Canadian Society of Forensic Science Journal

ISSN: 0008-5030 (Print) 2332-1660 (Online) Journal homepage: http://www.tandfonline.com/loi/tcsf20

Dog Hair Comparison: A Preliminary Study

Thomas William Suzanski

To cite this article: Thomas William Suzanski (1988) Dog Hair Comparison: A
Preliminary Study, Canadian Society of Forensic Science Journal, 21:1-2, 19-28, DOI:
10.1080/00085030.1988.10756958

To link to this article: http://dx.doi.org/10.1080/00085030.1988.10756958

Published online: 22 Nov 2013.

Submit your article to this journal

Article views: 5

View related articles

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tcsf20

Download by: [137.189.171.235] Date: 13 May 2016, At: 04:39

 Can. Soc. Forens. Sei. J. VoI. 21, Nos 1 & 2 (1988)

DOG HAIR COMPARISON: A PRELIMINARY STUDY

THOMAS WILLIAM SUZANSKI 1

ABSTRACT
Due to their close association with man, dog hairs are frequently recovered from ex-
hibits submitted to the forensic laboratory. Easily shed and exchanged, they could serve
as a source of associative evidence if it could be established that different dogs of the
same breed possess hair with sufficient variation of microscopic characteristics to per-
mit some degree of individualization. Known dog hair samples were obtained from 25
pure bred German Shepherds who shared common lineage. Questioned hairs selected
in the examiner's absence were compared macroscopically and microscopically to the
known samples. Six of fifteen questioned hairs were correctly assigned to their sample
of origin. No incorrect inclusions (Type 11 errors) occurred. The results of this prelimi-
nary study indicate that pure bred German Shepherd dogs possess some types of hair
with sufficient variation of microscopic characteristics to permit some degree of individu-
alization.
Downloaded by [] at 04:39 13 May 2016

RESUME
Du a leur association etroite avec I'homme les poils de chiens sont souvent preleves
sur des exhibits soumis au laboratoire de sciences judiciaires. Etant donne que leur
transfert est facile a effectuer i1s peuvent servir a des fins d'identification si on peut
demontrer que des chiens de mime race possede des poils avec une variation suffi-
sante nous permettant de les individualiser. Des echantillons d'origine connue prove-
nant de 25 bergers allemands de mime Iignee furent soumis pour expertises. Des poils
furent preleves en I'absence du specialiste qui par la suite compara, macroscopique-
ment et microscopiquement, ces prelevements a des echantillons d'origine connue. Six
des quinze poils en Iitige furent relies a leur echantillon original, alors qu'aucune fausse
relation fut etablie. Ces resultats indiquent que mime chez des chiens bergers alle-
mands de pure race on peut deceler une variation microscopique suffisante nous per-
mettant des les individualiser.

INTRODUCTION
Is a potential source of associative evidence being ignored?
Dogs share a close association with man. They possess a vast number of hairs which
are easily shed and exchanged. Primary hair transfer to the dog' s owner or it's environ-
ment and subsequent secondary hair transfer to a crime scene or murder victim is theoret-
ically very possible.
It is well known that human scalp hair possesses sufficient variation in microscopic charac-
teristics to permit a high degree of individualization. They are often used to associate two
people, a person and a crime scene, or a person and a weapon. If it could be established
that different dogs of the same breed possess hair with sufficient variation of microscopic
characteristics to permit some degree of individualization, then a type of hair long thought

l.
Hair & Fibre Section, Forensic Laboratory Regina. Royal Canadian Mounted Police, Box 6500, Regina,
Saskatchewan S4P 317.

