|Year : 2016 | Volume
| Issue : 4 | Page : 233-244
Evaluation and Examination of a Possible Shoe-polish Trace in a Hold-up Case
Line Gueissaz1, Tacha Hicks2, Cyril Muehlethaler3, Geneviève Massonnet1
1 Faculty of Law, Criminal Justice and Public Administration, School of Criminal Justice, University of Lausanne, Lausanne 1015, Switzerland
2 Faculty of Law, Criminal Justice and Public Administration, School of Criminal Justice, University of Lausanne, Lausanne 1015, ; Fondation Pour la Formation Continue Universitaire Lausannoise, Lausanne, 1015, Switzerland
3 Department of Chemistry, City College of the City University of New York, New York 10031, USA
|Date of Web Publication||9-Jan-2017|
School of Criminal Justice, Batochime – Quartier UNIL Sorge, University of Lausanne, Lausanne 1015
Source of Support: None, Conflict of Interest: None
In this article, we show how the Bayesian framework can be applied to a hold-up case involving a possible shoe-polish trace according to one of the parties. This article highlights the importance of interpreting data from the beginning of the examination through the preassessment steps. Once a set of alternative propositions in agreement with the information provided by the parties is chosen, one can establish what is needed in the case. Here, limited data were available to assign factors such as transfer and rarity of the traces. Consequently, we showed how specific case-tailored experiments provide meaningful data for evaluation. In this case, the police had observed a trace on the jacket of a person who reported to have been pushed with the offender's gun during the hold-up attempt. When the jacket was submitted to our laboratory, the exact nature of the trace was unknown. Particles from this trace were collected and analyzed by stereomicroscopy, microscopy, and infrared spectroscopy. The obtained results supported that this trace was waxy material. The literature dealing with the analysis of waxy materials generally uses solvent extraction-based methods. Here, as our analytical sequence allowed a good discrimination of different waxy products of known origin, we considered that this methodology was adequate. Moreover, it did not involve any extraction step that could lead to undesired compounds from the substrate (e.g., dyes and additives). This article therefore suggests an alternative analytical sequence for the analysis of such material in casework.
Keywords: Bayesian, evaluation, firearms, trace evidence, transfer
|How to cite this article:|
Gueissaz L, Hicks T, Muehlethaler C, Massonnet G. Evaluation and Examination of a Possible Shoe-polish Trace in a Hold-up Case. J Forensic Sci Med 2016;2:233-44
|How to cite this URL:|
Gueissaz L, Hicks T, Muehlethaler C, Massonnet G. Evaluation and Examination of a Possible Shoe-polish Trace in a Hold-up Case. J Forensic Sci Med [serial online] 2016 [cited 2019 May 22];2:233-44. Available from: http://www.jfsmonline.com/text.asp?2016/2/4/233/197930
| Introduction|| |
In this publication, we used a case report to show how one can apply the Bayesian framework in a given case. This case is about a hold-up attempt involving a possible shoe-polish trace and this article also presents a novel analytical sequence to deal with this kind of trace. We showed how case experiments can be conducted to help assign probabilities. These experiments are considered as original research specifically dedicated to help solving the case.
Our article is presented similarly to the statement that we would provide in casework, with an additional part regarding preassessment (that would be kept in the case file).
To understand the issue with which forensic science could help and also to choose the appropriate methods of analysis, we were given the following relevant case information:
An attempt of hold-up took place in a small shop. An armed person barged into the shop and forced the clerk to give him money. During this time, a customer (Ms Customer) entered into the shop and tried to escape when she realized what was happening. According to her testimony, the armed person threatened Ms Customer and ordered her to go back into the shop. Ms Customer complied immediately and headed toward the armed person, who pushed her in the back with his gun.
A few hours later, the police interviewed Ms Customer. When she explained her version of the events, the policeman noticed that a dark trace was present where she declared the armed person had pushed her. Ms Customer declared that this trace was not present previous to the events. Therefore, the police seized her jacket.
One month later, the police arrested a potential suspect, Mr X. The police seized a fake gun on him. Mr X did not deny that this was his gun.
Remark - Our approach to the examination and interpretation of the findings in this case was crucially dependent on the information made available to us. If any of this information had been incorrect or if further information was made available, we would have reconsidered our interpretation.
The authors were asked to compare the trace observed on Ms Customer's jacket with the fake gun seized on Mr X and to draw any useful information.
