Basicranial flexion relative brain size and facial kyphosis in Homo sapiens and some fossil hominids.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 98:575-593 (1995) Basicranial Flexion, Relative Brain Size, and Facial Kyphosis in Homo Sapiens and Some Fossil Hominids CALLUM ROSS AND MACIEJ HENNEBERG Biological Anthropology Research Programme, Department of Anatomy and Human Biology, University of the Witwatersrand, Parktown, Johannesburg, Republic of South Africa KEY WORDS Fossil hominids, Basicranial flexion, Computed tomography, Brain evolution ABSTRACT Comparative work among nonhominid primates has demonstrated t h a t the basicranium becomes more flexed with increasing brain size relative to basicranial length and a s the upper and lower face become more ventrally deflected (Ross and Ravosa [19931Am. J. Phys. Anthropol. 91:305324). In order to determine whether modern humans and fossil hominids follow these trends, the cranial base angle (measure of basicranial flexion), angle of facial kyphosis, and angle of orbital axis orientation were measured from computed tomography (CT) scans of fossil hominids (Sts 5, MLD 37/38, OH9, Kabwe) and lateral radiographs of 99 extant humans. Brain size relative to basicranial length was calculated from measures of neurocranial volume and basicranial length taken from original skulls, radiographs, CT scans, and the literature. Results of bivariate correlation analyses revealed that among modern humans basicranial flexion and brain sizehasicranial length are not significantly correlated, nor are the angles of orbital axis orientation and facial kyphosis. However, basicranial flexion and orbit orientation are significantly positively correlated among the humans sampled, as are basicranial flexion and the angle of facial kyphosis. Relative to the comparative sample from Ross and Ravosa (1993), all hominids have more flexed basicrania than other primates: Archaic Homo sapiens, Homo erectus, and Australopithecus africanus do not differ significantly from Modern Homo sapiens in their degree of basicranial flexion, although they differ widely in their relative brain size. Comparison of the hominid values with those predicted by the nonhominid reduced major-axis equations reveal that, for their brain sizehasicranial length, Archaic and Modern Homo sapiens have less flexed basicrania than predicted. H. erectus and A. africanus have the degree of basicranial flexion predicted by the nonhominid reduced major-axis equation. Modern humans have more ventrally deflected orbits than all other primates and, for their degree of basicranial flexion, have more ventrally deflected orbits than predicted by the regression equations for hominoids. All hominoids have more ventrally deflected orbital axes relative to their palate orientation than other primates. It is argued that hominids do not strictly obey the trend for basicranial flexion to increase with increasing relative brain size because of con- Received October 28, 1994; accepted August 15, 1995. Address reprint requests to Dr. Callum Ross, Dept. Anatomical Sciences, Health Sciences Center, SUNY Stony Brook, Stony Brook. NY 11794-8081. 0 1995 WILEY-LISS. INC. 576 C. ROSS AND M.HENNEBERG straints on the amount of flexion that do not allow it to decrease much below 90". Therefore, if basicranial flexion is a mechanism for accommodating a n expanding brain among non-hominid primates, other mechanisms must be at work among hominids. o 1995 Wiley-Liss, Inc. Homo sapiens have long been known to have highly flexed basicrania. Extreme basicranial flexion in humans has been claimed to provide better balance of the head on the vertebralcolumn(Wood Jones, 1917;Weidenreich, 1924, 1941; DuBrul, 19501, to create a pharynx shape suitable for the generation of vowel sounds used in human speech (Laitman, 1985),to reduce stresses in the anterior part of the cranial base due to loading of the jaw joint or occipital condyles (Demes, 1985), or to be the result of increases in relative brain size and decreases in the relative size of the masticatory apparatus (Biegert, 1963; see review in Ross and Ravosa, 1993). Basicranial flexion has also been invoked a s evidence in support of various hypotheses regarding the taxonomic relationships among hominoids (Dean and Wood, 1981, 1982, 1984; Shea, 1985). Elucidation of the functional and structural correlates of basicranial flexion is therefore of importance for the understanding of hominoid cranio-facial evolution. Previous comparative work on basicranial flexion in non-hominid primates found no correlation between the degree of basicranial flexion and habitual orthograde posture but confirmed Gould's (1977) suggestion that increases in brain size relative to basicranial length are correlated with increased basicranial flexion (as indicated by decreases in the cranial base angle [CBA]) (Ross and Ravosa, 1993). In addition, positive correlations were found between measures of basicranial flexion and the orientation ofthe orbital axes and palate, indicating that a s the basicranium becomes more flexed, the face becomes more ventrally deflected (i.e., more kyphotic or klinorhynch). The present study was carried out to determine whether the degrees of basicranial flexion and palatal and orbital kyphosis in hominids follow the trends documented by Ross and Ravosa (1993) for non-hominid primates. Specifically, we sought to test the fol- lowing null hypotheses: a ) that hominids have the degree of basicranial flexion expected for primates with their brain size relative to basicranial length, and b) that hominids have the orbit and palate orientation expected for primates with their degree of flexion. If hominids follow the trends of association exhibited by nonhominid primates, then the same causes of flexion may be postulated; if all or some of the hominids deviate from these trends, then different mechanisms may be invoked. MATERIALS AND METHODS Sample The sample of Homo sapiens used in this study consisted of 99 adult skulls from the Raymond A. Dart Collection of Human Skeletons in the Department of Anatomy and Human Biology at the University of the Witwatersrand. The specimens were selected to represent both sexes, a wide size-range and diverse ethnic origins. An attempt was made to measure five males and five females from all the major ethnic groups sampled in the Dart Collection, with the remaining skulls chosen to sample specimens of unusual origins (see Table 1).Lateral radiographs of the skulls were made in the Department of Radiology, Johannesburg General Hospital. The measurement techniques employed here (see below) require fairly complete and undistorted basicrania, a rare situation in fossil hominids. Moreover, of those hominids with complete basicrania, only a small proportion have been scanned by computed tomography, the East African hominids being the most important exceptions. As a result, the sample of fossil hominids available for this study was restricted to Sts 5, MLD 37/ 38 (Australopithecus africanus),OH9 (Homo erectus), and the Kabwe Skull (archaic Homo sapiens). Measurements of Sts 5 and MLD 37/38 were taken from computed tomography (CT) 577 HOMINID BASICRANIAL FLEXION TABLE 1. Composition of human sample (ethnicity and sex as recorded in Dart Collection records)‘ Ethnicity Zulu Xhosa Venda Indian Tswana Malawi Dama Fingoe Amazon Sotho