19
 to be useless by many forensic hair examiners would prove to represent a valuable form
of associative evidence.
Little significance has been attached to this type of evidence because of a perceived absence
of morphological variation upon which to make a valid comparison. As a result, little formal
work has been done, and no published literature exists regarding the significance of dog
hair comparison. Despite this, their value to the courts as associative evidence has not
been limited strictly to identification. In the Wayne Williams case (1), three highly quali-
fied hair examiners independently examined questioned dog hairs recovered from the bodies
of murder victims and each gave evidence indicating that the questioned hairs were con-
sistent with having originated from the accused's family dog.
Of particular relevance to this study was the testimony of B. D. Gaudette wherein one
experiment and two case studies were described. In connection with the above case, hairs
from 12 German Shepherd dogs were compared to a known sample obtained from the
Williams' dog. The results of the study were as follows:

- eight of the 12 dogs had no hairs which were indistinguishable from any of the hairs
from the Williams' dog.
- three of the 12 dogs had one variety (hereafter referred to as subtype) of hair which
was indistinguishable from one subtype of hair from the Williams' dog.
Downloaded by [] at 04:39 13 May 2016

- one of the 12 dogs had two subtypes of hair which were indistinguishable from two
subtypes of hair from the Williams' dog.
- no dog had more than two subtypes indistinguishable from the Williams' dog.

Secondly, in connection with a different murder case where 80 questioned dog hairs
of four distinct subtypes were found on various locations on the victim's body, examina-
tions of known samples from 20 different dogs showed only one of these dogs had any
hairs which were indistinguishable from the questioned hairs recovered from the body.
Only a few of these hairs matched, the great majority being dissimilar.
Finally, in another murder case (Regina vs Harold Smeltzer) where 26 dog hairs were
found on a body, 480 dogs were screened and 92 different dog hair samples were sub-
mitted to the laboratory. The examiner, K. E. Sanders, determined that of these 92 dogs,
only three were the possible source of the questioned dog hairs on the body. One of the
three dogs was later identified as being owned by the person who was later convicted of
the crime2 .
From the above and from additional testimony given by Gaudette and Deadman during
the Wayne Williams trial, the following important concepts emerged:

1. The principles of hair comparison are generally the same for dog hair comparison
as for human hair comparison. The relative weights to be put on certain characteristics
will be somewhat different.
2. It is possible for two different dogs to have hair which is indistinguishable. Dog hair
comparison, like human hair comparison, is not a positive means of demonstrating
association.
3. Dog hair can provide reasonably strong evidence demonstrating association, but should
not be considered as significant as association as a human hair match. There are fewer
2. K.E. Sanders, Personal Communication.

20
 characteristics and the variation in characteristics is not present between different dogs
in the same way that the variation and characteristics are present between different
humans.
4. Only guard hairs and intermediate hairs are suitable for comparison purposes.
5. Some hairs are very common and could have come from a number of dogs. Some hairs
exhibit more unusual characteristics.
6. Dog hair is likely to be found on most items in a dog's associated environment. Should
an object or person come in contact with this environment, one would expect transfer
to take place and dog hair to be present.

It should be noted that the above cases involved the transfer and recovery of large numbers
of dog hairs. It may be logical to conclude that because of their greater number and their
tendency to be easily shed, multiple comparisons to known samples will occur more fre-
quently with dog hair than with human hair. If dissimilar subtypes are found to be consis-
tent with a known sample, this would strengthen the significance of any associations made.
However, the present study involves the comparison of single questioned hairs of various
subtypes with known samples. The purpose of this study was to identify some of the problems
encountered in conducting dog hair comparisons, and to take a step towards establishing
their significance.
Downloaded by [] at 04:39 13 May 2016

MATERIALS AND METHODS

Hair samples were obtained from 25 pure bred German Shepherd dogs who shared
common lineage. Ages of the dogs ranged from six months to nine years. The population
consisted of 15 females and 10 males exhibiting two basic coat types: 12 black and tan
and 13 sable. Hairs were pulled from various parts of the body in a random fashion.
Known hair samples from each dog were sorted into various subtypes. A subsample
of each of the hair subtypes was selected from each known sample to represent the range
of hair exhibited by that particular dog. These hairs were then measured for length and
mounted on microscope slides using Permount as an embedding medium.
In the examiner's absence, 15 questioned hairs were selected from amongst the remaining
hairs in the known samples. The examiner had no knowledge of the samples from which
the questioned hairs originated, or if they had in fact originated from any of the known
samples.
Questioned hairs were subtyped, measured, mounted as before and then compared mac-
roscopically and microscopically to each of the 25 known samples using a Leitz Ortholux
11 comparison microscope. Microscopic comparisons were carried out almost exclusively
at 250x magnification.

