INVESTIGATING THE ROLE OF DIET IN U.S. CRANIOFACIAL SECULAR TRENDS By KATHERINE E. SKORPINSKI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014
2014 Katherine E. Skorpinski
To my parents, for their endless support and encouragement
4 ACKNOWLEDGMENTS This research would not have been possible without the collective efforts of many who have provided invaluable support, guidance, and sanity saving distractions. I would first like to thank my committee members, David Daegling, Michael Warren, John Krigbaum, and Sean Patrick Adams, for their always helpful input throughout this process. I am supremely grateful for the motivation, encouragement, understanding, and ever patient mentoring provided by Dr. Daegling. I am also forever indebted to Mike Warren for the guidance and opportunities he has provided me at the C.A. Pound Human Identification Laboratory throughout my time at the University of Florida , as well as trusting me with lab equipment that allowed me to complete this research. I would like to thank Dr. David Hunt for granting access to the Robert J. Terry Anatomical Skeletal Collection, and providing valuable assistance and suggestions regarding my research. I would also like to thank Dr. Dawnie Steadman and the research request review panel for providing access to the William M. Bass Donated Skeletal Collection at the University of Tennessee and assisting in acquiring approval for the use of radiographic equipment. This research was additionally made possible through financial support from the William R. Maples Award through the UF Department of Anthropology and the Doctoral Research Travel Award through the UF Graduate Sch ool. I am thankful for my fellow students at the CAPHIL, both past and present. I have had a great time learning, teaching, laughing, whining, and working with all of you. Special thanks go to Laurel, Carlos, Nicolette, and Traci, who have been tremendo us mentors and even better friends. I hope that someday, many years from now, we find a
5 way to play a round or two of MarioKart at an American Academy of Forensic Sciences Meeting. Last but certainly not least, I would like to thank my family. My parent s have provided unconditional love, support, and patience throughout this whole process. I am additionally grateful for the yearly box of muchneeded distraction that has appeared in the mail around my birthday from my brother Alex. I would also like to apologize for repeatedly graduating in August, when December would have been a much more ideal time to escape the lovely Illinois weather.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES ........................................................................................................ 12 ABSTRACT ................................................................................................................... 15 CHAPTER 1 INTRODUCTION .................................................................................................... 17 Secular Trends in the United States ....................................................................... 17 Possible Causes ..................................................................................................... 18 The Dietary Hypothesis ........................................................................................... 22 Research Question ................................................................................................. 23 Chapter Organization .............................................................................................. 23 2 U.S. DIETARY HISTORY ....................................................................................... 25 Changes in Food Processing .................................................................................. 25 Cultural and Historical Factors ................................................................................ 29 Population Movement ....................................................................................... 29 The Influence of War on Food Consumption .................................................... 30 The Great Depression ...................................................................................... 33 Nutrition Science and Dieting ........................................................................... 34 No n Diet Related Masticatory Activities ........................................................... 37 Effects of Socioeconomic Status ............................................................................ 39 Summary ................................................................................................................ 42 3 CRANIOFACIAL FORM AND MASTICATORY FUNCTION ................................... 46 Craniofacial Development ....................................................................................... 46 The TMJ as a Load Bearing Structure .................................................................... 50 Direction of Muscle Force ................................................................................. 51 TMJ An atomy ................................................................................................... 52 Joint Inefficiency ............................................................................................... 54 Experimental Evidence in Favor of TMJ Loading ............................................. 55 Evidence Linking Masticatory Function to Craniofacial Form .................................. 58 Humans with Different Subsistence Strategies ................................................. 58 Non human Species with Variations in Diet ...................................................... 62 Variation in Function Due to Tooth Loss and Attrition ...................................... 64 Experimental Studies ....................................................................................... 64 Summary .......................................................................................................... 68
7 4 MATERIALS AND METHODS ................................................................................ 69 Approach ................................................................................................................ 69 Sample .................................................................................................................... 69 Morphometric Methods ........................................................................................... 71 Radiographic Methods ............................................................................................ 73 Tooth Loss and Dental Wear .................................................................................. 77 Temporomandibular Joint Pathology ...................................................................... 80 Bilateral Symmetry .................................................................................................. 82 Collection Effect ...................................................................................................... 83 Statistics ................................................................................................................. 84 Null Hypotheses ...................................................................................................... 85 Alternative Hypotheses ........................................................................................... 86 5 RESULTS ............................................................................................................. 100 Preliminary Tests .................................................................................................. 100 Nonhuman Primates ....................................................................................... 100 Tooth Loss ...................................................................................................... 101 Tooth Wear ..................................................................................................... 103 Variation Between Populations ....................................................................... 104 Variation Between Collections ........................................................................ 106 Change Over Time ............................................................................................... 107 Morphometric Variables .................................................................................. 107 Correlation with Skull Dim ensions .................................................................. 110 Radiographic Variables .................................................................................. 113 6 DISCUSSION ....................................................................................................... 204 Preliminary Tests .................................................................................................. 204 Nonhuman Primates ....................................................................................... 204 Tooth Loss ...................................................................................................... 206 Tooth Wear ..................................................................................................... 208 Variation Between Populations ....................................................................... 209 Variation Between Collections ........................................................................ 211 Change Over Time ............................................................................................... 211 Joint Dimensions and Eminence Slope .......................................................... 211 Covariation with Mandibular and Cranial Dimensions .................................... 215 Bone Density .................................................................................................. 217 Complicating Factors ............................................................................................ 219 7 SUMMARY AND CONCLUSIONS ........................................................................ 226 LIST OF REFERENCES ............................................................................................. 230 BIOGRAPHICAL SKETCH .......................................................................................... 245
8 LIST OF TABLES Table page 4 1 Sample composition for pre1920 and post 1920 groups. .................................. 89 4 2 Sample composition by age. ............................................................................... 89 4 3 Morphometric measurements recorded. ............................................................. 92 4 4 Percent pigment/ink percentages used to calibrate radiographs. ....................... 92 4 5 Descriptions of wear stages from Smith (1984). ................................................. 97 4 6 Paired t tests of temporomandibular joint variables comparing left and right sides. .................................................................................................................. 99 5 1 ANOVA comparing linear condylar measurements between monkey species. 116 5 2 Tukeyâ€™s Honestly Significant Difference test comparing mediolateral condylar index between monkey species. ....................................................................... 117 5 3 Descriptive statistics for condyle graysc ale values for monkey species. .......... 118 5 4 ANOVA comparing gr ayscale values of monkey species. ................................ 119 5 5 Independent samples t tests comparing joint dimensions and eminence slopes between groups with varying degrees of dental loss.. ........................... 120 5 6 Descriptive statistics for grayscale values by dental status. ............................. 126 5 7 ANCOVA results for grayscale values and dental status, with age as a covariate. .......................................................................................................... 127 5 8 MannWhitney U Tests comparing mean posterior tooth wear between preand post 1920 groups. ...................................................................................... 130 5 9 MannWhitney U Tests comparing mean anterior tooth wear between preand post 1920 groups. ...................................................................................... 130 5 10 Regression results for mean posterior wear and year of birth. ......................... 133 5 11 Regression results for mean anterior wear and year of birth. ........................... 133 5 12 ANOVA results compari ng TMJ morphology between populations. ................. 140 5 13 Sample sizes for examination of collection effect. ............................................ 147
9 5 14 Independent samples t tests comparing TMJ variables between collections for white females. ............................................................................................. 148 5 15 Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 black female groups. ............................... 148 5 16 Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 black male groups. .................................. 149 5 17 Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 white female groups. ............................... 149 5 18 Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 white male groups. .................................. 150 5 19 Regression results for mediolateral condylar width index and year of birth. ..... 163 5 20 Regression results for condylar lateral anteroposterior index and year of birth. ................................................................................................................. 163 5 21 Regression results for condylar central anteroposterior index and year of birth. ................................................................................................................. 164 5 22 Regression results for condylar medial anteroposterior index and year of birth. ................................................................................................................. 164 5 23 Regression results for eminence mediolateral width index and year of birth. ... 165 5 24 Regression results for temporal lateral anteroposterior index and year of birth. ................................................................................................................. 165 5 25 Regression results for temporal central anteroposterior index and year of birth. ................................................................................................................. 166 5 26 Regression results for temporal medial anteroposterior index and year of birth. ................................................................................................................. 166 5 27 Regression results for lateral eminence slope and year of birth. ...................... 167 5 28 Regression results for central eminence slope and year of birth. ..................... 167 5 29 Regression results for medial eminence slope and year of birth. ..................... 168 5 30 Summary of variables with statistically significant change over time for linear measurements and eminence slopes. .............................................................. 175 5 31 Significant r values for TMJ dimensions and cranial length index. .................... 176 5 32 Significant r values for TMJ dimensions and cranial breadth index. ................. 176
10 5 33 Significant r values for TMJ dimensions and cranial height index. .................... 177 5 34 Significant r values for TMJ dimensions and palate breadth index. ................. 1 77 5 35 Significant r values for TMJ dimensions and palate length index. .................... 178 5 36 Significant r values for TMJ dimensions and symphysis thickness index. ........ 178 5 37 Significant r values for TMJ dimensions and symphysis height index. ............. 179 5 38 Significant r values for TMJ dimensions and corpus height index at mental foramen. ........................................................................................................... 179 5 39 Significant r values for TMJ dimensions and corpus thickness index at M1/M2. 180 5 40 Significant r values for TMJ dimensions and corpus height index at M1/M2. ..... 180 5 41 Significant r values for TMJ dimensions and bigonial width index. ................... 181 5 42 Sig nificant r values for TMJ dimensions and bicondylar breadth index. ........... 181 5 43 Significant r values for TMJ dimensions and minimum ramus breadth index. .. 182 5 44 Significant r values for TMJ dimensions and ramus height index. .................... 182 5 45 Significant r values for TMJ dimensions and mandibular length index. ............ 183 5 46 Summary of changes over time in covariation between TMJ dimensions and skull dimensions. .............................................................................................. 183 5 47 ANCOVA results for grayscale values for black females, with age as a covariate. .......................................................................................................... 184 5 48 ANCOVA results for grayscale values for black males, with age as a covariate. .......................................................................................................... 185 5 49 ANCOVA results for grayscale values for white females, with age as a covariate. .......................................................................................................... 186 5 50 ANCOVA results for grayscale values for white males, with age as a covariate. .......................................................................................................... 187 5 51 Tests of interaction between age and time period. ........................................... 188 5 52 Regression results for lateral condylar grayscale mean and year of birth. ....... 194 5 53 Regression results for central condylar grayscale mean and year of birth. ...... 194 5 54 Regression results for medial condylar grayscale mean and year of birth. ....... 195
11 5 55 Regression results for overall condylar grayscale mean and year of birth. ....... 195 5 56 Regression results for ulna grayscale mean and year of birth. ......................... 196 5 57 Summary of variables with statistically significant change over time for grayscale variables. .......................................................................................... 203
12 LIST OF FIGURES Figure page 2 1 Timeline of U. S. dietary history ........................................................................... 45 4 1 Samp le composition by year of birth ................................................................... 89 4 2 Mediolateral measurement a nd position of slopes and anteroposterior dimensions of t he temporal aspect of the joint .................................................... 90 4 3 Example of a tracing of a slope .......................................................................... 91 4 4 Example illustrating the procedure for recording coordinates in Im ageJ ............ 91 4 5 Example illustrating the selection of pigment/ink percentages in Adobe Photoshop ......................................................................................................... 93 4 6 Example illustrating the histogram of grayscale values using the curves tool in Adobe Photoshop. ....................................................................................... 94 4 7 Example illustrating the adjustment of radiographs using the curves tool in Adobe Photoshop ........................................................................................... 95 4 8 Selection areas used to acquire mean grayscale values .................................... 96 4 9 Diagram of the Eichner classification for tooth loss. ........................................... 98 5 1 Scatter plot of mediolateral con dylar index of monkey species. ....................... 116 5 2 Plot of overall condylar graysc ale values for monkey species. ......................... 119 5 3 Plot of condylar mediolateral width index by dental status. .............................. 121 5 4 Plot of condylar central anteroposterior index by dental status. ........................ 122 5 5 Plot of eminence mediolateral width index by dental status. ............................ 123 5 6 Plot of temporal central anteroposterior index by dental status. ....................... 124 5 7 Plot of central eminence slope by dental status. ............................................... 125 5 8 Plot of overall condylar grayscale mean by dental status. ................................ 128 5 9 Plot of ulna grayscale mean by dental status. .................................................. 129 5 10. Boxplots of mean post erior tooth wear by time period ....................................... 131 5 11. Boxplots of mean anterior to oth wear by time period ......................................... 132
13 5 12 Plot of wear scores for black females by year of birth. ..................................... 134 5 13 Plot of wear scores for black males by year of birth. ........................................ 135 5 14 Plot of wear scores for white females by year of birth. ..................................... 136 5 15 Plot of wear scores for white males by year of birth. ........................................ 137 5 16 Bootstrap results for mean poster ior tooth wear for black females ................... 138 5 17 Bootstrap results for mean anteri or tooth wear for black females ..................... 139 5 18 Boxplots of condylar width index by group. ...................................................... 141 5 19 Boxplots of condylar anteroposterior central index by group. ........................... 142 5 20 Boxplots of eminence width index by group. .................................................... 143 5 21 Boxplots of temporal anteroposterior central index by group. ........................... 144 5 22 Boxplots of central eminence slope by group. .................................................. 145 5 23 Boxplots of overall condylar grayscale mean by group. .................................... 146 5 24 Boxplots of ulna grayscale mean by group. ...................................................... 147 5 25 Boxplots of mediolateral condylar width index by time period. ......................... 151 5 26 Boxplots of condylar lateral anteroposterior index by time period. .................... 152 5 27 Boxplots of condylar central anteroposterior index by time period. ................... 153 5 28 Boxplots of condylar medial anteroposterior index by time period. ................... 154 5 29 Boxplots of eminence mediolateral width index by time period. ....................... 155 5 30 Boxplots of temporal lateral anteroposterior index by time period. ................... 156 5 31 Boxplots of temporal central anteroposterior index by time period. .................. 157 5 32 Boxplots of temporal medial anteroposterior index by time period. .................. 158 5 33 Bootstrap results for the temporal anteroposterior central index for black females ............................................................................................................. 159 5 34 Boxplots of lateral eminence slope by time period. ........................................... 160 5 35 Boxplots of central eminence slope by time period. .......................................... 161 5 36 Boxplots of medial eminence slope by time period. .......................................... 162
14 5 37 Plot of eminence mediolateral width index for white males by year of birth. ..... 169 5 38 Plot of temporal anteroposterior lateral index for white females by year of birth. ................................................................................................................. 170 5 39 Plot of temporal anteroposterior central index for black males by year of birth. 171 5 40 Plot of temporal anteroposterior central index for white males by year of birth. 172 5 41 Plot of temporal anteroposterior medial index for white females by year of birth. ................................................................................................................. 173 5 42 Plot of temporal anteroposterior medial index for white males by year of birth. 174 5 43 Plot of lateral eminence slope for white males by year of birth. ........................ 175 5 44 Boxplots of lateral condylar grayscale mean by time period. ............................ 189 5 45 Boxplots of central condylar grayscale mean by time period. ........................... 190 5 46 Boxplots of medial condylar grayscale mean by time period. ........................... 191 5 47 Boxplots of overall condylar grayscale mean by time period. ........................... 192 5 48 Boxplots of ulna grayscale mean by time period. ............................................. 193 5 49 Plo t of lateral condylar grayscale mean for black males by year of birth. ......... 197 5 50 Plot of central condylar grayscale mean for black males by year of birth. ........ 198 5 51 Plot of central condylar grayscale mean for white males by year of birth. ........ 199 5 52 Plot of medial condylar grayscale mean for black males by year of birth. ........ 200 5 53 Plot of overall condylar grayscale mean for black males by year of birth. ........ 201 5 54 Plot of ulna grayscale mean for white females by year of birth. ........................ 202 5 55 Plot of ulna grayscale mean for white males by year of birth. ........................... 203
15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INVESTIGATING THE ROLE OF DIET IN U.S. CRANIOFACIAL SECULAR TRENDS By Katherine E. Skorpinski August 2014 Chair: David Daegling Major: Anthropology Secular trends in the skulls of p opulations in the United States have occurred since the 1850s, and include an incr ease in cranial vault height and length and a narrowing of the cranial vault and face. Studies of the influence of dietary consistency on craniofacial development suggest th at softer diets are associated with smaller facial dimensions and narrower crania. This study investigates the possibility that advancements in food processing and other historical changes in diet have altered the biomechanical environment associated with mastication, and have influenced craniofacial development. Changes over time in morphometric and radiographic data from the temporomandibular joint (TMJ) were examined in black and white U.S. populations with dates of birth ranging from the 1850s to t he present. Variables include mediolateral and anteroposterior dimensions of the condyle and temporal aspect of the joint, the slope of the articular eminence, and condylar bone density. Tooth wear was scored for all individuals to provide morphological evidence for dietary change over time. Preliminary studies to assess the connection between TMJ morphology and masticatory behavior
16 were also performed using nonhuman primate species that vary in ingestive behavior, as well as comparisons of individuals w ith and without teeth. Preliminary studies indicate a relationship between TMJ morphology and mastication. Nonhuman primate species with tougher diets exhibited relatively wider condyles, and edentulous individuals exhibited a reduction in condylar bone d ensity. Degree of tooth wear decreased over time for all populations, supporting a shift to a softer and less demanding diet. Dimensions of the TMJ did not exhibit historical change aside from an increase in the anteroposterior dimension of the temporal aspect of the joint. Slope of the eminence decreased at the lateral aspect of the joint for white males and females only. Additionally, trabecular density in the condyle was shown to decrease over time relative to measures of systemic b one density. This study indicates some connection between U.S. dietary changes and the morphology of the temporomandibular joint. However, the subtlety of these morphological changes suggests that historical changes in dietary habits have not affected the morphology of th e rest of the skull.
17 CHAPTER 1 INTRODUCTION Change over time in skeletal morphology has been examined in a variety of populations over various temporal spans ( e.g. , Carlson and Van Gerven 1977; Goose 1981; Kaifu 1997; Little et al. 2006) . Ultimately, the underlying theme of these studies is to examine factors influencing skeletal growth and development, including genetic contributions, changes in health and nutrition, and variation in the biomechanical environment. T he relative contribut ion of these factors is dependent upon population history, which may involve changes in lifestyle, technological advancements, and migration of populations . Therefore, when studying secular trends in the skeleton, concurrent changes i n factors that influence skeletal morphology must be considered in order to understand the driving factors for these trends. Morphological changes in craniofacial form can be complex in that different dimensions of the skull may exhibit a combination of tr ends over time, with certain dimensions increasing while others decrease. This is the case for recent secular trends in the United States, which suggests that several different factors may be influencing the skeletal change in these populations. This study examines craniofacial secular trends observed in the United States in relation to one causal factor in particular: changes in masticatory function resulting from changes in diet. In general, this study aims to provide some insight into the degree of cr aniofacial plasticity in U.S. populations. Secular Trends in the United States Studies of U.S. secular trends have examined morphological change in male and female European and African Americans since the mid19th century using several
18 different appr oaches, including linear measurements on dry skulls from skeletal collections or recent forensic cases (Angel 1976; Jantz 2001; Jantz and Meadows Jantz 2000; Martin and Danforth 2009; MooreJansen 1989) , measurements on head radiographs (Hunter and Garn 1969; Smith et al. 1986) , or a two dimensional morphometric approach (Wescott and Jantz 2005) . Overall, these studies found that the degree of change in vault dimensions is greater than that observed in the face. All groups increase d in vault height, and d ecreased in the breadth of the vault and face. However, the increase in vault height in African A mericans occurred only recently, and i ncreases in cranial length and facial height were only significant in Europeans (Angel 1976; Jantz 2001; Jantz and Meadows Jantz 2000; Smith et al. 1986; Wescott and Jantz 2005) . Additionally, Wescott and Jantz (2005) clarified that the increases observed in vault height were mostly due to increases in cranial base height. Measures of facial depth (the anteroposterior dim ensions of the face) were found to increase in Europeans and decrease in African Americans (Hunter and Garn 1969; Jantz 2001; Moore Jansen 1989; Smith et al. 1986) . Mandibular changes generally mirror ed those observed in the cranium, and include a decreas e in breadth dimensions, an increase in anteroposterior dimensions, and a possible increase in vertical dimensions (Martin and Danforth 2009; Smith et al. 1986) . Therefore, overall, secular trends in the skull suggest an increase in vault height and lengt h, a narrowing of the vault and face, and an increase in vertical height in the middle and upper face. Possible Causes In general, explanations for secular change involve a combination of genetic and environmental factors, with the relative influence of t hese fact ors being populationspecific. Studies of trends in the U.S. propose several different contributing factors
19 influencing skeletal change. Some have suggested that genetic changes due to a reduction in infant and childhood mortality could affect s keletal morphology. This relaxation of selection would result in a greater variance in growth potential, and would possibly account for observed secular changes (Jantz 2001; Jantz and Meadows Jantz 2000; Martin and Danforth 2009; Wescott and Jantz 2005) . However, this assumes the existence of genetic differences between survivors and nonsurvivors with respect to craniofacial growth potential, the likelihood of which is unclear. Additionally, while a reduction in child mortality rates could explain a greater variance in morphology within a population, unidirectional change due to this factor is implausible unless selection was previously acting against certain craniofacial shapes . Admixture between populations has also been proposed to contribute to fluc tuations in skeletal form, but often appears to be insufficient to explain secular change (Angel 1976; Jantz and Meadows Jantz 2000; Wescott and Jantz 2005) . Since this explanation predicts that morphology of populations living in the same area would become more similar with increasing admixture, it does not account for instances where these populations exhibit change in the same direction. The authors of mos t studies of U nited States secular trends suggest that the most likely contributor to skeletal change involves improvements in nutrition and health over time (Jantz 2001; Jantz and Meadows Jantz 2000; MooreJansen 1989; Wescott and Jantz 2005) . In general , poor nutrition is associated with a decline in growth rates and stature (Larsen 1995) . Trends in stature in the United States , however, indicate an increase over time since the mid19th century, particularly in the length of the distal portions of the l imbs (Meadows Jantz and Jantz 1999; Trotter and Gleser 1951) .
20 Cranial vault height has been noted to follow a similar secular trend as the long bones , suggesting that their growth and development are being influenced by similar forces (Jantz and Meadows J antz 2000) . Therefore, if stature is assumed to respond to changes in nutrition and health, cranial vault height may be doing the same. Jantz and Meadows Jantz (2000) do not present comparisons of trends in stature to other craniofacial dimensions, howev er, suggesting that these measurements may not pattern as similarly to changes in the long bones. Further, no explanation is offered for the narrowing of the face and vault, which is not expected to occur with overall improvements in health. It is logic al to assume that changes over time in factors that influence skeletal growth, such as health and nutrition, would play a role in driving craniofacial secular trends. However, this suggestion is proposed without much explanation as to what exactly these health and nutritional changes entail, and why they would have such effects on the craniofacial skeleton. The rationale for this position includes a reduction in infant mortality, but it is not clear if and how this reduction has affected overall growth potential (Jantz and Meadows Jantz 2000; Wescott and Jantz 2005) . Wescott and Jantz (2005 : 242 ) state that â€œdiet has improved to the point where overnutrition has surpassed undernutrition as our most s erious malnutrition problem.â€ However, overnutrition d oes not necessarily equate to better nutrition. With developments in food processing, there has been an increase in consumption of refined sugar, vegetable oils, salt, and fatty meats which has decreased the overall nutritional quality of the U.S. diet (C ordain 2007; Cordain et al. 2005) . The effects of these nutritional changes on craniofacial growth are not clear.
21 Aside from the limitations involved with this explanation, th e re are practical difficulties in investigating the effects of altered nutrition and health on the skeletal morphology in the context of secular trends. A vailable skeletal samples in the United States are composed of individuals of low socioeconomic status who would not necessarily hav e been fortunate enough to benefit from advanceme nts in health care (Christensen 2006; Hunt and Albanese 2005) . Specific details of individual nutrition history are also largely unavailable for these collections. Any kind of chemical analysis on the bone is also not possible , since use of these collect ions must be nondestructive. A potential option for these collections would be to record skeletal indicators of stress, such as linear enamel hypoplasias (LEH), and examine if the prevalence of these indicators has changed over time. It has been stated that measurements of LEH relative to other aspects of the tooth can broadly estimate the timing when a stress event occurred , which would be useful in determining the effects of growth disruptions at different points during childhood on craniofacial morphology (Hillson 1996; Martin et al. 2008) . However, the accuracy of these methods are questionable, and standards may need to be population specific (Hillson 1996; Martin et al. 2008) . Another option would be to examine change over time using other pathological conditions. Unfortunately, many conditions do not leave any evidence on the skeleton, and evidence of disease or poor health at the time of death may not be an indicator of the health of that individual during growth and development. Even if evidence for a pathological condition is present , the same disease might affect craniofacial form differently depending on the time of onset of the disease. One could abandon the use of these collections in favor of clinical data, but this information would not be available
22 f or temporally older populations and would not include associated information on craniofacial form aside from the possibility of head radiographs. The Dietary Hypothesis Another suggestion presented to explain secular change in the U.S. is related to changes in diet ary consistency over time (Jantz and Meadows Jantz 2000; Wescott and Jantz 2005) . Specifically, this biomechanical explanation suggests that reduced masticatory stresses resulting from consumption of softer foods could alter craniofacial morphology . This has been proposed to account for craniofacial change when comparing shifts in subsistence strategies from hunting and gathering to agriculture (Carlson and Van Gerven 1977; Larsen 1995; Sardi et al. 2006), as well as in the context of more recent change concurrent with industrialization (Goose 1962; Jantz and Meadows Jantz 2000; Lavelle et al. 1971; Little et al. 2006; Martin and Danforth 2009; Rando et al. 2014) . The idea that a shift to a diet of softer consistency could co ntribute to craniofacial form in the U.S. implies significant changes in diet in the past 150 years. The nature and extent of these changes are detailed in Chapter 2. Studies linking masticatory function to craniofacial morphology have used experimental animal models ( e.g. , Beecher and Corruccini 1981a; Beecher and Corruccini 1981b; Bouvier and Hylander 1981) , comparisons of animal species with different diets ( e.g. , Taylor 2005; Taylor 2006; Taylor and Vinyard 2008) , comparisons of human groups with different subsistence strategies ( e.g. , Carlson and Van Gerven 1977; Sardi et al. 2006) , and comparisons of humans with reduced function due to tooth loss or attrition to normal dentate individuals ( e.g. , Giesen et al. 2003; Granados 1979; Hinton 1981) . Additi onally, a range of variables have been used for studies of this type, including morphometric, radiographic (or CT/MRI), and histological. These studies ,
23 which are discussed in more detail in Chapter 3, provide a strong foundation for making predictions fo r what one would expect to see in a population shift to the consumption of a lighter, more processed diet. Specifically, softer foods and a reduction in masticatory function tend to be associated with smaller facial dimensions and narrower crania. This i s consistent with the reduction in facial and cranial breadth observed in the United States, which are dimensional changes that cannot be explained by improved health and nutrition. Research Question As described above, c hange over time in cranial dimensio ns of United States populations over the past 150 years has been well documented (Angel 1976; Hunter and Garn 1969; Jantz 2001; Jantz and Meadows Jantz 2000; Martin and Danforth 2009; Moore Jansen 1989; Smith et al. 1986; Wescott and Jantz 2005). However, these studies focus on detailing the trends themselves, rather than testing factors that may inform underlying causes of this morphological change. As craniofacial form has been linked to dietary consistency in a variety of contexts , this study investigates the possibility that changes in diet, in particular advancements in food processing, have contributed to the observed craniofacial secular trends in the United States . The connection between craniofacial trends and changes in diet will be exp lored by examining trends in the internal and external morphology of a structure closely associated with the process of mastication: the temporomandibular joint. Chapter Organization This dissertation is composed of six chapters. Chapter 1 provides a brie f overview of secular trends in the United States and outlines the research question central to this study . Chapter 2 discusses factors influencing changes in diet since the
24 mid 1800s, including advancements in food processing, demographic changes, and hi storical and cultural events. Chapter 3 covers previous work investigating the link between craniofacial form and masticatory function. In Chapter 4, the sample composition, data collection methods, hypotheses, and statistical analyses used in this study are outlined. Chapter 5 presents the results of the analyses performed, while Chapter 6 discusses the implications of these results . A summary of this research is presented in Chapter 7.