Description of the received items
Ms Customer's jacket (Item 1) and the fake gun seized on Mr X (Item 2) were sent to the laboratory in sealed and separate paper bags. The items had been packaged 1 month apart, the gun being seized 1 month later. The bag containing the jacket was opened first on a clean table covered with new protective paper. The same precautions were taken for the bag containing the fake gun, which was opened afterward on a different table. The observation of the items led to the following results:
Item 1: Ms Customer's jacket
The jacket was light beige and included a belt. According to information provided by the label, this jacket was 45% acrylic and 55% cotton and of size “M.” No information about the brand was provided. This jacket presented no damages. An elongated dark trace in the back was visible although slightly marked. This dark trace was located at the level of the left shoulder blade.
Item 2: Fake gun seized on Mr X
The fake gun was covered with a dark material. The deposition of this dark material was not homogeneous. In some places, it was possible to observe the substrate of the gun, which was made of transparent plastic. A great amount of dark traces were observed inside the paper bag used by the police to pack the fake gun. The manner in which this dark material was deposited on the gun and the fact that it easily transferred indicated that the coating was not original (This was subsequently confirmed by Mr X, who declared to have covered his fake gun with shoe polish. This information was provided after the authors had proceeded with the analyses).
It was assumed that the delay between the alleged actions and seizure of the fake gun did not preclude the comparison between the trace and the dark material covering the gun. Stated otherwise, we assumed that the gun was in the same state as it was a month earlier.
Before conducting any analysis, the authors preassessed the case as recommended by Cook et al. This step allows ensuring that appropriate methods are used. In addition, it has the great advantage of obliging the scientist to assign the relevant factors without knowing the results – this is especially important to avoid “post hoc rationalization” or so-called bias.
The first step of the preassessment is to understand the issue and help formulate useful mutually exclusive propositions that are exhaustive in the context of the case. Here, the purpose was to analyze and assess the value of the trace recovered from Ms Customer's jacket. When phenomena such as primary, secondary (or tertiary) transfer, contamination, or fortuitous presence of such material in the environment affect the case evaluation of the findings, as it was the case here, activity level propositions should be considered as recommended in the guideline for evaluative reporting from the European Network of Forensic Science Institutes. Therefore, the results were preassessed given the following propositions:
- Mr X pushed Ms Customer with his fake gun
- An unknown person pushed Ms Customer with another gun.
One can note that the alleged activities are similar; however, transfer probabilities may be different given each propositions, thus the importance of an evaluation of the results given activity level propositions.
The second step in preassessment consists in listing the possible (and most expected) outcomes, before carrying out any examination or analysis. In general, two to four possible outcomes is a good number as it allows to cover most situations without spending too much time. In addition, depending on the outcomes, one will use a different likelihood ratio (LR) formula. It is thus important to only regroup categories where the same model (i.e., LR formula) would be used. We therefore chose the following possible outcomes:
- A dark trace visible to the naked eye that is not differentiated from the material of the seized gun
- A dark trace visible to the naked eye that is differentiated from the material of the seized gun.
The outcome “no dark trace” was not considered here because the authors knew that the police had observed a dark trace on Ms Customer's jacket. For both outcomes, we have considered the important factors to be assessed given each proposition. In the following sections, we will first explore the situation (a) where the trace and the material from the seized gun are not differentiated and then the situation and (b) where the trace has been differentiated.
The trace and the material from the seized gun are not differentiated
Here, we consider the first outcome, where the trace is not differentiated from the material from the seized gun. The questions the scientist needs to answer are as follows:
What is the probability of finding one trace that is not differentiated from the material of the seized gun given the case information and given that Mr X pushed Ms Customer with his fake gun?
This sentence can be formalized as Pr (E ǀ Hp, I), where “E” denotes the forensic results, “Hp” the prosecution proposition, and “I” the relevant information in the case. Let us explore how to assign this probability. If Mr X pushed Ms Customer with his fake gun and one recovers a trace that is not differentiated from the material of the gun, and then there are two possibilities:
Either (i) the fake gun transferred material, it is not differentiated from the trace and there was no background or (ii) there was no transfer, there was one trace as background that is adventitiously not differentiated from the material of the seized gun. We denote the probability of the events as follows:
- The probability of recovering this material given that there was transfer is denoted t
- The probability of recovering no material given that there was no transfer is denoted t0
- The probability of the trace not being differentiated is assigned as one given that there was transfer (we do not account for analytical errors)
- f is the probability of observing the analytical characteristics given that the trace is present as background
- The probability of no trace being present as background (i.e., for unknown reasons) is denoted b0
- The probability of a trace being present as background is denoted b1.