RESULTS
Cursory examination of known samples which contained large amounts of hair, indi-
cated that some discrimination between the coat types black and tan, and sable was poten-
tially possible. However, closer examination revealed that although some distribution
differences were apparent, no clear distinctions could be made. Each dog demonstrated
similar types and subtypes of hair as follows:

21
 Types
1. guard hairs
2 . intermediate hairs
3 . wool hairs (white, brown and black)

Subtypes*
1. banded
2. black tipped (long, medium and short)
3. black
4. white
5. brown tipped
*See Discussion for an explanation of subtypes.
The 15 questioned hairs were subtyped and compared to appropriate subtypes in the
known samples. Results are shown in Table 1. Six of the 15 questioned hairs were cor-
rectly associated with the known sample from which they originated: three banded hairs
and three black tipped hairs (one long and two medium). Five Type I errors (incorrect
eliminations) were made. Two of these Type I errors were due to examiner error which
was determined after reviewing known samples. Two of the Type I errors occurred because
the questioned hairs were not representative of the known sample from which they origi-
Downloaded by [] at 04:39 13 May 2016

nated. The fifth Type I error occurred because questioned hair #9 was atypical. Since no
TABLE 1

Questioned Known
Hair Number, Type Sample of Examiner's
Subtype Origin Result

I. Guard, banded Eliminated Type I error

(examiner error)
2. Guard, banded 2 Correctly identified
3. Guard, banded 23 Correctly identified
4. Wool hair, black 23 Unsuitable
5. Guard, black tipped 23 Correctly identified
(long)
6. Guard, white II Unsuitable
7. Guard, banded II Correctly identified
8. Guard, black tipped 6 Correctl y identified
(medium)
9. Guard, black 25 Eliminated Type I error
(atypical hair)
10. Intermediate, black 25 Unsuitable
tipped (short)
11. Intermediate, black 16 Eliminated Type I error
(not representative of
known sample)
12. Guard, black tipped 16 Eliminated Type I error
(medium) (not representative of
known sample)
13. Guard, black 21 Eliminated Type I error
(examiner error)
14. Guard, black tipped
(medium) 8 Correctly identified
15. Guard, white 8 Unsuitable

22
 hairs similar to it could be found in any of the known samples, a microscopical compar-
ison could not even be attempted. The remaining 4 questioned hairs (#4, #6, #10 and #15)
when considered individually were thought to lack suitable individualizing microscopic
characteristics upon which to make a meaningful comparison. It was concluded that they
could have originated from any number of dogs and that an association based solely on
these types of hairs (wool, white and black tipped - short) would have little significance.
No Type II errors occurred in this study.

DISCUSSION

German Shepherd Hair Subtypes

Banded Guard Hair

Banded hair possesses a color band, described by Moore (2) as being a band of color
on a hair which is sharply separated from adjacent areas of different colors above and
below the band.
In general only the 4 small areas illustrated in Figure 1 presented enough suitable
microscopic characteristics to enable meaningful comparisons to be made. Pigment color,
Downloaded by [] at 04:39 13 May 2016

size, distribution, density and texture were the major characteristics used for discrimina-
tion. The majority of the shafts presented few useful characteristics for comparison because

OPAQUE WHITE OPAQUE" '"

"ZONE ZONE ZONE \

/~
~ I
,,-"
\. .,-..... ~
J
--- \ ~ '\
\ I I
.... ~ ..... \ / I
,
... - I
(1) Root and (2) Proximal (3) Distal (4) Tip
Proximal Shaft Transition Zone Transition Zone and
Distal
Shaft

Figure I. Banded guard hair - Note, only four small areas present characteristics suitable for comparison.

the hair was either too opaque (due to a high concentration of pigment) or as in the white
zone, completely devoid of pigment.

