AN EVALUATION OF THREE STERNAL RIB END AGING METHODS ON
AN AMERICAN SAMPLE
By
Nicole Lynne Geske

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Forensic Science – Master of Science
2013

ABSTRACT
AN EVALUATION OF THREE STERNAL RIB END AGING METHODS ON
AN AMERICAN SAMPLE
By
Nicole Lynne Geske
The estimation of age is essential in the fields of physical and forensic anthropology;
therefore the accuracy of these methods needs to be evaluated. Currently, there are multiple
methods available for age estimation, including sternal rib end aging, which was specifically
evaluated in this research.
Three sternal rib aging techniques, İşcan et al. (1984a; 1985), Hartnett-Fulginiti (2007;
2010), and Oettlé and Steyn (2000), were the focus of this study. The main objective was to
determine the overall accuracy of each method, as well as examine differences in rates of
accuracy between European-Americans and African-Americans. Data was collected from the
Hamann-Todd and WM Bass collections. The sample consisted of male and female right fourth
sternal rib ends from 411 individuals of both European and African ancestry, ranging in age from
16 to 97 years.
Results demonstrate that all three methods perform equally well, although the HartnettFulginiti (2007; 2010) method is a slightly better predictor of age. Because the Oettlé and Steyn
(2000) method was designed specifically on individuals of African ancestry from South Africa, it
is not recommended for use within the United States. Additionally, due to similarities in
accuracy for the Hartnett-Fulginiti (2007; 2010) and İşcan et al. (1984a; 1985) methods, both
methods are a valuable age estimation technique in the field of physical anthropology and should
be utilized based on level of experience and preference.

ACKNOWLEDGEMENTS
This thesis would not have been possible without the support and guidance from many
individuals. First of all, I would like to thank my thesis committee, Dr. Todd Fenton, Dr. Norman
Sauer, and Dr. Sanja Kutnjak Ivkovich, for their comments and suggestions.
I would also like to thank the collections managers who allowed access of the skeletal
material used for this research. This includes Mr. Lyman Jellema and the Cleveland Museum of
Natural History for allowing me to utilize the Hamann-Todd Osteological Collection. Also, Dr.
Dawnie Steadman and the University of Tennessee, Knoxville for permitting me to use the
William M. Bass Donated Skeletal Collection.
Thank you to all of my fellow graduate students. Specifically, Ashley Kendell for her
comments and for being there when I needed. Also, Julie Fleischman, who gave me great
feedback and helped me tremendously in this process. I would have been lost without you.
Additionally, thanks to my family and friends. Especially, my mother for support
throughout my entire academic career and my father for giving me much needed statistical
advice. I would also like to thank my mentor and friend, Robin Lillie, who gave me a chance and
more experience than I could have ever imagined.
Finally, I would like to thank Michigan State University’s Graduate School and Forensic
Science Masters Program for their financial assistance of this research.

iii	
  

TABLE OF CONTENTS
LIST OF TABLES ....................................................................................................................... vi
LIST OF FIGURES ...................................................................................................................... x
INTRODUCTION ........................................................................................................................ 1
Pubis ................................................................................................................................... 1
Auricular Surface ............................................................................................................. 3
Cranial Sutures ................................................................................................................. 4
Sternal Rib......................................................................................................................... 5
Research Objectives.......................................................................................................... 6
Research Questions........................................................................................................... 7
Research Hypotheses ........................................................................................................ 8
LITERATURE REVIEW ............................................................................................................ 9
Calcification of the Costal Cartilage ............................................................................... 9
Costal Cartilage Aging ................................................................................................... 11
Sternal Rib End Aging ................................................................................................... 12
Sex Biases......................................................................................................................... 15
Ancestral Biases ..............................................................................................................18
Intercostal Variation.......................................................................................................21
First Rib Aging................................................................................................................23
Accuracy of the Method .................................................................................................24
Applicability of the Method ...........................................................................................26
Improvements of the Method.........................................................................................27
Modifications of the Method ..........................................................................................29
MATERIALS AND METHODS ...............................................................................................32
Materials ..........................................................................................................................32
The Hamann-Todd Subsample..............................................................................34
The William M. Bass Subsample...........................................................................36
Methods............................................................................................................................37
The İşcan et al. Method.........................................................................................38
The Hartnett-Fulginiti Method .............................................................................39
The Oettlé and Steyn Method................................................................................40
Intra-Observer Error ............................................................................................40
Statistical Analysis ..........................................................................................................41
Descriptive Statistics.............................................................................................42
Correlation............................................................................................................42
Sample Proportion ................................................................................................45
Cross-tabulation ...................................................................................................46
Inaccuracy and Bias .............................................................................................46
Intra-Observer Error ............................................................................................47

iv	
  

RESULTS ....................................................................................................................................48
Overall Accuracy ............................................................................................................48
Correlation............................................................................................................49
Cross-tabulation ...................................................................................................52
Accuracy for European-Americans ..............................................................................54
Correlation............................................................................................................55
Cross-tabulation ...................................................................................................57
Accuracy for African-Americans ..................................................................................60
Correlation............................................................................................................60
Cross-tabulation ...................................................................................................62
Accuracy Based on Ancestry .........................................................................................65
Correlation (İşcan et al. Method) .........................................................................65
Correlation (Hartnett-Fulginiti Method)..............................................................67
Correlation (Oettlé and Steyn Method) ................................................................70
Accuracy Based on Sex...................................................................................................72
Correlation (İşcan et al. Method) .........................................................................72
Correlation (Hartnett-Fulginiti Method)..............................................................74
Correlation (Oettlé and Steyn Method) ................................................................76
Correlation (All Methods) ....................................................................................78
Accuracy for All Categories...........................................................................................82
Inaccuracy .......................................................................................................................83
Bias ...................................................................................................................................86
Intra-observer Error ......................................................................................................90
Correlation............................................................................................................91
Cross-tabulation ...................................................................................................92
DISCUSSION ..............................................................................................................................95
Accuracy for European-Americans ..............................................................................95
Accuracy for African-Americans ..................................................................................96
Accuracy Based on Ancestry .........................................................................................98
Accuracy Based on Sex...................................................................................................98
Inaccuracy and Bias........................................................................................................99
Intra-observer Error .....................................................................................................100
Limitations......................................................................................................................100
CONCLUSION ..........................................................................................................................102
APPENDICES ............................................................................................................................107
Appendix A: İşcan et al. Male Phase Descriptions and Photographs .......................108
Appendix B: İşcan et al. Female Phase Descriptions and Photographs ...................113
Appendix C: Hartnett-Fulginiti Male Phase Descriptions and Photographs ..........118
Appendix D: Hartnett-Fulginiti Female Phase Descriptions and Photographs .......122
Appendix E: Oettlé and Steyn Male Phase Descriptions and Photographs .............126
Appendix F: Oettlé and Steyn Female Phase Descriptions and Photographs..........131
REFERENCES...........................................................................................................................136

v	
  

LIST OF TABLES
Table 1: Distribution of total sample ........................................................................................32
Table 2: Distribution of sample by ancestry and sex...............................................................33
Table 3: Distribution of the Hamann-Todd subsample ..........................................................36
Table 4: Distribution of the WM Bass subsample ...................................................................37
Table 5: Number of individuals for the İşcan et al. method by phase ...................................38
Table 6: Number of individuals for the Hartnett-Fulginiti method by phase.......................39
Table 7: Number of individuals of phases for the Oettlé and Steyn method by phase ........40
Table 8: Sample used for intra-observer error ........................................................................41
Table 9: Correlation for the total sample .................................................................................50
Table 10: Significance of the correlation coefficients for the total sample............................51
Table 11: Binomial distribution for the total sample ..............................................................52
Table 12: Cross-tabulation of the estimated and actual İşcan et al. phases for the total
sample ..........................................................................................................................................53
Table 13: Cross-tabulation of the estimated and actual Hartnett-Fulginiti phases for the
total sample..................................................................................................................................53
Table 14: Cross-tabulation of the estimated and actual Oettlé and Steyn phases for the total
sample ..........................................................................................................................................54
Table 15: Correlation for European-Americans......................................................................56
Table 16: Significance of the correlation coefficients for European-Americans ..................56
Table 17: Binomial distribution for European-Americans.....................................................57
Table 18: Cross-tabulation of estimated and actual İşcan et al. phases for EuropeanAmericans ....................................................................................................................................58
Table 19: Cross-tabulation of estimated and actual Hartnett-Fulginiti phases for EuropeanAmericans ....................................................................................................................................58

vi	
  

Table 20: Cross-tabulation of estimated and actual Oettlé and Steyn phases for EuropeanAmericans ....................................................................................................................................59
Table 21: Correlation for African-Americans .........................................................................61
Table 22: Significance of the correlation coefficients for African-Americans ......................61
Table 23: Binomial distribution for European-Americans.....................................................62
Table 24: Cross-tabulation of estimated and actual İşcan et al. phases for AfricanAmericans ....................................................................................................................................63
Table 25: Cross-tabulation of estimated and actual Hartnett-Fulginiti phases for AfricanAmericans ....................................................................................................................................63
Table 26: Cross-tabulation of estimated and actual Oettlé and Steyn phases for AfricanAmericans ....................................................................................................................................64
Table 27: Correlation based on ancestry for the İşcan et al. method ....................................66
Table 28: Significance of the correlation coefficients for the İşcan et al. method based on
ancestry (phase)...........................................................................................................................66
Table 29: Significance of the correlation coefficients for the İşcan et al. method based on
ancestry (age)...............................................................................................................................67
Table 30: Correlation based on ancestry for the Hartnett-Fulginiti method........................68
Table 31: Significance of the correlation coefficients for the Hartnett-Fulginiti method
based on ancestry (phase)...........................................................................................................69
Table 32: Significance of the correlation coefficients for the Hartnett-Fulginiti method
based on ancestry (age)...............................................................................................................69
Table 33: Correlation based on the ancestry for the Oettlé and Steyn method....................70
Table 34: Significance of the correlation coefficients for the Oettlé and Steyn method based
on ancestry (phase) .....................................................................................................................71
Table 35: Significance of the correlation coefficients for the Oettlé and Steyn method based
on ancestry (age) .........................................................................................................................71
Table 36: Correlation based on sex and ancestry for the İşcan et al. method ......................73
Table 37: Significance of the İşcan et al. method based on sex (phase) .................................73

vii	
  

Table 38: Significance of the İşcan et al. method based on sex (age) .....................................74
Table 39: Correlation based on sex and ancestry for the Hartnett-Fulginiti method ..........75
Table 40: Significance of the Hartnett-Fulginiti method based on sex (phase) ....................75
Table 41: Significance of the Hartnett-Fulginiti method based on sex (age) ........................76
Table 42: Correlation based on sex and ancestry for the Oettlé and Steyn method ............77
Table 43: Significance of the Oettlé and Steyn method based on sex (phase) .......................78
Table 44: Significance of the Oettlé and Steyn method based on sex (age) ...........................78
Table 45: Significance of the correlation coefficients for males .............................................79
Table 46: Significance of the correlation coefficients for females ..........................................80
Table 47: Significance of the correlation coefficients for African-American males.............80
Table 48: Significance of the correlation coefficients for European-American males .........81
Table 49: Significance of the correlation coefficients for African-American females..........81
Table 50: Significance of the correlation coefficients for European-American females ......82
Table 51: Correlation of estimated and actual ages for the entire sample ............................83
Table 52: Phase inaccuracy for all methods .............................................................................84
Table 53: Age inaccuracy for all methods ................................................................................85
Table 54: Phase bias for all methods.........................................................................................87
Table 55: Age bias for all methods ............................................................................................89
Table 56: Correlation of estimate 1 and estimate 2 for the entire sample.............................91
Table 57: Cohen’s Kappa between estimate 1 and estimate 2 for the entire sample ...........91
Table 58: Cross-tabulation of estimate 1 and estimate 2 for the İşcan et al. method for the
entire sample ...............................................................................................................................92
Table 59: Cross-tabulation of estimate 1 and estimate 2 for the Hartnett-Fulginiti method
for the entire sample ...................................................................................................................93

viii	
  

Table 60: Cross-tabulation of estimate 1 and estimate 2 for the Oettlé and Steyn method for
the entire sample .........................................................................................................................94

ix	
  

LIST OF FIGURES
Figure 1: Distribution of age by ancestry for the total sample ...............................................34
Figure 2: Phase inaccuracy for all methods .............................................................................84
Figure 3: Age inaccuracy for all methods.................................................................................86
Figure 4: Phase bias for all methods .........................................................................................88
Figure 5: Age bias for all methods ............................................................................................90
Figure 6: İşcan et al. (1984a) Male Phase 0 (16 years and younger) .....................................108
Figure 7: İşcan et al. (1984a) Male Phase 1 (17–19 years) .....................................................108
Figure 8: İşcan et al. (1984a) Male Phase 2 (20–23 years) .....................................................109
Figure 9: İşcan et al. (1984a) Males Phase 3 (24–28 years)....................................................109
Figure 10: İşcan et al. (1984a) Male Phase 4 (26–32 years) ...................................................110
Figure 11: İşcan et al. (1984a) Male Phase 5 (33–42 years) ...................................................110
Figure 12: İşcan et al. (1984a) Male Phase 6 (43–55 years) ...................................................111
Figure 13: İşcan et al. (1984a) Male Phase 7 (54–64 years) ...................................................111
Figure 14: İşcan et al. (1984a) Male Phase 8 (65 years and older) ........................................112
Figure 15: İşcan et al. (1985) Female Phase 0 (13 years and younger) .................................113
Figure 16: İşcan et al. (1985) Female Phase 1 (14–15 years)..................................................113
Figure 17: İşcan et al. (1985) Female Phase 2 (16–19 years)..................................................114
Figure 18: İşcan et al. (1985) Female Phase 3 (20–24 years)..................................................114
Figure 19: İşcan et al. (1985) Female Phase 4 (24–32 years)..................................................115
Figure 20: İşcan et al. (1985) Female Phase 5 (33–46 years)..................................................115
Figure 21: İşcan et al. (1985) Female Phase 6 (43–58 years)..................................................116
Figure 22: İşcan et al. (1985) Female Phase 7 (59–71 years)..................................................116

x	
  

Figure 23: İşcan et al. (1985) Female Phase 8 (70 years and older) ......................................117
Figure 24: Hartnett-Fulginiti (2007; 2010) Male Phase 1 (18–22 years) ..............................118
Figure 25: Hartnett-Fulginiti (2007; 2010) Male Phase 2 (21–28 years) ..............................118
Figure 26: Hartnett-Fulginiti (2007; 2010) Male Phase 3 (27–37 years) ..............................119
Figure 27: Hartnett-Fulginiti (2007; 2010) Male Phase 4 (36–48 years) ..............................119
Figure 28: Hartnett-Fulginiti (2007; 2010) Male Phase 5 (45–59 years) ..............................120
Figure 29: Hartnett-Fulginiti (2007; 2010) Male Phase 6 (57–70 years) ..............................120
Figure 30: Hartnett-Fulginiti (2007; 2010) Male Phase 7 (70–97 years) ..............................121
Figure 31: Hartnett-Fulginiti (2007; 2010) Male Variant Phase (27–69 years) ...................121
Figure 32: Hartnett-Fulginiti (2007; 2010) Female Phase 1 (18–22 years)...........................122
Figure 33: Hartnett-Fulginiti (2007; 2010) Female Phase 2 (24–27 years)...........................122
Figure 34: Hartnett-Fulginiti (2007; 2010) Female Phase 3 (27–38 years)...........................123
Figure 35: Hartnett-Fulginiti (2007; 2010) Female Phase 4 (39–49 years)...........................123
Figure 36: Hartnett-Fulginiti (2007; 2010) Female Phase 5 (47–58 years)...........................124
Figure 37: Hartnett-Fulginiti (2007; 2010) Female Phase 6 (60–73 years)...........................124
Figure 38: Hartnett-Fulginiti (2007; 2010) Female Phase 7 (65–99 years)...........................125
Figure 39: Oettlé and Steyn (2000) Male Phase 0 (Prephase seen in children)....................126
Figure 40: Oettlé and Steyn (2000) Male Phase 1 (18–21 years) ...........................................126
Figure 41: Oettlé and Steyn (2000) Male Phase 2 (22–24 years) ...........................................127
Figure 42: Oettlé and Steyn (2000) Male Phase 3 (25–29 years) ...........................................127
Figure 43: Oettlé and Steyn (2000) Male Phase 4 (30–37 years) ...........................................128
Figure 44: Oettlé and Steyn (2000) Male Phase 5 (38–46 years) ...........................................128
Figure 45: Oettlé and Steyn (2000) Male Phase 6 (47–55 years) ...........................................129

xi	
  

Figure 46: Oettlé and Steyn (2000) Male Phase 7 (52–69 years) ...........................................129
Figure 47: Oettlé and Steyn (2000) Male Phase 8 (63–77 years) ...........................................130
Figure 48: Oettlé and Steyn (2000) Female Phase 0 (Prephase seen in children) ................131
Figure 49: Oettlé and Steyn (2000) Female Phase 1 (12–17 years) .......................................131
Figure 50: Oettlé and Steyn (2000) Female Phase 2 (16–26 years) .......................................132
Figure 51: Oettlé and Steyn (2000) Female Phase 3 (23–27 years) .......................................132
Figure 52: Oettlé and Steyn (2000) Female Phase 4 (28–32 years) .......................................133
Figure 53: Oettlé and Steyn (2000) Female Phase 5 (33–43 years) .......................................133
Figure 54: Oettlé and Steyn (2000) Female Phase 6(a) (44–59 years)...................................134
Figure 55: Oettlé and Steyn (2000) Female Phase 6(b) (44–59 years)...................................134
Figure 56: Oettlé and Steyn (2000) Female Phase 7 (56–79 years) .......................................135

xii	
  

INTRODUCTION
The estimation of age in the field of forensic anthropology is essential for the
identification of human skeletal remains in medico-legal death investigations. This estimation is
conducted through the use of a biological profile, consisting of the estimation of age, sex,
ancestry, and living stature in order to reconstruct the identity of descendants (Byers, 2005).
Currently, there are multiple techniques available to evaluate age-at-death, all of which vary
significantly between adults and subadults. For instance, methods for subadult aging are based
on growth and development, such as dental eruption and epiphyseal union. Comparatively, due
to the maturation of the skeleton, adult aging methods are often more difficult, as they must
instead evaluate more subtle changes, often the degeneration of elements (İşcan and Loth, 1989).
Although the use of multiple techniques is ideal for aging an individual, problems of
preservation or postmortem loss may inhibit this; thus the accuracy of these methods needs to be
evaluated. Presently, there are four major elements primarily used for the morphoscopic
determination of age in adults: the pubis, the auricular surface of the ilium, cranial sutures, and
the fourth right sternal rib end. These four elements, as well as the aging techniques associated
with them, will be discussed below in further detail.

Pubis
The pubic symphysis is generally thought to be the most accurate of these techniques due
to the method’s reliability and easily visible age-related changes (Meindl and Lovejoy, 1989).
The first method for aging the pubic symphyses was developed by Todd (1920), who created a
ten-stage phase system for European-American males based on changes in morphology that
occur with age. Additional methods for European-American females and African-American

1	
  

males and females were developed the following year due to differences seen in the rates of
aging (Todd, 1921). Generally it was found that the pubic symphyseal face contains ridges in
young age, and as age increases, the ridges diminish with an overall degradation in bone quality
(Todd, 1920).
Brooks (1955) later evaluated Todd’s (1920; 1921) methods using Todd’s original
sample as well as an archaeological sample. The author found a tendency to over-age
individuals, and therefore proposed modified age ranges in order to correct for these inaccuracies
(Brooks, 1955). McKern and Stewart (1957) also tested Todd’s (1920) methods and found that it
did not take into account the range of variation that was observed. Therefore the authors
developed a component analysis system based on a sample of males who had perished during the
Korean War. Three features were used, consisting of the dorsal plateau, the ventral rampart, and
the symphyseal rim. To apply the method, each feature is assigned a numerical score of 0
through 5 in order to obtain a composite score, corresponding to an overall age range (McKern
and Stewart, 1957).
Because the McKern and Stewart (1957) method was developed only on males, the
technique was also tested on females, in which it was found to significantly underestimate age
for females over 40 years (Gilbert, 1973). Due to these variations, Gilbert and McKern (1973)
developed an additional component analysis system for aging females using the same criteria
defined by McKern and Stewart (1957). The authors found morphological developmental
differences for females in that they generally appear older and may suffer from childbirth
trauma, further complicating analysis (Gilbert and McKern, 1973).
Suchey (1979) later tested the Gilbert and McKern (1973) method on eleven females and
found it to be generally unreliable, as a large amount of variation was found between the

2	
  

observer responses. Additionally, Katz and Suchey (1986) and Suchey et al. (1986) evaluated
both the Todd (1920) and McKern and Stewart (1957) methods using a modern, autopsy sample.
Results indicated that the Todd (1920) method overestimated age while the McKern and Stewart
(1957) method was found to be less accurate for older individuals (Katz and Suchey, 1986;
Suchey et al., 1986; Klepinger, 2006).
Based on these findings, Brooks and Suchey (1990) developed a revision of the Todd
(1920; 1921) methods, which was based on a modern sample consisting of multiple ancestries
and both sexes. This technique, commonly known as the Suchey-Brooks method, consists of six
phases for each sex and is thought to be the most reliable of all aging methods (Brooks and
Suchey, 1990; Klepinger, 2006).

Auricular Surface
The auricular surface of the ilium is less prone to postmortem damage than is the pubic
symphysis and therefore may be utilized more frequently (Klepinger, 2006). And although the
degenerative changes are more complex, several occur beyond the age of 50, unlike the pubic
symphyses (Meindl and Lovejoy, 1989). The first method for aging from the auricular surface
was developed by Lovejoy et al. (1985) and was based on correlations that were seen between
age and degeneration in the features of the auricular surface, the apex, and the retroauricular
surface of the ilium. The authors developed an eight-phase system, establishing standards that
could be applied to both sexes (Lovejoy et al., 1985).
Buckberry and Chamberlain (2002) later revised the Lovejoy et al. (1985) method. These
revisions made the method easier to apply and more accurate and reliable with increased age
ranges and lowered rates of inter- and intra-observer error. The revised technique is based on a

3	
  

quantitative scoring system, in which multiple features are examined individually in order to
obtain a composite score, corresponding to an overall age range. Tests of this revised method
have shown that the method is equally applicable to males and females of both European and
African ancestry (Mulhern and Jones, 2005). The revised method is also more accurate for
younger individuals and easier to apply than the original method (Mulhern and Jones, 2005).
Correlation values were also found to be higher than that of the Suchey-Brooks (1990) method,
possibly indicating that the auricular surface may be more applicable, as its survival rate is also
higher than that of other elements used for aging (Buckberry and Chamberlain, 2002).

Cranial Sutures
Although somewhat controversial, the cranium has also been used to estimate age by
examining rates of ectocranial, endocranial, and maxillary suture closure (McKern and Stewart,
1957; Meindl and Lovejoy, 1985). The first modern study was conducted by Todd and Lyon
(1924; 1925a;b;c), which examined rates of endocranial and ectocranial suture closure. In order
to develop the method, crania were separated by sex and ancestry, and although differences were
found, the authors believed that developing separate standards was not necessary (Todd and
Lyon, 1925b).
Brooks (1955) later evaluated the technique, finding it to be generally unreliable, and
therefore suggested that it should only be used to confirm other aging techniques. Due to these
findings, Meindl and Lovejoy (1985) also conducted a test of the Todd and Lyon (1924;
1925a;b;c) method and developed a revision using a numerical scale of 0 (open) through 3
(obliterated) to evaluate suture closure at ten specific sites on the ectocranium (Meindl and
Lovejoy, 1985). According to Garvin and Passalacqua (2012), this method is now the most

4	
  

commonly utilized for suture closure, although it still should be used with caution or in
conjunction with other methods due to ancestral- and sex-based variations in the rate of closure
which are not accounted for in the method (Meindl and Lovejoy, 1985).
Additionally, maxillary suture closure has also been studied. A method developed by
Mann et al. (1987; 1991) evaluated rates of this closure the based on the obliteration of four
palatal sutures. Although tests of this method have shown that maxillary sutures are a useful age
estimator for older individuals (Ginter, 2005), the authors advise that it is only applicable for
placing individuals into general age categories, separating commingled remains, or when the
maxilla is the only element present (Mann et al., 1987; 1991; Gruspier and Mullen, 1991).
Finally, an additional segmented approach for aging from the closure of maxillary sutures has
also been developed by Beauthier et al. (2010), which also suffers from the same issues as the
Mann et al. (1987; 1991) method, in that the age of young individuals was often overestimated
and the age of older individuals was often underestimated.

Sternal Rib
The sternal end of the right fourth rib, which was evaluated in this research and is
discussed later in detail, was developed by İşcan, Loth, and Wright (1984a;b; 1985), and is
commonly known as the İşcan and Loth or the İşcan et al. method. In the development of this
technique, the authors examined the morphological features of the sternal end of right fourth rib
and determined that there were consistent, degenerative changes. Therefore, they were able to
develop a phase system to document these changes, as well as link this degeneration with age
based on changes “in the form, shape, texture, and overall quality of the sternal rib” (İşcan et al.,
1984a:1096).

5	
  

Research Objectives
The Scientific Working Group for Forensic Anthropology (SWGANTH) defines best
practices for the field, including standards for age estimation. As detailed by SWGANTH (2010),
the estimation of age should begin by first assessing the physiological age of the remains in order
to associate it with a chronological, or actual, age. This is done through the examination of
degenerative or developmental traits, which represent the biological or physiological age, as
reflected in the skeleton. For adults, degenerative traits are used for this process, using methods
that have been developed and tested on modern individuals (SWGANTH, 2010).
SWGANTH (2010) also states that when estimating the age of unknown individuals,
standards that have been developed on individuals of similar sex and ancestry to the decedent
should be utilized. Therefore, the objective of this research was to evaluate three sternal rib aging
techniques, the İşcan et al. (1984a; 1985), Hartnett-Fulginiti (2007; 2010), and Oettlé and Steyn
(2000) methods, for overall accuracy rates. Additionally, although the İşcan et al. (1984a; 1985)
method has been previously tested, the Hartnett-Fulginiti (2007; 2010) and Oettlé and Steyn
(2000) methods have not been previously evaluated and were therefore an important part of this
research.
A second purpose of this study was to evaluate any possible differences in accuracy
between European-Americans and African-Americans. Ancestral and population based
differences have previously been demonstrated in the aging process. For example, İşcan et al.
(1987) found differences in the rate and pattern of morphological age-related changes in AfricanAmericans compared to European-Americans. Oettlé and Steyn (2000) found similar features in
the degenerative changes in South Africans of African ancestry. Therefore it was important to
examine if any differences in accuracy were evident between African-Americans and European-

6	
  

Americans, for if significant differences are found, it may require the development of separate
rib aging standards based on ancestry.
In addition to assessing the accuracy of African-Americans, all three methods were tested
on European-Americans. Although the İşcan et al. (1984a; 1985) European-American methods
are currently the most accepted technique for sternal rib aging, the Hartnett-Fulginiti (2007;
2010) method was developed as a revision of the technique. Both methods also utilized a mostly
European-American population, therefore it was important to examine which method would be a
better assessor of age for these individuals.