The probability of finding one trace that is not differentiated from the material of the gun given the case information and given that Mr X pushed Ms Customer with his fake gun can thus be written as: Pr (E ǀ Hp, I) = t b0 + t0b1f.
Let us now consider the results given the defense proposition.
What is the probability of finding one trace that is not differentiated from the material of the seized gun, given the case information, but this time considering that an unknown person pushed Ms Customer with another gun?
If another gun was used, we need to introduce an additional factor f'. This represents the probability of observing the analytical characteristics given that the trace comes from an unknown gun (fake or not) that would transfer material. This probability is conditioned on the material's capacity of transferring as we can imagine that transfer and analytical characteristics are not independent.
Again, there are two possibilities that could explain the results:
Either (i) the unknown gun transferred material (this probability is denoted t'), it is not differentiated from the trace (f'), and there was no background (this probability is denoted b0) or (ii) there was no transfer (this probability is t'0), there was background (this probability is denoted b1) and adventitiously this material present as background is not differentiated from the material of the seized gun (given it is background, this probability is denoted f). This can be written as: Pr(EǀHd, I) = t' f b0 + t'0b1f.
The trace and the material from the seized gun are differentiated
We now consider the possibility that the trace is differentiated from the material of the seized gun.
As previously, we first focus on:
What is the probability of finding one trace differentiated from the material of the seized gun given that Mr X pushed Ms Customer with that gun and given the case information?
Here, there is only one possibility:
There was no transfer (this probability is denoted t0), there was background (this probability is denoted b1), and it is differentiated (given it is background, this probability is denoted 1 − f). This can be formalized as: Pr(EǀHp, I) = t0b1 (1 − f).
We then considered the same outcome but given defense proposition.
What is the probability of finding one trace differentiated from the material of the seized gun given that an unknown person pushed Ms Customer with another gun and given the case information?
To explain these observations there are two possibilities:
Either (i) the unknown gun transferred material (this probability is denoted t'), it is differentiated (1 − f') from the material of the seized gun, and there was no background (this probability is b0) or (ii) there was no transfer (this probability is t'0), there was background (this probability is b1) and this material present as background is differentiated from the material of the seized gun (given it is background, this probability is denoted 1 − f). This can be written as: Pr(EǀHd, I) = t' (1 − f') b0 + t'0b1 (1 − f).
[Table 1] summarizes the important factors given each proposition and the LR formulae associated with each outcome.
|Table 1: The defined outcomes and their related likelihood ratio formulae|
Click here to view
We see from the formulae as shown in [Table 1] that we need to assign:
- t0 and t
- t'0 and t'
- b0 and b1
- f and f'
Let us first begin with the background probabilities (i.e., material that is present for unknown reasons). Ms Customer declared that the dark trace found on her jacket was not present when she put it on before going to the shop. According to her testimony, Ms Customer did not perform any activity, which could have led to such a trace, between the alleged facts and the seizure of her jacket by the police (i.e., about 1–2 h later). Moreover, the jacket presented no damages. According to this information, it was judged that if any trace visible by the naked eye was found on Ms Customer jacket, and then the probability of its presence for reasons unconnected with the alleged activities would be very low. In other words, it was considered that one would not expect to find a trace as background – this means that the probability that a trace would be present for activities not linked to the case was close to zero.
The following values were assigned based on our experience and the case information [Table 2]:
The probabilities sum to one and are therefore coherent. As the trace was recovered from the victim's jacket, background probabilities are similar given each proposition.
Regarding transfer, several factors are important, for example, the type of activity and applied force, nature of transferred material, and type of substrate. However, in our case, for both propositions, the activity is similar (i.e., to push someone with a gun) and we judged that the force needed for this activity could be considered equivalent for any person (i.e., adult in good health). The main factor considered here was therefore the capacity of guns to transfer material.
We also took into account persistence and recovery. The persistence of the trace was judged as high as Ms Customer's jacket was seized a few hours after the alleged facts. Moreover, Ms Customer did not perform any activity that could have led to important losses. Likewise, the recovery of the trace was judged as good.
To assign transfer probabilities given the prosecution proposition, we took into account the fact that when we opened the bag where the fake gun was stored, we observed multiple dark traces in the interior of the bag. This indicates that the material covering the gun transfers easily. We therefore assigned a probability of 0.9 that this gun would transfer material.
