thicker skin. If the source is inaccessible, the bouts are longer with a frequency as low as of one sniff every 2 or 3 seconds. Notice the long deep sniffs a dog takes when sniffing at the crack of a closed door to an inaccessible room.

Figure 2 The internal aerodynamics of canine olfaction for the left nostril. Top: The olfactory recess (yellowish-brown) on the right. Flow paths during inha­lation are distinct for respiration and olfaction. Bottom: Velocities for the same flow paths show high velocity (red) in the front of the nose and low velocity (blue) in the olfactory recess. (From Craven 2008. Courtesy of Dr. B.A. Craven.)

Part of the inhaled air is used for respiration and part is directed into the olfactory recess (Figure 2) where hundreds of millions of receptor cells are located (I would like to know who counted them). Tus, olfactory and respiratory airflows are separated, each with a distinct flow path and velocity through the nasal cavity.

On exhalation, an interior fap of skin at the front of the nose closes and slits on the side of the nose cause air jets to be directed to the sides of the dog’s nose and downward (Settles et al. 2002). This nearly eliminates mixing of the exhaled air with the inhaled air so that fresh uncontaminated scent is introduced into a dog’s nose during sniffing. Laboratory experiments have shown that dust particles smaller than about 100 μm can be made airborne and subsequently inhaled (Figure 3). This indicates that these jets can also dislodge particulates, skin fakes, and, probably, adsorbed volatile organic compounds (VOCs) from surfaces.

Figure 3 Surface dust particles disturbed by exhalation jets from the dog’s nose. Flow directions for sources on the surface show that these air jets blow particles in a cloud to both sides of the nose and back in the photo. Some of the airborne particles can then be inhaled. (Courtesy of Dr. G.S. Settles 2002a.)

Observations of CDs show that dogs cast about, typically cross wind, when searching for a buried source or a hidden one on the ground surface. On detecting the source, their movements become more directed as they fol­low the scent plume. As they approach the source, their noses are close to the ground surface and they typically pass over the source while sniffing and then back up slightly until their noses are directly over the source or down­wind of it. The dogs attempt to put their noses as close as possible to the source before giving their alert, a desirable behavior. This method maximizes the ability of the dogs to detect scent since they can detect scent molecules in the air, cause particulates to become airborne and then inhaled, and actively dislodge scent molecules from surfaces which can be inhaled. Not all dogs use this method exactly. There are differences between dogs and between different breeds of dogs (e.g. hounds and some dogs tend to work with their noses always close to the ground).

This abbreviated description of the scenting process of dogs indicates that dogs smell VOCs in the gas phase and from particulates (since these may emit VOCs) that contact the receptors in their nose. They cannot smell particulate materials such as skin fakes and dust particles directly. Skin fakes (Figure 4) contain volatile glandular secretions (skin oils) and bac­teria which produce VOCs that dogs can smell (Syrotuck 1972). Dust par­ticles may contain VOCs from explosives, drugs, and cadavers since, under dry conditions, dust particles adsorb VOCs on their surfaces. With the dog’s head nearly vertical (Figure 3), the slits on the sides of the nose direct the exhaled air jets downward and backward along the surface which causes par­ticulates to become airborne and allows the dog to inhale some of them. On inhalation, the particulates are brought into the warm and humid environ­ment of the nose. For dust particles with adsorbed VOCs on their surfaces, the humid environment provides water molecules that replace the adsorbed VOCs and allows their transfer to the receptors where they can be detected.

Figure 4 Surface of the skin (magnified 2410 times) which shows several partly detached skin fakes and small fragments of detached skin. (From Clark 1974. Used with permission of Cambridge University Press.)

McLean and Sargisson (2005) proposed a variation of the effects of the air jets observed by Settles et al. (2002). When sniffing dry soil surfaces, the dog’s moist exhaled air jets may cause VOCs adsorbed on the dry dust par­ticles to be replaced by water molecules and be released into the air where the dog can inhale and detect them directly. While this hypothesis is plausible, Phelan and Webb (2002) suggest that it may not be necessary because of the robust scenting ability of dogs.

It appears that the scenting methods used by SDs to detect sources involve direct sniffing of source VOCs and airborne particulates. Dogs can also use exhalation air jets to remove VOCs and particulates from surfaces which can then be inhaled. VOCs adsorbed on the particulates can then be desorbed in the humid and warm air in their noses which frees them for detection. These methods are part of the sniffing process which dogs use to detect scent. In addition, conditions dictate which methods are most efficient and dogs appear to learn to select the appropriate methods for those conditions.

It is of interest to know whether dogs use their eyes in addition to their noses to locate sources. EDs have been tested under both virtually dark (very low light intensity) and full light conditions in controlled (indoor) and uncontrolled (field) environments (Gazit and Terkel 2003b). The main sense used by these dogs for detection was their scenting ability, not only when vision was difficult (in very low light intensity) but also in full light. Neither the presence nor the absence of light influenced the dogs’ detection ability. However, in addition to scent, search and rescue (SAR) dogs use their eyes to locate visible sources or subjects. Some hounds that trail with their noses to the ground learn to look for subjects. This difference between EDs and SAR dogs appears to be due to their training. EDs are trained almost entirely with hidden sources that can only be found using their scenting ability. SAR dog training generally includes subjects and sources that are sometimes visible, especially at short distances, so that the dogs learn to look for them.

There is little information on limits to the ability of CDs to detect VOCs from buried decomposing bodies, but experiments with EDs have shown that their sensing thresholds for explosives can be lower than those of labo­ratory instrumentation (Phelan and Webb 2002 or 2003). There were differ­ences in the sensitivities of the dogs. Not all dogs could reach these low levels and some dogs could not sense even high levels. There were differences in the sensitivity of an individual dog on sequential days and in the reliability of an individual dog at a given level of VOCs. Variations in training history and methods also resulted in different capabilities between dogs. This suggests it would be desirable to test the ability of candidate SDs to detect low scent lev­els and to select a dog with a suitable nose before investing the time, money, and effort in training and to define and use optimal training methods.

Dogs usually start to pant at temperatures in the 80s°F, but this depends on age, physical conditioning, level of work being performed, and genetic factors. Panting is thought to reduce a dog’s scenting ability because, with their mouths open, more air is drawn directly into their lungs through their mouth rather than through their nose. Experience with hardworking air scenting hunting dogs indicates that they appear to work better at air tem­peratures in the 60s°F or less. However, search dogs and hunting dogs appear to pant while running and are still able to scent the target source, indicating that some of the inhaled scent that enters through the nostrils moves into the olfactory recess.

Studies of racing sled dogs, hunting dogs, and other working dogs (Grandjean and Clero 2011; Reynolds 2017) discuss diet (nutrition) and exer­cise (physical training) requirements for optimum physical performance. A study by Altom et al. (2003) investigated the effects of diet and exercise on olfactory acuity (scenting ability). A group of dogs was exercised 30 min/day, 3 days per week on a treadmill, while another group was exercised 10 min/ day, 1 day per week. After 12 weeks, the two groups were subjected to a stress test (treadmill) for 1 hour, followed by measurements of their olfactory acuity (ability to detect an odorant). The exercise group did not show a reduction in their olfactory acuity from their pretest baseline. However, the non-exercise group showed a 64% reduction in olfactory acuity following the physical stress test. This data shows that a moderate physical conditioning program can help SDs maintain their scenting ability during periods of intense work. Gazit and Terkel (2003a) found that increased panting resulted in a signifi­cant decrease in explosives detection, but that the dogs were eventually able to adjust to working in extreme conditions. Grandjean and Clero (2011) point out that nutrition and training are interrelated and provide extensive recom­mendations for various types of working dogs.

The sense of smell in dogs and other animals has evolved by natural selection for the ability to use chemical cues (scent molecules) in activities associated with finding food, reproduction, and survival. Consequently, they are extremely effective at detecting and discriminating odors associated with these activities. This implies that the threshold for detecting some odorants may depend on their importance for survival to the dog. Some sources (e.g. explosives) that dogs are required to detect do not meet this requirement. However, dogs quickly learn that finding these sources enhances activities that are a normal part of their lives (obtaining food and bonding with pack members, handlers, and others).

The above description of the olfactory system for dogs suggests that the size of the system and the number of receptors would indicate the ability of the system to detect and discriminate scent. Rats, mice, and dogs have about the same number of functional olfactory genes and their olfactory systems are comparable in their ability to detect and discriminate odorants. However, dog olfactory recesses vary widely in size (compare Bloodhounds and Rottweilers to Beagles and Terriers) and all are much larger than those of rats and mice, a seeming contradiction. This suggests that other factors may be involved in the sense of smell in mammals.

2 Scent

Training and using SDs efficiently require an understanding of scent char­acteristics, movement, and interaction of scent with the environment. While information on scent characteristics has been scant in the past, there has been a rapid expansion of information over the last two decades. On the other hand, information on scent movement and the interaction of scent with the environment is sparse. This section reviews some of the information on scent characteristics and scent during training. Scent and Wind through Water Searches are concerned with scent movement and its interaction with the environment.

2.1 Properties

Scent molecules are defined by their physical and chemical characteristics at the molecular level, especially those that allow them to bind to receptors. Scent is the dog’s learned perception in the brain as a result of processing signals from the receptors and their experience.