Ndebele European Chinese Griqua Bushman Kalanga Totals Unknown sex Specimen Males 5 5 Females Total 6 5 11 10 3 8 10 5 1 3 4 4 5 5 5 1 1 1 5 5 1 1 1 5 5 8 10 18 TABLE 2. Computed tomography scans used to take measurements 6 1 3 46 35 1 10 10 8 10 9 1 99 ‘Zulu = Zulu-speaking; Xhosa = Xhosa-speaking; Venda = Vendaspeaking; Indian = individuals classified by South African apartheid laws as Indian (probably related to peoples originating in Indian suhcontinent);Tswana = Tswana-speakingpeoples;Malawi = Malawians; Dama = from Damaraland in northern Namibia; Fingoe = South African tribe; Amazon = Amazonian Indian; Sotho = Sotho-speaking peoples; Ndebele = Ndebele-speaking peoples; European = individuals classified as “White” by South African apartheid laws; Chinese = immigrants from China to South Africa; Griqua = individuals from old frontier community mostly of Khoi-San descent with European admixture; Kalanga = tribe from Botswana, Zimbabwe, and northern South Africa. Sts 5 MLD 37/38 Kabwe skull OH9 Tape location Scan numbers Transvaal Museum Pretoria, SA Transvaal Museum Pretoria, SA Philips HQ Best, Netherlands Philips HQ Best, Netherlands 41 (mid-sagittal) 32 & 51 (orbit) 5 (mid-sagittal) All scansL 201 (mid-sagittal) ‘Measurements of mid-sagittal variables were taken from the midsagittal section of a three-dimensional reconstruction of t h e Kabwe Skull. (i.e., older) measures. As noted previously, the advantage of Ross and Ravosa’s measure of the cranial base angle is that the points used to define the orientation of the occipital clivus both lie on the clivus, and those used to define the orientation of the planum sphenoideum both lie on the planum sphenoideum. Thus, our cranial base angle measures only the relative orientation of the endocranial surfaces of these two bones, without the confounding effects introduced by using points such a s nasion or foramen cecum which lie off these bones. Angles. The cranial base angle (CBA) is a n attempt to quantify the relative orientations of the endocranial surfaces of the clivus ossis occipitalis (CO) and the planum sphenoideum (PS)(Fig. 1).The orientation of the endocranial surface of CO is represented by a line from basion (B in Fig. 1)to the point on clivus where dorsum sellae begins to curve away superiorly (D). This latter point is a n attempt to estimate the position of the endocranial edge of the spheno-occipital synchrondrosis, which defines the anterior edge of the occipital clivus. The orientation of PS is represented by a line from the apex of the declivity above the sulcus chiasmatis (P) to the apex of the sloping posterior surface of the pit in which the cribriform plate is set Measures (A). The angle between CO and PS is the The measures taken in this study are CBA. those described by Ross and Ravosa (1993). The angle of facial kyphosis (AFK) meaAlthough a plethora of measures has been sures the Orientation of the palate to CO (Fig. used in the study of basicranial flexion, we 1).The plane of the palate is represented by believe these new measures to be more ap- a line between anterior (AN) and posterior propriate for determining the relative orien- (PN) nasal spines. This line is extended postations of the anterior and posterior portions teriorly to reach the plane of clivus: the angle of the basicranium than more “traditional” between these lines is the AFK. Identificascans (1.5mm slice thickness) of these fossils stored on tapes at the Transvaal Museum in Pretoria. They were displayed on the console of the CT scanner in the Department of Radiology, Johannesburg General Hospital, where measurements were taken. Hard copies of the scans were also made to enable comparison with the original fossils. CT scans (1.5 mm slice thickness) of the Kabwe (Broken Hill) Skull and OH9, in the care of the Foundation for Hominid Palaeoradiology, were measured on a console a t Philips Headquarters, Best, and University Hospital, Utrecht, Holland. Table 2 lists the scan slices used for measurement of fossil hominids. 578 C. ROSS AND M. HENNEBERG Fig. 1. Diagram illustrating points and planes used to measure cranial base angle (LBA) and angle of facial kyphosis (AFK). B, basion; D, point where dorsum sellae curves away; P, top of declivity above sulcus chiasmatis; A, top of slope above cribriform plate; PN, posterior nasal spine; AN, anterior nasal spine. tion of the posterior nasal spine on the radiographs was facilitated by comparisons with the skulls themselves. These comparisons with the original skulls were performed because in the replications performed to evaluate measurement error, these comparisons were not made and the measurement error of the AF'K was higher than that for the other measures (see below, Table 2). The angle of orbital axis orientation (AOA) was measured according to the method defined by Ravosa (1988). The orbital axis is drawn in the following way (see Fig. 2). On lateral radiographs, the distance (X) from the center of the optic canal to the anterior point of overlap of the orbital roof and the contour of the anterior cranial fossa (A) was measured. The point on the inferior orbital border a t distance (X) was then marked (B) and a line drawn between point B and point A. The line AB defines the plane of the orbital aperture. The line bisecting this plane and passing through the center of the optic canal is the orbital axis (OA). The angle between OA and CO was measured as AOA (Fig. 2). Linear measures. Linear measures of endocranial basicranial length (BL) were taken Fig. 2. Diagram illustrating points and planes used to measure angle of orbital axis orientation (AOA). AOA is the angle between orbital axis (OA) and clivus (CO). A is point of overlap of the orbital roof and the contour of the anterior cranial fossa. X is distance from optic canal to A. B is point on inferior orbital margin at distance X from optic canal. HOMINID BASICRANIAL FLEXION 579 Fig. 3. Photo of midsagittal CT scan of Sts 5 illustrating positions of planum sphenoideum (PS) and clivus (CO). The angle between these planes is the CBA. from the radiographs of the humans (Fig. l ) , from the mid-sagittal scans of MLD 37/38 and the Kabwe Skull, from the original of Sts 5 (Fig. 31, from a cast of OH9 in the Palaeoanthropology Research Unit at the University of the Witwatersrand, and from Maier and Nkini’s (1984) reconstruction of OH9. (In the latter instance the scale of the drawing [Maier and Nkini, 1984: Fig. 11 was assessed by measuring the width of the foramen magnum on the cast and on the drawing.) BL is estimated by measuring three segments: a) basion to pituitary point (Zuckerman, 19551, b) pituitary point to the posterior point on PS (as defined above), and c) posterior to anterior point of PS. For each individual, a, b, and c were summed. The linear measures taken from each radiograph were then corrected by multiplying them by the ratio prosthion-basion length measured on the radiopraDh prosthion-basion length measured on the skull. Measurements from scans of Sts 5, MLD 37/38,0H9, and the Kabwe Skull. Angu- lar and linear measurements of the basicrania of Sts 5, MLD 37/38, OH9, and the Kabwe Skull were taken from CT images. Sts 5. Hard copies of Images 27-55 were made and the mid-sagittal scan identified as Image 41 (Fig. 3) by comparison with the original specimen. To control for slight distortion in the specimen, the