If another unknown gun was used to push Ms Customer, we first need to think of what kind of gun we have to consider (i.e., fake or real gun). The case information provided did not allow us to know what kind of gun was used by the offender of the hold-up. Thereby, we considered all guns (i.e., fake and real guns) under the defense proposition.
No literature was found on the capacity of transfer of material from guns (fake or not) to a substrate for any kind of activities. We therefore conducted tests to help us assign probabilities of transfer in such circumstances. The procedure of these tests was as follows: A dummy was pushed in the back with a gun. The activity of pushing was done such as to simulate the person moving in one direction. This was done with considerable force. Eighteen tests were conducted: with three fake guns (tests 1–3) and with 15 real guns (tests 4–18). To conduct the tests, the guns were taken from the armory and used in their actual state (no preparation). Details about these guns are shown in [Appendix 1 [Additional file 1]]. The dummy was fixed standing up and wore a white T-shirt to ensure best contrast between potential traces and the substrate. Three new T-shirts of the brand “Chîcorée,” 100% cotton (information provided by the label), were used. These were observed before conducting the simulations to ensure that no dark trace was present in the background. Three tests were performed on each side of the T-shirts. After each simulation, the beginning and the end points of contact between the gun muzzle and the T-shirt were annotated with stickers. After the first three simulations per T-shirt, the front side was turned around to the back of the dummy to perform the last three tests. Once the six tests were performed, the T-shirt was cut on the side seams and each side was stored flat on separate paper sheets. After the simulations, any trace visible to the naked eye was recorded.
To assign transfer probabilities, we used the methodology presented in Biedermann et al. On the 18 tests, four presented a dark trace visible by the naked eye and for the others, no trace was visible, leading to probabilities of 0.25 and 0.75 (using prior counts of one, [Appendix 2 [Additional file 2]]). It is interesting to highlight that among the four guns that left a dark trace, three (numbers 5, 6, and 10) had been recently used for shooting. The last one (number 2) is a fake gun. An observation of its trace by stereomicroscopy clearly showed that the trace was gray with shiny particles, and that the muzzle of this gun was coated with metallic gray paint, while the other three traces were formed by small dark particles. The transfer probabilities are displayed in [Table 3].
|Table 3: Assigned probabilities for the transfer given the alternative propositions for the two possible outcomes|
Click here to view
We then assigned the probability of observing the analytical characteristics, given that the trace is present as background (f), and given that the trace comes from an unknown gun (f'). For preassessment, as we had not yet analyzed the trace, we took a value based on the discriminating power of the techniques – we assigned a value of 1% to f and of 5% for f'. Indeed, if the trace comes from an unknown gun (f') and without knowledge about the nature of the trace at hand, we thought that the probability that this unknown gun would present dark material with similar characteristics was low (0.05), but higher than for background (0.01). Indeed, if the trace is present because a gun was used, other guns are more likely to present similar material. By reporting the values of the different factors as assessed in [Table 2] and [Table 3], we obtained the results as shown in [Table 4].
|Table 4: Values used to assess the probability of the two possible outcomes given both propositions|
Click here to view
Following the methodology presented in Cook et al., if the prosecution proposition is true, our probability to have observations that would provide moderate (according to verbal scale published by Marquis et al.) support for the prosecution compared with the defense was assigned as 0.88. Our probability that the observations would provide strong support for the defense proposition compared with the prosecution proposition (i.e., misleading results) was assigned as 0.002.
If the defense proposition is true, our probability to have results that provide strong support for the defense (our LR is about 125 in favor of defense proposition compared with the prosecution proposition) (as it is difficult to grasp numbers such as 0.008, one can also inverse the propositions and say that our LR is equal to the probability of the results given the defense proposition and case information divided by the probability of the results given the prosecution proposition and case information) is 0.25 and our probability to have findings that provide moderate support for the prosecution proposition compared with the defense proposition is 0.01. In the great majority of cases, we would indeed expect to find no particles if the defense proposition was true. This situation is not included in [Table 4] because we knew that it was not a possible result in this case (i.e., as a trace had been observed by the police).
| Materials and Methods|| |
When the jacket was submitted for analysis to our laboratory, the exact nature of the trace was unknown. As we were able to collect large particles from this trace using a tip, these were analyzed according to the analytical sequence we use for paint traces  going from the general to the particular (from optical examinations to chemical analysis).
Optical examinations and sampling
The following procedures were followed:
- Observations with the naked eye, followed by stereomicroscopic examination (Leica M205C)
- Sampling of several particles of the trace on the jacket and of the dark material covering the fake gun
- Microscopic examinations of the sampled particles that were deposited and flattened on separate glass slides (Leica DM6000 – bright field, dark field and crossed polar).