Research Questions
Using a sample of human remains from multiple collections, this research addressed the
following questions:
1. Will the Hartnett-Fulginiti (2007; 2010) method more accurately assess age-at-death than
the İşcan et al. (1984a; 1985) method and the Oettlé and Steyn (2000) method?
2. Will the Hartnett-Fulginiti (2007; 2010) method more accurately assess the age-at-death
of European-Americans than the İşcan et al. (1984a; 1985) method and the Oettlé and
Steyn (2000) method?
3. Will the Oettlé and Steyn (2000) method more accurately assess the age-at-death of
African-Americans than the İşcan et al. (1984a; 1985) method and the Hartnett-Fulginiti
(2007; 2010) method?
4. Will there be differences in accuracy based on ancestry for each method?
5. Will there be differences in accuracy based on sex for each method?
6. What are the rates of intra-observer error?

7	
  

Research Hypotheses
This research had multiple hypotheses.
1. The Hartnett-Fulginiti (2007; 2010) method will more accurately assess age of all
individuals within the sample than the İşcan et al. (1984a; 1985) method and the Oettlé
and Steyn (2000) method.
2. The Hartnett-Fulginiti (2007; 2010) method will more accurately assess age of EuropeanAmericans than the İşcan et al. (1984a; 1985) method and the Oettlé and Steyn (2000)
method.
3. The Oettlé and Steyn (2000) method will more accurately assess age of AfricanAmericans than the Hartnett-Fulginiti (2007; 2010) method and the İşcan et al. (1984a;
1985) method.
4. There will be significant differences in accuracy between European-Americans and
African-Americans.
5. There will be significant differences in accuracy between males and females.
6. The intra-observer error rates will be insignificant.
	
  
	
  

8	
  

LITERATURE REVIEW
The impetus for the research conducted by İşcan, Loth, and Wright was Kerley’s (1970)
observation that there are morphological age-related changes in the sternal end of the rib, as well
as from degenerative changes seen in other “radiographic, histological, and osteological studies”
(İşcan et al., 1984b:147). Additionally, the thorax was favored as a location for age estimation, as
there is limited stress applied to the region. This is in contrast to other aging methods, such as the
pubic symphysis, which can suffer from weight bearing, locomotive, and childbirth stress (İşcan
and Loth, 1986; İşcan, 1991; Hartnett, 2007). Therefore the sternal end of the rib was
demonstrated to be an appropriate region to study age-related deterioration, as it is easily
accessible, has consistent degenerative changes, and has minimal stress applied to the area.

Calcification of the Costal Cartilage
Although researchers have been unable to precisely determine the specific means and
causes for the calcification of the costal cartilage, it is generally accepted that this process
develops “in relation to the vascular canals … and increase[s] with age” (Sokoloff, 1983:110). It
is the failure of this mechanism that alters the rate of calcification of the costal cartilage, affected
by both mechanical and systemic factors (Horner, 1949; Garvin, 2010).
Mechanical factors affecting calcification of the costal cartilage include both
biomechanical and physiological stresses (e.g. King, 1939; Sedlin et al., 1963; Semine and
Damon, 1975), with differences in the rate and pattern of calcification based on sex and ancestry
(e.g. Michelson, 1934; Elkeles, 1966; Saunders, 1966; Semine and Damon, 1975; McCormick,
1980; McCormick and Stewart, 1983; İşcan and Loth, 1986a;b; İşcan et al., 1987; Rao and Pai,
1988). For instance, King (1939) found that the calcification of the costal cartilage is a

9	
  

continuous process that occurs with age and is a consequence of the strain from intercostal
muscle attachments. This can especially be seen in males, in which calcification begins at the
superior and inferior margins, specifically where the intercostal muscles attach (Garvin, 2010).
Activity patterns may also affect the metamorphosis of the sternal rib ends (e.g. Rösing et
al., 2007), as well as stress from normal respiration (e.g. Semine and Damon, 1975; İşcan and
Loth, 1986c; McCormick and Stewart, 1988; Barchilon et al., 1996; Yoder et al., 2001).
Specifically, it was determined that the upper ribs suffer from more stress due to their attachment
to the sternum, therefore experiencing more torsional stress (Semine and Damon, 1975; Yoder et
al., 2001). In addition, the first costal cartilage was found to have a different and more variable
rate and pattern of calcification than that seen in the lower ribs (McCormick and Stewart, 1998).
Barchilon et al. (1996) found similar results from the examination of the calcification patterns of
the first rib, specifically that extreme physiological stress was one factor in early calcification of
the first rib cartilage.
Additional systemic factors, such as abnormalities, pathologies, nutrition, and genetics,
also affect age-related degenerative changes of the rib (e.g. Riebel, 1929; King, 1939; Vastine et
al., 1948; Horner, 1949; Eichelberger and Roma, 1954; Sokoloff, 1983; Senac et al., 1985;
Garvin, 2010; Webb et al., 2010; Ea et al., 2011). Webb et al. (2010) examined abnormal spinal
curvature and found that it affected sternal rib aging by distorting the ribs, making the sternal
ends appear more youthful or flattened. Therefore Webb et al. (2010) suggested that when aging
individuals with scoliosis, shape criteria should not be a major factor. Instead, porosity and
pitting should be examined in more detail (Webb et al., 2010).
In an additional study, Jordanoglou (1969) examined rib movement and found that
individuals suffering from ankylosing spondylitis (the inflammation of joints in the spine, which

10	
  

leads to fusion) and kyphoscoliosis (abnormal spinal curvature) had rib movement that was
reduced and limited to the rib-neck axis, possibly altering changes in the degeneration of the
costal cartilage. Senac et al. (1985) also found an increased rate of costochondral calcification in
adolescents with hyperthyroidism, linking the disease with increased bone maturation. Still,
additional factors should be taken into account for individuals with hyperthyroidism, as they may
also affect the rate of calcification, such as length of time the hormone level is elevated and
nutritional stress, especially in regards to calcium (Semine and Damon, 1975; Senac et al., 1985).
The results of these studies demonstrated that although both mechanical and systemic
factors affect both the rate and pattern of the degeneration of the sternal ends of the ribs, the area
does not suffer from a large amount of stress, suggesting that the region may be an appropriate
location for age estimation.

Costal Cartilage Aging
As a result of the studies mentioned above, which established that the calcification of the
costal cartilage was a consequence of age, many researchers attempted to develop age estimation
methods based on specific patterns seen in this degeneration process (e.g. Semine and Damon,
1975; McCormick, 1980; McCormick and Stewart, 1988; Barrès et al., 1989). For example, a
study conducted by Semine and Damon (1975) examined the rate of costochondral calcification
in five populations. The authors numerically scored the rate of calcification from 0 (none) to 4
(severe) from chest radiographs on the first rib and the lower ribs separately. The results
indicated that calcification of the rib cartilage was correlated with age, and if further
investigated, could prove to be an accurate age estimation technique. Additional sex and

11	
  

population differences were also noticed, demonstrating that standards should be separated based
on these factors (Semine and Damon, 1975).
In another study, McCormick (1980) examined radiographs of 20 chest plates from an
autopsy sample and confirmed that calcification of the costal cartilages was age-related.
McCormick and Stewart (1983) later expanded on this research by examining 651 radiographs of
chest plates from autopsy cases of known ancestry, sex, and age. Typical male and female
patterns of age related changes were noted, with the calcification of costal cartilages increasing
with age (McCormick and Stewart, 1983). Finally, McCormick and Stewart (1988) developed a
method based on matching known patterns of calcification to those found in each individual.
Additionally, a linear regression equation was developed, which was found to be reasonably
predictive of age for younger individuals, but was less reliable for older ages (McCormick and
Stewart, 1988).
Barrès et al. (1989) also developed a multiple linear regression equation based on the
correlations seen between the calcification of the costal cartilage and age. A qualitative scoring
system was used due to the multiple features seen in radiographs of chest plates. The method was
later tested by Garvin (2010) on a modern sample and was found not to be reliable, although the
author was able to revise the technique by applying a binomial scoring system. From this revised
system, Garvin (2010) was able to obtain more accurate age estimations.

Sternal Rib End Aging
Previous studies concluded that the costal cartilage undergoes degenerative changes
affecting the morphology of the sternal end of the rib. From these studies it was known that the
rib end general starts out relatively blunt in young individuals and becomes sharper and more

12	
  

cup-shaped with age (Kerley, 1970; Ubelaker, 1978; McCormick and Stewart, 1988).
Additionally, the positioning and function of the ribs were found not to be significantly affected
by function or stress, with minimal respiration strain applied to the region, making the sternal
end of the rib an appropriate location for age estimation (Semine and Damon, 1975; İşcan and
Loth, 1986; Loth and İşcan, 1989; İşcan, 1991; Hartnett, 2007). Due to previous studies, such as
these, the main objective of this area of research shifted from developing methods based on the
calcification of the cartilage to the remodeling of the sternal end of the rib.
A method for age estimation based on the sternal end of the right fourth rib was first
developed as a preliminary component analysis study (Loth et al., 1983), which was later tested
and published by İşcan, Loth, and Wright (1984b). A total of 93 modern European-American
males were used to develop this component analysis method, resembling that used by McKern
and Stewart (1957). The technique was based on features that were noticed in the sternal ends of
the ribs, consisting of the pit depth, pit shape, and rim and wall configurations (İşcan et al.,
1984b). In order to apply the component analysis method, the pit depth was measured, while
scoring of the pit shape consisted of morphological changes in the pit formation, such as taking
on a V- and then a U-shape. Finally, the rim was scored based on edge morphology and texture.
These three components were individually scored and placed into one of six stages, with the
composite score of all three components corresponding to an overall age range (İşcan et al.,
1984b).
This method was only found to be generally reliable for younger individuals as it became
less accurate with age (İşcan et al., 1984b). Therefore, the technique was later replaced by a more
accurate and simpler phase method, modeled after the Todd (1920) pubic symphysis method,
which reduced the more complicated four steps into one step (Wright et al., 1984). This change

13	
  

may have improved accuracy, as it was found that the component methods often perform poorly
due to “their failure to allow full integration of all age-related variables” (Meindl et al.,
1990:349).
The İşcan, Loth, and Wright (1984a) study was conducted on a modern, autopsy sample
of 118 European-American males, ranging in age from 17 to 85 years of known age, sex, and
ancestry (Appendix A). Morphological age related changes were not seen in males younger than
the age of 16, indicating that rib growth ends at this age; therefore individuals younger than this
were excluded from the study. It was generally found that in males after the development of the
rib is complete, the sternal end of the right fourth rib appears nearly flat with billowing (Phase
0). The degeneration of the rib begins with a pit on the medial articular surface (Phase 1), later
taking on a V-shape as the pit deepens (Phases 2 and 3), which gradually widens into a U-shape
(Phases 3 through 8). As age increases, the pit widens and deepens. Additionally, the rim boarder
becomes irregular and the overall texture of the articular surface becomes brittle and more
porous (İşcan et al., 1984a).
Tests of the method were conducted at the 1984 American Academy of Forensic Sciences
meeting using 15 European-American male ribs from a medical examiner’s office, with judges
consisting of students and professionals, as well as the authors (İşcan and Loth, 1986b). No
descriptions of the phases were given; therefore phase classification was only determined based
on the provided photographs. Results indicated that the changes in morphology were more
clearly seen below Phase 4, as older phases become more difficult to age. Additionally, interobserver error due to varying degree of experience was found to be minimal. Therefore the
authors concluded that age estimation from the right fourth sternal rib end could be estimated to

14	
  

fit within the correct phase, and that the accuracy based on experience was not a large issue
(İşcan and Loth, 1986b).

Sex Biases
Earlier studies showed sexual dimorphic differences in the calcification of costal cartilage
(e.g. Fischer, 1955; Elkeles, 1966; Saunders, 1966; Navanti et al., 1970; Semine and Damon,
1975; Stewart and McCormick, 1983; Stewart and McCormick, 1984; McCormick et al., 1985;
Rao and Pai, 1988; Torwalt and Hoppa, 2005; Rejtarová et al., 2009a; 2009b), as well as
specifically in the sternal end of the rib (e.g. Cöloğlu et al., 1998; Wiredu et al., 1999; Koçak,
2003; Gavit, 2009). Fischer (1955) noticed specific sex-based patterns to the calcification of the
costal cartilage, and Horner (1949) observed that menstrual disorders affect the rate and pattern
of this calcification. Navani et al. (1970) also noted these patterns in a sample of 1000 chest
radiographs and therefore proposed that these typologies could be used to predict sex by placing
individuals into one of three categories, representing a specific pattern of calcification. From
these typologies, Navani et al. (1970) determined that 95 percent of males and 93 percent of
females could be correctly sexed.
Saunders (1966) and McCormick and Stewart (1983) also found similar male and female
patterns of calcification of the costal cartilage. The authors found that males displayed linear or
marginal calcification, which begins at the superior and inferior borders of the ribs. Females
instead displayed calcification in the central area of the rib, often containing ossific nodules or
areas of irregular calcification (Saunders, 1966; McCormick and Stewart, 1983). Additionally,
Saunders (1966) found that when females displayed male patterns of calcification, most had
undergone pelvic surgery, often hysterectomies. Therefore it was concluded that the rate and

15	
  

pattern of calcification of the costal cartilage is sex dependent and at least partially based on
hormonal and genetic influences (Elkeles, 1966; Saunders, 1966; Semine and Damon, 1975;
McCormick and Stewart, 1983; Rao and Pai, 1988).
Due to variations seen between males and females in the rate and pattern of degenerative
changes, separate standards were developed for European-American females by İşcan, Loth, and
Wright in 1985 in order to more accurately reflect the female aging process. The sample
consisted of 86 European-American females ranging in age from 14 to 90 years that were
collected from modern autopsies. Morphological age-related changes were not seen in samples
of females younger than 14 years old, indicating that rib growth ends at this age; therefore
individuals younger than this were excluded from the study. It was generally found that female
development was slightly ahead of males, with the gap widening to ten years in later phases.
Differences between the sexes were mostly found in the shape and depth of the pit on the medial
articular surface and differences in the changes from the V- to U-shape of the pit (İşcan et al.,
1985; İşcan, 1991).
The findings of this study resulted in the creation of a separate set of standards for
European-American females (Appendix B). For females it was found that after development of
the rib is complete, the sternal end of the right fourth rib appears nearly flat with billowing
(Phase 0). The degeneration of the rib begins with a pit on the medial articular surface (Phase 1)
that takes on a V-shape as the pit deepens (Phases 2 though 6), which gradually widen into a Ushape (Phases 3 through 8). As age increases, the pit does as well, becoming both wider and
deeper. A scalloping pattern develops on the rim edges (Phases 2 through 4) and the rim border
turns irregular (Phases 5 through 8), with the overall texture of the articular surface becoming
brittle and porous with age (Phases 4 through 8). Additionally, females were found to have ribs

16	
  

that were thinner and less dense (İşcan and Loth, 1986c). Females also displayed a central arc of
bone reaching from the anterior to the posterior walls and bony extensions within the pit (İşcan et
al., 1985).
Tests of the method were conducted on a sample of ten European-American females from
a medical examiner’s office at the 1985 American Academy of Forensic Sciences meeting, with
judges consisting of both students and professionals. Photographic standards for each phase were
given to the observers, with the addition of two rib phase examples, in order to aid in the
estimation of age. The tests indicated the need for written descriptions when applying the
method, as sometimes one phase may resemble another if small nuances are not noticed (İşcan
and Loth, 1986a).
Differences between European-American males and females were further tested by İşcan
(1985). A modern, autopsy sample consisting of 144 European-American male and 86 EuropeanAmerican female fourth right sternal rib ends were used. Measurements were taken for the
maximum superior-inferior height, the maximum anterior-posterior breadth, and the maximum
pit depth. Significant differences were found between the sexes, indicating that sexual
dimorphism can be reliably assessed metrically; therefore a discriminant function formula was
developed in order to predict sex of the unknown rib (İşcan, 1985). Allen (1997) also tested
sexual dimorphism on the ribs on a modern African-American sample, consisting of 110 males
and 52 females. Discriminant function formulae were developed on ribs one through seven in
order to determine sex, with a maximum accuracy of 83 percent for rib four, although these
formulae are both population and rib specific (Allen, 1997).

17	
  

Ancestral Biases
Although the estimation of ancestry is an important part of the biological profile, it is
often the most controversial. Recently, the term “ancestry” has mostly replaced “race” within the
field of forensic anthropology due to negative connotations of word. This has prompted some
researchers to believe that the concept of race should be discarded completely in the field of
anthropology, as it is purely a social construct (Armelagos and Goodman, 1998; Smay and
Armelagos, 2000). Others, however, believe that although the concept is not valid, it may be
necessary in forensic contexts (Sauer, 1992; Kennedy, 1995).
Race is often defined on the basis of phenotypic, or observable, variations, which are a
result of adaptive responses to the environment. These variations can be seen in visible
morphological features and in the skeleton, especially in the facial area. It is from these
observable traits, such as skin color and facial features, that social categories of race have been
developed. Therefore, anthropologists must often translate their estimation of ancestry into a
racial category for identification purposes. This can be problematic, as racial and social identities
may differ from biology. Still, the estimation of ancestry within the field is useful, especially
when dealing with the identification of unknown decedents, as it allows for the estimation of an
individual’s ancestral origins.
Previous studies determined that ancestral variation in the rate of aging was present in the
costal cartilage and sternal end of the ribs (e.g. Michelson, 1934; Semine and Damon, 1975;
İşcan et al., 1987; Loth, 1990; Russell et al., 1993; Yoder et al., 2001; Love, 2005). Michelson
(1934) examined ancestral differences in the rate of first cartilage calcification found that it was
more rapid in African-Americans than in European-Americans. Semine and Damon (1975) also
noted population-based differences in the extent of the calcification of the costal cartilage.

18	
  

Due to these known variations, the applicability of the sternal rib aging on AfricanAmericans was evaluated as well (İşcan et al., 1987; Loth and İşcan, 1988). The İşcan et al.
(1984a; 1985) methods for European-Americans were tested on a sample of 73 modern AfricanAmericans, consisting of 49 males and 14 females of known age and sex. Results indicated
differences in both the rate and the pattern of morphological changes indicative of aging between
European-Americans and African-Americans, which were attributed to genetic influences (İşcan
et al., 1987). Specifically, the authors found an earlier maturation rate for African-Americans.
Degenerative changes occurred at a similar rate for both European-Americans and AfricanAmericans until the late twenties, with nearly identical mean ages per phase. After the age of
thirty, African-Americans were found to look older than European-Americans and were
consistently over-aged by three to ten years due to the presence of “pointy rims and pronounced
superior or inferior projections” (İşcan et al., 1987:455). Additionally, scalloping that was found
in Phase 2 for European-Americans was found to be inconsistent and less pronounced in AfricanAmericans. The anterior and posterior articular rim, which was usually rounded or flattened in
European-Americans, was pointed, squared off, or flattened in African-Americans. These same
projections appeared earlier and with greater frequency in African-Americans compared to
European-Americans. These findings indicated that ancestral differences might affect the
estimation of age. Therefore the authors proposed modifications of the phases to be applied to
African-American males and females (İşcan et al., 1987; Loth and İşcan, 1989; İşcan, 1991),
although according to Garvin and Passalacqua (2012), these standards are often not utilized.
Loth (1990) also examined differences between African-Americans and EuropeanAmericans using the İşcan et al. (1984a; 1985) European-American methods on third, fourth, and
fifth ribs of 136 African-American individuals from the Terry Collection. Results indicated that

19	
  

the patterning of morphological changes seen in the African-Americans were similar to those
seen in the İşcan et al. (1987) modern sample, with ribs of African-American individuals
remaining more dense and solid with less deterioration than those of European-American
individuals. Additionally, measurements of the ribs showed that African-Americans had smaller
anterior-posterior and superior-inferior dimensions, indicating that there are morphological
variations in the shape and degeneration of the sternal rib end between the two ancestral groups
(Loth, 1990).
Russell et al. (1993) also examined both European-American and African-American
males from the Hamann-Todd Osteological Collection. The İşcan et al. (1984a) method for
European-American males was tested on two samples; the first consisted of 23 EuropeanAmerican males and the second consisted of 80 males: 41 European-Americans and 39 AfricanAmericans with similar age distributions. A general non-significant tendency to under-age
African-American males was found, which is contradictory to what İşcan et al. (1987) predicted.
No statistically significant differences were found between the ancestries; therefore the authors
concluded that the European-American male standards could be used to estimate age for both
European-American and African-American males. Additionally, it was determined that although
the method was a useful indicator, it should only be used in conjunction with other aging
methods (Russell et al., 1993).
Finally, Love (2005) tested the İşcan et al. (1984a; 1985) European-American methods
on a modern autopsy sample consisting 50 African-Americans and European-Americans of both
sexes. The results found that 37 percent were assigned to the correct phase, with AfricanAmerican males significantly over-aged. Due to the small sample size used for this research,
Love (2005) also suggested that further research be conducted as to possible ancestral based

20	
  

differences in the rate and pattern of aging between European-Americans and African-Americans
at the sternal end of the rib.

Intercostal Variation
In addition to the mentioned mechanical and systemic factors that may affect the
reliability and repeatability of the method, another difficulty of the rib aging methods is that they
were developed specifically on the fourth right sternal rib end. Although this rib was selected for
its ease of acquisition during an autopsy (İşcan et al., 1984a; 1985), obtaining this specific rib
may be difficult in both forensic and archaeological contexts due to poor preservation or
postmortem loss. For these reasons, many researchers have addressed issues of intercostal
variation by developing techniques that help to determine the side and rib number in order to
better identify the fourth right rib. These techniques are based variation in the morphology of ribs
had already been established, especially for the first rib and between the lower and upper ribs
(e.g. Semine and Damon, 1975; İşcan and Loth, 1986c; McCormick and Stewart, 1988;
Barchilon et al., 1996; Yoder et al., 2001). Additionally, tests of the İşcan et al. (1984a; 1985)
methods were conducted on non-fourth ribs to determine the accuracy of the method when this
rib is not available.
Mann (1993) described an approach to side and sequence ribs by using the features of the
maximum rib length, size, and shape of the articular facets, the length between the facets and rib
angle, and the height of the rib heads. Additionally, Dudar (1993) developed a method to identity
the rib based on morphological changes seen in typical and atypical ribs, specifically changes
seen in the angle in both the horizontal and vertical planes. The author then tested the İşcan et al.
(1984a; 1985) European-American methods on ribs two through nine, and found no significant

21	
  

differences between non-fourth and fourth ribs. Therefore it was concluded that “the fourth rib
standards can be cautiously applied to ribs two through nine when rib number cannot be
identified, or when rib four is not preserved” (Dudar, 1993:796–797).
İşcan et al. (1989) tested ribs three, four, and five on a sample of 70 males and females in
order to determine if they had similar accuracy rates. Results indicated that all three ribs were
correctly phased in the majority of the cases and only minor differences between the ribs were
observed. Additional tests of the third and fifth sternal rib ends were on African-Americans from
the Terry Collection (Loth, 1990), the Spitalfields Collection (Loth, 2005), and a modern
forensic sample (Loth et al., 1994). Specifically, Loth (1990) examined 136 African-Americans
from the Terry Collection and compared those to modern a sample of 268 modern EuropeanAmericans and 89 African-Americans. Intercostal variation was examined using metric
measurements of superior-inferior height and maximum anterior-posterior breadth. The results
showed that most of ribs three and five were aged similarly to that of rib four, indicating that
these ribs could be used as a proxy for when the fourth rib is not present (Loth, 1990).
Loth et al. (1994) also examined ribs three, four, and five of 135 European-American
males and females in order to determine if rib aging standards could be used on adjacent ribs.
Results showed that ribs three and five were typically phased correctly, with all three ribs placed
in the same age range for 79 percent of the cases and 98 percent within one phase (Loth et al.,
1994). From these studies it was demonstrated that ribs three and five could be used as a
substitute in cases when the fourth rib was not present or damaged (Loth, 1990; Loth et al., 1994;
Aykroyd et al., 1999; Loth, 2005).
Alsup (2007) additionally evaluated the İşcan et al. (1984a; 1985) European-American
methods on second, fourth, and eighth ribs using a modern forensic sample of 156 individuals

22	
  

from the William M. Bass Donated Collection. The sample was divided into males and females,
consisting of mostly European-Americans, but also a few African-Americans and Hispanics.
Statistically significant differences were found on the right second and eighth ribs, with the
second ribs overestimating and the eight ribs underestimating the age of the individuals (Alsup,
2007). Yoder et al. (2001) also evaluated the accuracy of the İşcan et al. (1984a; 1985)
European-American methods on both left and right second through ninth ribs. The sample
consisted of 231 European-American and African-American males and females from the Terry
Collection, the Maxwell Museum Collection, and the William M. Bass Donated Collection. It
was found that for ribs four through nine, the left side does not differ significantly from the right
side, although statistically significant differences between ribs do appear when sex and ancestry
are taken into account. Additionally, ribs two and three were found to have significant
differences between the right and left sides, suggesting that the sides are not subjected to the
same amount of stress. The authors also found that the right third and right fifth through ninth
ribs do not significantly differ from the right fourth rib, suggesting these ribs could be
substituted, although they should be used with caution. Finally, the authors recommend a
composite scoring method of all ribs when possible due to variations in accuracy based on rib
number and side (Yoder et al., 2001).

First Rib Aging
The first rib has also been examined, as it was found to ossify at a more advanced rate
than the lower ribs (Semine and Damon, 1975; Barchilon et al., 1996). Kunos et al. (1999)
developed an age estimation technique from the first rib using a sample from the Hamann-Todd
Osteological Collection. This technique involves measurements of rib length and the costal face

23	
  

thickness, as well as morphological features seen in the costal face, rib head, and tubercle (Kunos
et al. 1999).
Kurki’s (2005) tests of the Kunos et al. (1999) method found it to be relatively accurate
and precise, although Schmitt and Murail (2004) found it to be subjective, with only 55 percent
of the sample correctly classified. Advantages of first rib aging methods are that the first rib is
easily identifiable, less fragile, and subjected to limited mechanical stress. Additionally, these
first rib aging methods take degenerative changes of the entire rib into account rather than just
the sternal end (Kunos et al., 1999; Kurki, 2005). A modification of the Kunos et al. (1999)
method has also been proposed by using the same criteria with additional categorical traits using
a Bayesian analysis (DiGangi et al., 2009). It was found that two morphological traits were agedependent: the geometric shape of the costal face and surface texture or the tubercle facet
(DiGangi et al., 2009).

