THE RECOVERY AND ANALYSIS OF DNA FROM FIRED CARTRIDGE CASINGS
By
Alicya Orlando

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Forensic Science
2012

ABSTRACT
THE RECOVERY AND ANALYSIS OF DNA FROM FIRED CARTRIDGE CASINGS
by
Alicya Orlando
Firearms, particularly handguns, are used in the commission of various crimes in
the United States, though frequently the only evidence that a gun was used is cartridge
casings. Thus, deeper investigation of the utility of DNA analysis for fired cartridge
casings has been attempted. Previous studies have explored the probability of developing
latent fingerprints and generating STR profiles from unfired and fired cartridge casings in
attempt to identify the individual who loaded the gun. In each of these instances casings
were washed to remove DNA from prior handlers. In reality, however, criminals will not
wash the cartridges before firing them, thus the primary focus of this study was utilizing
un-cleaned casings. A cumulative swabbing method was compared to a single swabbing
method to determine which generated higher DNA yields and more complete STR
profiles consistent with the loader. A consensus method for determining STR profiles
was also compared to the two swabbing methods. Thirty volunteers handled 10
cartridges each and loaded them into the magazine of a gun; cartridges were fired and
casings were collected and swabbed. DNAs were extracted, quantified, and amplified for
STR typing. There was no difference in DNA yield between the swabbing methods,
though cumulatively swabbing resulted in the greatest percentages of consistent alleles
and profiles. The consensus profiling method had a higher rate of consistency than the
single swabbing method but did not outperform the cumulative swabbing method.

ACKNOWLEDGMENTS
I would like to acknowledge the instruction, support and tireless editing of Dr.
David Foran, without whom this project would not have been possible. I would also like
to thank my thesis committee members: Drs Sanja Kutnjak Ivkovich and Ruth Smith as
well as Sergeant Ryan Larrison of the Michigan State Police Firearms Unit, for their time
and patience.
I would like to acknowledge Officer Grel Rousseau of the Michigan State Police
Firearms Unit for his help in organizing collection of casing samples. All of the
volunteers who participated in the study deserve special thanks without whose
participation I would have had no casings to analyze.
I received statistical help from Drs. Robert Templeman of the MSU College of
Agriculture and Christina DeJong of the MSU School of Criminal Justice to whom I owe
a great debt of gratitude. I would like to thank the MSU Graduate School for financial
support.
Lastly, I would like to thank my peers in the MSU Forensic Biology Laboratory
as well as my friends and family back home, with whose unwavering support I found the
strength to finish this project.

iii

TABLE OF CONTENTS
List of Tables ...................................................................................................................................v
List of Figures ................................................................................................................................ vi
Introduction......................................................................................................................................1
History of Firearms ..................................................................................................................1
Techniques used for DNA Analysis .........................................................................................3
Processing of Cartridge Cases by the Michigan State Police...................................................6
DNA Recovery from Fired Cartridge Casings .........................................................................8
Goals of this Study ...................................................................................................................9
Materials and Methods...................................................................................................................10
Cartridge Casing Collection ...................................................................................................10
Cumulative and Single Swabbing of Collected Cartridge Casings ........................................10
DNA Extraction and Isolation ................................................................................................11
DNA Quantitation ..................................................................................................................12
STR Typing of Cartridge Casing and Buccal Swab DNAs....................................................12
Determination of Profile Categories.......................................................................................13
Construction of Consensus STR Profiles using Single Swab Allele Calls ............................14
Statistical Analyses of DNA Yields, Profile Types, and Allelic Consistency .......................14
Results............................................................................................................................................16
Cartridge Casing Collection ...................................................................................................16
DNA Yields: Cumulative versus Single Swabbing................................................................16
STR Profile Designation ........................................................................................................17
Consistent and Inconsistent Allele Calls Based on Swabbing Method..................................19
Differences in Allele Calls Generated from 8 and 12 Second Injections...............................21
Discussion ......................................................................................................................................23
Conclusion .....................................................................................................................................36
Appendices.....................................................................................................................................37
Appendix A. DNA quantities recovered from cartridge casings............................................38
Appendix B. Allele calls and loci categorizations .................................................................41
References......................................................................................................................................58

iv

LIST OF TABLES
Table 1: Allelic Differences between 8 and 12 Second Injections ................................................21
Table A.1: DNA quantities recovered using the cumulative swabbing method............................39
Table A.2: DNA quantities recovered using the single swabbing method ....................................40
Table B.1: Alleles obtained at each locus for casing set #1 ..........................................................42
Table B.2: Alleles obtained at each locus for casing set #2 ..........................................................42
Table B.3: Alleles obtained at each locus for casing set #3 ..........................................................43
Table B.4: Alleles obtained at each locus for casing set #5 ..........................................................43
Table B.5: Alleles obtained at each locus for casing set #6 ..........................................................44
Table B.6: Alleles obtained at each locus for casing set #7 ..........................................................44
Table B.7: Alleles obtained at each locus for casing set #8 ..........................................................45
Table B.8: Alleles obtained at each locus for casing set #9 ..........................................................45
Table B.9: Alleles obtained at each locus for casing set #10 ........................................................46
Table B.10: Alleles obtained at each locus for casing set #11 ......................................................46
Table B.11: Alleles obtained at each locus for casing set #12 ......................................................47
Table B.12: Alleles obtained at each locus for casing set #13 ......................................................47
Table B.13: Alleles obtained at each locus for casing set #15 ......................................................48
Table B.14: Alleles obtained at each locus for casing set #16 ......................................................48
Table B.15: Alleles obtained at each locus for casing set #17 ......................................................49
Table B.16: Alleles obtained at each locus for casing set #18 ......................................................49
Table B.17: Alleles obtained at each locus for casing set #19 ......................................................50
Table B.18: Alleles obtained at each locus for casing set #20 ......................................................50
Table B.19: Alleles obtained at each locus for casing set #21 ......................................................51

v

Table B.20: Alleles obtained at each locus for casing set #22 ......................................................51
Table B.21: Alleles obtained at each locus for casing set #23 ......................................................52
Table B.22: Alleles obtained at each locus for casing set #24 ......................................................52
Table B.23: Alleles obtained at each locus for casing set #25 ......................................................53
Table B.24: Alleles obtained at each locus for casing set #26 ......................................................53
Table B.25: Alleles obtained at each locus for casing set #27 ......................................................54
Table B.26: Alleles obtained at each locus for casing set #28 ......................................................54
Table B.27: Alleles obtained at each locus for casing set #29 ......................................................55
Table B.28: Alleles obtained at each locus for casing set #30 ......................................................55
Table B.29: Alleles obtained at each locus for casing set #31 ......................................................56
Table B.30: Alleles obtained at each locus for casing set #32 ......................................................56
Table B.31: Alleles obtained at each locus for casing set #33 ......................................................57

vi

LIST OF FIGURES
Figure 1: Anatomy of a Modern Bullet Casing ...............................................................................1
Figure 2: Example of STR Profile from IdentifilerTM Plus .............................................................4
Figure 3: Example of Dropout .........................................................................................................5
Figure 4: Example of Stutter............................................................................................................6
Figure 5: Example of Potential DNA Loss using Cumulative Swabbing Method ..........................7
Figure 6: Cumulative versus Single Swabbing Techniques ..........................................................11
Figure 7: Histogram Depicting DNA Yields between Swabbing Methods...................................17
Figure 8: Comparative Distribution of Profile Type based on Method .........................................18
Figure 9: Histogram of Percent Distribution of Consistent, Inconsistent, No Allele Calls ...........20

vii

INTRODUCTION
History of Firearms
Carman (1955) defined a firearm as any device that expels a projectile through the use of
sudden explosion. The projectile and explosive are contained within the ammunition, which in
turn is contained in the firearm. The primary explosive for years has been a combination of
chemicals, which collectively generate high heat and pressure in a very short amount of time.
Gunpowder as first developed was a mixture of sulphur, charcoal, and saltpeter (potassium
nitrate), intended for use in firecrackers but soon proved to be an extremely powerful tool in
combination with projectiles (Carman, 1955; Peterson, 1961). Gunpowder generates up to sixtimes its mass in gaseous pressure. This expansion of gases inside the chamber of the firearm is
used to drive projectiles, such as cannonballs, lead pellets, and bullets, at a high rate of speed in a
specific direction (Peterson, 1961). Figure 1 shows the anatomy of a modern cartridge. The
bullet and propellant are contained within the cartridge case. The rim contains the primer which
when struck by the firing pin of a gun generates a spark to ignite the propellant, which in turn
displaces the bullet as gaseous pressure builds up behind it.