- Infrared spectroscopy measurements were conducted on a Thermo Nicolet 5700 FT-IR spectrometer coupled to a Nicolet microscope IR Continuum with a 32x infinity Reflachromat objective and a mercury cadmium telluride detector (MCT/A). The OMNIC software, version 9.2.41 (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA), was used
- Measurements were carried out in transmittance on specimens flattened on a potassium bromide (KBr) pellet. At least two replicates were collected for each specimen in the 4000–650 cm -1 domain with a resolution of 4 cm -1 and 32 co-added scans.
Evaluation of the within variability and comparison
Seven particles from the trace recovered from Ms Customer's jacket were sampled and analyzed. The results were compared to assess the within variability of the trace. Similarly, five specimens of the dark material covering the gun were analyzed and compared. Finally, the results of the particles sampled on the jacket and the specimens from the gun were compared. These comparisons were carried out qualitatively after each examination.
| Results and Discussion|| |
Optical examinations and sampling
Item 1: Ms Customer's jacket
The trace located on the left shoulder blade of Ms Customer's jacket measured about 15 cm in length and presented a maximum width of about 1 cm. The observation of this trace under the stereomicroscope showed that it was formed by a large amount of small dark particles (several hundreds). The majority of these particles was present at the surface of the threads on the external side of the fabric. The length of the biggest particles was <0.2 mm [Figure 1].
|Figure 1: Illustrations of dark particles of the dark trace on Ms Customer's jacket (stereomicroscope Leica M205C)|
Click here to view
Seven particles were collected from different locations of the trace, placed separately on glass slides, and flattened. The particles were easily flattened and the material looked greasy. These preparations were observed under the microscope. In bright field, these particles appeared as a dark brown matrix (which darkness depended on thickness) with small pink or blue particles that formed smears within the matrix [Figure 2]a. Of the seven particles, one presented additionally an orange smear [Figure 2]c. Under crossed polar illumination, the colored particles were slightly birefringent within the dark matrix. Dark field did not add additional information. Except the orange smear, the seven particles were not differentiated by microscopy. The dark material constituting the trace was not homogeneous: One could observe particles of different color (pink, blue, or orange) that were unevenly dispersed, but this inhomogeneity was consistently observed on all sampled particles, except the orange smear.
|Figure 2: Particles of the trace (a and c) and of the dark material of the fake gun (b and d) (microscope Leica DM6000 (a and b) and Leica DMR (c and d) – bright field with objective ×40, ~400)|
Click here to view
Item 2: Fake gun seized on Mr X
The observation of the dark material covering the fake gun under the stereomicroscope showed that it was a shiny substance. Its deposition was inhomogeneous. The gun was sampled five times by scraping the dark material with a scalpel. During this sampling procedure, the authors noticed that this dark material looked greasy and sticky. Five specimens were sampled at different locations to cover the surface of the gun, including barrel and muzzle. These five specimens were placed on separate glass slides and flattened. Similarly, to the particles recovered from the trace, the five specimens were easily flattened. The observation of these five preparations in bright field showed that they were constituted of a dark brown matrix that varied in darkness depending on the thickness. In the same way as the trace, lots of small pink or blue particles forming smears within the matrix were observed [Figure 2]b. Within one specimen, an orange smear was observed [Figure 2]d. These colored particles showed a slight birefringence in crossed polar illumination and no particularity was observed in dark field. The dark material covering the fake gun was not homogeneous, but it was constant on the entire gun, with the exception again of the orange smear.
Based on the microscopic observations, the particles from the trace recovered from Ms Customer's jacket and the particles sampled on the dark material covering the fake gun were not differentiated.
Item 1: Ms Customer's jacket
Given that the seven particles sampled from the trace were not differentiated by microscopy, only three of them were analyzed using infrared spectroscopy. Two measurements per particle were performed due to their small size. The zone measured was mostly composed of a brown matrix with small amounts of colored particles. For each particle, the two replicates showed a good repeatability [Figure 3] for the major bands supporting a weak within variability. Some differences of weak intensity were observed. This was not unexpected as there was thickness variation, as well as inhomogeneity of particles in the zone measured (e.g., different amounts of pink, blue, or orange particles in the area measured). The results supported a weak within variability for the trace particles. Based on the infrared spectra, the three particles were not differentiated.
|Figure 3: Illustration of the infrared spectra (in absorbance) of the three analyzed particles of the trace from 4000 to 650 cm-1 with a zoom on the 1800–650 cm-1 area (particle one below and in blue, particle two middle and in green, particle three above and in red)|
Click here to view
Item 2: Fake gun seized on Mr X
Three of the five specimens sampled from the dark material covering the fake gun were analyzed using infrared spectroscopy. Two measurements per specimen were performed and each of them showed a good repeatability for the major bands. Again, differences of weak intensity were observed; however, the results supported a weak within variability for these particles. In the same way as the trace, the three particles were judged to be non-differentiable from one another on the basis of their infrared spectra.