Accuracy of the Method
In order to achieve an accurate age estimate, multiple aging techniques should be utilized.
Unfortunately, some elements may be missing or damaged and therefore the number of
techniques available becomes limited. Due to these issues, the accuracy of all aging methods
should be tested on multiple populations and contexts, as well as in comparison to other aging
techniques.
Galera et al. (1995) tested multiple macroscopic aging methods, including the İşcan et al.
(1984a; 1985) European-American methods on the Terry Collection on left and right fourth and
seventh ribs. Results indicated that for all methods, there were no statistically significant
differences among recorders, demonstrating low rates of inter-observer error, indicating that the

24	
  

methods are easy to apply and minimal experience is needed. Other tests by Loth and İşcan
(1989) also found low rates of inter-observer error for the İşcan et al. (1984a; 1985) EuropeanAmerican methods. These results are in contrast to Dudar et al. (1993), who found high rates of
inter-observer error within the method. Kimmerle et al. (2008) also tested inter-observer error of
multiple methods, including the İşcan et al. (1984a; 1985) European-American methods, on a
sample of individuals from Kosovo. Observers disagreed by one phase, with most discrepancies
in Phases 3 through 7. This was determined to be a result of the variations seen within these
phases, such as with the presence and amount of rim projections (Kimmerle et al., 2008).
In another study, İşcan et al. (1992) tested the İşcan et al. (1984a; 1985) EuropeanAmerican standards in comparison to the Angel’s (1980) modification of Todd’s (1920) pubic
symphysis aging technique, as it is often thought to be of the most accurate and reliable aging
methods. The right pubic bone and right fourth rib of an autopsy sample of 80 EuropeanAmericans of known ancestry, sex, and age were used for the sample. The results showed that
the rib has less variation than the pubic symphyses. Additionally, the İşcan et al. method was
able to estimate individuals within one phase 93 percent of the time, compared to 61 percent for
the pubis, with females that were aged more accurately than males for both methods. The authors
believe that the main reason for the difference is that the pelvis often suffers from weight bearing
forces, which applies additional functional stress that may make the bone appear older (İşcan et
al., 1992). Comparatively, these stresses are not applied to the ribs, or are less of a factor, and
therefore morphological changes as seen in the rib have less variation and can give narrower age
ranges (İşcan et al., 1992; Loth and İşcan, 1990; Aykroyd et al., 1999).
Baccino et al. (1999) examined four techniques to estimate the age of death in adult
skeletal remains, including the İşcan et al. (1984a; 1985) European-American methods on a

25	
  

sample of 19 adults of known age and sex collected from autopsies in France. Tests indicated
that the İşcan et al. (1984a; 1985) methods tended to overestimate age; therefore the authors
suggested that multiple age estimators should be used in conjunction when possible, due to better
rates of accuracy (Baccino et al., 1999). Martrille et al. (2007) also tested four aging methods,
including the İşcan et al. (1984a; 1985) sternal rib end method on 218 individuals from the Terry
collection, consisting of 98 European-Americans and 120 African-Americans. The results
showed that all methods tested tend to over-age young individuals and under-age older
individuals, although the İşcan et al. (1984a; 1985) method displayed the least amount of
inaccuracy. It was also concluded that as many age estimators as possible should be used and the
final evaluation should consider methods that have the highest accuracy (Martrille et al., 2007).
Finally, Brown (2009) tested multiple aging methods, including the İşcan et al. (1984a)
European-American male method, to determine accuracy and reliability rates. For the İşcan et al.
(1984a) method, an overall tendency to under-age males was found. When individuals were
assigned into the closest two phases, accuracy increased to 100 percent, possibly indicating that
the confidence intervals for this method are too small (Brown, 2009).

Applicability of the Method
Many studies demonstrated that the İşcan et al. (1984a; 1985) European-American
methods were accurate in forensic contexts; therefore the applicability of the method to
archaeological and historic contexts was also examined by additional researchers. Loth (2005)
tested the İşcan et al. (1984a; 1985) methods on the archeological Spitalfields cemetery
collection, consisting of 74 adult individuals: 36 males and 38 females. Results showed that this
sample displayed the same age-related degenerative patterns to those of modern European-

26	
  

Americans. An additional test of the method by Potter (2010) examined differences between
historic (reference) and modern (recent) collections. Multiple methods were tested for this
research, including the İşcan et al. (1984a; 1985) European-American methods. Results did not
indicate differences in the rate of aging between the historic and modern collections, and the
İşcan et al. (1984a; 1985) methods produced strong and highly significant correlations, indicating
that the method is accurate and reliable.
In contrast, Saunders et al. (1992) applied multiple aging techniques, including the İşcan
et al. (1984a; 1985) European-American methods, to a documented archaeological sample.
Results showed that rates of intra- and inter-observer error were an issue for the İşcan et al.
(1984a; 1985) methods, suggesting that the technique needs to be further revised. Additional
issues, such as postmortem damage due to disinterment trauma or taphonomic processes were
also determined to be a pitfall of this method (Saunders et al., 1992).
Finally, Ritz-Timme et al. (2000) gave recommendations for when it is appropriate to use
specific age estimation techniques, including the İşcan et al. (1984a; 1985) European-American
methods. The authors state that the method is appropriate for cadavers, human remains, and
historic and archaeological remains. It is also most accurate for individuals younger than 40
years, as the method becomes less accurate with age. Additional factors of sex, ancestry, and
socio-economic status should also be taken into account, as they may also influence the rate of
these degenerative changes (Ritz-Timme et al., 2000).

Improvements of the Method
Based on results of the tests of the İşcan et al. (1984a; 1985) European-American
methods, Hartnett (2007; 2010) developed a modification of the technique, known as the

27	
  

Hartnett-Fulginiti method. The study was first conducted as a test of the method as a part of
Hartnett’s (2007) dissertation, which was later modified and published (Hartnett, 2010). The test
was performed on a modern autopsy sample located at the Maricopa County Forensic Science
Center, consisting of individuals of both sexes and of multiple ancestries. The results indicated
moderately high correlation values, suggesting that the method does work, but could be
improved. However, the cross-tabulation results were contradictory, as they resulted in poor
agreement between the estimated and actual phase numbers, with less than half of the sample
assigned to the correct phase, indicating that the method is reliable, but not accurate. Therefore in
order to improve accuracy of the method, the Hartnett-Fulginiti method revised descriptions of
phases and summary statistics and included a disclaimer that bone quality may affect phase
classification (Harnett, 2007; 2010).
The Hartnett-Fulginiti (2007; 2010) method was developed on the same sample from the
Maricopa County Forensic Science Center. Ribs were separated by sex and then seriated based
on age-related morphological changes. Three features were considered to be the most important
in age determination: the pit depth, the rim edges, and bone quality. The method consists of
seven phases and an additional male variant phase, consisting of complete calcification of the
costal cartilage that is sometimes seen in males (Appendices C and D). In general the new phase
classification begins with morphological changes that start as a shallow and flat U-shaped pit
(Phase 1). An indentation begins within the pit, which then takes on a V-shape with a welldefined rim (Phase 2), becoming slightly irregular (Phase 3). The pit then takes on a U-shape
(Phases 3 through 7). As age further increases, porosities begin to form inside the pit (Phases 3
through 6), with the bone becoming course and brittle and bony growths present within the pit

28	
  

(Phase 7). Although males and females have separate standards, the basic age changes remain the
relatively the same (Hartnett, 2007; 2010).

Modifications of the Method
The İşcan et al. (1984a; 1985) European-American methods have also been applied and
modified to reflect specific populations. One such study applied the İşcan et al. (1984a; 1985)
methods to a Turkish sample of 150 males and 144 females of known age from a modern,
forensic context in order to evaluate the accuracy. Results indicated that variability increased
with age, especially after Phase 5, possibly due to small sample size of older individuals.
Therefore the method was determined to be accurate for Turkish individuals (Yavuz et al., 1998).
A Balkan population was also used to evaluate the İşcan et al. (1984a; 1985) EuropeanAmerican methods (Đonić et al., 2005; Stuart and Konigsberg, 2004). These studies were in
response to the İşcan et al. (1984a; 1985) aging standards being used for age estimation of
Balkan war dead, although the accuracy of these methods had not yet been tested. Therefore
Đonić et al. (2005) evaluated the İşcan et al. (1984a; 1985) methods on a Balkan sample and
found significant differences between the actual and estimated phases, especially in older
individuals.
Stuart and Konigsberg (2004) also tested the reliability of the İşcan et al. (1984a; 1985)
methods using an American and Balkan sample. The American sample included a total of 188
individuals from the William B. Bass Donated Collection, Forensic Skeletal Collections, and the
Forensic Data Bank, and the Balkan sample consisted of 529 males and 78 females. Cumulative
probit and Bayesian approaches were proposed in order to improve accuracy. In addition,
advantages of these more complex statistical models are that they are less subjective, give more

29	
  

realistic error estimates, and do not force truncated age estimates, such as 50 or older (Stuart and
Konigsberg, 2004).
Fanton et al. (2010) examined the three variables used for the original component
analysis method: pit depth, pit shape, and rim and wall configurations. A modern French sample
was used, consisting of 59 males. Reproducibility for pit depth was found to be low, due to
inaccuracies in measurement location. Pit shape and rim and wall configurations were found to
have poor reproducibility, attributed to the descriptions. It was determined that the variables are
good indicators of age, but that they need to analyzed with more detail and with additional
variables, as well as the development of a multivariate statistical analysis (Fanton et al., 2010).
Verzeletti et al. (2010) developed a revision of the component analysis method on 49
Italian males. Three components were used: the articular surface, anterior-posterior walls, and
superior-inferior edges. Each variable is scored and then input into a linear regression formula.
Results indicated that for this formula, age was more accurately assessed for individuals younger
than 50 (Verzeletti et al., 2010).
Finally, Oettlé and Steyn (2000) tested and applied the İşcan et al. (1984a) EuropeanAmerican methods on modern South Africans of African ancestry. A total of 339 rib ends were
used: 265 males and 74 females, ranging in age between four and 94 years in males and eleven to
77 years in females, all of known age, sex, and ancestry. The sample consisted of right fourth
ribs that were removed during postmortem examinations from the State Mortuary and from the
cadaver collection at the Department of Anatomy, University of Pretoria.
The authors found that, generally, young males were often under-aged and older males
were often over-aged, possibly indicating that in this population age related changes might take
longer to appear in males. The method appeared less reliable for females, with no tendency to

30	
  

over- or under-age females. Features that were found only in African-Americans, such as more
squared off and pointy scallops and an earlier development of bony projections and window
formation were also found in the South African sample, although it was found that these
degenerative features did not necessarily indicate older age. Additionally, plaque in the interior
of the pit was also found more frequently in this population (Appendices E and F). These
differences may have been the result of genetic influences, as well as disease, physical activity
level, cultural variations, and socioeconomic and nutritional differences (Oettlé and Steyn,
2000).
Therefore, due to the unreliability of the İşcan et al. (1984a; 1985) methods on this
population, new phases were developed by seriating the ribs according to age-related
morphological changes. The new age ranges for each phase were developed within the 95
percent confidence level. For this method it was generally found that the fourth right sternal rib
end begins as a billowy with a regular rim with an epiphyseal ring (Phase 0). Age related
changes begin with an indication on the articular surface, starting as a round and regular pit
(Phase 1), then deepening into a V-shape with scalloping edge that may be “pointed squared off,
or flattened, or may be regular and receding” (Phase 2). The pit later takes on a U-shape with
edges becoming more irregular (Phases 3 through 8). Age ranges for these phases were modified,
as seen in timing of age related morphological changes (Oettlé and Steyn, 2000).
	
  

31	
  

MATERIALS AND METHODS
Materials
This research evaluated the accuracy of the İşcan et al. (1984a; 1985), the HartnettFulginiti (2007; 2010), and the Oettlé and Steyn (2000) sternal rib aging techniques. The sample
included modern skeletal remains from the William M. Bass (WM Bass) Donated Skeletal
Collection and historic skeletal remains from the Hamann-Todd Osteological Collection. Both
collections were combined due to the limited amount of modern skeletal material available for
research.
The entire sample consisted of right fourth ribs from 411 individuals, ranging in age from
24 to 97 years. The skeletal sample included a total of 209 African-Americans and 202
European-Americans. Of these, 143 were African-American males, 66 were African-American
females, 149 were European-American males, and 53 were European-American females. The
average age of the total sample was 42 years, with a median of 38 years and a mode of 40 years
(Table 1).
Table 1: Distribution of total sample
Count

% of
Total

African-American Males
African-American Females
Total African-Americans

143
66
209

34.8%
16.1%
50.9%

37
33
36

35
31
33

35
24
24

15.12
11.26
14.00

18 – 84
16 – 66
16 – 84

European-American Males
European-American Females
Total European-Americans

149
53
202

36.3%
12.9%
49.2%

46
53
48

45
54
46

40
38
40

15.87
19.08
16.98

20 – 91
16 – 97
16 – 97

Total Males
Total Females
Total Sample

292
119
411

71.0%
29.0%
100%

42
42
42

40
38
38

40
38
40

16.13
18.27
16.84

18 – 91
16 – 97
16 – 97

Entire Sample

Mean Median Mode Standard
Age
Age
Age
Age Deviation Range

32	
  

For the purposes of this study, only European-Americans and African-Americans of
known age, sex, and ancestry were included in order to determine any possible differences in
accuracy between these two ancestral groups. Many individuals were excluded from sampling
due to postmortem damage. This included individuals that had undergone autopsies, as ribs are
cut near the sternal end in this process, which can prevent the identification of the right fourth rib
in later analysis. Additional individuals suffering from other postmortem damage, such as
fragmentation of the right fourth rib, were also excluded. Therefore all individuals of African and
European ancestry that contained a usable right fourth rib were included in analysis. Biological
profile information consisting of age, sex, and ancestry were recorded from the databases
provided by each collection, with only sex and ancestry known to the author at the time of
examination.
As demonstrated in Table 2, the distribution of ancestry of the total sample is relatively
even, with 50.9 percent of the sample classified as African-Americans and 49.1 percent classified
as European-Americans. Additionally, the WM Bass sample contained mostly EuropeanAmericans, while the Hamann-Todd sample contained slightly more African-Americans.
Table 2: Distribution of sample by ancestry and sex
African-Americans

European-Americans

Total

11
4
15

37
29
66

48
32
81

Hamann-Todd Males
Hamann-Todd Females
Total Hamann-Todd

132
62
194

112
24
136

244
66
330

Total Males
Total Females
Total Sample

143
66
209

149
53
202

292
119
411

WM Bass Males
WM Bass Females
Total WM Bass

33	
  

Age for the entire sample ranged from 16 to 97 years, with a majority of individuals
classified as young to middle aged (Figure 1). African-Americans have a slightly younger
distribution, with a mean age of 36 years and a standard deviation of 14 years, while EuropeanAmericans had a mean age of 48 years with a standard deviation of 16.98 years. Although
African-Americans contained a skewed sample of younger individuals, the European-American
sample approached a normal distribution.

Count

Figure 1: Distribution of age by ancestry for the entire sample
90
80
70
60
50
40
30
20
10
0

African-Americans
European-Americans

Age-at-death

The Hamann-Todd Subsample
A majority of the data was collected from historic individuals from the Hamann-Todd
Osteological Collection curated at the Cleveland Museum of Natural History. This collection
consists of over three thousand individuals of African and European ancestry born between the
years of 1823 and 1934 (Potter, 2010). Carl August Hamann, an anatomy professor at the
Western Reserve University in Cleveland, Ohio, began the collection in 1893 in order to develop
an anatomical teaching museum. Many of these remains were first used for dissection at the

34	
  

university and then later cleaned and curated for the osteological collection (Cobb, 1981;
Mensforth and Latimer, 1989; Wescott, 1996; Quigley, 2001). In 1912, Thomas Wingate Todd
arrived at the university and expanded the collection by including unclaimed bodies from the
nearby morgue and hospitals in addition to cadaveric bodies. The collection persisted until
Todd’s death in 1938, with the collection finally moved in 1951 from the university to its current
location at the Cleveland Museum (Jones-Kern and Latimer, 1997; Quigley, 2001; Potter, 2010).
Documentation of individuals within the Hamann-Todd Osteological Collection includes
age, ancestry, sex, date and place of birth, occupation, and date and cause of death (Cobb, 1959;
Moore-Jansen, 1989). Within the collection, a majority of the individuals of European ancestry
were born abroad, while most of the African-Americans migrated to the area from other regions
within the United States (Cobb, 1952). This collection represents early twentieth century
Industrial America before the introduction of modern medicine. Therefore, many of these
individuals were of low socioeconomic status and may have been of poor health compared to the
rest of the population at the time (Cobb, 1981; Mensforth and Latimer, 1989; Wescott, 1996;
Kunos et al., 1999). Additionally, although this collection is very well documented for many
individuals due to provided death certificates, others have estimated ages that were given by the
family (Brooks and Suchey, 1990). This may be somewhat problematic, as ages for some
individuals may not be exact.
Data collection at the Hamann-Todd Osteological Collection occurred from May 14,
2012 to May 23, 2012 and consisted of 330 individuals, ranging in age from 16 to 80 years
(Table 3). The skeletal sample included 132 African-American males, 62 African-American
females, 112 European-American males, and 24 European-American females. The average age
of the Hamann-Todd sample was 38 years, with a median of 35 years and a mode of 40 years.

35	
  

Table 3: Distribution of the Hamann-Todd subsample
Count

% of
Total

African-American Males
African-American Females
Total African-Americans

132
62
194

40.0%
18.8%
58.8%

36
32
35

33
31
32

35
24
24

13.65
10.67
12.84

18 – 80
16 – 65
16 – 80

European-American Males
European-American Females
Total European-Americans

112
24
136

33.9%
7.3%
41.2%

43
40
42

40
40
40

40
38
40

14.52
13.49
14.33

20 – 80
16 – 62
16 – 80

Total Sample

330

100%

38

35

40

13.99

16 – 80

Hamann-Todd Sample

Mean Median Mode Standard
Age
Age
Age
Age Deviation Range

The William M. Bass Subsample
Additional data was collected from modern individuals at the William M. Bass (WM
Bass) Donated Skeletal Collection located at the University of Tennessee, Knoxville. The
collection resulted from a body donation program that began in 1981, in which many of the
remains were used for decomposition studies at the Anthropology Research Facility and were
then later cleaned and curated for the skeletal collection (Ubelaker and Hunt, 1995; Bassett et al.,
2003; Tersigni-Tarrant and Shirley, 2012). It is now regarded as the largest collection of modern
individuals in the United States, containing almost one thousand individuals born after 1982
(Tersigni-Tarrant and Shirley, 2012). Documentation of the remains consists of sex, age,
ancestry, cause and date of death, date of birth, stature, weight, handedness, and education level
(Potter, 2010). The collection contains mostly European-American males, but also individuals of
other ancestries from a variety of socioeconomic backgrounds, with a majority of individuals
originating from the Southeastern United States (Tersigni-Tarrant and Shirley, 2012).
Data collection at the WM Bass Donated Skeletal Collection was conducted from July
30, 2012 to August 2, 2012 and consisted of 81 individuals, ranging in age from 24 to 97 years
(Table 4). The skeletal sample included eleven African-American males, four African-American
36	
  

females, 37 European-American males, and 29 European-American females. The average age of
the WM Bass sample was 59.25 years, with a median of 59 years and a mode of 49 years. Higher
standard deviations, mean ages, and age ranges reflect the older age of individuals within the
donated WM Bass Collection.
Table 4: Distribution of the WM Bass subsample
WM Bass Sample

Count

% of
Total

Mean Median Mode Standard
Age
Age
Age
Age Deviation Range

African-American Males
African-American Females
Total African-Americans

11
4
15

13.6%
4.9%
18.5%

58.55
42.25
54.20

55.0
39.5
55.0

49
49

16.51
17.44
17.76

25 – 84
24 – 66
24 – 84

European-American Males
European-American Females
Total European-Americans

37
29
66

45.7%
35.8%
81.5%

57.86
63.62
60.39

58.0
61.0
60.0

57
73
38

14.42
16.51
15.52

26 – 91
36 – 97
26 – 97

Total Sample

81

100%

59.25

59.0

49

16.02

24 – 97

Methods
The objective of this research was to evaluate accuracy of the İşcan et al. (1984a; 1985),
Hartnett-Fulginiti (2007; 2010), and Oettlé and Steyn (2000) sternal rib aging methods. In
addition, differences in accuracy between European-Americans and African-Americans were
examined, due to indications of age-related variations based on ancestry, as seen in the literature.
Specifically, the Oettlé and Steyn (2000) method was evaluated for these differences, as well as
to determine if it is a viable method in the United States.
Although the literature has shown that the use of the left side or other ribs may give
similar results (e.g. İşcan et al., 1989; Loth, 1990; Dudar, 1993; Loth et al., 1994; Yoder et al.,
2001), only the right fourth sternal rib end was utilized for this research, as all three methods
were developed specifically on the right fourth rib. Additionally, an attempt was made to include

37	
  

at least 15 individuals per phase for both African-American and European-American categories,
although young individuals for both groups were limited. Finally, in order to prevent bias in the
assignment of phase numbers, each method was performed separately on all individuals.

The İşcan et al. Method
The İşcan et al. (1984a; 1985) method is a phase system developed on an autopsy sample
consisting of known individuals of European ancestry. To apply this method, the observer must
first determine the sex of the decedent, as there are separate standards for males and females.
Then the right fourth sternal rib is examined and matched to one of nine (0 through 8) phases.
Each phase contains descriptions and photographs of the typical age-related changes seen with
an associated age range. Due to the inclusion of a confidence interval, age ranges for each phase
typically overlap. Rib casts are also available, which give a three-dimensional image of these
age-related changes, as this method emphasizes morphological degenerative changes as the basis
for age. The photographs, descriptions, and age ranges for the method may be found in Appendix
A for males and Appendix B for females. Table 5 displays the distribution of individuals within
each phase for the İşcan et al. (1984a; 1985) method.
Table 5: Number of individuals for the İşcan et al. method by phase
0

1

2

3

4

5

6

7

8

African-American Males
African-American Females
Total African-Americans

0
0
0

3
0
3

24
3
27

19
13
32

20
20
40

32
23
55

27
4
31

9
3
12

9
0
9

European-American Males
European-American Females
Total European-Americans

0
0
0

0
0
0

7
3
10

8
0
8

16
3
19

37
12
49

39
14
53

15
10
25

27
11
38

Total Sample

0

3

37

40

59

104

84

37

47

38	
  

The Hartnett-Fulginiti Method
The Hartnett-Fulginiti (2007; 2010) method is also a phase system using the right fourth
sternal rib end, and was developed on a modern autopsy sample of known individuals of multiple
ancestries. Specifically, it was created as a revision to the İşcan et al. (1984a; 1985) method in
order to improve accuracy of the technique. The estimation of age involves matching an
unknown right fourth sternal rib end to one of eight phases (1 through 7), which give
descriptions and photographs of the typical age-related changes seen, with standards that are
separated on the basis of sex. Age ranges for each phase may overlap, due to the inclusion of a
confidence interval. Rib casts are not available for this method, which may make it more difficult
to visualize morphological features of each phase, although the emphasis for age determination is
instead placed on the overall bone quality. The photographs, descriptions, and age ranges for the
method may be found in Appendix C for males and Appendix D for females. Table 6 displays
the distribution of individuals within each phase for the Hartnett-Fulginiti (2007; 2010) method.
Additionally, individuals under the age of 18 were classified as a Phase 0, which is not a phase
listed for the method.
Table 6: Number of individuals for the Hartnett-Fulginiti method by phase
0

1

2

3

4

5

6

7

African-American Males
African-American Females
Total African-Americans

0
2
2

11
8
19

37
16
53

33
24
57

24
9
33

24
4
28

7
2
9

7
1
8

European-American Males
European-American Females
Total European-Americans

0
1
1

2
2
4

13
1
14

28
10
38

37
7
44

31
12
43

26
10
36

12
10
22

Total Sample

3

23

67

95

77

71

45

30

39	
  

The Oettlé and Steyn Method
Finally, the Oettlé and Steyn (2000) method is a phase system developed in South Africa
based on an autopsy sample of known individuals of African ancestry. Age estimation is
conducted by matching the unknown rib to one of nine phases (0 through 8 for males; 0 through
7 for females) based on descriptions and photographs of the typical age-related changes seen,
with standards are separated on the basis of sex. The method is similar to the İşcan et al. (1984a;
1985) method, as it places more emphasis on morphological age-related characteristics. Rib casts
are not available for this method, which may slightly inhibit the technique due to the emphasis
on visual characteristics, as one or two photographs of each phase may not be sufficient to cover
the variation within each phase. The photographs, descriptions, and age ranges for the method
may be found in Appendix E for males and Appendix F for females. Table 7 displays the
distribution of individuals within each phase for the Oettlé and Steyn (2000) method.
Table 7: Number of individuals of phases for the Oettlé and Steyn method by phase
0

1

2

3

4

5

6

7

8

African-American Males
African-American Females
Total African-Americans

0
0
0

14
1
15

20
11
31

21
14
35

27
10
37

23
21
44

21
6
27

10
3
13

7
7

European-American Males
European-American Females
Total European-Americans

0
0
0

4
0
4

6
3
9

10
2
12

28
1
29

33
12
45

24
15
39

27
20
47

17
17

Total Sample

0

19

40

47

66

89

66

60

24

Intra-Observer Error
The author reexamined a total of 105 individuals in order to determine rates of intraobserver error (Table 8). Random samples of 15 individuals of each sex and ancestry, totaling 60
individuals, were re-tested from the Hamann-Todd collection, while only a total of 45
40	
  

individuals from the WM Bass collection were re-tested due to the small sample size of AfricanAmericans. The total sample consisted of 26 African-American males, 19 African-American
females, 30 European-American males, and 30 European-American females. Each rib was
reassessed individually for all three methods. Additionally, rates of inter-observer error could not
be calculated due to the use of multiple collections.
Table 8: Sample used for intra-observer error
Hamann-Todd

WM Bass

Total

African-American Males
African-American Females
Total African-Americans

15
15
30

11
4
15

26
19
45

European-American Males
European-American Females
Total European-Americans

15
15
30

15
15
30

30
30
60

Total Sample

60

45

105

Statistical Analysis
The objective of this research was to evaluate the accuracy rates of three sternal rib aging
techniques: İşcan et al. (1984a; 1985), Hartnett-Fulginiti (2007; 2010), and Oettlé and Steyn
(2000). The rates of accuracy for the three methods were compared for the entire sample,
African-Americans, European-Americans, males, and females. Multiple statistical tests were
performed, including descriptive statistics, correlation, cross-tabulation, inaccuracy, bias, and
intra-observer error. All statistical analyses were conducted by hand or through the use of IMB
SPSS statistical software, version 20 and a Microsoft Excel spreadsheet available for download
from http://www.stat-help.com.

41	
  

Descriptive Statistics
Summary statistics were collected for each method, and were presented earlier in this
chapter. The counts; mean, median, and mode ages; standard deviations; and age ranges were
calculated for African-Americans, European-Americans, and the entire sample, with these
categories further divided by sex. Frequencies were also calculated and were separated on the
basis of ancestry and sex.