Black Tipped Guard Hair

These hairs are also known as bicolored hairs (2). Macroscopically they were of two
colors, white and black. Three varieties were distinguished on the basis of the length of
the pigmented tip: long, medium and short (Figure 2).
In these hairs, useful characteristics for comparison were mainly confined to the transi-
tion zone and tip regions of the shaft. The long and medium varieties generally showed
more characteristics useful for comparison. The short varieties were found to be unsuitable
for comparison as single questioned hairs since they possessed much finer pigment charac-
teristics, and since the examiner lacked confidence in the remaining characteristics. It was

23
 , -- - LONG

C:-:::=::'~!-,
,
--.........-

TRANSITIO~
~_
~" -' ,-' ~"'"' ~
ZONE '-~ V
"

"" - .... , SHORT

<:>-- ~-...,. ...:-- .~

... '

Figure 2. Black tipped guard hair (long, medium and short) - Characteristics suitable for comparison are
mainly confined to transition zones and tip regions.
Downloaded by [] at 04:39 13 May 2016

felt that these hairs could have originated from any number of pure bred German Shepherds.

Black Guard Hair

Macroscopically, these hairs were black from root to tip and presented only two regions
useful for comparison:

1. root and proximal shaft

2. tips

Otherwise, they were too opaque due to a high concentration of pigment along the
shaft.
Although no correct positive associations were made with black hairs, after reviewing
the known samples, it is believed that questioned hair #13 should have been correctly iden-
tified to its sample of origin. The other questioned black hair in this study (#11) also pos-
sessed some good characteristics for comparison. Although incorrectly eliminated, it was
later determined that this questioned hair was not representative of the known sample from
which it originated.

White Guard Hair

Macroscopically, these hairs were white from root to tip. In the absence of pigment
characteristics it was felt that these hairs could have originated from any number of pure
bred German Shepherd dogs and therefore were not suitable for comparison as single ques-
tioned hairs.

Brown Tipped Guard Hair

These hairs are similar to black tipped hairs except that macroscopically the tips appear
brown. They are found much less frequently as a subtype than black tipped hairs. Since
none of the questioned hairs was of this subtype, detailed microscopic examinations were
not made with respect to their suitablility for comparison.

24
 Wool Hair
No useful characteristics suitable for comparison were observed. Some distribution differ-
ences were observed. Sable coats possess much more white colored wool hair, while black
and tan coats possess much more black colored wool hair.

Comments on Unsuitable Subtypes

Summarizing, three subtypes of German Shepherd dog hair were found to be unsuitable
for purposes of comparison as single questioned hairs:

1. wool hair (white, brown or black)

2. white hair
3. black tipped hair (short) - white hairs with extremely small amounts of fine reddish
or brown pigment at the tip.

These findings are in disagreement with those of B.O. Gaudette3 who has found white
hairs to be generally useful for comparison with comparison characteristics including the
following:

1. degree of whiteness (grey white vs yellow white) as viewed microscopically

Downloaded by [] at 04:39 13 May 2016

2. presence and amount of pigment in the root area

3. extent, size, shape and distribution of fusi
4. amount of medullation (distribution and point of origin of the medulla as well as medul-
lary index)
5. diameter
6. characteristics of the tip

While these characteristics were noted in the examination of both white and black tipped
(short) hairs, it must be pointed out that there are a number of factors contributing to the
increased difficulty involved in comparing these kinds of hairs:

1. In general, the proportion of white hairs per known sample was found to be relatively
small in comparison to other hair subtypes.
~. These hair subtypes are more featureless.
3. Intrasample variation in characteristics complicates the comparison.
4. Intersample variation with respect to these minor characteristics is much less apparent
to the examiner.