Bullet
Bullet
Case
Propellant
Rim

Figure 1. Anatomy of a modern bullet cartridge. Taken from HowStuffWorks 2001.
http://static.howstuffworks.com/gif/revolver-bullet.gif. For interpretation of the references to
color in this and all other figures, the reader is referred to the electronic version of this thesis.

1

Following the development of gunpowder, firearms have advanced immeasurably;
gunsmiths developed mechanisms such as the matchlock, wheel lock, and flintlock, which
worked in similar manners, all for the purpose of generating enough spark to ignite the
gunpowder in the flash pan, ultimately resulting in an explosion and the firing of the projectile
(Ricketts, 1962). Since 1918 firearms have been improved upon by simplifying them, making
them more ‘user friendly’ (Roads and Hogg, 1983). After World War II large, ‘heavy hitting’
weaponry, such as the Boys Anti-Tank Rifle, ceased to exist as its civilian use could not be
rationalized. Instead, lightweight arms for military parachutists and other infantry specialists
became much more practical and were developed using a combination of already accepted
devices (Roads and Hogg, 1983). It was the availability of semi-automatic weaponry to the
public that seemed to change civilian dependency on firearms, especially handguns. Increased
civilian use translated into increased numbers of firearm related deaths.
As gun technology improved so did the ease of use, and self-protection became the
primary reason for owning a handgun in the US (OJJDP, 1996). In 2009, nearly 14,000 murders
occurred in the United States. Just over 9,000 (67.1%) of those involved the use of firearms
(FBI, 2010; Guardian News, 2011), 80% of which resulted from handguns, even though they
account for only 34% of firearms in the U.S. (LCAV, 2011). Aggravated assaults and robberies,
among other crimes, also involve firearms. In 2009, 42.6% of robberies and 20.9% of
aggravated assaults were committed using a firearm (FBI, 2010). This is likely because the
United States is one of the only modern industrial nations that is preserving a gun culture where
it is legal for large portions of the population to own firearms (OJJDP, 1996). However, since
peaking in the late 1980’s gun-related crimes have decreased; from 2005 to 2009 there was a

2

10% drop in firearm use related to violent crime, yet guns still pose a major threat to the safety of
all Americans (OJJDP, 1996; FBI, 2010).

Techniques Used for DNA Analysis
That guns are used to commit crimes is a given, however what is not so clear is how to
successfully identify the people loading and/or firing weapons to commit those crimes. The
ability of law enforcement to identify an assailant may lay in their capacity to utilize DNA.
When an individual holds or loads a gun skin cells may be deposited on the grip, hammer,
trigger, magazine, or cartridges; any surface that the ‘shooter’ comes in contact with has the
potential to harbor touch DNA. Degraded and low copy number (LCN) DNA (or the presence of
less than 100pg of DNA, Gill et al. 2000) are often encountered in forensic work. The most
important tool used in these instances is the polymerase chain reaction (PCR), which amplifies a
specific segment of DNA, resulting in over a billion copies and ensuring enough DNA for
downstream procedures. Today, this entails amplification of short tandem repeats (STRs), which
are the standard method used to produce highly discriminating DNA profiles. STRs are regions
of DNA with repeating sequence units 2 – 6 base pairs long. The number of repeats in each STR
marker is variable among individuals, and it is this variability that is used to differentiate and
identify individuals. A clean and easily interpretable STR DNA profile is shown in Figure 2. If
this profile were obtained from a piece of evidence it would be compared to a known or suspect
sample to see if the evidence profile and the suspect profile are both 10, 11 at the D7 locus, 14,
15 at the D3 locus, etc. After a profile has been compared at each locus a random match
probability is determined, calculating the probability that the evidence sample belongs to

3

someone other than the suspect. Often the random match probability from a full STR profile
exceeds one in a hundred trillion.

Figure 2. An example of an easily interpretable STR profile generated from control DNA, using
an IdentifilerTM Plus PCR Amplification Kit. The x-axis represents the size of the amplicon (bp)
and the y-axis represents relative fluorescent units (signal intensity). Applied Biosystems
IdentifilerTM Plus Manual, 2009. The text is not meant to be readable, but is for visual reference
only.
However, such a profile is often not obtained in forensic casework because the DNA is too
degraded or is LCN. Degradation frequently results from exposure to the elements or other
environmental conditions.
Though researchers have had great success performing STR analysis and generating
DNA profiles from LCN DNA, there are several difficulties that may be encountered. In
general, three artifacts often occur in STR profiles from LCN DNA. The first is ‘drop-in’, or the
presence of alleles that are not native to the sample. Allelic drop-in can create the appearance of

4

a mixture, which may cause incorrect profile interpretation. The second artifact is ‘drop-out’
(Figure 3) or the absence of alleles that should be present but have failed to amplify as a result of
unequal sampling of the two alleles from a heterozygous individual, which is often encountered
when few DNA molecules are utilized during amplification (Walsh et al. 1992). Similar to dropin, drop-out can be difficult to interpret. For instance, if an individual is heterozygous but one
allele did not amplify then the analyst can easily interpret that individual as homozygous at that
locus.

D8

D21

D8

D7

D21

D7

CSF

CSF

Figure 3. Example of four loci from a known sample (defendant) and the same four loci from an
evidence sample. The D21 and CSF loci both show the same STR alleles, 28, 31 and 11, 12
respectively, between the defendant and evidence samples. The D8 and D7 loci, however, are
exhibiting allelic dropout (red arrow), where the defendant is a 12, 13 at D8 but the evidence
sample only shows a 12, similarly, the defendant is a 7, 12 at D7 but the evidence shows only a
7, indicating that the 13 and 12, respectively, were not amplified during PCR. The majority of
text is not meant to be readable, for visual reference only. Taken from Applied Biosystems 2009.
http://marketing.appliedbiosystems.com/images/all_newsletters/forensic_1109/articles.html
The last of the three common artifacts is ‘stutter,’ defined by the presence of peaks four bases (1
repeat unit) smaller (or sometimes larger) than an allele in the STR electropherogram. Stutter,
exemplified in Figure 4, results from DNA slippage during DNA replication (Hauge and Litt,
1993) and can be much more extreme in LCN DNA, with stutter peaks equal to or exceeding the
size of the alleles native to that individual. Similar to drop-in, stutter peaks can give the false
appearance of mixtures.

5

Figure 4. An example of stutter in an STR profile. The tall green peaks are alleles that belong to
the handler. The red arrow points to a stutter peak for the first set of alleles. Taken from Butler
2005. http://www.cstl.nist.gov/biotech/strbase/training.htm. Text is not meant to be readable, but
for visual reference only.
Another challenge encountered when working with forensic samples is various
substances that inhibit PCR, resulting in little or no DNA amplification. Inhibitors can obstruct
cell lysis in preparation for DNA extraction, can degrade the DNA, or may compete or interfere
with elements of the PCR reaction, such as the polymerase, thereby preventing enzymatic
activity and amplification (Wilson, 1997). Overall, there is a loss of alleles and thus a loss of
genetic information.
Processing of Cartridge Cases by the Michigan State Police
Cartridge casings are often collected from crime scenes where a firearm has been used,
and are placed in evidence bags for transport to the crime laboratory (Sgt. Ryan Larrison,
personal communication). Once received into evidence by a Michigan State Police (MSP)
forensic laboratory, casings are generally submitted to the DNA unit for analysis (Sgt. Ryan
Larrison, personal communication). There, DNA analysts practice a cumulative swabbing
technique, where presumably related casings are swabbed successively using a single pair of

6

swabs (David Bicigo, personal communication). DNA is purified from the swabs via organic
®

extraction, quantified with Plexor HY, and commercial kits ProfilerTM Plus and CofilerTM are
used for STR amplification. Casings are returned to the evidence bag and no further processing
is pursued if amplification does not occur (Heather Clark, personal communication).
There are potential advantages and disadvantages with the cumulative swabbing method
utilized by the MSP. It is logical that the DNA yield from swabbing several casings at once
would be greater than swabbing casings individually, thereby resulting in more informative STR
profiles. Swabbing multiple casings at one time has the added benefit of being more cost and
time effective than processing casings individually. Alternatively, this swabbing technique may
lead to increased PCR inhibition due to an accumulation of inhibitors on the swabs. Further, if
casings collected from a scene were not all handled by a single DNA contributor, there is
potential for cross-contamination leading to an increased chance of mixtures. Lastly, DNA may
be left behind on casings that are swabbed consecutively, illustrated in Figure 5.