Correlation
All three aging methods utilized for this research were phase systems, meaning a phase
number, corresponding to an overall age range, was assigned to every individual. For the
purposes of statistical analysis, the estimated phase number was then compared to the actual
phase number, as determined by the individual’s actual age-at-death. Although the data collected
for this research was categorical, each phase was assigned a number, e.g. Phase 1 was assigned a
1, so that the data could be compared on an ordinal scale. Additionally, these categories were
mutually exclusive, had a logical order, and were scaled, therefore a nonparametric correlation
analysis was required for the analysis, as these types of tests do not require normal or
homogeneous data (Hinkle et al., 1988; Drennan, 2004; Hartnett, 2010; Sokal and Rohlf, 2012).
Finally, Pearson’s r correlation was excluded from analysis, as it assumes a normal distribution.
The Chi-squared test was also excluded, as the amount of zeros obtained would have violated the
test.
Therefore, in order to find the level of association between the observed phase and the
actual phase, a Spearman’s coefficient of rank correlation was employed, as this test is highly
robust, does not require normally distributed data, and does not rely on means or standard

42	
  

deviations or require other modifications (Drennan, 2004). Correlation coefficients produced by
Spearman’s rank describe the extent of the relationship between two sets of data (Hinkle et al.,
1988). The values indicate the general linear pattern of the data between -1.0 and 1.0, with the
sign indicating the direction of the relationship (Hinkle et al., 1988; Drennan, 2004). In order to
determine the magnitude of the association, the absolute value of this number must be used, with
1.0 indicating a perfect correlation and 0 indicating no relationship between the variables (Hinkle
et al., 1988).
Spearman’s coefficient of rank correlation was calculated for comparisons between the
estimated and actual phase numbers and between the estimated and actual ages. For each method
a phase number was assigned to each individual based on the available casts and written
descriptions from the original articles (İşcan and Loth, 1993; İşcan et al., 1984a; 1985; Hartnett,
2007; 2010; Oettlé and Steyn, 2000). This observed phase number was then compared to the
actual phase number, as determined by the individual’s actual age-at-death.
Comparisons were also made between the actual and estimated ages, which were
conducted by converting the estimated phase number for each individual to the mean age for that
specific phase. For example, for the Hartnett-Fulginiti (2007; 2010) method, if a female was
estimated to be a Phase 1, that individual had an estimated age range of 18 to 22 years and
therefore was approximately 19.57 years, which is the mean age for that phase. The estimated
age and the individual’s actual age-at-death were then compared through Spearman’s coefficient
of rank correlation.
In order to determine statistical significance in accuracy between the methods, the
differences between the correlation coefficients must be evaluated, but since they are ordinal and
not normally distributed, they cannot be directly compared (Hinkle et al., 1988). Therefore, in

43	
  

order to determine the statistical significance of the correlation coefficients obtained, Fisher’s z
and Steiger’s z transformations were used, as these tests allow for the comparison of two
correlation coefficients by converting the r-values to normally distributed z-scores (Cohen and
Cohen, 1983). These tests determine how much the normal curve deviates in order to conclude if
the correlation coefficients are equal, with values that are greater than |1.96| indicating statistical
significance for the 2-tailed test (Cohen and Cohen, 1983).
More specifically, Fisher’s z transformations require two independent correlation
coefficients, in order to determine if the two samples are homogeneous in the population (Cohen
and Cohen, 1983). Therefore these tests were only utilized for comparisons of categories within
methods, e.g. African-Americans and European-Americans within the Oettlé and Steyn (2000)
method, as these groups are independent of each other. However, comparisons between methods,
e.g. overall accuracy for the İşcan et al. (1984a; 1985) method and the Hartnett-Fulginiti (2007;
2010) method, cannot be calculated with Fisher’s z, as the coefficients are correlated due to the
methods utilizing the same individuals. Thus, a Steiger’s z transformation was used, as this test
allows for comparisons between correlation coefficients from the same sample. Additionally,
Steiger’s z transformations were only used for comparisons between the actual and estimated
ages, as comparisons between actual and estimated phases cannot be conducted due to differing
actual phase values for each method. For example, a female may be 24 years, and would
therefore have an actual phase of 2 for the Hartnett-Fulginiti (2007; 2010) method but an actual
phase of 3 for the İşcan et al. (1984a; 1985) method. These differences would violate statistical
analysis. Therefore only correlation coefficients of the estimated and actual ages could be used,
as every individual had the same actual age-at-death for every method. The calculation of
Fisher’s z and Steiger’s z transformations were conducted through a downloaded Microsoft

44	
  

Excel spreadsheet from http://www.stat-help.com, which used Cohen et al.’s (2003) formula for
the Fisher’s z transformation and Steiger’s (1980) formula for the Steiger’s z transformation.

Sample Proportion
Finally, a binomial distribution was utilized to determine how well each method could
predict age. This test is based on two mutually exclusive results, in which each observation was
either estimated correctly, defined as a success, or incorrectly, defined as a failure. Specifically, a
population proportion was used, because the results of the entire population are unknown, as
only a proportion of the population was sampled. Due to these reasons and a large sample size, a
binomial distribution was approximated by a normal distribution, known as the large sample
approximation to the binomial distribution. This test states that even if the data is not normally
distributed, the distribution of the sample approaches a normal distribution as sample size
increases (Freund et al., 2010).
This test was conducted in order to determine the number of successes in the sample,
which was then used to estimate the proportion of successes within a larger population. For each
method, the proportion of successes, standard error, and confidence intervals were determined.
The standard error was calculated, as it is an index of precision and indicates the amount of error,
with a larger standard error indicating that the method is less precise (Peers, 1996). Confidence
intervals were also calculated as they fall symmetrically around the sample proportion and
therefore give a range of values and probable limits in which the parameter encompasses (Hinkle
et al., 1988; Peers, 1996). Therefore, the method that displays the highest probability of
successes indicates that it is a better predictor of age, with non-overlapping confidence intervals
indicating significance.

45	
  

Cross-tabulation
Cross-tabulation tables, adapted from Hartnett (2007), were also used to compare the
estimated phase to the actual phase, giving a visual reference to the distribution of the data.
Tables demonstrate the number of instances that each phase was estimated correctly, as well as
the estimated phases in relation to the actual phases. Correct and incorrect phase classifications
were calculated and percentages are given for correct phase, one-phase differences, and twophase differences.

Inaccuracy and Bias
Inaccuracy and bias were also calculated for all groups and all methods between both the
estimated and actual phase numbers and the actual and estimated ages. Rates of inaccuracy, also
known as the absolute mean error, were calculated for all methods through the equation Σ
|estimated phase - actual phase| / N, which represents the sums of the error from the actual phase
number, divided by the number of individuals, without regard to whether the error is positive or
negative (Russell et al., 1993; Martrille et al., 2007; Fleischman, 2011; Nawrocki, 2011).
Inaccuracy was also calculated in terms of age through the equation of Σ |mean age - actual age| /
N, which represents the sums of the absolute values of the difference between the mean ages of
the estimated phases and the chronological, or actual, ages, divided by the number of individuals
(Russell et al., 1993; Martrille et al., 2007). As mentioned earlier, due to the nature of the phase
method, mean ages of each phase were used in substitute for an estimated age (Russell et al.,
1993; Martrille et al., 2007; Fleischman, 2011; Nawrocki, 2011).
Bias is the tendency to either over- or under-age an individual and was calculated for
both the estimated and actual phase numbers and the actual and estimated ages for each method.

46	
  

Phase bias was calculated through the equation Σ (estimated phase - actual phase) / N, which is
the sum of the estimated and actual phase numbers, divided by the number of individuals, while
bias in terms of age was also calculated through the equation Σ (estimated age - actual age) / N,
which is the sum of the estimated and actual ages, divided by the number of individuals. Again,
due to the nature of the phase method, mean ages of each phase were used in substitute for an
estimated age (Russell et al., 1993; Martrille et al., 2007; Nawrocki, 2011).

Intra-Observer Error
Finally, rates of intra-observer error were calculated for all three methods in order to
measure the amount of precision, or the closeness of the repeated estimations. For this test, the
author reevaluated a random selection of 105 individuals from the entire sample. Data was
statistically analyzed for each method using Spearman’s coefficient of rank correlation and
Cohen’s Kappa measure of agreement, as both tests describe the relationship between two sets of
data, with 1.0 indicating a perfect correlation and 0 indicating no relationship between the
variables (Hinkle et al., 1988). Additionally, Cohen’s Kappa is a measure of association and is
often used for determining rates of reliability and intra-observer error, as it specifically it
measures agreement by examining the expected frequencies within contingency tables and also
corrects for chance correlations (Sheskin, 2004).
	
  
	
  
	
  
	
  

47	
  

RESULTS
In order to test the accuracy of the İşcan et al. (1984a; 1985), the Hartnett-Fulginiti
(2007; 2010), and the Oettlé and Steyn (2000) sternal rib aging methods, each technique was
performed separately in order to avoid bias in the assignment of phase numbers. The entire
sample was first evaluated using the İşcan et al. (1984a; 1985) method, followed by the HartnettFulginiti (2007; 2010) method, and finally the Oettlé and Steyn (2000) method. All evaluations
were conducted through the use of the original published articles, which contained written
descriptions and photographs of characteristic age-related features found within each phase, with
the addition of rib casts for the İşcan et al. (1984a; 1985) method (İşcan and Loth, 1993).
Before statistical analysis was conducted, descriptive statistics were calculated;
consisting of the counts; mean, median, and mode ages; standard deviations; and age ranges for
African-Americans, European-Americans, and the total sample. Frequencies were also further
divided into subsamples based on sex and ancestry.

Overall Accuracy
The first aim of this research was to evaluate the İşcan et al. (1984a; 1985), HartnettFulginiti (2007; 2010), and Oettlé and Steyn (2000) methods for accuracy and how well each
method could predict age-at-death. The entire sample was utilized in order to calculate rates of
accuracy for all individuals of both European and African ancestry. It was hypothesized that the
Hartnett-Fulginiti (2007; 2010) method would be a more accurate assessor of age for the entire
sample, as the method was developed as a revision of the İşcan et al. (1984a; 1985) method in
order to improve accuracy.

48	
  

Correlation
In order to test the accuracy of the three sternal rib aging methods, Spearman’s
coefficient of rank correlation was utilized to make comparisons between the estimated and
actual phase numbers and the estimated and actual ages for all individuals within the sample. In
accordance with each method, a phase number corresponding to an age range was assigned to
each individual based on the observed morphological features. This estimated phase number was
then compared to the actual phase number, as determined by the individual’s actual age-at-death.
For the purposes of statistical analysis, additional comparisons between the estimated and actual
age were also made. This was conducted by converting the estimated phase number to the mean
age for that phase. For example, for the Hartnett-Fulginiti (2007; 2010) method, a Phase 1 for a
female has an estimated age range of 18 to 22 years and would therefore have an estimated age
of 19.57 years, which is the mean age for that phase.
The author examined a total of 411 right fourth sternal rib ends and scored them
according to the available casts and written descriptions from the original articles (İşcan and
Loth, 1993; İşcan et al., 1984a; 1985; Hartnett, 2007; 2010; Oettlé and Steyn, 2000). Table 9
displays the number of individuals, the correlation coefficients (rs), and the significance for all
three methods. When comparing the estimated and actual phase numbers the Hartnett-Fulginiti
method had the strongest coefficient (rs = 0.836), followed by İşcan et al. method (rs = 0.785),
and then Oettlé and Steyn method (rs = 0.751). Similar correlation coefficients were also found
when comparing the estimated and actual ages. The Hartnett-Fulginiti method displayed the
strongest coefficient (rs = 0.793), followed by İşcan et al. method (rs = 0.745), and then the
Oettlé and Steyn method (rs = 0.741). Although the Hartnett-Fulginiti (2007; 2010) method had

49	
  

the strongest correlations, all three techniques had strong associations (rs = 0.7–0.89) that were
highly significant (p < 0.0001).
Table 9: Correlation for the total sample
Method
İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

Count

Est vs. Actual Phase
Spearman’s r

411
411
411

0.780
0.832
0.746

Est vs. Actual Age
Spearman’s r
0.745
0.793
0.741

Significance
< 0.0001
< 0.0001
< 0.0001

Due to all three methods that displayed strong correlations, further analysis was required
to determine if there were any statistically significant differences in accuracy between the
methods. Statistical analysis of the correlation coefficients between the estimated and actual
phases were not possible, as for some individuals the actual phase numbers differ depending on
the method. Therefore comparisons could only made between the estimated and actual ages, as
the actual ages for each individual remain the same for every method. This analysis was
conducted through the use of a Steiger’s z transformation. This test allows for comparisons
between correlated samples by converting the r-scores to normally distributed z-scores, with a
score greater than |1.96| indicating statistical significance.
Table 10 demonstrates the test performed; the correlation coefficients (rs) of the first
method, the second method, and between the estimated ages of the first and second methods; the
z-scores; and the significance. When comparing the accuracy between the methods, two
statistically significant differences were found. The first was between the Hartnett-Fulginiti
(2007; 2010) method and the İşcan et al. (1984a; 1985) method (p = 0.022). The second was
between the Hartnett-Fulginiti (2007; 2010) method and the Oettlé and Steyn (2000) method (p =

50	
  

0.011). These results indicate that the Hartnett-Fulginiti (2007; 2010) method was the most
accurate at assessing age for all individuals within the sample.
Table 10: Significance of the correlation coefficients for the total sample
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.745
0.745
0.741

0.793
0.741
0.793

0.725
0.738
0.743

- 2.294
0.182
- 2.539

0.022*
0.855
0.011*

* p < 0.05; 2-tailed test, n = 411
Additional tests of the methods were conducted using a binomial distribution, which
determines how well each method can predict age. This is based on the amount of correct phase
classifications, defined as successes, and the number of incorrect classifications, defined as
failures. Due to the large sample size, a sample proportion was used, known as the normal
approximation of the binomial distribution. Ninety-five percent confidence intervals were
calculated and compared for each method, with non-overlapping confidence intervals indicating
statistical significance.
Table 11 demonstrates the mean, the proportion of successes (^), standard error, variance,
p
and confidence interval of each method. The Harnett-Fulginiti (2007; 2010) method had the
strongest p-value (^ = 0.4842) with a confidence interval between 0.4359–0.5325. Comparisons
^
p
between the Hartnett-Fulginiti (2007; 2010) method and the İşcan et al. (1984a; 1985) method
showed overlapping confidence intervals, indicating non-significance between the methods. In
contrast, the Oettlé and Steyn (2000) method displayed a smaller and non-overlapping
confidence interval (0.3160–0.4090) when compared to the Hartnett-Fulginiti method (0.4359–
0.5325). This suggests that the Hartnett-Fulginiti (2007; 2010) method was more accurate at
assessing age than the Oettlé and Steyn (2000) method. Additionally, the Harnett-Fulginiti

51	
  

(2007; 2010) method had the strongest p-value (^ = 0.4842), suggesting that it was a better
^
p
predictor of age in comparison to the other techniques.
Table 11: Binomial distribution for the total sample
Method

Mean

^-value
p

Stand Error

Variance

Confidence Interval

İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

168.02
199.01
148.99

0.4088
0.4842
0.3625

0.0242
0.0261
0.0237

0.000588
0.000608
0.000562

0.3613–0.4838
0.4359–0.5325
0.3160–0.4090

n = 411
Cross-tabulation
Cross-tabulation tables, modeled after Hartnett (2007), were also used to visually
compare the estimated and actual phase numbers for each method. The darkest shaded diagonal
region of the table designates when the actual and estimated phase numbers matched, while the
lighter shaded regions indicates a difference of one phase either above or below the actual phase
number.
Table 12 demonstrates the cross-tabulation table for the İşcan et al. (1984a; 1985)
method, with 168 of 411 individuals (40.9%) that were assigned to the correct phase. An
additional 179 (43.6%) were assigned within one phase from the actual phase number, meaning
either one phase over (n = 69) or one phase under (n = 110). Finally, 52 individuals (12.7%)
were assigned within two phases over (n = 18) or under (n = 34) the actual phase number.

52	
  

Table 12: Cross-tabulation of estimated and actual İşcan et al. phases for the total sample
0

Estimated
Phase

0
1
2
3
4
5
6
7
8

1

2

3

0
0
0
0
0
0
0
0
0

0
1
1
1
0
0
0
0
0

0
4
21
9
3
0
0
0
0

0
0
9
20
8
3
0
0
0

Actual Phase
4
5
0
2
0
12
22
17
6
0
0

0
0
1
6
32
45
16
4
0

6

7

8

0
1
0
0
13
27
33
9
1

0
0
0
0
0
6
9
13
9

0
0
0
1
1
6
9
17
13

Table 13 demonstrates the cross-tabulation table for the Hartnett-Fulginiti (2007; 2010)
method, with 199 of 411 individuals (48.4%) that were assigned to the correct phase number. An
additional 169 individuals (41.1%) were estimated within one phase over (n = 115) or one phase
under (n = 54). Finally, 36 individuals (8.8%) were correctly assigned within two phases over (n
= 26) or under (n = 10) the correct phase number.
Table 13: Cross-tabulation of estimated and actual Hartnett-Fulginiti phases for the total
sample
Actual Phase
0
1
2
3
4
5
6
7
0
0
0
0
0
0
0
0
0
1
3
3
1
0
0
0
0
0
2
0
13
36
9
0
0
0
0
3
0
5
24
43
11
2
0
0
Estimated
4
0
1
4
30
35
9
3
0
Phase
5
0
1
2
10
25
46
16
5
6
0
0
0
3
6
13
19
8
7
0
0
0
0
0
1
7
17

53	
  

Table 14 demonstrates the cross-tabulation table for the Oettlé and Steyn (2000) method.
The author assigned a total of 149 of 411 individuals (36.3%) to the correct phase, while 180
individuals (43.8%) were assigned within one phase over (n = 81) or under (n = 99) of the actual
phase number. A total of 58 individuals (14.1%) were estimated within two phases over (n = 13)
or under (n = 45) the correct phase number.
Table 14: Cross-tabulation of estimated and actual Oettlé and Steyn phases for the total
sample
Actual Phase
0
1
2
3
4
5
6
7
8
0
0
0
0
0
0
0
0
0
0
1
0
10
8
2
1
0
0
0
0
2
0
3
10
1
1
2
0
0
0
3
0
3
18
24
16
6
2
1
0
Estimated
4
0
2
1
16
28
24
15
3
1
Phase
5
0
1
2
1
18
37
28
17
4
6
0
0
1
3
2
14
12
16
4
7
0
0
0
0
0
5
8
19
6
8
0
0
0
0
0
1
1
4
9

The first hypothesis was accepted, as results showed that the Hartnett-Fulginiti (2007;
2010) method was the most accurate at assessing age for the entire sample. This technique was
also a slightly better predictor of age (^ = 0.4842), had the strongest correlation coefficients (rs =
p
0.832; 0.793), and displayed statistically significant differences in accuracy in comparison to the
İşcan et al. (1984a; 1985) method and Oettlé and Steyn (2000) method.

Accuracy for European-Americans
The second research question addressed which method would more accurately predict
and assess the age of European-Americans within the sample. The Hartnett-Fulginiti (2007;

54	
  

2010) method was developed as a revision in order to improve accuracy rates for all individuals.
Additionally, the technique is the most contemporary and was developed on a sample consisting
of mostly Caucasian individuals. It was therefore hypothesized that the Hartnett-Fulginiti (2007;
2010) method would be more accurate for European-Americans compared to the İşcan et al.
(1984a; 1985) method and Oettlé and Steyn (2000) method.

Correlation
To determine the accuracy of the İşcan et al. (1984a; 1985), Hartnett-Fulginiti (2007;
2010), and Oettlé and Steyn (2000) methods on European-Americans within the sample,
comparisons of the actual and estimated phase numbers and actual and estimated ages were made
through the use of Spearman’s coefficient of rank correlation. Table 15 illustrates the number of
individuals, the correlation coefficients (rs), and the significance for all methods. For
comparisons between the estimated and actual phase numbers, the Hartnett-Fulginiti method
displayed the highest correlation coefficient (rs = 0.783), followed by İşcan et al. method (rs =
0.731), and then Oettlé and Steyn method (rs = 0.705). Similar results were obtained with
comparisons between the estimated and actual ages. The Hartnett-Fulginiti method displayed the
strongest correlation coefficient (rs = 0.731), followed by İşcan et al. method (rs = 0.716), and
then Oettlé and Steyn method (rs = 0.666). Although the Hartnett-Fulginiti had the strongest
correlations, all three techniques had strong associations (rs = 0.7–0.89) that were highly
significant (p < 0.0001).

55	
  

Table 15: Correlation for European-Americans
Count
İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

Est vs. Actual Phase
Spearman’s r

202
202
202

0.727
0.780
0.701

Est vs. Actual Age
Spearman’s r
0.716
0.731
0.666

Significance
< 0.0001
< 0.0001
< 0.0001

Further analysis was required to determine if there were any significant differences in
accuracy between the methods. This was conducted through the use of a Steiger’s z
transformation, as it allows for comparisons between related samples and includes an additional
correlation coefficient between the estimated ages of the two methods compared. Table 16
demonstrates the test performed; the correlation coefficients (rs) of the first method, the second
method, and between the estimated ages of the first and second methods; the z-scores; and the
significance for each test. There were no statistically significant values found between any of the
methods, although two almost significant results were obtained. The first was between the
Hartnett-Fulginiti (2007; 2010) method and İşcan et al. (1984a; 1985) method (p = 0.066). The
second was between the Hartnett-Fulginiti (2007; 2010) method and Oettlé and Steyn (2000)
method (p = 0.062). These results demonstrated that all three methods were able to assess age
equally well on European-Americans within the sample.
Table 16: Significance of the correlation coefficients for European-Americans
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.716
0.716
0.731

0.731
0.666
0.666

0.709
0.705
0.727

- 0.438
1.369
1.868

0.066
0.171
0.062

2-tailed test, n = 202

56	
  

Additional tests of accuracy of the methods for European-Americans were compared
using a binomial distribution. Ninety-five percent confidence intervals were calculated for each
method and were then compared to determine if any overlapping confidence intervals were
present, with non-overlapping intervals indicating significance. Table 17 demonstrates the mean,
the proportion of successes (^), standard error, variance, and confidence interval of each method.
p
As can be seen from the table, all confidence intervals overlapped, demonstrating no statistical
significance in accuracy between the methods, although the Hartnett-Fulginiti (2007; 2010)
method proved to be a slightly better predictor of age for European-Americans as it had the
highest p-value (^ = 0.4901).
^
p
Table 17: Binomial distribution for European-Americans
Method

Mean

^-value
p

Stand Error

İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

83.00
48.52
71.99

0.4109
0.4901
0.3564

0.0346
0.0346
0.0332

Variance Confidence Interval
0.0012
0.0012
0.0011

0.3430–0.4788
0.4222–0.5580
0.2904–0.4224

n = 202
Cross-tabulation
Cross-tabulation tables were also used in order to visually compare the estimated and
actual phase numbers. Table 18 demonstrates the cross-tabulation table for European-Americans
for the İşcan et al. (1984a; 1985) method, with 83 of 202 individuals (41.1%) assigned to the
correct phase. An additional 91 individuals (45.1%) were assigned within one phase from the
actual phase number, either one phase over (n = 34) or one phase under (n = 57). Finally, 21
individuals (10.4%) were assigned two phases over (n = 2) or under (n = 19) the actual phase
number.

57	
  

Table 18: Cross-tabulation of estimated and actual İşcan et al. phases for EuropeanAmericans
Actual Phase
1
2
3
4
5
6
7
8
1
0
1
0
0
0
1
0
0
2
0
6
2
1
0
0
0
0
3
0
3
5
4
1
0
0
1
Estimated
4
0
0
1
6
13
6
0
1
Phase
5
0
0
0
9
25
18
4
5
6
0
0
0
0
9
22
5
7
7
0
0
0
0
1
5
9
14
8
0
0
0
0
0
1
7
10

Table 19 demonstrates the cross-tabulation table for European-Americans for the
Hartnett-Fulginiti (2007; 2010) method. A total of 99 of 202 individuals (49.0%) were assigned
to the correct phase number, with another 79 individuals (39.1%) estimated either one phase over
(n = 52) or one phase under (n = 27) the correct phase. Finally, an additional 22 individuals
(10.9%) were assigned within two phases over (n = 15) or under (n = 7) the correct phase
number.
Table 19: Cross-tabulation of estimated and actual Hartnett-Fulginiti phases for EuropeanAmericans
Actual Phase
0
1
2
3
4
5
6
7
0
0
0
0
0
0
0
0
0
1
1
3
0
0
0
0
0
0
2
0
1
6
2
0
0
0
0
3
0
0
7
14
6
0
0
0
Estimated
4
0
0
1
13
20
5
3
0
Phase
5
0
0
0
7
12
26
9
4
6
0
0
0
2
6
11
17
5
7
0
0
0
0
0
1
7
13

58	
  

Table 20 demonstrates the cross-tabulation table for European-Americans for the Oettlé
and Steyn (2000) method. The author assigned 72 of 202 individuals (35.7%) to the correct
phase, while 90 individuals (44.6%) were assigned either one phase over (n = 35) or under (n =
55) of the actual phase. Finally, 30 individuals (14.9%) were estimated either two phases over (n
= 7) or under (n = 23) the correct phase number.
Table 20: Cross-tabulation of estimated and actual Oettlé and Steyn phases for EuropeanAmericans
Actual Phase
0
1
2
3
4
5
6
7
8
0
0
0
0
0
0
0
0
0
0
1
0
2
1
1
0
0
0
0
0
2
0
1
4
0
0
1
0
0
0
3
0
1
4
6
5
1
1
0
0
Estimated
4
0
0
0
4
12
15
7
1
1
Phase
5
0
0
0
0
11
18
16
12
4
6
0
0
0
1
1
5
8
13
2
7
0
0
0
0
0
4
6
17
5
8
0
0
0
0
0
1
1
4
5

Although the Hartnett-Fulginiti (2007; 2010) method had the highest correlation
coefficients (rs = 0.780; 0.731) of all three methods, no statistically significant differences in
accuracy were found between the three techniques. Therefore, results rejected the second
hypothesis, which predicted that the Hartnett-Fulginiti (2007; 2010) method would more
accurately assess age of European-Americans than the İşcan et al. (1984a; 1985) method and the
Oettlé and Steyn (2000) method. Instead it was found that all three methods performed equally
well at assessing age for this population, although the Hartnett-Fulginiti (2007; 2010) method
proved to be a slightly better predictor of age for European-Americans (^ = 0.4901).
p

59	
  

Accuracy for African-Americans
The third research question addressed if the Oettlé and Steyn (2000) method would more
accurately assess and predict the age of African-Americans in the sample. Although this
technique was developed on South Africans, it was hypothesized that this technique would be
more accurate for African-Americans than European-Americans, as it was developed on a
population of individuals of African ancestry. Testing the accuracy of the Oettlé and Steyn
(2000) method on an American population was also a main concern of this research, as if this
method proved to be reliable and accurate, it could possibly become a viable method within the
United States.