25 CHAPTER 2 U.S. DIETARY HISTORY Craniofacial morphology can be influenced by a variety of factors, including sex, ancestry, and age. Outside of these influences, a prominent focus of study has been the effect of mastication on the craniofacial skeleton. This process can affect skeletal morphology by placing a variet y of stre sses on the craniofacial region, which depend on factors such as dietary consistency and frequency of mastication. Variation in dietary consistency has been demonstrated experimentally to affect craniofacial morphology in a variety of species ( e. g. , Bouvier and Hylander 1984; Ciochon et al. 1997; Dias et al. 2011) . Similar to food consistency, t he frequency of mastication may also affect temporomandibular joint morphology and craniofacial morphology overall (McGraw et al. 2011; Rubin et al. 2001; Rubin et al. 2002; Scott et al. 2014) . Eating more food more frequently places greater stress on the temporomandibular joint. In the United States, advancements in food processing and other historical and cultural factors have affected the types and amo unts of food consumed since 1850. This chapter outlines these dietary changes over time. Since a majority of the individuals included in this study are of low socioeconomic status, the impact of dietary changes on different socioeconomic groups is also discussed. Changes in Food Processing The most prominent changes to dietary consistency during this period are related to advances in food processing. Between 1859 and 1899 alone, the food processing industry expanded fifteenfold in the United States (Gab accia 1998) . P rocedures for canning fruits, meats, and vegetables were developed in France in the early 1800s, but did not boom in the U.S. until the end of that century with the introduction of Heinz
26 canning in 1876, FrancoAmerican in 1887, and Campbell â€™s condensed soup in 1898 (Bowers 2000; Dyson 2000; Gabaccia 1998; Hooker 1981) . During the Civil War, canned goods went to soldiers, sailors, and the wounded, and subsequently became better known to the general public (Hooker 1981) . Campbellâ€™s soup sales increased from a half million cans per week to 18 million cans per week between 1900 and the early 1920s (Hooker 1981) . Franco American merged with Campbellâ€™s in 1921 to get nationwide distribution (Gabaccia 1998) . Canned strained and pureed baby foods were also introduced in the 1920s (Hooker 1981) . The trend in increased canned food consumption continued through the 1900s, with Americans eating 50% more canned food and dried fruits and vegetables in 1940 than in 1930 (Dyson 2000) . Advances in mill ing also changed the texture and consistency of grain consumed during this period by removing the portions of the grain that were more difficult to process (Storck and Teague 1952) . Before the Industrial Revolution, which started in the mid1700s, cereals were ground with the use of stone milling tools, which left the entire contents of grain unless it was sieved (Cordain et al. 2005; Storck and Teague 1952) . Near the end of the 19th century, the invention of mechanized steel roller mills and automated si fting devices removed germ and bran from the grain, leaving uniformly small particulate endosperm (Cordain et al. 2005; Hooker 1981; Nelson 1985; Storck and Teague 1952) . These developments also helped to remove dirt, small sticks and stones, and other foreign matter from the grain more completely and efficiently, and allowed for different species of wheat to be utilized that were formerly too difficult to process (Storck and Teague 1952) . Storck and Teague (1952: 196) summarize these developments succinc tly:
27 In the United States during the decades after 1860 flour milling completely altered its character. It became a thoroughly mechanized, largescale industry, drawing its ever more varied wheats from ever more distant sources, submitting them to a cleaning and reduction program whose broad outlines were thoroughly standardized, producing flours and feeds of remarkable purity and uniformity, and disposing of a constantly increasing volume of production in ever expanding markets. This mechanization of techniques and concentration of milling in larger towns , such as Minneapolis and St. Paul, solved technical problems for storing, cleaning, and transporting grain and paved the way for mass produced goods (Gabaccia 1998; Storck and Teague 1952) . For ex ample, in the 1920s The Taggart Company developed Wonder Bread , which became â€œthe first bread available ready sliced as well as uniformly white and spongelike [and] rapidly became the national standardâ€ (Gabaccia 1998: 57). Aside from such mass produced g oods, the uniformly soft flour was sold directly to domestic markets starting in the 1880s and 1890s; specifically, large milling companies began packaging flour in smaller quantities for home use (Storck and Teague 1952) . The 1920s involved the formatio n of the first conglomerate food companies , including General Foods and Standard Brands, set new standards for organized food processing, which prompted a rapid expansion in food retailing (Gabaccia 1998) . Until 1920, most people in rural areas bought food from general stores that sold local produce, meat, and grain; in cities, markets sold relatively local goods, while small grocery stores sold some mass produced items (Gabaccia 1998) . With the increase in chain grocery stores dealing in mass produced it ems, the cost of most foods declined. This led to a decrease in consumption of fresh locally produced goods in favor of more refined sugar, bread, and starch products, particularly for the poor (Dyson 2000; Hooker
28 1981) . By 1956, large supermarkets that sold processed foods accounted for 63% of all grocery sales (Levenstein 2003) . Other technological advances in the latter half of the 20th century, including crock pots, electric blenders and mixers, electric carving knives, food processors, and juicers, allowed for further food processing to occur within the household (Bowers 2000; Hooker 1981) . The number of women entering the work force started increasing with changes in womenâ€™s rights in the 1920s, including gaining the right to vote. The subsequent reduction in womenâ€™s time at home led many people to take advantage of the efficiency of new technology and more processed foods (Bowers 2000) . Companies began producing foods that were quickly and easily prepared, such as TV dinners that were introduced in the 1950s. After World War II, when the number of women in the work force greatly increased, food processing companies marketed such â€œconvenience foodsâ€ and other products that made cooking easier and more efficient to working mothers by stressing the amount of time saved by using these products (Levenstein 2003) . Although these new â€œconvenience foodsâ€ decreased preparation time in the home, â€œtechniques for processing, preserving, precooking, and packaging had one thing in common: they made food lose th eir taste, texture, and normal appearanceâ€ (Levenstein 2003: 109). These problems were counteracted by the use of additives in these products, which aimed to adjust the taste and appearance of foods to a more natural and familiar condition. Between 1949 and 1959 alone, chemists produced over 400 new additives for food processing companies to improve their products (Levenstein 2003) . Vitamins and minerals were also added to these processed foods; by 1943, approximately 75% of the bread made in America used enriched flour (Levenstein 2003;
29 Levenstein 1996) . Although these processed foods became more prevalent over time, utilization of these products depended on location as well as socioeconomic status, as described further below. Cultural and Historical Factors Aside from technological advancements in food processing, various cultural and historical factors have influenced the types and amount of food consumed over time in the United States. Population movement, two World Wars, the Great Depression, and a dvances in nutrition science have all contributed to changes in the American diet. Other nondiet related masticatory activities, including tobacco and gum chewing, have also changed in frequency over time since the 1850s. Population Movement The utilizat ion of these new methods of food processing was highly dependent upon location. The movement of people from rural to urban areas resulted in a significant shift in consumption of homegrown foods to more processed storebought foods (Bowers 2000; Gabaccia 1998; Hooker 1981) . In 1900, 60% of the U.S. population lived in rural areas, in contrast to the current 25% , with the shift to an urban majority occurring by 1920 (Beale 2000) . The primary consequence of moving from a primarily agricultural society to an industrial one involves a more sedentary lifestyle. Less manual labor would require less energy, and thus less eating overall (Levenstein 1988) . The diet of 19th century rural Americans consisted of multiple heavy meals per day centered around meat, potatoes, and grain; when vegetables were consumed, they were cooked to excess (Dyson 2000; Gabaccia 1998; Hooker 1981) . By 1910, per capita calorie consumption began to decline due to a drop in consumption of cereals,
30 breads, and potatoes, as well as an overall decrease in manual labor. In addition, meals at breakfast continued to become smaller in quantity and were often eliminated altogether (Hooker 1981; Viet 2013) . The rising popularity of different kinds of foods also contributed to this overall decline in calorie consumption. For example, pancakes, or buckwheat cakes , became very popular around the end of the 19th century, which helped replace heavy meals involving steaks ( Hooker 1981) . During the 20th century, the gap between rural and urban diets narrowed over time as advertising reached more homes throughout the country. Canned food became high status food in rural areas in the 1930s, and â€œeven those who could little af ford them sacrificed to purchase themâ€ (Levenstein 2003: 27). By 1932, time spent preparing food in homes in the city and in rural areas was largely the same; any additional time spent in the kitchen on farms was due to making homemade bread (Levenstein 2003) . The Influence of War on Food Consumption Times of war over the past 150 years have impacted the American diet through food shortages and restrictions . The Civil War in the 1860s brought food shortages, particularly in the South (Hooker 1981) . In 1 917, the United Stated Food Administration was created, headed by Herbert Hoover, to manage and ship food supplies to aid those involved in World War I; the formation of this association was also the first attempt by the U.S. to manage food on a national s cale (Viet 2013) . R estrictions were placed on what the public should eat in order to save red meat, butter, wheat flour, and sugar . The consumption of such foods was discouraged by promoting w heatless Mondays and Wednesdays, meatless Tuesdays, and p orkle ss Thursdays and Saturdays. To compensate, the consumption of potatoes, oatmeal, and beans was promoted instead (Hooker 1981; Levenstein 1996; Viet 2013) . While the pleasure of eating was
31 downplayed during this time, â€œrationalâ€ food choices that fulfilled basic nutritional needs for low cost were emphasized. For example, peanut butter and cottage cheese served as cheap sources of protein and calcium. Such â€œsubstitute foodsâ€ were suggested as a result of developments in nutrition science in the early 1900s, and the subsequent spread of knowledge about the vitamin and mineral content of different foods (Levenstein 1996; Viet 2013) . In general, people embraced the opportunity to make â€œrationalâ€ food choices (Viet 2013) . In 1917 and 1918, compared to six years before the start of the war, consumption of restricted foods all decreased, while consumption of the encouraged substitute foods increased. Although food restrictions and substitutions were voluntary, many Americ ans participated; approximately 70% of American families were involved in food conservation during the war. If able, African Americans embraced the food conservation movement in an effort to dispel racial prejudice and prove their patriotism. The presidents of African American colleges led local food conservation efforts in this population (Viet 2013). The enthusiasm to participate in food conservation efforts was due in part to social pressures to comply with these restrictions. A lack of participation was associated with a lack of both self control and patriotism (Viet 2013) . The newfound association between eating less and good health was also incorporated into the food conservation campaign; eating less would not only help the food conservation mov ement, but it was a matter of public health. Food administrators enlisted the help of preachers, rabbis, and priests to help spread the idea that food conservation was important and a moral thing to do. Through these efforts of the U.S. Food
32 Administrati on, obesity became associated with laziness, gluttony, and immorality, and was a clear sign to others that participation in the food conservation movement was lacking (Levenstein 2003; Viet 2013) . Food conservation strategies were much different during Wor ld War II. There was less emphasis on conserving food, and more encouragement for Americans to â€œeat more vitamins and minerals to improve their mental and physical alertnessâ€ (Levenstein 2003: 66). This movement had much less of an impact compared to the food conservation movement during World War I since the new Office of Defense Health and Welfare Services relied more on nutritional and social scientists rather than those from the advertising industry, as was the strategy of the U.S. Food Administration. Certain foods were rationed during the war, such as canned meats and fish, but this prompted the public to hoard these items. False rumors of rationing certain foods also led to shortages due to hoarding, which in turn caused those products to actually be rationed. In general, Americans were much less likely to comply with any effort to conserve or ration food compared to World War I (Levenstein 2003) . When food s such as meat, butter, and sugar were scarce during World War II, many returned to home can ning and home grown vegetables ( i.e. , â€œvictory gardensâ€) , as well as processed foods that saved time and effort (Bowers 2000; Hooker 1981; Levenstein 2003) . Meat shortages led to the increased consumption of eggs, milk, beans, and macaroni and cheese (Lev enstein 2003) . The lower class often ate better during the war due to increased incomes associated with war work, as well as rationing and price controls that made more nutritious foods more accessible to these groups. Between 1936 and 1948, consumption of protein increased by 30%, consumption of
33 calcium increased by 60%, and consumption of iron increased by 61% for the poorest third of the urban population; these increases were much greater compared to those in the middle and upper class (Levenstein 2003) . After the war, people resumed a heavy consumption of meat, but their use of fresh fruits and vegetables declined in favor of canned and frozen varieties (Hooker 1981; Levenstein 2003; Putnam 2000) . For example, in 1919, 150 pounds per capita of fresh fruit were consumed compared to 30 pounds per capita of processed fruit; in 1998, the amount of fresh fruit consumed decreased to 140 pounds per capita, while the amount of processed fruit increased to 155 pounds per capita (Putnam 2000) . The Great Depression During the Great Depression, between and of the U.S. population was left unemployed and many turned to soup kitchens, breadlines, and garbage dumps for nourishment (Hooker 1981; Levenstein 2003) . Between 1929 and 1933, the average family income in America dropped by 40%. Farmers and smaller food processing companies were unable to lower the prices of their products without going under, which left larger companies to produce goods that could be affordable during this time. Due to this drop in pric e, Americans ate 50% more processed fruit and vegetables in 1940 compared to 1930 (Levenstein 2003) . People who were already poor prior to 1929 were the most affected by the Depression, including southern sharecroppers, migrant farm workers in the west, co mmunities in the Appalachians, and the urban poor. There was political opposition to giving aid to the poor during this time; many cited the prevailing idea that malnutrition was a result of ignorance of nutrition rather than low income (Levenstein 2003) . Toward the end of the 1930s, however, people began to realize that income greatly influenced
34 peopleâ€™s food choices, and that rising income was associated with the consumption of more healthy foods. In 1939, the food stamp program was developed to aid the poor and improve nutrition, which mostly helped welfare cases in urban areas, and school lunch programs were started that mainly helped those in rural southern areas (Levenstein 2003) . For the â€œDepression poor,â€ meaning the newly unemployed, the number of meals per day was cut from three to two, there was a reduction of consumption in milk and fresh fruit, and eggs took the place of meat. â€œFamilies fell back on their historic staples: pasta and beans for ItalianAmericans, corn meal for Southern blacks and whites, and beans and pancakes for northern nativeborn whitesâ€ (Levenstein 2003: 61). Thus, the Depression caused many to turn to affordable processed foods and cheaper substitute staple foods, and reduce consumption overall. Nutrition Science and Di eting The latter half of 19th century involved an increased knowledge of nutrition with the discovery of vitamins and minerals (Hooker 1981). Around the turn of the century when more people transitioned into a more sedentary lifestyle, nutritionists urged Americans to scale back large heavy meals (Bowers 2000; Dyson 2000; Hooker 1981) . In the early 1900s, nutrition scientists compared foods from different cultures, regions, and classes, and demonstrated similarity based on nutrition content; this served as the basis for the food substitutions suggested during World War I in order to fu nnel resources overseas (Viet 2013) . With this new knowledge of nutrition, associations were also made between being overweight and negative health effects. Mortality studies sponsored by life insurance companies found that higher weight was linked to a shorter life expectancy. Chronic diseases such as heart disease and diabetes, which first surpassed epidemic diseases as the leading cause of death in Americans in the
35 late 1910s, were now linked to living habits. Prior to these discoveries, excess fat w as thought to be due to heredity and glandular issues rather than diet (Levenstein 2003; Viet 2013) . Due to these advancements in nutrition science, the â€œthin idealâ€ started to become popular in the 1910s, and there was an explosion of weight loss cultur e during and after World War I (Viet 2013) . Part of this new ideal body type stemmed from the food restrictions during the war, since those who were overweight were deemed â€œunpatrioticâ€ for not doing their part for the food conservation movement (Viet 201 3) . In the 1920s and 1930s, as more people became aware that weight gain resulted from consuming more calories than expended, weight loss increasingly became a matter of willpower (Viet 2013) . As visually oriented print media became more popular througho ut the 1900s, it impacted physical ideals and encouraged women in particular to lose weight; advertising doubled in the 1920s alone. Fashion changes in the 1920s, including the popularity of the thin â€œflapperâ€ style among women, involved more revealing and form fitting clothing for both men and women (Hooker 1981; Viet 2013) . By this time, most people bought clothing ready made by size rather than making clothes custom at home, which allowed people to associate different clothing sizes with different body types (Viet 2013) . In order to capitalize on the desire to lose weight, companies often promoted fad diets involving their products. For example, the United Fruit Company helped popularize a diet that focused on the consumption of bananas and skim mil k (Levenstein 2003) . Although emphasis on dieting was reduced during World War II, it regained popularity after the war. While there was a concurrent resurgence of gourmet
36 cooking after World War II, this was likely to have involved more affluent households that could afford the cost and time associated with this trend (Bowers 2000) . A return to home cooking was also spurred by questions about the safety and quality of processed foods and concerns about environmental issues and labor practices (Levenstei n 2003; Levenstein 1996; Viet 2013) . Food processing companies attempted to alleviate fears starting in the 1970s by marketing foods as â€œall natural , â€ regardless of the actual ingredients and additives used in the products (Levenstein 2003; Levenstein 1996) . Dieting gained more popularity after a 1977 report by the Senate Nutrition Committee entitled Dietary Goals for the United States (Levenstein 2003) . This report emphasized eating less of what was thought to have negative health consequences, such as f at, sugar, and salt. Meat, butterfat, eggs, and other highcholesterol foods were discouraged, while the population was urged to eat more poultry, fish, fruits, vegetables, whole grains, and nonfat milk. Consumption of these items did change in favor of these recommendations, but mainly only among the middle and upper classes (Levenstein 2003) . Despite the emphasis on dieting, obesity has been an ever increasing problem in recent decades. In 1985, the National Institutes of Health reported that 28% of a dults were considered overweight. More recently in 2000, the Centers for Disease Control and Prevention indicated an increase in obesity rates among all sociodemographic groups, especially the poor, African American, and Hispanic populations (Levenstein 2003) . Thus, while some groups have been affected by dieting trends over time, others have experienced an increase in diet related disease.
37 Non Diet Related Masticatory Activities Change over time in nondiet related activities that involve mastication, such as tobacco chewing and gum chewing, have also occurred in the United States. Since the colonial period, tobacco use varied depending on location and social class (Pampel 2004; Robert 1949) . Pipes were popular in general, while aristocrats tended to use snuff and those in rural areas used chew. Snuff involves the use of preground tobacco placed between the lip and gums or inhaled through the nostrils; chewing tobacco is moved around the mouth and held between the cheek and gums, but few vigorously chew it (Rogozinski 1990) . Only the destitute used cigarettes during the colonial period by rolling discarded bits of tobacco. In the early 19th century, snuff became unfashionable, while pipe smoking and chew remained common; overall, chewing tobacco was more popular than smoking during the mid1800s (Gately 2001; Pampel 2004; Parker 2005; Robert 1949) . Cigarettes were used during the Civil War, but the habit was not continued after the war as their use was viewed as effeminate. During this time, chew was increasingly viewed as unsanitary in urban areas, and was mostly only popular in rural areas (Gat ely 2001; Pampel 2004) . Cigars, which were more expensive, became popular among the affluent business class after the war. During the 1870s and 1880s, automation of cigarette manufacturing resulted in cigarettes becoming the dominant tobacco product by the early 20th century (Pampel 2004) . World War I helped to dispel the notion that cigarette smoking was an effeminate behavior by associating smoking with patriotic fighting. The explosion in popularity of cigarette smoking during the 1900s resulted in a steady decline in the use of snuff and chew over time, even in rural areas (Courtless 1994; Parker 2005; Robert 1949) . The use of tobacco products differed among women, however, as they generally
38 used pipes or snuff and avoided chewing tobacco entirely. As these products declined in popularity by the turn of the century, women were less likely to use any tobacco products, as tobacco use started to be viewed as a largely masculine activity (Pampel 2004) . Although the activity of gum chewing has existed pri or to its use in the United States, John Bacon Curtis of Maine was the first to have the idea to package a nd market chewing gum in 1848 (Gustaitis 1998; Robinson 2004) . European settlers in the United States adopted the habit of chewing on spruce tree res in from Native Americans, and some collected and sold the resin for extra income. However, Curtis was the first to package the resin for sale in stores. He built the worldâ€™s first chewing gum factory in Portland in 1852, which manufactured flavored paraf fin instead of spruce gum, which was dwindling in supply and lacking in good taste (Gustaitis 1998) . By the 1870s, gum was made out of chicle, and sold in both stick and gumball form. The 1880s brought the development of gumball machines, although their initial poor design tended to be unreliable and resulted in gumballs that quickly grew stale; the modern gumball machine was not introduced until 1918 (Gustaitis 1998) . William Wrigley Jr., who was perhaps the most famous entrepreneur of chewing gum, int roduced Juicy Fruit and Spearmint in 1893; Spearmint became the most popular gum in the U.S. by 1910 (Gustaitis 1998; Robinson 2004) . In the early 1920s, Wrigley made a multi year deal to advertise on billboards, posters, subways, and street cars, further spreading the popularity of gum chewing. By the end of the 1920s, Wrigley was the fifth largest advertiser in the United States (Robinson 2004) . In 1928, Frank Henry Fleer introduced bubble gum, which appealed more to children, and was soon
39 paired with the sale of trading cards in the 1930s (Gustaitis 1998; Robinson 2004) . Gum was also marketed by being wrapped in comic strips, such as Bazooka Joe, which was introduced in 1953. Rations of sugar and flavorings during World War II resulted in a decrease in chewing gum use due to lack of supply, but there was a resurgence in the demand for gum once the war ended (Gustaitis 1998) . Therefore, this period of dietary change since the 1850s has also involved the development, marketing, and rise in popularity of chewing gum. Effects of Socioeconomic Status When considering the evolution of the American diet , socioeconomic status has had a significant effect on dietary patter n s of different populations . Specifically, it has influenced the amount of processed foods consumed as opposed to more fresh, unprocessed foods, as well as the frequency and amount of food consumed overall. Some occupational differences can also be observed, but many of these are related to differences in income. Different socioeconomic groups were also differentially affected by social pressures associated with eating. The change in cost that accompanied advancements in food processing, as well as the availability and distribution of processed foods, significantly influenced the ability of different socioeconomic groups to exploit these new food sources. With new standards for organized food processing and a rapid expansion in food retailing, the cost of most foods declined in the 1920s. This led to increased availability and consumption of processed foods, particularly in refined sugar, bread, and starch products, as well as canned foods with a softer consistency (Dyson 2000; Gabaccia 1998; Jekanowski and Binkley 2000) . Th e poor were less likely to be able to afford processed food until this time, while those of higher socioeconomic status could take
40 advantage of technological advances in food processing soon after they were introduced in the late 19th century, such as highly refined grains (Cordain et al. 2005; Hooker 1981; Jekanowski and Binkley 2000) . However, the upper class was also able to afford the fresh but more expensive alternatives to products such as canned foods (Jekanowski and Binkley 2000). The poor also had limited access to in home technology that makes it possible to pr ocess and preserve food (Larkin 1988; Miller 1987) . More affluent households would have better ovens and cookware, whereas in poor households, â€œthe bread was baked in the ashesâ€ (Larkin 1988:29). Such differences extended to the simplest of utensils, such as the availability of forks and knives. Each p erson in more affluent households would have their own set of utensils, while those in poorer households often had to share (Larkin 1988) . In addition, t hose who could afford iceboxes could preserve fresh food for much longer instead of eating salted meat s (Larkin 1988; Levenstein 1988) . O ccupational differences in food consumption often depended on disparities in income. The â€œworking classâ€ in urban areas encompassed a wide range of incomes, with skilled workers earning roughly five times more than unski lled workers (Levenstein 1988) . Those that were more prosperous living in urban areas had a plentiful supply of food year round. In rural areas, the food supply was often dependent on the seasons, and could be affected by environmental conditions (Dirks 2003; Larkin 1988) . This seasonality would also affect the urban poor due to the fluctuations in prices of fresh foods associated with availability. In general , when comparing populations in the same area, the poor would eat less overall (Larkin 1988) . When they did eat, they tended to consume less meat and more meal (Gabaccia 1998) . However, rural farmers still often
41 ate better than lower paid urban workers, since a greater quantity of local fresh foods were readily available. The urban poor would oft en have to fall back on a limited diet of potatoes, cabbage, and turnips (Levenstein 1988) . Social expectations and fashionable trends related to eating also affected those of different socioeconomic status differently. In more affluent households, wome n were expected to be daintier eaters (Levenstein 1988) . Fashionable diet trends would also be more likely to affect more affluent populations, since they could afford such luxuries without sacrificing their health (Levenstein 1988) . For example, larger figures in women were the more ideal body type at the end of the 19th century, whereas the thin â€œflapperâ€ style became popular among women in the 1920s (Hooker 1981; Levenstein 1988) . However, those of lower socioeconomic status did not elude diet related social pressures. During World War I, the poor were largely unable to participate in food conservation efforts, as they had very little to give up. Additionally, the foods that were suggested as substitutes for restricted foods tended to be more expensi ve (Viet 2013). However, the food conservation movement attempted to reach the lower class through religious means by enlisting clergy to stress the moral importance of avoiding wastefulness, though it is not clear what proportion of those in lower socioeconomic groups actually participated in these efforts. Nutrition studies also found that â€œpoor Americans were more often moved by considerations of status than by economy in choosing and preparing foodsâ€ (Levenstein 2003: 70). Therefore, although diet re lated trends and ideal body types may not have necessarily affected those of lower socioeconomic status, these groups were not immune to social pressures related to eating.
42 The willingness to try unfamiliar foods also varied between different groups. â€œC ul tural conservatism,â€ which was most common in immigrants and the poor, involved a reluctance to consume ready to eat foods in favor of foods more traditional to their culture and country of origin (Gabaccia 1998) . However, immigrants started Americanizing diets in the mid1900s, largely due to pressure from their children born and raised in the U.S. (Levenstein 2003) . Many Americans were also reluctant to incorporate ethnic and foreign foods into the American diet in the late 1800s and early 1900s. While this is due, in part, to unfamiliarity with these foods and an aversion to stronger spices, the eugenics movement in the early 1900s played a role in changing peopleâ€™s ideas about the effects of food and nutrition on the body (Viet 2013) . The new discipline of euthenics suggested that environmental factors, and diet in particular, could affect physical and mental development; since different races ate different foods, these differences in diet were presumed to play a role in supposed racial inequality. I n short, some people were reluctant to try foreign foods out of fear of negative developmental effects associated with â€œlowerâ€ races (Viet 2013) . This view changed over time, however, as spaghetti and pasta became integrated into the American diet in the 1920s and 1930s, and interest in other ethnic and foreign foods increased in the 1970s (Levenstein 2003) . Summary Since the 1850s, food consumption in the United States has changed in several ways, which are summarized in Figure 21. Developments in food processing involved the mass production of canned goods and highly refined grains, and the invention of inhome technological advances, such as crock pots and food processors. Population movement to urban areas resulted in a decrease in overall food consumption that
43 accompanied a more sedentary lifestyle, as well as greater access to mass produced foods. Historical events, including two World Wars and the Great Depression, resulted in food shortages and restrictions. Advancements in nutrition science led to an increase in dieting and the persistence of the â€œthin ideal.â€ Changes in non diet related masticatory activities have occurred since the 1850s, including an increase in popularity of gum chewing and a reduction in chewing tobacco use. However, the impact of changes in chewing tobacco habits on temporomandibular joint morphology was likely small in comparison to diet related changes, since there is little chewing involved in its use. In combination, these various changes have generally resulted in a shift to a lighter diet consisting of more processed foods, although the extent of this shift varied by socioeconomic status. As can be seen in Figure 21 and will be explained further in Chapter 4, the 1920s involved a number of events in the U.S. relat ed to a shift to a more processed diet which were followed by the Great Depression in the 1930s. Therefore, in this study, the year 1920 was used as a turning point to a lighter diet consisting of more processed foods . In terms of its effect on skeletal growth and development, a shift to softer diet and a reduction in food consumption over time would both affect temporomandibular joint morphology in the same way. Specifically, a shift from frequent heavy meals to fewer meals consisting of both a greater proportion of processed food and a smaller quantity of food would result in fewer chewing cycles and a reduction in bite force. These changes would place the temporomandibular joint under lower stress, which would affect bone growth and remodeling. The nature of the morphological effects of these shifts in masticatory function is outlined in Chapter 3.
44 Since different socioeconomic groups have experienced different patterns of dietary changes over time, socioeconomic factors need to be considered in this study. The Terry Collection at the Smithsonian National Museum of Natural History provided individuals born in the mid19thand early 20thcenturies. The majority of the skeletons in the collection are from unclaimed cad avers used in medical school anatomy classes at Washington University in St. Louis , which tend to be individuals from les s well off segments of society (Hunt and Albanese 2005) . The contemporary portion of the sample was composed of individuals from the W illiam M. Bass Donated Collection housed at the University of Tennessee, Knoxville. Historically, many of these donations are also unclaimed cadavers from local Medical Examinerâ€™s Offices, which would also represent individuals of low socioeconomic status (Christensen 2006) . However , this collection has had an influx of self donations over the past decade. Therefore, when interpreting recent skeletal change involving individuals from the Bass Collection, this possible variation in socioeconomic status wi ll be considered.
45 Figure 21. Timeline of U.S. dietary history. Chewing tobacco popular Civil War Advances in milling Bazooka Joe Campell's soup Wrigley introduced Juicy Fruit and Spearmint WWI food conservation Women gained the right to vote -population shift to urban majority -Franco American/Campbell's soup merge -"flapper" style popular Wonder Bread -canned baby food -conglomorate food companies formed -increase in chain grocery stores Great Depression WWII victory gardens -food stamp and school lunch programs Bubble gum introduced Processed food additives Natural food movement Heinz canning Franco American canning Flour sold to domestic markets Increased in home technology TV dinners Per capita calorie consumption began to decline -popularization of thin ideal Increased knowledge of nutrition Dietary Goals for the United States 28% of U.S. adults overweight Cigarette manufacturing First chewing gum factory 1850 1870 1890 1910 1930 1950 1970 1990
46 CHAPTER 3 CRANIOFACIAL FORM AND MASTICATORY FUNCTION The idea that changes in dietary consistency and ingestive behaviors over time have affected the morphology of the skulls of U.S. populations relies on the connection between craniofacial morphology and biomechanical forces due to masticatory behavior. As this study aims to investigate this idea by examining the morphology of the temporomandibular joint, the function of this joint in the process of mastication needs to be addressed. T his chapter discusses the role of biomechanical and genetic influences in craniofacial development, the idea that the temporomandibular joint transmits masticatory forces between the mandible and the cranium, as well as the literature linking craniofacial and temporomandibular joint morphology to variations in masticatory fu nction. Due to the consistency of results in this literature, predictions can be made regarding how temporomandibular joint morphology should change with a shift to a softer diet. Craniofacial Development In order to understand how variation masticatory behavior may influence the morphology of the skull, one needs to consider the process of craniofacial growth and how development in one aspect of the skull may affect the development in another. The mechanism s of craniofacial growth and development have be en postulated by Melvin Moss and Donald Enlow, who maintained contrasting views on how this process occurs. Melvin Mossâ€™s contribution to the study of craniofacial growth is called the Functional Matrix Hypothesis, which suggests that bone development rel ies on the presence and activity of the surrounding soft tissues and functional spaces ( i.e. , the functional matrix) , and that skeletal form is a result of the developmental and biomechanical circumstances
47 in separate localized regions (Moss 1962; Moss 1973; Moss and Salentijn 1969a; Moss and Salentijn 1969b; Moss and Young 1960; Watson 1982). In contrast to Moss, Enlow asserts that no component of the craniofacial complex is developmentally self contained; normal growth and development is a composite proc ess involving concurrent change in different aspects of the skull , with growth in one area affecting adjacent skeletal structures (Enlow 1968; Enlow 1990; Enlow and Hans 1996) . The assertion that masticatory function influences craniofacial form relies on both the concept of the functional matrix, as well as the idea that growth in one area of the skull may affect the development and form of another. For example, morphology of different portions of the mandible should reflect the functions of its matrices, and include the following: alveolar (teeth), coronoid (temporalis), angular (masseter and medial pterygoid), condyloid (cartilage, lateral pterygoid), and basal (neurovascular bundle) (Moss 1962) . As the functional matrix of the condyle also includes j oint reaction force, the morphology of this structure should reflect alterations in masticatory force. These alterations should also be evident in areas of masticatory muscle attachment, s ince bone grows and remodels in response to mechanical demands (Martin et al. 1998; Rando et al. 2014; Weijs and Hillen 1984) . However, it is unclear if stress and strain due to mastication can directly affect other aspects of the cranium. In vivo masticatory strains have been found to be higher at sutures in both the f ace and vault compared to the surrounding bone (Herring and Teng 2000) . It has been suggested that the amount of strain at the sutural margins could influence growth, and thus alter the overall morphology of the face and vault (Behrents et al. 1978; Kiliaridis 1995) . However, the implicit assumption that higher stresses and strains at the cranial sutures could
48 influence growth may be incorrect, since the differences in magnitude of these stresses may simply reflect differences in the mechanical properties of the sutures compared to the surrounding bone (Enlow and Hans 1996) . Further, the distribution of strain throughout the skull varies greatly during mastication, with the highest strains occurring along the mandible and zygomatic arches, and the lowest strains occurring along the frontal superior to the orbits (Hylander and Johnson 1992; Sun et al. 2004) . Therefore, variation in diet may influence certain structures directly via the functional matrix, such as the temporomandibular joint and areas of mas ticatory muscle attachment. However, such variation may also influence the morphology of regions not directly related to mastication ( e.g. , the frontal) via Enlowâ€™s assertion that skeletal development in one region can influence the development of adjacent structures . The existence of species specific morphology suggests that skeletal development involves a significant genetic component, which bolsters the idea that craniofacial form does not depend completely upon function. Turnerâ€™s (1998:399 ) def inition of bone adaptation helps to explain this phenomenon succinctly: â€œBone cells begin with the genetic blueprint and sculpt it until the skeletal design m eets the loading requirements . â€ Although this may be oversimplified since development involves both gene expression and the epigenetic regulation of this expression, it is reasonable to conlcude that normal development and gross anatomical features are largely influenced by genetic factors (Carter and Beaupr 2001; Martin et al. 1998) . If this were not true, there would be no foundation for Darwinian evolution. M orphology, as well as the norm of reaction for bone growth and remodeling, must be heritable in order for selection to act upon it ; otherwise, organisms could skeletally adapt to any environm ental situation.
49 Most studies of heritability of human craniofacial traits find that craniofacial dimensions have a moderate to high heritability, though these studies vary widely in sample population and statistical methods employed in data analysis (Martinez Abadias et al. 2009) . Other studies have found that craniofacial variation patterns with phylogenetic history between related species ( i.e. , species that are more closely related have similar morphology), and that closely related species tend to share patterns of sexual dimorphism (Ackermann 2002; Ackermann and Cheverud 2000; Plavcan 2002) . Since phylogenetic relatedness implies genetic relatedness, similarity in morphology for closely related species provides evidence for the genetic influence of c raniofacial form. Aside from the influence of genetics, the mechanism and process of bone metabolic activity provides some insight into the relative stability of craniofacial form. Specifically, bone cells have been shown to accommodate to a routine loadi ng environment, which makes them less responsive to regular habitual loading signals. More succinctly, abnormal strains drive structural change (Turner 1998) . This is similarly expressed through Frostâ€™s mechanostat theory (Martin et al. 1998) , as well as the idea of a â€œlazy zoneâ€ for bone adaptation (Carter and Beaupr 2001) . Frostâ€™s mechanostat theory suggests that there is a minimum effective strain in bone that needs to be exceeded for an adaptive response to occur, and that there is a range of strain values that will not stimulate any adaptive response (Martin et al. 1998) . Others have referred to this range as the â€œlazy zone,â€ for which a range of tissue stresses will stimulate only slight bone apposition or resorption, whereas stresses outside this zone will result in rapid bone apposition or resorption (Carter and Beaupr 2001) .
50 When considering this information in the context of dietary changes in the United States, it seems unlikely that the resulting changes in masticatory forces would prompt considerable morphological craniofacial alterations. However, since bone adaptation and craniofacial plasticity serve to remodel skeletal structure to withstand the current loading environment, one may expect more subtle morphological differences to occur . When examining populations that vary in masticatory behaviors, even slight bone apposition and resorption may result in overall size and shape differences. Additionally, as highfrequency low level strain has been demonstrated to affect the density of t rabecular bone (Rubin et al. 2002) , subtle changes in masticatory forces may be more apparent in the internal trabecular structure of the masticatory complex. The TMJ as a Load Bearing Structure When investigating the influence of variation in masticatory forces on craniofacial morphology, one must consider structures that transmit these forces. The temporomandibular joint (TMJ), which serves as the articulation between the mandible and the cranium, is assumed to endure joint reaction forces during mastica tion. However, this assumption has not remained unchallenged. Arguments that the temporomandibular joint is not loadbearing are related to a debate between whether or not the mandible functions as a lever. In such a model, the temporomandibular joint w ould act as a fulcrum and bear a portion of the load produced during mastication. Arguments against lever action of the mandible revolve around several assertions. First, some argue that the resultant of the masticatory muscle forces always passes throug h the bite point, which would eliminate any joint reaction force at the temporomandibular joint. Second, the idea that the morphology of the temporomandibular joint would be able to withstand the forces generated during
51 mastication is disputed. Third, some suggest that since lever action of the mandible would be inefficient, such an arr angement should not exist. However, weaknesses in each of these arguments, in combination with more direct experimental evidence, suggest that the TMJ does indeed endure j oint reaction forces during mastication. Direction of Muscle Force One of the first arguments against lever action of the mandible was proposed by Wilson (1921) , who asserted that the resultant force of the masticatory muscles lies perpendicular to the occlusal plane. However, this does not mean that this resultant force necessarily passes through the bite point. In an attempt at a similar argument, Robinson (1 946) as well as Frankel and Burstein (1970) calculated that the resultant muscle force passes through the molars. However, based on this rationale, biting anywhere other than the molars would cause the mandible to act as a lever. Aside from this flaw, th e drawings used to formulate the resultant muscle force in both of these analyses used incorrect directions of the muscle forces, as well as incorrect positioning of the origin of the masseter muscle and of the dental arcade relative to the rest of the cra niofacial skeleton. The resultant muscle force would actually pass through the mandible posterior to the molars (Hylander 1975) . A major weakness in these and other attempts at determining the resultant muscle force using drawings is the subjectivity in the determination of the direction and magnitude of muscle action (Smith 1978) . In fact, even when informed estimates of these parameters is provided, determination of these aspects of individual muscles is complicated since different portions of the musc les may function at different times throughout mastication. A dditionally, electromyographic ( EMG ) activity of internal muscles like the medial pterygoid is difficult to measure (May et al. 2001; Throckmorton 1985) .