Figure 5. Example of potential DNA loss through the use of cumulative swabbing. The blue
represents skin cells on both casings and swabs. By the time the last casings is reached there is
an accumulation of DNA on that casing from previously swabbed casings, and less on the swab
that will be subjected to extraction.
DNA Recovery from Fired Cartridge Casings

7

Researchers have started to examine the utility of DNA recovery and analysis from spent
cartridge casings more rigorously because current DNA analysis and fingerprinting methods are
generally unsuccessful. Spear et al. (2011) tested how often they could obtain fingerprints from
unfired and fired cartridge cases, as well as the likelihood of obtaining a DNA profile from the
same casings. Forty-eight casings made of brass, nickel-plated brass, and aluminum were
examined. Bloody, eccrine, and oily prints were placed on the casings prior to firing or
processing. Bloody fingerprints were treated with amido black, and the eccrine and oily prints
were subjected to cyanoacrylate fuming. Casings were swabbed, and DNA was extracted and
amplified using ProfilerTM Plus. Overall, there were no useable prints on any of the smaller,
fired casings regardless of the type of print or processing method. In contrast, 6 of 36 bloody
prints were developed from the larger 9mm and 45 caliber casings, three oily prints were deemed
useable from unfired casings, while no useable eccrine prints were recovered. Only three DNA
profiles were obtained from the casings tested, all of which were developed from bloody prints,
two unfired and one fired.
Horsman-Hall et al. (2009) investigated whether amplifiable DNA could be obtained
from unfired and fired cartridge casings. The firearms and cartridges were first cleaned and then
handled for 30 seconds. A gloved firearms examiner loaded the cartridges into the chamber of a
gun, not using a magazine. As cartridges were fired the casings landed on a paper covered floor
and were collected using a broom and the wooden end of a sterile swab. Buccal swabs were
collected from all handlers for reference samples, and each casing was swabbed using a double
swab technique. DNA was extracted using a DNA IQTM Isolation System for Large Volumes
with 0.4% Sarkosyl, and quantified using commercial kits from Promega and Stratagene. STR
amplification was attempted using kits from Applied Biosystems (MiniFilerTM and IdentifilerTM)

8

®

and Promega (PowerPlex 16 BIO). The MiniFilerTM amplification kit resulted in a
®

significantly greater number of alleles, overall, than either Powerplex 16 BIO or IdentifilerTM,
the latter of which resulted in no alleles in any of the casing samples. The majority (54%) of
fired cartridge casings generated no profile, 26% produced less than half of the possible alleles,
19% resulted in more than half the alleles, and 1% resulted in a full profile.

Goals of this Study
Cartridges are often the only evidence that a firearm was used at a crime scene, and they
are handled for only a brief period of time during loading. Since casings are objects from which
touch DNA may potentially be collected, the goals of this study were to utilize un-cleaned
cartridges and different methodologies for recovering and analyzing DNA from fired cartridge
casings, in order to maximize the chances of obtaining useful genetic information from a
perpetrator. The effectiveness of two swabbing methods and the utility of DNA analysis for
identifying an individual who loaded or shot a firearm were also assessed. DNAs were
extracted, quantified, amplified using IdentifilerTM Plus, and STR profiles developed. Results
generated using the cumulative swabbing technique were compared to those developed from the
single swabbing method, to determine which maximized DNA yield and allelic consistency,
while limiting cross contamination (mixtures) and PCR inhibition. A consensus profiling
method, combining STR profiles from five single swabs, was also used to evaluate the two
swabbing methods based on the number of alleles consistent with the handlers’ profiles.

9

MATERIALS AND METHODS
Cartridge Casing Collection
®

Remington , 40-caliber, brass cartridges were purchased from a local sporting goods
store. Prior to cartridge distribution, ten cartridges were selected at random, swabbed, and DNA
was extracted to determine whether background DNA was present. Cartridges were divided into
sets of ten, each of which were placed in paper bags and labeled 1 – 33. Volunteers from the
Michigan State Police Bridgeport and Northville crime laboratories selected a number at random
and were provided the bag of cartridges corresponding to that number. The volunteers loaded all
ten cartridges into the magazine of a gun without excessive handling. The cartridges were fired
until each magazine was empty and casings were caught either in heat-seal grade plastic bags
contained in a wide rimmed container as they were ejected, or in the netting around the shooting
hole of a firing tank. Casings were then placed back into their respective paper bags. A new
heat-seal bag was used for each set of casings to avoid cross contamination. Volunteers selected
a letter and submitted a buccal swab that was labeled with the letter and placed in a culture test
tube; in total 33 sets of casings and corresponding buccal swabs were collected. The MSU
Institutional Review Board cleared all experiments for human participation (IRB#10-766).

Cumulative and Single Swabbing of Collected Cartridge Casings
Casings were swabbed using a double swab technique (Sweet et al. 1997) wetting the
first swab with 100µL of 5% SDS. Each set of casings was randomly divided in half. As
illustrated in Figure 6, five of the casings were cumulatively swabbed using one pair of swabs
consecutively, while the remaining five casings were swabbed individually, each with its own
pair of swabs.

10

Figure 6. Cumulative swabbing is shown on the left, where 5 casings were swabbed with a
single pair of swabs. The five un-grouped casings on the right were swabbed singly, each with
its own pair of swabs.
DNA Extraction and Isolation
Swab heads were clipped off with wire cutters and placed in a 1.5mL microcentrifuge
tube with 500µL of digestion buffer (20mM Tris—pH 7.5; 50mM EDTA, 0.1% SDS) and 6µL of
proteinase K (20mg/mL). The tubes were vortexed and incubated overnight at 56°C. Swab
heads were removed from the digestion solution and placed in spin baskets and centrifuged for 2
minutes at 1,000 relative centrifugal force (rcf). The liquid collected was added to the digestion
solution and swab heads were discarded.
DNA extractions were performed by adding 600µL of phenol to the digestion tube,
vortexing, and centrifuging at top speed (21,000 rcf) for 5 minutes. The aqueous layer was
removed and added to 600µL of chloroform, vortexed, and centrifuged at 21,000 rcf for 5
®

minutes. The aqueous layer was removed and placed in a 30Kd Millipore Amicon column.
Columns were centrifuged at 14,000 rcf for 10 minutes and filtrate was discarded. DNAs were
washed using two cycles of 300µL of TE (10mM Tris; 1mM EDTA, pH 7.5) and centrifuged at
14,000 rcf for 10 minutes. Columns were inverted over a clean 1.5mL Amicon tube and
centrifuged at 1,000 rcf for 2 minutes.

11

Zymo-SpinTM IV-HRC columns were prepared per manufacturer’s protocol. DNAs were
added to the column in a 1.5mL microcentrifuge tube. Columns were centrifuged at 8,000 rcf for
1 minute and filtered DNAs were stored at 4°C. Buccal swabs were extracted and cleaned in the
®

same manner, including the Zymo-Spin column to keep processing consistent, and were diluted
1:100 for downstream analysis. During the extraction process two casing samples, #4 and 14,
were compromised, as they were cross-contaminated.

DNA Quantitation
A QuantifilerTM Human DNA Quantification kit was used to determine DNA
concentration. DNA (1.2µL) was mixed with 6.3µL of primer mix and 7.5µL of reaction mix.
Tubes were vortexed and placed in the Bio-Rad iQ5 Thermal Cycler using cycling conditions:
95°C for 10 minutes, followed by 95°C for 15 seconds, and 60°C for 1 minute for 40 cycles.
The real-time PCR software was used to compare cycle threshold (CT) values for each reaction to
those of two sets of 8 standards, containing known DNA concentrations as per the
manufacturer’s instructions. A standard curve was produced and used to determine DNA
quantities for the casing reactions.

STR Typing of Cartridge Casing and Buccal Swab DNAs
Five microliters of each DNA extract, combined with 4µL of mastermix and 1µL of
®

primer mix, were amplified using an AmpFℓSTR IdentifilerTM Plus amplification kit in an AB
2720 Thermal Cycler with cycling parameters: 95°C for 11 minutes, cycled 28 times at 94°C for
20 seconds, 59°C for 3 minutes, and was held at 60°C for 10 minutes. One microliter of

12

amplified product was mixed with 9.85µL of HI-DI formamide and 0.15µL of AB GeneScanTM –
500 LIZTM Size Standard. Amplified products were electrophoresed on an AB 3500 Genetic
Analyzer with the following instrument protocol:
Oven Temperature: 60°C
Run Time: 1330 seconds
Run Voltage: 19.5 kVolts
Injection Time: 8 or 12 seconds
Injection Voltage: 1.6 kVolts
Capillary Length: 50 centimeters
The 8 second injections did not generate as many informative genetic profiles as the 12 second
injection, so the 8 second data are not represented in this study other then when the two injection
®

times were directly compared. Data were analyzed with GeneMapper Software Version 4.1 at
a threshold height of 50 RFUs for all dyes.
Determination of Profile Categories
Electropherograms were evaluated for completeness relative to the STR profile of the
reference buccal swab. The same categories used by Horsman-Hall et al. (2009) were applied:
no profile was the complete absence of alleles, weak partial profiles contained seven or fewer
loci with alleles, strong partial profiles contained eight or more loci with alleles, and a full profile
had alleles at all 16 loci. Note that profile designation was not based on whether the alleles were
consistent with the handler, just if alleles were present.