mid-sagittal measurements (CBA and AFK) were also made on Images 39,40, and 42 for comparison with those taken on Image 41 and found to differ by, a t most, one degree. The CBA was measured as shown in Figure 3. Figure 4A illustrates the positions of CO and PS on the original specimen: Figure 4B illustrates the position of the anterior edge of PS in more detail (APS). The AOA used here was defined by Ravosa (1988) for measurement from lateral radiographs and uses points in parasagittal planes projected onto the radiograph. Because measurements ofAOA in sts 5 were by necessity taken from parasagittal CT slices passing through the optic foramina, some modification of this technique was necessary. This was done by measuring the orientation of CO to the horizontal on Image 41 and the orientation of the orbital axes to the hori- 580 C. ROSS AND M. HENNEBERG A B Fig. 4. Stereophotos of interior of Sts 5 calvaria illustrating A) position of CO and PS, and B) position of anterior point on planum sphenoideum ( A P S ) . I zontal on Images 32 (left) and 51 (right), enabling the angles between the orbital axes and CO to be calculated trigonometrically. The optic foramen on the right side of Sts 5 is not preserved, so it was necessary to estimate its position using the optic foramen on the left side. This was done by positioning the cursor in the center of the optic foramen in Image 32 then bringing up Image 5 1 without moving the cursor. Although this technique does not take into account some minor distortion in the fossil, the results are deemed acceptable because the measurements obtained from the two sides of the skull are comparable (Table 5). The mean of these two values (144.8') was used. MLD 37138. In this specimen, the anterior portion of planum sphenoideum and most of the face are missing. Consequently, it was not possible to measure AOA, AFK, or BL. Nor was it possible to measure the CBA using the anteriormost point on PS a s defined above, a s the pit in which the cribriform plate sits is missing. However, because a significant portion of PS is preserved, it was possible to estimate the CBA, yielding a value not dissimilar to that obtained for Sts 5 (Table 5). An estimate of BL in MLD 37/38 was also attempted. The distances from basion to pituitary point, and from pituitary point to the posterior edge of PS were measured from the CT scan. The values for these measurements are similar to those for Sts 5, so the length of PS in Sts 5 was used a s the estimate of PS length used to calculate BL in MLD 37/ 38 (Table 5). OH9. The only skull of Homo erectus for which scans were available is OH9, the midline basicranium of which is badly damaged, missing much of the dorsal surface of clivus and dorsum sellae. Maier and Nkini (1984) have attempted a reconstruction of the basicranium (redrawn in Fig. 5). The position of basion can be reconstructed from the outline of the foramen magnum a s only several millimeters of bone appear to be missing. The position of tuberculum sellae on the left side has allowed Maier and Nkini to approximate the position of dorsum sellae; however, it is possible that dorsum sellae may have been lower than they suggest. The preserved HOMINID BASICRANIAL FLEXION 581 Fig. 5. Slightly parasagittal scan of OH9 with reconstruction of clivus by Maier and Nkini (1984). Redrawn from Maier and Nkini (1984). CO’A’ and CO’B’ represent our estimates of the likely upper and lower limits possible for the position of CO. MN is the position of CO as measured from Maier and Nkini’s reconstruction. portions of tuberculum sellae and lamina cribrosa can be used to reconstruct planum sphenoideum, although with limited confidence. One estimate of CBA (MN in Fig. 5) was taken from Maier and Nkini’s (1984) reconstruction, one was taken from a slightly modified version of this reconstruction, with the dorsum sellae slightly lower (CO’B’ in Fig. 5) and one was taken by rough estimate from the mid-sagittal scan stored at Philips Headquarters in Best (CO’A’ in Fig. 5). This latter reconstruction assumes that basion was further forward than Maier and Nkini reconstruct it. The three differing results, listed in Table 5, are presented with some circumspection. We have not examined the original skull itself, and would not be surprised if our estimates were significantly in error. Consequently, OH9 is given only cursory treatment in this study. Kabwe skull. Measurements of the Kabwe Skull were taken from the mid-sagittal section (CBA, AFK, BL) or parasagittal section (AOA) of a three-dimensional reconstruction of CT slices stored at Philips. All points on the midline basicranium used for measurements in this study are preserved. The method used for measuring AOA in Sts 5 was applied to the left side of the Kabwe Skull with the assistance of Prof. Zonneveld. Volumes. Estimates of overall brain size in the human sample were made by measuring the endocranial volume of the skulls with mustard seed. After plugging up foramina and holes due to breakage, the neurocranium was filled with mustard seed through the foramen magnum while gently shaking the skull from side to side. The skull was then tapped with a finger and more seed added. This procedure was repeated until no more mustard seed would fit into the braincase. The mustard seed was then poured into a graduated cylinder which was shaken until the column of seed had a flat surface a t the top. The volume was then read off the cylinder. The volumes of the neurocrania of OH9, Sts 5, and MLD 37/38 were taken from Holloway (1973a,b). Neurocranial volume in the Kabwe Skull was obtained from Pycraft (1928) rather than Holloway (1981) because Zonneveld’s (pers. comm.) estimate of neurocranial volume in Broken Hill using 3-dimensional reconstruction of the CT slices of the skull is identical to that obtained by Pycraft . Statistical analyses Reliability of the radiographic methods and the techniques for taking measurements 582 C. ROSS AND M. HENNEBERG TABLE 3. Measurement error: Results o f t h e analysis of variance i n measurements of CBA, AOA, and AFK repeated fiue times on fiue human skulls (n = 25, k = 51 CBA AOA AFK Mean Total variance Error variance Erroritotal 109.04 118.78 123.88 69.14 37.98 35.91 4.02 2.23 4.98 0.0582 0.0587 0.1386 from the radiographs were evaluated by taking five lateral radiographs of each of five skulls and measuring the CBA, AFK, and AOA once on each radiograph. The radiographs were all taken on one day; one radiograph for each specimen was measured each day for five days. One-way analysis of variance (ANOVA)( P < 0.05) indicated that the variance of the replicate measurements taken on single individuals is significantly less than the inter-individual variance. Error variance calculated from the one-way ANOVA is shown in Table 3. Descriptive statistics for all measurements taken from Homo sapiens are presented in Table 4. Sample sizes for H. sapiens are less than 99 because it was not possible to take measurements from all skulls that were radiographed. Index of Relative Encephalization 1(IREl)--cube root of neurocranial volume divided by basicranial length (Ross and Ravosa, 1993)-was calculated for each individual measured. Descriptive statistics for IREl are also given in Table 4. The values obtained for the fossil hominids are presented in Table 5, as are the means for the same measurements