All solid and liquid phases of matter emit gas molecules that are part of the scent that dogs detect. The process is often referred to as outgassing and can include evaporation (liquid to gas phase changes), sublimation (solid to gas phase changes), and desorption (removal of a substance that has been absorbed or adsorbed by a solid or liquid). We will use the general term, “vola­tilization”, to describe the processes by which scent molecules are transferred from liquids and solids to the gas phase. This transfer of a chemical into the atmosphere involves several steps, including not only phase transformations but also transport processes. The rate of volatilization depends primarily on the material’s vapor pressure. Materials with high vapor pressure at normal temperatures are said to be volatile (they outgas readily emitting a lot of mol­ecules, e.g. gasoline). Volatility of a material is measured by its equilibrium vapor pressure. This is done by placing the material in a closed container and monitoring the vapor pressure in the space (headspace) over the material and by more complex methods. The vapor pressure increases as more molecules leave the material than return to it. When the number of molecules leav­ing the material equals the number returning, the pressure becomes con­stant at the equilibrium vapor pressure. Volatility increases with temperature so that higher temperatures increase the amount of scent escaping from a source and available for detection. This makes volatility of the scent source an important factor in training, testing, and deploying SDs. Dogs can detect scent emitted from living and dead organisms and matter, both natural and manufactured, provided the scent source is sufficiently volatile or has a com­ponent that is volatile. Typical values of equilibrium vapor pressure in mm of mercury (Hg) near room temperature are: water vapor, 8; trinitrotoluene (TNT), 7.0E-6 (0.000007); acetic acid (heroin), 13; and dimethyl disulfide (human decomposition), 28. Normal atmospheric pressure is about 760 mm Hg (29.92 inches or 14.7 pounds per square inch).

Most of the scent from sources that SDs seek are VOCs. VOCs are com­pounds whose molecules contain carbon although some carbon containing compounds are classified as inorganic. VOCs have high vapor pressures at ordinary ambient temperatures that cause a significant number of molecules to volatilize from the liquid or from the solid form into the air. The scent of ethyl alcohol (C₂H₆O) is an example that even humans can easily detect. Examples of volatile inorganic compounds (VICs) include those containing nitrogen (e.g. ammonia, NH₃) and sulfur (e.g. hydrogen sulfide, H₂S).

Scent density refers to the “weight” of the scent molecules per unit volume and can be an important factor in scent movement. Density is not a significant factor for movement in an open system that is subject to air currents. However, it can be a factor in closed systems (e.g. contain­ers, autos, houses) with no internal air currents and in open systems with restricted pore space (e.g. soils, snow) that inhibit air movement. If the scent density is less than air (e.g. ammonia), the scent can be expected to rise, and if it is greater than air (e.g. dinitrotoluene, DNT), it will sink or flow downward. Almost all VOCs from explosives, drugs, and human remains are heavier than air so they would be expected, in the absence of wind, to drain downward and settle in low areas (depressions over graves, ditches, floor drains, car floors). In outdoor settings, scent can pool in these areas and, under calm and cloudy conditions, can remain there until removed by turbulent air conditions.

2.2 Surfaces

Surfaces in contact with airborne VOCs can interact with the VOCs and mod­ify the concentration of VOCs available to SDs. Common surfaces include natural ones (soil and rock, vegetation, wood) and manufactured ones (cloth, plastic, pavement, metal). VOCs in contact with surfaces can produce new chemical compounds that may help or hinder SDs. VOCs may be adsorbed (stick) to surfaces and later serve as sources by desorbing (volatilizing) by natural processes or by air jets from the dog’s nose. Some sources, especially drugs and explosives, produce microparticles that are “sticky” and adhere to surfaces where they volatilize or can be removed by the action of dog’s exha­lation air jets, producing scent for dogs to detect. This sticky nature of VOCs and microparticles can cause contamination of nearby surfaces such as other training aids, clothes, hair, hands, travel bags, and dust particles (Oxley et al. 2008; Goldblatt 2017). Dogs can detect these contaminated surfaces and dust particles (Phelan and Webb 2003).

When scent molecules contact surfaces, their behavior depends on their properties, properties of the surface, and external conditions (wind veloc­ity, temperature, humidity, and other factors). The amount of scent that volatilizes from skin secretions, decomposition fluids, explosives, drugs, and other sources depends on their volatility, surface area of the source, whether it is contained or buried, and environmental factors. A substance that has a larger surface area will evaporate at a faster rate, resulting in more scent (because there are more surface molecules in contact with the air). Wind increases evaporation rates from a source by enhancing the movement of scent molecules away from the surface into the air which increases its vola­tility. Increased temperatures increase volatility by making scent molecules more energetic which allow them to more easily escape a source.

The amount of scent and the time it remains on a surface also depends on the number of binding sites for the scent molecules. Metals have few binding sites while most natural surfaces such as vegetation, soil, and wood have many. VOCs attach readily to these natural surfaces at cooler tempera­tures and volatilize at warmer temperatures. This accounts for the daily and seasonal variations of VOCs found in the environment. For example, the air in forests normally emits more VOCs during the day than at night and more during warm seasons than colder seasons (Schade and Goldstein 2001).

Air humidity is also an important factor. VOCs readily attach to dust and very dry soils. With an increase in humidity, water molecules preferen­tially replace the VOCs on soil surfaces allowing the VOCs to escape into the air, resulting in dramatic increases in VOC emissions from the soil (Petersen et al. 1996).

Once scent has been deposited on surfaces, it has a remarkable ability to survive. Depending on the surface, it can survive washing, radiation, explo­sions, fire, and even attempts to mask its presence (Stockham et al. 2004; Goldblatt et al. 2009). Some sources (e.g. urine) may contain other chemicals, besides the scent molecules, that stabilize the scent and allow it to remain in the environment longer.

2.3 Residual Scent

The above discussion indicates that VOCs can attach to surfaces some dis­tance from the source. In addition, when a source is placed at a site and removed later, VOCs, microparticles, and liquid residuals may be left behind. These materials can be in the form of liquids from the source, liquids that have been in contact with the source, particulates from the source, and VOCs from it. Fingerprints consisting of liquids and particulates (skin fakes) are an example, and VOCs that accumulate on vegetation downwind of a source are another. VOCs from liquids, particulates, and the source, when it was present, can remain in the air and/or may be adsorbed on nearby objects around the former location of the source or downwind from it.

While there are numerous anecdotal examples of residual scent involving SDs, there are few studies of it. Scent from some sources such as explosives and drugs can be transferred to the skin and clothing of persons handling them and to objects they touch (notice airport security wiping baggage sur­faces). CDs have alerted on blood in vehicles that were involved in accidents a decade earlier (Osterkamp, unpublished), have alerted in vehicles that were used to transport bodies, and have found clandestine graves where the body had been removed. The FBI has shown that the scent of a person can linger in the neighborhood where their house was located for at least 6 months after they have moved. A trailing dog given a scent article from the person trailed a short distance to the house (Stockham et al. 2004). This could have been a result of VOCs on the nearby vegetation and other surfaces or perhaps the home with its many porous surfaces saturated with the person’s scent was acting as a source that attracted the dog.

Volatilized scent molecules from a source that has been removed can cause SDs to alert, but little is known about the conditions that need to be favorable for this to occur. In the short term (days to weeks), scent molecules can probably linger in the air at sites with restricted air circulation (e.g. car­pet, grass, leaf layer on the forest floor). In the long term (perhaps weeks to years), it appears that the scent molecules would have to be adsorbed on or absorbed in objects around the site. Changes in temperature, wind, humid­ity, the mechanical action of sniffing (air jets), or the dog’s movement may cause the scent molecules to dislodge or volatilize from the objects which makes them available for detection.

A study of the residual scent left by marijuana, hashish, amphetamine, cocaine, and heroin in unsealed plastic bags used the number of alerts of trained DDs to evaluate the persistence of scent at 24/48 hours after the drug was removed from the site of the hide (Jezeirsky et al. 2014). The residual scents of hashish and marijuana were the most persistent and that of heroin the least, although it has a high vapor pressure. The percent of dogs alerting on the former location of hashish sources in open plastic bags after 24/48 hours was 100/80% and on heroin 33/8%. The VOCs causing the alerts may have diffused through the plastic bags to the surface on which they were resting or may have passed directly to the air from the unsealed bags and then adsorbed on nearby surfaces. It is known that airtight containers used to transport sources to training sites may eventually develop contaminated exterior surfaces that leave residual scent wherever they are placed. These results suggest that care must be taken to allow time (several days or more) for residual scent to dissipate when using areas (buildings, vehicles, field sites) repeatedly during training and testing. They also suggest that an alert during a search where nothing is found may mean that a source was previ­ously present at the site or nearby. For most applications, it is desirable to train the dogs to offer their alert only when the source is present.

3 Training Aids

Training SDs requires that we have an example (training aid) of the target source to use to communicate to the dogs the scent we wish them to find. Training aids are commonly packaged and/or put in containers to prevent the dogs from coming in contact with the source. The packaging should not look like toys that are used to reward them since this may lead the dogs to pick them up, an undesirable behavior (Goldblatt et al. 2009). Aids should have the same scent expected to be found in searches. If the aids do not have the same scent, their scent should be sufficiently similar to the target source to allow the dogs to generalize and detect the source. It appears that the ability to generalize depends on the similarity of the scent of the trained and untrained target source, breadth of training with different variations of the source, and, possibly, the dog’s intelligence. If the target source is a single chemical compound the task is easier, but most sources contain many chemical compounds and it may not be known which chemical(s) triggers the dog’s alert.

The training aid could be some of the source material, something to which the source scent has been transferred, or a mimic (simulant) of the source. While the source material used with direct scenting is an ideal training aid, it can be difficult to obtain, inconvenient, sometimes requiring numerous source variations, and even dangerous or illegal to use (e.g. explosives, drugs, cadavers). Transferring the source scent to a sorbent material (e.g. gauze pad) alleviates these problems (this method is also used by trailing dog handlers to create human scent articles). The transfer can be accomplished by direct contact with the source (e.g. wiping a solid or transferring part of a liquid source onto the material), passively adsorbing scent from the source onto the material, and by actively drawing scent through sorbent material to which the scent molecules adhere. Wiping a source with sorbent material trans­fers some of the scent to the material. However, in some cases, wiping may destroy evidence such as fingerprints.