13

Construction of Consensus STR Profiles using Single Swab Allele Calls
Consensus profiles were generated using the five STR profiles from the individual casing
swabs based on consensus profile guidelines generated in the Forensic Biology Laboratory at
Michigan State University. Each locus in the single swab profiles was examined and ‘true’
peaks were separated from artifacts. The base (largest) peak at each locus was determined and
peak height was recorded; an allele was considered ‘Major’ if it was less than or equal to 75% of
the base peak height; ‘Minor 1’ alleles were between 33 – 74% of the base peak; and ‘Minor 2’
alleles were less than or equal to 32% of the base peak. ‘Major’ alleles were given 1 point;
‘Minor 1’ alleles were given 0.5 point, and ‘Minor 2’ alleles were given no points. For an allele
to be included in the consensus profile its point value from all five swabs had to be 2 or greater,
which was lowered from 3 in the original guidelines for this study.
Statistical Analyses of DNA Yields, Profile Types, and Allelic Consistency
DNA quantities were tested for normal distribution using R software (www.r-project.org)
and the Shapiro-Wilk test. A Mann-Whitney U test was performed using R at a significance
level of α=0.05 to determine whether the quantity of DNA obtained differed between the
cumulative or single swabbing method. Prior to statistical analysis DNA quantities that were
three or more standard deviations from the mean were removed; 11 values were withheld. Also
removed from swabbing comparisons were DNA quantities of 0.0pg/µL, in attempt to decrease
the variation. A Kolmogorov-Smirnov test using R was used to determine if there was a
difference between the two swabbing methods in terms of profile type: (full profile, strong
partial profile, weak partial profile, no profile).
STR profiles generated from the cartridge casings were compared to reference profiles to
assess if they were consistent with the volunteer who loaded the magazine. An inconsistent

14

allele call was noted in two situations: when none of the handler’s alleles were in the locus or
when a locus had extraneous alleles (e.g., if the handler’s profile was a 6, 9.3 but the casing
produced a 6, 8, 9.3). An inconsistent profile on the other hand, was when at least one locus was
composed only of erroneous alleles.
A Mann-Whitney U test was performed in R to determine whether there was a significant
difference (α=0.05) between the number of consistent and inconsistent allele calls among
swabbing methods or the consensus profiling method. The consensus profiling method and the
cumulative swabbing method were also compared to determine which generated a greater
number of alleles consistent with the handler.
Finally, STR profiles from the 8 and 12 second injection times were examined. The
number of alleles that differed between the two, as well as the number of alleles consistent with
the reference profiles, were compared. As these data were normally distributed, a t-test was
performed (α=0.05) using R, to evaluate whether the consistency of allele calls with the handler
was dependent on injection time.

15

RESULTS
Cartridge Casing Collection
Heat-sealed plastic bags held inside a large diameter circular container were better suited
for casing collection than allowing casings to fall in the net of a shooting tank. When using the
circular container new plastic bags for each individual reduced the chance of crosscontamination and casings were always caught as they were ejected from the weapon. The net
attached to the firing tank could not be changed so the risk of contamination from outside
sources was higher and occasionally casings fell out of the netting onto the floor during firing or
while transferring casings back into the paper bags.
DNA Yields: Cumulative Versus Single Swabbing
Initial PCR experiments on DNA extracted from cumulative and single casing swabs
showed inhibition (no amplification of the QuantifilerTM internal positive control), which was
overcome by cleaning DNAs with the Zymo-SpinTM column. The 10 randomly selected casings
that were tested for background DNA yielded 0.0pg/µL. Raw DNA yields for cumulative and
single swabbings are shown in appendix A. The cumulative swabbing method did not produce
significantly higher DNA yields than those generated using the single swabbing method (p =
0.88) when all 0.0pg/µL values and outliers were removed from the data set1. The mean DNA
yields for the cumulative swabs were 38.23 ± 141.52pg/µL as compared to the mean yield of
7.67 ± 9.55pg/µL for the single swabs (Figure 7).

1

There was no significant difference in DNA yields between cumulative and single swabbing
methods prior to values of 0.0pg/µL and outliers being removed.

16

Figure 7. Histogram depicting the mean DNA yields between cumulative (n=20) and single
(n=51) swabbing methods. The quantities represented are greater than 0.0pg/µL and less than
1000pg/µL as any quantity outside this range was at least three standard deviations from the
mean.
STR Profile Designation
Most of the STR results developed using the cumulative swabbing method were weak
partial profiles (67.7%), followed by strong partial profiles (25.8%), and no profiles (6.5%),
while no full profiles were generated (Figure 8). Similarly, the majority of results developed
using the single swabbing method were weak partial profiles (74.2%), followed by strong partial
profiles (15.5%), no profiles (7.1%), and full profiles (3.2%). Results from the consensus
profiling method showed a similar pattern, where 64.5% were weak partial profiles, 16.1% were
strong partial or no profiles, and 3.2% were full profiles. Three of the full profiles from single
swabs were from the same individual, and the lone full consensus profile was produced from that
individual as well. The frequency of each type of profile did not depend on swabbing method,
based on the test statistic of 0.06 being less than the critical value of 0.24. The same test

17

demonstrated no significant difference between profile type generated using the consensus
profiling method and the single swabbing or cumulative swabbing methods.

Figure 8. Histogram showing the total number of each type of profile obtained.
When examining both swabbing and the consensus profiling methods, the majority of
full, strong partial, and weak partial profiles were inconsistent with the handler (cumulative
swabbing 26.7% of the time, consensus profiling 33.3% of the time, and single swabbing 40% of
the time). On the other hand, several of the weak partial profiles consistent with the handler
contained minimal information, having single or few alleles. The majority of these occurred at
amelogenin, where an X call was always consistent with the handler, even if a Y should also
have been present (Appendix B). In contrast, only two of the five full profiles were consistent
with the handler, two remaining full profiles contained alleles consistent with the handler at 75%
of loci, and one had consistent alleles at fewer than 50% of loci. Twenty-eight percent of strong

18

partial profiles were consistent with the handler, all of which were developed using the
cumulative swabbing method. The greatest percentage (68.7%) of consistent profiles were in the
weak partial category; 23.8% of which were generated using cumulative swabbing, 15.7% were
generated using the single swabbing method, and 29.2% were developed using consensus
profiling. There was no significant difference in the frequency of consistent profiles between
full and strong partial profiles (p = 0.35), nor between full and weak partial profiles (p = 0.85).
There was a significant difference between strong partial and weak partial profiles, (p = 0.047).
Two of the high DNA quantities removed as outliers resulted in no STR profiles, the largest
outlier generated a strong partial profile, although with a majority of inconsistent alleles, and the
remaining eight produced weak partial profiles, one of which was consistent with the handler
(Appendix A).

Consistent and Inconsistent Allele Calls Based on Swabbing Method
The majority (56 – 61%) of profiles across all three swabbing/profiling methods had no
allele calls (Figure 9). Those consistent with the handler ranged from 22 – 31%, while
inconsistent calls ranged from 13 – 17%. The cumulative swabbing method produced
significantly more consistent alleles (p = 0.02), and significantly fewer inconsistent calls
(p = 0.05) than single swabs. In contrast, there were no significant differences in consistent or
inconsistent calls between cumulative swabbing and consensus profiles (p = 0.257 and 0.48
respectively), or between single swabs and consensus profiles (p = 0.269 and 0.34 respectively).