taken from the great apes. In order to evaluate the null hypotheses, Student’s t-tests were used to determine whether the mean values for Homo sapiens and the individual variates for the fossil hominids are significantly different from the values predicted from the nonhominid primate reduced major-axis (RMA) regression equations. Formulae for the standard errors used to calculate 95% confidence limits of the predicted individual values and the sample means were taken from Sokal and Rohlf (1981). RMA equations were used for prediction because we have no reason to believe that our x-values are measured without error. More importantly, many of our predictions pertain to x-values lying well outside the range of values used to generate the equation, a situation in which RMA performs better than least-squares (Draper and Smith, 1981; Ricker, 1984). RMA slopes were compared using a computer program written by Tim Cole; y-intercepts were compared using the “quick test” of Tsutakawa and Hewett (1977). Table 6 lists the RMA equations for nonhominids, nonhominoids and hominoid primates, along with the standard errors of b and the y-intercept. Bivariate plots of the comparisons listed above were created and the RMA lines for non-hominid primates were added (Figs. 6-9). These lines enable the position of hominids relative to the general nonhominid primate trends to be evaluated. In Figures 7-9, the RMA for hominoids was also added to illustrate the divergent trends seen in hominoids. Note that the RMAs for nonhominid and nonhominoid primates are very similar (Table 6). Loess (Lowess) curve-fitting was applied to the data on CBA and I R E l in order to describe the nature of the change in CBA with increasing relative brain size. Being a nonparametric regression technique, Loess has the advantage of making fewer assumptions about the form of the relationship be- TAELE 4. Descriptiue statistics for measurements taken on sample of Homo sapiensl Homo sapiens n Mean S.D. Min. Max. Neurocranial volume (cc) CBA (deg.) AFK (deg.) AOA (deg.) Na-S-Ba (deg.) IRE 1 BL (mm) 92 93 93 93 83 89 94 1,351 111.8 122.2 114.6 134.7 1.65 67.6 148.1 7.43 6.52 6.40 6.09 0.172 6.45 1,050 92 103 102 116 1.36 45.23 1,835 135 140 133 149 2.25 79.3 ‘CBA = cranial base angle; AFK = angle of facial kyphosis; AOA = angle of orbital axis orientation; Na-S-Ba = nasion-sella-basion angle; IREl = index of relative encephalization 1; BL = basicranial length. 583 HOMINID BASICRANIAL FLEXION TABLE 5. Mean values for measurements on great apes (from Ross, 19931 and values for measurements on fossil specimens Neuro volume AFK AOA (deg.1 BL (mm) IRE 1 167.0 161.0 163.0 L = 144.2 R = 145.5 71.62 70.85 77.22 60.67 1.03 1.04 1.02 1.30 110.5 58.80 1.29 99.0 92.7 104.0 128.0 88.92' 88.92 88.92 81.60 1.15 1.15 1.15 1.33 (cc1 CBA (deg.) (deg.1 Pongo pygmaeus Pan troglodytes Gorilla gorilZa Sts 5 396.8 398.3 489.6 485.0 135.0 152.0 148.0 114.0 162.0 156.0 150.0 141.0 MLD 37/38 OH9' (M&N) (CO'B') (CO'A) Kabwe Skull 435.0 1,067.0 1,067.0 1,067.0 1.280.0 n 125.0 127.0 'M&N, estimate of CO onentation taken from Maier and Nkini's (1984) reconstruction; CO'A, estimate of upper limit possible for orientation of CO; CO'B', estimate of lower hmit possible for onentation of CO. Estimate of basicranial length based on Maier and Nkini (1984) reconstruction, TABLE 6. Statistics for RMA regression eauations for nonhominid. nonhominoid. and hominoid urimates b n CBA x IREl Nonhominid primates Nonhominoid primates Hominoids AOA x CBA Nonhominid primates Nonhominoid primates Hominoids AFK X CBA Nonhominid primates Non h omin oid primates Hominoids AOA x AFK Nonhominid primates Nonhominoid primates Hominoids 64 58 11 66 60 8 66 60 9 65 59 8 sb - 132.30 12.776 - 138.92 15.239 -120.83 29.596 277.62 10.868 304.81 12.758 216.51 33.074 1.085 0.1049 1.204 0.8522 0.923 0.2035 -14,037 17.3364 -35.268 18.5827 25.198 29.2802 1.045 0.1138 1.140 0.1130 0.786 0.1861 -23.201 18.7914 -40.924 18.7962 36.670 26.4301 1.026 0.0667 1.050 0.0673 1.419 0.1246 12.145 9.9655 9.402 9.9932 -56.631 18.8118 tween CBA and IREl than standard parametric techniques (Efron and Tibshirani, 1991). Despite the fact that our measurement techniques are not ideal for intraspecific studies (Ross and Ravosa, 19931, the large size of our human sample and the low errors r P -0.649 *** -0.571 *:i* -0.677 0.633 *** 0.708 **:!: 0.841 *** 0.490 *** 0.650 *** 0.779 0.856 *** 0.875 *** 0.976 +** associated with our measurements of the human sample (Table 2) suggest that it would be of interest to determine whether the nonhominid primate trends discussed in the Introduction hold within humans as a group. Consequently, RMA equations and Pearson correlations were computed for the following C. ROSS AND M. HENNEBERG 584 190 Hylohatids 180 0 Great Apes 170 + Homo sapiens 4 Australopithecus africanus 160 Kahwe 150 140 I "\ \u 130 120 110 I00 X ## 90 .8 .6 1 1.2 1.4 1.6 2 1.8 2.2 Index of relative encephalization 1 Fig. 6. Bivariate plot of the cranial base angle (CBA) and index of relative encephalization 1 (IRE11 in primates. The latter variable is calculated as the cube root of neurocranial volumehasicranial length. The reduced major-axis (RMA) regression line for nonhominid primates is shown. The polygon surrounding the human mean defines the range of values obtained over the human sample 190 8 '3 Q 1 170 Y 's8 - 150 - Y 2o a9 130 - H F y . +I- 0 22 110 - 0 Non-hominoid primates W GreatApes + Homo sapiens 90 70 Hylohatids 0 Australopithecus africunus - Nonhominid RMA 1 . - - 1 -.. 0 1 . 9 - 1 -. - 1 Kabwe . . . , . . Fig. 7. Bivariate plot of the angle of orbital axis orientation (AOA) and the cranial base angle (CBA) in primates. The reduced major-axes (RMA) for nonhominid and hominoid primates are added. The polygon surrounding the human mean defines the range of values obtained over the human sample. bivariate comDarisons within the Homo saRESULTS piens sample:*CBA vs. IRE1; CBA vs. AOA; Homo sapiens CBA VS.AF'K; AOA vs. AF'K. Pearson correlation coefficients were acceDted as signifiDescriptive statistics for the measures taken o n the human sample are presented cant at P < 0.05. ~~ 585 HOMINID BASICRANIAL FLEXION 180 170 Hylobatids I0 Great Apes 120 I0 Homo sapiens Ausfrulopithecus africanus 110 0 I+ 6) $ Kabwe I I I 100 Cranial base angle (degrees) Fig. 8. Bivariate plot of the angle of facial kyphosis (AFK) and the cranial base angle (CBA) in primates. The reduced major-axes (RMA) for nonhominid and hominoid primates are added. The polygon surrounding the human mean defines the range of values obtained over the human sample. cu 0 90 I 100 I . 110 " I . 120 ' ' I . 130 . . I . 140 . ~ 150 l ' ' ' 160 I ' ' 170 ' I ~ ~ ' t 180 Angle of facial kyphosis (degrees) Fig. 9. Bivariate plot of the angle of orbital axis orientation (AOA) and the angle of facial kyphosis (AFK) in primates. The reduced major-axes (RMA) for nonhominid and hominoid primates are added. The polygon surrounding the human mean defines the range of values obtained over the human sample. in Table 4. Mean neurocranial volume and the standard deviation of the sample are very similar to those estimated for the human species by larger samples (approx. 