Passive adsorption involves placing the sorbent material in contact with the source or placing the source and the material, without contact, in a closed container. A significant time may be needed to transfer the scent and, if the material contacts the source, it may modify or destroy a fragile source. Actively drawing scent from the source through and onto sorbent material to create the training aid is a relatively fast process that still preserves the source for forensic analysis. The Scent Transfer Unit 100 (STU-100) is a portable, low power vacuum cleaner which uses this method of scent collection (Tolhurst 1991; Eckenrode et al. 2006).

Training aids may attempt to simulate the scent source by using the pure chemical, chemical mixtures, a chemical component(s) of the source that is a major portion of its vapor headspace, or by using a small amount of the active chemical in a nonhazardous matrix (e.g. silica, fiber, inert chemical). Another method separates the components of an explosive mixture but allows their gases to combine and form a merged scent (Lazarowski and Dorman 2014).

It has been shown that some simulants may not elicit an alert by SDs which indicates that not all the dominant components are the actual odor-ants used by the dogs to detect explosives and drugs (Macias et al. 2008; Macias 2009). Regardless of the method used to develop an aid, it should be confirmed that SDs trained using the aid recognize and alert on the tar­get source during field operations and that trained canines recognize and alert on the aid. Training procedures may also need to be modified to proof the dogs of scents introduced by the method. Also, the dogs still need to be trained on variations of the actual sources encountered while working.

Dogs are the gold standard for detecting sources under field conditions because of their scenting ability that allows them to distinguish the source from a diverse scent background, their mobility in difficult terrain, their use of highly efficient methods for detecting and precisely locating a source, their intelligence, and their rapport with humans. While electronic “noses” may be able to detect the presence of a source, they must be transported or directed around the search area by a human that understands the effects of weather (especially wind), terrain, and vegetation on scent movement; have a reliable way of precisely locating the source once it has been detected; be able to work through turbulence and discontinuous scent plumes, and be able to distin­guish the source from similar background scents. Also, an instrument can­not provide emotional support and rapport that may be desirable under the stressful conditions associated with finding some sources (e.g. explosives).

3.1 Explosives

The importance, difficulties, inconveniences, and dangers of using the actual explosives as training aids have resulted in extensive efforts to develop aids that are safe and convenient to use. These efforts have been hampered by the possible explosive nature of the aids, the large number of different explosives, possible improvised explosives, lack of knowledge about what chemicals trig­ger the dog’s alert, debate over training methods, and other factors. Table 2.1 shows some common explosives and their vapor pressures; many more exist.

Source: Data from Ewing, R.G. et al. 2013. Trends Anal Chem 42:35–48.a

1.9E−1 means the number with the decimal point moved one place left (0.19).

Generally, explosives in the top half of Table 2.1 are readily detected by dogs. Some explosives such as RDX and HMX have extremely low vapor pressures that indicate so few scent molecules are emitted from them that dogs are not likely to be able to detect the explosive component directly from the VOC emissions resulting from vapor pressure. Goldblatt (2017) suggests that stickiness of explosive microparticles may be as important as vapor pres­sure. The ability of dogs to use nasal air jets to dislodge and inhale particles of explosives and dust containing explosive VOCs may account for the ability of dogs to detect some explosives.

Vapor pressures of some explosives vary strongly with temperature. While there is considerable variability in the data, it appears that the vapor pressure of TNT increases about 10 times when the temperature increases from 25°C (77°F) to 40°C (104°F) (Oxley et al. 2005) indicating that much more scent would be available at higher temperatures.

The mol weight of dry air is 29 g/mol and that of humid air is less because the water vapor in it is just 18 g/mol. In the absence of wind or air currents, all the scent molecules from the explosives shown in Table 2.1 would eventu­ally drain to the lowest point in the area because their mol weight is greater than that of air. This suggests, when there is no wind or there is a downslope flow, it would be desirable to search for hazardous sources (e.g. buried explo­sives) on a slope by gridding across the slope and advancing the grid upslope.

Commercial explosives are often mixtures of other explosives and con­tain chemicals in addition to the parent explosive. These additional chemi­cals may be a result of the manufacturing process, weathering, chemical reactions between components, and other factors. They are commonly more volatile than the parent explosive and may account for most of the vapor in the headspace of the parent explosive (Goldblatt et al. 2009). For example, military TNT consists of about 99.8% of 2,4,6-TNT and less than 0.1% of 2,4-DNT but the vapor in the headspace consists of 58% 2,4,6-TNT and 35% 2,4-DNT, a result of DNT’s higher vapor pressure (Table 2.1). The primary component of US military C4 (cyclonite, RDX) is about 91% by weight, but it is not present in the headspace because of its low vapor pressure.

It seems logical to conclude that EDs would detect the more common chemicals in the vapor headspace. The problem for training is that while it may be thought that the dogs are detecting the parent explosive, they may be learning to detect a volatile component(s) rather than the parent explosive. Since TNT is found in about 80% of all land mines, this can have serious con­sequences. For example, a dog trained on TNT that contains DNT may learn to detect DNT. Russian and some other manufactured TNT does not contain DNT and neither does some weathered TNT which may result in a failure of the dog to alert on them (Goldblatt et al. 2009).

In a study to evaluate the ability of EDs to generalize from a pure potas­sium chlorate (PC) source to PC-based mixtures (Lazarowski and Dorman 2014), it was found that 87% of dogs trained with pure PC alone did not correctly alert to the presence of one or more of four PC-based explosive mixtures (PC plus a fuel). However, after training with a device that kept the components separated but allowed the vapors to combine, a significant improvement was seen in the number of dogs alerting to PC-based mixtures compared to training with PC alone. A similar inability of the dogs to gen­eralize from pure ammonium nitrate to ammonium nitrate-based explosives was also noted. If it is desired to have dogs reliably alert to a mixture of com­ponents X and Y, then it is not sufficient to train them on only one of the components. This is not surprising since for a mixture of two scents (X and Y), one may partially or completely mask the other or a totally new scent may be produced as noted previously.

Jezierski et al. (2012) conducted a study of 80 ED teams used by Polish police to determine the effectiveness of the teams in finding explosives in training and testing environments. The study was conducted using rooms known and unknown to the dogs and the exterior of cars outdoors (Table 2.2). Each search was limited to 10 minutes.

Dynamite (NG based) was the easiest explosive to find based on the shortest search time, most correct alerts (82%), least misses (2%), and less false alerts (17%). TNT was the most difficult explosive to find based on lon­ger than average search time, least correct alerts (49%), most misses (17%), and most false alerts (33%). These results may be primarily due to the higher vapor pressure of dynamite compared to TNT.

Source: Data from Jezierski, T. et al. 2012. Poster presented at the Canine Science Forum, Barcelona, Spain.

Teams were less successful searching known rooms compared to unknown rooms based on much longer search times, fewer correct alerts, and many more false alerts (although misses were fewer). The exteriors of cars outdoors were less difficult to search than the rooms although the rooms appear to have been much larger. Experienced teams that reexamined produced more cor­rect alerts and fewer misses but did not produce fewer false alerts.

3.2 Drugs

As with explosives, it is not necessary to use parent compounds for train­ing DDs. Research that includes studies of the compounds present in the vapor headspace coupled with studies of the response of dogs to these com­pounds indicates the dogs use the most abundant chemical component(s) in the vapor headspace to detect and find drugs (Furton and Myers 2001). Most of the VOCs in drugs have a high vapor pressure (Table 2.3).

Marijuana is a commonly used drug in the USA and is now legal in sev­eral states. Studies of the headspace volatiles have shown that there are many; however, α-pinene, β-pinene, myrcene, limonene, and β-caryophyllene occupy a major portion of the headspace (Hood et al. 1973). Additional stud­ies with canines indicate that α-pinene, β-pinene, and limonene are impor­tant for developing training aids for marijuana (Lai et al. 2008).

Cocaine was an early example of dogs alerting on the primary head-space component of the target source, methyl benzoate. Methyl benzoate is a decomposition product of cocaine with a high vapor pressure (Table 2.3). Field tests of 28 DDs showed that a threshold level of 1 μg (microgram) of methyl benzoate spiked with cocaine on US currency was required to initi­ate an alert by 50% of the dogs (Furton et al. 2002). However, the majority of dogs did not alert to pharmaceutical grade cocaine at levels as high as 1 g. Methyl benzoate is not found on circulated currency in sufficient quantities to cause an alert from a DD. Consequently, an alert on currency by a DD dog indicates that the currency was recently exposed to cocaine since no other drugs are commonly found on currency.

a Source: Data from Lai, L.H. et al. 2008. J Sep Sci 31:402–412.
b Source: Data from https://pubchem.ncbi.nlm.nih.gov/compound/.

Acetic acid is the primary volatile in the headspace of heroin and vinegar (Macias et al. 2008). Since vinegar may be commonly found in the search environment, acetic acid is not a good choice for a training aid for heroin and it will be necessary to find some other compound or to use the drug itself as a training aid.

Benzaldehyde is a primary scent volatile of methamphetamine (Vu 2001). Results from tests with a small number of DDs (6) showed that they alerted on street methamphetamine but did not alert on pharmaceutical grade methamphetamine.

Impurities in drugs because of the manufacturing process may produce volatiles and some manufacturing processes may result in the presence or absence of certain volatiles. Dogs trained on training aids developed from drugs manufactured by one process may not alert on similar drugs produced by another process. For example, the volatile piperonal is a recommended training aid for MDMA (Macias 2009). It has been shown that dogs trained to alert on piperonal will alert on MDMA and dogs trained on MDMA will alert on piperonal. However, piperonal occurs in widely varying amounts in MDMA depending on the manufacturing process, a condition that may cre­ate difficulties for dogs using piperonal as a training aid to detect MDMA.