52 TMJ Anatomy The idea that the anatomy of the temporomandibular joint cannot withstand a joint reaction force has been proposed by several individuals. Robinson (1946) and Taylor (1986) suggest that the mandibular fossa is too thin to bear the stress of mastication, and that the articular disc contains blood vessels and nerves that could not be placed under compression. While these assertions are true, the relevance of these facts to the loadbearing capabilities of the joint is based on a misunderstanding of the position of the condyle and ac tual point of articulation during biting. The stress bearing portion of the temporal side of the TMJ is actually on the articular eminence , which is made of thick trabecular bone and a dense cortical plate. The portion of the articular disc at this point is actually avascular, while the nerves and vessels lie more posteriorly in the disc where compression would not occur (Hylander 1975; Picq et al. 1987) . Taylor (1986) has pointed out that the condyle is in the fossa during unilateral clenching as an arg ument against the possibility of a joint reaction force. However, much of the joint reaction force is likely carried by the balancing side condyle ( i.e. , opposite the side of clenching), which is positioned up against the articular eminence during this ac tion, and any reaction force on the working side is dissipated through the condyle, disc, and base of the articular eminence (Picq et al. 1987) . Other evidence for a higher joint reaction force on the balancing side during unilateral biting is described below ( e.g. , Hylander 1979) . Both Robinson (1946) and Tattersall (1973) argued that the fibrocartilagenous disc in the temporomandibular joint does not have stress bearing qualities. However, dynamic compression applied to the disc suggests that the disc i s actually capable of deforming and dissipating peak loads that may damage the joint (Beek et al. 2001;
53 Stankovic et al. 2013; Tanaka and van Eijden 2003) . Tattersall (1973) asserted that the condylar neck is too weak to withstand joint reaction force and the associated shear and bending stresses that would be encountered in that region. Hylander (1975) tested this assertion by determining the second moment of area of the cross section of a human condylar neck. The second moment of area measures a struct ureâ€™s resistance to bending by looking at the amount and distribution of bone around a neutral axis (Daegling and Grine 2007) . These calculations, though erring in favor of weak mechanical properties of the condylar neck, found that this region is suffici ently strong enough to withstand masticatory forces (Hylander 1975) . Other anatomy related arguments against lever action of the mandible were proposed by Frank (1950) and Taylor (1986) . While he recognized the proper placement of the condyle during biti ng using lateral head radiographs, Frank (1950) concluded that the TMJ could not act as a fulcrum because the condyle was never in direct contact with the articular eminence . However, soft tissues are less clearly visible on radiographs, and the condyle is in contact with the eminence indirectly through the articular disc (Hylander 1975) . This argument is additionally flawed in that direct boneonbone contact in any joint is an indicator of osteoarthritis, rather than normal functioning of the joint. Taylor (1986) observed that mastication can still occur after a condylectomy and concluded that condyles are not necessary for masticatory function. However, while the mandible can still function, the efficiency in c hewing is diminished to the point where the mandibular incisors, canines, premolars, and anteriormost molars cannot be brought into occlusion with the maxillary teeth (Picq et al. 1987) . Just
54 because the mandible can function, albeit abnormally, without t he condyles does not preclude normal functioning of the mandible as a lever. Joint Inefficiency A third argument against the mandible acting as a lever focuses on its inefficiency. Tattersall (1973) asserted that such a system would never have evolved, or would have not been conserved by natural selection. This argument is very weak, however, in light of the inefficiency of the rest of our joints. The vast majority or the joints in the body function as third class lever s, which are the least mechanically efficient. Gingerich (1971; 1979) admits that the mandible can and does act as a lever, but maintains that it can also function as a more efficient link between adductor muscle force and bite force. His support for this places a large importance on the function of the temporalis muscle, since it is the largest elevator of the mandible. He argues that since the temporalis is fan shaped, different areas of the muscle will contract depending on the bite point, so that the angle of the force of this muscle will pass through the bite point. For example, the most posterior fibers of the muscle should contract during anterior biting since their angulation would align the muscle force to pass through the bite point. He asserts that such a system would be more efficient than a lever system that involves a joint reaction force (Gingerich 1971; Gingerich 1979) . While this may be true, his proposed muscle function has been refuted by EMG studies of the temporalis, which found that the anterior part of the temporal is is more active during biting with the incisors, and that the temporalis is actually less active during anterior biting compared to molar biting (Hylander 1975) . Further, excluding the action of the masseter and medial pterygoid entirely ignores a signi ficant portion of muscle force that is acting in a different direction and would alter the position of the result ant muscle force. Gingerichâ€™s (1971)
55 assessment is also based on incorrect information on the movement of the mandible ( i.e. , upward and backw ard) during biting, which would cause the orientation of the bite force to be superoposterior ( i.e. , in direct opposition to the function of the temporalis) rather than vertical. It has been shown that that the teeth actually move upward and forward, resulting in a bite force orientation in a similar direction (Hylander 1975) . Upon finding more concrete evidence of the bite force direction occurring in this manner, Gingerich (1979) revised his assessment of mandibular function by asserting that it functions as both a lever (by the action of the masseter and medial pterygoid) and a link (by the action of the temporalis) simultaneously. This is based on data showing that bite force directions can fall outside of the expected value for the mandible to functio n as a â€œpure leverâ€ (Gingerich 1979) . However, these bite force directions are still not remotely close to the expected direction for a â€œpure linkâ€ model, and thus would not be in direct line with any muscle fiber direction of the temporalis. Further, the assertion that the mandible functions as both a lever and a link simultaneously is nonsensical, since the mandible can only function as a link if there is no joint reaction force. Experimental Evidence in Favor of TMJ Loading Several lines of evidence s uggest that the condyle is loaded during mastication. One method involves the analysis of in vivo bone deformation around the temporomandibular joint, since direct measurement within the joint would interfere with normal mastication. Hylander (1979) used strain gages immediately below the condyles of macaques to examine if the joint is loaded, if loads differ between the two temporomandibular joints, and when maximum and minimum loadings take place. Patterns of strain support compression reaction force i n the condyle during the power stroke of mastication and incision, as well as during isometric biting at both the incisors
56 and molars, with the contralateral side ( i.e. , balancing side) strains being higher. Strains were observed throughout the entire chewing cycle with strain during wide opening being smaller than during power stroke or isometric biting, but greater than during opening stroke of mastication. Isometric biting strains were larger than mastication strains, and isometric biting with incisors produces smaller strains than biting with the molars (Hylander 1979) . Another more invasive study involving macaques used a force transducer cemented to the anterosuperior aspect of the condyle determined that the joint is loaded during chewing, drinking, incisal biting, and vocal aggression (Boyd et al. 1990) . Chewing harder food (animal biscuits) produced greater forces compared to chewing softer foods (orange), and there was little fluctuation in forces as the chewing progressed toward swallowing (Boy d et al. 1990) . More recently, in vivo masticatory related strains were examined using mini pigs with strain gages posterior and lateral to the condyle, as well as on the zygomatic process lateral to the articular eminence (Herring and Liu 2001) . The pri ncipal compressive axis on the condyle was found to be perpendicular to the occlusal plane, and the principal tensile strain was smaller than the compressive strain. Stimulation of the ipsilateral masseter mimicked mastication strain in condyle, which fur ther indicates that the strain in the condyle is due to joint reaction force during mastication. Strains were also observed to change throughout the chewing cycle, which is likely due to the complex movements in the joint during this action. Tension exceeds compression along the zygomatic arch, indicating that it is bent and twisted. Therefore, strains in both the mandibular and temporal aspect of the temporomandibular joint are indicative of loading in this joint during mastication (Herring and Liu 2001) .
57 Human subjects have also been used to demonstrate loading at the temporomandibular joint. Using an in vitro approach, Throckmorton and Dechow (1994) attached strain gages to the anterior, posterior, medial, and lateral aspects of the condylar neck of human mandibles dissected from cadavers. Force of the superficial masseter and medial pterygoid muscles was simulated using a compression device; action of the temporalis and lateral pterygoid were not included in this study. Strains in the condyle were measured at various bite points as well as various positions of the resultant muscle force. The balancing side condyle was shown to be more heavily loaded than the working side condyle, and joint force was higher when the resultant muscle force was closer to the joint, and the bite point was further from the joint (Throckmorton and Dechow 1994) . Using living subjects, Hylander (1978) measured incisal bit e force direction in ten humans in order to respond to the assertion that bite force is equal to and in the opposite direction of the resultant masticatory muscle force, which would leave the t emporomandibular joint unloaded. This analysis confirmed that bite force was directed vertically and anteriorly, rather than posteriorly as advocated by proponents of nonlever action of the mandible. Since this bite force direction is not opposite the muscle force direction, the temporomandibular joint must be loaded (Hylander 1978) . It seems apparent that arguments that the temporomandibular joint is not loadbearing are largely based on incorrect assumptions and oversimplifications. In vivo experiments using strain gages suggest that both aspects of the temporomandibular joint are capable of loadbearing and show consistent patterns of stress during mastication. Direct determination of bite force has refuted the possibility that this force
58 acts in opposition to muscle force, which means that the temporomandibular joint must carry part of the load. Additionally, the morphology of the temporomandibular joint has been shown to vary with dietary consistency and masticatory function, which is discussed further below. This evidence indicates that the temporomandibular joint is an appropriate structure to study when assessing the influence of diet and masticatory forces on craniofacial morphology. Evidence Linking Masticatory Function to Craniofacial Form The connection between masticatory function, dietary consistency, and craniofacial morphology has been studied in a variety of species and contexts. Several studies have compared both human and non human populations with naturally occurring variations in diet. Some investigate the relationship of craniofacial morpholog y to factors that may affect masticatory function, such as tooth loss and attrition . Others use a more experimental approach involving the comparison of groups of animals raised under different dietary and masticatory conditions. Humans with Different Subsistence Strategies Evidence citing a link between dietary consistency and human craniofacial form often involves comparison of craniofacial dimensions between hunter gatherer populations and agricultural populations. The premise of these studies is that agricultural populations are presumed to have softer diets compared to hunter gatherers, and thus dietary consistency may affect craniofacial growth and form. This conclusion stems largely from the reduction in tooth wear noted with the transition to agri culture, and subsequently to post Industrial Revolution diets (Larsen 1995; Rando et al. 2014; Smith 1984) . Degree of tooth wear depends on both characteristics of the food itself and techniques of preparation, and is more severe in diets involving less
59 p rocessed foods that may contain more grit (Addy and Shellis 2006; Larsen 1995; Powell 1988) . Increased use of cooking vessels and grinding stones softened the consistency of the food consumed; however, grinding stones may also have introduced abrasive elements that resulted in more rapid tooth wear (Larsen 1995; Smith 1984) . An increase in frequency of dental caries has also accompanied these dietary shifts, resulting mainly from an increased consumption of carbohydrates (Larsen 1995) . This may also be due in part to the tendency for agricultural populations to boil plants to a gruel like consistency. This practice may promote the growth of bacteria in fissures and grooves in the teeth that are more difficult to clean. The degree of this increase in fr equency of carious lesions varies by population, since certain foods available in certain areas may be more cariogenic than others (Larsen 1995) . Changes in craniofacial dimensions with shifts in subsistence strategies have been studied in a variety of contexts, and are generally consistent in their results. An investigation of the transition to agriculture in Nubia found a decrease in corpus lengt h, ramus height, symphysis height, and coronoid process height (Carlson and Van Gerven 1977) . In the Argentine Center West, agriculturalists exhibited an overall smaller craniofacial size, as well as localized reduction in structures related to the mastic atory complex, such as the size of the attachments of the masseter and temporalis (Sardi et al. 2006) . Similarly, a study by Paschetta et al. (2010) on Ohio Valley populations found that agriculturalists exhibited shorter and narrower palates and relatively smaller temporomandibular joint dimensions. More recent craniofacial change that involved a shift from an agricultural diet to a more modern Western diet was investigated by comparing individuals from the late mediaeval period (10501150) and post medi eval
60 period (15501850) in London (Rando et al. 2014) . For both males and females, reductions in all mandibular ramus dimensions and overall width of the mandible were found; these dimensions in particular are closely tied to areas of masticatory muscle a ttachment (Rando et al. 2014) . An extensive review of similar studies found a similar overall reduction in size and robusticity of the face and mandible (Larsen 1995) . Cultural changes in food preparation and the transition to agriculture have also been tied to a reduction in the size of the dentition, particularly in the molars (Brace and Mahler 1971; Calcagno 1986; LeBlanc and Black 1974) . However, as tooth dimensions are under greater genetic control compared to other skeletal dimensions, the functional influence on these changes is unclear (Larsen 1995) . In her 2011 study, Noreen von CramonTaubadel aimed to determine the influence of changing subsistence strategies on mandibular morphology relative to other factors, such as genetics, geography, and climate. Using eleven populations , six of which were considered agricultural and five of which were categorized as hunter gatherer , the author compared population distance matrices based on mandibular shape data against matrices based on neutral genetic, geographical, climate, and subsistence data. The results of this study suggest that mandibular morphology correlates best with subsistence strategy when controlled for population history and geography (von CramonTaubadel 2011) . Essentially, transition from hunting and gathering to agriculture erases morpholog ical evidence of genetic relationships in the mandible. Agricultural populations are shown to have shorter and broader mandibles with taller, more angled rami and coronoid processes (von CramonTaubadel 2011) .
61 This morphology conflicts with general reduc tion in craniofacial features found in other studies. The inconsistencies between studies may be due to some methodological issues involved with von CramonTaubadel â€™s (2011) approach, including a lack of specificity of the dietary compositions of the popul ations used. Although she considered six aspects of subsistence economy ( i.e. , gathering, hunting, fishing, animal husbandry, agriculture, and milking), she does not describe any specific aspects of food processing or exploitation for the groups involved. Including such varied regions and groups confounds the study by not addressing cultural differences in food use, as well as access to different foods in general due to geographical location. Smaller scale studies in restricted areas might better represe nt the effects of changes to agriculture, since restricting the population in question serves as a control for cultural variation and geographic availability of different foods (Larsen 1995) . Regardless of the conflicting morphological findings, von Cramon Taubadelâ€™s (2011) study supports the idea that mandibular morphology patterns less closely with genetic factors, and may be more influenced by environmental factors , such as subsistence strategy. Several studies involving human populations with various s ubsistence strategies have focused on TMJ morphology specifically . In a Nubian population, size of the glenoid fossa of the temporal was found to be smaller in agricultural populations when compared to hunter gatherer populations, despite an overall increase in cranial size over time (Hinton and Carlson 1979) . Hinton (1983) also investigated variation in size of both the mandibular condyle and the mandibular fossa in eight different populations with a range of subsistence practices. Among aboriginal groups, dimensions of the TMJ
62 were larger in hunter gatherer groups compared to agricultural groups, with the most marked difference occurring in the anteroposterior dimension of the condyle (Hinton 1983) . Similar results were found in a comparison of an 18th century London population to other European material spanning back to the Neolithic, in which the anteroposterior dimension of the mandibular condyle has reduced over time (Whittaker et al. 1990) . However, a more recent comparison of late mediaeval and post mediaeval period individuals from London found reductions in the mediolateral width of the condyle over time, but no change in the anteroposterior dimension (Rando et al. 2014) . Although these studies differ regarding statistically significant change in mediolateral and anteroposterior TMJ dimensions, all suggest a trend toward smaller TMJ dimensions with shifts to agricultural and more modern Western diets. Non human S pecies with Variations in Diet Aside from comparisons of human groups with different subsistence strategies, diet related morphological differences have been examined in nonhuman primate samples as well. Mandibular and temporomandibular joint variation in the great apes has been investigated by Taylor (2005; 2006) and Terhune (2011b; 2013). The diets of African apes are characterized by various degrees of folivory and frugivory, with chimps and bonobos consuming a greater proportion of fruit, and gorilla species consuming a more folivorous diet (Taylor 2005) . Bonob os tend to have a greater reliance on piths and leaves year round compared to chimps, but this may vary across feeding sites (Terhune 2013) . Orangutans are thought to consume a diet at least as challenging as gorillas (Terhune 2013) . Among orangutan spec ies, Sumatran orangutans spend more time feeding on fruit than do Bornean orangutans, which eat more bark and vegetation. Additionally, Sumatran orangutans have been observed to use tools to open certain
63 fruits, while Bornean orangutans use their jaws to access the seeds (Taylor 2006; Terhune 2013) . Bonobos have relatively larger condylar surface areas and more anteroposteriorly curved condyles compared to chimps, which exhibit larger anteroposterior dimensions of the temporal aspect of the TMJ (Terhune 2013) . According to Taylor (2005) , g orillas were found to have significantly higher mandibular rami, higher condyles relative to the occlusal plane, and greater condylar width compared to chimps. Within gori lla species, mountain gorillas are the extreme folivore, with significantly higher rami and condyles and greater condylar width compared to the more frugivorous lowland gorillas (Taylor 2005) . However, Terhune (2013) did not find any condylar width differences between the two gorilla species; condylar areas were significantly greater in bonobos compared to gorillas when scaled to the distance from the condyle to the first mandibular molar, but gorillas were found to be significantly greater in condylar area than chimps when this variable was scaled to mandibular length. Compared to Sumatran orangutans, Bornean orangutans were found to be significantly larger in corpus, symphysis, and condylar dimensions, condylar surface area, and the anteroposterior dimension of the temporal aspect of the TMJ, which suggests that these species are better suited to resist larger or more frequent loads than Sumatran orangutans (Taylor 2006; Terhune 2013) . The anteroposterior length of the articular eminence was found to be relatively longer in chimps, bonobos, and orang utans compared to gorillas; this may reflect a greater capacity for condylar translation associated with a greater frequency of processing foods with the anterior teeth in these species (Terhune 2013) . Additionally, gorillas and Bornean orangutans
64 exhibit more angled articular eminences compared to chimps, which consume a less challenging diet (Terhune 2011b) . Overall, species with a tougher, more folivorous diet were found to have significantly higher mandibular rami, higher condyles relative to the occl usal plane, greater angulations of the articular eminence, and greater condylar, corpus, and symphysis dimensions compared to more frugivorous species (Taylor 2005; Taylor 2006; Terhune 2011b) . Variation in Function Due to Tooth Loss and Attrition Morphology of the TMJ has been compared between human groups with varying degrees of antemortem dental loss and attrition. Edentulous individuals produce smaller forces during mastication, and would thus endure smaller joint reaction forces in the TMJ (Helkimo et al. 1977; Ikebe et al. 2010; Ikebe et al. 2005; Turner 1998). Those with attrition and complete edentulism tend to show the lowest angulations of the articular eminence of the temporal ( i.e. , shallower mandibular fossae), individuals with complete dent ition have steeper articular eminences, and those with partial dentition are in between (Granados 1979; Hinton 1981; Lawther 1956) . Edentulous individuals also have lower cancellous bone density in the mandibular condyle, as well as lower stiffness and st rength in axial loading of the condyle (Giesen et al. 2003) . These differences cannot be attributed to osteoporotic bone loss, since the increase in anisotropy of cancellous bone normally associated with osteoporosis was not observed (Ciarelli et al. 2000 ; Giesen et al. 2003) . Experimental Studies In general, experimental studies investigating the link between masticatory function and craniofacial form compare the morphology of different groups raised on different dietary consistencies or with other impair ments of normal masticatory function.
65 Animal models most often used include rats, rhesus macaques, squirrel monkeys, pigs, rabbits, and mice. The logic behind these studies is that the consumption of a soft diet or impaired functioning would result in sm aller masticatory and joint reaction forces. Since bone adaptation tends to occur under high and dynamic stress environments (Turner 1998) , factors such as a harder diet should result in differential bone growth compared to a soft diet. A review of the lit erature suggests that altered dietary consistency or other manners of impaired functioning can result in gross morphological changes to the craniofacial region. In the masticatory complex, animals raised on harder diets tend to exhibit wider dental arches (Beecher and Corruccini 1981a; Beecher and Corruccini 1981b; Corruccini and Beecher 1982; Mavropoulos et al. 2004) , larger condylar and corpus dimensions (Bouvier 1988; Bouvier and Hylander 1981; Bouvier and Hylander 1984; Ciochon et al. 1997; Ravosa et al. 2007) , larger symphysis dimensions (Ravosa et al. 2007) , larger bicondylar and bigonial breadth (Ciochon et al. 1997) , larger dimensions of the ramus (Ciochon et al. 1997) , and larger muscle mass and physiological cross sectional area of t he masseter (Taylor et al. 2006) . Elsewhere in the cranium, animals raised on a hard diet exhibit a more vertical frontal profile, greater thickness of the outer table of the frontal bone, greater zygomatic height, a greater cranial and bizygomatic breadt h, more globular neurocranium with a greater curvature of lateral vault walls, greater occipital height, rostrocaudally shorter basisphenoids, and dorsoventrally deeper pterygoid plates (Ciochon et al. 1997; Menegaz et al. 2010) . Internal structure of bone and characteristics of condylar cartilage have also been shown to vary with masticatory function. Hard diet groups exhibit thicker condylar
6 6 cartilage, greater trabecular bone volume and trabecular thickness in the condyle, a greater degree of condylar and symphyseal bone mineralization, and thicker cortical bone at the symphysis and in the corpus compared to soft diet groups (Bouvier 1988; Bouvier and Hylander 1981; Bouvier and Hylander 1984; Chen et al. 2009; Mavropoulos et al. 2004; Tanaka et al. 2007; Yamada and Kimmel 1991) . Using intravenous doses of stains to highlight areas of bone formation in pigs raised on different dietary consistencies, Dias and colleagues (2011) found that the hard diet group showed a significantly higher degree of bone depos ition on the condyle compared to the soft diet group. Condyles in the hard diet group were also found to have significantly thicker and more mineralized trabecular structure (Dias et al. 2011) . Another strategy used by various researchers involves raising animals on one experimental condition, then changing to another to examine the amount of â€œrecoveryâ€ in different morphological features. Chen and colleagues (2009) , Bouvier (1988) , and Bouvier and Hylander (1984) all found that groups initially raised on a soft diet then switched to a hard diet exhibited morphology more similar to hard diet groups, and were often not statistically significantly different from them. Specifically, condylar cartilage thickness (Bouvier and Hylander 1984; Chen et al. 2009) , trabecular density (Bouvier 1988) , and condylar dimensions (Bouvier and Hylander 1984) were all similar between hard diet groups and groups that switched from a soft to a hard diet. A more recent experimental study examined the effects of hard fallback fo ods on mandibular morphology to investigate the role of dietary seasonality on craniofacial form (Scott et al. 2014) . A group of rabbits fed on consistently tougher diets (hay and hard pellets) were found to have larger cross section areas of the condyle compared to animals fed
67 on consistently hard diets (pellets only) ; a â€œseasonalâ€ group, which was fed hard pellets and hay for 6 weeks then switched to an all pellet diet for the next 18 weeks, exhibited intermediate morphology but was most similar to the l esschallenging all pellet diet control group (Scott et al. 2014) . These studies indicate a degree of plasticity not only during early development, but throughout life, although bone formation has been demonstrated to be impaired with increasing age (Liang et al. 1992) . With similar goals but a different strategy to the above studies, a group of researchers examined the mandibular and condylar morphology of control mice and myostatin deficient mice (Nicholson et al. 2006; Ravosa et al. 2008) . Since my ostatin is a negative regulator of muscle growth, myostatin deficient mice ( i.e. , knockout mice) serve as a model for overuse. Masseter and temporalis muscles in these mice tend to be over 50% larger than in normal mice. Unlike the above studies, both groups were fed the same diet. The knockout mice exhibited larger dimensions of the condyle, corpus, symphysis, and jaw adductors, increased articular cartilage height, thicker cortical bone along symphysis articular surface, and greater levels of biomineral ization in symphysis and temporomandibular joint, including the condylar neck (Nicholson et al. 2006; Ravosa et al. 2008) . Although this morphology is consistent with that of groups raised on a hard diet, it is not clear if this is due to the larger musculature and subsequent increase in bite force alone or secondary to the genetic alterations involved . Alternatively, low level passive muscular activity has been shown to increase the rate of bone formation (Rubin et al. 2001) . Larger muscles would produc e greater low level strains that would induce a greater bone growth response.
68 Summary The evidence for a link between craniofacial form and masticatory loading is consistent regardless of the design of the study (natural versus experimental) or the speci es of the samples involved. Based on these studies, one would expect a softer diet and a reduction in masticatory function to be associated with gracilization: i.e. , a reduction in craniofacial dimensions . In the temporomandibular joint specifically, l ow er masticatory loads are linked to smaller joint dimensions, shallower mandibular fossae, and thinner, less mineralized trabecular structure in the condyle.