This preliminary study suggests these hairs are unsuitable in that by themselves, a finding
of association would be of little significance and open to question since there are fewer
useful characteristics for purposes of comparison. Obviously, a great deal more research
needs to be done with respect to single questioned white dog hairs.
These unsuitable subtypes are certainly useful for purposes of elimination where the
known sample does not contain these particular subtypes of hair. Secondly, they may be
useful as additional points of comparison in multiple hair comparisons, where many ques-
tioned hairs of common origin (if this can be established) are being compared to a known
sample. Finally, additional study may indeed show that meaningful comparisons are pos-
sible with single questioned white dog hairs.

3.
B. D. Gaudette, Personal Communication.

25
 Atypical Hairs
The particular atypical questioned hair (#9) in this study possessed the macroscopic charac-
teristics illustrated in Figure 3. This hair contained good microscopic characteristics but
was unsuitable for comparison in that it was unrepresentative of any of the other hairs
in the 25 known samples. Only three other atypical hairs were found: one hair banded
white, brown and black respectively from root to tip (known sample #17) and two hairs
banded white, black and brown respectively from root to tip (known sample #13).

~
~II:::::J -~

Figure 3. An atypical hair - small white zone occurs in proximal shaft, otherwise black from root to tip.

Type I Errors
In any hair comparison, a Type I error occurs when an examiner incorrectly eliminates
a questioned hair (3). Three possible causes of Type I errors have been identified by Gaudette
Downloaded by [] at 04:39 13 May 2016

(4):

1. examiner error
2. deficiencies in the known sample:
(a) not large enough
(b) not representative
(c) contains incomplete hairs
(d) large amount of time between date of offence and date when known sample was
obtained
3. atypical hairs

All three types occurred in this study; two due to examiner error, two due to deficien-
cies in the known sample and one due to an atypical hair.

Known Samples
What is an adequate known sample? How many hairs should be collected to provide
an adequate representation of the range of characteristics present on a dog? While it is
widely accepted that a known sample of 100 human scalp hair represents the range of
characteristics present in a person's scalp, no such guidelines exist when dealing with dog
hair. Obviously, a known sample must contain representative amounts of each subtype
present on a dog.
Unfortunately, it is impossible to know in advance what the distribution of subtypes
on a particular dog will be and in the collection of a known sample, it is necessary to
make some basic assumptions. If we assume that 100 hairs of each subtype would ade-
quately represent the range of characteristics exhibited by that subtype on a particular dog,
and if we further assume that there are four common subtypes of guard hair (black, white,
banded, and black tipped) on a dog, and finally, that these subtypes are equally distributed
(which mayor may not be true), it can be concluded that a minimum of 400 to 500 guard
hairs should be taken in a random fashion from each dog to adequately represent the range
of characteristics present on a particular German Shepherd dog.

26
 Human Scalp Hair Comparison vs Dog Hair Comparison
The principles of dog hair (German Shepherd) comparison are similar to those of human
hair comparisons with some important differences. In humans, somatic origins can be deter-
mined due to the discontinuous nature of hair distribution in man. Human pubic hair, scalp
hair, body hair and so on can be thought of as subtypes. Only the same subtypes are com-
pared to each other, just as in dog hair comparison. In dogs, due to the continuous nature
of the coat, determination of somatic origin is much more general and imprecise, but there
appear to be certain characteristics which may give clues. For example, guard hairs of
a wavy nature seem to be characteristic of the tail.
Secondly, there is a wide range of characteristics in human scalp hairs which occur along
the entire length of the shaft. Almost all of the determinations in this study were based
on the color, density, distribution, size of pigment and the texture of the hairs in the roots,
transition zones and tips. A large portion of the shaft had little value for the purposes of
comparison. Other characteristics useful for comparison were:

- length,
- width,
- nature of the medulla at the tip,
- presence of medullary pigment in the transition zones,
Downloaded by [] at 04:39 13 May 2016

- presence of diffuse yellow pigment about the medulla in the white zones of banded hairs,
- the approximate length and degree of whiteness of the white zone in banded hairs as
viewed macroscopically,
- the amount of taper and color of the tip,
- presence and size of cortical fusi in the root area.