1,350 cc and 157 cc, respectively [Henneberg, 1990; Tobias, 19941); other sample statistics obtained in the present study may therefore also be good estimates of the degree of human variation. Similarly, the values for neurocranial volume obtained for the great apes 586 C. ROSS AND M. HENNEBERG (Table 5) are comparable with those given by Tobias (1994). Figure 6 is a plot of CBA (cranial base angle) against I R E l (Index of Relative Encephalization 1) with the RMA regression line for nonhominid primates added. Homo sapiens has a much larger brain size (estimated by neurocranial volume) relative to basicranial length than do other primates (Fig. 6): the mean IREl for humans lies outside the range for nonhominid primates. Also, a s has long been known, modern humans have more flexed basicrania than other primates (Fig. 6). The range of values for the CBA in modern humans extends from 135", equal to the mean for the Pongo sample, to 92" (three standard deviations from the mean). The correlation between CBA and IREl that was observed across nonhominid primates (Ross and Ravosa, 1993) was not seen within the human sample. One-tailed t-tests reveal that the mean value for the cranial base angle in humans is significantly higher than the mean predicted by the nonhominid RMA. The entire range of values for the angle of orbital axis orientation (AOA) recorded for Homo sapiens lies outside the range of mean values for other primates: humans have more ventrally deflected orbital axes than all other primates (Fig. 7). Modern humans also have more ventrally deflected orbital axes than expected for a hominoid with their CBA, despite the fact that the human mean was included in the calculation of the hominoid RMA (Fig. 7 ) !However, the mean value for AOA among humans does not differ significantly from the mean value predicted by the nonhominid RMA regression equation (Fig. 7). Within the human sample, the correlation between AOA and CBA was moderate (r = 0.705) while the resulting RMA line (r = 0 . 8 4 8 ~+ 19.768) is not significantly different from the RMA for AOA and CBA across hominoids. The mean value for the angle of facial kyphosis (AFK) for the human sample is equalled or exceeded by the means of all but four other primates (Figs. 8 and 9). Thus, modern humans have comparatively ventrally deflected palates. However, the mean value for the AFK among modern humans is significantly higher (one-tailed t-test) than the mean value estimated by the nonhominid RMA equation (Fig. 8).The correlation between AFK and CBA observed in interspecific comparisons across nonhominid primates (Ross and Ravosa, 1993) is observed within the human sample (r = 0.492) with the RMA equation (y = 0 . 8 7 6 ~+ 24.261) not being significantly different from that for all hominoids. Regression of AOA on AFK is illustrated in Figure 9. One-tailed t-tests reveal that, for their AFK, humans have more ventrally deflected orbital axes than predicted by the nonhominid RMA equation. AOA and AF'K are not significantly correlated within the human sample a s they are among nonhominid primates. Fossil hominids A. africanus and Archaic Homo sapiens resemble modern humans in having more flexed basicrania than non-hominid primates. Two-tailed t-tests suggest that the CBAs of Sts 5 (114"), MLD 37/38 (126"), and Kabwe (126") are not significantly different from the mean for the Homo sapiens sample. If any of our estimates of the CBA in H. erectus is correct, then, contra Maier and Nkini (19841, OH9 also had a degree of flexion not significantly different from the human mean. When CBA is regressed against IREl (Fig. 6) the Kabwe Skull is seen to resemble humans in having a significantly larger CBA 6 e . , a less flexed basicranium) than that predicted by the nonhominid RMA. The values for the CBA for OH9, Sts 5, and MLD 37/38 do not differ significantly from those predicted by the RMA equation for nonhominids. There are no significant differences between the slopes or y-intercepts of the RMA equations for the regression of CBA on IREl among nonhominid, nonhominoid, or hominoid primates (Table 6). Consequently, only the RMA for nonhominid primates is shown in Figure 6. A. africanus resembles H. sapiens in having comparatively ventrally deflected orbits: the only primates with smaller AOAs are Tarsius syrichta (139") and H. sapiens (Fig. 7). Sts 5 has a higher AOA than predicted by the nonhominid RMA equation (one-tailed ttest), having more dorsally deflected orbital HOMINID BASICRANIAL FLEXION 587 axes than predicted for a nonhominid pri- tion expected for a nonhominid primate with mate with its CBA. their degree of basicranial flexion. The orbit Comparison of the RMAs for hominoids orientation ofA. africanus is as predicted for and nonhominoids reveals the slopes to be hominoids with their CBA, but higher than not significantly different, but the y-inter- predicted for a nonhominid (Fig. 7). Comparcept of hominoids to be significantly higher ison of RMA regression lines indicates that than that of nonhominoids (Table 6). Thus, hominoids have more dorsally deflected although the AOA and CBA covary across palates than nonhominoids when CBA is hominoids in a manner similar to that seen taken into account. in nonhominoids, hominoids a s a group have All hominoids have more airorhynch more dorsally deflected palates relative to palates than predicted for nonhominoids their CBA than other primates. (and nonhominids) with their degree of basiFigure 8 illustrates the regression of AFK cranial flexion (Fig. 8).Modern humans have on CBA. Sts 5, the Kabwe Skull, and the fairly kyphotic palates in comparison with great and lesser apes resemble modern hu- other primates, however. Comparison of mans in having more dorsally deflected RMA regression lines indicates that homipalates than predicted by the nonhominid noids have more dorsally deflected palates RMA. Comparison of the hominoid RMA re- than nonhominoids when CBA is taken gression of AFK on CBA with the RMA for into account. nonhominoid primates reveals that the Hominoids, including humans, have more slopes are not significantly different (Table ventrally deflected orbital axes for their 6) but that the y-intercept for hominoids is palate orientation than other primates higher than that for other primates. Thus, (Fig. 9). hominoids have more dorsally deflected palates €or their CBA than other primates. DISCUSSION As in humans, AOA relative to AF'K in Sts Basicranial flexion 5 is significantly smaller than that predicted by the nonhominid RMA equation (Fig. 9). Humans have long been known to have Comparison of RMA regressions for homi- extreme degrees of basicranial flexion and noid and nonhominoid primates reveals that australopithecines have often been assumed their slopes are significantly different to have a degree of flexion intermediate be( P < 0.05): for the range of values covered tween that of humans and the great apes by primates, hominoids have more ventrally (e.g., Biegert, 1963). The results of the presdeflected orbital axes for their palate orien- ent study suggest that this is not the case: tation than other primates. Australopithecus africanus, represented by Sts 5 and MLD 37/38, has the degree of basicranial flexion seen in Modern and Archaic SUMMARY OF RESULTS Homo sapiens. If our estimates of basicranial Homo erectus, Australopithecus africanus, flexion in OH9 are correct, then this is also Archaic and Modern Homo sapiens have true of Homo erectus. These results confirm more flexed basicrania than other primates. observations