69 CHAPTER 4 MATERIALS AND METHODS Approach The goal for this study is to investigate whether dietar y changes have contributed to the observed secular trends in craniofacial morphology of U.S. populations through observations of temporomandibular joint (TMJ) morphology. The TMJ was chosen as the focus of study due to the direct relationship of this structure with masticatory function, and was examined using both radiographic data and morphometric methods . The morphometric variables allowed examination of changes in the overall size and shape of the joint, whereas the radiographic data were used to exami ne more subtle changes in trabecular bone density over time. Th e premise of this study is that the variables used actually reflect variation in dietary consistency and ingestive behavior. Therefore, prior to examining change over time, preliminary tests were performed to investigate the suitability of variables used in this study to assess the effects of increased food processing on TMJ morphology. This was examined using nonhuman primate species that vary in diet and ingestive behavior (described below), as well as humans with varying degrees of dental loss . Variation in TMJ morphology between populations and collections was also examined. Sample The human sample for this study consist ed of males and females from two documented skeletal collections. The Terry Collection at the Smithsonian National Mu seum of Natural History provided individuals born in the mid19th and early 20th centuries. The majority of the skeletons in the collection are from unclaimed cadavers used in medical school anatomy classes at Washington University in St. Louis. While
70 the collection is mostly composed of individuals from less well off segments of society, the completeness and preservation of the skeletons and associated demographic information make this coll ection an invaluable resource for skeletal research (Hunt and Albanese 2005) . The contemporary portion of the sample was composed of individuals from the William M. Bass Donated Collection housed at the University of Tennessee, Knoxville. Historically, many of these donations are also unclaimed cadavers from local Medical Examinerâ€™s Offices, which would also represent individuals of low socioeconomic status (Christensen 2006) . However, this collection has had an influx of self donations over the past fifteen years, which are more likely to involve individuals of higher socioeconomic status . â€œBlackâ€ and â€œwhiteâ€ ancestry groups are included in this study. However, racial categories documented in these coll ections reflect self identified social categorization rather than biological reality . The sample composition by year of birth is illustrated in Table 41 and Figure 41. Sample composition by age is presented in Table 42. Data was collected on an addit ional sample of 41 white males that were edentulous or nearly edentulous (21 from the Terry Collection, 20 from the Bass Collection) to examine the utility of the variables in this study to reflect differences in function. Edentulous individuals have been shown to produce smaller bite forces compared to dentate individuals (Helkimo et al. 1977; Ikebe et al. 2010; Ikebe et al. 2005; Turner 1998) , exhibit lower bone density in the mandibular condyle (Ciarelli et al. 2000; Giesen et al. 2003) , and exhibit low er angulations of the articular eminence of the temporal (Granados 1979; Hinton 1981; Lawther 1956). In this sample, all individuals were completely edentulous aside from one individual with several anterior mandibular teeth and one anterior maxillary tooth,
71 one individual with one tooth in both the maxilla and mandible that were nonoccluding, and two individuals with one maxillary tooth each. The nonhu man primate species examined are Procolobus badius (Western red colobus ; n=6 ), Colobus polykomos (King colobus ; n=6 ), Cercocebus atys (sooty mangabey ; n=6 ), and Cercopithecus d iana (Diana monkey ; n=6 ) from Ta Forest, Cte dâ€™Ivoire. These sympatric species differ in dietary habits and foraging behavior , with the sooty mangabeys exploiting hard objects and insects, Diana monkeys consuming large amounts of soft fruits and insects, and the colobus monkeys eating more leaves and seeds (Daegling and McGraw 2001; McGraw et al. 2011; McGraw and Zuberbuhler 2007) . W ithin the colobus species, C. polykomos presumably exploits a greater quantity of tougher foods than P. badius (Daegling and McGraw 2001; McGraw and Zuberbuhler 2007) . Morphometric Methods Linear measurements of the mandibular condyle and temporal aspect of the joint, as well as the average slope of the articular eminence, were taken for comparison with previous studies of diet and TMJ morphology. The mediolateral width was taken for each condyle. The anteroposterior dimension of the condyle was recorded at three points: the medial, central, and lateral aspects. The mediolateral dimension of the temporal aspect of the joint was taken from the posterior terminus of the sphenotemporal suture to the inferolateral most point on the eminence (Figure 4 2 ) . T he anteroposterior measurements of the temporal aspect of the joint are described below. The anteroposterior dimension of the temporal aspect of the joint and average slope of the eminence were found using a contour gage with 35 pins per inch. With the
72 cr anium positioned upside down, the long arm of the contour gage was held parallel to Frankfort horizontal. The contour of the eminence and mandibular fossa was taken at medial, middle, and lateral aspects of the joint ( Figure 4 2 ). The anteriormost aspect of the contour was defined by a line parallel to the mediolateral measurement of the eminence, but positioned anteriorly such that it intersects with the posterior most aspect of the zygomatic arch. The posterior most aspect of the joint was defined by t he post glenoid process. Contours were also taken perpendicular to the mediolateral dimension of the eminence. Once the contour was establish, it was outlined with a felt tip pen on a data sheet, such that the long arm of the contour gage was horizontal ( i.e. , horizontal lines on the paper represent Frankfort Horizontal). With the long arm of the calipers positioned horizontally, the anteroposterior dimension of each contour was recorded (Figure 4 3). The average slope for each eminence was acquired using ImageJ and Microsoft Excel. Each data sheet was scanned, and each eminence contour was saved as a separate image. In ImageJ, a grid with an area per point of 100 pixels2 was superimposed on each contour using the Grid Plugin. This created vertical and horizontal lines spaced at every 10 pixels. Using the Multipoint Selection tool, points were selected at every intersection of the vertical gridlines with the contour ( i.e. , every 10 pixels along the contour), as well as at each of the endpoints of the c ontour (Figure 4 4 ). These selections were then saved as a set of XY coordinates, which were transferred into Excel. Slopes were derived along the length of the contour using these XY coordinates. The mean of these slopes were acquired for the eminence portion of the contour, which was defined to extend from the most superior point in the fossa to
73 the most inferior point on the eminence ( i.e. , the positive slopes on left side contours and negative slopes on right side contours). However, since the origi n is positioned in the top left corner of the image in ImageJ, a plot of the XY coordinates produces an inverted image of each contour. Therefore, the negative slopes were averaged for the left sides, and the positive slopes were averaged for the right si des. Other measurements of the skull were taken following MooreJansen and colleagues (1994) , and are listed in Table 43 . A set of mandibular and cranial measurements was recorded to investigate the covariance of TMJ dimensions with other craniofacial dimensions, and how this covariance may change over time. In order to control for size, variation in linear measurements was examined using an index involving the ratio of the linear measurement to the maximum diameter of the femoral head. Radiographic Methods Radiographs of the mandibular condyle allow ed the examination of the density of the trabecular bone within this structure, with mean grayscale values serving as a proxy for bone density. Radiographs were taken using DEXIS digital dental radiograph Platinum Sensor and imaging software and a NOMAD Pro handheld portable x ray source. The NOMAD was propped on a Styrofoam stand to prevent movement of the x ray source and consequent blurring of radiographs. The sensor was affixed to the NOMAD using a bitewing ring, bitewing aiming bar, a sensor holder, and removable adhesive strips. The sensor was positioned in the center of the circular x ray source at a distance of 66 mm from the surface of the x ray source for all radiographs. Radiographs of the condyles were taken from posterior to anterior, with the sensor positioned anterior to the condyle. The mandible was positioned such that the posterior
74 border of the ramus was oriented vertically and the anterior aspect of the condyle was touching the r adiograph sensor. Additionally, in order to ensure consistency between radiographs, an aluminum step wedge was attached to the sensor with tape such that it covered approximately a third of the surface of the radiograph. Aluminum step wedges are used to calibrate x ray equipment and assess the consistency between radiographs. They are made of a substance of consistent density ( e.g. , aluminum), and increase in thickness from one end to the other to produce a range of grayscale values. In this study, a 38 mm x 9 mm step wedge was used ( i.e. , roughly the size of the dental x ray sensor), which ranged in thickness from 1 mm at one end to 9 mm at the other end; as such, 9 steps are present that increase in thickness in 1 mm increments of . The 1 mm thick step in the step wedge did not fit on the sensor and does not appear in the radiographs. Trabecular bone density was evaluated using the mean grayscale values in three regions of interest throughout the condyle, as well as in one large region spanning the entire condyle. To acquire these values, all images were first calibrated in Adobe Photoshop using the percent pigment/ink percentages of each step of the aluminum step wedge present in each radiograph. The pigment/ink percentage is a way of characterizi ng the grayscale value in an image, and measures the intensity of the pigment. Therefore, when using this measure of grayscale, larger pigment/ink percentages indicate darker grayscale values ( e.g. , 0% = white, 100% = black). All radiograph images were c alibrated to the control pigment/ink percentages in Table 4 4 . These values were chosen by acquiring pigment/ink percentages for each step in the step wedge in a sample of radiographs, and choosing the most consistent values for
75 each step. In general, these percentages were recorded for approximately 25 radiographs. The most frequent pigment/ink percentage observed at each step was chosen as the calibration value to adjust the entire sample of radiographs. Images were adjusted in Adobe Photoshop using the Eyedropper T ool and a 31 x 31 pixel sample area. The pigment/ink percentage for each step in the step wedge was obtained by pressing the Shift key while clicking in an area in each step (Figure 4 5) . Four areas were selected at a time, and the pigment/ink percentages for each of the four areas appear in the Info window. These percentages were adjusted using the curves tool, which presents a histogram of all grayscale values in the image. Since the step wedge covers roughly a third of the radio graph, a large proportion of this histogram represents the steps in the step wedge. Peaks in this histogram were consistently observed around the pigment/ink percentages associated with each step in the step wedge (Figure 46). The straight diagonal line through the graph represents the original image. The line was adjusted to the control pigment/ink percentage by clicking on the line at each peak in the histogram and dragging the line to the desired pigment/ink percentage for each step in the step wedge (Figure 4 7). The input values at each peak were the original pigment/ink percentages of the image, while the output values were the corresponding control values. Once adjusted, the radiographs were rotated in Adobe Photoshop such that the most mediall y and laterally projecting points on the articular surface of the condyle both intersected a horizontal line. Using the length of the line connecting these two points, each condyle was divided into thirds and demarcated by vertical lines drawn in ImageJ ( Rasband 1997 2011) . The Overlay tool was used for these lines to ensure that
76 they would not alter the grayscale measurements taken on the radiographs. Mean grayscale values were taken in ImageJ by drawing selection areas on the radiograph and using the H istogram tool. These grayscale values range from 0 to 255, with 0 corresponding to black and 255 corresponding to white. The same rectangular selection area was used for all three aspects of the condyle: the largest possible area was taken in the smalles t division that avoided subchondral bone and neck cortical bone. One additional large oval selection was taken over the entire area of the condyle. As with the smaller selections, this large oval selection avoided subchondral bone and the cortical bone i n the condylar neck. All selections were taken superior to the narrowing of the condyle into the neck (Figure 48 ). These radiographic data were also collected on a lateral radiograph of the distal ulna as a measure of systemic bone density, since this is a nonweight bearing element that would be less likely to be altered by biomechanical influences. In these radiographs, the region of interest was the largest oval selection on the head of the ulna that avoided subchondral bone, the styloid process, and any arthritic changes in the joint. The entire selection lies distal to the narrowing of the bone proximally into the shaft of the ulna. The use of radiographs in this study resulted in images that captured the three dimensional internal structure of the condyle in a two dimensional image, as opposed to providing a cross section of the internal morphology as would be captured by a CT scan. Therefore, condyles that are thicker in the anteroposterior dimension will automatically appear to have denser bone. T o account for variations in thickness of the bone, all grayscale values were corrected using regression. Specifically, adjusted Y -
77 values for both the condyle and the ulna were calculated using the regression of grayscale value on the anteroposterior dim ension of the mandibular condyle, and the mediolateral dimension of the ulna. Tooth Loss and Dental Wear When choosing variables to include in a study, it is important to consider and control for the effects of confounding factors . Since the focus of this study is the connection between masticatory function and temporomandibular joint morphology, factors other than dietary consistency that may affect morphology need to be considered. In particular, both antemortem tooth loss and dental wear have been demonstrated to affect some of the variables that will be included in this study. Overall, tooth loss and attrition seem to have the largest effect on the angle of the articular eminence and the density and mechanical properties of the trabecular bone in the condyle. Hinton (1981) examined the slope of the articular eminence in hunter gatherer, agricultural, and urban populations in relation to degree of tooth loss and attrition. For higher wear scores, particularly in the molars, hunter gatherer populations exhibited a lower angulation of the eminence ( i.e. , a shallower mandibular fossa). The agricultural and urban populations exhibited a similar decrease in angulation with increased tooth wear and molar loss (Hinton 1981) . Granados (1979) found similar r esults in a study of 103 skulls from both agricultural and urban populations, but attributed at least some of this variation to temporomandibular joint disease. The connection between dental wear and temporomandibular joint pathology has been investigated, as well, but the results appear mixed. Some have found no association between degree of tooth wear and TMJ pathology (Eversole et al. 1985; Sheridan et al. 1991) , while others have found the
78 opposite (Hodges 1991) . Since individuals with evidence of TM J pathology were excluded from the current study as explained further below, this possible association will not affect the results of this investigation. However, dental wear will be considered when interpreting any trends involving the slope of the artic ular eminence. Using lateral radiographs of living individuals, Lawther (1956) also found that edentulous individuals have a decreased slope of the articular eminence and a decrease in height of the mandibular fossa, though these measurements still fell wi thin the range of the standard deviation of the dentate group. However, since all of the edentulous individuals in the study wore dentures, joint force during mastication may have been only slightly lower for this group compared to the dentate group (Lawt her 1956) . Mongini ( 1972) suggests that partially edentulous individuals with tooth loss in the molar region tend to exhibit a unique condylar shape due to remodeling on the superior surface of the condyle. In addition to differences in gross morphology of the temporomandibular j oint, edentulous individuals have also been found to have lower trabecular bone density in the mandibular condyle, as well as lower stiffness and strength i n axial loading of the condyle (Giesen et al. 2003) . S ince the increase in anisotropy of trabecular bone normally associated with osteoporosis was not observed, these condylar differences are likely due to differences in function (Ciarelli et al. 2000; Giesen et al. 2003) . Despite these effects, measurements on individuals with tooth loss and dental wea r were used in this study. Since e dentulous individuals have been shown to produce smaller forces during mastication, and would thus endure smaller joint reaction forces in the temporomandibular joint (Helkimo et al. 1977; Ikebe et al. 2010; Ikebe et
79 al. 2005) , these individuals can be used to verify that the variables used in this study can reflect variation in temporomandibular joint morphology due to differences in masticatory function. In particular, the radiographic methods and approach used to assess the slope of the articular eminence differ from methods used in previous studies. Individuals with dental wear were included in this study because changes in the degree tooth wear over time would inform changes in diet. Specifically, a decrease in toot h wear would reflect a decrease in grit in the diet and indicate a shift to the consumption of softer and more processed foods (Addy and Shellis 2006; Crothers 1992; Powell 1988) . Dental attrition was scored following Smith (1984) , which is described in Ta ble 45 . Using this scoring system, replicability is about 90% for molars, and 85% for more anterior teeth, with any errors of more than one wear stage occurring extremely rarely (Smith 1984) . A mean wear score was calculated separately for posterior (molars and premolars) and anterior (canines and incisors) teeth. Wear scores were summed, then divided by the number of teeth present. Tooth loss was scored using the Eichner Index (Ikebe et al. 2010) . This index categorizes degree of tooth l oss in terms of occlusal support between opposing teeth ( e.g. , upper and lower molar contact). Premolars, molars, and anterior teeth are each counted as separate support zones. Group A includes 4 posterior support zones, group B includes one to three pos terior support zones or contact in the anterior area only, and group C has no support zone, though a few teeth may be present (Figure 4 9 ). Each of these groups is further divided into subgroups. Individuals in group A1 have no missing teeth, those in gr oup A2 have at least one tooth missing in either the maxilla or the
80 mandible, and those in group A3 are missing at least one tooth in both the mandible and maxilla. Individuals classified in groups B1, B2, and B3 have three, two, and one posterior support zones, respectively, while group B4 has occlusal support in anterior teeth only. Although those in group C have no occlusal support zones, individuals in group C1 have at least one tooth both the mandible and maxilla, individuals in group C2 have at leas t one tooth in either the mandible or the maxilla, and individuals in group C3 are completely edentulous (Ikebe et al. 2010) . The categories of the Eichner I ndex have been shown to be significantly associated with maximum occlusal force and overall masti catory performance (Ikebe et al. 2010) . Maximum occlusal force was measured using pressure sensitive sheets, while masticatory performance was measured by having subjects chew on a gummy jelly for 30 chewing cycles and calculating the remaining surface ar ea of the jelly (Ikebe et al. 2012) . Both maximum occlusal force and masticatory performance decreased significantly with an increase in loss of occlusal contacts. Reduced oral functions were observed even if removable prostheses w ere used to restore the teeth . Maximum occlusal force was not significantly different between groups A2 to B2, although group A1 produced a significantly larger maximum occlusal force than these groups; masticatory performance did not significantly differ between groups A1 to B 1 (Ikebe et al. 2010) . In the interest of including as many individuals as possible without tooth loss being a confounding factor, only individuals classified into groups A1 to B2 were included in the analysis of change over time. Temporomandibular Joint Pathology Aside from tooth loss and wear, pathology at the temporomandibular joint also alters the variables used in this study. The progression of the skeletal manifestations of
81 temporomandibular joint disease occurs as follows. Once cartilage destruction reaches the bone surface, an inflammatory response initiates skeletal tissue regeneration (Carter and Beaupr 2001). In areas of low stress ( i.e. , the lateral edges of the condyle and around the margins of the posterior slope of the articular eminence) bone will form directly, resulting in osteophyte development. Areas of high compressive loading ( i.e. , the articular surface of the condyle, and the posterior slope of the articular eminence) promote cartilage formation. Once formed, exposure o f this new cartilage to the degenerated joint surface results in further destruction, erosion, and removal of the cartilage by inflammatory cells (Carter and Beaupr 2001). The lesions resulting from this process are usually more severe and appear earlier on the temporal surface of the joint (Mann and Hunt 2005). Early osteoarthritis generally involves porosity in the middle of the fossa, articular eminence anterior to the fossa, and the articular surface of the mandibular condyle. With increasing severi ty, larger areas of bone in these regions are eroded, resulting in an enlarged mandibular fossa and flattened condylar heads w ith osteophytic development. Most severe cases can involve the complete destruction of the condylar articular surface (Aufderheid e and Rodrguez Martn 1998; Mann and Hunt 2005; Rando and Waldron 2012). The morphological change associated with temporomandibular joint disease precludes the use of individuals exhibiting pathology in this study. On the condyle, any porosity or osteophytic development would disrupt both linear measurements and radiographic data. Porosity and osteophytic development on the articular eminence and in the mandibular fossa might alter the contours used to assess the slope of the eminence. Therefore, where pathological changes were present, variables that were
82 likely to be affected by this pathology ( e.g. , eminence slope, condylar density) were not recorded. Bilateral Symmetry Radiographic and morphometric variables in the skull were collected for both lef t and right sides, when applicable and possible. Mastication can occur bilaterally, in which chewing occurs on both right and left sides, or unilaterally, in which one side is preferred over the other. Unilateral chewing may occur for various reasons, including dental loss, the presence of dental work or dental prostheses, asymmetry of occlusal contact area, or unilateral pain in facial muscles or in the TMJ (Farias Gomes et al. 2010) . If asymmetry in mastication was present in the sample used in this st udy, comparisons of measurements taken only on one side may be misleading. Therefore, the degree and direction of bilateral asymmetry is investigated here. Ulnar variables and maximum femoral head diameter were recorded for the left side only. However, the right side was substituted in the event that postmortem damage or arthritic changes could affect the variables examined. Variables were not recorded in the event that postmortem damage or pathological changes interfered with data collection. Diffe rences between left and right sides for temporomandibular joint variables were examined using pairedsamples t tests (Table 46). Significant differences were found between sides for the following variables: lateral and medial condylar grayscale means, anteroposterior dimension of the condyle at the lateral and medial aspects, eminence width, and anteroposterior dimension of the temporal at the medial aspect of the joint. No significant differences between sides were found for eminence slopes. Where sign ificant differences were found, one side was not consistently larger or denser than the other. To examine if side differences may be related to dental status, these
83 pairedsamples t tests were performed again including only those individuals classified in Group A using the Eichner Index ( i.e. , 4 posterior support zones are present), since these individuals would not favor one side over the other due to tooth loss. Significant differences were found between sides for the same variables listed above. Theref ore, there is no connection with dental status and no clear indication of side favoritism during chewing. In order to condense the data for ease of interpretation, the mean of right and left side s was calculated for all bilateral variables for use in subs equent statistical analyses. Collection Effect If secular trends in TMJ morphology are found, other factors that may account for these trends need to be considered. Specifically, the individuals in the collections used in this study come from different geographical locations. The Terry Collection ori ginates from the St. Louis area, while individuals in the Bass Collection are mostly from Tennessee. These areas may differ in terms of the history of the settlement of immigrant populations. Further, the Bass Collection is partially composed of self don ations, which would likely involve individuals of higher socioeconomic status compared to those in the Terry Collection. Therefore, change over time might actually represent morphological differences between these two skeleton collections due to other fac tors like geographic area or socioeconomic status . To ensure that this is not the case, collection effects were examined using independent samples t tests and individuals with dates of birth ranging from 1910 to 1940, which encompasses the temporal overlap between the two collections .
84 Statistics Statistics were calculated using IBM SPSS Version 22. Preliminary analyses that assess the efficacy of the variables in this study in reflecting masticatory function involved the comparison of nonhuman primate g roups with variations in ingestive behaviors, and the comparison of individuals with and without teeth. For the nonhuman primates, morphometric variables were compared between groups using Analysis of Variance (ANOVA), and controlling for size by using the ratio of condylar dimensions to mandibular length. Radiographic variables for these species were analyzed using ANOVA, and were collected using the methods described above. The human samples with and without teeth were compared using independent samples t tests for morphometric variables, and Analysis of Covariance (ANCOVA) for radiographic variables to consider the effects of age, since bone density has been shown to decrease with age (Riggs et al. 2004). Population differences were characterized using ANOVA; and variation between collections were investigated using independent samples t tests. Analysis of secular trends in the TMJ was structured by considering historical data and socioeconomic factors. As the skeletal collections to be used in this study generally consist of individuals of lower socioeconomic status, effects of food processing and other dietary changes need to be considered with this group in mind. Specifically, although new procedures for milling and companies that mass produced canned goods were introduced in the late 1800s, affordable products stemming from these developments, such as the uniformly soft Wonder B read and canned strained and pureed baby foods, did not become available until the 1920s (Dyson 2000; Hooker 1981). A population shift to an urban majority also occurred by 1920; this shift to city life
85 is associated with a more sedentary lifestyle, as well as greater access to grocery store chains with mass produced goods (Beale 2000). As the 1920s were followed by the Great Depression in the 1930s and food shortages during World War II, the year 1920 appears to be a reasonable turning point from the heavy meat centered meals of the 1800s and early 1900s to lighter diet consisting of more processed foods. As such, in order to assess secular trends in relation to diet, morphology of individuals born before 1920 were compared with those born after 1920 usi ng independent samples t tests for morphometric variables, and ANCOVA for radiographic variables to consi der the effects of age . As tooth wear was not normally distributed, MannWhitney U tests were used to assess change over time in this variable. Addi tionally, trends over time were analyzed for each variable using regression analysis using year of birth as the predictor variable. Due to morphological differences between populations, change over time was analyzed separately for each population. Null H ypotheses The premise that TMJ morphology reflects masticatory function was first examined using monkey species as well as humans with varying degrees of tooth loss. The null hypothesis for all comparisons between monkey species is that there are no diffe rences in TMJ morphology between species once sizeeffects are considered . Since these species exploit foods of various toughness and hardness , this finding would suggest either no link between diet and TMJ morphology or the presence of a confounding vari able, which would likely be an effect of phylogenetic history . The null hypothesis for the comparison between humans of varying dental status is that there are no differences in TMJ morphology between people with different degrees of tooth loss. This wou ld also suggest a lack of a connection between masticatory function and
86 TMJ morphology since these variations in tooth loss vary predictably with bite force and joint reaction force. The null hypothesis for the analysis of secular trends in humans is that there is no change in TMJ morphology over time. This would suggest one of three things: 1) dietary changes over time have not contributed to secular trends observed in the skull; 2) if there are also no differences between monkey species, this may provide further support that TMJ morphology does not actually vary with differences in diet; 3) developments in food processing and other changes in quantity and frequency of food consumption have not been extreme enough to alter skeletal morphology. If dietary changes have not contributed to secular trends in the skull, other factors such as changes in population genetics or nutrition and health may be driving these trends. Alternative Hypotheses Significant differences in TMJ morphology between monkey species may be observed. Presumably, the four species utilized in this study can be ranked by the amount of effort needed to consume their respective diets; sooty mangabeys would require the most effort, Diana monkeys the least, and colobus monkeys in between, w ith C. polykomos exploiting a tougher diet than P. badius . If species with harder and tougher diets possess larger condylar dimensions and denser trabecular bone, these findings reinforce a link between TMJ morphology and dietary consistency. In this cas e, the observed differences could be used as a basis for interpreting secular trends in humans. However, if significant differences are observed that are not consistent with previous studies of TMJ morphology, these results may reflect other species diffe rences. For example, differential intake and retention of calcium would affect bone
87 density, or other more minor aspects of dietary behavior may influence TMJ morphology . Significant differences may also be observed when comparing individuals of varying dental status. A significant degree of tooth loss in humans is associated with a reduction of dental function, and thus smaller bite forces and joint reaction forces . Therefore, compared to dentate individuals, one would expect the TMJs of edentulous in dividuals to be smaller with less dense trabecular bone in the condyle, and the articular eminence should be shallower . Significant differences observed between edentulous and dentate individuals that are not consistent with this pattern would again indic ate the presence of a confounding variable. In the analysis of secular trends, significant differences may be found between the pre1920 group and post 1920 group. If a shift to a lighter, more processed diet is contributing to these trends, one would expect the post 1920 group to have a smaller TMJ , less dense trabecular bone in the condyle, and a shallower articular eminence. However, since trabecular bone density may respond to changes in dietary consistency throughout life, individuals born before 1920 and lived for a significant amount of time after this year may also exhibit less dense trabecular bone. For all variables, regression analysis should demonstrate morphology consistent with a shift to a lighter, more processed diet. If TMJ morphology changes over time in a manner inconsistent with such a shift, other factors unrelated to masticatory function, such as changes in health, nutrition or genetics, are likely driving secular trends. Investigating the covariance of other mandibular dimensions wi th dimensions of the TMJ over time can also provide some insight into the likelihood that changes in food
88 processing have contributed to secular trends. If this is the case, one would expect the covariance between TMJ dimensions and other mandibular dimensions to decrease over time. Access to a greater variety of processed foods would result in a relaxation of functional demands, which in turn would increase the variation in mandibular morphology, which itself might indicate decreasing covariance among mandibular variables. Additionally, changes in health, nutrition, and standard of living over time may affect TMJ morphology. These factors have been suggested to affect the increase in stature observed in the U.S. over the same time period investigated her e (Meadows Jantz and Jantz 1999) . Since better health and nutrition is associated with an increase in size, one would expect TMJ dimensions to increase and mirror the change observed in stature, rather than decrease, which is expected for a more processed diet. As alterations in nutrition should have a systemic effect on the skeleton, trends in bone density and mineralization were investiga ted by taking radiographs of a nonload bearing element unrelated to mastication ( i.e. , ulnar head), to be compared with trends in TMJ density. This will help clarify if changes over time in bone density are localized to the condyle.
89 Table 41. Sample composition for pre1920 and post 1920 groups. Year of b irth W hite males W hite females B lack ma les B lack females 1850 1920 42 30 30 27 1921 2000 57 37 21 5 Total 99 67 51 32 Figure 41. Sample composition by year of birth. Table 42. Sample composition by age. Age White m ales White f emales Black m ales Black f emales 16 20 2 0 0 2 21 30 7 2 8 6 31 40 23 11 11 12 41 50 33 14 18 5 51 60 24 17 10 4 61 70 10 15 2 2 71 80 0 5 2 1 81 90 0 3 0 0 Total 99 67 51 32
90 Figure 42. Mediolateral measurement (red) and position of slopes and anteroposterior dimensions (green) of the temporal aspect of the joint.
91 Figure 43. Example of a tracing of a slope, with the anteroposterior dimension of the joint indicated by the red line. Anterior is to the right. Figure 44. Example illustrating the procedure for recording coordinates in I mageJ. Anterior is to the right.
92 Table 43. Morphometric measurements recorded. Element Sample(s) Variable Unit Condyle Humans and m onkeys Lateral A P d imension mm Central A P d imension mm Medial A P d imension mm Mediolateral d imension mm Temporal Humans Lateral A P d imension mm Central A P d imension mm Medial A P d imension mm Mediolateral eminence d imension mm Lateral average s lope mm/mm Central average s lope mm/mm Medial average s lope mm/mm Femur Humans Femoral head d iameter mm Mandible Humans Corpus thickness at mental f oramen mm Corpus height at mental f oramen mm Corpus t hickness at M 1 /M 2 j unction mm Corpus h eight at M 1 /M 2 j unction mm Symphyseal h eight mm Symphyseal t hickness mm Bicondylar b readth mm Bigonial w idth mm Minimum ramus b readth mm Mandibular l ength mm Ramus h eight mm Cranium Humans Cranial l ength mm Cranial b readth mm Cranial h eight mm Palate l ength mm Palate b readth mm Ulna Humans Me diolateral dimension of ulnar h ead mm Mandible Monkeys Mandibular l ength mm Table 44. Percent pigment/ink percentages used to calibrate radiographs. Step t hickness Pigment/ink p ercentage 2 mm 82% 3 mm 67% 4 mm 56% 5 mm 47% 6 mm 39% 7 mm 33% 8 mm 28% 9 mm 23%
93 Figure 45. Example illustrating the selection of pigment/ink percentages in Adobe Photoshop. Circles with corresponding colors indicate the step wedge selection and its corresponding pigment/ink percentage in the Info window. Based on the required percentages listed in Table 44, the pigment/ink percentages need to be reduced one percentage point aside from the second selection (blue circle). The red rectangle indicates the selection area for the Eyedropper tool.
94 Figure 46. Example illustrating the histogram of grayscale values using the curves tool in Adobe Photoshop. The red arrows indicate the eight peaks in the histogram that represent the eight steps in the step wedge.
95 Figure 47. Example illustrating the adjustment of radiographs using the curves tool in Adobe Photoshop. To adjust the pigment/ink percentage, click on the horizontal line in the histogram box at each peak corresponding to the selected ste p in the step wedge (red circle), and drag the line so that the output percentage matches the appropriate value in Table 44. Red rectangles indicate the original input value (24%) and adjusted output value (23%).
96 Figure 48. Selection areas used to acquire mean grayscale values, outlined in yellow.
97 Table 45. Descriptions of wear stages from Smith (1984). Molars Premolars Incisors and c anines 0. Missing or cannot be coded 0. Missing or cannot be coded 0. Missing or cannot be coded 1. Unworn to polished or small facets (no dentin exposure) 1. Unworn to polished or small facets (no dentin exposure) 1. Unworn to polished or small facets (no dentin exposure) 2. Moderate cusp removal (blunting). No more than one or two pinpoint exposures 2. Moderate cusp removal (blunting) 2. Point or hairline of dentin exposure 3. Full cusp removal and/or some dentin exposure, pinpoint to moderate 3. Full cusp removal and/or moderate dentin patches 3. Dentin line of distinct T hickness 4. Several large dentin expos ures, still discrete 4. At least one large dentin exposure on one cusp 4. Moderate dentin exposure no longer resembling a line 5. Two dentinal areas coalesced 5. Two large dentin areas (maybe slight coalescence) 5. Large dentin area with enamel rim complete 6. Three dentinal areas coalesced, or four coalesced with enamel island 6. Dentinal areas coalesced, enamel rim still complete 6. Large dentin area with enamel rim lost on one side or very thin enamel only 7. Dentin exposed on entire surface, enamel rim largely intact 7. Full dentin exposure, loss of rim on at least one side 7. Enamel rim lost on two sides or small remnants of enamel remain 8. Severe loss of crown height, breakdown of enamel rim; crown surface takes on shape of roots 8. Severe loss of crown height; crown surface takes on shape of roots 8. Complete loss of crown, no enamel remaining: crown surface takes on shape of roots
98 Figure 49 . Diagram of the Eichner classification for tooth loss. Horizontal line indicates antemortem tooth loss. Gray indicates no c ontact with antagonist tooth. Group C3 represents completely edentulous individuals and is not shown. Figure adapted from Ikebe et al. (2010).
99 Table 46. Paired t tests of temporomandi bular joint variables comparing left and right sides. Mean Std. d eviation Std. e rror m ean 95% Confidence i nterval of the d ifference t df Sig. (2 tailed) Lower Upper Grayscale v ariables L lat. mean â€“ R l at . m ean 3.5321 8.1892 .5762 4.6682 2.3959 6.130 201 < .00 1 L cent. m ean â€“ R c ent . m ean .0360 7.3720 .5200 .9894 1.0614 .069 200 .945 L med. m ean â€“ R m ed . m ean 5.3114 8.3947 .5907 4.1467 6.4760 8.992 201 < .00 1 L overall mean â€“ R o verall m ean .4824 5.4429 .3839 .2746 1.2395 1.257 200 .210 Linear m easurements L cond. width R cond. w idth .0154 1.4554 .0928 .1673 .1982 .166 245 .868 L c on d. AP med. R. cond. AP m ed. .4402 .8251 .0531 .5449 .3356 8.283 240 < .00 1 L cond. AP cent. R cond. AP c ent. .0573 .8940 .0578 .0566 .1712 .991 238 .323 L cond. AP lat. R cond. AP l at. .1822 .8595 .0554 .0731 .2912 3.290 240 .001 L em. w idth R em. w idth .2030 1.1429 .0747 .3502 .0558 2.717 233 .007 L temp. l at . AP R temp. l at . AP .0972 1.2066 .1161 .3274 .1329 .837 107 .404 L temp. c ent . AP R temp . c ent . AP .1459 1.1609 .1051 .3540 .0622 1.388 121 .168 L temp. m ed . AP R temp. m ed . AP .3474 1.2928 .1200 .1097 .5852 2.894 115 .005 Eminence s lopes L lateral â€“ R l ateral .0238 .1361 .0140 .0515 .0039 1.704 94 .092 L m edial â€“ R m edial .0042 .1910 .0189 .0417 .0333 .223 101 .824 L c entral â€“ R c entral .0025 .17 80 .0172 .0316 .0366 .147 106 .884
100 CHAPTER 5 RESULTS Chapter 5 presents the results of all analyses performed for this study. Several preliminary tests are presented first, which involve a comparison of nonhuman primates, a comparison of humans with variations in dental status, an analysis of change over time in too th wear, an analysis of TMJ variation between populations, and a test for morphological bias due to the use of multiple skeletal collections. Following these preliminary results, change over time in morphometric variables and radiographic variables are pr esented. This chapter ends with an analysis of change over time in the correlation between TMJ dimensions and dimensions of the cranium and mandible. Interpretations of these results are presented in Chapter 6 . Preliminary Tests Nonhuman Primates Suitability of the variables used in this study to assess the effects of increased food processing on TMJ morphology was first examined by comparing nonhuman primate species that vary in diet and ingestive behavior. The null hypothesis for this test is as follo ws: H0: No differences in condylar dimensions or bone density will be observed between species. Condylar dimensions of the nonhuman primate species were examined using the ratio of each dimension to mandibular length to control for body size. Species dif ferences were examined with an ANOVA (Table 51). Although there were no differences between species in any anteroposterior condylar dimension, the mediolateral width of the condyle differed significantly between species (p < .001). Figure 51 displays a plot of the mediolateral index (ratio of mediolateral width to
101 mandibular length) for all individuals for each species. When comparing the species individually using a Tukeyâ€™s Honestly Significant Difference test, Cercopithecus Diana was significantly smaller than all other groups in the mediolateral width of the condyle (p < .001 for both colobus species, p = .010 for C. atys ) , and Cercocebus atys was significantly smaller than Procolobus badius (p = .017) ; the two colobus s pecies were not significantly different from each other (Table 52). No significant differences in grayscale values were found between species (Table 53, Table 54). Procolobus badius and Cercocebus atys exhibited the highest grayscale values. Cercopith ecus diana exhibited the lowest values for all measures of grayscale except within the lateral aspect of the condyle. Figure 52 displays a plot of overall grayscale values for each species; individual plots of medial, central, and lateral grayscale values are not shown due to similarity in patterning to overall grayscale values. Therefore, for comparisons of nonhuman monkey species, the null hypothesis that there are no differences between species was not rejected for mean grayscale values and the anteroposterior dimensions of the condyle, but was rejected for the mediolateral condylar dimension. Tooth Loss Individuals with varying degrees of dental loss were also compared to examine the utility of the variables in this study to reflect differences in function. Specifically, edentulous individuals produce lower bite forces and subsequently endure smaller joint reaction forces in the TMJ compared to individuals with teeth. Only white males were included in this analysis since the demographic compositions of the skeletal collections used involve a greater proportion of this group, and a subsample of edentulous
102 individuals could be found more efficiently. The null hypothesis for this test is as follows: H0: No differences in temporomandibular joint dimensio ns or bone density will be observed between edentulous and dentate groups. Linear dimensions of the condyle and temporal aspect of the TMJ were examined using the ratio of each dimension to maximum femoral head diameter to control for body size. Independent samples t t ests were used to compare groups with and without teeth for all linear dimensions, as well as the slope of the articular eminence (Table 55). No significant differences were found for any dimension of the condyle or temporal aspect of the joint. Similarly, no significant differences were found in eminence slope between groups. Figures 53 â€“ 5 7 present the frequencies of condylar mediolateral width index, condylar central anteroposterior index, eminence mediolateral width index, temporal central anteroposterior index, and central eminence slope as examples of the similarity in morphology between edentulous and dentate groups. When comparing mean grayscale values in the mandibular condyle between individuals with and without teeth, and considering age as a covariate, edentulous individuals exhibited significantly lower mean grayscale values in all regions of the condyle (p < .001; Tables 56 and 57). Figure 58 presents the frequencies of overall mean grayscale values for the condyle by dental status; plots of the lateral, central, and medial grayscale means are similar in distribution and are not shown. Unlike the condyle, no significant differences were found for the mean grayscale value taken from the ulna (Figure 5 9). Age was not a significant factor in any of these analyses . Therefore, the null hypothesis that there are no differences between edentulous and dentate groups was not rejected for linear dimensions of the condyle, linear dimensions
103 of the temporal aspect of the joint, e minence slope, or mean grayscale value in the ulna. The null hypothesis was rejected, however, for all grayscale values in the mandibular condyle. Tooth Wear In order to help characterize changes in diet over time in the populations examined, tooth wear w as scored for all teeth present for each individual. Specifically, assuming a similarity in age, a larger degree of tooth wear would indicate a greater amount of grit in the diet . Therefore, a change in the degree of tooth wear over time would reflect changes in the food consumed. The null hypothesis for this test is as follows: H0: No change over time in degree of tooth wear will be observed for all groups. Due to the nonnormal distribution of tooth wear scores for all groups, MannWhitney U tests were used to examine differences between preand post 1920 groups. Table 58 displays the results for mean posterior tooth wear. For all populations, the post 1920 group exhibits a significantly lower degree of posterior tooth wear compared to the pre1 920 group (p = .008 for black females, p < .001 for all other groups). Trends in anterior tooth wear are generally similar and involve a decrease in severity of tooth wear over time (Table 59), although preand post 1920 black females are not significan tly different. Post1920 black males (p = .001), white females (p = .022), and white males (p < .001) all exhibit a significantly lower degree of tooth wear. Figures 510 and 511 illustrate the differences between preand post 1920 groups using boxplot s for mean posterior tooth wear and anterior tooth wear, respectively. Change over time was also examined using regression and year of birth for posterior tooth wear (Table 510) and anterior tooth wear (511). For both posterior and
104 anterior tooth wear, regressions were significant at the p < .001 level for all groups except black females. Regressions were nevertheless statistically significant for black females for both posterior (p = .006) and anterior tooth wear (p = .042. R squared values ranged fro m .227 to .575 for posterior tooth wear and from .130 to .478 for anterior tooth wear. Figures 512 â€“ 5 15 present scatter plots of wear scores by year of birth for black females, black males, white females, and white males, respectively. Due to the small post 1920 sample size for black females (n=5), tooth wear for this group was reanalyzed using bootstrapping. These results are depicted for posterior tooth wear and anterior tooth wear in Figures 516 and 517, respectively. The histogram in each of these figures represents 10,000 resamples of 5 individuals from the pre1920 sample , and the dotted lines represent the 95% confidence interval. The post 1920 group falls within the 95% confidence interval of the resamples of the pre1920 group for both post erior and anterior tooth wear. This is consistent with the results of the Mann Whitney U test for anterior tooth wear, but conflicts with the significant change over time found with both the MannWhitney U test and regression analysis for posterior tooth wear. Therefore, the observed significant decrease in posterior tooth wear for black females should be interpreted with caution. Based on the MannWhitney U tests, regression analysis, and reanalysis of the black female group, the null hypothesis is rej ected for all groups for both posterior and anterior mean tooth wear, except for anterior tooth wear for black females. Variation Among Populations Differences among black females, black males, white females, and white males were examined using an ANOVA. For variables in which multiple measurements were taken throughout the joint ( e.g. , medial, central, and lateral anteroposterior dimension of
105 the condyle), only the central measurement was used in this analysis to provide a general idea of morphological di fferences between groups. Similarly, only the overall mean condylar grayscale value was used to assess differences in condylar bone density between groups. The null hypothesis for this analysis is as follows: H0: No variation in internal and external m orphology of the temporomandibular joint will be observed between groups. Table 512 presents the ANOVA results for all variables examined. Significant differences between groups were found at the p < .001 level for all variables except central eminence slope (p = .005). Figures 5 18 â€“ 5 24 display boxplots of each variable by group, as well as the significant results from a Tukeyâ€™s Honestly Significant Difference test which examined significant differences between individual groups. Since all linear me asurements were sizecorrected, these results reflect differences that are relative to body size rather than absolute size differences. For dimensions of the condyle, black females were significantly larger in condylar width index for white females (p = . 035) and white males (p < .001); black females were also significantly larger in the anteroposterior dimension than all other groups at the p < .001 level. For mediolateral and anteroposterior t emporal dimensions, both female groups were significantly lar ger than both males groups. These comparisons were significant at p < .001 except for the comparison of white females and black males for eminence width index (p = .004). White males exhibited a significantly larger central eminence slope than black females (p = .022) and white females (p = .037). Black males and females both exhibited significantly higher grayscale values in the condyle compared to white males and females; these comparisons were all significant at p < .001. Group differences for grays cale values in the ulna exhibited a
106 different pattern, however. Black males exhibited significantly higher grayscale values compared to all other groups (p < .001). White males exhibited significantly higher values than black females (p = .015) and white females (p < .001). Within females, black females exhibited significantly higher values than white females (p < .001). Thus, the null hypothesis is rejected for all variables analyzed for variation between populations. Variation Between Collections The samples for this study were collected from two separate documented skeletal collections from different areas in the country. In order to assess whether morphological differences exist between these two samples, collection effects were examined using indep endent samples t tests and individuals with dates of birth ranging from 1910 to 1940. However, due to limited temporal overlap between collections, sample sizes were very small for all groups except white females (Table 5 13). Sample sizes were further r educed in cases where certain variables could not be collected for these individuals due to pathology or postmortem damage. Therefore, only results for white females will be presented here. The null hypothesis for this analysis is as follows: H0: No differences in internal and external morphology of the TMJ will be observed between collections for white females. Table 5 14 lists the means for the Terry and Bass Collection by measurement for white females, as well as the pvalues for all t tests exami ning the possible effect of collection. White females did not exhibit a significant collection effect for any linear measurement, eminence slope, or measure of bone density in the condyle or ulna. Therefore, the null hypothesis is not rejected.