The comparisons between the infrared spectra of the particles from the trace (recovered from Ms Customer's jacket) and of the particles sampled of the dark material covering the fake gun were not differentiated [Figure 4] either.
|Figure 4: Illustration of the comparison of the infrared spectra (in absorbance) of the three analyzed particles of the trace (above and in red) and the three analyzed specimens from the dark material covering the fake gun (below and in blue) - from 4000 to 650 cm-1 with a zoom on the 1800–650 cm-1 area|
Click here to view
The infrared spectra of the trace and of the dark material covering the fake gun were searched in several commercial databases. The best candidates were found within the spectra of waxes (bees or carnauba) from the database HR Comprehensive Forensic Fourier Transform Infrared (FT-IR) Collection (Thermo Fisher Scientific Inc., for Nicolet FT-IR materials). The police were therefore informed that the trace and the material covering the gun had characteristics similar to those of a waxy material. Both the trace and material covering the gun could be, for example make-up, automotive polish and shoe polish. In a subsequent interview, Mr X declared that he had used shoe polish to cover the fake gun.
Additional work to help with the assessment of the probability of the analytical characteristics
Given that the dark trace recovered from Ms Customer's jacket presented correspondences with waxy material, we decided to analyze additional products of this category to evaluate whether they could be differentiated or not by the analytical sequence used. Published literature ,,,, dealing with this kind of material generally uses solvent extraction techniques followed by microscopy, microspectrophotometry, or chromatography (thin-layer chromatography or gas chromatography). As there is to our knowledge no literature focused on the analysis of waxy materials with the analytical sequence used in the present study, we tested its discriminating power. If this analytical sequence proved to be able to discriminate most wax-based products, it would be more suitable as it is a solvent-free extraction technique. It would avoid extraction of undesired compounds from the substrate (e.g., dyes, additives). The goal of this additional work was also to gather more information to help us assign the factor f'.
Thereby, ten black shoe polishes, one black automotive polish, one black stick for automotives, two black makeups for face painting, and two greases for firearms were analyzed by microscopy and FT-IR [Table 5].
For the 16 products as shown in [Table 5], a given amount was deposited on glass slides and dried in the open air for several days. Then, a small portion was flattened on another glass slide and observed by microscopy in bright field, dark field, and crossed polar. Unless mentioned otherwise, the results described below regard bright field illumination.
Most of the products (n = 7) presented an inhomogeneous granular black and transparent matrix without other colored particles: shoes polishes (N° 1, 5, 9), automotive polish and stick (N° 11 and 12), and two makeups (N° 15 and 16) contained additional shiny particles visible in crossed polar illumination. Shoe polishes N° 8 and 10 also presented a black and transparent matrix but smoother, with glitter/shiny particles in crossed polar illumination (N° 8) and red particles in dark field and crossed polar illuminations (N° 10). Three shoe polishes presented a grey-blue matrix with (N° 4) or without (N° 3, 6) small blue or red particles like small crystals. Two shoe polishes presented a dark brown matrix (N° 2 and 7) without colored particles. The matrix of the two greases for firearms (N° 13 and 14) was generally transparent with big black particles.
Although shoe polishes N° 2 and 7 presented a brown matrix not differentiated from the trace recovered from Ms Customer's jacket, the absence of pink/blue and orange particles/smears allowed to differentiate them. The others products were easily differentiated from the recovered trace.
The 16 products were then analyzed by infrared spectroscopy following the same procedure used for the trace and the black material covering the fake gun. A small portion was flattened on KBr pellet and several measures per specimen were performed. The spectra of each product were visually compared to evaluate the within-source variability. They were then compared to the trace recovered from Ms Customer's jacket to evaluate whether they were different or not. All products were finally compared two by two to estimate the discrimination power of the techniques.
The spectra of the 16 products presented a good repeatability and within-source variability was low. The 16 products showed infrared spectra that were different from the trace. Among all possible pairwise comparisons, only the two greases for firearms presented nondifferentiable infrared spectra.