19

Figure 9. Histogram portraying the percentage of consistent (22 – 31%), inconsistent (13 –
17%), and no allele (56 – 61%) calls for the cumulative swabbing method (n=31), the single
swabbing method (n=155), and the consensus profiling method (n=31).
Of all the profiles inconsistent with the handler 46 (21.2%) were different by only one allele 32
(14.7%) were different by two alleles, and 23 (10.6%) were different by three alleles and 165
(78.2%) were different by four or more alleles. Among the single allele inconsistencies, 7
(15.2%) were generated using the cumulative swabbing method, 27 (58.7%) were produced
using the single swabbing method, and 12 (26.1%) from the consensus profiling method.
Cumulative swabbing produced 7 (21.9%) profiles inconsistent by two alleles, single swabbing
generated 24 (75%) and consensus profiling produced 1 (3.1%). One (4.3%), 16 (69.6%), and 6
(26.1%) profiles with three allelic inconsistencies were developed using cumulative, single and
consensus profiling, respectively.

20

Differences in Allele Calls Generated from 8 and 12 Second Injections
The 12 second injections resulted in more allele calls than 8 second injections, wherein
41 of 180 single/cumulative swabs had no results for 8 second injections, yet had allele calls
after the longer injection. Allele calls changed between the two injection times in 39 of the 180
samples (Table 1), 10 of which were derived from the cumulative swabbing method, and 29 from
the single swabbing method. The 12 second injection aided in ‘picking-up’ a handler’s allele in
19 instances, however in 13 instances a consistent allele was lost. Likewise, one or more new
inconsistent alleles appeared after a 12 second injection in 24 loci, most of which were in FGA.
Overall there was no significant difference in consistent allele calls between the two injection
times (p = 0.117).

Table 1. Allelic differences observed between 8 and 12 second injection times. Shown are the
samples and loci where allelic differences existed between the two injection times and what
alleles were called after each injection time. The numbers highlighted in yellow indicate alleles
that were consistent with the handlers’ profile.
Sample
Locus 8 second inj.
12 second inj.
2C
FGA
29, 46.2
23
5F
FGA
25.2, 22
17
9A
FGA
23, 25.2
22
9C
FGA
19, 32
17, 22
11F
FGA
20, 47.2
20, 22
8C
FGA
17, 18
21, 25
17A
FGA
22, 28.2
22, 23
17E
FGA
46.2
23
17F
FGA
46.2
18, 28.2
12A
FGA
43.2
17, 18
20D
FGA
16
23, 23.2
16D
FGA
32
22, 28.2
22A
FGA
24
17, 44.2
18E
FGA
48.2
16, 16.2
23A
FGA
25, 45.2
17, 17.2
21B
FGA
27, 49.2
23.2
29D
FGA
30.2
24, 25
29E
FGA
24.2, 30
25

21

Table 1 (cont'd)
26F
FGA
11A
D5
8D
D5
8E
D5
3D
D5
6B
D5
11C
D21
9C
D21
8C
D21
17A
D3
19F
D3
26A
D3
7C
D19
8D
D19
17A
D8
1D
D8
8F
D7
7C
D7
8C
D16
8F
TPOX
12A
vWA
None
D18
THO1
D13
D2
CSF

18, 34.2
11, 13
13
7
12
12, 13
31
29, 30
27, 30
14, 16
16, 17
14, 16
13, 14.2
10.2, 11
14, 15
14
10
8, 10
9, 12
8, 11
14, 17
0
0
0
0
0

22

24
11, 15
11, 12
11, 12
11, 13
9, 12
28, 32
27
29
18
15
17
14.2, 15
12, 13
10
13, 17
8, 9, 12
8, 11
11
8, 12
16
0
0
0
0
0

DISCUSSION
Cartridge casings are a common form of evidence found at crime scenes in the United
States, but standard methods for identifying the loader or shooter of the firearm are rarely
successful. Historically, crime laboratory attempts to enhance latent prints from cartridge
casings have met with little or no success, creating a need for different methods of analysis.
Researchers have found that, under very controlled conditions, obtaining DNA profiles from
fired cartridge casings is possible (Horsman-Hall et al. 2009; Spear et al. 2011). The present
study aimed to utilize and load cartridges directly from the store shelf, as a shooter would do,
and analyze the DNA remaining on the casing after it was fired. The recovery method currently
used by the DNA unit of the MSP, cumulatively swabbing casings (Heather Clark, personal
communication), was compared to an alternative recovery method, single swabbing and
development of a consensus profile, in an attempt to determine which technique was more
efficient for retrieving DNA from fired cartridge casings as well as which maximizes the genetic
information obtained.
When cartridge casings are collected at a crime scene and subjected to DNA analysis, the
goal is to gain as much genotypic information as possible. The fact that the type of profile (full,
partial, etc.) obtained in this study did not correlate with the swabbing method used indicates that
the recovery method is not as important as other factors in gathering a substantial amount of
genetic information. In this study, the mean DNA yield from cumulative swabbing was much
greater than single swabbing (Figure 7); however the standard deviations for the two methods
were so large that the mean yield based on swabbing method was not statistically different.
Alternative variables in obtaining genetic data range from the propensity of a person to leave
cells on a cartridge to PCR inhibitors, all of which would affect the quantity of DNA recovered

23

and/or the quality of resultant STR profiles. Given that greater DNA yields would be expected
to result in a stronger, more complete STR profile, it was surprising that there was no association
between DNA yield and profile quality. In fact, the few swabs that produced improbably high
DNA quantities generated only weak partial or no profiles, with the exception of a single strong
partial profile (which was not consistent with the handler) thus those quantitative values may
have been unreliable. In contrast, both strong and weak partial profiles were generated from
some DNAs that quantified at zero, and it was not uncommon to have no STR profile generated
from DNAs that quantified at greater than zero. Full profiles, however, were only derived from
DNAs that quantified above zero, yet all were well below the threshold of what is considered
LCN DNA. This indicates that the amount of DNA from casings estimated using QuantifilerTM
did not translate well to the number or quality of alleles generated. The amplicon size of the
QuantifilerTM target locus is smaller (62 bases) than those amplified using IdentifilerTM Plus
(100+ bp); therefore, QuantifilerTM should overestimate DNA quantity due to shorter DNA
templates being more abundant in degraded DNA. This was apparently not the case in this
study. A possible explanation is that DNA does not degrade uniformly and that the QuantifilerTM
locus is more susceptible to degradation than the STR loci; a factor that has been noted
previously for certain loci in degraded DNA (Foran, 2006). In this regard, it may be worthwhile
quantifying DNAs using larger sized amplicons that better mimic STR loci, and in particular,
target one or more loci that actually exist in IdentifilerTM, as these could better estimate the
concentration of the loci to be assayed.
The number of casing profiles consistent with a handler correlated with the profile type.
In this study, weak partial profiles, which were obtained most frequently, resulted in the largest
number and percentage of handler profiles. In general this was because a profile that contained a

24

small number or even a single handler allele was categorized as completely consistent, even
though the genetic information was minimal. Likewise, a profile that had a large number of
handler alleles was classified as inconsistent owing to a few or even a single allele that did not
come from the handler.
An important finding from this research was that the majority of profiles inconsistent
with the handlers, no matter the recovery method, were only dissimilar by one, two, or three
alleles. This is noteworthy because these profiles could still be probative given that the majority
of information within them was consistent with the handler, and thus could be used for
comparison to known profiles. Further, all results completely inconsistent with the handler were
from weak partial profiles, which often resulted from as little as a single allele. This indicates
that a greater variety of profile categories need to be developed, as the four used here are too
vague and encompass too wide a variety of results, making them rather uninformative.
Attempting to fit profile classification into only two categories (consistent or inconsistent with a
handler) likely created a misleading dichotomy, wherein relatively good profiles were
categorized as poor, while those with virtually no information were classified as good. However,
since continuums of alleles were consistent/inconsistent with the handlers’ profiles, attempting to
place each result into a precise category would be unwieldy (or perhaps impossible), thus was
not attempted in this research. An alternative approach to analyzing this type of data may be
developing a random match probability (RMP) for each profile. If every profile were evaluated
in this manner then a greater number of handler alleles would result in a stronger profile. On the
other hand, erroneous alleles and mixtures would make the RMP weaker; overall, the RMP is
more informative in terms of each profile than lumping profiles into different ‘blanketing’
categories.