Correlation
In order to test the accuracy of the İşcan et al. (1984a; 1985), Hartnett-Fulginiti (2007;
2010), and Oettlé and Steyn (2000) methods for African-Americans, comparisons of the actual
and estimated phase numbers and estimated and actual ages were made using Spearman’s
coefficient of rank correlation. Table 21 illustrates the number of individuals, the correlation
coefficients (rs), and the significance for all methods. For comparisons between the actual and
estimated phase numbers, the Hartnett-Fulginiti method had the highest correlation coefficient (rs
= 0.790), followed by İşcan et al. method (rs = 0.769) and then Oettlé and Steyn method (rs =
0.704). Similar results were obtained when the estimated and actual ages were compared. The
Harnett-Fulginiti method displayed the strongest correlation coefficient (rs = 0.739), followed by
İşcan et al. method (rs = 0.704) and then Oettlé and Steyn method (rs = 0.692). For AfricanAmericans within the sample, the Hartnett-Fulginiti (2007; 2010) method had the strongest

60	
  

correlation coefficients, although all three techniques had strong (rs = 0.7–0.89) and highly
significant associations (p < 0.0001).
Table 21: Correlation for African-Americans
Count
İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

Est vs. Actual Phase
Spearman’s r

209
209
209

0.762
0.784
0.696

Est vs. Actual Age
Spearman’s r
0.704
0.739
0.692

Significance
< 0.0001
< 0.0001
< 0.0001

Further analysis was required to determine if the differences in accuracy between the
methods were significant. Therefore correlation coefficients between the actual and estimated
ages were compared through the use of a Steiger’s z transformation. Table 22 demonstrates the
test performed; the correlation coefficients of the first method, the second method, and between
the estimated ages of the first and second methods; the z-scores; and the significance. No
statistically significant results were found, indicating that all three methods were able to assess
age equally well for African-Americans within the sample.
Table 22: Significance of the correlation coefficients for African-Americans
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.704
0.704
0.739

0.739
0.692
0.692

0.662
0.697
0.641

- 0.978
0.334
1.268

0.328
0.738
0.205

2-tailed test, n = 209
Additional tests of accuracy of the methods for African-Americans were compared using
a binomial distribution. Ninety-five percent confidence intervals were calculated for each method
and were then compared to determine if any overlapping confidence intervals were present, with
non-overlapping intervals indicating significance. Table 23 demonstrates the mean, the
61	
  

proportion of successes (^), standard error, variance, and confidence interval of each method. As
p
can be seen from the table, all confidence intervals overlap, demonstrating no statistical
significance in accuracy between the methods, although the Hartnett-Fulginiti (2007; 2010)
method proved to be a slightly better predictor of age for European-Americans as it had the
strongest p-value (^ = 0.4737).
^
p
Table 23: Binomial distribution for European-Americans
Method
İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

Mean

^-value
p

Stand Error

Variance

Confidence Interval

85.00
99.00
77.00

0.4067
0.4737
0.3684

0.0340
0.0344
0.0332

0.0012
0.0012
0.0011

0.3727–0.4407
0.4390–0.5078
0.3352–0.4016

n = 209
Cross-tabulation
Cross-tabulation tables were also used in order to visually compare the estimated and
actual phase numbers. Table 24 demonstrates the cross-tabulation table for African-Americans
for the İşcan et al. (1984a; 1985) method. This table shows that 85 of 209 individuals (40.7%)
were assigned to the correct phase, with an additional 88 individuals (42.1%) assigned within
one phase from the actual phase number, meaning either one phase over (n = 35) or one phase
under (n = 53). Finally, 32 individuals (15.3%) were assigned within two phases over (n = 16) or
under (n = 16) the correct phase number.

62	
  

Table 24: Cross-tabulation of estimated and actual İşcan et al. phases for AfricanAmericans
Actual Phase
0
1
2
3
4
5
6
7
0
0
0
0
0
0
0
0
0
1
0
1
3
0
2
0
0
0
2
0
1
15
7
0
1
0
0
3
0
1
6
15
8
5
0
0
Estimated
4
0
0
3
7
16
19
7
0
Phase
5
0
0
0
3
8
20
9
2
6
0
0
0
0
6
7
11
4
7
0
0
0
0
0
3
4
4
8
0
0
0
0
0
0
0
2

8
0
0
0
0
0
1
2
3
3

Table 25 demonstrates the cross-tabulation table for African-Americans for the HartnettFulginiti (2007; 2010) method, with 99 of 209 individuals (47.4%) that were assigned to the
correct phase number. Another 91 individuals (43.5%) were assigned either one phase over (n =
63) or one phase under (n = 28) the correct phase. Finally, 14 individuals (6.7%) were correctly
assigned either two phases over (n = 11) or under (n = 3) the actual phase number.
Table 25: Cross-tabulation of estimated and actual Hartnett-Fulginiti phases for AfricanAmericans
Actual Phase
0
1
2
3
4
5
6
7
0
0
0
0
0
0
0
0
0
1
2
0
1
0
0
0
0
0
2
0
12
30
7
0
0
0
0
3
0
5
17
29
5
2
0
0
Estimated
4
0
1
3
17
14
5
0
0
Phase
5
0
1
2
3
13
20
7
1
6
0
0
0
1
0
2
2
3
7
0
0
0
0
0
0
0
4

Table 26 demonstrates the cross-tabulation table for African-Americans for the Oettlé
and Steyn (2000) method. A total of 77 of 209 individuals (36.8%) were assigned to the correct

63	
  

phase, while another 90 individuals (43.1%) were assigned within one phase over (n = 46) or
under (n = 44) the correct phase. Finally, 28 individuals (13.4%), age was assigned estimated
two phases over (n = 6) or under (n = 22) the actual phase number.
Table 26: Cross-tabulation of estimated and actual Oettlé and Steyn phases for AfricanAmericans
Actual Phase
0
1
2
3
4
5
6
7
8
0
0
0
0
0
0
0
0
0
0
1
0
8
7
1
1
0
0
0
0
2
0
2
6
1
1
1
0
0
0
3
0
2
14
18
11
5
1
1
0
Estimated
4
0
2
1
12
16
9
8
2
0
Phase
5
0
1
2
1
7
19
12
5
0
6
0
0
1
2
1
9
4
3
2
7
0
0
0
0
0
1
2
2
1
8
0
0
0
0
0
0
0
0
4

Results rejected the third hypothesis, which predicted that the Oettlé and Steyn (2000)
method would more accurately assess age of African-Americans than the Hartnett-Fulginiti
(2007; 2010) method and the İşcan et al. (1984a; 1985) method. Instead it was found that all
three methods were able to assess age equally well for African-Americans within the sample.
Although the Hartnett-Fulginiti (2007; 2010) method displayed the highest correlation
coefficients (rs = 0.784; 0.739), no statistically significant differences were found between any of
the methods. Additionally, the Hartnett-Fulginiti (2007; 2010) method was found to be a slightly
better predictor of age for African-Americans (^ = 0.4737).
p

64	
  

Accuracy Based on Ancestry
An additional aim of this research was to determine if there were any differences in
accuracy between African-Americans and European-Americans. This analysis was conducted
based on previous studies that had demonstrated ancestral based differences in the rate and
pattern of morphological age-related characteristics for African-Americans and South African
blacks (İşcan et al., 1987; Oettlé and Steyn, 2000). From these results, it was expected that there
would be statistically significant differences between the two ancestral groups. Analysis was
conducted by examining the previously calculated correlation coefficients for African-Americans
and European-Americans in order to determine if there were any statistically significant
differences in accuracy between the groups. If major differences are found, it may require the use
of separate standards on the basis of ancestry.

Correlation (İşcan et al. Method)
Correlation coefficients were calculated for three categories for the İşcan et al. (1984a;
1985) method, consisting of the entire sample, African-Americans, and European-Americans.
Table 27 demonstrates the number of individuals, the correlation coefficients (rs), and the
significance for all groups for the İşcan et al. (1984a; 1985) method. For comparisons between
the estimated and actual phase numbers, the entire sample had the highest correlation coefficient
(rs = 0.780), followed by European-Americans (rs = 0.762), and then African-Americans (rs =
0.727). Similar results were obtained from comparisons between the estimated and actual ages.
The entire sample displayed the highest correlation coefficient (rs = 0.745), followed by
European-Americans (rs = 0.716), and then African-Americans (rs = 0.704). The entire sample

65	
  

had the highest correlation coefficients for the İşcan et al. (1984a; 1985) method, although all
categories had strong (rs = 0.7–0.89) and highly significant associations (p < 0.0001).

Table 27: Correlation based on ancestry for the İşcan et al. method
Count
Entire Sample
African-Americans
European-Americans

Est vs. Actual Phase
Spearman’s r

411
209
202

0.780
0.727
0.762

Est vs. Actual Age
Spearman’s r
0.745
0.704
0.716

Significance
< 0.0001
< 0.0001
< 0.0001

Due to correlation coefficients that were all strong and highly significant, further analysis
was required to determine if there were statistically significant differences in accuracy between
the methods. This was conducted through a Fisher’s z transformation, as all comparisons were
made within the same method. This test allows for the comparison of two correlation coefficients
by converting the r-values to normally distributed z-scores, with values that are greater than
|1.96| indicating statistical significance. Table 28 demonstrates the correlation coefficients
between the estimated and actual phase numbers for the first category and the second category,
the z-scores, and the significance for each test. No significant values between ancestries were
found for the İşcan et al. (1984a; 1985) method.
Table 28: Significance of the correlation coefficients for the İşcan et al. method based on
ancestry (phase)
Test

r(1)

z-score

Significance

0.780
0.780
0.727

Entire Sample
/ African-Americans
Entire Sample
/ European-Americans
African-Americans / European-Americans

r(2)
0.727
0.762
0.762

1.439
0.514
- 0.791

0.150
0.608
0.429

2-tailed test, entire sample (n = 411), African-Americans (n = 209), European-Americans (n
= 202)

66	
  

Additional comparisons of the correlation coefficients between the estimated and actual
ages were also made. Table 29 demonstrates the correlation coefficients between the estimated
and actual ages for the first category and the second category, the z-scores, and the significance
for each test. Again, no significant values were found between ancestries for the İşcan et al.
(1984a; 1985) method, indicating that the technique was able to assess age equally well for both
ancestral groups.
Table 29: Significance of the correlation coefficients for the İşcan et al. method based on
ancestry (age)
Test

r(1)

z-score

Significance

0.745
0.745
0.704

Entire Sample
/ African-Americans
Entire Sample
/ European-Americans
African-Americans / European-Americans

r(2)
0.704
0.716
0.716

1.011
0.720
- 2.435

0.312
0.472
0.808

2-tailed test, entire sample (n = 411), African-Americans (n = 209), European-Americans (n
= 202)
No statistically significant values were found for any of the comparisons for the İşcan et
al. (1984a; 1985) method. This demonstrates that the method was able to assess age equally well
for both African-Americans and European-Americans.

Correlation (Hartnett-Fulginiti Method)
Correlation coefficients for the Hartnett-Fulginiti (2007; 2010) method were calculated
for three categories, consisting of the entire sample, African-Americans, and EuropeanAmericans. Table 30 demonstrates the number of individuals, the correlation coefficients (rs),
and the significance for all groups for the Hartnett-Fulginiti (2007; 2010) method. For
comparisons between the estimated and actual phase numbers, the entire sample had the highest
correlation coefficient (rs = 0.832), followed by African-Americans (rs = 0.784), and then
67	
  

European-Americans (rs = 0.780). Similar results were obtained from comparisons between the
estimated and actual ages. The entire sample displayed the highest correlation coefficient (rs =
0.793), followed by African-Americans (rs = 0.739), and then European-Americans (rs = 0.731).
The entire sample has the strongest correlation coefficients for the Hartnett-Fulginiti (2007;
2010) method, although all categories had strong (rs = 0.7–0.89) and highly significant
associations (p < 0.0001).
Table 30: Correlation based on ancestry for the Hartnett-Fulginiti method
Count
Entire Sample
African-Americans
European-Americans

Est vs. Actual Phase
Spearman’s r

411
209
202

0.832
0.784
0.780

Est vs. Actual Age
Spearman’s r
0.793
0.739
0.731

Significance
< 0.0001
< 0.0001
< 0.0001

Due to all correlations that were strong and highly significant, a Fisher’s z transformation
was utilized in order to determine if there were any statistically significant differences in
accuracy between the groups. Table 31 demonstrates the correlation coefficients between the
estimated and actual phase numbers for the first category and the second category, the z-scores,
and the significance for each test. No statistically significant values were found between the
groups, indicating that the technique was able to assess age equally well for both ancestral
groups.

68	
  

Table 31: Significance of the correlation coefficients for the Hartnett-Fulginiti method
based on ancestry (phase)
Test

r(1)

z-score

Significance

0.832
0.832
0.784

Entire Sample
/ African-Americans
Entire Sample
/ European-Americans
African-Americans / European-Americans

r(2)
0.784
0.780
0.780

1.625
1.726
0.104

0.104
0.084
0.917

2-tailed test, entire sample (n = 411), African-Americans (n = 209), European-Americans (n
= 202)
Correlation coefficients between the estimated and actual ages were also compared. Table
32 demonstrates the correlation coefficients between the estimated and actual ages for the first
category and the second category, the z-scores, and the significance for each test. Again, no
significant values between ancestries were found for the Hartnett-Fulginiti (2007; 2010) method,
indicating that the technique was able to assess age equally well for both ancestral groups.
Table 32: Significance of the correlation coefficients for the Hartnett-Fulginiti method
based on ancestry (age)
Test

r(1)

z-score

Significance

0.793
0.793
0.739

Entire Sample
/ African-Americans
Entire Sample
/ European-Americans
African-Americans / European-Americans

r(2)
0.739
0.731
0.731

1.535
1.719
0.175

0.125
0.086
0.861

2-tailed test, entire sample (n = 411), African-Americans (n = 209), European-Americans (n
= 202)
No statistically significant values were found for any of the comparisons for the HartnettFulginiti (2007; 2010) method. This demonstrates that the method was able to assess age equally
well for both African-Americans and European-Americans.

69	
  

Correlation (Oettlé and Steyn Method)
Correlation coefficients were calculated for three categories for the Oettlé and Steyn
(2000) method, consisting of the entire sample, African-Americans, and European-Americans.
Table 33 demonstrates the number of individuals, the correlation coefficients (rs), and the
significance for all groups for the Oettlé and Steyn (2000) method. For comparisons between the
estimated and actual phase numbers, the entire sample had the highest correlation coefficient (rs
= 0.746), followed by European-Americans (rs = 0.701), and then African-Americans (rs =
0.696). Similar results were obtained from comparisons between the estimated and actual ages.
The entire sample displayed the highest correlation coefficient (rs = 0.741), followed by
European-Americans (rs = 0.692), and then African-Americans (rs = 0.666). The entire sample
contained the strongest correlation coefficients, although all categories had strong (rs = 0.7–0.89)
and highly significant associations (p < 0.0001).
Table 33: Correlation based on ancestry for the Oettlé and Steyn method
Count
Entire Sample
African-Americans
European-Americans

Est vs. Actual Phase
Spearman’s r

411
209
202

0.746
0.696
0.701

Est vs. Actual Age
Spearman’s r
0.741
0.692
0.666

Significance
< 0.0001
< 0.0001
< 0.0001

All three groups displayed correlation coefficients that were all strong and highly
significant. Therefore in order to determine if there was a difference in accuracy between the
groups, a Fisher’s z transformation was utilized. Table 34 demonstrates the correlation
coefficients between the estimated and actual phase numbers for the first category and the second

70	
  

category, the z-scores, and the significance for each method. No significant values were found,
indicating that the Oettlé and Steyn (2000) method was able to assess age equally well for both
ancestral groups.
Table 34: Significance of the correlation coefficients for the Oettlé and Steyn method based
on ancestry (phase)
Test

r(1)

z-score

Significance

0.746
0.746
0.696

Entire Sample
/ African-Americans
Entire Sample
/ European-Americans
African-Americans / European-Americans

r(2)
0.696
0.701
0.701

1.221
1.094
- 0.098

0.222
0.274
0.922

2-tailed test, entire sample (n = 411), African-Americans (n = 209), European-Americans (n
= 202)
Correlation coefficients between the estimated and actual ages were also compared. Table
35 demonstrates the correlation coefficients between the estimated and actual ages for the first
category and the second category, the z-scores, and the significance for each test. Again, no
significant values between ancestries were found for the Oettlé and Steyn (2000) method,
indicating that the method was able to assess age equally well for both ancestral groups.
Table 35: Significance of the correlation coefficients for the Oettlé and Steyn method based
on ancestry (age)
Test

r(1)

z-score

Significance

0.741
0.741
0.692

Entire Sample
/ African-Americans
Entire Sample
/ European-Americans
African-Americans / European-Americans

r(2)
0.692
0.666
0.666

1.167
1.745
0.485

0.243
0.081
0.627

2-tailed test, entire sample (n = 411), African-Americans (n = 209), European-Americans (n
= 202)
No statistically significant values were found for any of the comparisons for the HartnettFulginiti (2007; 2010) method. This demonstrates that the method was able to assess age equally
well for both African-Americans and European-Americans within the sample.

71	
  

For all three methods, no statistically significant differences were found between AfricanAmericans and European-Americans, demonstrating that each technique was able to assess age
equally well for both ancestral groups. These results therefore reject the hypothesis that there
would be significant differences between African-Americans and European-Americans within
the sample. This suggests that although differences in the rate and pattern of aging may be
present, separate standards for African-Americans and European-Americans may not be
necessary.

Accuracy Based on Sex
The fifth research question addressed the rates of accuracy between males and females
for each method. Each method contains a separate set of standards for males and females due to
differences in degenerative changes of the fourth sternal right rib that were noticed. It was
therefore hypothesized that would be differences in accuracy for each sex, especially when
divided on the basis of ancestry.

Correlation (İşcan et al. Method)
Spearman’s coefficient of rank was used to calculate accuracy rates for all categories,
including African-American males, European-American males, total males, African-American
females, European-American females, and total females. Table 36 shows the correlation of the
estimated and actual phase numbers and estimated and actual ages for the İşcan et al. (1984a;
1985) method. African-American males had the highest correlation (rs = 0.835; 0.823) and
African-American females had the lowest (rs = 0.617; 0.571). All categories, except AfricanAmerican females, had strong (rs = 0.7–0.89) and highly significant correlations (p < 0.0001),
72	
  

although the moderate correlation for African-American females may have been due to small
sample size.
Table 36: Correlation based on sex and ancestry for the İşcan et al. method
Count

Est vs. Act Phase Est vs. Act Age
Spearman’s r
Spearman’s r

Significance

African-American Males
European-American Males
Total Males

143
149
292

0.835
0.702
0.800

0.823
0.681
0.788

< 0.0001
< 0.0001
< 0.0001

African-American Females
European-American Females
Total Females

66
53
119

0.617
0.795
0.755

0.571
0.792
0.735

< 0.0001
< 0.0001
< 0.0001

Total Sample

411

0.780

0.745

< 0.0001

To determine if the differences between the correlation coefficients for each category
were significant for the İşcan et al. (1984a; 1985) method, a Fisher’s z transformation was used,
as comparisons were made within the same method. Table 37 shows the tests, correlation
coefficients between the estimated and actual phase numbers for the first and second categories,
z-scores, and significance for all comparisons for the İşcan et al. (1984a; 1985) method. No
statistically significant values were found, indicating that the İşcan et al. (1984a; 1985) method
was able to assess age equally well for both males and females within the sample.
Table 37: Significance of the İşcan et al. method based on sex (phase)
Test

r(1)

r(2)

z-score

Significance

Entire Sample / Total Males
Entire Sample / Total Females
Total Males
/ Total Females

0.780
0.780
0.800

0.800
0.755
0.755

- 0.692
0.579
1.038

0.489
0.563
0.299

2-tailed test, entire sample (n = 411), total males (n = 292), total females (n = 119)

73	
  

Correlation coefficients between the estimated and actual ages were also compared. Table
38 demonstrates the correlation coefficients between the estimated and actual ages for the first
second categories, the z-scores, and the significance for each test. Again, no significant values
between ancestries were found for the İşcan et al. (1984a; 1985) method, indicating that the
technique was able to assess age equally well for males and females.
Table 38: Significance of the İşcan et al. method based on sex (age)
Test

r(1)

r(2)

z-score

Significance

Entire Sample / Total Males
Entire Sample / Total Females
Total Males
/ Total Females

0.745
0.745
0.788

0.788
0.735
0.735

- 1.259
0.210
1.152

0.174
0.834
0.249

2-tailed test, entire sample (n = 411), total males (n = 292), total females (n = 119)
No statistically significant values were found for any of the comparisons for the İşcan et
al. (1984a; 1985) method. This indicates that the method was able to assess age equally well for
both males and females within the sample.

Correlation (Hartnett-Fulginiti Method)
Spearman’s coefficient of rank was used to calculate accuracy rates for all categories for
the Hartnett-Fulginiti (2007; 2010) method. Table 39 shows the correlation coefficients between
the estimated and actual phase numbers and the estimated and actual ages. European-American
females had the highest correlations (rs = 0.854; 0.844) and African-American females had the
lowest (rs = 0.645; 0.672). All categories, except African-American females, had strong (rs =
0.7–0.89) and highly significant correlations (p < 0.0001), although the moderate correlation for
African-American females may have been due to small sample size.

74	
  

Table 39: Correlation based on sex and ancestry for the Hartnett-Fulginiti method
Count

Est vs. Act Phase Est vs. Act Age
Spearman’s r
Spearman’s r

Significance

African-American Males
European-American Males
Total Males

143
149
292

0.829
0.747
0.825

0.777
0.705
0.783

< 0.0001
< 0.0001
< 0.0001

African-American Females
European-American Females
Total Females

66
53
119

0.645
0.854
0.833

0.672
0.844
0.844

< 0.0001
< 0.0001
< 0.0001

Total Sample

411

0.832

0.793

< 0.0001

In order to determine if there were any statistically significant differences in accuracy
between each category for the Hartnett-Fulginiti (2007; 2010) method, a Fisher’s z
transformation was used, as the comparisons were made within the same method. Table 40
shows the tests, correlations coefficients between the estimated and actual phase numbers for the
first and second categories, z-scores, and significance for all comparisons of the HartnettFulginiti (2007; 2010) method. No statistically significant values were found, indicating that the
Hartnett-Fulginiti (2007; 2010) method was able to assess age equally well for males and
females.
Table 40: Significance of the Hartnett-Fulginiti method based on sex (phase)
Test

r(1)

r(2)

z-score

Significance

Entire Sample / Total Males
Entire Sample / Total Females
Total Males
/ Total Females

0.832
0.832
0.825

0.825
0.833
0.833

0.290
- 0.031
- 0.233

0.772
0.975
0.816

2-tailed test, entire sample (n = 411), total males (n = 292), total females (n = 119)
Correlation coefficients between the estimated and actual ages were also compared. Table
41 demonstrates the correlation coefficients between the estimated and actual ages for the first

75	
  

category and the second category, the z-scores, and the significance for each test. Again, no
significant values between ancestries were found for the Hartnett-Fulginiti (2007; 2010) method,
demonstrating that the method was able to assess age equally well for males and females.
Table 41: Significance of the Hartnett-Fulginiti method based on sex (age)
Test

r(1)

r(2)

z-score

Significance

Entire Sample / Total Males
Entire Sample / Total Females
Total Males
/ Total Females

0.793
0.793
0.783

0.783
0.844
0.844

0.343
- 1.477
- 1.654

0.731
0.140
0.098

2-tailed test, entire sample (n = 411), total males (n = 292), total females (n = 119)
No statistically significant values were found for any of the comparisons for the HartnettFulginiti (2007; 2010) method. This demonstrates that the method was able to assess age equally
well for both males and females within the sample.

Correlation (Oettlé and Steyn Method)
Spearman’s coefficient of rank was used to calculate accuracy rates for all categories for
the Oettlé and Steyn (2000) method. Table 42 shows the correlation coefficients between the
estimated and actual phase numbers and estimated and actual ages for the Oettlé and Steyn
(2000) method. Total males had the highest correlations for both comparisons (rs = 0.762;
0.763). African-American females had the lowest for comparisons between phases (rs = 0.598),
and European-American females had the lowest for comparisons between ages (rs = 0.533). All
categories, except African-American females and European-American females, had strong (rs =
0.7–0.89) and highly significant correlations (p < 0.0001). Again, these moderate and low
correlations may have been due to small sample size.

76	
  

Table 42: Correlation based on sex and ancestry for the Oettlé and Steyn method
Count

Est vs. Act Phase Est vs. Act Age
Spearman’s r
Spearman’s r

Significance

African-American Males
European-American Males
Total Males

143
149
292

0.741
0.708
0.762

0.737
0.704
0.763

< 0.0001
< 0.0001
< 0.0001

African-American Females
European-American Females
Total Females

66
53
119

0.598
0.668
0.755

0.583
0.533
0.671

< 0.0001
< 0.0001
< 0.0001

Total Sample

411

0.746

0.741

< 0.0001

Due to similar correlation coefficients for many categories, further tests were required in
order to determine if there are any statistically significant differences in accuracy for each
category for the Oettlé and Steyn (2000) method. This was conducted through the use of a
Fisher’s z transformation, as comparisons were made within the same method. Table 43 shows
the tests, the correlation coefficients between the estimated and actual phase numbers for the first
and second categories, z-scores, and significance for all comparisons of the Oettlé and Steyn
(2000) method. No statistically significant values were found, indicating that the Oettlé and
Steyn (2000) method was able to assess age equally well for males and females within the
sample.
Table 43: Significance of the Oettlé and Steyn method based on sex (phase)
Test

r(1)

r(2)

z-score

Significance

Entire Sample / Total Males
Entire Sample / Total Females
Total Males
/ Total Females

0.746
0.746
0.762

0.762
0.755
0.755

- 0.482
- 0.196
0.145

0.629
0.845
0.881

2-tailed test, entire sample (n = 411), total males (n = 292), total females (n = 119)

77	
  

Correlation coefficients between the estimated and actual ages were also compared. Table
44 demonstrates the correlation coefficients between the estimated and actual ages for the first
category and the second category, the z-scores, and the significance for each test. Again, no
significant values between ancestries were found for the Oettlé and Steyn (2000) method,
demonstrating that the technique was able to assess age equally well for males and females.
Table 44: Significance of the Oettlé and Steyn method based on sex (age)
Test

r(1)

r(2)

z-score

Significance

Entire Sample / Total Males
Entire Sample / Total Females
Total Males
/ Total Females

0.741
0.741
0.763

0.763
0.671
0.671

- 0.659
1.332
1.736

0.560
0.183
0.083

2-tailed test, entire sample (n = 411), total males (n = 292), total females (n = 119)
No statistically significant values were found for any of the comparisons for the HartnettOettlé and Steyn (2000) method. This demonstrates that the method was able to assess age
equally well for both males and females.
The fifth research hypothesis addressed differences in accuracy based on sex. It was
expected that there would be significant differences between each sex for every method. No
significant differences between males and females were found for any method, rejecting this
hypothesis. This suggests that separate standards based on sex are necessary and that this
separation does not appear impact accuracy rates for either sex.

Correlation (All Methods)
Comparisons of accuracy based on sex were also conducted between methods using the
categories of males and females conducted through the use of a Steiger’s z transformation. Table
45 demonstrates the correlation coefficients for males of the first and second methods, the

78	
  

correlation coefficient of the estimated ages between the first and second methods, the z-scores,
and the significance. When examining the differences in significance of accuracy rates between
males for all three methods, no statistically significant differences were found, indicating that all
methods were able to assess age equally well for males within the sample.
Table 45: Significance of the correlation coefficients for males
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.788
0.788
0.783

0.783
0.763
0.763

0.785
0.815
0.784

0.230
1.195
0.891

0.818
0.232
0.373

2-tailed test, n = 292
Comparing the correlations of all three methods for females was also conducted using a
Steiger’s z transformation. Table 46 demonstrates the correlation coefficients of the first and
second methods, the correlation coefficient of the estimated ages between the first and second
methods, the z-scores, and the significance. Two statistically significant differences were found.
The first was between the Hartnett-Fulginiti (2007; 2010) method and the İşcan et al. (1984a;
1985) method (p = 0.004). The second was between the Hartnett-Fulginiti (2007; 2010) method
and the Oettlé and Steyn (2000) method (p = 0.000). This demonstrates that the HartnettFulginiti (2007; 2010) method was able to more accurately assess age for females in the sample
in comparison to the other two methods.
Table 46: Significance of the correlation coefficients for females
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.735
0.735
0.844

0.844
0.671
0.671

0.678
0.620
0.658

- 2.847
1.237
4.118

0.004*
0.216
0.000*

* p < 0.05; 2-tailed test, n = 119

79	
  

Sex was further assessed on the basis of ancestry. The correlation coefficients for
African-American males between methods were first compared for each method using a
Steiger’s z transformation. Table 47 demonstrates the correlation coefficients of the first and
second methods, the correlation coefficient of the estimated ages between the first and second
methods, the z-scores, and the significance. Statistically significant differences were found
between the İşcan et al. (1984a; 1985) method and the Oettlé and Steyn (2000) method (p =
0.006). This indicates that the İşcan et al. (1984a; 1985) method was able to assess age more
accurately than the Oettlé and Steyn (2000) method for African-American males.
Table 47: Significance of the correlation coefficients for African-American males
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.823
0.823
0.777

0.777
0.737
0.737

0.772
0.780
0.712

1.512
2.736
1.065

0.130
0.006*
0.287

* p < 0.05; 2-tailed test, n = 143
Comparisons of European-American males between methods were also made using a
Steiger’s z transformation. Table 48 demonstrates the correlation coefficients of the first and
second methods, the correlation coefficient of the estimated ages between the first and second
methods, the z-scores, and the significance. No statistically significant differences were found
between methods, indicating that all three methods were able to assess age equally well for
European-American males.