A study of 164 DD teams in 1219 searches for marijuana, hashish, amphet­amine, cocaine, and heroin evaluated the abilities of the dogs to detect and locate these drugs as shown in Table 2.4 (Jezierski et al. 2014). Street drugs were used for searches in rooms familiar to the dogs, rooms unknown to them, outside rooms (stables, storerooms), exteriors and interiors of cars, and lines of luggage. Each search was limited to 10 minutes.

Marijuana was the easiest drug to detect based on the shortest search time, 92% correct alerts, only 4% misses, and 4% false alerts. Heroin was the most difficult drug to detect with the longest search time, 70% correct alerts, 12% misses, and 18% false alerts. This difficulty occurs even though the pri­mary scent volatile in heroin is acetic acid (Table 2.3) which has a high vapor pressure that should make it relatively easy to detect. The ranking from the easiest to the most difficult was marijuana, hashish, amphetamine, cocaine, and heroin. The areas searched had more than 83% correct alerts except for the interiors (58%) and exteriors (64%) of cars. False alerts were generally less than 15% except for the interiors (36%) and exteriors (22%) of cars. The large fraction of false alerts inside cars was thought to be caused by the odor plume that was distributed throughout the interior space. In the US, some DD handlers close car doors with their dogs inside and other handlers open all car doors to help localize the scent plume but there does not appear to be any research to verify either method. Misses were less than 8% except for the exteriors of cars (15%) where the large fraction of misses was thought to be caused by air turbulence around them.

Source: Data from Jezeirsky, T. et al. 2014. Forensic Sci Int 237:112–118.

Comparison of the results for explosives and drugs indicates that explo­sives are much more difficult to detect and locate compared to drugs based on 72% vs 88% correct alerts and 20% vs 5% false alerts for explosives and drugs, respectively. This may be caused by the much lower vapor pressures of explosives and subsequent reduced scent availability compared to drugs. However, the question remains open as does the reason for the large percent­age of false alerts by the ED teams.

3.3 Human Scent

Human scent from live persons consists of VOCs produced by bacteria acting on the skin and skin fakes (including skin fakes from breath), volatilization of secretions (oils) on skin and skin fakes, and VOCs from breath. Except for breath, this scent is carried upward by body air currents (Figure 5) and exits the body at the top of the head and on horizontal surfaces. The scent is carried away from the body by the wind which produces a human scent plume. It is this airborne scent plume that SDs detect. See Trails and Trailing for a detailed description of scent movement on and away from the body.

Figure 5 Drawing of the rising boundary layer and human scent plume from a person in calm air.

CDs search for decomposing bodies, body parts (large ones like a head or limbs to small ones like teeth or a piece of bone), bones, tissue, blood, body fluids, and trace (residual) evidence of these in all stages of decay. It is not feasible for handlers to obtain and store such a wide variety of aids; they usually resort to using a dozen or so types of aids in varying stages of decay.

Scent from human remains is emitted from the remains as gases or is readily volatilized from liquids and solids. There are nearly 500 VOCs from decomposing bodies that comprise the total VOC profile for them (Vass et al. 2004, 2008), but the specific compounds that elicit an alert by CDs are unknown. Some of the VOCs are present throughout the decomposition pro­cess and others are limited to certain stages of decomposition.

3.3.1 Decomposition

Decomposition processes include autolysis, putrefaction, liquefaction and disintegration, skeletonization, and diagenesis. These processes are repre­sented by stages of decomposition (e.g. fresh, early decomposition, advanced decomposition, skeletonization, and decomposition of skeletal material) that are largely based on the visual appearance of the remains (Rebmann et al. 2000; Vass 2001; Dent et al. 2004).

The effects of decomposition processes are to reduce the body to elemen­tal solids, liquids, and gases. These materials and gases evolving from them eventually pass into the soil and/or to the atmosphere where the gases can be detected. Bodies exposed to air decompose roughly eight times faster than buried bodies and about twice as fast as submerged bodies although many factors can modify these estimates. The slow decomposition rates of buried cadavers compared to those on the ground surface are generally a result of cool soil temperatures, soil moisture, lack of insects and scavengers, and other factors. For buried bodies, except for shallow graves and/or coarse-grained soils, decay is mostly anaerobic because of oxygen depletion in graves.

3.3.1.1 Autolysis Autolysis is defined as self-digestion. Without oxygen, cells begin to decay a few minutes after death (Vass 2001). Decay proceeds from within the cells with enzymes that dissolve the cells, eventually causing them to rupture and release fluids that are rich in nutrients. Bacteria in the respiratory and digestive systems multiply rapidly. Autolysis becomes appar­ent in a few days when fluid-filled blisters appear on the skin and sheets of skin slough of the body. When enough cells have ruptured, the nutrient-rich fluids become available and putrefaction occurs.

3.3.1.2 Putrefaction Putrefaction is the consumption of the soft tissues of the body by microorganisms such as bacteria, fungi, and protozoa (Vass 2001). The microorganisms feed on the nutrient-rich fluids produced dur­ing autolysis. Body tissues are converted into gases, liquids, and simple mol­ecules. There is a greenish discoloration of the skin and it is easily detached. The gases cause tissues in all areas of the body to swell or bloat (including organs and the circulatory system), which increases body volume. Gas and fluid accumulations in the intestines eventually purge from the body, usually from the rectum. Further, protein and fat decomposition produce putrescine, cadaverine, and other volatile fatty acids.

Saponification, freezing, and mummification are the processes that retard decomposition. Saponification produces a kind of soap under favorable con­ditions (warm and moist environments in the presence of bacteria) called adipocere, which is formed from the adipose fat layer lying just beneath the skin. It is a rancid, greasy, soapy, wax-like, and malodorous substance and its formation inhibits further decomposition. A saponified body recovered from a deep, water-filled South African cave 10 years after the person drowned looked like it was in good condition but smelled badly (Zimmermann 2006). Mummification is the process of desiccating tissue into a leathery form under conditions of dry heat or very cold temperatures and low humidity which tend to preserve the tissue. This process occurs naturally in deserts and Arctic regions. Freezing a body at the time of death preserves it by stop­ping the decomposition process. Permafrost (perennially frozen ground with temperatures less than 0°C (32°F)) underlies about 20 to25% of the exposed land surface of the earth (Osterkamp and Burn 2002). Bodies buried in per­mafrost for more than a century have been found almost intact, and animals buried in permafrost for tens of thousands of years have been found remark­ably preserved. Likewise, bodies encased in glacier ice for several thousand years have been found well preserved.

3.3.1.3 Liquefaction and Disintegration The body’s tissues and organs soften during decomposition and degenerate into a mass of unrecognizable tissue that becomes liquefied with continued decomposition. Liquefaction products may exude from the natural orifices. In the case of a body buried directly in the soil, a mucus sheath may form around the body consisting of liq­uid body decomposition products and fine soil (Dent et al. 2004). Ultimately, liquefaction and disintegration of the soft tissues leave behind skeletonized remains. At this stage, the skeletonized body will be held together by liga­ments and surrounded by a putrid, liquefying mass. Eventually the lique­faction products may be incorporated within percolating water and enter surrounding soil and groundwater.

3.3.1.4 Skeletonization and Diagenesis Skeletonization refers to the removal of soft tissue from bone. Skin, muscle, and internal organs are gen­erally lost to the environment well before a skeleton becomes disarticulated. In dry or freezing environments and in water where saponification has occurred, skeletonization may be incomplete. Bone may be broken down over time by diagenesis and physical breaking. Diagenesis of bone exposed to the environment refers to the alteration of bone composition by several complex processes (Dent et al. 2004).

3.3.2 Scent Sources

3.3.2.1 Chemical Compounds Decomposition odor increases with time and then decreases through the skeletonization phase until it may no longer be detectable by humans. During putrefaction, microorganisms destroy soft tissues in the intestines and elsewhere, producing gases such as hydrogen sulfide, carbon dioxide, methane, ammonia, sulfur dioxide, hydrogen, and fatty acids (Vass 2001). Putrescene and cadaverine are significant decompo­sition products that have been used to train cadaver dogs but are toxic and should not be used since other training aids are available.

VOCs emanating from buried cadavers in shallow graves that are not in coffins are under investigation in long-term experiments (Vass et al. 2004, 2008). These field experiments are valuable because the environmental conditions (soil, flora, fauna, moisture, thermal, and atmospheric) can add to, reduce, or eliminate VOCs from decomposition scent. Decomposition compounds occur in the following classes: acids, alcohols, halogens, ketones, aldehydes, cyclic hydrocarbons, sulfides, and nitrogen-containing com­pounds. The best represented class is cyclic hydrocarbons, with toluene and p-xylene being reported regularly. Putrescene and cadaverine have not been detected and extremely light VOCs (ammonia, hydrogen, carbon dioxide, methane) could not be detected with the methodology used.

It required 17 days for the first compounds to be detected at the sur­face from a 1.5 f burial depth with most compounds apparent after the first month. These findings have implications for training, evaluating, and search­ing with CDs (Vass et al. 2004, 2008). Depth of burial, class of compound, and season of burial influenced this timing.

Soil temperatures are also important and BADDs (burial accumulated degree days) are used to calculate the cumulative effects of variable soil temperatures over time. These factors are not all independent since BADDs depend on the depth of burial and the season of burial. A cadaver buried at a depth where the average soil temperature was 5°C for 1 day, 7°C for 1 day, and 10°C for 2 days would have BADDs = 32 degree-days for these 3 days. An estimate of average annual BADDs for shallow burials near Phoenix is about 8,000 degree-days. Estimates for areas where soils at the depth of burial freeze, must have thawed soil temperature data available to estimate BADDs since cadavers do not decompose at temperatures below freezing. Where bodies are buried in permafrost (e.g. near Fairbanks), BADDs would be near zero. These estimates can vary significantly from year to year, from site to site, and from region to region even when at the same latitude because of the influence of soil properties, elevation, and distance to the ocean or large water bodies on soil temperatures.