by Ashton et al. (1975) that baWhen relative brain size is taken into ac- sicranial flexion measured endocranially is count, Archaic and Modern Homo sapiens similar in Sts 5 to that in Homo sapiens. have significantly less flexed basicrania However, previous studies of basicranial than predicted by the nonhominid RMA. H. flexion using measurements on the exterior erectus and A. africanus have the degree of of the skull found no similarity between Sts flexion predicted by the RMA regression 5 and Homo sapiens (Laitman et al., 1979). equation for nonhominids (Fig. 6). This suggests that endocranial and exocraModern humans have more ventrally de- nial flexion of the basicranium bear little if flected orbits than all other primates and, any relationship to each other. for their CBA, have more ventrally deflected Although A. africanus, H. erectus, and Arorbits than predicted by the RMA for homi- chaic and Modern humans have similar denoids. However, they have the orbit orienta- grees of basicranial flexion, relative brain 588 C. ROSS AND M. HENNEBERG size in the four groups is very different: the may select for increased basicranial flexion humans have a much greater brain size rela- with increasing relative brain size. If these explanations for the correlation tive to basicranial length than A. africanus and H. erectus. In the context of non-human between increasing relative brain size and primates, A. africanus and H. erectus have increasing flexion are correct, why do huthe degree of flexion for their relative brain mans not obey the nonhominid primate size predicted by the RMA, but Homo sapiens trend? Basicranial flexion is correlated with have less flexed basicrania than predicted. orbit and palate orientation, both across all Thus, anatomically modern humans do not primates and among modern humans, i.e., follow the nonhominid primate trend €or ba- increasing basicranial flexion is associated sicranial flexion to increase with increasing with increased ventral deflection of the orbrain size relative to basicranial length. Be- bital axes and palate relative to clivus. If fore addressing the deviation of humans humans had the CBA predicted for a primate from the non-hominid primate trend, why of their relative brain size, and values for does this trend exist in other primates? AFK and AOA predicted for a primate with As noted by Ross and Ravosa (1993), in- that CBA, their faces and basicrania would creasing flexion of the basicranium effec- be so closely approximated as to occlude the tively increases the volume of the skull uti- airway and disturb the functional relationlized as neurocranium without increasing ships in the masticatory apparatus. This is the length of the skull “in a fashion analo- clearly untenable. It would seem that, given gous to increasing the proportion of a sphere the relationships between basicranial which is utilized without increasing the flexion and the orientation of the upper and sphere’s diameter” (Ross and Ravosa, 1993, lower face demonstrated here, basicranial p. 319). Basicranial flexion thereby enables flexion cannot be reduced much below 90” the brain to be enlarged without changing without radical changes in the entire craits “spherical” shape and without increasing nial architecture. the diameter of the skull. This is advantaIt is therefore likely that, because of spageous for two reasons. First, although the tial-packing problems, the degrees of basibrain becomes larger, distances between the cranial flexion seen in A. africanus and H. different parts of the brain are minimized. erectus are close to the extreme value possiIf the brain is enlarged by becoming longer, ble and that modern humans do not exhibit then the distance between rostra1 and caudal significantly more flexed basicrania because poles of the cerebrum, for example, would it would be functionally impossible. This imexpand, increasing the time that it takes for plies that brain expansion in the human linneuronal impulses to travel between the two eage beyond that seen in A. africanus must points. Given that many brain functions are have been accommodated in the skull via not localized but are dispersed over various mechanisms other than basicranial flexion. areas of cortex, this may be an important One possible mechanism is suggested below functional constraint on brain shape. Sec- (see Facial Orientation). There are also good reasons for believing ond, a spherical brain results in a spherical brain case. Spherical shell structures not that there are upper limits to the angle of only enclose the greatest amount of volume flexion measured by the CBA. Retroflexion for a given surface area, but they also pro- of the basicranium would not only limit the vide greater strength than cylindrical shell amount of space available for the brain, but structures, enabling them to be thinner- it would also reverse the primary flexure of walled (Demes, 1985). For example, Demes the brain that occurs during primate ontog(1985) has argued that the thick wall of the eny. Thus, the few values for CBA that exHomo erectus skull may be necessary for ceed 180“ (Cheirogaleus major 181“,Loris strengthening a skull that is elongated in tardigradus 181”,Alouatta belzebul 186“,A. shape; the more spherical braincase of Homo palliata 188”, and Pithecia pithecia 182”) sapiens facilitates a thinner-walled calvaria. probably approach the upper limits for the Thus, there is reason to suggest that selec- CBA. If this is correct, then the line most tion for a strong and economical brain case appropriate for describing the relationship HOMINID BASICRANIAL FLEXION 100 1 0.5 I I I 1.O I .5 2.0 Index of relative encephalization 1 Fig. 10. Plot of cranial base angle (CBA) against neurocranial volumehasicranial length (IRE11 with Loess (Lowess) regression line added. between CBA and IRE1 is not a reduced major-axis (or least squares regression), but a logistic curve with asymptotes of approximately 90" and 180". At present, however, only the flattening out of the bottom part of the distribution can be demonstrated. Figure 10 illustrates a Loess regression line fitted to all the data from Figure 6 except OH9, this fossil being excluded because of its fragmentary condition. The decrease in the slope of the curve as it approaches Homo sapiens is apparent. It is notable that the most flexed basicranium recorded among humans has a CBA of 92", slightly above the proposed lower limit of 90". Facial orientation When interpreting these results, it is important to note that the angle of orbital axis orientation, the angle of facial kyphosis, and the cranial base angle measure the orientation of different planes (orbit, palate, planum sphenoideum) to the same plane-clivus ossis occipitalis. Consequently, changes in the orientation of clivus alone will result in apparently correlated changes in these angles. However, changes in the cranial base angle are not necessarily accompanied by changes in orbit orientation; across strepsirhines, CBA and AOA are not correlated 589 (Ross and Ravosa, 1993). Moreover, measures of orbit orientation that do not use clivus as a reference plane confirm that orbit orientation across haplorhines covaries with the degree of flexion; the degree of orbital frontation, or the degree of verticality of the orbital margins relative to a line from nasion to inion, covaries with the