80	
  

Table 48: Significance of the correlation coefficients for European-American males
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.681
0.681
0.705

0.705
0.704
0.704

0.733
0.800
0.767

- 0.588
- 0.640
0.027

0.556
0.522
0.979

2-tailed test, n = 149
African-American females were also compared between methods using a Steiger’s z
transformation. Table 49 demonstrates the correlation coefficients of the first and second
methods, the correlation coefficient of the estimated ages between the first and second methods,
the z-scores, and the significance. No statistically significant differences were found for AfricanAmerican females between the methods, indicating that all three methods were able to assess age
equally well for African-American females.
Table 49: Significance of the correlation coefficients for African-American females
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.571
0.571
0.672

0.672
0.583
0.583

0.465
0.599
0.521

- 1.091
- 0.137
1.014

0.275
0.891
0.311

2-tailed test, n = 66
Finally, European-American females were also compared between methods using a
Steiger’s z transformation. Table 50 demonstrates the correlation coefficients of the first and
second methods, the correlation coefficient of the estimated ages between the first and second
methods, the z-scores, and the significance. Two statistically significant values were found. The
first was between the Hartnett-Fulginiti (2007; 2010) method and the Oettlé and Steyn (2000)
method (p = 0.000). The second was between the İşcan et al. (1984a; 1985) method and Oettlé
and Steyn (2000) method (p = 0.007). This indicates that for European-American females, the

81	
  

Hartnett-Fulginiti (2007; 2010) method and İşcan et al. (1984a; 1985) method were more
accurate assessors of age than the Oettlé and Steyn (2000) method.
Table 50: Significance of the correlation coefficients for European-American females
Test
İşcan et al.
/ Hartnett-Fulginiti
İşcan et al.
/ Oettlé and Steyn
Hartnett-Fulginiti / Oettlé and Steyn

r(1)

r(2)

r(3)

z-score

Significance

0.792
0.792
0.844

0.844
0.533
0.533

0.726
0.422
0.623

- 1.022
2.716
4.245

0.306
0.007*
0.000*

* p < 0.05; 2-tailed test, n = 53
It was found that for African-American males, the İşcan et al. (1984a; 1985) method
more accurately assessed age than the Oettlé and Steyn (2000) method. No statistically
significant differences were found between methods for European-American males and AfricanAmerican females. For European-American females it was found that the Hartnett-Fulginiti
(2007; 2010) method and the İşcan et al. (1984a; 1985) method were both more accurate
assessors of age than the Oettlé and Steyn (2000) method.

Accuracy for All Categories
Table 51 displays the number of individuals and correlation coefficients for each category
of each method. The values in bold indicate the methods that were statistically more accurate for
each category. As can be seen from the table, the Hartnett-Fulginiti (2007; 2010) method most
accurately assessed age for almost all categories.

82	
  

Table 51: Correlation of estimated and actual ages for the entire sample
Count

İşcan et al.

143
66
209

0.823*
0.571
0.704

0.777
0.672
0.739

0.737
0.583
0.692

European-American Males
149
European-American Females 53
Total European-Americans
202

0.681
0.792
0.716

0.705
0.844
0.731

0.704
0.533
0.666

Total Males
Total Females
Total Sample

0.788
0.735
0.745

0.783
0.844
0.793

0.763
0.671
0.741

African-American Males
African-American Females
Total African-Americans

292
119
411

Hartnett-Fulginiti Oettlé and Steyn

* Only significant between the İşcan et al. and Oettlé and Steyn methods
Inaccuracy
Inaccuracy was calculated for all groups and all methods in order to further demonstrate
the accuracy of all three methods. Inaccuracy, also known as the absolute mean error, was
calculated for multiple categories and for every method. This was conducted through the use of
the equation Σ |estimated phase - actual phase| / N, which measures the differences between the
estimated and actual phases. It represents the average error from the actual phase number without
regard to whether the error is positive or negative (Russell et al., 1993; Martrille et al., 2007).
Table 52 displays the number of individuals and phase inaccuracy of each method for
multiple categories. The least amount of inaccuracy is in bold for each subsample. The HartnettFulginiti (2007; 2010) contained the least amount of inaccuracy for each category, while the
Oettlé and Steyn (2000) method displayed the most amount of inaccuracy.

83	
  

Table 52: Phase inaccuracy* for all methods
Count

İşcan et al.

143
66
209

0.748
0.864
0.785

0.573
0.788
0.641

0.916
0.879
0.914

European-American Males
149
European-American Females 53
Total European-Americans
202

0.859
0.604
0.792

0.664
0.566
0.639

0.946
0.755
0.901

Total Males
Total Females
Total Sample

0.805
0.748
0.788

0.620
0.689
0.640

0.938
0.824
0.905

African-American Males
African-American Females
Total African-Americans

292
119
411

Hartnett-Fulginiti Oettlé and Steyn

* Inaccuracy (Absolute Mean Error): Σ |estimated phase - actual phase| / N
Figure 2 shows a visual representation of this data. All three methods contained an
inaccuracy of less than one phase from the actual phase number, indicating that most methods
accurately assessed age within one phase.
Figure 2: Phase inaccuracy for all methods

Phase Inaccuracy

İşcan and Loth

Hartnett-Fulginiti

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

84	
  

Oettlé and Steyn

Inaccuracy was also calculated in terms of the age by specifically examining error in
terms of years. Age inaccuracy is determined through the equation Σ |mean age - actual age| / N
and is evaluated by the sums of the absolute values of the difference between the mean ages of
the estimated phases and the chronological, or actual, ages, divided by the number of individuals.
In order to calculate inaccuracy in regards to age, the sample was divided by method and
into multiple subsamples based on ancestry and sex. Table 53 displays the number of individuals
and the age inaccuracy for all three methods. The least amount of inaccuracy is in bold for each
subsample. Rates of inaccuracy ranged from 6.74 to 9.37 years for the İşcan et al. (1984a; 1985)
method, with the lowest rates for African-American males. Rates of inaccuracy for the HartnettFulginiti (2007; 2010) method ranged from 6.87 to 9.54 years, with the lowest rates for AfricanAmerican females. Finally, rates of inaccuracy ranged from 7.52 to 14.55 years for the Oettlé
and Steyn (2000) method, with the lowest rates for African-American males.
Table 53: Age inaccuracy* for all methods
Count

İşcan et al.

143
66
209

6.74
9.33
7.56

7.20
6.87
7.12

7.52
7.61
7.55

European-American Males
149
European-American Females 53
Total European-Americans
202

9.37
9.16
9.31

9.54
8.45
9.25

9.49
14.55
10.82

Total Males
Total Females
Total Sample

8.08
9.25
8.42

8.39
7.57
8.16

8.53
10.70
9.16

African-American Males
African-American Females
Total African-Americans

292
119
411

Hartnett-Fulginiti Oettlé and Steyn

* Inaccuracy (Absolute Mean Error): Σ |mean age - actual age| / N

85	
  

For the purposes of comparison, inaccuracy was plotted for the three methods for all
subsamples and is displayed in Figure 3. For the total sample, African-Americans, and EuropeanAmericans, the Hartnett-Fulginiti (2007; 2010) method had the lowest amount of inaccuracy.
Figure 3: Age inaccuracy for all methods

Age Inaccuracy

İşcan and Loth

Hartnett-Fulginiti

Oettlé and Steyn

16
14
12
10
8
6
4
2
0

Bias
Bias was also calculated for the entire sample for each method. While inaccuracy
measures the difference between the estimation and actual phases, bias instead represents the
tendency to either overage or underage an individual. It is calculated from the difference between
the estimated and actual phase numbers and is calculated through the equation Σ (estimated
phase - actual phase) / N. Phase bias demonstrates the average error between the estimated and
actual phase.

86	
  

Table 54 demonstrates the number of individuals and the phase bias of each method for
multiple categories. The numbers in bold represent the lowest amount of bias for each category
in terms of phase, with all methods that over or underestimated each category by less than one
phase.
Table 54: Phase bias* for all methods
Count

İşcan et al.

143
66
209

- 0.301
0.197
- 0.144

0.266
0.455
0.325

- 0.259
0.212
- 0.110

European-American Males
149
European-American Females 53
Total European-Americans
202

- 0.456
- 0.302
- 0.416

0.262
0.151
0.233

- 0.315
- 0.453
- 0.351

Total Males
Total Females
Total Sample

- 0.380
- 0.025
- 0.277

0.264
0.319
0.280

- 0.288
- 0.084
- 0.229

African-American Males
African-American Females
Total African-Americans

292
119
411

Hartnett-Fulginiti Oettlé and Steyn

* Bias: Σ (estimated phase - actual phase) / N
Figure 4 gives a visual representation of this data. All three methods contained a bias of
less than one phase than the actual phase, indicating all methods accurately assessed age within
one phase.

87	
  

Figure 4: Phase bias for all methods
İşcan and Loth

Hartnett-Fulginiti

Oettlé and Steyn

0.6

Phase Bias

0.4
0.2
0
-0.2
-0.4
-0.6

Bias on the basis of age was also calculated using the equation Σ (estimated age - actual
age) / N. This demonstrates the average error in age between the estimated and actual age with
regards to the direction of the error. In order to determine for inaccuracy and bias in terms of age,
mean ages of each phase were used in substitute for an estimated age, as each phase contains an
age range. For example, for the Hartnett-Fulginiti (2007; 2010) method, if a female is estimated
to be a Phase 1, that individual has an estimated age range of 18 to 22 years; therefore she is
approximately 19.57 years, which is the mean age for that phase (Russell et al., 1993; Martrille et
al., 2007).
Table 55 demonstrates the number of individuals and the age bias of each method for
multiple categories. For the İşcan et al. (1984a; 1985) method bias ranged from 3.93 years under
to 2.03 years over the actual ages, with the lowest amount for total females. Bias for the Hartnett88	
  

Fulginiti (2007; 2010) method ranged from 1.27 to 3.31 years over the actual ages, with the
lowest amount of bias for African-American males. Finally, bias for the Oettlé and Steyn (2000)
method ranged from 11.14 years under to 0.79 years over the actual ages, with the lowest amount
of bias for African-American females.
Table 55: Age bias* for all methods
Count
African-American Males
African-American Females
Total African-Americans

İşcan et al.

Hartnett-Fulginiti Oettlé and Steyn

143
66
209

3.73
4.03
3.83

1.27
3.31
1.91

- 3.13
0.79
- 1.89

European-American Males
149
European-American Females 53
Total European-Americans
202

- 3.93
- 3.82
- 3.90

2.28
1.34
2.03

- 3.67
- 11.14
- 5.63

Total Males
Total Females
Total Sample

- 3.48
0.53
- 2.32

1.78
2.43
1.97

- 3.40
- 4.52
- 3.73

292
119
411

* Bias: Σ (estimated age - actual age) / N
Figure 5 displays the tendency to over- and underage each category for all methods. The
Oettlé and Steyn (2000) method and the İşcan et al. (1984a; 1985) method tended to
underestimate age for most categories, while the Hartnett-Fulginiti (2007; 2010) method
overestimated age for all categories. Additionally, the Hartnett-Fulginiti (2007; 2010) method
had the least amount of bias for almost all categories.

89	
  

Figure 5: Age bias for all methods
İşcan and Loth

Hartnett-Fulginiti

Oettlé and Steyn

6
4
2
Age Bias

0
-2
-4
-6
-8
-10
-12

Intra-observer Error
Intra-observer error rates were calculated for all three methods on a total of 105
individuals. The author re-examined a selection of sixty individuals from the Hamann-Todd
Osteological Collection, consisting of 15 African-American males, 15 African-American
females, 15 European-American males, and 15 European-American females. Additionally, a total
of 45 individuals from the William M. Bass Donated Skeletal Collection were re-examined,
consisting of eleven African-American males, four African-American females, 15 EuropeanAmerican males, and 15 European-American females.

90	
  

Correlation
The İşcan et al. (1984a; 1985), Hartnett-Fulginiti (2007; 2010), and Oettlé and Steyn
(2000) methods were examined for intra-observer error rates using a Spearman’s coefficient of
rank correlation. Table 56 displays the number of individuals, the correlation coefficients (rs),
and the significance for the three methods. All correlations were strong (rs = 0.7–0.89),
indicating a low intra-observer error rate for the author. The highest intra-observer agreement
was found for the Hartnett-Fulginiti method (rs = 0.888), followed by the Oettlé and Steyn
method (rs = 0.882), and then the İşcan et al. method (rs = 0.831).

Table 56: Correlation of estimate 1 and estimate 2 for the entire sample
Count
İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

Spearman’s r

Significance

105
105
105

0.831
0.888
0.882

< 0.0001
< 0.0001
< 0.0001

Additional tests using Cohen’s Kappa measure of agreement were also used. Table 57
displays the number of individuals, the Kappa value (κ), the standard error, and the significance
of all methods. Results displayed a moderate association (κ = 0.4–0.75) for the Hartnett-Fulginiti
(2007; 2010) method and Oettlé and Steyn (2000) method, while the İşcan et al. (1984a; 1985)
displayed a low association (κ = < 0.4).

91	
  

Table 57: Cohen’s Kappa between estimate 1 and estimate 2 for the entire sample
Count
İşcan et al.
Hartnett-Fulginiti
Oettlé and Steyn

Kappa

Std Error

Significance

105
105
105

0.354
0.582
0.448

0.057
0.056
0.059

< 0.0001
< 0.0001
< 0.0001

Cross-tabulation
Cross-tabulation tables were also used in order to visually compare the first estimation
and the second estimation. Table 58 demonstrates the 105 individuals that were re-examined for
the İşcan et al. (1984a; 1985) method. A total of 48 cases (45.7%) were given the same phase
estimate as the first estimation. Estimations within one phase of the original estimate included 42
individuals (40.0%), either one phase over (n = 23) or one phase under (n = 19). Therefore the
author successfully reassessed 85.7% of individuals within one phase of the original estimate.
Finally, 13 individuals (12.4%) were within two phases over (n = 8) or under (n = 5) the original
estimate.
Table 58: Cross-tabulation of estimate 1 and estimate 2 for the İşcan et al. method for the
entire sample
Estimate 2
0
1
2
3
4
5
6
7
8
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
2
0
0
1
2
0
0
0
0
0
3
0
2
3
1
1
2
1
0
0
Estimate 1
4
0
0
1
3
11
2
0
0
0
5
0
0
0
2
3
11
4
2
0
6
0
0
0
1
2
9
6
5
1
7
0
0
0
0
0
0
3
7
5
8
0
0
0
0
0
0
1
2
11

92	
  

Table 59 demonstrates the 105 individuals that were re-examined for the HartnettFulginiti (2007; 2010) method. A total of 69 cases (65.7%) were given the same phase estimate
as the first estimation. Estimations within one phase of the original estimate included 29
individuals (27.6%), either one phase over (n = 13) or one phase under (n = 16). Therefore the
author successfully reassessed 93.3% of individuals within one phase of the original estimate.
Finally, 7 individuals (6.7%) were within two phases over (n = 2) or under (n = 5) the original
estimate.
Table 59: Cross-tabulation of estimate 1 and estimate 2 for the Hartnett-Fulginiti method
for the entire sample
Estimate 2
0
1
2
3
4
5
6
7
0
0
0
0
0
0
0
0
0
1
0
0
1
0
0
0
0
0
2
0
0
3
3
1
0
0
0
3
0
0
1
15
5
1
0
0
Estimate 1
4
0
0
0
4
10
4
2
0
5
0
0
0
1
3
13
2
1
6
0
0
0
0
1
3
12
1
7
0
0
0
0
0
0
2
16

Table 60 demonstrates the 105 individuals that were re-examined for the Oettlé and Steyn
(2000) method. A total of 57 cases (54.3%) were given the same phase estimate as the first
estimation. Estimations within one phase of the original estimate included 42 individuals
(40.0%), either one phase over (n = 18) or one phase under (n = 24). Therefore the author
successfully reassessed 93.3% of individuals within one phase of the original estimate. Finally, 6
individuals (5. 7%) were within two phases over (n = 2) or under (n = 4) the original estimate.

93	
  

Table 60: Cross-tabulation of estimate 1 and estimate 2 for the Oettlé and Steyn method for
the entire sample
Estimate 2
0
1
2
3
4
5
6
7
8
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
0
0
2
0
0
0
1
0
0
0
0
0
3
0
0
1
12
2
1
0
0
0
Estimate 1
4
0
0
0
3
10
7
1
0
0
5
0
0
0
1
8
12
5
0
0
6
0
0
0
0
0
3
8
4
1
7
0
0
0
0
0
1
0
8
5
8
0
0
0
0
0
0
0
3
7

For all three methods tests indicated a high level of correlation (rs = 0.7–0.89) and a
moderate level of agreement (κ = 0.4–0.75). These results show that all methods do not suffer
from a significant amount of intra-observer error.

	
  
	
  
	
  
	
  

94	
  

DISCUSSION
This research addressed the accuracy of three sternal rib aging techniques: the İşcan et al.
(1984a; 1985), the Hartnett-Fulginiti (2007; 2010), and the Oettlé and Steyn (2000) methods.
Results indicated that the three methods are reliable and accurate, all demonstrating strong and
highly significant correlations between the actual and estimated ages (p < 0.0001). Specifically,
when the correlation values for the entire sample were examined, the Hartnett-Fulginiti (2007;
2010) method contained the strongest associations (rs = 0.831; 0.793). The method was also
determined to be a slightly better predictor of age compared to the other techniques analyzed (^ =
p
0.4842). Further analysis of overall accuracy rates showed statistical significance for the
Hartnett-Fulginiti (2007; 2010) method, indicating that it was the most accurate at assessing age
for the entire sample (p = 0.011, 0.022).
Although all methods had strong correlation coefficients, cross-tabulation tables showed
a relatively low accuracy rate for correct phase association, with less than half of the sample that
was correctly assigned for each method. Most of the sample was assigned within two phases
either above or below the actual phase number. This may indicate that the confidence intervals
and age ranges for each phase are too narrow and therefore should be expanded in order to
include more individuals.

Accuracy for European-Americans
Additional tests of accuracy were conducted on European-Americans within the sample.
These individuals were specifically addressed as part of this research due to the majority of
methods that have been developed exclusively or mostly on individuals of European ancestry.
This research especially addressed the İşcan et al. (1984a; 1985) method, which was developed

95	
  

on only European-Americans, and the Hartnett-Fulginiti (2007; 2010) method, which was
developed on individuals of multiple ancestries, consisting of mostly Caucasians. It was expected
that the Hartnett-Fulginiti (2007; 2010) method would more accurately assess age for individuals
of European ancestry, as it was developed as a modification and replacement of the İşcan et al.
(1984a; 1985) method.
Statistical analysis was conducted on the European-American subsample (n = 202). All
three methods displayed correlation coefficients that were strong and highly significant. (p <
0.0001). Further analysis was conducted in order to determine if there were any statistically
significant differences in accuracy among the methods. No significant differences were found,
indicating that all three methods were able to assess age equally well for European-Americans
within the sample. Additionally, each method was tested to determine how well it could predict
age. Although no statistically significant differences were found, the Hartnett-Fulginiti (2007;
2010) method was determined to be a slightly better predictor of age when compared to the other
techniques analyzed (^ = 0.4901).
p
Finally, cross-tabulation tables were also used to give a visual representation of the data.
Again, a relatively low accuracy rate for classifying ribs into the correct age phase was found for
each method. The Hartnett-Fulginiti (2007; 2010) method displayed the highest accuracy, with
49.3 percent of the sample correctly assigned. These results suggest that all three methods can be
utilized on a European-American sample, as they all performed equally well at assessing age for
this population.

96	
  

Accuracy for African-Americans
In addition to determining accuracy rates of all methods for the entire sample, rates of
accuracy for African-Americans within the sample were also addressed per differences in
accuracy based on ancestry as seen in the literature. Specifically, İşcan et al. (1987) developed a
rib aging method for African-Americans due to differences found in the morphological agerelated characteristics. Additionally, Oettlé and Steyn (2000) found poor accuracy rates of the
İşcan et al. (1984a; 1985) methods for South African blacks, prompting a new technique to be
developed for this population. Due to similar features found in the degeneration process of the
sternal end of the rib for both South African blacks and African-Americans, a main objective of
this research was to determine if there were differences in accuracy between the methods for
African-Americans. Based on these observations, it was expected that the Oettlé and Steyn
(2000) method would improve the technique’s accuracy rates for African-Americans.
Statistical analysis was conducted on the African-American subsample (n = 209). All
three methods displayed correlations that were strong and highly significant (p < 0.0001).
Further tests were utilized to determine if there were any statistically significant differences in
accuracy between the methods. No statistically significant values were found, indicating that all
methods were able to accurately assess age for African-Americans in the sample, although the
Hartnett-Fulginiti (2007; 2010) method was determined to be a slightly better predictor of age
when compared to the other techniques analyzed (^ = 0.4737).
p
Finally, cross-tabulation tables were also used to give a visual representation of the data.
A relatively low accuracy rate for classifying ribs into the correct age phase was found for each
method. The Hartnett-Fulginiti (2007; 2010) method displayed the highest accuracy, with 47.9

97	
  

percent of the sample phased correctly. These results suggest that all three methods can be used
on African-Americans, as they all performed equally well as assessing age for this population.

Accuracy Based on Ancestry
An additional research question addressed whether the differences in the rates of
accuracy for African-Americans and European-Americans were statistically significant. As
previously mentioned, the literature had indicated possible ancestral differences in the rate of
aging, prompting new standards for specific populations. Therefore an additional aim of this
research was to determine if there were any differences in the rates of accuracy, which would
provide a basis for separate rib aging standards for African-Americans and European-Americans.
Statistical analysis was conducted on the African-American subsample (n = 209) and the
European-American subsample (n = 202). Comparisons were made in order to determine if there
were differences in accuracy between the correlation coefficients obtained for AfricanAmericans and those obtained for European-Americans for each method. No statistically
significant differences were found for any method, indicating that if differences in the rate of
aging are present, they are not significant enough to impair accuracy rates or require separate
standards.

Accuracy Based on Sex
Comparisons between males and females were examined within each method, with no
statistically significant differences found between males and females for any method. Correlation
coefficients were also compared for males and females between each method. When examining
males, no statistically significant values were found, indicating that for males, all methods were

98	
  

able to assess age equally well. When examining females, statistically significant values were
found for the Hartnett-Fulginiti (2007; 2010), indicating that for females, the Harnett-Fulginiti
(2007; 2010) method was able to assess age more accurately than the other techniques.
Sex was then further divided on the basis of ancestry in order to make additional
comparisons within and between all three methods. No statistically significant differences were
found between methods for European-American males and African-American females,
indicating that all three methods were able to assess age equally well for these groups. It was also
found that for African-American males, the İşcan et al. (1984a; 1985) method was able to assess
age more accurately than the Oettlé and Steyn (2000) method. Finally, for European-American
females it was found that the Hartnett-Fulginiti (2007; 2010) method and the İşcan et al. (1984a;
1985) method were able to assess age more accurately than the Oettlé and Steyn (2000) method.

Inaccuracy and Bias
Rates of inaccuracy and bias were also calculated for multiple categories for each
method. Phase inaccuracy, or the average error from the actual phase number without regard to
the direction of the error, was determined for all methods (Russell et al., 1993; Martrille et al.,
2007). Less than a one phase difference was found for each technique, indicating that all
methods can correctly estimate an individual’s age within one phase of the actual phase number.
Age inaccuracy was also calculated, which determines the average error in terms of age. The
Hartnett-Fulginiti (2007; 2010) method and İşcan et al. (1984a; 1985) method showed the least
amount of inaccuracy, overestimating age between six and ten years. The Oettlé and Steyn
(2000) method displayed the most inaccuracy, overestimating age between seven and fifteen
years.

99	
  

Bias was also calculated for both phase and age differences. All three methods displayed
a phase bias for less than one phase difference either above or below the actual phase. For age
bias, the Hartnett-Fulginiti (2007; 2010) method displayed the least amount of bias for almost all
categories while the Oettlé and Steyn (2000) method displayed the least amount of bias for
African-Americans and African-American females.

Intra-observer Error
Finally, intra-observer error rates were calculated for each method on a subsample of 105
individuals. Statistical tests included Spearman’s coefficient of rank correlation and Cohen’s
Kappa measure of agreement. Tests indicated a high level of correlation (rs = 0.7–0.89) and a
moderate level of agreement (κ = 0.4–0.75) for all three methods. These results show that all
methods do not suffer from a significant amount of intra-observer error.

Limitations
One limitation of this study was that rates of inter-observer error were not calculated.
This limitation can be attributed to the use of multiple collections; therefore, it was not be
possible for the same observers to analyze both collections. This also prevented further
conclusions addressing the level of experience needed to perform each method. Additionally,
although the author had practiced each method extensively, intra-observer error rates may be a
more accurate reflection of the author’s experience than of the method itself.
An additional limitation comes from the conversion of a phase number to a mean age for
the purposes of comparison. This was conducted for both Spearman’s rank correlation and for
inaccuracy and bias. Although this was conducted in accordance with the literature, it may bring

100	
  

a source of error to this research. This is because a phase method is designed to give an age range
in order to include many individuals. Therefore the mean for each phase may significantly underor over-estimate the age of that individual. To accommodate for this error, comparisons between
the estimated and actual phase numbers were also included in the analysis for both the
Spearman’s coefficient of rank correlation and for the calculation of inaccuracy and bias. This
comparison was made between the estimated phase number, as determined by the author’s
evaluation, and the actual phase number, as determined by the individual’s actual age-at-death.
A third limitation of this research was that the three methods have problems of sample
bias. When testing a method it is important to make an attempt to match the distribution of the
original sample. Unfortunately, this was not possible, as this research utilized the same sample of
individuals for the three methods. Therefore some error may have been introduced into the
analysis.
Finally, it was a goal to obtain at least 15 individuals for each phase to maintain
statistically accuracy. This was not possible based on the material available, therefore many of
the younger phases contain less than 15 individuals, while the older and middle-aged phases may
contain significantly more. Again, this may bias the results, as more emphasis may be placed on
the accuracy of the older phases.