Assigning f and f'
If the trace came from an unknown gun, the question one should ask is: among all possible guns (fake or not), what is the probability to encounter one covered with a dark waxy material with similar characteristics to the trace of our case? To assess f', one should ideally analyze dark traces left by guns on clothes of someone having been pushed and compare these traces to the trace at hand. However, as highlighted in our small study on the transfer, most of guns did not leave a dark trace (t'0 = 0.75, see Appendix 2). To get enough data, we would have had to conduct dozens of simulations and this was not possible in the time at disposal for our casework. That is why we proceeded to the analysis of 16 dark waxy products bought in commerce to help us assess f'.
Based on our experiments on transfer with guns, we observed, using stereomicroscopy, that among the guns that left a dark trace (n = 4), one had a waxy appearance. The probability of the characteristics (i.e., waxy appearance) among guns that transferred traces (f'1) can be assessed as 0.33 [Appendix 3 [Additional file 3]]. Of the 16 dark waxy materials analyzed, none of them presented similar characteristics to our trace. The probability of these characteristics among dark waxy materials (f'2) can thereby be assessed as 0.06 (=1/18 with prior counts of one [Appendix 3]). This probability is conditioned on the f'1 probability. Thus, these probabilities (f'1 and f'2) can be combined and lead to f' = 0.02 (that is: 0.06 × 0.33).
Evaluation of the findings
Thanks to our detailed preassessment, it was straightforward to evaluate our results by considering outcome one of [Table 4], while adapting the factor f'. The probability of the results given that Mr X pushed Ms Customer is in the order of 0.9. The probability of the results given that an unknown person pushed Ms Customer with another gun is in the order of 0.005 (by adapting f' from 5% in preassessment to 2%). Our results are therefore in the order of 200 times more probable given the first proposition than the second. The reader can see that our LR was multiplied by two between the preassessment and evaluation of the findings and this can be surprising. This is due to the fact that during the preassessment step, the authors had less information for f' because they had not analyzed the trace. Given that the trace recovered from Ms Customer's jacket was particular, especially due to the pink, blue, and orange smears, the probability to find another gun (fake or not) covered with this kind of material decreased drastically. Contrariwise, if the trace could not be differentiated from grease for weapons, f' would have drastically increased leading to a smaller LR. Indeed, in this case, it would be more probable to find others guns with grease on their surface that could not be differentiated from our trace.
Communication of results is crucial. Thus, in the following section, we report our results as we would in our statement.
Evaluation of the results
We have evaluated the obtained results (a dark trace visible by the naked eye and recovered from the jacket that is not differentiated from the material covering the seized fake gun) in the context of the alleged activities as described in the case information. As our approach to the examination and interpretation of the observations in this case is crucially dependent on the information available (e.g., where the victim was pushed, absence of the trace previous to the event), we would like to outline that if any of the information given is incorrect or if further information is made available, it will be necessary for us to reconsider the evaluation of our results.
The testimony of Ms Customer (i.e., the trace was not present previous to the event and she did not have activities before/after the seizure of her jacket, which could lead to such a trace), and where the trace was recovered (left shoulder blade) are elements that indicate that the probability of the trace being present as background (for a reason that is not connected to the incident) is small. Therefore, the presence of the trace recovered from the jacket can be explained by two main activities: either Mr X pushed Ms Customer with his fake gun, or some unknown person pushed her with another gun and Mr X has nothing to do with the incident. We can help the court with this issue by assessing how much more probable the results are, given each of the two propositions.
To be balanced, we assess the value of our results given the two propositions above. Let us first evaluate the results given that Mr X pushed Ms Customer with his gun. If Ms Customer was pushed in the back (on her left shoulder blade) with Mr X's fake gun, we would expect to find (as it was the case here) a trace visible to the naked eye not differentiated from the material covering the fake seized gun where she was pushed. Indeed, the material covering the fake gun transfers very well and in such amounts that traces are visible by the naked eye when touching surfaces. Moreover, the delay, between the incident and when the jacket was seized by the police, is short (about 2 h). We therefore assigned the probability of the results as 0.9. This means that on average, if we simulated this case, we would expect to observe these findings in about 9 out of 10 experiments.
If now we consider that the trace was recovered because some other gun had been used to push Ms Customer, what would be our probability of our results? In that case, the alternative gun would have to transfer material in a quantity visible to the naked eye. In addition, this material would have to be non-differentiable from the trace using all methods used.