25

It is interesting to examine profile consistency at the allelic level, as the number of alleles
from a handler is critical in real world applications. In this study allele consistency was
significantly greater when casings were swabbed cumulatively than when swabbed singly, but
this method only improved allele consistency slightly over consensus profiling. Since
cumulative swabbing generated a greater percentage of handler alleles it is logical that it also
resulted in the smallest percentage of inconsistent alleles. This indicates that swabbing five
casings consecutively results in a higher percentage of the handlers’ alleles being recovered. It is
also possible that the increased recovery of handlers’ alleles through cumulative swabbing
swamps out the small amount of background DNA during amplification, which may not be the
case with single swabbing since drop-in occurred with greater prevalence in the profiles
developed using that method.
The purpose of consensus profiling was to filter out inconsistent alleles and the results
illustrate that this technique generated a higher percentage of handler alleles, and a lower
percentage of inconsistent alleles than the single swabbing method. Therefore, consensus
profiling was effective at removing some, but not all, of the inconsistent alleles recovered
through single swabbing. This method, however, was outperformed by cumulative swabbing for
two possible reasons. First, because consensus profiling scored allele peaks based on intensity,
‘Major’ peaks occurring two or more times were automatically placed into the profile. This
means that drop-in occurring twice as a ‘Major’ peak, was interpreted as part of the handlers’
profiles resulting in a greater number of inconsistent profiles. Second, the score required for
alleles to be included in the consensus may have been too low. The score was dropped to 2 for
this study because alleles belonging to the handler often showed up only twice among the five
single swab profiles. The consensus rules did lessen the subjectivity inherent in STR analysis,

26

though empirical rules developed to increase profile consistency would ideally be applicable to
any type of STR profile, while still being modifiable depending on the amount and type of
evidence obtained. The consensus profiling methodology applied in this study was developed
using an entirely different sample type: post-deflagration IED components. The consensus
criteria were, therefore, modified for these data based on the amount of genetic information
obtained from the cartridge casings. Since profiles generated in this study often had few alleles,
the number of alleles required for placement into the consensus profile was based on the most
common number of alleles, two, present at each locus across five single swabs. Given that the
consensus profile rules are influenced by sample type and size it would be worthwhile
reinvestigating the consensus methodology using fired cartridge casings, to see whether the
consensus method can be improved upon.
STR profiles containing fewer non-handler alleles would be forensically advantageous
since drop-in or mixtures decrease a RMP. This could potentially be accomplished by reamplifying extracts and performing multiple injections of the same sample, which should result
in an increase in handler alleles and a decrease of inconsistent alleles (Taberlet et al. 1996;
Whitaker et al. 2001). The more times an extract is injected onto the capillary and the results of
those individual injections are combined, the better the chance of obtaining true alleles from the
handler.
Regardless of the method used to generate STR profiles, it is important to understand the
origin of extraneous alleles. Contamination is always a concern when working with LCN DNA
because it has the potential to be amplified when LCN DNA is not (Gill and Kirkham, 2004).
DNA other than the handler’s may have been introduced post-firing, while they were being
collected and put back into paper bags, but could also have been introduced during DNA

27

processing, though alleles were not consistent with the researchers involved. Since cartridges
were not washed prior to firing the potential for contamination from pre-firing sources was
higher than in previous studies that pre-cleaned cartridges before loading (Horsman-Hall et al.
2009), however if this were the case, contaminating DNA would likely have been detected on the
cartridges that were swabbed for background DNA. Given that no alleles were generated from
such swabs, contamination from outside sources was not likely the cause of extra (inconsistent)
alleles. Although contamination may have occurred when casings fell on the floor, inconsistent
alleles could not be traced back to any specific individual. This raises questions about real-world
crime scenes where casings may be collected from a variety of locations, which could easily
increase contamination, given the environmental conditions in which casings existed, and result
in mixture profiles and erroneous allele calls during downstream analysis. One question that
remains upon completion of this research is how various environmental conditions, such as
precipitation, temperature, ground cover, etc., affect the quantity and quality of DNA recovered
from fired cartridge casings. A study exploring the effects of environmental factors and
collection methods on cartridge casing evidence would be useful for the advancement of forensic
DNA analysis.
The fact that the same extraneous alleles were not present among profiles and that
variable loci were affected indicates that they resulted from random drop-in rather than
contamination. The effects of drop-in are often quite dramatic in extracts containing very small
amounts of DNA, wherein sporadic peaks can be present at the same intensity as alleles
belonging to the handler, giving the impression of a mixture (Taberlet et al. 1996; Gill et al.
2000). In this study, sporadic peaks frequently occurred at the FGA and D5 loci, which
combined resulted in over 60% of inconsistent alleles. These loci are in the red channel in

28

IdentifilerTM, which also regularly shows the most noise among dyes, thus noise may be called as
allele peaks when the RFU threshold is set very low, as was the case here (50 RFU). A possible
remedy would be to increase the threshold in the red channel to greater than 50 RFU, which
would decrease the amount of noise being called, but would also result in the loss of handler
alleles. If the RFUs were increased to 100, more than half of the allelic data from the red
channel in this study would have been lost.
Allelic drop-out was also common in this study, and undoubtedly caused interpretation
problems. Using amelogenin as an example, there were several instances in weak partial profiles
where only one allele existed, which was frequently an uninformative X, even though the
individual was male. The reverse situation also occurred, where the Y locus amplified and X
dropped out. Another common occurrence was X and Y peaks existing in a profile indicating
that the handler was male, when in actuality the handler was female, thus the Y peak dropped in.
For those reasons, peaks at amelogenin were not particularly informative, and of course, the
same was often true of other loci.
Drop-in and drop-out were best demonstrated during the 8 and 12 second injection
comparisons. In numerous instances alleles, both consistent and inconsistent with the handler,
were called after each of the injections, though in most cases, alleles present from 8 second
injections were not the same as those from 12 second injections. Three different scenarios
existed; the 8 second injection had a handler allele that dropped out with the 12 second injection
and an inconsistent allele dropped in, the 8 second injection had inconsistent alleles that dropped
out and a handler allele was picked-up with the 12 second injection, or inconsistent alleles
present after the 8 second injection dropped out after the 12 second injection and different
inconsistent alleles dropped in. These circumstances are cause for concern considering different

29

results were obtained when analyzing the same sample but modifying only capillary injection
time. If handler’s alleles had been called consistently after going from the 8 second to the 12
second injection then it would seem that an increased injection time is a solution to obtaining
handler allele’s more often. This was not the case however, which indicates a need for
optimization and a means of separating handler alleles from erroneous drop-in.
Perhaps the best approach for recognizing and eliminating drop-in in LCN samples was
suggested by Taberlet et al. (1996), who recommended multiple amplifications of the same DNA
extract. This was based on the probability of a given allele dropping in being approximately 5%,
which meant that the probability that the same erroneous allele occurring in two separate
amplifications was less than 1%. They and others have advised that replicates be evaluated to
develop a “composite profile” where alleles are considered ‘true’ if they are present in each of
the replicates (Taberlet et al. 1996; Whitaker et al. 2001). Development of a composite profile
from several amplifications of a single extract should result in a decreased number of erroneous
alleles, similar to the consensus profiling method used in this research. However, the consensus
method was somewhat different because it looked at repeatability among DNA extracts rather
than within a single extract. Using the consensus method, drop-in from cartridge casings handled
by a single individual were often effectively filtered out. Drop-out, on the other hand, could
potentially be addressed by strengthening the signal for that allele, including increasing DNA
injection time or voltage during electrophoresis (Budowle et al. 2001; Horsman-Hall et al. 2009).
The increased injection times tested in this study generally resulted in more handler alleles, but
as noted, some drop-out and new inconsistent alleles also occurred.
Overall the methods tested in this study did slightly improve on results from earlier
research of forensic DNA analysis from cartridge casings. Given that the majority of handler

30

profiles were partial, the forensic utility of swabbing cartridge casings becomes a concern, but in
fact partial profiles can be very informative. As an example, they can be used for calculating
RMPs as discussed previously, making each profile informative to a certain extent. Even if the
probability of the evidence coming from the suspect or someone other than the suspect is 1 in
100 or 1 in 10,000 (as opposed to 1 in 100s of trillions found in a full profile), any such number
can have value in court (Budowle et al. 1998). The more the guilt of the suspect can be
supported by STR data, the more worth it has in the minds of the jurors. If quantitative
information can be presented even without a full profile then it makes even limited STR data
useful, as compared to fingerprinting or firearms examination where analysis cannot be upheld
with numbers or statistics.
Partial profiles can also be used to search databases. The Combined DNA Index System
(CODIS) is a US DNA database available to crime laboratories. It has two primary indices:
convicted offenders and forensic casework. For profiles to be uploaded into the forensic
casework index they must have at least 10 CODIS loci with genetic information, while the full
13 are required for convicted offenders (Budowle et al. 1998; FBI, 2011). Profiles then, remain
in the database acting as a foundation to which others are compared.
Searching CODIS requires far fewer alleles than uploading a profile; in fact data from a
single locus can theoretically be interrogated. This means that any of the results obtained in this
study could have some utility. When a search is performed in CODIS using partial profiles, a
partial match between two single source profiles may result, where the known profile contains all
of the alleles of the questioned one, while the questioned profile contains a subset of the known
alleles (FBI, 2011). Although partial profiles are not ideal, they may provide investigative leads
or be useful for exclusions. In this regard, the single swabbing method could prove preferable to