101	
  

CONCLUSION
The purpose of this research was to evaluate the accuracy of three sternal rib aging
techniques: İşcan et al. (1984a; 1985), Hartnett-Fulginiti (2007; 2010), and Oettlé and Steyn
(2000), as well as to evaluate any differences in accuracy between methods when applied to
African-Americans and European-Americans. The results of this research were expected to
support all hypotheses. Particularly important was the hypothesis that there would be statistically
significant differences in the accuracy of the aging techniques between African-American and
European-American specimens.
The first and second hypotheses of this research addressed the Hartnett-Fulginiti (2007;
2010) method. The first stated that the Hartnett-Fulginiti (2007; 2010) method would more
accurately assess age for all individuals. According to Garvin and Passalacqua (2012), the most
accepted method for sternal rib aging is the İşcan et al. (1984a; 1985) method. The HartnettFulginiti (2007; 2010) method is a more recent technique developed as a revision in order to
improve accuracy rates. It was therefore expected that the Hartnett-Fulginiti (2007; 2010)
method would be the most accurate at assessing age for the entire sample. This research
supported this hypothesis, as the Hartnett-Fulginiti (2007; 2010) method contained a
significantly higher association between the estimated and actual phase numbers compared to the
other methods in question.
The second hypothesis also addressed the Hartnett-Fulginiti (2007; 2010) method,
specifically if it would more accurately assess age for European-Americans within the sample. It
was expected that this method would more accurately assess age for this group, as it was
developed on a collection consisting mostly of individuals classified as Caucasian (Hartnett,
2007). Results rejected this hypothesis, with no significant differences in accuracy found

102	
  

between the three methods for European-Americans, although the Hartnett-Fulginiti (2007;
2010) method was determined to be a slightly better predictor of age.
The third hypothesis stated that the Oettlé and Steyn (2000) method would more
accurately assess the age of African-Americans within the sample, as this method was designed
specifically on South African blacks. Due to similar features that were found in the degeneration
process of both South African blacks and African-Americans, it was expected that the Oettlé and
Steyn (2000) method would be the most accurate at assessing age for African-Americans in the
sample. This hypothesis was rejected, as no significant differences in accuracy were found
between the three methods for African-Americans, although the Hartnett-Fulginiti (2007; 2010)
method was a slightly better predictor of age.
An additional goal of this research was to determine if there were any differences in
accuracy between African-Americans and European-Americans, which was addressed in the
third hypothesis. No statistically significant differences were found between the two groups for
any method, demonstrating that if differences in the rate of aging are present, they do not appear
to be significant enough to impair accuracy rates or require separate standards.
The fourth hypothesis examined differences in accuracy based on sex. Sex-based
differences were first examined within each method, with no significant differences found,
therefore rejecting this hypothesis. Additional differences in accuracy were examined for males
and females between the three methods. For males, no statistically significant differences in
accuracy were found, indicating that all three methods can assess age equally well for males. In
contrast, for females, the Hartnett-Fulginiti (2007; 2010) method performed the best in
comparison to both the İşcan et al. (1984a; 1985) and Oettlé and Steyn (2000) methods.

103	
  

Sex was further divided on the basis of ancestry for each method in order to address the
fifth hypothesis. No statistically significant differences were found between methods for
European-American males or African-American females. For African-American males, the İşcan
et al. (1984a; 1985) method more accurately assessed age than the Oettlé and Steyn (2000)
method. For European-American females, the Hartnett-Fulginiti (2007; 2010) method and the
İşcan et al. (1984a; 1985) method were able to assess age equally well and more accurately than
the Oettlé and Steyn (2000) method.
Finally, the sixth research hypothesis addressed the rates of intra-observer error, stating
that for all methods, the rates would be low. Intra-observer error was calculated for all methods,
as well as specifically for the subsamples of African-Americans and European-Americans within
each method. All three methods demonstrated strong correlations and moderate agreement,
indicating low levels of intra-observer error. Additionally, no differences in the rates of intraobserver error were found between African-Americans and European-Americans for each
method.
Although strong correlation coefficients were obtained for all methods, relatively low
accuracy rates for classifying ribs into the correct phase were found, with less than half of the
sample correctly assigned for each method. Instead, most of the sample was assigned within two
phases, which may suggest that the confidence intervals and age ranges for each phase are too
narrow. Rib aging techniques have incredibly small age ranges; some phases are as limited as
two years (e.g. İşcan et al., 1984a). Therefore it may be useful to either categorize a rib into the
two closest phases or expand the age ranges for each phase in order to include more individuals.
Additionally, it may be beneficial to revise the rib end aging methods further to either a nonphase system or to a general young, middle, old age classification system.

104	
  

In order to enhance this study, further research should be conducted on the HartnettFulginiti (2007; 2010) method and the İşcan et al. (1984a; 1985) method using a sample that
includes a larger quantity of modern individuals. Tests of inter-observer error, especially for the
Hartnett-Fulginiti (2007; 2010) method, should also be conducted, as it was not possible for this
research. More research should also be conducted on specific populations, including AfricanAmericans to further test if separate aging standards are necessary. Finally, examining the aging
process of the costal cartilage and sternal rib in more depth may also be helpful, as the current
sternal rib aging techniques rely purely on segregating ribs into phases based on morphological
features seen and ignore the biomechanical processes behind the changes.
To conclude, this research demonstrated that the İşcan et al. (1984a; 1985), HartnettFulginiti (2007; 2010), and Oettlé and Steyn (2000) methods performed well on the samples of
both European-Americans and African-Americans. Specifically for African-Americans, the
Oettlé and Steyn (2000) method performed similarly to the other three methods and did not
significantly improve accuracy rates. Therefore the method, or a variation of it, is not
recommended for use within the United States. Additionally, although no significant differences
were found between African-Americans and European-Americans for any method, differences in
the aging process were noticed between the two ancestral groups during examination. These
differences were seen specifically in African-American males, in that they typically retained
scalloping patterns and bone quality throughout age compared to their European-American
counterparts. These findings indicate that even if differences are present between individuals of
African and European ancestry, these differences are do not appear to be significant enough to
require the use of separate standards, although further testing should be conducted. This may also

105	
  

suggest that variation within ancestries is just as great as variation between ancestries, which
may have prevented these differences from showing significance.
Additionally, the Hartnett-Fulginiti (2007; 2010) method was determined to be a better
assessor of age when compared to the İşcan et al. (1984a; 1985) method, which is currently the
most accepted method in the field (Garvin and Passalacqua, 2012). For the author, the HartnettFulginiti (2007; 2010) method was easier to apply than the other two methods. And although
casts are not available as they are for the İşcan et al. (1984a; 1985) method, they were not needed
since this method places more emphasis on bone quality that cannot be demonstrated in a bone
cast. During examination there were several instances where the morphological features did not
correspond to the texture and quality of the bone, and for all cases, the bone quality was always a
more accurate reflection of age. Therefore, it is recommended that bone quality be the most
important factor in the determination of age from the sternal end of the rib.
Finally, although accuracy rates were slightly enhanced using the Hartnett-Fulginiti
(2007; 2010) method, they were not drastically improved. Therefore, a total replacement of the
Hartnett-Fulginiti (2007; 2010) method is unrealistic. Instead it is recommended that when
utilizing a rib aging technique, practitioners should decide which method to use based on their
own level of experience and preference.

	
  

106	
  

APPENDICES

107	
  

Appendix A: İşcan et al. Male Phase Descriptions and Photographs
Figure 6: İşcan et al. (1984a) Male Phase 0 (16 years and younger)

“The articular surface is flat or billowy with a regular rim and rounded edges. The bone itself is
smooth, firm, and very solid” (İşcan et al. 1984a:1096).

Figure 7: İşcan et al. (1984a) Male Phase 1 (17–19 years)

“There is a beginning amorphous indentation in the articular surface, but billowing may also still
be present. The rim is rounded and regular. In some cases scallops may start to appear at the
edges. The bone still firm, smooth, and solid” (İşcan et al. 1984a:1096).

108	
  

Figure 8: İşcan et al. (1984a) Male Phase 2 (20–23 years)

“The pit is now deeper and has assumed a V-shape appearance formed by the anterior and
posterior walls. The walls are thick and smooth with a scalloped or slightly wavy rim with
rounded edges. The bone is firm and solid” (İşcan et al. 1984a:1096).

Figure 9: İşcan et al. (1984a) Male Phase 3 (24–28 years)

“The deepening pit has taken on a narrow to moderately U-shape. Walls are still fairly thick with
rounded edges. Some scalloping may still be present but the rim is becoming more irregular. The
bone is still quite firm and solid” (İşcan et al. 1984a:1096).

109	
  

Figure 10: İşcan et al. (1984a) Male Phase 4 (26–32 years)

“Pit depth is increasing, but the shape is still a narrow to moderately wide U. The walls are
thinner, but the edges remain rounded. The rim is more irregular with no uniform scalloping
pattern remaining. There is some decrease in the weight and firmness of the bone, however, the
overall quality of the bone is still good” (İşcan et al. 1984a:1096,1099).

Figure 11: İşcan et al. (1984a) Male Phase 5 (33–42 years)

“There is little change in pit depth, but the shape in this phase is predominantly a moderately
wide U. Walls show further thinning and the edges are becoming sharp. Irregularity is increasing
in the rim. Scalloping pattern is completely gone and has been replaced with irregular bony
projections. The condition of the bone is fairly good, however, there are some signs of
deterioration with evidence of porosity and loss of density” (İşcan et al. 1984a:1099).

110	
  

Figure 12: İşcan et al. (1984a) Male Phase 6 (43–55 years)

“The pit is noticeably deep with a wide U-shape. The walls are thin with sharp edges. The rim is
irregular and exhibits some rather long bony projections that are frequently more pronounced at
the superior and inferior borders. The bone is noticeably lighter in weight, thinner, and more
porous, especially inside the pit” (İşcan et al. 1984a:1099).

Figure 13: İşcan et al. (1984a) Male Phase 7 (54–64 years)

“The pit is deep with a wide to very wide U-shape. The walls are thin and fragile with sharp,
irregular edges and bony projections. The bone is light in weight and brittle with significant
deterioration in quality and obvious porosity” (İşcan et al. 1984a:1099).

111	
  

Figure 14: İşcan et al. (1984a) Male Phase 8 (65 years and older)

“In this final phase the pit is very deep and widely U-shaped. In some cases the floor of the pit is
absent or filled with bony projections. The walls are extremely thin, fragile, and brittle with
sharp, highly irregular edges and bony projections. The bone is very lightweight, thin, brittle,
friable, and porous. “Window” formation is sometimes seen in the walls” (İşcan et al.
1984a:1099).

112	
  

Appendix B: İşcan et al. Female Phase Descriptions and Photographs
Figure 15: İşcan et al. (1985) Female Phase 0 (13 years and younger)

“The articular surface is nearly flat with ridges or billowing. The outer surface of the sternal
extremity of the rib is bordered by what appears to be an overlay of bone. The rim is regular with
rounded edges, and the bones itself is firm, smooth, and very solid” (İşcan et al. 1985:855).
Figure 16: İşcan et al. (1985) Female Phase 1 (14–15 years)

“A beginning, amorphous indentation can be seen in the articular surface. Ridges or billowing
may still be present. The rim is rounded and regular with a little waviness in some cases. The
bone remains solid, firm, and smooth” (İşcan et al. 1985:855).

113	
  

Figure 17: İşcan et al. (1985) Female Phase 2 (16–19 years)

“The pit is considerably deeper and has assumed a V-shape between the thick, smooth anterior
and posterior walls, Some ridges or billowing may still remains inside the pit. The rim is wavy
with some scallops beginning to form at the rounded edge. The bone itself is firm and solid”
(İşcan et al. 1985:855).

Figure 18: İşcan et al. (1985) Female Phase 3 (20–24 years)

“There is only slight if any increase in pit depth, but the V-shape is wider, sometimes
approaching a narrow U as the walls become a bit thinner. The still rounded edges now show a
pronounced, regular scalloping pattern. At this stage, the anterior or posterior walls or both may
first start to exhibit a central, semicircular arc of bone. The rim is firm and solid” (İşcan et al.
1985:855).

114	
  

Figure 19: İşcan et al. (1985) Female Phase 4 (24–32 years)

“There is a noticeable increase in the depth of the pit, which now has a wide V- or narrow Ushape with, at time, flared edges. The walls are thinner but the rim remains rounded. Some
scalloping is still present, along with the central arc; however, the scallops are not as well
defined and the edges look somewhat worn down. The quality of the bone is fairly good but there
is some decrease in density and firmness” (İşcan et al. 1985:855).

Figure 20: İşcan et al. (1985) Female Phase 5 (33–46 years)

“The depth of the pit stays about the same, but the thinning walls are flaring into a wider V- or
U-shape. In most cases, a smooth, hard, plaque-like deposit lines at least part of the pit. No
regular scalloping remains and the edge is beginning to sharpen. The rim is becoming more
irregular, but the central arc is still the most prominent projection. The bone is noticeably lighter
in weight, density and firmness. The texture is somewhat brittle” (İşcan et al. 1985:858).

115	
  

Figure 21: İşcan et al. (1985) Female Phase 6 (43–58 years)

“An increase in pith depth is again noted, and its V- or U-shape has widened again because of
pronounced flaring at the end. The plaque-like deposit may still appear but is rougher and more
porous. The walls are quite thin with sharp edges and an irregular rim. The central arc is less
obvious and, in many cases, sharp points project from the rim of the sternal extremity. The bone
itself is fairly thin and brittle with some signs of deterioration” (İşcan et al. 1985:858).
Figure 22: İşcan et al. (1985) Female Phase 7 (59–71 years)

“The depth of the predominantly flared U-shaped pit not only shows no increase, but actually
decreases slightly. Irregular bony growths are often seen extruding from the interior of the pit.
The central arc is still present in most cases but is now accompanied by pointed projections,
often at the superior and inferior borders, yet may be evidenced anywhere around the rim. The
very thin walls have irregular rims with sharp edges. The bone is very light, thin, brittle, and
fragile, with deterioration inside the pit” (İşcan et al. 1985:858).

116	
  

Figure 23: İşcan et al. (1985) Female Phase 8 (70 years and older)

“The floor of the U-shaped pit in this final phase is relatively shallow, badly deteriorated, or
completely eroded. Sometimes it is filled with bony growths. The central arc is barely
recognizable. The extremely thin, fragile walls have highly irregular rims with very sharp edges,
and often fairly long projections of bone at the inferior and superior borders. “Window”
formation sometimes occurs in the walls. The bone itself is in poor condition – extremely thin,
light in weight, brittle, and fragile” (İşcan et al. 1985:858).
	
  
	
  

117	
  

Appendix C: Hartnett-Fulginiti Male Phase Descriptions and Photographs
“Bone quality is a major determining factor in placing each specimen in a group. An individual
may be moved up or down phases based solely on bone quality” (Hartnett, 2007:373).
Figure 24: Hartnett-Fulginiti (2007; 2010) Male Phase 1 (18–22 years)

“The pit is shallow and flat, and there are billows in the pit. The pit is shallow U-shaped in
cross-section. The bone is very firm and solid, smooth to the touch, dense, and of good quality.
The walls of the rim are thick. The rim may show the beginnings of scalloping” (Hartnett,
2010:1156). For interpretation of the references to color in this and all other figures, the reader is
referred to the electronic version of this thesis.
Figure 25: Hartnett-Fulginiti (2007; 2010) Male Phase 2 (21–28 years)

“There is an indentation to the pit. The pit is V-shaped in cross-section, and the rim is well
defined with round edges. The rim is regular with some scalloping. The bone is firm and solid,
smooth to the touch, dense, and of good quality. There is no flare to the rim edges; they are
parallel to each other. The pit is still smooth inside, with little to no porosity” (Hartnett,
2010:1156).

118	
  

Figure 26: Hartnett-Fulginiti (2007; 2010) Male Phase 3 (27–37 years)

“The pit is V-shaped, and there is a slight flare to the rim edges. The rim edges are becoming
undulating and slightly irregular, and there may be remnants of scallops, but they look worn
down. There are no bony projections from the rim. There is porosity inside the pit. The bone
quality is good; it is firm, solid, and smooth to the touch. The rim edges are rounded, but sharp”
(Hartnett, 2010:1156).

Figure 27: Hartnett-Fulginiti (2007; 2010) Male Phase 4 (36–48 years)

“The pit is deep and U-shaped. The edges of the pit flare outwards, expanding the oval area
inside the pit. The rim edges are not undulating or scalloped but are irregular. There are no long
bony projections from the rim, and the rim edges are thin, but firm. The bone quality is good but
does not feel dense or heavy. There is porosity inside the pit. In some males, two distinct
depressions are visible in the pit” (Hartnett, 2010:1156).
119	
  

Figure 28: Hartnett-Fulginiti (2007; 2010) Male Phase 5 (45–59 years)

“There are frequently small bony projections along the rim edges, especially at the superior and
inferior edges of the rim. The pit is deep and U-shaped. The rim edges are irregular, flared,
sharp, and thin. There is porosity inside the pit. The bone quality is fair; the bone is coarse to the
touch and feels lighter than it looks” (Hartnett, 2010:1156).

Figure 29: Hartnett-Fulginiti (2007; 2010) Male Phase 6 (57–70 years)

“The bone quality is fair to poor, light in weight, and the surfaces of the bone feel coarse and
brittle. There are bony projections along the rim edges, especially at the superior and inferior
edges, some of which may be over 1 cm long. The pit is deep and U-shaped. The rim is very
irregular, thin, and fragile. There is porosity inside the pit. In some cases, there may be small
bony extrusions inside the pit” (Hartnett, 2010:1156).

120	
  

Figure 30: Hartnett-Fulginiti (2007; 2010) Male Phase 7 (70–97 years)

“The bone is very poor quality, and in many cases, translucent. The bone is very light, sometimes
feeling like paper, and feels coarse and brittle to the touch. The pit is deep and U-shaped. There
may be long bony growths inside the pit. The rim is very irregular with long bony projections. In
some cases, much of the cartilage has ossified and window formation occurs” (Hartnett,
2010:1156).

Figure 31: Hartnett-Fulginiti (2007; 2010) Male Variant Phase (27–69 years)

“In some males, the cartilage has completely or almost completely ossified. The ossification
tends to be a solid extension of bone, rather than a thin projection. All of the bone is of very good
quality, including the ossification. It is dense, heavy, and smooth. In these instances, bone quality
should be the determining factor. There are probably other factors, such as disease, trauma, or
substance abuse that caused premature ossification of the cartilage. When the individual is truly
very old, the bone quality will be very poor. Be aware of these instances where a rib end may
appear very old because of ossification of the cartilage but is really actually a young individual,
which can be ascertained by bone quality. In these cases, consult other age indicators in
conjunction with the rib end” (Hartnett, 2010:1156).

121	
  

Appendix D: Hartnett-Fulginiti Female Phase Descriptions and Photographs
“Bone quality is a major determining factor in placing each specimen in a group. An individual
may be moved up or down phases based solely on bone quality” (Hartnett, 2007:373).
Figure 32: Hartnett-Fulginiti (2007; 2010) Female Phase 1 (18–22 years)

“The pit is shallow and flat, and there are billows in the pit. The pit is shallow U-shaped in crosssection. The bone is very firm and solid, smooth to the touch, dense, and of good quality. The
walls of the rim are thick. The rim may show the beginnings of scalloping” (Hartnett,
2010:1156).

Figure 33: Hartnett-Fulginiti (2007; 2010) Female Phase 2 (24–27 years)

“There is an indentation to the pit. The pit is V-shaped in cross-section, and the rim is well
defined with round edges. The rim is regular with some scalloping. The bone is firm and solid,
smooth to the touch, dense, and of good quality. There is no flare to the rim edges; they are
parallel to each other. The pit is still smooth inside, with little to no porosity. In females, the
central arc, which manifests on the anterior and posterior walls as a semicircular curve, is
visible” (Hartnett, 2010:1156).

122	
  

Figure 34: Hartnett-Fulginiti (2007; 2010) Female Phase 3 (27–38 years)

“The pit is V-shaped, and there is a slight flare to the rim edges. The rim edges are becoming
undulating and slightly irregular, and there may be remnants of scallops, but they look worn
down. There are no bony projections from the rim. There is porosity inside the pit. The bone
quality is good; it is firm, solid, and smooth to the touch. The rim edges are rounded, but sharp.
In many females, there is a build-up of bony plaque, either in the bottom of the pit or lining the
interior of the pit, creating the appearance of a two-layer rim. An irregular central arc may be
apparent” (Hartnett, 2010:1156).

Figure 35: Hartnett-Fulginiti (2007; 2010) Female Phase 4 (39–49 years)

“The pit is deep and U-shaped. The edges of the pit flare outwards, expanding the oval area
inside the pit. The rim edges are not undulating or scalloped but are irregular. There are no long
bony projections from the rim, and the rim edges are thin, but firm. The bone quality is good but
does not feel dense or heavy. There is porosity inside the pit… In females, the central arc may be
present and irregular; however, the superior and inferior edges of the rim have developed,
decreasing the prominence of the central arc” (Hartnett, 2010:1156).

123	
  

Figure 36: Hartnett-Fulginiti (2007; 2010) Female Phase 5 (47–58 years)

“There are frequently small bony projections along the rim edges, especially at the superior and
inferior edges of the rim. The pit is deep and U-shaped. The rim edges are irregular, flared,
sharp, and thin. There is porosity inside the pit. The bone quality is fair; the bone is coarse to the
touch and feels lighter than it looks” (Hartnett, 2010:1156).

Figure 37: Hartnett-Fulginiti (2007; 2010) Female Phase 6 (60–73 years)

“The bone quality is fair to poor, light in weight, and the surfaces of the bone feel coarse and
brittle. There are bony projections along the rim edges, especially at the superior and inferior
edges, some of which may be over 1 cm long. The pit is deep and U-shaped. The rim is very
irregular, thin, and fragile. There is porosity inside the pit. In some cases, there may be small
bony extrusions inside the pit. In females, the central arc is not prominent” (Hartnett,
2010:1156).

124	
  

Figure 38: Hartnett-Fulginiti (2007; 2010) Female Phase 7 (65–99 years)

“The bone is very poor quality, and in many cases, translucent. The bone is very light, sometimes
feeling like paper, and feels coarse and brittle to the touch. The pit is deep and U-shaped. There
may be long bony growths inside the pit. The rim is very irregular with long bony projections. In
some cases, much of the cartilage has ossified and window formation occurs. In some females,
much of the cartilage in the interior of the pit has ossified into a bony projection extending more
than 1 cm in length” (Hartnett, 2010:1156).
	
  

125	
  

Appendix E: Oettlé and Steyn Male Phase Descriptions and Photographs
Figure 39: Oettlé and Steyn (2000) Male Phase 0 (Prephase seen in children)

“The articular surface is flat or billowy with a regular rim and rounded edges. The bone itself is
smooth, firm, and very solid, with an epiphyseal ring” (Oettlé and Steyn 2000:1073).

Figure 40: Oettlé and Steyn (2000) Male Phase 1 (18–21 years)

“An amorphous indentation in the articular surface is beginning. Parts of the pit can show more
advanced features than the rest. The rim is rounded and regular. In some cases scallops may start
to appear at the edges” (Oettlé and Steyn 2000:1073).

126	
  

Figure 41: Oettlé and Steyn (2000) Male Phase 2 (22–24 years)

“The pit is deeper and has a V-shaped appearance formed by the anterior and posterior walls.
The walls are thick and smooth and can exhibit a scalloped or slightly wavy rim with rounded
edges. The scalloping features may be pointed, squared off, or flattened, or may be regular and
receding” (Oettlé and Steyn 2000:1073).

Figure 42: Oettlé and Steyn (2000) Male Phase 3 (25–29 years)

“The deepening pit has taken on a narrow to moderately U-shape. Walls are still fairly thick with
rounded edges. The rim may be unscalloped, smooth, and regular, or may manifest some
scalloping which are becoming more irregular. Bony projections from the rim and, occasionally,
a plaque-like deposit from the interior of the pit may already be seen” (Oettlé and Steyn
2000:1073).

127	
  

Figure 43: Oettlé and Steyn (2000) Male Phase 4 (30–37 years)

“Pit depth is increasing. The walls are thinner, but the edges remain rounded. The rim is more
irregular with no uniform scalloping pattern remaining. There is some decrease in the weight and
firmness of the bone; however, the overall quality of the bone is still good” (Oettlé and Steyn
2000:1073).

Figure 44: Oettlé and Steyn (2000) Male Phase 5 (38–46 years)

“The pit is predominantly a moderately wide U-shape. Walls show further thinning and the edges
are becoming sharp. The scalloping pattern disappears and the rim is irregular with bony
projections, or has a flattened and pointy appearance. The condition of the bone is fairly good,
although there are some signs of deterioration with evidence of porosity and loss of density.
Small bony growths from the pit interior are sometimes present” (Oettlé and Steyn
2000:1073,1075).

128	
  

Figure 45: Oettlé and Steyn (2000) Male Phase 6 (47–55 years)

“The pit is noticeably deep with a wide U-shape. The walls are thin with sharp edges. The rim is
irregular and exhibits some rather long bony projections that are frequently more pronounced at
the superior and inferior borders. The bone is noticeably lighter in weight, thinner, and more
porous, especially inside the pit” (Oettlé and Steyn 2000:1075).

Figure 46: Oettlé and Steyn (2000) Male Phase 7 (52–69 years)

“The pit is deep with a wide to very wide U-shape. The walls are thin and fragile with sharp,
irregular edges and bony projections. The bone is light in weight and brittle with significant
deterioration in quality and obvious porosity” (Oettlé and Steyn 2000:1075).

129	
  

Figure 47: Oettlé and Steyn (2000) Male Phase 8 (63–77 years)

“The pit is very deep and widely U-shaped. In some cases the pit is absent or filled with bony
projections. The walls are extremely thin, fragile, and brittle with sharp, highly irregular edges
and bony projections. The bone is very lightweight, thin, brittle, friable, and porous. “Window”
formation is sometimes seen in the walls” (Oettlé and Steyn 2000:1075).

130	
  

Appendix F: Oettlé and Steyn Female Phase Descriptions and Photographs
Figure 48: Oettlé and Steyn (2000) Female Phase 0 (Prephase seen in children)

“The articular surface is nearly flat with ridges or billowing. The outer surface of the sternal
extremity of the rib is bordered by what appears to be an overlay of bone, the epiphyseal growth
plate. The rim is regular with rounded edges, and the bone itself is firm, smooth, and very solid”
(Oettlé and Steyn 2000:1075).

Figure 49: Oettlé and Steyn (2000) Female Phase 1 (12–17 years)

“An amorphous indentation is beginning in the articular surface. Ridges or billowing may still be
present. The rim is rounded and regular with a little waviness in some cases. An epiphyseal
growth plate or ring could still be noticed” (Oettlé and Steyn 2000:1075).

131	
  

Figure 50: Oettlé and Steyn (2000) Female Phase 2 (16–26 years)

“The amorphous indentation becomes deeper at more than one point, and then continues to form
a V-shaped pit between the thick, smooth anterior and posterior walls. Some ridges or billowing
may still remain inside the pit. The rim is wavy with some scallops beginning to form at the
rounded edge” (Oettlé and Steyn 2000:1075,1077).

Figure 51: Oettlé and Steyn (2000) Female Phase 3 (23–27 years)

“There is only slight or no increase in pit depth, but the V-shape is wider, sometimes
approaching a narrow U as the walls become a bit thinner. The rounded edges now show a
pronounced, regular, scalloping pattern, or the rim may exhibit an irregular scalloping pattern,
which gives it a pointy appearance” (Oettlé and Steyn 2000:1077).

132	
  

Figure 52: Oettlé and Steyn (2000) Female Phase 4 (28–32 years)

“The pit now has a narrow U-shape with at times, flared edges. The pit interior may contain
plaque-like deposits. The walls are thinner and the rim remains rounded, but may also be flat and
pointed. Some scalloping is still present, but this is not as well defined and the edges look
somewhat worn down. There is some decrease in density and firmness of the bone” (Oettlé and
Steyn 2000:1077).