If somebody pushed another person with a gun (fake or not), in most cases (15 of 20), we would expect no material to be transferred to the person. Indeed, the transfer of dark material in such a quantity that it is visible to the naked eye would be a rare event. One could expect to find grease on the gun as this material is used for the maintenance of firearms. However, the material recovered from the jacket of Ms Customer was differentiated from the two greases tested.
If another firearm was used, it would have to be covered with a material that transfers easily that would also present characteristics non-differentiable from the trace. The techniques that we used have a high power of discrimination for waxy materials. Indeed, the 16 dark waxy products analyzed were all differentiated by the analytical sequence applied, except the two greases for firearms which were not differentiated between them. The probability of observing characteristics non-differentiable from the trace if another gun was used was assigned as 2%.
Consequently, given the defense proposition, we have considered that it would be very rare to observe a dark trace visible to the naked eye that would not be differentiated from the material covering the gun. This probability has been assigned as 0.005.
In conclusion, we are therefore of the opinion that the results are in the order of 200 times more probable given that Mr X pushed Ms Customer than given that an unknown person pushed Ms Customer with another gun. If using a verbal qualifier (that is by essence subjective), we could say that the results offer strong support for the prosecution proposition compared with the defense. To convey the impact of the forensic results on the probability of each propositions, it is possible to use a table as advised by Marquis et al.
| Conclusion|| |
In this article, we have presented how one can apply the Bayesian framework to an uncommon case when data are scarce. We have shown that when performed thoroughly, starting from the very beginning of the case approach, pre-assessment helps ask oneself the good questions and acquire the most relevant data. We have shown through this case report that how one can acquire limited data specific to the case at hand by performing case-based experiments, in order to assess the value of the findings. In the last part, we have demonstrated that once the pre-assessment is performed and relevant data are acquired, the evaluation of the findings is facilitated and risks of unforeseen situations or bias are avoided.
From a methodological point of view, a new sequence of analysis was developed for waxy material. It enables analyzing only part of the trace (leaving material in situ for possible counter-expertise and avoiding the extraction of undesired compounds) contrary to a method that would involve an extraction of the entire trace with a solvent. Our methodology is therefore considered as less destructive and more appropriate. Moreover, this analytical sequence is promising regarding its capacity to discriminate waxy materials. In the future, we intend to further explore the value of this sequence to distinguish material that can be found in casework, such as shoe polish.
The authors would like to warmly thank Damien Rhumorbarbe and Denis Werner from the School of Criminal Justice for their help with the experiments conducted with guns.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Cook R, Evett IW, Jackson G, Jones PJ, Lambert JA. A model for case assessment and interpretation. Sci Justice 1998;38:151-6.
Biedermann A, Taroni F, Bozza S. Implementing statistical learning methods through Bayesian networks. Part 1: A guide to Bayesian parameter estimation using forensic science data. Forensic Sci Int 2009;193:63-71.
Cook R, Evett IW, Jackson G, Jones PJ, Lambert JA. Case pre-assessment and review in a two-way transfer case. Sci Justice 1999;39:103-11.
Marquis R, Biedermann A, Cadola L, Champod C, Gueissaz L, Massonnet G, et al.
Discussion on how to implement a verbal scale in a forensic laboratory: Benefits, pitfalls and suggestions to avoid misunderstandings. Sci Justice 2016;56:364-370.
Muehlethaler C, Gueissaz L, Massonnet G. Forensic paint analysis. In: Siegel JA, Saukko PJ, editors. Encyclopedia of Forensic Sciences. 2nd
ed., Vol. 2. London: Waltham Academic Press; 2013. p. 265-72.
Byrne LM, Cole MD, Milligan F, Thorpe JW. Shoe polish stains on fabric – A comparison of different shoe polish types. J Forensic Sci Soc 1994;34:53-60.
Cole MD, Thorpe JW. The analysis of black shoe polish marks on clothing. J Forensic Sci Soc 1992;32:237-44.
Griffin RM, Doolan K, Campbell M, Hamill J, Kee TG. Analysis of wax-based products by capillary gas chromatography mass spectrometry. Sci Justice 1996;36:229-43.
Ismail D, Daeid NN. Comparison of smears of wax-based products using thin-layer chromatography and microspectrophotometric detection. J Forensic Identification 2011;61:136-46.
Sahajpal V, Garg RK. Thin-layer chromatography of black shoe polish stains on fabric. J Forensic Identification 2006;56:339-43.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]