31

the cumulative swabbing method in that the latter provides only one ‘chance’ for detecting
alleles belonging to the handler of cartridges, whereas the single swabbing method provides
several opportunities for handler alleles to show up multiple times among the swabs. This could
be particularly important because excluding an individual requires only a single inconsistent
allele between evidence and suspect profiles. If, for example a set of cartridge casings were
submitted as evidence and an analyst was using the cumulative swabbing method, and the result
was a profile containing alleles inconsistent with the suspect, there is no ‘second chance’ to
obtain alleles consistent with the handler. It is possible; however, that a single swab of a casing
would have picked up the true alleles and the suspect and known profiles would match at all loci.
Thus, with regard to databases such as CODIS, partial STR profiles, though not ideal, can be
helpful in generating leads in criminal cases and in exclusion of suspects.
In their study of DNA analysis from fired cartridge casings, Horsman-Hall et al. (2009)
compared profile types between fired and unfired casings handled by five individuals. One of
the unfired cartridges produced a full profile and the remaining 4 produced strong partial
profiles. The fired casings, on the other hand, resulted in no full profiles (the authors did not
address partial profiles in this section of the study, strong or otherwise). The authors also
assessed DNA recovery from 5 types of fired ammunition using MinifilerTM and demonstrated a
distribution of profile types, similar to the profile distribution of this research: after injecting
DNA from 75 samples for 12 seconds onto an AB 3130, greater than 50% of casings resulted in
no profiles, followed by weak partial profiles (26%), strong partial profiles (19%) and full
profiles (1%). IdentifilerTM produced no alleles for any of the 5 ammunition types. It is not
surprising that MinifilerTM produced better results because it has an enhanced buffer to help
overcome PCR inhibition, which also exists in IdentifilerTM Plus that was used in the current

32

study. It is surprising, however, that in the majority of cases no profiles were obtained. In this
study, volunteers handled cartridges for a much shorter time, yet weak partial profiles were often
obtained. Unfortunately, Horsman-Hall et al. (2009) did not state whether their results were
consistent or inconsistent with the handlers, so profile type results are the only possible
comparison to the research presented here.
Horsman-Hall et al. (2009) also examined the recovery of DNA from firearm surfaces.
This portion of their study does not relate directly to the experiments presented here, but it does
raise an interesting point. They tested three surfaces, the ejection port, breechface, and barrel,
which were swabbed in between sets of 3 shooters, for a total of nine swabs. When the shooter
profiles were compared to the profiles obtained from the firearm surfaces, allelic drop-in
consistent with a previous shooter was observed. This indicates that cartridges may pick up
contaminating DNA from the gun surface that was left by a previous shooter upon loading or
casing ejection. This could affect the current study because shooters shared guns and magazines
without cleaning in between. Transfer of shooter DNA to other casings would not be surprising,
especially when using the firing tank, where every cartridge casing was caught in the same net.
The results of this study raise several new questions that could be investigated.
Optimizing protocols and instrumentation would be highly beneficial in increasing the numbers
of handlers’ alleles recovered and obtaining profiles with less noise and drop-in. To achieve
such goals, cycle number during STR amplification should be adjusted to yield the greatest
amount of DNA product possible; although this is difficult since stochastic effects are inherent
with low copy number samples. Similarly, due to stochastic sampling and drop-in, multiple
injections of the same DNA extract should be explored and applied in a manner similar to
consensus profile analysis in an effort to eliminate noise and drop-in. With the results of this

33

study being relatively variable with regard to the 8 and 12 second injection times, capillary
injection times for DNA extracts should be optimized as well. The longer injection time
produced a greater number of handler alleles, but it is unclear if a longer injection is always
better, or only if a sample is LCN. If capillary injection time affects overall allele recovery then
it should be investigated in more detail.
Further research could also be performed on the DNA quantitation and identification
methodologies themselves. One important undertaking, especially for LCN DNA work, is the
development of a quantitative assay that uses STR loci rather than the current QuantifilerTM
locus, as DNA quantitation would be more representative of the loci being assayed.
Alternatively, testing the effectiveness of alternative identification methods, such as
mitochondrial DNA analysis, could prove valuable in collecting data from fired cartridge
casings.
Maximizing cell collection is imperative in optimizing DNA yields and STR profiling,
and various collection techniques should be investigated. For instance, a larger study comparing
DNA recovery through cumulative or single swabbing would be helpful, particularly since the
tendency of individuals to shed skin cells is highly variable (Lowe et al. 2002). Likewise, a
soaking method, where fired cartridge casings are placed in a soaking solution to release all
DNA from the surface, could increase DNA yields. However; PCR inhibition caused by gunshot
residue from the inside of cartridge or the metal casing reacting with the reagents in the soaking
solution might result. All of these could be explored.
As mentioned, the categorization system used to analyze the profiles in this study may
have been too broad to be highly informative. Therefore, it would be helpful to explore a variety
of ways to pull pertinent information out of the data obtained from the STR profiles. Using

34

random match probability calculations to determine the frequency of a DNA profile would be
more informative than determining whether the profile fell into categories, such as weak or
strong or full. Another means of indicating how useful a profile is would be to calculate the
percent of the profile that was recovered, be it 100%, 95%, 50%, etc. There are likely several
other ways to establish how genetically or forensically informative an STR profile is, but these
two would work well when applied to the results obtained here.
Lastly, since this was a controlled experiment that occurred indoors away from the
elements, the results may not be applicable to cartridge casings found at an outdoor crime scene.
Therefore, the next step would be to investigate the effects of different environmental conditions,
such as precipitation, ground condition (e.g., gravel, sand, asphalt), temperature, etc., on the
quantity and quality of DNA recovered from cartridge casings under those circumstances.
Overall, profiles containing handler alleles, even if few in number, are useful and can aid
in criminal investigations. Crime laboratories must recognize that successfully recovering and
analyzing LCN DNA can be difficult but highly valuable. As criminals find ways to circumvent
existing forensic technologies and analytical methods, it is important for researchers to continue
pushing the limits. The development of new, more sensitive, and effective methods for
analyzing LCN DNA are imperative in achieving greater success in producing genetic profiles
from fired cartridge casings. Data from this study show that using IdentifilerTM Plus, in
combination with the swabbing and consensus profiling methods, generated handler alleles more
often than extraneous alleles. The fact that cumulative, single swabbing, and consensus profiling
all produced handler alleles demonstrates their utility and applicability to forensic casework.

35

CONCLUSION
DNA recovered from fired cartridge casings ranged from zero to LCN, which indicated
that, although there is an ideal quantity of DNA for optimum STR results, lower quantities can
still result in genetically informative profiles. Cumulatively swabbing cartridge casings resulted
in greater DNA yields on average, although this did not correlate with the number of handler
alleles recovered. No relationship existed between swabbing method and the resulting profile
type: full, strong partial, weak partial, or none. However, cumulative swabbing generated a
significantly higher percentage of handler’s alleles than did individual swabbing.
Given that the recovered DNAs were LCN in nature and were subject to PCR inhibition,
contamination, drop-in, and drop-out, steps should be taken to address sources of inconsistent
alleles. Utilizing a DNA purification kit that removes PCR inhibitors and an STR amplification
kit designed for LCN DNA would likely augment results. Multiple amplifications of extracts in
combination with consensus profiling should be used to address the remaining sources of allelic
inconsistency. Instituting these measures will assure that the greatest numbers of handler alleles
are obtained. Though cross-contamination is a concern, based on the results of this study, the
cumulative swabbing method appears to be best for recovering and analyzing DNA from fired
cartridge casings.

36

APPENDICES

37

APPENDIX A. DNA QUANTITIES FOR CUMULATIVE AND SINGLE SWABS
Extracts identified with an “A” indicate quantities recovered using the cumulative
swabbing method. Extracts identified with “B – F” indicate quantities recovered using the single
swabbing method. All DNA yields are represented in pg/µL. Quantities highlighted in yellow
indicate values that were removed as outliers, as they were three or more standard deviations
from the mean DNA yield.

38

Table A.1. DNA quantities recovered from five fired cartridge casings using the cumulative
swabbing method.