Figure 53: Oettlé and Steyn (2000) Female Phase 5 (33–43 years)

“The pit depth increases, and the thinning walls are flaring into a wider U-shape. A smooth, hard,
plaque-like deposit lines at least part of the pit in some ribs. No regular scalloping pattern
remains and the edge is beginning to sharpen. The rim is becoming more irregular. The bone is
noticeably lighter in weight, density, and firmness. The texture is somewhat brittle” (Oettlé and
Steyn 2000:1077).

133	
  

Figure 54: Oettlé and Steyn (2000) Female Phase 6(a) (44–59 years)

“An increase in pit depth is again noted, and its U-shape has widened again because of
pronounced flaring at the end. The plaque-like deposits may still appear but are rougher and
more porous. The walls are quite thin with sharp edges and an irregular rim. In many cases, sharp
points or bony projections project from the rim of the sternal extremity. The bone itself is fairly
thin and brittle with some signs of deterioration” (Oettlé and Steyn 2000:1077–1078).

Figure 55: Oettlé and Steyn (2000) Female Phase 6(b) (44–59 years)

The pit depth, although not increasing, does not decrease. Obvious bony growths from the pit
interior can be seen. No central arc of bone is present. Pointed projections, often at the superior
and inferior borders, or anywhere around the rim, are found. The very thin walls have irregular
rims and sharp edges” (Oettlé and Steyn 2000:1078).

134	
  

Figure 56: Oettlé and Steyn (2000) Female Phase 7 (56–79 years)

“The floor of the U-shaped pit in this final phase is relatively shallow, badly deteriorated, or
completely eroded. Sometimes it is filled with bony growths. The central arc is barely
recognizable. The extremely thin, fragile walls have highly irregular rims with very sharp edges.
The bone itself is in a poor condition—extremely thin, light in weight, brittle, and fragile”
(Oettlé and Steyn 2000:1077).

135	
  

REFERENCES

136	
  

REFERENCES
2010. Age estimation. Scientific Working Group for Forensic Anthropology (SWGANTH).
http://swganth.startlogic.com/Age%20Rev0.pdf.
Allen MB. 1997. Osteometric analysis of intercostal variation and sexual dimorphism in the
sternal end of the rib of modern American Blacks. Masters Thesis, Florida Atlantic University.
Alsup BK. 2007. Investigation of second, fourth, and eighth sternal rib end variation related to
age estimation. Masters Thesis, The University of Tennessee, Knoxville.
Angel JL. 1980. Physical anthropology: determining sex, age, and individual features. In:
Cockburn A, Cockburn E. Mummies, disease, and ancient cultures. New York: Cambridge
University Press. p. 241–257.
Armelagos GJ, Goodman AH. 1998. Race, racism, and anthropology. In: Goodman AH,
Leatherman T. Building a new biocultural synthesis: political-economic perspectives on human
biology. Ann Arbor: University of Michigan Press. p. 359–377.
Aykroyd RG, Lucy D, Pollard AM, Roberts CA. 1999. Nasty, brutish, but not necessarily short:
a reconsideration of the statistical methods used to calculate age at death from adult human
skeletal and dental age indicators. American Antiquity 64(1):55–70.
Baccino E, Ubelaker DH, Hayek LC, Zerilli A. 1999. Evaluation of seven methods of estimating
age at death from mature human skeletal remains. Journal of Forensic Sciences 44(5):931–936.
Barchilon V, Hershkovitz I, Rothschild BM, Wish-Baratz S, Latimer B, Jellema LM, Hallel T,
Arensburg B. 1996. Factors affecting the rate and pattern of first costal cartilage ossification.
American Journal of Forensic Medicine and Pathology 17(3):239–247.
Barrès DR, Durigon M, Paraire F. 1989. Age estimation from quantitation of features of “chest
plate” x-rays. Journal of Forensic Sciences 34(1):228–233.
Bassett HE, Spradley MK, Jantz LM. 2003. The William M. Bass Donated Collection at the
University of Tennessee in Knoxville. Abstract, Proceedings of the American Academy of
Forensic Sciences 9:243–244.
Beauthier J, Lefevre P, Meunier M, Orban R, Polet C, Werquin J, Quatrehomme G. 2010.
Palatine sutures as an age indicator: a controlled study in the elderly. Journal of Forensic
Sciences 5(1):153–158.
Brooks S. 1955. Skeletal age at death: the reliability of cranial and pubic age indicators.
American Journal of Physical Anthropology 3(4):567–597.
Brooks S, Suchey JM. 1990. Skeletal age determination based on the os pubis: a comparison of
the Acsádi-Nemeskéri and Suchey-Brooks methods. Human Evolution 5(3):227–238.

137	
  

Brown C. 2009. Uncertainty in skeletal aging: a retrospective study and test of skeletal aging
methods at the Joint POW/MIA Accounting Command Central Identification Laboratory.
Masters Thesis, California State University, Chico.
Buckberry JL, Chamberlain AT. 2002. Age estimation from the auricular surface of the ilium: a
revised method. American Journal of Physical Anthropology 119(3):231–239.
Byers SN. 2005. Introduction to forensic anthropology. Third edition. Boston, MA: Pearson
Education, Inc.
Cobb WM. 1952. Skeleton. In: Lansing AI. Cowdry’s problems of ageing, biological and
medical aspects. Third edition. Baltimore: The Williams and Wilkins Company.
Cobb WM. 1959. Thomas Wingate Todd. Journal of the National Medical Association
51(3):233–246.
Cobb WM. 1981. Thomas Wingate Todd, 1885–1938. American Journal of Physical
Anthropology 56(4):517–520.
Cohen J, Cohen P. 1983. Applied multiple regression/correlation analysis for the behavioral
sciences. Second edition. New Jersey: Lawrence Erlbaum Associates.
Cohen J, Cohen P, West SG, Aiken LS. 2003. Applied multiple regression/correlation analysis
for the behavioral sciences. Third edition. Hillsdale, New Jersey: Lawrence Erlbaum Associates.
Cöloğlu AS, İşcan MY, Yavuz MF, Sari H. 1998. Sex determination from the ribs of
contemporary Turks. Journal of Forensic Science 43(2):273–276.
DiGangi EA, Bethard JD, Kimmerle EH, Konigsberg LW. 2009. A new method for estimating
age-at-death from the first rib. American Journal of Physical Anthropology 138(2):164–176.
DeCoster J, Iselin A-M. 2009. Comparing correlation coefficients. Downloaded Microsoft Excel
spreadsheet from http://www.stat-help.com.
Drennan RD. 2004. Statistics for archaeologists: a commonsense approach. Interdisciplinary
Contributions to Archaeology. New York: Springer.
Đonić D, Đurić M, Babić D, Popović Đ. 2005. Reliability of the individual age assessment at the
time of death based on sternal rib-ends’ morphology in Balkan population. Vojnosanit Pregl
62(6):441−446.
Dudar CJ. 1993. Identification of rib number and assessment of intercostal variation at the sternal
rib end. Journal of Forensic Sciences 38(4):788–797.

138	
  

Dudar CJ, Pfeiffer S, Saunders SR. 1993. Evaluation of morphological and histological adult
skeletal age-at-death estimation techniques using ribs. Journal of Forensic Sciences 38(3):677–
685.
Ea H, Nguyen C, Bazin D, Bianchi A, Guicheux J, Reboul P, Daudon M, Lioté F. 2011.
Articular cartilage calcification in osteoarthritis: insights into crystal-induced stress. Arthritis and
Rheumatism 63(1):10–18.
Eichelberger L, Roma M. 1954. Effects of age on the histochemical characterization of costal
cartilage. American Journal of Physiology 178(2):296–304.
Elkeles A. 1966. Sex differences in the calcification of the costal cartilages. Journal of the
American Geriatrics Society 14(5):456–462.
Falconer B. 1938. Calcification of hyaline cartilage in man. The American Journal of Pathology
18:942–955.
Fanton L, Gustin M, Paultre U, Schrag B, Malicier D. 2010. Critical study of observation of the
sternal end of the right 4th rib. Journal of Forensic Science 55(2):467–472.
Fischer E. 1955 Verkalkungsformen der rippenknorpel. Fortschritte auf dem Gebiete der
Röntgen. Strahlen und der Nuklear Medizin 82:474–481.
Fleischman JM. 2011. A comparative evaluation of the Chen et al. and Suchey-Brooks pubic
bone aging methods on a North American sample. Masters Thesis, Michigan State University.
Freund RJ, Wilson WJ, Mohr DL. 2010. Statistical methods. Third edition. Amsterdam; Boston:
Elsevier.
Galera V, Ubelaker DH, Hayek LC. 1995. Interobserver error in macroscopic methods of
estimating age at death from the human skeleton. International Journal of Anthropology
10(4):229–239.
Garvin HM. 2010. Limitations of cartilage ossification as an indicator of age at death. In: Latham
KE, Finnegan M. Age estimation of the human skeleton. Springfield, IL: Charles C. Thomas
Publisher. p. 118–133.
Garvin HM, Passalacqua NV. 2012. Current practices by forensic anthropologists in adult
skeletal age estimation. Journal of Forensic Science 57(2):427–433.
Gavit KR. 2009. Estimating sexual dimorphism from sternal rib ends. Masters Thesis, San
Marcos, Texas.
Gilbert BM. 1973. Misapplication to females of the standard for aging the male Os pubis.
American Journal of Physical Anthropology 38(1):39–40.

139	
  

Gilbert BM, McKern TW. 1973. A method for aging the female Os pubis. American Journal of
Physical Anthropology 38(1):31–38.
Ginter JK. 2005. A test of the effectiveness of the revised maxillary suture obliteration method in
estimating adult age at death. Journal of Forensic Sciences. 50(6):1303–1309.
Gruspier KL, Mullen GJ. 1991. Maxillary suture obliteration: a test of the Mann method. Journal
of Forensic Sciences 36(2):512–519.
Hartnett KM. 2007. A re-evaluation and revision of pubic symphysis and fourth rib aging
techniques. Ph.D. Dissertation, Arizona State University.
Hartnett KM. 2010. Analysis of age-at-death estimation using data from a new, modern autopsy
sample – part II: sternal end of the fourth rib. Journal of Forensic Sciences 55(5):1152–1156.
Hass GM. 1943. A morphologic and chemical analysis of aging human costal cartilage. Archives
of Pathology 35:275–284.
Hinkle DE, Wiersma W, Jurs SG. 1988. Applied statistics for the behavioral sciences. Boston:
Houghton Mifflin Company.
Horner JL. 1949. Premature calcification of the costal cartilages: its frequent association with
symptoms of non-organic origin. American Journal of the Medical Sciences 218(2):186–193.
IBM Corp. 2011. IBM SPSS Statistics for Macintosh, Version 20.0. Armonk, NY: IBM Corp.
İşcan MY. 1985. Osteometric analysis of sexual dimorphism in the sternal end of the rib. Journal
of Forensic Sciences 30(4):1090–1099.
İşcan MY. 1991. The aging process in the rib: an analysis of sex- and race-related morphological
variation. American Journal of Human Biology 3(6):617–623.
İşcan MY, Loth SR. 1986a. Determination of age from the sternal rib in white females: a test of
the phase method 31(3):990–999.
İşcan MY, Loth SR. 1986b. Determination of age from the sternal rib in white males: a test of
the phase method 31(1):122–132.
İşcan MY, Loth SR. 1986c. Estimation of age and determination of sex from the sternal rib. In:
Reichs, KJ. Forensic osteology: advances in the identification of human remains. Springfield, IL:
Charles C. Thomas. p. 68–89.
İşcan MY, Loth SR. 1989. Osteological manifestations of age in the adult. In: İşcan, MY,
Kennedy KAR. Reconstruction of life from the skeleton. New York: Alan R. Liss. p. 23–40.

140	
  

İşcan MY, Loth SR. 1993. Casts of age phases from the sternal end of the rib for white males and
females. Fort Collins, CO: France Casting.
İşcan MY, Loth SR, Scheuerman EH. 1989. Assessment of intercostal variation on the
estimation of age from the sternal end of the rib. Abstract. American Journal of Physical
Anthropology 78:245.
İşcan MY, Loth SR, Scheuerman EH. 1992. Age estimation from the sternal end of the rib and
pubic symphysis: a systematic comparison. Anthropologie 30(1):40–44.
İşcan MY, Loth SR, Wright RK. 1984a. Age estimation from the rib by phase analysis: white
males. Journal of Forensic Sciences 29(4):1094–1104.
İşcan MY, Loth SR, Wright RK. 1984b. Metamorphosis at the sternal rib end: a new method to
estimate age at death in white males. American Journal of Physical Anthropology 65(2):147–
156.
İşcan MY, Loth SR, Wright RK. 1985. Age estimation from the rib by phase analysis: white
females. Journal of Forensic Sciences 30(3):853–863.
İşcan MY, Loth SR, Wright RK. 1987. Racial variation in the sternal extremity of the rib and its
effect on age determination. Journal of Forensic Sciences 32(2):452–466.
Jones-Kern K, Latimer B. 1997. Skeletons out of the closet. The Explorer 38(1–2):26–28.
Jordanoglou J. 1969. Rib movement in health, kyphoscoliosis, and ankylosing spondylitis.
Thorax 24(4):407–414.
Katz D, Suchey JM. 1986. Age determination of the male os pubis. American Journal of Physical
Anthropology 69(4):427–435.
Kennedy KAR. 1995. But professor, why teach race identification if races don’t exist? Journal of
Forensic Sciences 40:797–800.
Kerley ER. 1970. Estimation of skeletal age: after about 30 years. In: Stewart TD. Personal
identification in mass disasters. Washington, DC: National Museum of Natural History. p. 57–
70.
Kimmerle EH, Prince DA, Berg GE. 2008. Inter-observer variation in methodologies involving
the pubic symphysis, sternal ribs, and teeth. Journal of Forensic Sciences 53(3):594–600.
King JB. 1939. Calcification of the costal cartilages. British Journal of Radiology 12(133):2–12.
Klepinger LL. 2006. Age estimation. In: Cartmill M, Brown K. Fundamentals of forensic
anthropology. Hoboken, NJ: Wiley-Liss. p. 42–63.

141	
  

Koçak A, Aktas EÖ, Ertürk S, Aktas S, Yemisçigil A. 2003. Sex determination from the sternal
end of the rib by osteometric analysis. Legal Medicine 5(2):100–104.
Kunos CA, Simpson SW, Russell KF, Hershkovitz I. 1999. First rib measurements: its possible
utility for human age-at-death estimation. American Journal of Physical Anthropology
110(3):303–323.
Kurki H. 2005. Use of the first rib for adult age estimation: a test of one method. International
Journal of Osteoarchaeology 15(5):342–350.
Loth SR. 1990. A comparative analysis of the manifestation of age, sex, and race in the sternal
extremity of the rib: a consideration of human skeletal variation. Masters Thesis, Florida Atlantic
University.
Loth SR. 2005. Age assessment of the Spitalfields cemetery population by rib phase analysis.
American Journal of Human Biology 7(4):465–471.
Loth SR, İşcan MY. 1988. Skeletal aging techniques: the good, the bad, and the equivocal.
American Journal of Physical Anthropology 75:241.
Loth SR, İşcan MY. 1989. Morphological assessment of age in the adult: the thoracic region. In:
İşcan MY. Age markers in the human skeleton. Springfield, IL: Charles C Thomas. p.105–135.
Loth SR, İşcan MY. 1990. A systematic comparison of the accuracy of age estimation from the
rib and pubis. American Journal of Physical Anthropology 81:260.
Loth SR, İşcan MY, Scheuerman EH. 1994. Intercostal variation at the sternal end of the rib.
Forensic Science International 65(2):135–143.
Loth SR, İşcan MY, Wright RK. 1983. Estimation of age by component analysis of sternal rib
end metamorphoses. American Journal of Physical Anthropology 60(2):220.
Love JC. 2005. Evaluation of the sternal rib end age estimation technique using a modern
medical examiner sample. Abstract, Proceedings of the American Academy of Forensic Sciences
11:323.
Lovejoy CO, Meindl RS, Pryzbeck TR, Mensforth RP. 1985. Chronological metamorphosis of
the auricular surface of the ilium: a new method for the determination of adult skeletal age at
death. American Journal of Physical Anthropology 68(1):15–28.
Mann RW. 1993. A method for siding and sequencing human ribs. Journal of Forensic Sciences
38(1):151–155.
Mann RW, Jantz RL, Bass WM, Willey PS. 1991. Maxillary suture obliteration: a visual method
for estimating skeletal age. Journal of Forensic Sciences 36(3):781–791.

142	
  

Mann RW, Symes SA, Bass WM. 1987. Maxillary suture obliteration: aging the human skeleton
based on intact or fragmentary maxilla. Journal of Forensic Sciences 32(1):148–157.
Martrille L, Ubelaker DH, Cattaneo C, Seguret F, Tremblay M, Baccino E. 2007. Comparison of
four skeletal methods for the estimation of age at death on white and black adults. Journal of
Forensic Sciences 52(2):302–307.
McCormick WF. 1980. Mineralization of the costal cartilages as an indicator of age: preliminary
observations. Journal of Forensic Sciences 25(4):736–741.
McCormick WF, Stewart JH. 1983. Ossification patterns of costal cartilages as an indicator of
sex. Archives of Pathology and Laboratory Medicine 107(4):206–210.
McCormick WF, Stewart JH. 1988. Age related changes in the human plastron: a
roentgenographic and morphologic study. Journal of Forensic Sciences 33(1):100–120.
McCormick WF, Stewart JH, Langford LA. 1985. Sex determination from chest plate
roentgenograms. American Journal of Physical Anthropology 68(2):173–195.
McKern TW, Stewart TD. 1957. Skeletal age changes in young American males. Technical
Report E–P 45. Headquarters, Quartermaster Research and Development Command, Natick,
Massachusetts.
Meindl RS, Lovejoy CO. 1985. Ectocranial suture closure: a revised method for the
determination of skeletal age at death and blind tests of its accuracy. American Journal of
Physical Anthropology 68(1):57–66.
Meindl RS, Lovejoy CO. 1989. Age changes in the pelvis. In: İşcan MY. Age markers in the
human skeleton. Springfield, IL: Charles C. Thomas. p. 137–168.
Meindl RS, Russell KF, Lovejoy CO. 1990. Reliability of age at death in the Hamann-Todd
collection: validity of subselection procedures used in blind tests of the summary age technique.
American Journal of Physical Anthropology 83(3):349–357.
Mensforth RP, Latimer BM. 1989. Hamann-Todd collection aging studies: osteoporosis fracture
syndrome. American Journal of Physical Anthropology 80(4):461–479.
Michelson N. 1934. The calcification of the first costal cartilage among whites and negroes.
Human Biology 6:543–557.
Moore-Jansen PH. 1989. Multivariate craniometric analysis of secular change and variation
among recent North American populations. Ph.D. Dissertation, The University of Tennessee,
Knoxville.
Mulhern DM, Jones EB. 2005. Test of revised method of age estimation from the auricular
surface of the ilium. American Journal of Physical Anthropology 126(1):61–65.

143	
  

Navani S, Shah JR, Levy PS. 1970. Determination of sex by costal cartilage calcification.
American Journal of Roentgenology, Radium Therapy, and Nuclear Medicine 108(4):771-774.
Nawrocki SP. 2011. The nature and sources of error in the estimation of age at death from the
skeleton. In: Latham KE, Finnegan M. Age estimation of the human skeleton. Springfield, IL:
Charles C. Thomas Publisher. p. 79–101.
Oettlé AC, Steyn M. 2000. Age estimation from sternal ends of ribs by phase analysis in South
African blacks. Journal of Forensic Sciences 45(5):1071–1079.
Peers I. 1996. Statistical analysis for education and psychology researchers. Bristol, PA: Taylor
& Francis.
Potter WE. 2010. Evidence for a change in the rate of aging of osteological indicators in
American documented skeletal samples. Dissertation, The University of New Mexico.
Quigley C. 2001. Skulls and skeletons: human bone collections and accumulations.
Jefferson, NC: McFarland & Company.
Rao NG, Pai LM. 1988. Costal cartilage calcification pattern – a clue for establishing sex
identity. Forensic Science International 38(3–4):193–202.
Rejtarová O, Hejna P, Rejtar P, Bukač J, Slížová D, Krs O. 2009a. Sexual dimorphism of
ossified costal cartilage. Radiograph scan study on Caucasian men and women (Czech
population). Forensic Science International 191(1–3):110.e111-110.e115.
Rejtarová O, Hejna P, Soukup T, Kuchař M. 2009b. Age and sexually dimorphic changes in
costal cartilages. A preliminary microscopic study. Forensic Science International 193(1–3):7278.
Riebel F. 1929. Ossification of costal cartilages: their relation to habitus and disease. American
Journal of Roentgenology and Radium Therapy 21:44–47.
Ritz-Timme S, Cattaneo C, Collins MJ, Waite ER, Schütz HW, Kaatsch HJ, Borrman HIM.
2000. Age estimation: the state of the art in relation to the specific demands of forensic practise.
International Journal of Legal Medicine 113(3):129–136.
Rösing FW, Graw M, Marré B, Ritz-Timme S, Rothschild MA, Rötzscher K, Schmeling A,
Schröder I, Geserick G. 2007. Recommendations for the forensic diagnosis of sex and age from
skeletons. HOMO – Journal of Comparative Human Biology 58(1):75–89.
Russell KF, Simpson SW, Genovese J, Kinkel MD, Meindl RS, Lovejoy CO. 1993. Independent
test of the fourth rib aging technique. American Journal of Physical Anthropology 92(1):53–62.

144	
  

Sauer NJ. 1992. Forensic anthropology and the concept of race: if races don’t exist, why are
forensic anthropologists so good at identifying them? Social Science and Medicine 34(2):107–
111.
Saunders CF. 1966. Sexing by costal cartilage calcification – correspondence. The British
Journal of Radiology 39:233.
Saunders SR, Fitzgerald C, Rogers T, Dudar C, McKillop H. 1992. A test of several methods of
skeletal age estimation using a documented archaeological sample. Canadian Society of Forensic
Science Journal 25(2):97–117.
Schmitt A, Murail P. 2004. Is the first rib a reliable indicator of age at death assessment? Test of
the method developed by Kunos et al (1999). Homo 54(3):207–214.
Semine AA, Damon A. 1975. Costochondral ossification and aging in five populations. Human
Biology 47(1):101–116.
Senac MO, Lee FA, Gilsanz V. 1985. Early costochondral calcification in adolescent
hyperthyroidism. Radiology 156(2):375–377.
Sheskin DJ. 2004. Handbook of parametric and nonparametric statistical procedures. Third
edition. Boca Raton, FL: Chapman & Hall/CRC Press.
Smay D, Armelagos G. 2000. Galileo wept: a critical assessment of the use of race in forensic
anthropology. Transforming Anthropology 9(2):19–29.
Sokal RR, Rohlf FJ. 2012. Biometry: the principles and practice of statistics in biological
research. Fourth edition. New York: W.H. Freeman and Company.
Sokoloff L. 1983. Aging and Degenerative Diseases Affecting Cartilage. In: Hall BK. Cartilage,
volume 3: biomedical aspects. The University of Michigan: Academic Press. p. 109–141.
Steiger JH. 1980. Tests for comparing elements of a correlation matrix. Psychological Bulletin,
87(2):245–251.
Stewart JH, McCormick WF. 1983. The gender predictive value of sternal length. The American
Journal of Forensic Medicine and Pathology 4(3):217–220.
Stewart JH, McCormick WF. 1984. A sex- and age-limited ossification pattern in human costal
cartilages. American Journal of Clinical Pathology 81(6):765–769.
Stuart J, Konigsberg L. 2004. Sternal rib standards for age estimation in Balkan populations: an
evaluation of U.S. standards using alternative statistical methods. Abstract, Proceedings of the
American Academy of Forensic Sciences 10:279.

145	
  

Suchey JM. 1979. Problems in the aging of females using the Os pubis. American Journal of
Physical Anthropology 51(3):467–470.
Suchey JM, Wiseley DV, Katz D. 1986. Evaluation of the Todd and McKern-Stewart methods
for aging the male os pubis. In: Reichs KJ. Forensic osteology: advances in the identification of
human remains. Springfield, IL: Charles C. Thomas. p 33–67.
Tersigni-Tarrant MA, Shirley NR. 2012. Brief history of forensic anthropology. In: TersigniTarrant MA, Shirley NR. Forensic anthropology: an introduction. Boca Raton, FL: CRC Press. p.
1–16.
Todd TW. 1920. Age changes in the pubic bone: I, the white male pubis. American
Journal of Physical Anthropology 3(3):285–334.
Todd TW. 1921. Age changes in the pubic bone. American Journal of Physical
Anthropology 4(1):1–70.
Todd TW, Lyon DW. 1924. Endocranial suture closure: its progress and age relationship, part I.
American Journal of Physical Anthropology 7(3):325–384.
Todd TW, Lyon DW. 1925a. Endocranial suture closure: its progress and age relationship, part
II. American Journal of Physical Anthropology 8(1):23–45.
Todd TW, Lyon DW. 1925b. Endocranial suture closure: its progress and age relationship, part
III. American Journal of Physical Anthropology 8(1):47–71.
Todd TW, Lyon DW. 1925c. Endocranial suture closure: its progress and age relationship, part
IV. American Journal of Physical Anthropology 8(2):149–168.
Torwalt, C, Hoppa RD. 2005. A test of sex determination from measurements of chest
radiographs. Journal of Forensic Sciences 50(4):785-790.
Ubelaker DH. 1978. Human skeletal remains: excavation, analysis, interpretation. Third edition.
Chicago, IL: Aldine Publishing Company.
Ubelaker D, Hunt D, Owsley D, Marks M, Berryman H, Smith O, Symes S, McKeown A,
Bennett J, Jantz R. 1995. The influence of William M. Bass III on the development
of American forensic anthropology. Journal of Forensic Sciences 40(5):729–734.
Vastine JH, Vastine MF, Arango O. 1948. Genetic influence on osseous development with
particular reference to the deposition of calcium in the costal cartilages. American Journal of
Roentgenology and Radium Therapy 59(2):213–221.
Verzeletti A, Cassina M, Micheli L, Conti A, De Ferrari F. 2010. Age estimation from the rib by
components method analysis in white males. American Journal of Forensic Medicine and
Pathology 3(1):27–33.

146	
  

Webb NM, Walsh-Haney HA, Shepard KL, Herrick CE, Butler AL, Coburn MU, Arruza M.
2010. Dead man’s curve: how scoliosis affects rib aging. Abstract, Proceedings of the American
Academy of Forensic Sciences 16:403–404.
Wescott DJ. 1996. The effect of age on the sexual dimorphism in the adult cranial base and
upper cervical region. Masters Thesis, Wichita State University.
Wiredu EK, Kumoji R, Seshadri R, Biritwum RB. 1999. Osteometric analysis of sexual
dimorphism in the sternal end of the rib in a West African population. Journal of Forensic
Sciences 44(5):921–925.
Wright RK, İşcan MY, Loth SR. 1984. Determination of age from the rib in white males.
American Journal of Physical Anthropology 63:236–237.
Yavuz MF, İşcan MY, Çöloğlu AS. 1998. Age assessment by rib phase analysis in Turks.
Forensic Science International 98(1–2):47–54.
Yoder C, Ubelaker D, Powell JF. 2001. Examination of variation in sternal rib end morphology
relevant to age assessment. Journal of Forensic Sciences 46(2):223–227.

147