Thirty cyclic compounds that were detectable at the soil surface over bur­ied cadavers have been identified as key markers of decomposition and there­fore of importance for CDs. They were grouped according to those found throughout the time of burial, only early in the burial (<7,300 BADDs), and those persisting until all soft tissue was gone (<18,000 BADDs). Compounds present throughout the decomposition process would be desirable choice for training aids. However, there is still too little known about the decomposi­tion compounds that dogs use to detect human remains to state this as more than a hypothesis.

The bad news is that this group of compounds (including toluene, eth­ane, sulfides, methane, aldehydes) is not unique and can be found in many outdoor samples taken virtually anywhere. Their ubiquitous presence may account for some of the false alerts by cadaver dogs.

3.3.2.2 Compounds in Near-Surface Soils A global study of the near-surface chemical compounds of soils associated with graves (60+ years old) and the downslope surface and subsurface plumes in different environments identified more than 50 human decomposition compounds detectable in the near-surface soils as shown in Table 2.5 (Vass 2012). Their presence depends on environmental conditions and their properties (sol­ubility, density, molecular weight, etc.) determine whether a compound is detected in close association with the remains (corpse) and/or in the downslope chemical plume.

The vapor pressures of the VOCs in Table 2.5 are high, which indicates that their volatility is high and that dogs could easily detect them if the grave soils are sufficiently porous to allow the scent to reach the surface and they are not removed by chemical or biological processes in the soil.

Decomposition in the grave environment was found to be aerobic (involves oxygen) if it was a surface or near-surface event, the remains were loosely wrapped in clothing, in a non-airtight container above ground, or covered by <0.3 m (1 f) of loose material. The environment was anaerobic (doesn’t involve oxygen) if the remains were buried a minimum of 0.61 m (2 f) deep in clay textured soils, 0.76 m (2.5 f) in peat or silty textured soils, 1.07 m (3.5 f) deep in sand or loam soils, or the remains were wrapped or placed in an airtight container or matrix. These results indicate that training aids for CDs used to detect graves should include materials that have decom­posed both aerobically and anaerobically.

Some compounds tend to concentrate near the remains rather than in the plume as the gravesite ages. Thus, when searching for clandes­tine graves, the compounds that appear predominantly near or at the corpse as opposed to those present in the plume may be more valuable as training aids. These include benzene, some fluorinated halogens, sulfur compounds (dimethyl disulfide, dimethyl trisulfide), and a few aldehydes (heptanal, hexanal, nonanal, octanal). Using this information requires that we know which VOCs are present in training aids used for CDs. None of the VOCs associated only with human decomposition (Table 2.5) were present in a study of training aids (Hoffman et al. 2009). Unfortunately, not enough is known about the effects of age, storage conditions (tem­perature, humidity, aerobic or anaerobic decomposition), and other fac­tors on training aids to predict the specific VOCs that would be present in the aids.

Source: Data from Vass, A.A. 2012. Forensic Sci Int 222:234–241.

3.3.3 Live vs Dead

Decomposition scent differs from live scent since it is thought to be generic. The scent specific to an individual is eventually replaced by decomposition scent consisting of elemental compounds thought to be common to all cadav­ers since it involves the decomposition of proteins, fats, and carbohydrates. It has been shown there is less variation in VOCs among recent cadavers than among the living, and decomposition may reduce the differences among cadavers (DeGreef and Furton 2011). However, each living person’s scent is unique, specific to the individual, which indicates that some of the chemicals that make up scent may be controlled by the individual’s genetics. It is also known that dogs can detect bones, teeth, and hair in older burials and that these materials retain their DNA. This raises the possibility that dogs may be capable of distinguishing between individual cadavers (skeletons) by the unique scent of their bones, teeth, and hair.

The time frame for decomposition to occur is highly variable and depends on temperature, availability of oxygen, moisture, types of available fauna (bacteria), soils, diet, medicines, obesity, and other factors. Typical times required would be a few weeks to a few months. Under hot and moist condi­tions, the time can be significantly shorter. At the time of death, breath and metabolism cease but scent from glandular secretions and bacteria remain for an undetermined period. Dogs appear to be able to recognize when a person is dead, but it is not known how they do this. They may be detecting a lack of visual cues of life, lack of breath in the scent bouquet, a scent associ­ated with the death of cells, and/or some other effects. Some nurses report that people who are about to die have a certain smell, but this does not appear to have been verified.

Carrion flies, when present and under favorable environmental condi­tions, have been observed to appear on a corpse in a matter of minutes after death (Haskel et al. 1997). The flies are attracted to the nose and mouth most likely because of the odors emanating from these two sites that attract them. These observations suggest that there may be a scent associated with death, an “odor mortis,” that appears within minutes of dying or possibly before.

In response to a suspected murder investigation, an attempt was made to answer two questions raised by law enforcement (Oesterhelweg et al. 2008). How long would a deceased body have to be in contact with a material for the scent to be detectable by a CD, and how long would detectable scent remain on the material? Carpet squares were placed under two cadavers wrapped in a thin cotton blanket at postmortem intervals < 3 hours and removed after 2 min for one and 10 min for the other. Tree CDs alerted on the carpet squares up to 35 days (2-min exposure) and 65 days (10-min exposure). The authors also concluded that a person dead for 2 hours could be detected by CDs.

3.3.4 Cadaver Scent and Discrimination

CD handlers have long believed that dogs can discriminate between human and animal remains, and VOC profiles from human and animal remains have been shown to be different (DeGreef and Furton 2011). CDs can detect blood from cadavers at concentrations of 1 ppm and can distinguish it from distractions including blood in urine, swine blood, and dog menstrual blood (Riezzo et al. 2014). CD handlers also believe that dogs can discriminate between fresh human and animal blood stains and this has been confirmed in two instances by laboratory analysis (Osterkamp unpublished). Human blood stains on automobile upholstery can be detected for a decade under favorable conditions (Osterkamp unpublished).

3.3.5 Specific Training Aids

Desirable training aids for CDs would be bodies, body parts, bones, blood, and other body fluids in various stages of decay including saponification, mummification, and under different decomposition conditions (e.g. wet, dry, aerobic, anaerobic). Collection and storage of these aids would be a daunt­ing task, so handlers must resort to using human remains that may be law­fully obtained. Handlers obtain human remains for training aids from other handlers, coroners, medical examiners, morgues, hospitals, and elsewhere. Several universities in the US, Canada, and Australia, offer classes for CDs that include detection of full body cadavers (i.e. the full VOC scent profile for deceased humans).

An assumption in the use of aids is that the dog will generalize from the limited number of compounds that it is trained to alert on. Also, it will per­form its alert when encountering larger cadaver parts or the whole cadaver with the full VOC profile (as modified by local conditions). The basis for making this assumption is not well established. Handlers compensate by using as many different training aids as possible. Since these training aids are limited, it is unknown if they contain sufficient components of the VOC profile to enable the dog to generalize to the odors that they encounter when finding human remains. However, while we do know that CDs trained on a few small amounts of human remains do find bodies, it would be desirable to have a better understanding of what the dogs smell and what is needed to train them.

Historically, CDs trained with putrescene or cadaverine have been suc­cessful in finding cadavers in the field (Rebmann et al. 2000). These com­pounds were not found at the ground surface in the grave study. Studies of EDs suggest that dogs trained on limited varieties of smokeless powders were only able to reliably detect the specific types on which they were trained. Buried bodies may present more difficult conditions because the soil environ­ment (which includes chemicals and bacteria in the soil) can add to, remove, and/or modify the VOCs emanating from a decomposing buried body.

In a study of training aids contributed by CD handlers (Hofman et al. 2009), 33 VOCs occurred in the headspace of 14 decomposing training aids that included blood, tissue, skin, fat, adipocere, bone, and teeth. No informa­tion was given about the storage conditions (e.g. aerobic, anaerobic, tempera­ture, humidity) during decomposition of these aids. Comparing these results to 30 key VOCs present at the ground surface over graves (Vass et al. 2008), only four of the compounds (dimethyl disulfide, toluene, nonanal, and tet­rachloroethylene) were found in the training aids. It is of interest to cadaver dog trainers and handlers that the first three compounds occurred in only one of the training aids, body fat attached to skin. However, the effects of decomposition and environment on the compounds emanating from body fat and skin are not known. Tetrachloroethylene was found in blood, muscle, skin, adipocere, bone, and teeth.

None of the 14 training aids contained VOCs known to be specific to human decomposition only (pentane, carbon tetrachloride, decane, undec­ane) (Table 2.5). The training aids did contain seven of the compounds found in the global grave study. Six were found in bone; five in muscle; four each in blood, body fat attached to skin, adipocere, and fat; and three each in skin and teeth.

Disarticulated bodies can be widely scattered by animals, dispersed downhill under the influence of gravity, and transported by water. Since searches may be conducted specifically for cadaver parts that can yield DNA (bones, teeth, and hair), it is desirable to store training aids for them sepa­rately. Bones can be obtained from companies that sell bones and even com­plete skeletons although little is known about how these were processed and handled. Some appear to have been chemically cleaned. Teeth are especially important since they are commonly used for identification of the body. In some states, dentists may be able to provide teeth that have not been cleaned or autoclaved. Hair can be obtained from barbers, hair dressers, and friends.