107 Change O ver Time Change over time in TMJ morphology was examined separately by group ( i.e. , black females, black males, white females, and white males). The null hypothesis for the analysis of change over time is as follows: H0: No change over time in internal and external morphology of the temporomandibular joint will be observed for all groups. Morphometric V ariables L inear dimensions of the condyle and temporal aspect of the TMJ were examined using the ratio of each dimension to maximum femoral head diameter to control for body size. Independent samples t t ests were used to compare pre and post 1920 groups for all linear dimensions, as well as the slope of the articular eminence. Tables 515 â€“ 5 18 present the results of these t t ests for black females, bl ack males, white females, and white males, respectively. Figures 525 â€“ 5 32 present boxplots of all linear measurements by time period and population. For all populations, there was no significant difference between preand post 1920 groups for any condylar dimension. Similarly, for all groups, there was no significant difference between preand post 1920 groups for the mediolateral width of the eminence or the anteroposterior dimension of the temporal at the lateral aspect of the joint. However, the post 1920 group exhibited a significantly larger anteroposterior dimension of the temporal at the central aspect of the joint for black females (p = .019), black males (p=.006), and white males (p < .001) compared to the pre1920 group. Although trends ar e similar for all populations at the medial aspect of the temporal, only the post 1920 white male group exhibited a significantly larger anteroposterior dimension compared to the pre1920 group (p = .001).
108 Similar to the analysis of tooth wear, the signifi cant results for black females were reanalyzed using bootstrapping due to small sample size. These results are depicted for the anteroposterior central index of the temporal in Figure 533. The histogram represents 10,000 resamples of 5 individuals from the pre1920 sample , and the dotted lines represent the 95% confidence interval. Although the mean of the post 1920 group lies toward the upper end of the distribution of the resamples of the pre1920 group, it still falls within the 95% confidence interv al. Therefore, the significant increase in the central anteroposterior dimension of the temporal for black females should be interpreted with caution. Figures 534â€“ 5 36 present boxplots of all eminence slopes by time period and population. Post 1920 whi te males exhibit significantly shallower slopes of the eminence at the lateral aspect of the joint compared to the pre1920 sample (p < .001). White females exhibit the same trend in morphology for this measure of eminence slope, though it is not statisti cally significant (p = .109). No other measures of eminence slope are significantly different between preand post 1920 groups for all populations. However, the tendency for a shallower eminence over time is present in the rest of the joint for white males, particularly at the central aspect (p = .089). Change over time was also examined for linear dimensions and eminence slopes using regression by year of birth. Tables 5 19 â€“ 5 29 present the results of these regressions for all populations separately by each morphometric variable. Scatter plots of significant regressions are shown in Figures 537 â€“ 5 43 . All regressions for condylar dimensions were not statistically significant and exhibited very low r squared values ranging from .000 to .038.
109 The regression for mediolateral eminence width index was significant for white males (p = .037), which indicates an increase in this dimension over time; however, the r squared value was very low (r2 = .048). No other groups exhibited statistically signi ficant changes in the mediolateral width of the condyle. White females exhibit a significant decrease in the anteroposterior dimension of the temporal at the lateral aspect of the joint (p = .046), but the r squared value was also low ( r2 = .090). No other groups exhibited statistically significant changes over time in the lateral anteroposterior dimension of the temporal. A statistically significant increase in the central anteroposterior dimension of the temporal was observed for black males (p = .007; r2 = .170) and white males (p < .001; r2 = .231). White females exhibited a similar tendency with a lower r squared value (p = .050; r2 = .075). Black females did not exhibit a statistically significant trend for the central anteroposterior dimension of the temporal. A statistically significant increase was also observed in the medial anteroposterior dimension of the temporal for white females (p = .014; r2 =.115) and white males (p < .001; r2 = .148). Significant trends were not observed for black fem ales or males in the medial anteroposterior dimension of the temporal. The regression for lateral eminence slope was significant for white males (p < .001; r2 = .223), and indicated a decrease in this slope over time. White females exhibited a similar tendency with a lower r squared value (p = .059; r2 = .079). Black females and males did not exhibit a significant trend in the slope of the eminence at the lat eral aspect of the joint. A decrease in eminence slope was also observed at the central aspect of the joint for white males, though this trend was not statistically significant (p = .054; r2 = .044). Significant trends were not observed for the central
110 e minence slope for the other three populations. Regressions for the slope of the eminence at the medial aspect of the joint were not significant for any group, and exhibited extremely low r squared values ranging from .004 .008. Thus, for all populations , the null hypothesis was not rejected for all condylar dimensions. Based on regression analysis and year of birth, the null hypothesis was rejected for the mediolateral dimension of the eminence for white males only and the lateral anteroposterior dimens ion of the temporal for white females only; however, neither of these results was supported by the analysis of change over time using t tests. The null hypothesis was rejected for the central anteroposterior dimension of the temporal for black and white m ales only based on t tests and regression analysis. For the medial anteroposterior dimension of the temporal, the null hypothesis was rejected for white males based on t tests and regression analysis, and for white females based on regression analysis onl y. The null hypothesis was rejected for the lateral eminence slope for white males only based on t tests and regression analysis, but was not rejected for the central and medial eminence slope for any group. Table 530 lists all morphometric variables for which significant change over time was observed, as well as the direction of change for each variable. Correlation with Skull Dimensions Change over time in the covariation between TMJ dimensions and other cranial and mandibular dimensions was examined b ecause a decrease in covariation over time could indicate a relaxation in functional demands. The null hypothesis for this test is as follows: H0: No change over time in the strength of correlation between TMJ dimensions and other skull dimensions will be observed for all populations.
111 These tests were performed by comparing the r values for pairs of variables for both preand post 1920 groups to determine if the strength of the correlation between these variables changed over time. All variables were divided by femoral head diameter to create an index that controls for body size. Additionally, only the mediolateral width and central anteroposterior dimensions of the condyle and temporal were used. Therefore, four TMJ indices were compared with each s kull index. As this resulted in a large number of comparisons, the significance level for these tests were set at p = .0125 (.05/4) to reduce Type I error. Whenever a significant correlation was found, the 95% confidence interval of the r values was used to determine if there was any change over time in the strength of the correlation. For example, if the r value of the pre1920 group is significant but the r value of the post 1920 group falls within the 95% confidence interval for the pre1920 group, then there is no change over time in the strength of that correlation. Tables 531 â€“ 5 45 present the significant r values for the correlations between TMJ dimensions and each skull dimension for preand post 1920 populations, and indicates whether these si gnificant correlations became stronger (â€œincreaseâ€), weaker (â€œdecreaseâ€), or did not change over time (â€œnoneâ€). In the event that a pair of variables was not significantly correlated for both preand post 1920 groups, the r values were omitted; â€œn.s.â€ in dicates that the correlation was not significant. In cases where only one time period resulted in a significant r value, r values, pvalues, and the sample sizes for both time periods are presented for comparison. No significant correlations were found f or the corpus thickness index at the mental foramen; thus the table presenting the results for these comparisons is omitted.
112 No change over time in the strength of correlation was found between TMJ indices and cranial breadth index, cranial height index, s ymphysis thickness index, symphysis height index, and corpus thickness index at M1/M2. Out of all comparisons, black females only exhibited an increase in covariation between the anteroposterior dimension of the condyle and mandibular length. Black males mostly exhibited a decrease in covariance between TMJ dimensions and mandibular dimensions. These decreases included the covariation of condylar width and both bigonial width and bicondylar breadth, the covariation of the anteroposterior dimension of the condyle and mandibular length, and the covariation of eminence width and corpus height at the mental foramen. However, the anteroposterior dimension of the temporal was more strongly correlated with both cranial length and palate length over time for black males. White females exhibited a decrease in covariation between minimum ramus breadth and condylar width, as well as between palate length and eminence width. White males exhibited a decrease in covariation between condylar width and both palate brea dth and ramus height. However, there was an increase in covariation between skull dimensions and the anteroposterior dimensions of the condyle and temporal. The dimensions that covaried with the condyle included palate breadth, corpus height at the mental foramen and M1/ M2, and mandibular length; the dimensions that covaried with the temporal included corpus height at the mental foramen and M1M2, bicondylar breadth, and ramus height. Therefore, the null hypothesis was rejected for correlations between TMJ dimensions and cranial length, palate length and breadth, corpus height at the mental foramen and M1M2, bigonial width, bicondylar breadth, minimum ramus width, ramus
113 height, and mandibular length. The null hypothesis was not rejected for correlations between TMJ dimensions and cranial breadth, cranial height, symphysis thickness and height, and corpus thickness index at the mental foramen and M1/M2. Changes over time in covariation between TMJ dimensions and skull dimensions are summarized in Table 546. Radiographic Variables Comparisons of preand post 1920 groups for m ean grayscale values in the mandibular condyle and ulna were performed using ANCOVA using age as a covariate. These results are displayed in Tables 547 â€“ 5 50 for black females, black males, white females, and white males, respectively. In cases where a significant effect of the covariate is observed in addition to significant change over time, the ANCOVAs were rerun to test the assumption of homogeneity of regression slopes; these results are presented in Table 551 . Essentially , this examines the interaction between age and time period to ensure that the relationship between age and grayscale values are similar for both preand post 1920 groups (Field 2013) . Figures 544 â€“ 5 48 pr esent boxplots of all grayscale measurements by time period and population. For black females and white females, no significant differences were found between preand post 1920 groups for all measures of grayscale in the condyle and ulna. However, post 1 920 black males exhibited significantly lower mean grayscale values compared to the pre1920 group in the lateral condyle (p = .031; r2 = .113), central condyle (p = .004; r2 = .182), medial condyle (p = .001; r2 = .259), and overall condyle (p = .005; r2 = .180). There is no significant difference between preand post 1920 black male groups for grayscale in the ulna, however (p = .677; r2 = .024). Although age was significant in the model for medial grayscale mean for black males,
114 the interaction betwee n age and time period is not significant (p = .548). Post 1920 white males also exhibited significantly lower mean grayscale values compared to the pre 1920 group in the central condyle (p = .001; r2 = .107), medial condyle (p = .002; r2 = .121), and over all condyle (p = .003; r2 = .096). There is no significant difference between preand post 1920 white male groups for grayscale in the lateral condyle (p = .312; r2 = .013) or ulna (p = .162; r2 = .034). Age was a significant factor for white males, but the interaction between age and time period was not significant for central grayscale mean (p = .999), medial grayscale mean (p = .522), or overall condylar grayscale mean (p = . 803). Change over time for m ean grayscale values in the mandibular condyle and ulna were also examined using regression and year of birth; the results of these regressions are presented in Tables 552 â€“ 5 56. Scatter plots of significant regressions are shown in Figures 549 â€“ 5 55. Black males exhibited a significant decrease ov er time in mean grayscale values for the lateral condyle (p = .012; r2 = .125), central condyle (p = .001; r2 = .195), medial condyle (p = .003; r2 = .172), and overall condyle (p = .003; r2 = .168). However, there was no significant relationship between year of birth and mean grayscale value for the ulna for black males (p = .896; r2 = .000). For white males, there was a significant decrease over time in mean grayscale values for the centr al condyle (p = .047; r2 = .041), as well as in the ulna (p = .042; r2 = .044), although the r squared values for these regressions are low. No other significant relationships between year of birth and grayscale values were found for white males. There w as no significant change over time in mean grayscale values in the condyle for white females; however, there was a significant increase in the mean grayscale value in the ulna (p = .009; r2 =
115 .109). Similar to the comparisons of preand post 1920 groups, there was no significant relationship between any measure of grayscale in the condyle or ulna and year of birth for black females. Thus, the null hypothesis was rejected for all measures of condylar grayscale values for black males based on ANCOVA and regression analysis. For white males, the null hypothesis was rejected for central condylar grayscale values based on ANCOVA comparing pre and post 1920 groups and regression analysis by year of birth; the null hypothesis was also rejected for medial and overall condylar grayscale values based on ANCOVA only. For mean grayscale of the ulna, the null hypothesis was rejected for white males and white females based on regression analysis, but the direction of change differed between the two groups. Th e null hypothesis was not rejected for any grayscale variables in the condyle or ulna for black females. Table 557 lists all grayscale variables for which significant change over time was observed, as well as the direction of change for each variable.
116 Ta ble 51. ANOVA comparing linear condylar measurements between monkey species. Sum of s quares df Mean s quare F Sig. ML i ndex Between g roups .012 3 .004 17.455 <.001 Within g roups .004 18 .000 Total .016 21 Medial AP i ndex Between g roups .000 3 .000 .572 .641 Within g roups .003 18 .000 Total .003 21 Central AP i ndex Between g roups .001 3 .000 1.535 .240 Within g roups .002 18 .000 Total .003 21 Lateral AP i ndex Between g roups .000 3 .000 1.110 .371 Within g roups .001 18 .000 Total .002 21 Figure 51. Scatter plot of mediolateral condylar index of monkey species. The index for the monkey species is the ratio of the condylar dimension to mandibular length.
117 Table 52. Tukeyâ€™s Honestly Significant Difference test comparing mediolateral condylar index between monkey species. (I) Species (J) Species Mean d ifference (I J) Std. e rror Sig. 95% Confidence i nterval Lower b ound Upper b ound Colobus polykomos Procolobus badius .00606 .00858 .893 .0303 .0182 Cercocebus atys .02281 .00858 .069 .0014 .0471 Cercopithecus diana .05740 .00960 <.001 .0303 .0845 Procolobus badius Colobus polykomos .00606 .00858 .893 .0182 .0303 Cercocebus atys .02888 .00858 .017 .0046 .0531 Cercopithecus diana .06347 .00960 <.001 .0363 .0906 Cercocebus atys Colobus polykomos .02281 .00858 .069 .0471 .0014 Procolobus badius .02888 .00858 .017 .0531 .0046 Cercopithecus diana .03459 .00960 .010 .0075 .0617 Cercopithecus diana Colobus polykomos .05740 .00960 <.001 .0845 .0303 Procolobus badius .06347 .00960 <.001 .0906 .0363 Cercocebus atys .03459 .00960 .010 .0617 .0075
118 Table 53. Descriptive statistics for condyle grayscale values for monkey species. Grayscale values range from 0255, with 0 corresponding to black and 255 corresponding to white. N Mean Std. dev Std. error Minimum Maximum Medial Colobus polykomos 6 71.02407 15.717203 6.416521 51.463 93.376 Procolobus badius 6 82.04426 12.919720 5.274454 69.648 101.795 Cercocebus atys 6 73.19015 18.640767 7.610061 48.717 94.388 Cercopithecus diana 4 64.91312 8.396197 4.198099 58.307 77.223 Total 22 73.50925 15.101162 3.219579 48.717 101.795 Central Colobus polykomos 6 69.21319 11.309561 4.617109 52.884 85.075 Procolobus badius 6 78.64223 15.995429 6.530106 66.023 107.782 Cercocebus atys 6 81.18817 15.992331 6.528842 57.426 100.538 Cercopithecus diana 4 66.78021 9.592643 4.796321 56.430 79.244 Total 22 74.60829 14.227393 3.033290 52.884 107.782 Lateral Colobus polykomos 6 67.61467 12.758012 5.208437 47.558 80.983 Procolobus badius 6 75.83259 16.036443 6.546851 57.859 95.868 Cercocebus atys 6 69.64218 16.971545 6.928604 40.955 86.928 Cercopithecus diana 4 68.38365 15.561230 7.780615 53.644 90.052 Total 22 70.54869 14.653364 3.124108 40.955 95.868 Overall Colobus polykomos 6 68.81725 8.227319 3.358789 61.736 84.969 Procolobus badius 6 76.57438 10.072684 4.112156 65.964 91.046 Cercocebus atys 6 71.94515 13.519250 5.519211 52.709 87.066 Cercopithecus diana 4 64.84297 8.200569 4.100285 60.502 77.135 Total 22 71.06330 10.540792 2.247304 52.709 91.046
119 Table 54. ANOVA comparing grayscale values of monkey species. Sum of squares df Mean square F Sig. Medial Between g roups 770.319 3 256.773 1.150 .356 Within g roups 4018.628 18 223.257 Total 4788.947 21 Central Between g roups 777.164 3 259.055 1.342 .292 Within g roups 3473.629 18 192.979 Total 4250.793 21 Lateral Between g roups 242.848 3 80.949 .342 .796 Within g roups 4266.294 18 237.016 Total 4509.143 21 Overall Between g roups 371.937 3 123.979 1.138 .360 Within g roups 1961.337 18 108.963 Total 2333.274 21 Figure 52. Plot of overall condylar grayscale values for monkey species. This value is taken from one large oval selection encircling the entire area of the condyle.
120 Table 55. Independent samples t tests comparing joint dimensions and eminence slopes between groups with varying degrees of dental loss. The index for the human sample is the ratio of the TMJ dimension to femoral head diameter. T df Sig. (2 tailed) Mean d ifference Std. error d ifference 95% Confidence interval of the d ifference Lower Upper Condyle ML width i ndex .159 135 .874 .00130 .00820 .01490 .01751 AP lat. i ndex 1.081 135 .282 .00412 .00381 .00342 .01166 AP cent. i ndex .136 135 .892 .00067 .00491 .00903 .01037 AP med. i ndex 1.214 135 .227 .00607 .00500 .01595 .00381 Temporal ML width i ndex .482 129 .631 .00401 .00832 .02048 .01246 AP lat. i ndex .685 112 .495 .00459 .00670 .01786 .00868 AP c ent . i ndex .759 118 .449 .00484 .00637 .00778 .01745 AP med. i ndex .271 115 .787 .00183 .00676 .01523 .01157 Eminence s lope Lateral .775 112 .440 .024315 .031392 .086515 .037884 Central 1.203 118 .231 .053924 .044830 .142699 .034851 Medial .526 115 .600 .024901 .047340 .118672 .068871
121 Figure 53. Plot of condylar mediolateral width index by dental status.
122 Figure 54. Plot of condylar central anteroposterior index by dental status.
123 Figure 55. Plot of eminence mediolateral width index by dental status.
124 Figure 56. Plot of temporal central anteroposterior index by dental status.
125 Figure 57. Plot of central eminence slope by dental status.
126 Table 56. Descriptive statistics for grayscale values by dental status. N Mean Std. deviation Std. error Minimum Maximum Lateral mean Dent 97 95.6171 9.21053 .93519 63.42 119.16 Edent 41 86.2767 11.29018 1.76323 58.81 106.36 Total 138 92.8421 10.72450 .91293 58.81 119.16 Central mean Dent 97 104.6449 11.01060 1.11796 71.78 129.07 Edent 41 92.8347 11.34651 1.77203 66.35 121.22 Total 138 101.1360 12.32403 1.04909 66.35 129.07 Medial mean Dent 97 92.4759 10.81262 1.09786 65.97 119.63 Edent 41 83.2431 11.20514 1.74995 62.78 111.93 Total 138 89.7328 11.68394 .99460 62.78 119.63 Overall mean Dent 97 99.0340 9.29684 .94395 69.33 118.34 Edent 41 89.1436 10.92186 1.70571 65.89 112.12 Total 138 96.0956 10.76895 .91671 65.89 118.34 Ulna mean Dent 95 128.8531 10.73851 1.10175 101.21 152.26 Edent 41 127.9804 10.96790 1.71290 93.62 149.73 Total 136 128.5900 10.77489 .92394 93.62 152.26
127 Table 57. ANCOVA results for grayscale values and dental status, with age as a covariate. Source Type III sum of s quares df Mean s quare F Sig. R s quared Lateral m ean Corrected m odel* 2519.561 2 1259.780 12.848 < .001 .160 Age 5.297 1 5.297 .054 .817 Dental s tatus 2125.239 1 2125.239 21.674 <.001 Error 13237.474 135 98.055 Total 1205269.221 138 Corrected t otal 15757.035 137 Central m ean Corrected m odel* 4146.819 2 2073.410 16.800 <.001 .199 Age 127.154 1 127.154 1.030 .312 Dental s tatus 3062.070 1 3062.070 24.811 <.001 Error 16660.987 135 123.415 Total 1432340.341 138 Corrected t otal 20807.806 137 Medial m ean Corrected m odel* 2855.800 2 1427.900 12.164 <.001 .153 Age 399.161 1 399.161 3.401 .067 Dental s tatus 1545.908 1 1545.908 13.170 <.001 Error 15846.669 135 117.383 Total 1129874.581 138 Corrected t otal 18702.469 137 Overall m ean Corrected m odel* 2992.979 2 1496.489 15.667 <.001 .188 Age 173.934 1 173.934 1.821 .179 Dental s tatus 2026.663 1 2026.663 21.218 <.001 Error 12894.956 135 95.518 Total 1290228.966 138 Corrected t otal 15887.935 137 Ulna m ean Corrected m odel* 79.512 2 39.756 .339 .713 .005 Age 57.702 1 57.702 .492 .484 Dental s tatus 49.091 1 49.091 .419 .519 Error 15593.745 133 117.246 Total 2264486.038 136 Corrected t otal 15673.257 135 *Corrected m odel is the fit of the model overall ( ag e, dental status, and intercept ) (Field 2013) .
128 Figure 58. Plot of overall condylar grayscale mean by dental status.
129 Figure 59. Plot of ulna grayscale mean by dental status.
130 Table 58. Mann Whitney U Tests comparing mean posterior tooth wear between preand post 1920 groups. N Posterior wear m ean MannWhitney U Standard e rror Sig. Black f emales Pre 1920 27 1.68 18.500 19.247 .008 Post 1920 5 1.14 Black m ales Pre 1920 30 2.05 46.000 51.974 <.001 Post 1920 21 1.17 White f emales Pre 1920 30 1.82 271.500 76.980 <.001 Post 1920 36 1.28 White m ales Pre 1920 41 3.20 124.500 138.487 <.001 Post 1920 57 1.26 Table 59. Mann Whitney U Tests comparing mean anterior tooth wear between preand post 1920 groups. N Anterior wear m ean Mann Whitney U Standard e rror Sig. Black f emales Pre 1920 27 2.30 44.000 19.254 .241 Post 1920 5 1.92 Black m ales Pre 1920 28 2.49 128.500 49.328 .001 Post 1920 21 1.87 White f emales Pre 1920 29 2.59 349.500 75.577 .022 Post 1920 36 2.02 White m ales Pre 1920 41 3.70 254.500 134.833 <.001 Post 1920 55 2.25
131 Figure 510. Boxplots of mean posterior tooth wear by time period. The box represents the middle 50% of scores, or the interquartile range (IQR), and the line within the box represents the median. The distance between the top of the box and upper whisker represents t he top 25% of scores, and the distance between the bottom of the box and the lower whisker represents the bottom 25% of scores. The circles represent individuals that are 1.5 times the IQR above the upper quartile (top of the box), or below the lower quar tile (bottom of the box). The stars represent individuals that are 3 times the IQR above the upper quartile, or below the lower quartile. In such cases, the top and bottom 25% are calculated excluding these extreme values (Field 2013) .
132 Figure 5 11. Boxplots of mean anterior tooth wear by time period.
133 Table 510. Regression results for mean posterior wear and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression 1.762 1 1.762 8.801 .006 .476 .227 Residual 6.007 30 .200 Total 7.769 31 Black m ales Regression 13.500 1 13.500 40.468 <.001 .673 .452 Residual 16.346 49 .334 Total 29.846 50 White f emales Regression 7.690 1 7.690 19.159 <.001 .480 .230 Residual 25.689 64 .401 Total 33.380 65 White m ales Regression 93.550 1 93.550 130.073 <.001 .759 .575 Residual 69.044 96 .719 Total 162.594 97 Table 5 11. Regression results for mean anterior wear and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression 1.908 1 1.908 4.489 .042 .361 .130 Residual 12.748 30 .425 Total 14.656 31 Black m ales Regression 6.922 1 6.922 19.525 <.001 .542 .293 Residual 16.662 47 .355 Total 23.584 48 White f emales Regression 9.665 1 9.665 16.200 <.001 .452 .205 Residual 37.584 63 .597 Total 47.249 64 W hite m ales Regression 58.146 1 58.146 86.127 <.001 .691 .478 Residual 63.461 94 .675 Total 121.607 95
134 Figure 512. Plot of wear scores for black females by year of birth.
135 Figure 513. Plot of wear scores for black males by year of birth.
136 Figure 514. Plot of wear scores for white females by year of birth.
137 Figure 515. Plot of wear scores for white males by year of birth.
138 Figure 516. Bootstrap results for mean posterior tooth wear for black females. Red line represents mean of the post 1920 group.
139 Figure 517. Bootstrap results for mean anterior tooth wear for black females. Red line represents mean of the post 1920 group.
140 Table 512. ANOVA results comparing TMJ morphology between populations. Sum of s quares df Me an s quare F Sig. Condylar width i ndex Between g roups .035 3 .012 6.210 <.001 Within g roups .443 234 .002 Total .478 237 Condylar AP central i ndex Between g roups .032 3 .011 16.500 <.001 Within g roups .153 234 .001 Total .185 237 Eminen ce width i ndex Between g roups .114 3 .038 23.184 <.001 Within g roups .370 226 .002 Total .485 229 Temporal AP central i ndex Between g roups .070 3 .023 19.709 <.001 Within g roups .243 205 .001 Total .313 208 Central eminence s lope Between g roups .414 3 .138 4.376 .005 Within g roups 6.469 205 .032 Total 6.883 208 Overall grayscale m ean Between g roups 14772.216 3 4924.072 39.580 <.001 Within g roups 29360.605 236 124.409 Total 44132.821 239 Ulna grayscale m ean Between g roups 37346.447 3 12448.816 89.766 <.001 Within g roups 32451.460 234 138.681 Total 69797.907 237
141 Figure 518. Boxplots of condylar width index by group.
142 Figure 519. Boxplots of condylar anteroposterior central index by group.
143 Figure 520. Boxplots of eminence width index by group.
144 Figure 521. Boxplots of temporal anteroposterior central index by group.
145 Figure 522. Boxplots of central eminence slope by group.
146 Figure 523. Boxplots of overall condylar grayscale mean by group.
147 Figure 524. Boxplots of ulna grayscale mean by group. Table 513. Sample sizes for examination of collection effect. Black f emales Black m ales White f emales White m ales Total Terry Collection 6 3 7 5 21 Bass Collection 1 5 13 1 20 Total 7 8 20 6 41
148 Table 514. Independent samples t tests comparing TMJ variables between collections for white females. t df Sig. (2 tailed) Mean d ifference Std. error d ifference 95% Confidence interval of the d ifference Lower Upper Condyle ML width i ndex 1.175 18 .255 .02325 .01979 .01833 .06482 AP lat. i ndex .527 17 .605 .00503 .00954 .02517 .01511 AP cent. i ndex 1.994 18 .061 .02206 .01106 .04529 .00118 AP med. i ndex 1.867 18 .078 .01632 .00874 .03469 .00205 Temporal ML width i ndex 1.233 19 .233 .02206 .01789 .01539 .05950 AP lat. i ndex .573 15 .575 .01194 .02083 .03246 .05635 AP c ent . i ndex .833 16 .417 .01254 .01505 .04445 .01937 AP med. i ndex .258 17 .800 .00358 .01386 .03282 .02567 Eminence s lope Lateral .519 14 .612 .027667 .053269 .086584 .141917 Central 1.246 15 .232 .057071 .045787 .040522 .154665 Medial 1.181 16 .255 .070584 .059788 .056160 .197329 Grayscale Lateral m ean 1.668 18 .113 7.47399 4.48028 1.93872 16.88670 Central m ean 1.554 19 .137 9.44732 6.08057 3.27946 22.17409 Medial m ean 1.621 19 .122 11.31916 6.98469 3.29997 25.93829 Overall m ean 1.699 19 .106 8.99378 5.29437 2.08746 20.07502 Ulna m ean 1.445 19 .165 6.90896 4.78019 3.09610 16.91403 Table 515. Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 black female groups. t df Sig. (2 tailed) Mean d ifference Std. error d ifference 95% Confidence interval of the d ifference Lower Upper Condyle ML width i ndex .512 30 .612 .01383 .02701 .04132 .06899 AP lat. i ndex .726 30 .473 .01148 .01580 .04375 .02080 AP cent. i ndex .192 30 .849 .00247 .01285 .02377 .02870 AP med. i ndex .075 30 .941 .00072 .00959 .02031 .01888 Temporal ML width i ndex .860 29 .397 .01481 .01722 .05004 .02041 AP lat. i ndex .417 26 .680 .01021 .02447 .06051 .04010 AP c ent . i ndex 2.500 28 .019 .05316 .02126 .09670 .00961 AP med. i ndex .712 27 .482 .01452 .02039 .05637 .02732 Eminence s lope Lateral .931 25 .361 .048273 .051863 .155086 .058541 Central .579 27 .567 .035458 .061247 .090209 .161126 Medial .735 26 .469 .060833 .082713 .230851 .109185
149 Table 516. Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 black male groups. t df Sig. (2 tailed) Mean d ifference Std. error d ifference 95% Confidence interval of the d ifference Lower Upper Condyle ML width i ndex 1.285 48 .205 .01500 .01167 .03847 .00847 AP lat. i ndex .019 48 .985 .00011 .00590 .01176 .01198 AP cent. i ndex 1.657 48 .104 .01264 .00763 .00270 .02799 AP med. i ndex .837 48 .407 .00576 .00688 .00807 .01959 Temporal ML width i ndex 1.485 46 .144 .01849 .01245 .04355 .00657 AP lat. i ndex 1.001 35 .324 .01162 .01161 .01195 .03518 AP c ent . i ndex 2.878 40 .006 .03030 .01053 .05158 .00902 AP med. i ndex .315 36 .755 .00376 .01196 .02801 .02049 Eminence s lope Lateral .064 35 .949 .002981 .046383 .091181 .097144 Central .280 40 .781 .013730 .049057 .112878 .085418 Medial .297 36 .768 .017585 .059283 .137816 .102646 Table 517. Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 white female groups. t df Sig. (2 tailed) Mean d ifference Std. error d ifference 95% Confidence interval of the d ifference Lower Upper Condyle ML width i ndex 1.080 58 .285 .01212 .01122 .01035 .03458 AP lat. i ndex 1.311 57 .195 .00736 .00561 .00388 .01860 AP cent. i ndex .460 58 .647 .00309 .00671 .01034 .01651 AP med. i ndex .752 58 .455 .00434 .00577 .00721 .01589 Temporal ML width i ndex .844 57 .402 .00827 .00979 .02787 .01134 AP lat. i ndex 1.616 43 .113 .01871 .01158 .00464 .04205 AP c ent . i ndex 1.339 50 .187 .01050 .00784 .02625 .00525 AP med. i ndex 1.467 50 .149 .01097 .00748 .02600 .00405 Eminence s lope Lateral 1.636 44 .109 .058138 .035541 .013490 .129767 Central .221 51 .826 .009196 .041586 .092684 .074293 Medial .230 51 .819 .009421 .040909 .091549 .072707
150 Table 518. Independent samples t tests comparing joint dimensions and eminence slopes between preand post 1920 white male groups. t df Sig. (2 tailed) Mean d ifference Std. error d ifference 9 5% Confidence i nterval of the d ifference Lower Upper Condyle ML width i ndex .679 94 .499 .00571 .00841 .02241 .01099 AP lat. i ndex .389 94 .698 .00163 .00419 .00996 .00670 AP cent. i ndex 1.046 94 .298 .00528 .00505 .01531 .00474 AP med. i ndex 1.811 94 .073 .00871 .00481 .01826 .00084 Temporal ML width i ndex 1.973 90 .052 .01759 .00892 .03530 .00012 AP lat. i ndex .095 79 .925 .00068 .00713 .01351 .01486 AP c ent . i ndex 4.882 83 <.001 .02981 .00611 .04195 .01766 AP med. i ndex 3.449 83 .001 .02402 .00696 .03787 .01017 Eminence s lope Lateral 4.098 79 <.001 .125705 .030678 .064641 .186768 Central 1.722 83 .089 .080743 .046890 .012519 .174005 Medial .602 83 .549 .031357 .052118 .072303 .135017
151 Figure 525. Boxplots of mediolateral condylar width index by time period.
152 Figure 5 26. Boxplots of condylar lateral anteroposterior index by time period.
153 Figure 527. Boxplots of condylar central anteroposterior index by time period.
154 Figure 528. Boxplots of condylar medial anteroposterior index by time period.
155 Figure 529. Boxplots of eminence mediolateral width index by time period.
156 Figure 530. Boxplots of temporal lateral anteroposterior index by time period.
157 Figure 531. Boxplots of temporal central anteroposterior index by time period.
158 Figure 532. Boxplots of temporal medial anteroposterior index by time period.
159 Figure 533. Bootstrap results for the temporal anteroposterior central index for black females. Red line represents mean of the post 1920 group.