the CBA across haplorhines but not strepsirhines (Ross, 1995). Finally, although AOA and AFK covary with each other across primates with a n RMA slope not significantly different from 1.00, a s expected ifchanges in clivus orientation account for the correlation between the two (Table 6 ) ,the nature of this relationship can change; the slope of the hominoid RMA is significantly different not only from the nonhominoid line, but also from a slope of 1.00. Thus, although it is possible for changes in clivus orientation alone to produce correlated changes in the CBA, AOA, and AFK, it is clear that more complex morphological changes are being described by our measurements. Hominoid primates (great apes, lesser apes, and hominids) follow the general trends for palate and orbital axis orientation to covary with basicranial flexion. However, they generally have more airorhynch faces (palates and orbits) than other primates, a s reflected i n the transposition of the hominoid RMA lines above that for other primates in Figures 7 and 8. These data support the hypothesis that this is the primitive condition for hominoids (Shea, 1985). Humans are a significant exception to these trends. Their orbits are more ventrally deflected than is expected for a nonhominid primate with their degree of palatal kyphosis and more ventrally deflected than expected for a hominoid with their degree of flexion. However, they do have the palate orientation expected for a hominoid with their degree of basicranial flexion. Why do humans have such ventrally deflected orbits in comparison with other primates with similar degrees of flexion (i.e., Australopithecus africanus)? Why do they display the orbit orientation expected for a non-hominoid primate with their degree of flexion, rather than that expected for a hominoid? Brain expansion in the post-australopithecine lineage leading to Homo sapiens proba- 590 C.ROSS AND M.HENNEBERG bly occurred primarily via addition of mass to the periphery of the brain, to the neocortex. The brain ofHomo habilis, which is “appreciably” more encephalized than that of the australopithecines (Tobias, 19871, exhibits “marked transverse expansion of the cerebrum, especially the frontal and parietooccipital parts . . . and increased bulk of the frontal and parietal lobes” (%bias, 1987, p. 741). We hypothesize that addition of cortex to the frontal lobes of the brain was accomplished by a n anterior and inferior movement of the frontal pole relative to basicranial and facial structures. As a result, the orbits became ventrally deflected relative to the basicranium, explaining why humans have the value for the orbital axis orientation predicted for non-hominid primates rather than hominoids. In other words, modern humans have more ventrally deflected orbital axes than predicted for a hominoid with their CBA because of the need to accommodate brain expansion without basicranial flexion. This may also explain why the hominoid RMA for the regression of AOA on AFK is steeper than that for nonhominids (Fig. 9). It was noted above that although palate and orbit orientation covary across hominoids, they do so in a manner different from that in other primates; i.e., hominoids tend to have more ventrally deflected orbital axes than nonhominoids with equivalent degrees of palatal kyphosis. It is possible that this enables brain expansion to be accommodated via ventral deflection of the upper face, while the associated ventral rotation of the palate is minimized to avoid possibly detrimental effects to the masticatory apparatus and airway, Humans already have one of the most kyphotic palates relative to clivus orientation of any primates (Fig. 81, confirming Laitman et al.’s (1979) measurements using exocranial landmarks. It is difficult to imagine how further restriction of the oropharynxdue to approximation of clivus and palatemight be accommodated. CONCLUSIONS Gould (1977) hypothesized that increased brain size relative to basicranial length is the most important cause of the extreme ba- sicranial flexion and other “paedomorphic” features characteristic of the human skull. Ross and Ravosa (1993) found, as predicted by Gould‘s hypothesis, that increasing basicranial flexion among nonhominid primates is significantly correlated with increasing brain size relative to basicranial length. The present study demonstrates that living and fossil hominids do indeed have larger brains relative to basicranial length than other primates, a s well a s having more flexed basicrania. However, when the values for hominid basicranial flexion are compared with those predicted by the nonhominid primate reduced major-axis equation, it is clear that Archaic and Modern Homo sapiens have less flexed basicrania than predicted for their relative brain sizes. Moreover, the considerable variability in relative brain size among hominids is not correlated with differences in the degree of basicranial flexion. These results suggest that although Gould‘s (1977) hypothesis is generally true across all primates, increased basicranial flexion in post-australopithecine hominids is not attributable to increases in relative brain size. Hominids probably achieved the most extreme basicranial flexion possible early in their evolution; subsequent expansion in relative brain size was not accompanied by increases in basicranial flexion. This suggests that if basicranial flexion serves a s a mechanism for accommodating a n expanding brain, then brain expansion among hominids must have been accommodated by mechanisms other than basicranial flexion. One possible mechanism is ventral deflection of the upper face relative to both the basicranium and the palate; another is lateral expansion of the braincase above the petrous parts of the temporal bone. Historical changes in patterns of morphological correlation Despite the admittedly restricted sample of fossil hominids used here, we feel that the patterns of correlation discussed here (and in Ross and Ravosa, 1993) can be hypothesized to have arisen as the result of certain evolutionary changes in anthropoid skull morphology. In Figure 11,these changes are mapped onto a n accepted phylogeny of the Hominoidea; their consequences for patterns HOMINID BASICRANIAL FLEXION Apes A. africanus 591 Homo sapiens 0 Lower limit of Other’ anthropoids Fig. 11. Phylogeny ofhominoid primates with evolutionary events discussed in the text mapped onto it. of interspecific and evolutionary correlation are depicted in the “correlation-line’’ diagram to the right. Extensive approximation of the bony orbits below the olfactory tract in the haplorhine stem lineage resulted i n morphological integration of the orbits and anterior basicranium (Ross and Ravosa, 1993; Ross, 1994). As a result, interspecific changes in orbital axis orientation among anthropoids were accompanied by correlated increases in basicranial flexion (and vice versa). Subsequently, the hominoid stem line a g e - o r a predecessor-acquired increased airorhynchy (dorsal deflection of the face relative to the basicranium), accounting for the higher y-intercept of the hominoid RMA in Figure 7 (Shea, 1985, 1988). This is symbolized in the right side of Figure 11 by the transposition of the hominoid “correlation line” above that of other anthropoids. It is notable, however, that the correlation between AOA and CBA still obtains