The large numbers of VOCs produced during decomposition and the lack of information on what compounds the dogs use to detect decomposi­tion scent pose significant problems for developing training aids for CDs. A viable alternative to using human remains for training aids would be to transfer scent from the remains to a material that would become the training aid. DeGreef and Furton (2011) used the STU-100 (with Dukal cotton gauze (Dukal Corporation, Syosset, NY) for the scent collection pads) to obtain VOCs from cadavers in a morgue and a crematorium, and from gauze con­taining decomposition fluid, adipocere, bone residue, or blood. These scent pads were stored in glass jars and used later in tests of trained CDs. Te majority of dogs alerted on the aids with different concentrations and from different odor sources in every scenario tested.

Another way to collect cadaver scent would be to place sterile gauze pads in contact with a cadaver for a length of time similar to the method used by Oesterhelweg et al. (2008). The experience of trailing dog handlers indicates this would be about 10 to 30 minutes. Store the pads in clean, sealed glass vials or jars for later use. The method requires access to cadaver sources in different stages of decay but does not require an expensive STU-100 unit. This method is not intrusive so that morgues and crime scene investigators may be willing to allow this type of scent collection. The advantages are that handling the pads would not be as hazardous as cadaver materials and it does not appear that there are legal restrictions to possessing them.

Artificial scents such as Sigma Pseudo Corpse I and II (Sigma Aldrich Chemical Corporation) have been used to train cadaver dogs. While the scents appear to have a limited scent profile, CDs trained with them have found bodies (Rebmann et al. 2000), but it is not known if the handlers observed a full alert or a CB (change of behavior). Training aids have been made from chemical solutions of the most commonly occurring VOCs found in human remains. These aids were used in trials with cadaver dogs, but the dogs showed inconsistent interest in them.

3.4 Other Sources

It appears that training aids for most other types of SDs rely on some of the source material to communicate to the dog the scent that results in being rewarded. A novel training aid for SDs used to search for bed bugs was devel­oped by immersing the bugs in solvents (pentane, acetone, methanol, and water) (Pfester et al. 2008). It was determined beforehand that the dogs did not alert on these solvents. The solvents were then sealed in containers. For training, 1 ml of the extract (equivalent to five bed bugs) was placed on fil­ter paper inside a vial. All dogs trained on bed bugs alerted on the pentane extract but not on the other extracts indicating that the scent of the bed bugs was transferred to that solvent.

Training aids for dogs used to detect human diseases, especially types of cancer, have used breath from affected individuals, even though the cancer may not directly affect their respiratory system (Lippi and Cervellin 2011).

3.5 General Recommendations

Research is currently being conducted to develop training aids for EDs, DDs, and CDs so that any recommendations may soon be out of date. ED and DD handlers usually have access to scent kits and source materials from their organizations. Given the success of the STU-100 for making training aids for CDs, it would seem obvious to use it for making aids for EDs and DDs. CD handlers generally use whatever scent sources they can obtain. The STU-100 is too expensive for many teams, and until new information and/or products are available, handlers will need to continue using a wide variety of cadaver materials in different stages of decay or scent pads that have been in contact with them. Commonly used aids include placentas with blood and body fluids, skin, tissue, blood, adipocere, body fat, bone, and teeth. Soil from under a decomposing body (hanging, on the surface, or in a grave) and clothes or articles that have been in contact with a decom­posing body can be also used. Body fat attached to skin, blood and body flu­ids, muscle, skin, bones, and teeth may be especially important in searches for clandestine graves. It is important to expose CDs to aids of varying sizes and types.

3.5.1 Storage

Glass containers with glass stoppers that are not exposed to ultraviolet light have been shown to offer greater stability for stored human scent samples and probably for other scent samples as well (Hudson et al. 2009). Glass jars with plastic, perforated lids have also been used. If canning jars are used, decom­position chemicals and chemicals from other sources may cause the metal lids to rust. Rust can be detected by dogs (Schoon et al. 2014) and its presence contaminates the stored scent source. Freezing training aids for long times can cause them to desiccate, possibly changing their properties, because the duty cycle of a freezer causes moisture to migrate from the source to the inside walls of the storage container.

Training aids for all SDs should be carefully handled, stored, and trans­ported to prevent contamination. Contamination can occur by transfer of trace fluids, microparticles, and VOCs from the aids to other aids and surfaces nearby. It can also be a result of background scents in the location where they were prepared, contact with humans (primarily from skin secre­tions, perfumes, lotions, smoke, skin fakes), type of clothing and gloves used by the preparer, the environment (vehicle fuel and fumes, oils, grease, buildings, sewage) and animals present such as pets and other SDs. Cross contamination can result when aids are in close proximity while in semi­permeable containers (e.g. paper or canvas bags) during handling, shipping, storage, and placement. Cross contamination can also occur if two different drugs or explosives are handled with the same pair of gloves. The problem is that it may be thought that the dog is being trained for one source, but the dog is alerting to the presence of another.

Handling aids with clean tongs or forceps while wearing clean medi­cal examination gloves helps to keep human and other scent of them. SD handlers often use these gloves to handle their aids. However, if the outside of the fingers of the gloves are touched by their skin when putting them on or if any object is touched that has previously been touched by bare hands

(e.g. door knob, scent container) then human scent will be transferred to the gloves and aids. Do not hide training aids where dogs can touch the aids with their wet noses or mouth them. Training aids hidden for a long time or repeatedly in the same place may cause the location to be contaminated with residual scent.

It can be argued that ED and DD training aids (explosives and drugs) will commonly contain the scent of the people that packaged them (from skin fakes, secretions, breath, lotions, perfumes, etc.), so that human con­tamination should not create problems. However, limited information indicates EDs may prefer the scent of one person over that of others to the point that it influences their success during training (Goldblatt 2017). Te question may be more important for training than for searches, but it remains open.

4 Scent in Training

4.1 Scent Availability

Scent availability refers to the quantity of scent (concentration of scent mol­ecules) available for a dog to detect. It has usually been assumed that scent availability depends on the amount (volume or mass) of the source, but there are other factors. Lotspeich et al. (2012) have shown that scent availability in containers also depends on the container volume, explosive vapor pres­sure, and temperature. The primary consideration for reliable detection of a source by SDs is for the headspace of the container to be saturated with scent molecules volatilized from the source. Once the headspace is saturated any increase in the amount of sample will not result in an increase in scent availability.

Calculations and studies with EDs have shown that the amount of sample that is needed to saturate different size containers normally used for train­ing is relatively small. For nitromethane, nitroethane, and nitropropane, the amounts recommended to saturate a container and allow for stable concen­trations over time are 10 μl (microliters) for 20 ml (milliliters) headspace vials, 100 μl for quart cans and 1000 μl for gallon cans (1000 μl = 1 ml = 3.38E-2 oz).

Generally, factors influencing scent availability include the charac­teristics of the container (scent containment, volume, materials) and the source (phase, type, amount, surface area, diffusivity and vapor pres­sure (includes the effects of temperature)). Adhesion of some VOCs and microparticles to the container surface may also be an important factor (Schade and Goldstein 2001; Goldblatt 2017). Scent availability depends critically on the degree of scent containment provided by the container (packaging). Source containers can be completely closed, diffusion lim­ited, partially open, or open.

Completely closed containers do not allow scent molecules to pass through their walls. Common examples include those made of metal (anti­personnel mines), glass, tires, and some plastics. While plastics generally allow gas transport through them, from an operational viewpoint the flux rate of VOCs through the container material does not have to be zero; it just needs to be below the limit of detection by dogs. A potential problem is that the VOCs of some sources may degrade the container material (e.g. corrode some metals, chemically break down plastics).

Diffusive containers are those that allow scent molecules to pass through the container material by diffusion or have such small openings that scent escapes through them by mass flow, but very slowly. Diffusion is such a slow process that it becomes important only over long time periods or with very thin container walls. Slow mass flow (such as through small holes) can appear to behave like diffusion but the scent flux would be much larger than for diffusion. Examples of diffusive containers include thin plastic and rubber packaging (food storage bags, balloons) that allow scent molecules to diffuse through their thin walls. Windproof clothing consists of materials that allow a small amount of air and scent to flow through them but restrict the passage of wind (they are not truly diffusive).

Diffusive training aids have been developed for explosives, drugs, cadavers, and other scent sources called COMPS (Controlled Odor Mimic Permeation System) (Furton and Harper 2017). In use, some of the source or a mimic is placed in the solid form or, for a liquid, on a sterile gauze pad in a plastic bag that is then sealed. The thickness of the bag is chosen to allow scent to diffuse through it at a relatively constant and reproducible rate that produces a fixed amount of scent available to the dogs. Scent availability can be controlled by the diffusion rate of the scent molecules through the bag which is determined by the type of material, thickness of the bag, and the properties of the scent molecules. The bags can be stored and used as train­ing aids later. Additional details and variations of the method for making training aids for EDs, DDs, and CDs can be found in Macias (2009) and DeGreef (2010).

Partially open sources are those that have restricted access to the exter­nal air. These include unpackaged sources in containers that are open at the top (cans and jars) and containers that have significant holes in them (lug­gage, boxes, back packs, canvas bags). The National Odor Recognition Test (NORT) for explosives is such an example. It has the source in a small can with holes in the lid that is placed in an open quart can that is placed in an open gallon can.

If more scent is available than needed to completely saturate a par­tially open container, the excess scent may overflow the container and flow to low levels in the surrounding area because the VOCs are typically heavier than the air. This means that in training, testing, and searching, SDs should be encouraged to search the lowest areas around a known or suspected position of the source. For sources in closed rooms, scent would be expected to exit at the bottom of the doors, and for luggage to be pres­ent at the bottom edge. In outdoor areas, scent pools in depressions near the source until turbulence mixes it with the air and it is transported away by the wind.