39

Table A.2. DNA quantities recovered from five fired cartridge casings, B – F, using the single
swabbing method

40

APPENDIX B: ALLELE CALLS AT EACH LOCUS FOR ALL SETS OF FIRED
CARTRIDGE CASINGS
Allele calls are shown at each locus for all sets of fired cartridge casings. The letter A
indicates profiles developed using the cumulative swabbing method, while B – F denote those
profiles generated using the single swabbing method. Profiles B – F were combined to develop
the consensus profile for each set of 5 casings that were singly swabbed, these are shown in the
last column of each table. The bolded numbers denote allele calls that were consistent with the
handlers’ profiles. (-) indicates that no alleles from the five single swabs were placed in the
consensus profile. It should be noted that casing set #12 has allele calls but no numbers are
bolded as there was no STR amplification from its corresponding buccal swab.

41

Table B.1. Alleles obtained at each locus for casing set #1.

Table B.2. Alleles obtained at each locus for casing set #2.

42

Table B.3. Alleles obtained at each locus for casing set #3.

Table B.4. Alleles obtained at each locus for casing set #5.

43

Table B.5. Alleles obtained at each locus for casing set #6.

Table B.6. Alleles obtained at each locus for casing set #7.

44

Table B.7. Alleles obtained at each locus for casing set #8.

Table B.8. Alleles obtained at each locus for casing set #9.

45

Table B.9. Alleles obtained at each locus for casing set #10.

Table B.10. Alleles obtained at each locus for casing set #11.

46

Table B.11. Alleles obtained at each locus for casing set #12. The known handler specimen for
individual #12 did not generate any profile data therefore it cannot be known which of the allele
calls from the swabbed casing were consistent with the handler.

Table B.12. Alleles obtained at each locus for casing set #13.

47

Table B.13. Alleles obtained at each locus for casing set #15.

Table B.14. Alleles obtained at each locus for casing set #16.

48

Table B.15. Alleles obtained at each locus for casing set #17.

Table B.16. Alleles obtained at each locus for casing set #18.

49

Table B.17. Alleles obtained at each locus for casing set #19.

Table B.18. Alleles obtained at each locus for casing set #20.

(NC) – indicates the allele was not called by the GeneMapperTM Software

50

Table B.19. Alleles obtained at each locus for casing set #21.

Table B.20. Alleles obtained at each locus for casing set #22.

51

Table B.21. Alleles obtained at each locus for casing set #23.

Table B.22. Alleles obtained at each locus for casing set #24.

52

Table B.23. Alleles obtained at each locus for casing set #25.

Table B.24. Alleles obtained at each locus for casing set #26.

53

Table B.25. Alleles obtained at each locus for casing set #27.

Table B.26. Alleles obtained at each locus for casing set #28.

54

Table B.27. Alleles obtained at each locus for casing set #29.

Table B.28. Alleles obtained at each locus for casing set #30.

55

Table B.29. Alleles obtained at each locus for casing set #31.

Table B.30. Alleles obtained at each locus for casing set #32.

56

Table B.31. Alleles obtained at each locus for casing set #33.

57

REFERENCES

58

REFERENCES
®

Applied Biosystems. 2006. AmpFℓSTR Identifiler™ Plus PCR Amplification Kit User’s
Manual, Applied Biosystems, Foster City, CA.
Applied Biosystems. 2009. The Impact of Data Quality on Data Interpretation. Forensic News.
Applied Biosystems, Foster City, CA.
http://marketing.appliedbiosystems.com/images/all_newsletters/forensic_1109/articles.html
Bicigo D. Michigan State Police DNA Unit: Personal Communication
Budowle B, Moretti TR, Niezgoda SJ, and Brown BL. 1998. CODIS and PCR-Based Short
nd
Tandem Repeat Loci: Law Enforcement Tools. Proceedings of the 2 European Symposium on
Human Identification. Pp. 73 – 88.
Budowle B, Hobson DL, Smerick JB, and Smith JAL. 2001. Low Copy Number—
th
Consideration and Caution. Proceedings of the 12 International Symposium on Human
Identification.
Butler JM. 2005. STR Training Materials. http://www.cstl.nist.gov/biotech/strbase/training.htm
Carman WY. 1955. A History of Firearms: From Earliest Times to 1914. 1 – 14.
Clark H. Michigan State Police DNA Unit: Personal Communication
FBI. 2010. Crime Statistics.
www.fbi.gov/news/stories/2010/november/ucrtool_112910/ucrtool_112910
FBI. 2011. Frequently Asked Questions on the CODIS Program and the National DNA Index
System. www.fbi.gov/about-us/lab/codis/codis-and-ndis-fact-sheet.
Foran, DR. 2006. Relative Degradation of Nuclear and Mitochondrial DNA: An Experimental
Approach. Journal of Forensic Sciences. 51: 769 – 770.
Gill P, Whitaker J, Flaxman C, Brown N, and Buckleton J. 2000. An Investigation of the rigor
of interpretation rules for STRs derived from less than 100 pg of DNA. Forensic Science
International, 112: 17 – 40.
Gill P and Kirkham A. 2004. Development of a simulation model to assess the impact of
contamination in casework using STRs. Journal of Forensic Sciences, 49: 485 – 491.
Guardian News and Media Limited. 2011. Gun Crime Statistics by US State.
www.guardian.co.uk/news/datablog/2011/jan/10/gun-crime-us-state.

59

Hauge XY and Litt M. 1993. A study of the origin of ‘shadow bands’ seen when typing
dinucleotide repeat polymorphisms by the PCR. Human Molecular Genetics, 2: 411 – 415.
Horsman-Hall KM, Orihuela Y, Karczynski SL, Davis AL, Ban JD, and Greenspoon SA. 2009.
Development of STR profiles from firearms and fired cartridge cases. Forensic Science
International: Genetics. 3: 242 – 250.
How Stuff Works. 2001. The Modern Bullet Cartridge.
http://static.howstuffworks.com/gif/revolver-bullet.gif.
Larrison R. Michigan State Police Firearms Unit: Personal Communication
Legal Community Against Violence (LCAV). 2011. Gun Violence Statistics.
www.lcav.org/statistics-polling/gun_violence_statistics.asp.
Lowe A, Murray C, Whitaker J, Tully G, and Gill P. 2002. The propensity of individuals to
deposit DNA and secondary transfer of low level DNA from individuals to inert surfaces.
Forensic Science International. 129: 25-34.
Office of Juvenile Justice and Delinquency Prevention (OJJDP). 1996. Section I: Gun Violence
in the United States: The Nature of the Problem and Current Trends.
www.ojjdp.gov/pubs/gun_violence/sect01.html.
Peterson HL. 1961. A History of Firearms. 1 – 20.
R Project for Statistical Computing. 2011. www.r-project.org.
Ricketts H. 1962. Firearms: Period Firearms.
www.silcom.com/~vilman/isles/scriptorium/firearm/firearm.html.
Roads CH and Hogg I. 1983. Pollard’s History of Firearms: General Survey of Firearms 1918
to the Present Day. 387 – 408.
Spear T, Clark J, Giusto M, Khoshkebari N, Murphy M, and Rush J. 2011. Fingerprints and
Cartridge Cases: How Often are Fingerprints Found on Handled Cartridge Cases and Can These
Fingerprints Be Successfully Typed for DNA? California Criminalistics Institute.
Sweet D, Lorente M, Lorente JA, Valenzuela A, and Villanueva E. 1997. An improved method
to recover saliva from human skin: the double swab technique. Journal of Forensic Sciences,
52: 320 – 322.
Taberlet P, Griffin S, Goossens B, Questiau S, Manceau V, Escaravage N, Waits LP, and Bouvet
J. 1996. Reliable Genotyping of Samples with Very Low DNA Quantities Using PCR. Nucleic
Acids Research, 24: 3189 – 3194.

60

Walsh PS, Erlich HA, and Higuchi R. 1992. Preferential PCR amplification of alleles:
mechanisms and solutions. Genome Research. 1: 241 – 250.
Whitaker JP, Clayton TM, Urquhart AJ, Millican ES, Downes TJ, Kimpton CP, and Gill P.
1995. Short tandem repeat typing of bodies from a mass disaster: high success rate and
characteristic amplification patterns in highly degraded samples. BioTechniques, 18: 670 – 677.
Whitaker JP, Cotton EA, and Gill P. 2001. A Comparison of the characteristics of profiles
®

produced with the AMPFlSTR SGM Plus™ multiplex system for both standard and low copy
number (LCN) STR DNA analysis. Forensic Science International, 123: 215 – 223.
Wilson IG. 1997. Inhibition and Facilitation of Nucleic Acid Amplification. Applied and
Environmental Microbiology. 63: 3741 – 3751.

61