160 Figure 534. Boxplots of lateral eminence slope by time period.
161 Figure 535. Boxplots of central eminence slope by time period.
162 Figure 536. Boxplots of medial eminence slope by time period.
163 Table 519. Regression results for mediolateral condylar width index and year of bi rth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .003 1 .003 .872 .358 .168 .028 Residual .090 30 .003 Total .093 31 Black m ales Regression .001 1 .001 .582 .449 .109 .012 Residual .081 48 .002 Total .082 49 White f emales Regression .001 1 .001 .303 .584 .072 .005 Residual .109 58 .002 Total .110 59 White m ales Regression .001 1 .001 .805 .372 .092 .008 Residual .157 94 .002 Total .158 95 Table 520. Regression results for condylar lateral anteroposterior index and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .000 1 .000 .103 .751 .058 .003 Residual .032 30 .001 Total .032 31 Black m ales Regression .000 1 .000 .014 .906 .017 .000 Residual .020 48 .000 Total .020 49 White f emales Regression .001 1 .001 1.908 .173 .180 .032 Residual .026 57 .000 Total .027 58 White m ales Regression .000 1 .000 .143 .706 .039 .002 Residual .039 94 .000 Total .039 95
164 Table 521. Regression results for condylar central anteroposterior index and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .001 1 .001 1.056 .312 .184 .034 Residual .020 30 .001 Total .021 31 Black m ales Regression .001 1 .001 1.906 .174 .195 .038 Residual .035 48 .001 Total .036 49 White f emales Regression .000 1 .000 .126 .724 .047 .002 Residual .038 58 .001 Total .039 59 White m ales Regression .000 1 .000 .552 .459 .076 .006 Residual .057 94 .001 Total .057 95 Table 522. Regression results for condylar medial anteroposterior index and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .000 1 .000 .334 .568 .105 .011 Residual .012 30 .000 Total .012 31 Black m ales Regression .000 1 .000 .600 .443 .111 .012 Residual .028 48 .001 Total .028 49 White f emales Regression .000 1 .000 .040 .841 .026 .001 Residual .038 58 .001 Total .039 59 White m ales Regression .001 1 .001 2.548 .114 .162 .026 Residual .052 94 .001 Total .053 95
165 Table 523. Regression results for eminence mediolateral width index and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .000 1 .000 .065 .800 .047 .002 Residual .037 29 .001 Total .037 30 Black m ales Regression .004 1 .004 2.360 .131 .221 .049 Residual .084 46 .002 Total .088 47 White f emales Regression .003 1 .003 2.558 .115 .207 .043 Residual .075 57 .001 Total .079 58 White m ales Regression .008 1 .008 4.490 .037 .218 .048 Residual .159 90 .002 Total .166 91 Table 524. Regression results for temporal lateral anteroposterior index and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .000 1 .000 .036 .851 .037 .001 Residual .064 26 .002 Total .064 27 Black m ales Regression .000 1 .000 .396 .533 .106 .011 Residual .042 35 .001 Total .042 36 White f emales Regression .006 1 .006 4.235 .046 .299 .090 Residual .061 43 .001 Total .067 44 White m ales Regression .000 1 .000 .111 .740 .037 .001 Residual .078 79 .001 Total .078 80
166 Table 525. Regression results for temporal central anteroposterior index and year of birth. Sum of s quares d f Mean s quare F Sig. R R s quared Black f emales Regression .005 1 .005 2.394 .133 .281 .079 Residual .059 28 .002 Total .064 29 Black m ales Regression .009 1 .009 8.179 .007 .412 .170 Residual .046 40 .001 Total .056 41 White f emales Regression .003 1 .003 4.034 .050 .273 .075 Residual .038 50 .001 Total .041 51 White m ales Regression .019 1 .019 24.922 <.001 .481 .231 Residual .063 83 .001 Total .082 84 Table 526. Regression results for temporal medial anteroposterior index and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .000 1 .000 .076 .785 .053 .003 Residual .039 27 .001 Total .039 28 Black m ales Regression .000 1 .000 .003 .956 .009 .000 Residual .048 36 .001 Total .048 37 White f emales Regression .004 1 .004 6.505 .014 .339 .115 Residual .033 50 .001 Total .037 51 White m ales Regression .014 1 .014 14.424 <.001 .385 .148 Residual .081 83 .001 Total .095 84
167 Table 527. Regression results for lateral eminence slope and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .000 1 .000 .001 .972 .007 .000 Residual .283 25 .011 Total .283 26 Black m ales Regression .009 1 .009 .461 .502 .114 .013 Residual .647 35 .018 Total .655 36 White f emales Regression .052 1 .052 3.754 .059 .280 .079 Residual .606 44 .014 Total .658 45 White m ales Regression .389 1 .389 22.630 <.001 .472 .223 Residual 1.357 79 .017 Total 1.745 80 Table 528. Regression results for central eminence slope and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .004 1 .004 .276 .603 .101 .010 Residual .420 27 .016 Total .424 28 Black m ales Regression .002 1 .002 .077 .782 .044 .002 Residual 1.002 40 .025 Total 1.004 41 White f emales Regression .003 1 .003 .133 .716 .051 .003 Residual 1.146 51 .022 Total 1.149 52 White m ales Regression .171 1 .171 3.804 .054 .209 .044 Residual 3.721 83 .045 Total 3.891 84
168 Table 529. Regression results for medial eminence slope and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression .003 1 .003 .129 .722 .070 .005 Residual .619 26 .024 Total .623 27 Black m ales Regression .004 1 .004 .127 .724 .059 .004 Residual 1.171 36 .033 Total 1.175 37 White f emales Regression .006 1 .006 .295 .589 .076 .006 Residual 1.093 51 .021 Total 1.099 52 White m ales Regression .037 1 .037 .670 .415 .089 .008 Residual 4.625 83 .056 Total 4.662 84
169 Figure 537. Plot of eminence mediolateral width index for white males by year of birth.
170 Figure 538. Plot of temporal anteroposterior lateral index for white females by year of birth.
171 Figure 539. Plot of temporal anteroposterior central index for black males by year of birth.
172 Figure 540. Plot of temporal anteroposterior central index for white males by year of birth.
173 Figur e 541. Plot of temporal anteroposterior medial index for white females by year of birth.
174 Figure 542. Plot of temporal anteroposterior medial index for white males by year of birth.
175 Figure 543. Plot of lateral eminence slope for white males by year of birth. Table 530. Summary of variables with statistically significant change over time for linear measurements and eminence slopes. Group t t est Regression Direction of c hange Eminence ML width i ndex WM Not s ignificant Significant Increase Temporal AP lateral i ndex WF Not s ignificant Significant Decrease Temporal AP central i ndex BM Significant Significant Increase WM Significant Significant Increase Temporal AP medial i ndex WF Not s ignificant Significant Increase WM Significant Significant Increase Lateral eminence s lope WM Significant Significant Decrease
176 Table 531. Significant r values for TMJ dimensions and cranial length index. BF BM WF WM Pre Post r Ch ange Pre Post r C hange Pre Post r C hange Pre Post r C hange Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r .304 .457 p n.s. n.s. None n.s. n.s. None .132 .007 None n.s. n.s. None n 26 34 Em. w id th i ndex r .640 .650 .516 .562 p < .00 1 .235 None n.s. n.s. None .010 <.001 None n.s. n.s. None n 26 5 24 35 Temp. AP i ndex r .594 .802 .593 .857 .435 .628 .470 .544 p .002 .102 None .003 .000 Increase .038 < .00 1 None .004 < .00 1 None n 25 5 23 19 23 29 35 50 Table 532. Significant r values for TMJ dimensions and cranial breadth index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r .388 .344 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .012 .012 None n 41 53 Cond. AP i ndex r .181 .368 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .257 .007 None n 41 53 Em. w id th i ndex r .611 .031 .586 .265 .362 .425 .504 .387 p .001 .961 None .001 .246 None .082 .011 None .001 .004 None n 26 5 27 21 24 35 37 53 Temp. AP i ndex r .619 .357 .607 .644 .403 .591 p .001 .555 None .002 .003 None n.s. n.s. None .018 < .00 1 None n 25 5 23 19 34 49
177 Table 533. Significant r values for TMJ dimensions and cranial height index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Em. w id th i ndex r .513 .113 .572 .380 .372 .362 p .007 .857 None n.s. n.s. None .004 .024 None .022 .007 None n 26 5 24 35 38 54 Temp. AP i ndex r .592 .111 .556 .524 .418 .540 .428 .489 p .002 .859 None .006 .021 None .047 .003 None .010 < .00 1 None n 25 5 23 19 23 29 35 50 Table 534. Significant r values for TMJ dimensions and palate breadth index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r .420 .113 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .007 .415 Decrease n 40 54 Cond. AP i ndex r .032 .441 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .843 .001 Increase n 40 54 Em. w id th i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Temp. AP i ndex r .607 .089 p .002 .887 None n.s. n.s. None n.s. n.s. None n.s. n.s. None n 23 5
178 Table 535. Significant r values for TMJ dimensions and palate length index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Em. w id th i ndex r .628 .271 .326 .342 p n.s. n.s. None n.s. n.s. None .001 .115 Decrease .046 .011 None n 24 35 38 54 Temp. AP i ndex r .462 .833 .529 .425 .412 .578 p n.s. n.s. None .026 < .00 1 Increase .009 .021 None .014 < .00 1 None n 23 19 23 29 35 50 Table 536. Significant r values for TMJ dimensions and symphysis thickness index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r .270 .394 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .083 .003 None n 42 54 Em. w id th i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Temp. AP i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n
179 Table 537. Significant r values for TMJ dimensions and symphysis height index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Em. w id th i ndex r .298 .485 p n.s. n.s. None n.s. n.s. None .157 .003 None n.s. n.s. None n 24 35 Temp. AP i ndex r .638 .738 .389 .486 .226 .355 p n.s. n.s. None .001 < .00 1 None .067 .008 None .191 .011 None n 23 19 23 29 35 50 Table 538. Significant r values for TMJ dimensions and corpus height index at mental foramen. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r .005 .388 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .976 .004 Increase n 42 54 Em. w id th i ndex r .545 .190 p n.s. n.s. None .003 .411 Decrease n.s. n.s. None n.s. n.s. None n 27 21 Temp. AP i ndex r .610 .695 .356 .555 .231 .471 p n.s. n.s. None .002 .001 None .095 .002 None .182 .001 Increase n 23 19 23 29 35 50
180 Table 539. Significant r values for TMJ dimensions and corpus thickness index at M1/M2. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r .525 .414 .336 .344 p n.s. n.s. None .003 .062 None n.s. n.s. None .030 .011 None n 29 21 42 54 Em. w id th i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Temp. AP i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Table 540. Significant r values for TMJ dimensions and corpus height index at M1/M2. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r .452 .184 p .018 .767 None n.s. n.s. None n.s. n.s. None n.s. n.s. None n 27 5 Cond. AP i ndex r .541 .387 .070 .440 p n.s. n.s. None n.s. n.s. None .004 .024 None .661 .001 Increase n 26 34 42 54 Em. w id th i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Temp. AP i ndex r .532 .460 .065 .507 p n.s. n.s. None .009 .048 None n.s. n.s. None .709 < .00 1 Increase n 23 19 35 50
181 Table 541. Significant r values for TMJ dimensions and bigonial width index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r .475 .020 p n.s. n.s. None .011 .932 Decrease n.s. n.s. None n.s. n.s. None n 28 21 Cond. AP i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Em. w id th i ndex r .552 .509 .311 .388 p .003 .381 None n.s. n.s. None n.s. n.s. None .057 .004 None n 26 5 38 54 Temp. AP i ndex r .575 .378 .419 .341 p .003 .531 None n.s. n.s. None n.s. n.s. None .012 .015 None n 25 5 35 50 Table 542. Significant r values for TMJ dimensions and bicondylar breadth index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r .683 .302 .772 .356 .585 .500 .579 .576 p < .00 1 .621 None < .00 1 .113 Decrease .002 .003 None < .00 1 < .00 1 None n 27 5 29 21 25 34 42 54 Cond. AP i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Em. w id th i ndex r .775 .807 .586 .525 .533 .500 .514 .634 p <.001 .099 None .001 .015 None .009 .002 None .001 < .00 1 None n 26 5 27 21 23 35 38 54 Temp. AP i ndex r .501 .413 .545 .469 .357 .681 p .011 .490 None n.s. n.s. None .009 .010 None .035 < .00 1 Increase n 25 5 22 29 35 50
182 Table 543. Significant r values for TMJ dimensions and minimum ramus breadth index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r .553 .015 p n.s. n.s. None n.s. n.s. None .003 .932 Decrease n.s. n.s. None n 26 34 Cond. AP i ndex r .472 .322 .379 .459 .335 .441 p n.s. n.s. None .010 .154 None .056 .006 None .030 .001 None n 29 21 26 34 42 54 Em. w id th i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Temp. AP i ndex r .403 .596 .400 .411 p n.s. n.s. None .057 .007 None n.s. n.s. None .017 .003 None n 23 19 35 50 Table 544. Significant r values for TMJ dimensions and ramus height index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r .019 .355 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .907 .008 Decrease n 42 54 Cond. AP i ndex r .539 .227 .142 .350 p n.s. n.s. None n.s. n.s. None .005 .198 None .369 .010 None n 26 34 42 54 Em. w id th i ndex r .547 .204 p .004 .741 None n.s. n.s. None n.s. n.s. None n.s. n.s. None n 26 5 Temp. AP i ndex r .565 .362 .015 .493 p n.s. n.s. None n.s. n.s. None .005 .054 None .931 < .00 1 Increase n 23 29 35 50
183 Table 545. Significant r values for TMJ dimensions and mandibular length index. BF BM WF WM Pre Post r Change Pre Post r Change Pre Post r Change Pre Post r Change Cond. width i ndex r p n.s. n.s. None n.s. n.s. None n.s. n.s. None n.s. n.s. None n Cond. AP i ndex r .276 .953 .484 .087 .381 .592 p .164 .012 Increase .008 .706 Decrease n.s. n.s. None .013 < .00 1 Increase n 27 5 29 21 42 54 Em. w id th i ndex r .420 .184 p n.s. n.s. None n.s. n.s. None n.s. n.s. None .009 .183 None n 38 54 Temp. AP i ndex r .484 .886 .389 .519 p .014 .046 None n.s. n.s. None n.s. n.s. None .021 < .00 1 None n 25 5 35 50 Table 546. Summary of changes over time in covariation between TMJ dimensions and skull dimensions. BF BM WF WM Skull d imension Change Skull d imension Change Skull d imension Change Skull d imension Change Cond. w idth None N/A Bigonialw idth Decrease Min. ram. b readth Decrease Palate b readth Decrease Bicond. b readth Decrease Ramus h eight Decrease Cond. AP Mand. l e ngth Increase Mand. l ength Decrease None N/A Palate b readth Increase Corp. ht. ment. f or. Increase Corp. h t. M 1 M 2 Increase Mand. l ength Increase Em. w idth None N/A Corp. ht. ment. f or. Decrease Palate l ength Decrease None N/A Temp. AP None N/A Cranial l ength Increase None N/A Corp. ht. ment. f or. Increase Palate l ength Increase Corp. h t. M 1 M 2 Increase Bicondylar b reath Increase Ramus h eight Increase
184 Table 547. ANCOVA results for grayscale values for black females, with age as a covariate. Source Type III sum of s quares df Mean s quare F Sig. R s quared Lateral m ean Corrected m odel 266.564 2 133.282 .977 .389 .063 Age 19.290 1 19.290 .141 .710 Time p eriod 230.107 1 230.107 1.687 .204 Error 3956.430 29 136.429 Total 400128.003 32 Corrected t otal 4222.994 31 Central m ean Corrected m odel 29.888 2 14.944 .107 .899 .007 Age 8.898 1 8.898 .064 .802 Time p eriod 23.742 1 23.742 .170 .683 Error 4047.951 29 139.585 Total 480222.047 32 Corrected t otal 4077.840 31 Medial m ean Corrected m odel 430.011 2 215.006 1.368 .271 .086 Age 226.509 1 226.509 1.441 .240 Time p eriod 249.127 1 249.127 1.585 .218 Error 4558.755 29 157.198 Total 385499.837 32 Corrected t otal 4988.766 31 Overall m ean Corrected m odel 95.148 2 47.574 .430 .654 .029 Age 5.521 1 5.521 .050 .825 Time p eriod 93.377 1 93.377 .845 .366 Error 3206.011 29 110.552 Total 435338.407 32 Corrected t otal 3301.158 31 Ulna m ean Corrected m odel 30.488 2 15.244 .081 .922 .006 Age 11.524 1 11.524 .061 .806 Time p eriod 22.004 1 22.004 .117 .735 Error 5459.418 29 188.256 Total 478582.006 32 Corrected t otal 5489.907 31
185 Table 548. ANCOVA results for grayscale values for black males, with age as a covariate. Source Type III sum of s quares df Mean s quare F Sig. R s quared Lateral m ean Corrected m odel 934.513 2 467.257 2.984 .060 .113 Age 174.770 1 174.770 1.116 .296 Time p eriod 776.309 1 776.309 4.958 .031 Error 7359.753 47 156.590 Total 617411.219 50 Corrected t otal 8294.266 49 Central m ean Corrected m odel 1490.588 2 745.294 5.234 .009 .182 Age 207.833 1 207.833 1.460 .233 Time p eriod 1306.095 1 1306.095 9.172 .004 Error 6692.582 47 142.395 Total 665224.216 50 Corrected t otal 8183.170 49 Medial m ean Corrected m odel 2331.039 2 1165.520 8.233 .001 .259 Age 682.807 1 682.807 4.823 .033 Time p eriod 1696.855 1 1696.855 11.986 .001 Error 6653.530 47 141.564 Total 562068.558 50 Corrected t otal 8984.569 49 Overall m ean Corrected m odel 1341.882 2 670.941 5.150 .010 .180 Age 280.695 1 280.695 2.154 .149 Time p eriod 1086.050 1 1086.050 8.335 .006 Error 6123.733 47 130.292 Total 625461.284 50 Corrected t otal 7465.615 49 Ulna m ean Corrected m odel 214.024 2 107.012 .586 .561 .024 Age 178.258 1 178.258 .976 .328 Time p eriod 32.144 1 32.144 .176 .677 Error 8581.262 47 182.580 Total 979716.410 50 Corrected t otal 8795.285 49
186 Table 549. ANCOVA results for grayscale values for white females, with age as a covariate. Source Type III sum of s quares df Mean s quare F Sig. R s quared Lateral m ean Corrected m odel 517.950 2 258.975 1.750 .183 .058 Age 497.326 1 497.326 3.360 .072 Time p eriod 7.855 1 7.855 .053 .819 Error 8436.063 57 148.001 Total 519329.660 60 Corrected t otal 8954.013 59 Central m ean Corrected m odel 627.717 2 313.858 1.305 .279 .043 Age 626.309 1 626.309 2.604 .112 Time p eriod 78.228 1 78.228 .325 .571 Error 13948.631 58 240.494 Total 614377.027 61 Corrected t otal 14576.348 60 Medial m ean Corrected m odel 727.990 2 363.995 1.691 .193 .055 Age 659.338 1 659.338 3.063 .085 Time p eriod 249.605 1 249.605 1.160 .286 Error 12483.753 58 215.237 Total 469023.597 61 Corrected t otal 13211.743 60 Overall m ean Corrected m odel 626.478 2 313.239 1.879 .162 .061 Age 618.053 1 618.053 3.707 .059 Time p eriod 108.704 1 108.704 .652 .423 Error 9669.948 58 166.723 Total 555639.421 61 Corrected t otal 10296.426 60 Ulna m ean Corrected m odel 1552.887 2 776.443 7.800 .001 .212 Age 927.282 1 927.282 9.315 .003 Time p eriod 207.499 1 207.499 2.084 .154 Error 5773.725 58 99.547 Total 673716.866 61 Corrected t otal 7326.611 60
187 Table 550. ANCOVA results for grayscale values for white males, with age as a covariate. Source Type III sum of s quares df Mean s quare F Sig. R s quared Lateral m ean Corrected m odel 105.457 2 52.728 .617 .542 .013 Age 1.691 1 1.691 .020 .888 Time p eriod 88.313 1 88.313 1.033 .312 Error 8038.591 94 85.517 Total 894980.001 97 Corrected t otal 8144.048 96 Central m ean Corrected m odel 1241.360 2 620.680 5.612 .005 .107 Age 553.583 1 553.583 5.005 .028 Time p eriod 1197.419 1 1197.419 10.826 .001 Error 10397.045 94 110.607 Total 1073841.299 97 Corrected t otal 11638.405 96 Medial m ean Corrected m odel 1361.638 2 680.819 6.489 .002 .121 Age 943.281 1 943.281 8.991 .003 Time p eriod 1113.231 1 1113.231 10.611 .002 Error 9861.987 94 104.915 Total 840746.600 97 Corrected t otal 11223.625 96 Overall m ean Corrected m odel 800.196 2 400.098 5.016 .009 .096 Age 427.767 1 427.767 5.363 .023 Time p eriod 739.917 1 739.917 9.277 .003 Error 7497.209 94 79.758 Total 959647.474 97 Corrected t otal 8297.405 96 Ulna m ean Corrected m odel 367.780 2 183.890 1.616 .204 .034 Age 7.315 1 7.315 .064 .800 Time p eriod 225.903 1 225.903 1.985 .162 Error 10471.877 92 113.825 Total 1588135.686 95 Corrected t otal 10839.657 94
188 Table 551. Tests of interaction between age and time period. Group Source Type III sum of s quares df Mean s quare F Sig. Central m ean WM Corrected m odel 1241.360 3 413.787 3.701 .014 Time p eriod 63.034 1 63.034 .564 .455 Age 552.907 1 552.907 4.946 .029 Time period * A ge .000 1 .000 .000 .999 Error 10397.045 93 111.796 Total 1073841.299 97 Corrected t otal 11638.405 96 Medial m ean BM Corrected m odel 2383.656 3 794.552 5.537 .002 Time p eriod 13.649 1 13.649 .095 .759 Age 673.502 1 673.502 4.693 .035 Time period * A ge 52.616 1 52.616 .367 .548 Error 6600.914 46 143.498 Total 562068.558 50 Corrected t otal 8984.569 49 WM Corrected m odel 1405.251 3 468.417 4.437 .006 Time p eriod 1.474 1 1.474 .014 .906 Age 956.150 1 956.150 9.057 .003 Time period * A ge 43.613 1 43.613 .413 .522 Error 9818.374 93 105.574 Total 840746.600 97 Corrected t otal 11223.625 96 Overall m ean WM Corrected m odel 805.231 3 268.410 3.332 .023 Time p eriod 16.374 1 16.374 .203 .653 Age 430.456 1 430.456 5.343 .023 Time period * A ge 5.034 1 5.034 .062 .803 Error 7492.175 93 80.561 Total 959647.474 97 Corrected t otal 8297.405 96
189 Figure 544. Boxplots of lateral condylar grayscale mean by time period.
190 Figure 545. Boxplots of central condylar grayscale mean by time period.
191 Figure 546. Boxplots of medial condylar grayscale mean by time period.
192 Figure 547. Boxplots of overall condylar grayscale mean by time period.
193 Figure 548. Boxplots of ulna grayscale mean by time period.
194 Table 552. Regression results for lateral condylar grayscale mean and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression 104.296 1 104.296 .760 .390 .157 .025 Residual 4118.697 30 137.290 Total 4222.994 31 Black m ales Regression 1032.812 1 1032.812 6.827 .012 .353 .125 Residual 7261.454 48 151.280 Total 8294.266 49 White f emales Regression 5.186 1 5.186 .034 .855 .024 .001 Residual 8948.827 58 154.290 Total 8954.013 59 White m ales Regression 160.631 1 160.631 1.911 .170 .140 .020 Residual 7983.417 95 84.036 Total 8144.048 96 Table 553. Regression results for central condylar grayscale mean and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression 15.708 1 15.708 .116 .736 .062 .004 Residual 4062.131 30 135.404 Total 4077.840 31 Black m ales Regression 1591.824 1 1591.824 11.592 .001 .441 .195 Residual 6591.346 48 137.320 Total 8183.170 49 White f emales Regression 12.117 1 12.117 .049 .825 .029 .001 Residual 14564.230 59 246.851 Total 14576.348 60 White m ales Regression 477.531 1 477.531 4.065 .047 .203 .041 Residual 11160.873 95 117.483 Total 11638.405 96
195 Table 554. Regression results for medial condylar grayscale mean and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression 16.509 1 16.509 .100 .754 .058 .003 Residual 4972.257 30 165.742 Total 4988.766 31 Black m ales Regression 1548.238 1 1548.238 9.994 .003 .415 .172 Residual 7436.331 48 154.924 Total 8984.569 49 White f emales Regression 74.697 1 74.697 .335 .565 .075 .006 Residual 13137.046 59 222.662 Total 13211.743 60 White m ales Regression 275.058 1 275.058 2.387 .126 .157 .025 Residual 10948.567 95 115.248 Total 11223.625 96 Table 555. Regression results for overall condylar grayscale mean and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression 3.085 1 3.085 .028 .868 .031 .001 Residual 3298.074 30 109.936 Total 3301.158 31 Black m ales Regression 1255.737 1 1255.737 9.706 .003 .410 .168 Residual 6209.878 48 129.372 Total 7465.615 49 White f emales Regression 6.034 1 6.034 .035 .853 .024 .001 Residual 10290.392 59 174.413 Total 10296.426 60 White m ales Regression 273.873 1 273.873 3.243 .075 .182 .033 Residual 8023.532 95 84.458 Total 8297.405 96
196 Table 556. Regression results for ulna grayscale mean and year of birth. Sum of s quares df Mean s quare F Sig. R R s quared Black f emales Regression 19.508 1 19.508 .107 .746 .060 .004 Residual 5470.398 30 182.347 Total 5489.907 31 Black m ales Regression 3.175 1 3.175 .017 .896 .019 .000 Residual 8792.111 48 183.169 Total 8795.285 49 White f emales Regression 797.028 1 797.028 7.202 .009 .330 .109 Residual 6529.583 59 110.671 Total 7326.611 60 White m ales Regression 475.202 1 475.202 4.264 .042 .209 .044 Residual 10364.455 93 111.446 Total 10839.657 94
197 Figure 549. Plot of lateral condylar grayscale mean for black males by year of birth.
198 Figure 550. Plot of central condylar grayscale mean for black males by year of birth.
199 Figure 551. Plot of central condylar grayscale mean for white males by year of birth.
200 Figure 552. Plot of medial condylar grayscale mean for black males by year of birth.
201 Figure 553. Plot of overall condylar grayscale mean for black males by year of birth.
202 Figure 554. Plot of ulna grayscale mean for white females by year of birth.
203 Figure 5 55. Plot of ulna grayscale mean for white males by year of birth. Table 557. Summary of variables with statistically significant change over time for grayscale variables. Group ANCOVA Regression Direction of c hange Lateral grayscale m ean BM Significant Significant Decrease Central grayscale m ean BM Significant Significant Decrease WM Significant Significant Decrease Medial grayscale m ean BM Significant Significant Decrease WM Significant Not s ignificant Decrease Overall grayscale m ean BM Significant Significant Decrease WM Significant Not s ignificant Decrease Ulna grayscale m ean WF Not s ignificant Significant Increase WM Not s ignificant Significant Decrease
204 CHAPTER 6 DISCUSSION The primary goal of this study was to assess how diet related masticatory change may have influenced morphological change in the skulls of U.S. populations from the 19th and 20th centuries. This was achieved by examining change over time in the temporomandibular joint, which is a structure essential to the process of mastication. Prior to this investigation of change over time, two different samples were used to assess the utility of the variables used in this study to reflect differences in fu nction. Change over time in the severity of tooth wear was examined as morphological evidence for a dietary shift. Population variation and the possibility of morphological differences between skeletal collections were also investigated. This chapter pr esents the interpretation of the results of these analyses, and discusses other factors that may have influenced historical changes in the skeletons of populations in the United States. Preliminary Tests Nonhuman Primates When examining condylar morphology of monkey species with different ingestive behaviors, condylar dimensions were examined relative to mandibular length as a control for size. Although the femora were not available for this sample to use the femoral head as a control for size, mandibular length has been used previously as a biomechanical size proxy (Bouvier 1986) . Although no significant differences in the anteroposterior dimension of the condyle were found, the mediolateral width of the condyle differed significantly between species. Sp ecifically, Cercopithecus diana was significantly smaller than all other groups, and Cercocebus atys was significantly
205 smaller than Procolobus badius . The two colobus species, which have the toughest diets of the four species examined, exhibited the wides t condyles and were not significantly different from each other. This result is somewhat consistent with previous studies of dietary consistency and condylar dimensions. Other comparisons of species with naturally occurring variations in diet found that species with tougher, more folivorous diets tend to have wider condyles (Taylor 2005; Taylor 2006) , although this is not consistently found in all studies (Terhune 2013) . Comparisons of human groups with different subsistence strategies have found that s ofter and more processed diets are associated with reductions in the anteroposterior dimension of the condyle only (Hinton 1983; Whittaker et al. 1990) , though a more recent study examining the effects of industrialization on populations in London found reductions over time in the mediolateral width of the condyle only (Rando et al. 2014) . More controlled experimental studies have found a reduction in all dimensions of the condyle with softer diets (Bouvier 1988; Bouvier and Hylander 1984; Ravosa et al. 2007) . The greater width of condyles in species with tougher diets may be a reflection of the repetitive chewing necessary to process these foods. This hypothesis has recently been supported by an experimental study examining the effects of hard fallback foods on mandibular morphology (Scott et al. 2014) . Groups fed on consistently tougher diets (hay and hard pellets) were found to have larger cross sectional areas of the condyle compared to groups fed on consistently hard diets (pellets only). The great er condylar width may be a result of repetitive compressive loading on the lateral aspect of the condyle during chewing (Terhune 2011a; Terhune 2013) . The current study supports
206 the idea that repetitive loading and longer load durations may have a larger influence on condylar morphology than the magnitude of bite force since Cercocebus atys , which has the hardest diet, exhibits relatively narrower condyles compared to both colobus species. This study found no significant differences bet ween nonhuman primate species in bone density. However, Cercopithecus diana generally exhibited the lowest grayscale values in the condyle. This is consistent with the expected differences between these species in that the species with the softest diet ( C. diana) displayed the lowest bone density. The fact that Procolobus badius exhibited consistently higher bone density than Colobus polykomos is unexpected, since C. polykomos is thought to ingest a tougher diet than P. badius . Specifically , C. polykomos gnaws through the thick woody pods of Pentaclethra macrophylla, whereas P. badius eats softer seeds before they mature (Daegling and McGraw 2001; Koyabu and Endo 2009; McGraw and Zuberbuhler 2007) . However, this is consistent with previous studies of ma ndibular corpus robusticity, in that P. badius was found to have more robust corpora at the second molar compared to C. polykomos (Daegling and McGraw 2001) . These morphological differences may be related to systemic differences in bone density or rates o f bone deposition and resorption. Larger sample sizes may clarify if these observed patterns in bone density represent any biological significance. Tooth Loss The other method used to examine the appropriateness of the variables used in this study involved a comparison of white males with and without teeth. No significant differences were found in dimensions of the condyle or temporal aspect of the joint between edentulous and dentate groups. Although joint dimensions have been shown
207 to vary predictably with differences in function ( e.g. , dietary consistency), these studies involve differences in function that are present since birth ( e.g. , Ciochon et al. 1997; Ravosa et al. 2007) . Dental loss occurs later in life, generally long after growth and development has ceased. Therefore, any additional bone deposition or resorption due to change in function after growth is complete may not be significant enough to im pact overall dimensions of the joint. Although previous studies have demonstrated differences in the slope of the eminence with dental status (Granados 1979; Hinton 1981; Lawther 1956) , this study found no significant differences were found in eminence s lope at any aspect of the joint between edentulous and dentate groups. This disparity in result s may reflect differences in data collection methods compared to previous studies. While the study by Granados (1979) used clay to take an impression of the em inence and mandibular fossa in the center of the joint, Hinton (1981) used tracings of lateral skull photographs, and Lawther (1956) used tracings of lateral head radiographs . Additionally, unlike the present study, individuals with evidence of TMJ diseas e were not necessarily excluded from these previous studies. Therefore, some of these previous results may be a reflection of arthritic changes at the lateral aspect of the joint. Individuals without teeth exhibited significantly lower bone density throughout the condyle compared to those with teeth. This is consistent with previous studies of bone density variation and dental status (Giesen et al. 2003) . However, unlike previous work, this study incorporated an examination of bone density in a more â€œbiomechanically neutralâ€ element as a measure of general bone density throughout the body. This consideration is important in that people without teeth may be unable to maintain a
208 similar diet as those with teeth due to limitations in the ability to proces s foods. This may result in nutritional differences that could affect bone density and confound functional interpretations in the TMJ. The fact that this study found no significant differences in the bone density of the distal ulna between edentulous and dentate groups indicates that the density differences observed in the TMJ are due to functional rather than nutritional differences. Tooth Wear Degree of tooth wear was shown to decrease signifi cantly over time for all groups. A lthough significant result s for black females should be interpreted with caution due to a small post 1920 sample size, the consistency of this trend for all groups suggests that a similar result would be obtained with a larger sample of contemporary black females . This decrease in tooth wear over time provides evidence of a change in diet occurring since the 1850s. Specifically, the presence of tooth wear is associated with unrefined diets, and has decreased in populations since the Industrial Revolution due to the consumption of softer and easier to masticate foods (Addy and Shellis 2006) . Tooth wear generally occurs through the interaction between three processes: attrition, which involves the contact of teeth with each other; abrasion, which results from contact of teeth with e xtraneous substances such as grit; and erosion resulting from acidic substances (Addy and Shellis 2006; Powell 1988) . â€œThe rate of abrasion is directly influenced by the general consistency of the diet, which is a function both of food processing techniques and of the texture of foods as they are customarily consumedâ€ (Powell 1988) . Therefore, a reduction in tooth wear over time consistently observed across populations is indicative of a significant shift to a softer and more processed diet.