Completely open sources are those that are exposed with unrestricted access to the external air. Examples are sources attached to surfaces (explosives) and those lying on the ground surface (spilled gunpowder, shell casings, drugs, bodies, and teeth). Scent availability depends on their size, surface area, and volatility (vapor pressure which depends on tem­perature). On volatilization from the source, scent molecules enter the air where any wind moves them away from the source in a scent plume detect­able by dogs.

It is possible that the scent from a training aid may remain constant with increasing size of the source, but this may not be true for all sources. For some sources, the perceived nature of the scent is thought to change with size. CD handlers have observed that cadaver dogs trained on small sources may fail to indicate a body or may not perform their trained alert on one. A similar situation seems to occur with EDs and DDs that encounter very large sources.

The performance of CDs seems to improve once they have found a body or two. So, it is desirable to use large sources to train the dog although these may be difficult to find. One solution may be to put all the sources used by a group of handlers into an open or ventilated box and work the dogs on it at a group training or seminar. This may not be the best procedure due to the differing ages of the sources, cross contamination of the sources, and other factors.

A curious fact is that agency certifications generally use relatively small sources. The certified teams are then deemed qualified to search for large explosive and drug sources and for bodies without ever having demonstrated in either training or testing that they will alert on them. The assumption that, if the team can find and alert on small sources, then they can find and alert on large ones, does not appear to have been evaluated.

4.2 Use of Aids

4.2.1 Preexposure and Prescenting

Preexposure is exposing an untrained dog to a source prior to initial training. A kind of preexposure occurs with some training methods that use a scented toy as the first step in training. Hall et al. (2013) have shown that preexposure increases the success rate during initial training. Prescenting is exposing a trained dog to a source prior to performing a search for that source (Papet and Minhinnick 2016). Prescenting in the present context is used by scent discriminating trailing dog handlers prior to starting a trail to communicate to the dog the person the dog is to follow. It has also been used by a few CD handlers prior to land and water searches, especially when the search is for specific items (hair, bone, teeth) or for specific stages of decomposition.

4.2.2 Type of Reward

Play/retrieve as a reward (reinforcement) has been widely used. It is thought that dogs trained with this method are highly motivated to find scent. However, some agencies use food and select dogs for their food drive (Department of Agriculture). Food rewards require less skill to deliver, allow more trials per day, and it is easier to find food-motivated dogs. Some dogs are trained every day with their food reward for finds being their only source of food. A potential disadvantage is that in some situations (disaster, crime scenes) food at the site may become a distraction and even ingested by the dogs. Food rewards used in training are often dropped by the handler or the dog so that any dogs working the same hide afterward may be distracted by the food and ingest it.

Using multiple rewards during training can be helpful when working the dog in hazardous (along a highway) or difficult (steep terrain) conditions. Rewards can include food, praise, touch (petting, scratching), games (keep away), tugging, retrieving, combinations, and others. The handler then has the option to select the reward best suited to conditions.

4.2.3 Number of Aids

During the initial training, the number of aids hidden from the dog (hides) should be large to establish initial discrimination and variable to keep the dog from getting used to finding a fixed number of aids (Goldblatt et al. 2009). As the dog becomes reasonably proficient at detecting and locating the scent and the handler proficient in working the dog, the number can be systematically reduced. For example, a line of ten blocks could contain 3 to 4 sources initially and eventually be reduced to 0 to 2.

It is necessary to use a null line (no source present) regularly. It appears that dogs that always make a find each time they are trained are more prone to false alerts. For an operational dog, the number of aids needs to be large enough to keep the dog motivated to search. It is usually recommended (e.g.

J. Joyce in seminars) for dogs used to search large areas to occasionally search a large null area.

4.2.4 Reward Schedule

During initial training, SD trainers generally recommend rewarding the dog for every correct response. However, most SD handlers continue to reward their dog in training every time it detects, finds, and alerts on a source. They do not reward them for a positive response on a search since it is not known if a source is present. For SDs, it is desirable to use a variable schedule of reinforcement in training to condition them to not having a reward every time for a find (Goldblatt et al. 2009). A typical reinforcement schedule in training might be 80% (4 of 5) of the correct responses, although there are considerable variations between trainers. SDs training on multiple sources (e.g. graves in a cemetery) may be worked on 5 or more sources before they get their reward. However, their handlers must be acutely aware of any reduction in focus and motivation to avoid developing problems with their alert.

4.2.5 Multiple Component Sources

SDs may be required to find many different sources with each source consist­ing of multiple chemical components. For example, EDs may be required to find different types of explosives, DDs, different types of drugs, and CDs, different body parts. Each one of these sources consists of multiple chemi­cal components. Two distinct methods have been developed for introducing these sources to EDs and DDs (Goldblatt et al. 2009). One combines sources to produce a scent mixture that is introduced to the dogs. Once the dogs rec­ognize and alert on the mixture, the sources are separated for the rest of the training and never recombined. However, Goldblatt et al. (2009) recommend introducing each source separately either sequentially or during the same time frame.

4.2.6 Distractions

Distractions include sights, sounds, and scents that distract a SD while work­ing. A SD is not fully trained until it works reliably in different places in the presence of distractions. Other dogs or animals that SDs see or hear while working are common distractions. Distraction scents are any non-source scents that cause interest or an alert by the dog which confuses the issue of whether a source is present. Blood, bones, and other remains of animals, fish, and birds can be distractions for CDs. Food, fish bait, trash, animal scents, and human and animal urine and feces are other common distractions encountered on searches (there are many others). SDs should be discouraged from showing an interest in or alerting on distraction odors although some are problematical (feces, urine, and semen for CDs) as noted previously. This can be done for any SD by routinely including distractions in yard work and in field problems and discouraging any interest in them. Distractions may also have an impact on the handler and anything that distracts the handler may be expected to influence handler-dog interactions and thereby affect the performance of the team. Tus, the handler should also become accustomed to training in the presence of distractions that may be encountered during searches. A potential distractor is the state of stress of the handler which increases their anxiety and may result in a change in the dog’s performance during testing or searching (Zubedat et al. 2014).

This idea was evaluated with handlers of military EDs. Superior officers of the handlers told the handlers prior to a test that they were to be reassigned and/or were present at the tests where they pointed at the han­dlers and appeared to write notes about them. The results showed that the handlers’ stress decreased their attention and elevated their anxiety level but increased the activity level of the dogs. Surprisingly, this improved the dog’s performance as shown by the reduced time to detect the explosives. It was thought that the external stress disturbed the handler’s focus which resulted in reduced control of the dog and allowed the dogs to work in a less handler-dependent manner. These results emphasize the importance of SD-handler interactions in all searches and suggest that reduced control of the dog (e.g. of leash vs on leash) may be desirable. Trainers have long recognized this and often recommend that handlers “let the dog work” (i.e. reduce their control).

4.3 Source Placement

If a certain type or part of a regularly used search area never has a source, a dog’s motivation to search it and the dog’s POD (probability of detec­tion) when a source is present may be reduced. In a study of EDs (Gazit et al. 2005b), experienced handlers and dogs trained on two different roads on alternate days for 20 days. Road A had sources every time and road B never had a source present. Sources were then placed on road B at a frequency of one every third day, but the dogs detected only about 50% of them whereas their success on road A was much higher (about 90%). The dogs were then retrained on road B with sources placed there every day but the dog’s moti­vation to search road B never recovered. Additional research confirmed this result (i.e. reduced motivation and reduced ability to detect a source in an area despite continued reinforcement).

These results appear likely to apply to other SDs. It is suggested that SDs be trained in as many different environments as possible and that source placement be varied so that the dogs find sources in all types and parts of the areas being searched, especially regularly used training areas. This includes buildings with different types of rooms (e.g. bathrooms, bedrooms, kitchens, living rooms, etc.) and outdoor settings with different terrain and vegetation. A way to avoid problems here would be to keep a record of where aids are placed in each area, to regularly review this information, and to make sure that all subsets of an area are used.

4.4 Potential Problems

Papet (2016) has reviewed the use of training aids for DDs and EDs, includ­ing aspects of selection, packaging, handling, storage, and transporting aids. Contamination can occur between aids stored in proximity and by human scent in all aspects of use. CD handlers have observed that their dogs may not come to a full alert on aids used by other handlers which probably indicates contamination. Aids can be contaminated by materials in the area where they are placed as hides and can also contaminate these areas with residual scent as noted (Jezeirski et al. 2014). Cross contamination of sources during use, especially storage, by VOCs and microparticles can be a potential prob­lem for SDs. This indicates that it is desirable to store training aids separately to avoid cross contamination.

Contaminating aids with human scent can occur during packaging and handling. Using medical examination gloves that are put on by handling the wrist section only and clean metal tongs can be helpful. Skin fakes fall­ing on an aid or its container and particulates from sources are problems that are difficult to eliminate. Complete separation of all types of sources during all phases of use is desirable. The primary handler problem appears to be contamination with skin secretions since we leave scented skin oils (fingerprints) on everything we touch. An example in training occurs when a container contaminated with live human scent is used to package the training aid.

In addition to scent from the surroundings, there are seven possible scents: source, container, human, and their four combinations. We would like the dogs to alert on the source scent and any combination that includes it but not on the container or human scent or both. However, experience with EDs, DDs, and CDs shows that dogs will often alert on the container with live human scent on it and no source. This makes it important to include containers with human scent on them, but no source, in yard and field work and to ignore interest or alerts on them. Since contamination and deteriora­tion of training aids due to ageing and weathering can create training prob­lems, Papet (2016) recommends that training aids for EDs and DDs be put on a regular replacement schedule where they are cycled sequentially through usage conditions such as initial training, testing, routine training, training in contaminated environments, and then replaced.

As noted above, not enough is known about the scents from specific sources that cause a dog to alert, and the initial training of many SDs is done with training aids with an unknown scent profile. Special care should be taken at this stage to make sure that the aids are not contaminated (Goldblatt et al. 2009; Papet 2016). It is also necessary to use many variations (types, sizes) of the aids to help the dogs generalize and to eventually train them on the sources found in the field of operations.