The Significance of Dog Hair Comparisons

Can dog hairs provide reasonably strong evidence of association? Clearly, this study
has shown that in closely related German Shepherds, much more variation in microscopic
characteristics is present than was originally thought to exist.
It has been suggested4 that because of their relative length and color variation, German
Shepherd dog hairs are better for comparison purposes than most dog hairs. As a result,
the expected rate of Type 11 errors in this study would be lower than for dog hairs in general.
On the other hand, because this study dealt with closely related dogs of the same breed,
the expected probability of Type 11 errors would be artificially higher than in casework
situations where the suspect population is hairs of all dogs of all breeds. Finally, that these
two effects could be expected to balance out, thereby making this study generalizable to
other dogs in casework situations.
However, for several reasons, dog hair comparisons are less significant as evidence of
association than human hair comparisons:

1. There are less characteristics and thus less variation.

2. Only small portions of the shaft are suitable for comparison.
3. Little research has been done.

Intermediate Hairs
Personal observation of intermediate hairs indicates that these hairs possess sufficient

4. B. D. Gaudette, Personal Communication.

27
 characteristics to attempt comparisons. In fact, because pigment density in these hairs is
much less than in the largely opaque guard hairs, greater portions of the shaft may be
suitable. It must be remembered however that for the purposes of this study, all of the
questioned hairs were known to have originated from a member of the dog family. One
must bear in mind that in a situation where the origins of the hairs are unknown, all hairs
with the exception of guard hairs, may not be suitable for identification let alone compar-
ison. Therefore, when doing a comparison, the questioned hair should ideally be:

1. guard hair
2. banded

CONCLUSIONS
1. Results of this preliminary study indicate that with German Shepherds, some degree
of individualization based on hair examination is possible.
2. In contrast to human scalp hairs, there are fewer characteristics useful for compar-
ison and only small segments of the hairs are generally suitable for comparison.
3. Certain types and subtypes are more suitable for purposes of comparison than others.
Banded guard hairs are particularly promising while white and black tipped (short)
Downloaded by [] at 04:39 13 May 2016

guard hairs are extremely difficult to compare.

4. It is suggested that 400 to 500 hairs would represent a minimum acceptable sample
size in order to account for the range of characteristics present for each commonly
occurring subtype of hair on a German Shepherd dog.

ACKNOWLEDGEMENTS
I would like to express my deep appreciation to Staff Sergeant D. Hepworth for sug-
gesting this study, to B. D. Gaudette for his insights and advice, and to the members of
the German Shepherd Dog Drill Team of Saskatchewan and Velhandus for providing me
with known samples.

REFERENCES
1. Transcript, State of Georgia vs Wayne Bertram Williams, 1982, Criminal Action File No. A-56186.
2. Moore, T. D., Spence, L. E. and Dugnolle, C. E. Identification of the Dorsal Guard Hairs of Some
Mammals of Wyoming. Wyoming Game and Fish Department, 1974, Bulletin #14, pp. 1-177.
3. Gaudette, B. D. A Supplementary Discussion of Probabilities and Human Hair Comparisons. J. For.
Sci. 1982; 27 (2): 279-289.
4. Gaudette, B. D. Strong, Negative Conclusions in Hair Comparison - A Rare Event. Can. Soc. Forensic
Sci. J. 1985; 18 (I): 32-37.

28