209 Variation Am ong Populations When examining population differences in temporomandibular joint morphology, significant differences were found for both morphometric and radiographic variables. Although the underlying reasons for this variation are unclear, these differences support the approach of this study in analyzing secular trends in the TMJ separately for each population. All linear measurements in these analyses were scaled to body size using the maximum diameter of the femoral head. For condylar dimensions, black females were demonstrated to have relatively larger condyles compared to all other groups, although condylar width was not significantly different from black males. More straightforward sexual dimorphism was observed in the temporal aspect of the joint , with both females groups exhibiting relatively larger dimensions compared to both male groups. Although examined relative to cranial dimensions rather than femoral head, Hinton (1983) found a similar pattern of relative temporomandibular joint size. Namely, rather than exhibiting the typical pattern of sexual dimorphism ( i.e. , males being larger than females), the examination of relative joint size reveals that females may exhibit relatively larger joints compared to males. This was particularly evident in a population of Eskimo males and females, but was less pronounced in populations of American white males and females (Hinton 1983) . White males exhibited significantly steeper eminence slopes compared to both female groups. However, the range of vari ation for this variable was much larger for white males, and overlapped entirely with the range of variation for both females groups, as well as black males. Due to this large amount of overlap, it is unlikely that this statistical difference represents any significant biological difference between groups.
210 The four populations differed in internal morphology of the condyle as well, with both black males and females exhibiting significantly denser bone compared to both white males and females. Black females exhibited the densest bone, while white females were at the other end of the spectrum. However, the male groups were significantly denser in the ulna compared to female groups, although there was considerable overlap in the range of variation between white males and black females. Denser bone in African Americans has been previously observed in studies of variation in bone density within the skeleton (Broman et al. 1958; Trotter et al. 1959) , in a study of histological aging in the rib used in forensi c anthropology (Cho et al. 2002) , general observations of ancestry related differences in bone texture in the skull (Bass 1995) , and studies of population differences in prevalence of fractures related to osteoporosis (Barret Connor et al. 2005) . In a stu dy of the difference of fracture rates in the hip, distal forearm, proximal humerus, and ankle between black and white geriatric populations, all four types of fractures were more common in whites (Baron et al. 1994) . These differences in bone mineral density are evident during childhood and adolescence, and may be a result of a combination of genetic, hormonal, and environmental factors (Braun et al. 2007; Pollitzer and Anderson 1989) . For example, Braun and colleagues (2007) have demonstrated that in a sample of adolescent females, black females exhibited a significantly higher average calcium retention compared to white females at a variety of calcium intake levels. Thus, the variation in bone density between populations observed in this study is consi stent with previous observations of density variation.
211 Variation Between Collections Unfortunately, due to small sample sizes, differences between collections could only be investigated using white females. These small samples sizes resulted from the nec essity to observe individuals from both collections during the same time period; however, there is a general lack of temporal overlap between collections. This lack of overlap was exacerbated by the requirement for dentate individuals in this study. Thes e individuals tended to be younger, which resulted in more recent dates of birth for those in the Bass Collection that were less likely to overlap with the Terry Collection. Using the white female sample, however, no significant differences were found bet ween collections for all measures of temporomandibular joint morphology as well as for bone density in the ulna. This supports the idea that the two collections used in this study represent a homogenous population despite originating from two separate loc ations. Thus, these samples can be used to assess change over time without concern for a significant effect of collection. Change Over Time Joint Dimensions and Eminence Slope Change over time was examined for dimensions of the condyle and eminence using an index involving the maximum diameter of the femoral head as a control for body size. This control was included in this study due to previous demonstrated changes over time i n stature and length of the long bones in the upper and lower limbs (Meadows Jantz and Jantz 1999; Trotter and Gleser 1951) . Therefore, differences observed in temporomandibular joint dimensions over time, as defined here, are not due to differences in ov erall body size.
212 For all populations, no changes over time were observed in dimensions of the condyle relative to body size. Previous studies associating condylar dimensions with dietary consistency have found that larger condyles are associated with a harder and more diff icult to process diet. This has been demonstrated using experimental studies (Bouvier 1988; Bouvier and Hylander 1981; Bouvier and Hylander 1984; Ciochon et al. 1997; Ravosa et al. 2007) , observations of species with natural variations i n diet (Taylor 2005; Taylor 2006) , and comparisons of humans with different subsistence strategies (Hinton 1983; Hinton and Carlson 1979; Rando et al. 2014; Whittaker et al. 1990) . This lack of a relative decrease in dimensions of the condyle over time is inconsistent with the idea that a shift to a softer and more processed diet has affected condylar morphology. Further, if no change over time has occurred in the morphology of the condyle, which is a structure directly involved in the process of masticat ion, it appears unlikely that dietary changes have altered the dimensions of the rest of the cranium and mandible. Although the external morphology of the condyle has not changed over time, this study indicates some morphological change in the temporal aspect of the joint. The mediolateral with of the eminence increased significantly over time for white males. However, the low r squared value associated with this result suggests that very little variation in this dimension is due to year of birth. A dditionally, this significant result may be due to several individuals from the Bass Collection with unusually large mediolateral eminence width indices. These considerations, in combination with the lack of change over time in this dimension in all other populations observed, suggest that this statistically significant change for white males may not hold any biological significance.
213 The anteroposterior dimension of the temporal increased over time for all groups, aside from the lateral aspect of the joint for white females which decreased over time. However, the r squared value for this decrease in white females is very low, which suggests that very little variation in this dimension is due to year of birth. The increase in the anteroposterior dimension of the temporal was most evident for all groups in the central aspect of the joint; white males and females also exhibited increases at the medial aspect of the joint. This change in morphology does not support the idea that this morphological change is due to a softer diet, since one would expect the joint to be smaller rather than larger (Hinton 1983; Hinton and Carlson 1979) . Lengthening of the anteroposterior dimension of the temporal aspect of the joint has been associated with increased sagittal sli ding of the condyle and a larger gape in nonhuman primates (Terhune 2011a; Terhune 2013; Wall 1995; Wall 1999) , but the need for an increased gape over time in humans seems unlikely based on the known dietary changes in the United States. However, this ch ange over time is consistent with increase in anteroposterior dimensions of the rest of the skull, including an increase in cranial length and facial depth (Angel 1976; Hunter and Garn 1969; Jantz 2001; Jantz and Meadows Jantz 2000; Martin and Danforth 2009; Moore Jansen 1989; Smith et al. 1986; Wescott and Jantz 2005) . This increase in the temporal aspect of the joint may be a byproduct of the larger anteroposterior dimensions observed in the rest of the skull. Future clarification regarding the nature o f this change may be required, since the anteroposterior measurement used in this study included the mandibular fossa for ease of standardization of measurement landmarks. It is unclear from the present study if the increase in this dimension occurs evenl y throughout the joint.
214 For white males, eminence slope has decreased over time, and is most evident at the lateral aspect of the joint. The eminenc e slope is shallower for whit e females over time as well, but this is only evident at the lateral aspect of the joint. The slope at the medial aspect of the joint has not exhibited change over time for any group. A decrease in eminence slope has been previously associated with a decrease in function, specifically with a greater amount of tooth loss and deg ree of tooth wear (Granados 1979; Hinton 1981; Lawther 1956) . A smaller slope has also been noted in nonhuman primates with less resistant diets (Terhune 2011b) . Since individuals with substantial tooth loss were not included in this study and tooth wear has decreased over time in this population, these factors are not confounding the decrease in slope found here. Although this is consistent with a shift to a softer diet, it is not consistent throughout the joint, and is not apparent in black males or females. The greater decrease in slope in the lateral aspect of the joint is not unexpected, since the lateral aspect of the joint is thought to be more heavily loaded during chewing (Hylander 1979) . Therefore, a decrease in force in the joint would affect this part of the joint more than the medial side. Differences between ancestral groups may be due to a less pronounced dietary shift for the black population; however, the consistency in decrease in tooth wear over time suggests this is not the case. Population differences in bone resorption and turnover rates may play a role if the decreased slope is due to resorption of bone on the surface of the joint. African Americans have been demonstrated to exhibit lower rates of bone resorption, as evidenced by lower urinary calcium levels, as well as slower bone turnover rates and greater rates of subperiosteal deposition of bone (Cho et al. 2006) . These differences are most pronounced at younger ages, which is
215 relevant for the current study in that the vast m ajority of black males and females in this sample were below 50 years of age. Covariation with Mandibular and Cranial Dimensions Change over time in how closely TMJ dimensions were correlated with other dimensions of the mandible and cranium was also inves tigated. A decrease over time in covariation between the morphology of the TMJ and dimensions of the skull would suggest a relaxation in selection due to a reduction of functional demands. While the significance of this change over time in covariation depends on the population and dimensions involved, a different pattern of change is observed for the anteroposterior and mediolateral dimensions of the TMJ. For the anteroposterior dimension, increases in covariation with several dimensions of the mandible and palate were observed for all groups except white females. Additionally, there was a decrease in covariation between the anteroposterior dimension of the condyle and mandibular length for black males. All groups aside from black females exhibited dec reases in covariation over time between the width of the TMJ and several dimensions of the mandible and palate. No change in covariation between TMJ dimensions and cranial vault dimensions are observed for any group other than black males; the correlation between the anteroposterior dimension of the temporal and cranial length increased over time. Thus, the covariation between dimensions of the mandible and palate have increased over time with respect to the anteroposterior dimension of the TMJ and decreased over time for the mediolateral dimension. It is possible that the mediolateral dimension of the joint may be more tied to dietary variations, since this dimension of the condyle was observed to vary with toughness of the diet when comparing nonhuman primates with
216 variations in ingestive behaviors. However, there was no change over time observed in this dimension for the human sample that would suggest a shift to a softer diet. The reasons for the increase in covariation with the anteroposterior dimensions of the joint are not clear, although the anteroposterior dimension of the temporal was the only TMJ dimension to exhibit change over time. This change may have resulted in greater covariance with other skull dimensions that have exhibited historical change. Such increases in covariation support Enlow â€™ s model of craniofacial growth involving concurrent development in adjacent skeletal structures (Enlow 1968; Enlow 1990; Enlow and Hans 1996) . Growth in one aspect of the skull ( e.g ., the anteroposter ior dimension of the temporal aspect of the TMJ) is influenced by simultaneous growth elsewhere ( e.g ., the anteroposterior dimension of the cranial vault). As such, TMJ dimensions may be affected more by skeletal development in the rest the cranium and ma ndible rather than the functional matrix, as suggested by Melvin Moss ( e.g ., Moss and Young 1960). Greater developmental influence of the functional matrix of the TMJ , which involves muscular activity associated with mastication function, should have resu lted in decreased TMJ dimensions and a reduction in covariation with other skull dimensions over time. Considering secular trends in the skull in light of the results from experimental animal studies provides further indication that changes in dietary consistency and a reduction in muscular activity are not likely to be driving these trends. The consumption of softer food has been shown to r esult in reduced craniofacial growth and smaller dimensions of the mandible, cranial vault, and face (Beecher and Corruccini 1981a; Beecher and Corruccini 1981b; Bouvier 1988; Bouvier and Hylander 1981; Bouvier and
217 Hylander 1984; Ciochon et al. 1997; Corruccini and Beecher 1982; Mavropoulos et al. 2004; Menegaz et al. 2010; Ravosa et al. 2007). Although the skulls of U.S. populations have narrowed over time, which is consistent with morphology associated with a softer diet, the increases in cranial height and length and facial height conflict with these experimental studies. The lack of reduction in TMJ dimensions found in the current study reinforce the minimal influence of dietary shifts on secular trends in the skull, since softer diets are associated with smaller TMJ dimensions. Bone Density The internal morphology of the condyle observed via radiography exhibits a pattern of change over time for males that is consistent with a shift to a softer diet . Density of the condyle has decreased over time for both black and white males; this change is most evident at the central and medial aspects of the condyle. This reduced density is consistent with previous studies obs erving lower condylar trabecular density with reduction in function (Bouvier 1988; Bouvier and Hylander 1984; Chen et al. 2009; Dias et al. 2011; Giesen et al. 2003; Mavropoulos et al. 2004; Tanaka et al. 2007; Yamada and Kimmel 1991) . A lack of change over time in the density of the ulna, which served as a measure of systemic bone density, bolsters the idea that the decreased density in the condyle is a result of decreased stress in the joint. If this bone density change was a result of changes in nutrition or health over time, one would expect to observe a similar decrease in density of the ulna. The female groups present a different pattern of change for bone density. Black females do not exhibit change over time in the density of the condyle or the ulna. However, as with all other variables in this study, interpretation of this result is complicated by a very small contemporary sample. Therefore, a larger sample should
218 be obtained before any justifiable conclusions can be drawn for this group. White females do not exhibit change over time in the density of the condyle, but exhibit an increase in the density of the ulna. Therefore, relative to systemic bone density, the density of the condyle did exhibit a decrease over time. The increase in bone density in the ulna for white females was not observed for any other group, and may be related to the diagnosis and treatment for osteoporosis . Although the term â€œosteoporosisâ€ was first used in the 1820s by a French pathologist, it was not defined as a clinical syndrome until the late 1940s by an American endocrinologist (O'Neill 2005) . Inconsistencies in the bone mass threshold used to diagnose osteoporosis existed until 1994, when the World Health Organization defined the presence of osteoporosis as involving â€œbone mass which is less than 2.5 standard deviations from the mean value of bone mass in young adult womenâ€ (Oâ€™Neill 2005: iv33). The initial definition of the condition an d subsequent understanding of the causes of the disease likely increased public awareness of preventative measures, such as the intake calcium supplements. The use of vitamin and mineral supplements, including calcium, has increased in the latter half of the 20th century, particularly after concerns about the nutritional content of processed foods became more prevalent after World War II (Levenstein 2003; Millen et al. 2004) . Calcium enriched diets, especially during growth, have been demonstrated to incr ease bone mass and reduce risk of the osteoporotic fracture later in life (Bonjour et al. 1997; Lee et al. 1994; Ulrich et al. 1996) . However, as the exact diets of the populations used in the current study are unknown, it is not clear if this explains the increase in bone density observed in white females.
219 An additional consideration for the interpretation of these results involves socioeconomic status. Although the Terry Collection and early portion of the Bass Collection are both composed of individual s of low socioeconomic status, the most recent portion of the sample may involve those of higher socioeconomic status. This is due to a recent influx of self donations to the Bass Collection over the past two decades (Christensen 2006; Hunt and Albanese 2005) . Those of higher socioeconomic status would be able to afford more nutritious foods, as well as take advantage of vitamin supplements that at beneficial to skeletal health (Bonjour et al. 1997; Levenstein 2003) . Therefore, one would expect those of higher socioeconomic status to exhibit denser bone. The fact that a significant decrease in condylar density was observed despite the fact that socioeconomic status for the most recent individuals may be higher provides additional support for a functional interpretation of these results. Therefore, this study demonstrates some influence of food processing and changing dietary habits on the morphology of the temporomandibular joint. However, the lack of changes to the external morphology of the joint cons istent with a shift to a softer diet suggests that changes in food processing have not contributed to secular change in the rest of the skull. Morphological change over time in other aspects of the skull that are directly related to mastication ( e.g ., mas ticatory muscle attachment areas) should be investigated before the possible influence of changes in masticatory function is completely ruled out. Complicating Factors Although the dietary trends over the time period covered in this study generally involve a shift to a lighter diet of softer and more processed foods that would require less demanding masticatory activity , frequency of chewing gum use has increased since
220 it was first marketed in the U.S. in the mid19th century . This would result in more chewing cycles over time, and would particularly affect masticatory activity in children after the introduction of bubble gum in the 1930s (Gustaitis 1998; Robinson 2004) . More frequent masticatory loads, particularly early in development , may have counteracted the decrease in masticatory activity resulting from historical changes in diet. Although chewing gum use in the study population is unknown, the fact that decreases in bone density were observed over time indicates that changes in chewing gum use m ay not have significantly altered morphology. Decreases in condylar bone density without a concurrent reduction in joint dimensions were found in edentulous individuals compared to individuals with teeth, suggesting that this internal morphology can respond to changes in masticatory loads over a relatively short period of time ( e. g. , tooth loss occurring later in life). If increases in chewing gum use counteracted dietary changes, one would expect either no change or an increase in bone density over time. Additionally, a side from the increase in the anteroposterior dimension of t he temporal aspect of the joint which is likely a byproduct over other dimensional changes in the cranial vault, no morphological changes in the TMJ were consistent with a shift t o greater masticatory activity. A more complex issue related to this research is that changes in diet over time in the United States have not only affected the biomechanical environment associated with chewing. Changes in nutrition and health have also ac companied advancements in food processing, which may have influenced craniofacial growth. Specifically, there is a reduction in nutritional value associated with the consumption of processed foods (Cordain 2007; Cordain et al. 2005; Levenstein 2003) . Add itionally, the consumption of
221 less food over time, which accompanied a shift to a more sedentary lifestyle, would also result in the ingestion of fewer nutrients. However, particularly in the mid1900s, food processing companies started adding nutritional supplements and other additives to their products due to concerns from the public (Levenstein 2003; Levenstein 1996) . The influence of these shifts in nutrition on craniofacial form is not clear, as studies investigating this relationship appear to be few in number and limited to experimental animal studies. These studies have investigated the effects of malnutrition in rats (Pucciarelli 1981; Pucciarelli and Oyenhart 1987) , squirrel monkeys (Dressino and Pucciarelli 1997; Pucciarelli et al. 1990) , and pigs (Luke et al. 1979) . In general, these studies have found arrested growth in the skull as a whole and in the face in particular , as well as an associated reduction in tooth size. However, it is unknown if the additives used to counteract the negative nutritional effects of food processing would mitigate these developmental issues. Further complicating this analysis of diet and change over time is the fact that changes in health and nutrition also vary by location and socioeconomic stat us. Differences in stature between those living in rural and urban areas suggest that different conditions in these two areas can affect overall growth (Cuff 2005; Larkin 1988; Levenstein 1988) . For example, in a study examining the variation in stature of 20,000 Pennsylvanian Civil War soldiers, Cuff (2005) found that those born in rural areas were consistently taller than those from urban areas. This is consistent with a shift to a more sedentary lifestyle, since the closer and more crowded living conditions associated with an urban environment facilitates the spread of disease (Larsen 1995) . Living in urban areas also increases the likelihood of eating more processed foods , which involve
222 reduced nutritional value (Cordain 2007; Cordain et al. 2005; Levenstein 2003) . However, the food supply in rural areas was often dependent on the seasons and could be affected by environmental conditions (Dirks 2003; Larkin 1988) . Within locations, health disparities exist between the rich and the poor, which is ref lected in the differences in height between groups of different socioeconomic status; the poor tended to be shorter compared to more affluent individuals (Larkin 1988; Levenstein 1988) . Aside from being in better health due to a better diet, the rich were more likely to have access to medical care. Doctors were less often found in rural areas, which led people to turn to folk medicine and herbal remedies in times of sickness (Larkin 1988) . Thus, the health and nutrition intake of U.S. populations has var ied based on location and socioeconomic status, and have likely changed over time due to population movement from rural to urban areas. Influence of these changes on craniofacial growth and development are likely yet specific effects are unknown. An addit ional nutrition related factor that may be influencing craniofacial growth in recent years is the rise in rates of obesity, which has increased 30% over the last 50 years (Danubio and Sanna 2008) . This trend is even more pronounced for those in lower soci oeconomic groups (Anderson and Butcher 2006; Danubio and Sanna 2008; Levenstein 2003) . Currently, 34% of U.S. adults are obese, compared to 12% in 1991 (Eckel et al. 2011; Mokdad et al. 2003) . Although the childhood obesity epidemic did not start until t he 1980s, rates are now increasing as rapidly as that for obesity in adults (Anderson and Butcher 2006; Danubio and Sanna 2008) . Since the 1970s, daily caloric intake has increased by 150300 kcal/day, and part of this increase is due to a greater proport ion calories from beverages (Neeley and Gonzalez 2007; Popkin 2011) .
223 Similarly, the prevalence of type 2 diabetes has increased from 4.9% of U.S. adults diagnosed in 1990 compared to 11% currently (Eckel et al. 2011; Mokdad et al. 2003) . This increase is due in part to insulin resistance associated with obesity, and the increased rates of childhood obesity have led to the development of type 2 diabetes in children (Neeley and Gonzalez 2007) . In general, an increase in caloric intake coupled with a more s edentary lifestyle has contributed to these recent trends in weight gain and type II diabetes (Anderson and Butcher 2006) . This combination of changing activity patterns and energy intake may have had an influence on craniofacial development in the past 150 years. Obese individuals exhibit absolute denser bone compared to control groups, but less dense bone relative to their body weight (Clark et al. 2006; Leonard et al. 2004; Neeley and Gonzalez 2007; Rocher et al. 2008) . Thus, while increased weight i nvolves greater stress on the skeleton and triggers a bone growth response, this response is not proportional to the amount of extra weight carried by obese individuals (Leonard et al. 2004; Rocher et al. 2008) . Although type 1 diabetes is generally assoc iated with decreased bone mass, the effect of type 2 diabetes on the skeleton is more complicated (Auwerx et al. 1988; Bouillon 1991; Krakauer et al. 1995) . Since type 2 diabetes develops in obese subjects who tend to have greater bone mass, mixed results have been observed when investigating the connection between type 2 diabetes and bone density (Auwerx et al. 1988; Bouillon 1991) . Krakauer and colleagues (1995) have suggested that diabetes is associated with lower rates of bone formation and turnover. Therefore, individuals with type 1 diabetes would have lower bone mineral density due to early age of onset. Those with type 2 diabetes tend to develop the disease later in life compared to type 1
224 diabetics, and lower bone turnover rates would result in decreased bone loss (Krakauer et al. 1995) . The increased bone mass observed in obese individuals is consistent with studies of craniofacial abnormalities associated with obesity. Specifically, larger dimensions have been found in obese individuals for ma ndibular length, total facial height, and the width of the skull base, face, and mandible; however, a reduction in upper facial height has also been found (Neeley and Gonzalez 2007) . The increased height and mandibular length dimensions are consistent wit h observed craniofacial secular trends in the U.S., but a narrowing of the cranial vault and face has occurred over time. Since the increase in obesity rates has been relatively recent, the possible effects of this condition on craniofacial form may not b e observed in the current sample. If future studies of craniofacial trends find a recent reversal in the trend toward a narrowing of the skull, increased obesity rates may be implicated. Aside from the increased prevalence of diet related disease, other healthrelated changes have been associated with industrialization. Although the time period examined excludes more recent skeletal change, Lewis (2002) compared several populations from medieval and postmedieval England (8501859 AD) with the goal of a ssessing health effects of industrialization. The industrialized population exhibited reduced rates of dental enamel defects (enamel hypoplasias), evidence of nonspecific infections (periostitis), and maxillary sinusitis compared to nonindustrialized rural and urban populations, but similar rates of anemia (cribra orbitalia). However, enamel defects and cribra orbitalia both occur much earlier in life in the industrialized population, indicating a period of stress for young infants; many enamel defects formed
225 as early as 6 months old. Additionally, the presence of enamel defects in deciduous teeth suggests a period of stress in utero, which would result from poor maternal health (Lewis 2002) . T he presence of enamel defects in living populations has been linked to nutritional status (Larsen 1995) . As such, changes in maternal nutrition may have an effect on skeletal development in offspring. This has been tested directly in an experimental setting by Fernandes and colleagues (2008) by restricting maternal protein and energy consumption in rats and measuring the skull dimensions of their offspring. In this study, malnutrition of the mother was started at the birth of the offspring to investigate the effects of malnutrition during lactation. Despite the fact that malnutrition did not occur in utero, the offspring of the protein and energy restricted mothers exhibited reductions in all skull dimensions as well as lower body weights (Fernandes et al. 2008) . Although increases in height and anteroposterior length of the cranial vault and face have occurred in the United States, a lack of adequate nutrition associated with the greater consumption of processed foods may have contributed to the narrowing in the faces and vaults of U.S. populations. Thus, the influence of changing nutrition over time, involving the intake of fewer vitamins and minerals, an increase in caloric intake, and a decrease in energy expenditure, should be investigated in the context of skeletal morphological trends in order to better understand the causal factors involved in craniofacial change.
226 CHAPTER 7 SUMMARY AND CONCLUSIONS The connection between diet and craniofacial form has been examined in a variety of human populat ions that span a range of time periods and geographic locations, as well as in various animal species that vary in ingestive behaviors. The current study investigated this connection in the context of recent craniofacial change in the United States. Change over time in the temporomandibular joint was used to assess the possibility that a shift to a lighter, more processed diet resulting from advances in food processing and other diet related trends has altered craniofacial dimensions in U.S. populations. If changes in diet have altered craniofacial dimensions, one would expect dimensions of a structure so closely tied to mastication to change as well. Aside from examining gross morphological change in the joint, the trabecular structure of the condyle was also examined in the event that dietary changes have resulted in more subtle changes to its internal morphology. Although the skeletal collections used in this study are well documented, specific dietary information for these individuals is unknown. To better understand the dietary habits of the population used in this study, the severity of tooth wear was scored for all individuals. A decrease in tooth wear since the 1850s was observed for all populations, which supports the assumption that dietary consistency has changed over time. Specifically, this change has involved a shift to more processed food containing less grit, which would result in a decrease in stress in the temporomandibular joint. This finding supports the assumption that changes in temporomandibular joint morphology should be observed despite the relatively short time period involved in this study.
227 Analysis of temporomandibular joint morphology found that overall dimensions of the joint did not change, aside from the anteroposterior dimension of the temporal aspect of the joint, which increased. This is not consistent with previous studies that found that softer and more processed diets were associated with smaller TMJ dimensions. In contrast, the decrease in eminence slope over ti me noted in the lateral aspect of the joint for white males and females supports a shift to reduced forces in the TMJ. Since the joint is thought to be more heavily loaded at that aspect of the joint, it is expected that the slope on the lateral side woul d be the most affected by alterations in joint reaction force. Analysis of bone density was also consistent with a shift to a more processed diet. Specifically, bone density of the condyle decreased over time relative to systemic bone density, as assesse d relative to the density of the ulna. This decrease in condylar density without a concurrent decrease in density elsewhere in the skeleton strongly suggests a biomechanical explanation for this change. Thus, subtle differences in temporomandibular joint morphology have occurred that are likely related to dietary changes over time. However, since this biomechanical variation did not result in overall dimensional change in the joint, it appears unlikely that dietary change has altered the morphology of ot her aspects of the skull. This study was significant both in the context of investigation the link between craniofacial form and diet, and for the question of what factors are driving U.S. craniofacial secular change. Connections were found between mediol ateral condylar width and ingestive behavior in nonhuman primate species. Analysis of condylar bone density provided further evidence of the influence of masticatory forces on craniofacial morphology. Cercopithecus diana, which was the nonhuman primate s pecies with the
228 softest diet, was found to have the lowest bone density in the condyle. Bone density was also found to be reduced in the condyles of edentulous individuals with no corresponding decrease in systemic bone density. Since edentulous individuals have been shown to produce smaller bite forces, this reduction in bone density likely results from decreased force in the joint during chewing. As discussed above, change over time in condylar bone density was also found, which corresponds to a shift to a softer, more processed diet. Thus, this study provides evidence of subtle changes in morphology that accompany changes in masticatory function. Although the lack of gross morphological chang e in the TMJ suggests that that change over time in food consistency and the amount of food consumed have not altered craniofacial form in the United States, this finding provides several avenues for future research. Since changes in nutrition have accompanied advancements in food processing, a better understan ding of nutritional effects on craniofacial form may clarify the underlying causes for craniofacial change in the United States. This study was limited in that specific dietary information was not available for the populations used. However, analysis of dental calculus may provide some insight into the dietary habits of the individuals typically used for studies of change over time in the United States. If more specific dietary information for this population was known, a better assessment of nutritional changes over time could be made. These may include a reduction in the intake of certain vitamins and minerals, as well as change over time in overall caloric intake. The effects of these nutritional changes on craniofacial form could be investigated in an experimental setting, and results could be compared to the observed secular trends in craniofacial form.
229 Other changes in health and stress over time may provide insight into change over time in the skull. Growth disturbances may be evaluated through s keletal indicators of stress, such as linear enamel hypoplasias or Harris lines. Since these features are indicative of arrests in skeletal growth, it is reasonable to assume that craniofacial dimensions could be affected by these periods of stress . Chan ge over time in the number and extent of these growth disturbances would explain some portion of trends observed in the long bones as well as in the skull. However, the relationship between these stress indicators and craniofacial form must be better understood in order to assess their influence on trends in the skull. Additionally, since much of craniofacial development occurs in utero and within the first few years after birth, the effects of maternal health on craniofacial growth may also be of interes t in the context of secular trends in the skull. Investigations into the nutritional status of mothers, which has possibly changed over time, and how this may affect craniofacial growth in offspring may further elucidate the causes of craniofacial change over time in the United States .