Once scent recognition training has been done and the dog has an iden­tifiable and reliable alert that the handler is proficient at recognizing, it is important for the handler to start running blind problems regularly, includ­ing null problems. The reason is that, if the handler usually knows the loca­tion of the aid, they are likely to give involuntary cues to the dog. Cueing is a major problem when training SDs (Ensminger and Papet 2011a and b). When training on a line of scent containers, handlers may alter their pace, body posture, or eye position at a container with scent which cues the dogs to the location of the scent. In working field problems, handlers are likely to become lax in the control of the dog while searching until in the vicinity of the aid and this too is a cue to the dog. Furthermore, if the handler places the aid and happens to be in a hurry, there is a tendency to hurry the search, not covering the area thoroughly, and effectively lead the dog to the source to get finished, a sure recipe for failure since these actions are the opposite of what must be done during a real search. One of the most important factors when training SDs is to cover the area thoroughly just as in searching.

During training, it is important to make sure that the dog responds to the scent of the aid and is not responding to human, container scent, their combination, or to any other possible contaminant odor. Varying the type, size, and placement of the aids; running blind problems; and having distrac­tions present (scent, sight, and sound) are important.

SDs often learn to trail their handlers or others who place the source, not only in field and forest settings, but also in buildings. Some strategies to prevent trailing to the source are to throw the source (no scent trail), to criss­cross the area with a bewildering array of human scent trails, and to enter all the rooms in a building. For buried sources, it is desirable to place the sources several weeks before the dog is used to find them; sources buried for weeks to months are much more realistic for training. This requires a large area dedicated to training and several buried sources. Experience shows that dogs can remember the location of sources if they are worked more than every four months on them. Care in burying the sources is necessary because of possible ground surface contamination from source liquids or particles that the dogs can easily detect. Cadaver sources may need to be placed in a wire mesh cage for shallow burials (<0.30 m or 1 f) since roaming dogs and wild animals may dig them up. Excessive digging by CDs during training usually requires a long time for the site to return to its natural condition. With care, sites can be useful for years. A site with a placenta and a large handful of hair buried <0.30 m deep in 2005 can still be detected about 50% of the time 13 years later.

Assuming they are properly trained, the most common reasons for SDs not finding a source during searches are (i) the handler missed the change in behavior in the dog and (ii) the dog was never in scent, both handler errors. On searches, SDs that have never found the type and size of source and the amount of scent available from a source may not give their alert but rather some other change in behavior (CB). Consequently, handlers must be aware of changes in the dog’s behavior and be able to interpret them. If the dog was never in scent, it is likely a result of poor search methods during training. It is very important during all training to cover the search area thoroughly. Blind trials help to make sure of this. Variable winds may require changes in search plans to cover an area thoroughly. The idea of training like you would search is especially important in SD training. The dog should not be able to tell if they are training or searching. This is difficult to do because the dogs are so expert at reading the handler’s emotional, mental, and physical states.

When a dog or handler works harder and longer during a search than during training, the dog is more likely to give a false alert and the handler is more likely to make mistakes. A dog or handler that has only worked 30 min problems cannot search effectively for several hours. Trainers (e.g. J. Joyce in seminars) recommend that “nose time” (the amount of time that the dog and handler can work continuously and reliably) be determined during training. It is also important during training to condition the dog and handler for the longest searches they will be asked to do in the terrain or physical settings where they will do them. Occasionally doing long unknown null problems and problems where the dog will not make a find until after several hours of searching is helpful. For known problems, bringing a dog back to the area of a source repeatedly or hanging out there encouraging the dog to search and make the find may result in a tendency (it is a cue to the dog) to false alert.

A common question of SD handlers is how often the dogs should be trained. Research shows that daily training and encounters with scent sources by SDs improves their ability to recognize and discriminate between sources and background scents and lowers their threshold of detection (increases their sensitivity) for detecting weak sources.

Some agencies recommend 16 hours of training per month in their gen­eral guidelines but do not say if this is actual training time or includes travel or other time. If this is training time only, small-area searches for buried sources and some structure searches typically require about 30 min which would be more than 30 such searches per month, an improbable number. Some arson dogs are trained daily, and some EDs not worked for 3 days must be reevaluated before going back into service. Clearly, training almost daily is desirable, but few teams can do this using field problems entirely. Experience suggests that a mix of yard work (blocks, boxes, carousel, wall) and field training totaling four or five times a week seems to keep CDs and handlers sharp in detecting faint scents.

Do not cross -train SDs to detect articles with live human scent on them unless that is the application you desire. You may have law enforcement or other agencies dig up or recover some buried clothes or shoes instead of the desired source. Do not train CDs in cemeteries unless it can be verified that the bodies were not embalmed with formaldehyde. Plywood manufactured before the mid-1990s was made with glue containing formaldehyde. A dog trained in a cemetery on graves with embalmed bodies may alert on disinte­grating plywood made with formaldehyde glue. It appears that the emissions from the glue are similar enough to embalming fluid for the dog to generalize and alert.

5 Summary

Scent molecules are volatile organic compounds, VOCs. Scent is the dog’s perception of these molecules as a result of processing by the brain. During sniffing, exhalation causes air jets to be directed to the sides of the dog’s nose and downward. These air jets can dislodge particles of dust, skin fakes, and adsorbed VOCs from surfaces which can then be inhaled. Scent molecules on dust and skin fakes can be liberated in the dog’s warm and humid nose and then be detected by the dog. It may also be possible for the humid air jets impinging on dry soil surfaces to free VOCs adsorbed on dry soil particles to be inhaled for detection of buried sources such as land mines. The warmth of the air jets may also help free VOCs adsorbed on surfaces allowing them to be inhaled and detected.

Sources emit scent by evaporation, sublimation, and desorption. These processes are collectively referred to as volatilization which depends on the vapor pressure of the source which depends strongly on the temperature. For example, the amount of scent volatilizing from TNT increases about ten times when the temperature increases from 77°F to 104°F.

Drugs and explosives produce “sticky” microparticles that adhere to sur­faces and humans produce skin fakes that come to rest on surfaces where they can be dislodged by air jets from the dog’s nose to produce a detectable scent. This sticky nature of microparticles can contaminate nearby surfaces such as other training aids, dust particles, rooms, and people. For detection of explosives with very low vapor pressures, stickiness may be as important as vapor pressure. The ability of dogs to dislodge these particles by sniffing and inhaling them may account for the ability of EDs to detect explosives with very low vapor pressures such as RDX.

VOCs from explosives, drugs, human remains, and live persons are heavier than air and are expected, in the absence of wind, to drain down­ward and settle in low areas (depressions over graves, ditches, floor drains, car floors). In outdoor settings, scent can pool in these areas and, under calm and cloudy conditions, can remain there until removed by turbulent air con­ditions. This suggests, when there is no wind or there is a downslope flow, it would be desirable to search for buried explosives and other sources on a slope by gridding across the slope and advancing the grid up the slope.

It is not always necessary to use parent compounds for training EDs and DDs because the dogs use the most abundant chemical component(s) in the vapor headspace. Dogs trained on aids developed from explosives and drugs manufactured by one process may not alert on similar explosives and drugs produced by another process. For example, EDs trained on TNT that con­tains DNT may learn to detect the more volatile DNT. Russian, other manu­factured TNT, and some weathered TNT does not contain DNT, which may result in a failure of the dogs to detect the TNT.

Residual scent from sources can remain in hides for days or much longer and possibly result in alerts where nothing is found. Care must be taken to allow time (several days or more) for residual scent to dissipate when using areas (buildings, vehicles, field sites) repeatedly during training and testing. It is also desirable to train dogs to offer their alert only when the source is present.

Favorable times to search for sources buried in dry soil are when humid­ity is high, or when the ground has been recently wetted by dew, very light rain, or mechanically misted.

Variability in scenting thresholds by different dogs indicates it would be desirable to test the ability of candidate SDs to detect low scent levels in order to select a dog with a suitable nose.

SDs that were not physically conditioned showed a 64% reduction in olfactory acuity following a physical stress test. A moderate physical condi­tioning program can help SDs maintain their scenting ability during periods of intense work.

CDs can detect blood from cadavers at concentrations of 1 ppm and can distinguish between living and dead humans, human and animal remains, cadaver blood and swine blood, and between cadaver blood and dog men­strual blood. Human blood stains on automobile upholstery can be detected for a decade under favorable conditions.

Body fat attached to skin, bones, blood and body fluids in various stages of decomposition and aerobic and anaerobic decay, adipocere, and mummi­fied tissue are especially important training aids for CDs.

Glass containers with glass stoppers not exposed to ultraviolet light, glass jars with plastic, perforated lids, and canning jars are desirable (in this order) for storing aids. Scent availability from aids depends on the amount; presence of source particulates; vapor pressure; temperature; size of the containers; and whether the containers are closed, partially open, or open. Humidity can also be a factor.

During initial training, SDs should be rewarded for every correct response. A variable reward schedule is phased in for later training. SDs are not fully trained until they work reliably in different places in the presence of distractions (scents, sights, sounds). If a certain type or part of a regularly used search area never has a source, a dog’s motivation to search it decreases significantly and may never recover.

SDs and handlers should be conditioning for the longest searches they will do in the weather, terrain, and environments where they will search including occasional long unknown null problems. Do not cross train SDs to detect articles with live human scent unless they will be tasked to search for it. Daily training improves SDs’ ability to recognize sources, discrimi­nate between background scents, and lowers their threshold of detection for detecting weak sources. Do not train CDs in cemeteries unless it can be veri­fied that the bodies were not embalmed with formaldehyde.

Tom Osterkamp