across hominoids. Increases in flexion in the hominoid lineage continued until the lower limit for the CBA (90”) was reached, possibly by some members of the A. africanus popula- tion. Further increases in relative brain size in the lineage leading to modern humanswhich could not be accommodated by increased basicranial flexion-were accommodated by ventral deflection of the orbital axes, giving Homo sapiens the orbit orientation expected for a nonhominid primate with their degree of basicranial flexion. If this sequence of events is correct, the basicranium of Homo habilis should not be significantly different from that of humans (or A. africanus) in its degree of flexion. Whether the increased ventral deflection of the orbits posited for the humans lineage had already occurred in H. erectus or H. habilis remains to be determined. However, given the expansion of the neocortex that was well underway in Homo habilis, it might be expected that this taxon would exhibit some ventral deflection of the orbits relative to the basicranium. If the sequence of events postulated in Figure 11 is correct, then i t is important to explain why hominoids a s a group have more airorhynch faces than other primates; it is this increased airorhynchy that enabled 592 C. ROSS AND M. HENNEBERG Homo sapiens to accommodate subsequent neural expansion via ventral deflection of the orbital axes. It is also important to understand how palate and orbit orientation might become dissociated during evolution; it is only via this dissociation that humans could ventrally deflect their orbits without also ventrally deflecting their palates. Such questions regarding morphological integration and “disintegration” during hominoid evolution remain to be addressed. ACKNOWLEDGMENTS We thank Robin Van Der Riet for providing access to the facilities in the Department of Radiology at Johannesburg General Hospital, Elspeth Kruger for assistance in obtaining images of Sts 5 , and Berneice Eales for the lateral radiographs of the Dart Collection skulls. Dr. Frans Zonneveld kindly provided hospitality and assistance in obtaining access to CT scans in the curatorship of the Foundation for Hominid Palaeoradiology. At the Transvaal Museum, Dr. Francis Thackeray kindly granted permission to study Sts 5 and Mr. David Panagos expertly removed the calotte of Sts 5 . Thanks are also offered to Brigitte Demes and Charles Lockwood, whose comments improved the manuscript; Bill Jungers, who assisted with Loess regression; and David Strait, for productive discussions on basicranial evolution. C.R. was supported by a J.J. Smieszek Research Fellowship in the Department of Anatomy and Human Biology and by a Postdoctoral Research Fellowship from the University of the Witwatersrand. This research was supported by a SAFRD Research Grant to M.H. LITERATURE CITED Ashton EH, Flinn RM, and Moore WJ (1975) The basicranial axis in certain fossil hominids. J. Zool. Lond. 176:577-591. Biegert J (1963) The evaluation of characteristics of the skull, hands and feet for primate taxonomy. In SL Washburn (ed.):Classification and Human Evolution. Chicago: Aldine, pp. 116-145. Dean MC, and Wood BA (1981) Metrical analysis of the basicranium of extant hominoids and Austrulopithecus. Am. J . Phys. Anthropol. 59:53-71. Dean MC, and Wood BA (1982) Basicranial anatomy of Plio-Pleistocenehominids from East and South Africa. Am. J . Phys. Anthropol. 59:157-174. Dean MC, and Wood BA (1984) Phylogeny, neoteny and growth of the cranial base in hominoids. Folia Primatol. (Basel) 43:157-180. Demes B (1985)Biomechanics ofthe Primate Skull Base. Advances in Anatomy, Embryology and Cell Biology, No. 94. Berlin: Springer-Verlag. Draper N, and Smith H (1981)Applied Regression Analysis. New York: John Wiley and Sons. DuBrul EL (1950) Posture, locomotion and the skull in Lagomorpha. Am. J. Anat. 87:277-313. Efron B, and Tibshirani R (1991) Statistical data analysis in the computer age. Science 253:390-395. Gould SJ (1977) Ontogeny and Phylogeny. Cambridge: Belknap Press. Henneberg M (1990) Brain sizehody weight variability in Homo sapiens: Consequences for interpretinghominid evolution. Homo 39:121-130. Holloway RL (1973a) Endocranial volumes of early African hominids and the role of the brain in human mosaic evolution. J . Hum. Evol. 2:449-458. Holloway RL (1973b) New endocranial volumes for the East African early hominids. Nature 243.9-99. Holloway RL (1981)Exploring the dorsal surface of hominid brain endocasts by stereoplotter and discriminant analysis. Philos. Trans. R. SOC.Lond. [Biol.] 292: 155-166. Laitman J T (1985)Evolution of the hominid upper respiratory t r a c t the fossil evidence. In PV Tobias (ed.): Human Evolution: Past, Present and Future. New York: Alan R. Liss, pp. 281-286. Laitman JT, Heimbuch RC, and Crelin ES (1979) The basicranium of fossil hominids as an indicator of their upper respiratory systems. Am. J. Phys. Anthropol. 51t15-34. Maier N, and Nkini A (1984) Olduvai hominid 9: New results of investigation. In P Andrews and J L Franzen (eds.):The Early Evolution of Man. Cour. Forsch. Inst. Senckenberg Vol. 69, pp. 69-82. Pycraft WP (1928) Rhodesian Man and Associated Remains. London: British Museum (Natural History). Ravosa MJ (1988) Browridge development in Cercopithecidae: A test of two models. Am. J.Phys. Anthropol. 76:535-555. Ricker WE (1984) Computation and uses ofcentral trend lines. Can J . Zool. 62:1897-1905. Ross CF (1993)The Functions ofthe Postorbital Septum and Anthropoid Origins. Unpublished Ph.D. Dissertation, Duke University. Ross CF (1994) Craniofacial evidence for anthropoid and tarsier relationships. In J Fleagle and RF Kay (eds.): Anthropoid Origins. New York: Academic Press, pp. 469-457. Ross CF (1995) Allometric and functional influences on primate orbit orientation and the origins of the Anthropoidea. J . Hum. Evol. 29t201-227. Ross CF, and Ravosa MJ (1993)Basicranial flexion, relative brain size and facial kyphosis in nonhuman primates. Am. J.Phys. Anthropol. 91:305-324. Shea BT (1985) On aspects of skull form in African apes and orang-utans, with implications for hominoid evolution. Am. J. Phys. Anthropol. 68:329-342. Shea BT (1988) Phylogeny and skull form in the hominoid primates. In J H Schwartz (ed.):Orang-Utan Biology. Oxford: Oxford University Press, pp. 233-245. HOMINID BASICRANIAL FLEXION Sokal RF, and Rohlf FJ (1981) Biometry. New York: WH Freeman and Co. Tobias PV (1987) The brain of Homo habilis: A new level of organization in cerebral evolution. J . Hum. Evol. I6:74 1-761. Tobias PV (1994) The craniocerebral interface in early hominids. In RS Corrucini and RL Ciochon (eds.): Integrative Paths to the Past. Paleoanthropological Advances in Honor of F. Clark Howell. Advances in Human Evolution Series, Vol. 2. Englewood Cliffs, NJ: Prentice Hall, pp. 185-203. Tsutakawa RK, and Hewett J E (1977) Quick test for 593 comparing two populations with bivariate data. Biometrics 33:2 15-2 19. Weidenreich F (1924) Die Sonderform des Menschenschadels als Anpassung a n den aufrechten Gang. Z. Morphol. Anthropol. 24t157-189. Weidenreich F (1941)The brain and its role in the phylogenetic transformation of the human skull. Trans. Am. Phil. SOC.31:321-442. Wood Jones F (1917) Arboreal Man. London: Arnold. Zuckerman S (1955)Age changes in the basicranial axis of the human skull. Am. J. Phys. Anthropol. 13: 521-539.