230 LIST OF REFERENCES Ackermann RR. 2002. Patterns of covariation in the hominoid craniofacial skeleton: implications for paleoanthropological models. J Hum Evol 42:167187. Ackermann RR, and Cheverud JM. 2000. Phenotypic covariance structure in tamarins (genus Saguinus ): a comparison of variation patterns using matrix correlation and common principal component analysis. Am J Phys Anthropol 111:489 501. Addy M, and Shellis RP . 2006. Interaction between attrition, abrasion and erosion in tooth wear. In: Lussi A, editor. Dental Erosion: From Diagnosis to Therapy. Basel: Karger. p 1731. Anderson PM, and Butcher KF. 2006. Childhood obesity: trends and potential causes. Future Child 16:1945. Angel JL. 1976. Colonial to modern skeletal change in the U.S.A. Am J Phys Anthropol 45:723736. Aufderheide AC, and Rodrguez Martn C. 1998. The Cambridge Encyclopedia of Human Paleopathology. Cambridge: Cambridge University Press. Auwer x J, Dequeker J, Bouillon R, Gausens P, and Nijs J. 1988. Mineral metabolism and bone mass at peripheral and axial skeleton in diabetes mellitus. Diabetes 37:812. Baron JA, Barrett J, Malenka D, Fisher E, Kniffin W, Bubolz T, and Tosteson T. 1994. Racial differences in fracture risk. Epidemiology 5:4247. Barret Connor E, Siris ES, Wehren LE, Miller PD, Abbott TA, Berger ML, Santora AC, and Sherwood LM. 2005. Osteoporosis and fracture risk in women of different ethnic groups. J Bone Miner Res 20:185194. Bass WM. 1995. Human Osteology: A Laboratory and Field Manual , 4th Ed . Columbia, MO: Missouri Archaeological Society. Beale CL. 2000. A century of population growth and change. Food Rev 23:1622. Beecher RM, and Corruccini RS. 1981a. Effects of dietary consistency on craniofacial and occlusal development in the rat. Angle Orthod 51:6169. Beecher RM, and Corruccini RS. 1981b. Effects of dietary consistency on maxillary arch breadth in macaques. J Dent Res 60:68. Beek M, Aarnts MP, Koolstra JH, Feilzer AJ, and van Eijden TMGJ. 2001. Dynamic properties of the human temporomandibular joint disc. J Dent Res 80:876 880.
231 Behrents RG, Carlson DS, and Abdelnour T. 1978. In vivo analysis of bone strain about the sagittal suture in Macaca mulatta during masticatory movements. J Dent Res 57:904908. Bonjour J, Carrie A, Ferrari S, Clavien H, Slosman D, Theintz G, and Rizzoli R. 1997. Calcium enriched foods and bone mass growth in prepubertal girls: a randomized, doubleblin d, placebocontrolled trial. J Clin Invest 99:12871294. Bouillon R. 1991. Diabetic bone disease. Calcified Tissue Int 49:155160. Bouvier M. 1986. Biomechanical scaling of mandibular dimensions in New World monkeys. Int J Primatol 7:551567. Bouvier M. 1988. Effects of age on the ability of the rat temporomandibular joint to respond to changing functional demands. J Dent Res 67:12061212. Bouvier M, and Hylander WL. 1981. Effect of bone strain on cortical bone structure in macaques ( Macaca mulatta). J Morphol 167:112. Bouvier M, and Hylander WL. 1984. The effect of dietary consistency on gross and histologic morphology in the craniofacial region of young rats. Am J Phys Anthropol 170:117126. Bowers DE. 2000. Cooking trends echo changing roles of wom en. Food Rev 23:2329. Boyd RL, Gibbs CH, Mahan PE, Richmond AF, and Laskin JL. 1990. Temporomandibular joint forces measured at the condyle of Macaca arctoides . Am J Orthod Dentofac 97:472479. Brace CL, and Mahler PE. 1971. Post Pleistocene changes in the human dentition. Am J Phys Anthropol 34:191204. Braun M, Palacios C, Wigetrz K, Jackman LA, Bryant RJ, McCabe LD, Martin BR, McCabe GP, Peacock M, and Weaver CM. 2007. Racial differences in skeletal calcium retention in adolescent girls with varied c ontrolled calcium intakes. The Am J Clin Nutr 85:1657 1663. Broman GE, Trotter M, and Peterson RR. 1958. The density of selected bones of the human skeleton. Am J Phys Anthropol 16:197211. Calcagno JM. 1986. Dental reduction in Post Pleistocene Nubia. A m J Phys Anthropol 70:349363. Carlson DS, and Van Gerven DP. 1977. Masticatory function and post Pleistocene evolution in Nubia. Am J Phys Anthropol 46:495506.
232 Carter DR, and Beaupr GS. 2001. Skeletal Function and Form: Mechanobiology of Skeletal Development, Aging, and Regeneration. Cambridge: Cambridge University Press. Chen J, Sorensen KP, Gupta T, Kilts T, Young M, and Wadhwa S. 2009. Altered functional loading causes differential effects in the subchondral bone and condylar cartilage in the temporomandibular joint from young mice. Osteoarthr Cartilage 17:354361. Cho H, Stout SD, and Bishop TA. 2006. Cortical bone remodeling rates in a sample of African American and European American descent groups from the American Midwest: comparisons of age and sex in ribs. Am J Phys Anthropol 130:214226. Cho H, Stout SD, Madsen R W, and Streeter MA. 2002. Populationspecific histological age estimating method: a model for known AfricanAmerican and EuropeanAmerican skeletal remains. J Forensic Sci 47:1218. Christensen AM. 2006. Moral considerations in body donation for scientifi c research: a unique look at the University of Tennesseeâ€™s Anthropological Research Facility. Bioethics 20:136145. Ciarelli TE, Fyhrie DP, Schaffler MB, and Goldstein SA. 2000. Variations in threedimensional cancellous bone architecture of the proximal femur in female hip fractures and in controls. J Bone Miner Res 15:3240. Ciochon RL, Nisbett RA, and Corruccini RS. 1997. Dietary consistency and craniofacial development related to masticatory function in minipigs. J Cran Genet Dev Bio 17:96102. Clark EM, Ness AR, and Tobias JH. 2006. Adipose tissue stimulates bone growth in prepubertal children. J Clin Endocr Metab 91:25342541. Co rdain L. 2007. Implications of PlioP leistocene hominin diets for modern humans. In: Ungar PS, editor. Evolution of the H uman Diet: the Known, the Unknown, and the Unknowable. New York, NY: Oxford University Press. p 363383. Cordain L, Boyd Eaton S, Sebastian A, Mann N, Lindeberg S, Watkins BA, O'Keefe JH, and BrandMiller J. 2005. Origins and evolution of the Western diet : health implications for the 21st century. The Am J Clin Nutr 81:341 354. Corruccini RS, and Beecher RM. 1982. Occlusal variation related to soft diet in a nonhuman primate. Science 218:7476. Courtless JC. 1994. Trends in tobacco use. Fam Econ Rev 7:2839. Crothers AJR. 1992. Tooth wear and facial morphology. J Dent 20:333341.
233 Cuff T. 2005. The Hidden Cost of Economic Development: the Biological Standard of Living in Antebellum Pennsylvania. Burlington, VT: Ashgate Publishing Limited. Daegling DJ, and Grine FE. 2007. Mandibular biomechanics and the paleontological evidence for the evolution of the human diet. In: Ungar PS, editor. Evolution of the Human Diet: the Known, the Unknown, and the Unknowable. New York, NY: Oxford University Press. p 77105 . Daegling DJ, and McGraw WS. 2001. Feeding, diet, and jaw form in the West African Colobus and Procolobus . Int J Primatol 22:10331055. Danubio MA, and Sanna E. 2008. Secular changes in human biological variables in western countries: an updated review and synthesis. J Anthropol Sci 86:91112. Dias GJ, Cook RB, and Mirhosseini M. 2011. Influence of food consistency on growth and morphology of the mandibular condyle. Clin Anat 24:590 598. Dirks R. 2003. Diet and nutrition in poor and minority communities in the United States 100 years ago. Annu Rev Nutr 23:81100. Dressino V, and Pucciarelli HM. 1997. Cranial growth in Saimiri sciureus (Cebidae) and its alteration by nutritional factors: a longitudinal study. Am J Phys Anthropol 102:545554. Dyson LK. 2000. American cuisine in the 20th century. Food Rev 23:27. Eckel RH, Kahn SE, Ferrannini E, Goldfine AB, Nathan DM, Schwartz MW, Smith RJ, and Smith SR. 2011. Obesity and type 2 diabetes: what can be unified and what needs to be individualized? J Clin E ndocr Metab 96:16541663. Enlow DH. 1968. The Human Face. New York: Harper & Row. Enlow DH. 1990. Facial Growth. Philadelphia, PA: W.B. Saunders Company. Enlow DH, and Hans MG. 1996. Essentials of Facial Growth. Philadelphia, PA: W.B. Saunders Company. Eversole LR, Pappas JR, and Graham R. 1985. Dental occlusal wear and degenerative disease of the temporomandibular joint: a correlational study utilizing skeletal material from a contemporary population. J Oral Rehabil 12:401406. Farias Gomes SG, Custodio W, Moura Jufer JS, Del Bel Cury AA, and Rodrigues Garcia RCM. 2010. Correlation of mastication and masticatory movements and effect of chewing side preference. Braz Dent J 21:351355.
234 Fernandes RM, Abreu AV, Silva RB, Silva DF, Martinez GL, Babinski MA, and Ramos CF. 2008. Maternal malnutrition during lactation reduces skull growth in weaned rat pups: experimental and morphometric investigation. Anat Sci Int 83:123 130. Field A. 2013. Discovering Statistics Using IBM SPSS Statistics. London: Sage. Frank L. 1950. Muscular influence on occlusion as shown by x rays of the condyle. Dent Dig 56:484488. Frankel VH, and Burstein AH. 1970. Orthopaedic Biomechanics. Philadelphia, PA: Lea and Febiger. Gabaccia DR. 1998. We Are What We Eat: Ethnic Food and the Making of Americans. Cambridge, MA: Harvard University Press. Gately I. 2001. Tobacco: The Story of How Tobacco Seduced the World. New York, NY: Grove Press. Giesen EBW, Ding M, Dalstra M, and Van Eijden TMGJ. 2003. Reduced mechanical load decreases the density, stiffness, and strength of cancellous bone of the mandibular condyle. Clin Biomech 18:358363. Gingerich PD. 1971. Functional significance of mandibular translation in vertebrat e jaw mechanics. Postilla 152:310. Gingerich PD. 1979. The human mandible: lever, link, or both? Am J Phys Anthropol 51:135138. Goose DH. 1962. Reduction of palate size in modern populations. Arch Oral Biol 7:343 350. Goose DH. 1981. Changes in human face breadth since the Mediaeval period in Britain. Arch Oral Biol 26:757758. Granados JI. 1979. The influence of the loss of teeth and attrition on the articular eminence. J Prosthet Dent 42:78 85. Gustaitis J. 1998. The sticky history of chewing gum. Am Hist 33:3037. Helkimo E, Carlsson GE, and Helkimo M. 1977. Bite force and state of dentition. Acta Odontol Scand 35:297303. Herring SW, and Liu ZJ. 2001. Loading of the temporomandibular joint: anatomical and in vivo evidence from the bones. Cells Tissues Organs 169:193200.
235 Herring SW, and Teng S. 2000. Strain in the braincase and its sutures during function. Am J Phys Anthropol 112:575593. Hillson S. 1996. Dental Anthropology. Cambridge: Cambridge University Pres s. Hinton RJ. 1981. Changes in articular eminence morphology with dental function. Am J Phys Anthropol 54:439455. Hinton RJ. 1983. Relationships between mandibular joint size and craniofacial size in human groups. Arch Oral Biol 28:3743. Hinton RJ, and Carlson DS. 1979. Temporal changes in human temporomandibular joint size and shape. Am J Phys Anthropol 50:325 334. Hodges DC. 1991. Temporomandibular joint osteoarthritis in a British skeletal population. Am J Phys Anthropol 85:367 377. Hooker RJ. 198 1. Food and Drink in America: A History. Indianapolis, IN: The Bobbs Merrill Company, Inc. Hunt DR, and Albanese J. 2005. History and demographic composition of the Robert J. Terry anatomical collection Am J Phys Anthropol 127:406417. Hunter SW, and Gar n SM. 1969. Evidence for a secular trend in face size. Angle Orthod 39:320323. Hylander WL. 1975. The human mandible: lever or link? Am J Phys Anthropol 43:227 242. Hylander WL. 1978. Incisal bite force direction in humans and the functional significanc ce of mammalian mandibular translation. Am J Phys Anthropol 48:17. Hylander WL. 1979. An experimental analysis of temporomandibular joint reaction force in macaques. Am J Phys Anthropol 51:433456. Hylander WL, and Johnson KR. 1992. Strain gradients in the craniofacial region of primates. In: Davidovitch Z, editor. The Biological Mechanisms of Tooth Movement and Craniofacial Adaptation. Columbus, OH: The Ohio State University College of Dentistry. p 559569. Ikebe K, Matsuda K, Kagawa R, Enoki K, Okada T, Yoshida M, and Maeda Y. 2012. Masticatory performance in older subjects with varying degrees of tooth loss. J Dent 40:71 76.
236 Ikebe K, Matsuda K, Murai S, Maeda Y, and Nokubi T. 2010. Validation of the Eichner Index in relation to occlusal force and mas ticatory performance. Int J Prostodont 23:521524. Ikebe K, Nokubi T, Morii K, Kashiwagi J, and Furuya M. 2005. Association of bite force with ageing and occlusal support in older adults. J Dent 33:131 137. Jantz RL. 2001. Cranial change in Americans: 1850 1975. J Forensic Sci 46:784787. Jantz RL, and Meadows Jantz L. 2000. Secular change in craniofacial morphology. Am J Hum Biol 12:327 338. Jekanowski MD, and Binkley JK. 2000. Food spending varies across the United States. Food Rev 23:3843. Kaifu Y. 1997. Changes in mandibular morphology from the Jomon to Modern periods in Eastern Japan. Am J Phys Anthropol 104:227 243. Kiliaridis S. 1995. Masticatory muscle influence on craniofacial growth. Acta Odontol Scand 53:196 202. Koyabu DB, and Endo H. 200 9. Craniofacial variation and dietary adaptations of African colobines. J Hum Evol 56:525536. Krakauer JC, McKenna MJ, Buderer NF, Rao DS, Whitehouse FW, and Parfitt AM. 1995. Bone loss and bone turnover in diabetes. Diabetes 44:775782. Larkin J. 1988. The Reshaping of Everyday Life 17901840. New York, NY: Harper & Row, Publishers, Inc. Larsen CS. 1995. Biological changes in human populations with agriculture. Annu Rev Anthropol 24:185 213. Lavelle CLB, Flinn RM, and Lunt DA. 1971. Analysis of the changes in maxillary dental arch dimensions between AngloSaxon and modern times. J Dent Res 50:409413. Lawther WL. 1956. A roentgenographic study of the temporomandibular joint using a special head positioner. The Angle Orthod 26:2233. LeBlanc SA, and Black B. 1974. A long term trend in tooth size in the Eastern Mediterranean. Am J Phys Anthropol 41:417 422.
237 Lee WTK, Leung SSF, Wang S, Xu Y, Zeng W, Lau J, Oppenheimer SJ, and Cheng JCY. 1994. Doubleblind, controlled calcium supplementation and bone mineral accretion in children accustomed to a low calcium diet. Am J Clin Nutr 60:744750. Leonard MB, Shults J, Wilson BA, Tershakovec AM, and Zemel BS. 2004. Obesity during childhood and adolescence augments bone mass and bone dimensions . Am J Clin Nutr 80:514 523. Levenstein H A . 2003. Paradox of Plenty: A Social History of Eating in Modern America. Berkeley, CA: University of California Press. Levenstein HA. 1988. Revolution at the Table: the Transformation of the American Diet. New Yo rk, NY: Oxford University Press. Levenstein HA. 1996. The politics of nutrition in N orth America. Neurosci Biobehav R 20:7578. Lewis ME. 2002. Impact of industrialization: comparative study of child health in four sites from medieval and postmedieval England (A.D. 850 1859). Am J Phys Anthropol 119:211223. Liang CT, Barnes J, Seedor JG, Quartuccio HA, Bolander M, Jeffrey JJ, and Rodan GA. 1992. Impaired bone activity in aged rats: alterations at the cellular and molecular levels. Bone 13:435441. Litt le BB, Buschang PH, Pea Reyes ME, Kheng Tan S, and Malina RM. 2006. Craniofacial dimensions in chil dren in rural Oaxaca, s outhern Mexico: secular change, 19682000. Am J Phys Anthropol 131:127 136. Luke DA, Tonge CH, and Reid DJ. 1979. Metrical analysis of growth changes in the jaws and teeth of normal, protein deficient, and calorie deficient pigs. . J Anat 129:449457. Mann RW, and Hunt DR. 2005. Photographic Regional Atlas of Bone Disease. Spring field, IL: Charles C. Thomas. Martin DC, and Danforth M. 2009. An analysis of secular change in the human mandible over the last century. Am J Hum Biol 21:704 706. Martin RB, Burr DB, and Sharkey NA. 1998. Skeletal Tissue Mechanics. New York , NY : Springer Verlag. Martin SA, Guatelli Steinberg D, Sciulli PW, and Walker PL. 2008. Brief communication: comparison of methods for estimating chronological age at linear enamel formation on anterior dentition. Am J Phys Anthropol 135:362365.
238 Martinez Abadias N, Esparza M, Sjovold T, Gonzalez Jose R, Santos M, and Hernandez M. 2009. Heritability of human cranial dimensions: comparing the evolvability of different cranial regions. J Anat 214:1935. Mavropoulos A, Kiliaridis S, Bresin A, and Ammann P. 2004. Effect of different masticatory functional and mechanical demands on the structural adaptation of the mandibular alveolar bone in young growing rats. Bone 35:191197. May B, Saha S, and Saltzman M. 2001. A threedimensional methematical model of temporoma ndibular joint loading. Clin Biomech 16:489 495. McGraw WS, Vick AE, and Daegling DJ. 2011. Sex and age differences in the diet and ingestive behaviors of sooty mangabeys ( Cercocebus atys ) in the Tai Forest, Ivory Coast. Am J Phys Anthropol 144:140153. McGraw WS, and Zuberbuhler K. 2007. The monkeys of the Ta forest: an introduction. In: McGraw WS, Zuberbuhler K, and Noe R, editors. Monkeys of the Ta Forest: An African Primate Community. New York , NY : Cambridge University Press. p 148. Meadows Jantz L, and Jantz RL. 1999. Secular change in long bone length and proportion in the United States, 18001970. Am J Phys Anthropol 110:57 67. Menegaz RA, Sublett SV, Figueroa SD, Hoffman TJ, Ravosa MJ, and Aldridge K. 2010. Evidence for the influence of diet o n cranial form and robusticity. Anat Rec 293:630641. Miles AEW. 2001. The Miles method of assessing age from tooth wear revisited. J Archaeol Sci 28:973 982. Millen AE, Dodd KW, and Subar AF. 2004. Use of vitamin, mineral, nonvitamin, and nonmineral sup plements in the United States: the 1987, 1992, and 2000 National Health Interview Survey results. J Am Diet Assoc 104:942950. Miller DW. 1987. Technology and the ideal: production quality and kitchen reform in nineteenthcentury America. In: Grover K, editor. Dining in America: 18501900. Amherst, MA: The University of Massachusetts Press. p 4784. Mokdad AH, Ford ES, Bowman BA, Dietz WH, Vinicor F, Bales VS, and Marks JS. 2003. Prevalence of obesity, diabetes, and obesity related health risk factors, 2001. J Am Med Assoc 289:7679. Mongini F. 1972. Remodelling of the mandibular condyle in the adult and its relationship to the condition of the dental arches. Acta Anat 82:437 453.
239 Moore Jansen PH. 1989. A multivariate craniometric analysis of secular change and variation among recent North American populations . Dissertation, University of Tennessee, Knoxville. Moore Jansen PH, Ousley SD, and Jantz RL. 1994. Data Collection Procedures for Forensic Skeletal Material, 3rd Ed. Knoxville, TN: University of Tennessee. Moss ML. 1962. The functional matrix. In: Kraus BS, and Reidel R, editors. Vistas in Orthodontics. Philadelphia, PA: Lea & Febiger. p 8598. Moss ML. 1973. A functional cranial analysis of primate craniofacial growth. Symposium of the IVth In ternational Congress of Primatology Vol 3: Craniofacial Biology of Primates. Basel: Karger. p 191208. Moss ML, and Salentijn L. 1969a. The capsular matrix. Amer J Orthodontics 56:474 490. Moss ML, and Salentijn L. 1969b. The primary role of functional m atrices in facial growth. Amer J Orthodontics 55:566 577. Moss ML, and Young RW. 1960. A functional approach to craniology. Am J Phys Anthropol 18:281 292. Neeley WW, and Gonzalez DA. 2007. Obesity in adolescence: implications in orthodontic treatment. Am J Orthod Dentofac 131:581588. Nelson JH. 1985. Wheat: its processing and utilization. The Am J Clin Nutr 41:10701076. Nicholson EK, Stock SR, Hamrick MW, and Ravosa MJ. 2006. Biomineralization and adaptive plasticity of the temporomandibular joint i n myostatin knockout mice. Arch Oral Biol 51:3749. O'Neill TW. 2005. Looking back: developments in our understanding of the occurrence, aetiology and prognosis of osteoporosis over the last 50 years. Rheumatology 44:iv33iv35. Pampel FC. 2004. Tobacco I ndustry and Smoking. New York, NY: Facts on File, Inc. Parker J. 2005. Chewing tobacco. In: Goodman J, editor. Tobacco in History and Culture: An Encyclopedia. Detroit, MI: Charles Scribner's Sons. p 125129. Paschetta C, de Azevedo S, Castillo L, Martinez Abadias N, Hernandez M, Lieberman DE, and Gonzalez Jose R. 2010. The influence of masticatory loading on craniofacial morphology: a test case across technological transitions in the Ohio Valley. Am J Phys Ant hropol 141:297314.
240 Picq PG, Plavcan JM, and Hylander WL. 1987. Nonlever action of the mandible: the return of the hydra. Am J Phys Anthropol 74:305307. Plavcan JM. 2002. Taxonomic variation in the patterns of craniofacial dimorphism in primates. J Hum Evol 42:579608. Pollitzer WS, and Anderson JJB. 1989. Ethnic and genetic differences in bone mass: a review with a hereditary vs environmental perspective. The Am J Clin Nutr 50:12441259. Popkin BM. 2011. Is the obesity epidemic a national security iss ue around the globe? Curr Opin Endocinol Diabetes 18:328 331. Powell ML. 1988. Status and Health in Prehistory: A Case Study of the Moundville Chiefdom. Washington D.C.: Smithsonian Institution Press. Pucciarelli HM. 1981. Growth of the functional components of the rat skull and its alteration by nutritional effects: a multivariate analysis. Am J Phys Anthropol 56:3341. Pucciarelli HM, Dressino V, and Niveiro MH. 1990. Changes in skull components of the squirrel monkey evoked by growth and nutrition: an experimental study. Am J Phys Anthropol 81:535543. Pucciarelli HM, and Oyenhart EE. 1987. Effects of maternal food restriction during lactation on craniofacial growth in weanling rats. Am J Phys Anthropol 72:6775. Putnam J. 2000. Major trends in the U .S. food supply, 19091990. Food Rev 23:815. Rando C, Hillson S, and Antoine D. 2014. Changes in mandibular dimensions during the mediaeval to post mediaeval transition in London: A possible response to decreased masticatory load. Arch Oral Biol 59:7381 . Rando C, and Waldron T. 2012. TMJ osteoarthritis: a new approach to diagnosis. Am J Phys Anthropol 148:45 53. Rasband WS. 19972011. ImageJ, U.S. National Institutes of Health. Bethesda, MD, USA, http://imagej.nih.gov/ij/ . Ravosa MJ, Kunwar R, Stock S R, and Stack MS. 2007. Pushing the limit: masticatory stress and adaptive plasticity in mammalian craniomandibular joints. J Exp Biol 210:628641.
241 Ravosa MJ, Lopez EK, Menegaz RA, Stock SR, Stack MS, and Hamrick MW. 2008. Adaptive plasticity in the mammalian masticatory complex: you are what, and how, you eat. In: Vinyard CJ, Ravosa MJ, and Wall CE, editors. Primate Craniofacial Function and Biology. New York, NY: Springer. p 293328. Riggs BL, Melton III JM, Robb RA, Camp JJ, Atkinson EJ, Peterson JM, Rouleau PA, McCollough CH, Bouxsein ML, and Khosla S. 2004. Populationbased study of age and sex differences in bone volumetric density, size, geometry, and structure at different skeletal sites. J Bone Miner Res 19:19451954. Robert JC. 1949. The St ory of Tobacco in America. New York, NY: Alfred A. Knopf. Robinson DJ. 2004. Marketing gum, making meanings: Wrigley in North America, 18901930. Enterp Soc 5:4 44. Robinson M. 1946. The temporomandibular joint: theory of reflex controlled nonlever acti on of the mandible. J Am Dent Assoc 33:12601271. Rocher E, Chappard C, Jaffre C, Benhamou C, and Courteix D. 2008. Bone mineral density in prepubertal obese and control children: relation to body weight, lean mass, and fat mass. J Bone Miner Metab 26:7378. Rogozinski J. 1990. Smokeless Tobacco in the Western World: 15501950. New York, NY: Praeger. Rubin C, Turner AS, Bain S, Mallinckrodt C, and McLeod K. 2001. Low mechanical signals strengthen long bones. Nature 412:603604. Rubin C, Turner AS , Mallinckrodt C, Jerome C, K M, and Bain S. 2002. Mechanical strain, induced noninvasively in the highfrequency domain, is anabolic to cancellous bone, but not cortical bone. Bone 30:445452. Sardi ML, Novellino PS, and Pucciarelli HM. 2006. Craniofacial morphology in the Argentine Center West: consequences of the transition to food production. Am J Phys Anthropol 130:333343. Scott JE, McAbee KR, Eastman MM, and Ravosa MJ. 2014. Experimental perspective on fallback foods and dietary adaptations in earl y hominins. Biol Letters 10:20130789. Sheridan SG, Mittler DM, van Gerven DP, and Covert HH. 1991. Biomechanical association of dental and temporomandibular joint pathology in a Medieval Nubian population. Am J Phys Anthropol 85:201205. Smith BH. 1984. Patterns of molar wear in hunter gatherers and agriculturalists. Am J Phys Anthropol 63:39 56.
242 Smith BH, Garn SM, and Hunter WS. 1986. Secular trends in face size. Angle Orthod 56:196204. Smith RJ. 1978. Mandibular biomechanics and temporomandibular joint function in primates. Am J Phys Anthropol 49:341 350. Stankovic S, Vlajkovic S, Boskovic M, Radenkovic G, Antic V, and Jevremovic D. 2013. Morphological and biomechanical features of the temporomandibular joint disc: an overview of recent findings. Arch Oral Biol 58:14751482. Storck J, and Teague WD. 1952. Flour for Man's Bread. Minneapolis, MN: University of Minnesota Press. Sun Z, Lee E, and Herring SW. 2004. Cranial sutures and bones : growth and fusion in relation to masticatory strain. Anat Rec Part A 276A:150161. Tanaka E, Sano R, Kawai N, Langenbach GEJ, Brugman P, Tanne K, and Van Eijden TMGJ. 2007. Effect of food consistency on the degree of mineralization in the rat mandible. Ann Biomed Eng 35:16171621. Tanaka E, and van Eijden T. 2003. Biomechanical behavior of the temporomandibular joint disc. Crit Rev Oral Biol M 14:138 150. Tattersall I. 1973. Cranial anatomy of Archaeolemurinae (Lemuroidea Primates). Anthropol Pap Am Mu s 52:1110. Taylor AB. 2005. A comparative analysis of temporomandibular joint morphology in the African apes. J Hum Evol 48:555574. Taylor AB. 2006. Feeding behavior, diet, and the functional consequences of jaw form in orangutans, with implications for the evolution of Pongo . J Hum Evol 50:377393. Taylor AB, Jones KE, Kunwar R, and Ravosa MJ. 2006. Dietary consistency and plasticity of masseter fiber architecture in postweaning rabbits. Anat Rec Part A 288A:11051111. Taylor AB, and Vinyard CJ. 2008 . The relationship between jaw muscle architecture and feeding behavior in primates: treegouging and nongouging gummivorous callitrichids as a natural experiment. In: Vinyard CJ, Ravosa MJ, and Wall CE, editors. Primate Craniofacial Function and Biology. New York, NY: Springer. p 241262. Taylor RMS. 1986. Nonlever action of the mandible. Am J Phys Anthropol 70:417421.
243 Terhune CE. 2011a. Dietary correlates of temporomandibular joint morphology in New World primates. J Hum Evol 61:583596. Terhune CE. 2011b. Modeling the biomechanics of articular eminence function in anthropoid primates. J Anat 219:551564. Terhune CE. 2013. Dietary correlates of temporomandibular joint morphology in the great apes. Am J Phys Anthropol 150:260272. Throckm orton GS. 1985. Quantitative calculations of temporomandibular joint reaction forces II. The importance of the direction of the jaw muscle forces. J Biomech 18:453461. Throckmorton GS, and Dechow PC. 1994. In vitro strain measurements in the condylar process of the human mandible. Arch Oral Biol 39:853867. Trotter M, Broman GE, and Peterson RR. 1959. Density of cervical vertebrae and comparison with densities of other bones. Am J Phys Anthropol 1959:1925. Trotte r M, and Gleser GC. 1951. Trends in stature of American whites and Negroes born between 1840 and 1824. Am J Phys Anthropol 9:427440. Turner CH. 1998. Three rules for bone adaptation to mechanical stimuli. Bone 23:399407. Ulrich CM, Georgiou CC, Snow Ha rter CM, and Gillis DE. 1996. Bone mineral density in mother daughter pairs: relations to lifetime exercise, lifetime milk consumption, and calcium supplements. Am J Clin Nutr 63:7279. Viet HZ. 2013. Modern Food, Moral Food: Self Control, Science, and the Rise of Modern American Eating in the Early Twentieth Century. Chapel Hill, NC: The University of North Carolina Press. von CramonTaubadel N. 2011. Global human mandibular variation reflects differencesin agricultural and hunter gatherer subsistence st rategies. P N atl A cad S ci USA 108:1954619551. Wall CE. 1995. Form and Function of the Temporomandibular Joint in Anthropoid Primates. Ann Arbor, MI: State University of New York at Stony Brook. Wall CE. 1999. A model of temporomandibular joint function in anthropoid primates based on condylar movements during mastication. Am J Phys Anthropol 109:6788. Watson WG. 1982. The functional matrix hypothesis revisited. Amer J Orthodontics 81:7173.
244 Weijs WA, and Hillen B. 1984. Relationships between masticatory muscle crosssection and skull shape. J Dent Res 63:11541157. Wescott DJ, and Jantz RL. 2005. Assessing craniofacial secular change in American blacks and whites using geometric morphometry. In: Slice DE, editor. Modern Morphometrics in Physical Anthropology. New York: Kluwer Academic/Plenum Publishers. p 231245. Whittaker DK, Jones JW, Edwards PW, and Molleson T. 1990. Studies on the temporomandibular joints of an eighteenthcentury London population (Spitalfields). J Oral Rehabil 17:8997. Wi lson GH. 1921. The anatomy and physics of the temporomandibular joint. J Natl Dent Assoc 8:236241. Yamada K, and Kimmel DB. 1991. The effect of dietary consistency on bone mass and turnover in the growing rat mandible. Arch Oral Biol 36:129 138.
245 BIOGRAPHICAL SKETCH Katherine Skorpinski graduated with high honor from Michigan State University in 2006 with a Bachelor of Science in anthropology. During her undergraduate work, she became involved in bioarchaeological research involving aging a skeletal population from Albania using tooth wear and histomorphological analysis. She also spent the summers of 2005 and 2006 as an intern at the Field Museum of Natural History in Chicago, where she worked with both skeletal remains and material artifacts fr om Kish, an ancient Mesopotamian city that was occupied between 3200 B.C. and the 7th century A .D. Ms. Skorpinski earned her m asterâ€™s degree at the University of Florida in August 2009 with research examining secular trends in the mandibles of popul ations in the United States. Since starting graduate school at the University of Florida in August 2006, Ms. Skorpinski served as a Graduate Analyst in the C.A. Pound Human Identification Laboratory conducting forensic anthropological analyses, assisting with search and recovery efforts for hum an remains, and developing Standard Operating Procedures for the laboratory. She served as Laboratory Coordinator for the CAPHIL from February 2011 to August 2012, and subsequently as Quality Assurance Coordinator until she earned her doctorate in August 2014. During her time at the University of Florida, Ms. Skorpinski taught several semesters of a laboratory based course in human osteology and assisted with forensic anthropology workshops for law enforcement and medical examiner personnel.