AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 90:185-198 (1993) Dermatoglyphic Asymmetry and Testosterone Levels in Normal Males CHERYL SORENSON JAMISON, ROBERT J. MEIER, AND BENJAMIN C. CAMPBELL Anthropology Department, Indiana University, Bloomington, Indiana 47405 (C.S.J., R.J.M.) and Anthropology Department, Haruard University, Cambridge, Massachusetts 02138 (B.C.C.) KEY WORDS Finger prints, Palm prints, Salivary steroids, Geschwind Hypothesis ABSTRACT Dermatoglyphic prints and salivary samples were taken on a sample of 39 adult males. A statistical relationship between dermatoglyphic asymmetry and adult testosterone levels as measured in saliva was examined for seven dermatoglyphic variables by means of correlation, regression, and analysis of covariance, controlling for age and stature when necessary. The first two types of anlayses indicated a significant effect of testosterone level upon the asymmetry of three dermatoglyphic variables: a-b ridge count, palmar pattern intensity, and the combined pattern intensity of palm and digits. Analysis of covariance, which examined the effect of testosterone level as a categorical variable, while holding age or stature constant, demonstrated the asymmetry of five variables to be significantly affected by testosterone level: radial digital count, digital pattern intensity, palmar pattern intensity, total digital ridge count, and the combined palmar and digital intensity. Although there is as yet only associational evidence linking levels of prenatal and secondary testosterone, the results of the present study lend support to the hypothesis that prenatal testosterone levels may have a significant effect on the development of dermatoglyphics. 0 1993 Wiley-Liss, Inc The search for the determinants of dermatoglyphic variation has identified a number of potential alternatives, many of which could function interactively. Genetics, of course, plays a major role (e.g., Holt, 1956; Loesch et al., 1978), but the fact that even monozygotic twins differ to some degree in dermatoglyphic traits (e.g., Reed et al., 1977) indicates a strong prenatal environmental component in the expression of dermatoglyphic variation. Since dermatoglyphic characteristics are established before the 19th week of gestation (Cummins, 1929; Hale, 1949, 1952; Mulvihill and Smith, 1969; Babler, 1987), whatever factors that are hypothesized to affect ridge development must be in operation before or during this time period. After the 19th week, epidermal ridge configurations are set; injury and caustic substances 0 1993 WILEY-LISS, INC. may reduce the appearance of the ridges, but unless the skin is destroyed to a depth of at least one millimeter, the original pattern will return (Cummins and Midlo, 1961). A number of studies have elucidated the effects of various environmental factors on dermatoglyphic traits. Internal prenatal environment effects that have been investigated have included, for example: rubella (Achs et al., 1966), alcohol (Qazi et al., 1980; Dar and Jaffe, 19831, cytomegalovirus infec-. tion (Wright et al., 1972), nonspecific intrauterine insult (Rose et al., 1972), and “maternal effects” (Reed et al., 1979). The Received October 10,1991; accepted July 29,1992 Benjamin C. Campbell’s present address is Carolina Population Center, University of North Carolina, Chapel Hill, North Carolina 27516. 186 C. SORENSON JAMISON ET AL. external environment has also been hypothesized to have critical influence on epidermal features (e.g., Rosa, 1985; Loesch, 1986; Blangero, 1988). The present research follows a different line of investigation from those previously described. Here it is proposed that a critical variable, such as the level of a circulating hormone, specifically testosterone, has a genetic as well a s a prenatal environmental component. The prenatal testosterone to which a fetus is exposed arises from two sources: the hormone that the fetus itself produces has a (probably pleiotropic) genetic basis, while the testosterone that originates from the mother provides environmental exposure, as far a s the fetus is concerned. We have earlier reported the rate of maturation to be a determinant of dermatoglyphic variation (Meier et al., 1987; see also Livshits and Kobyliansky, 19871, and that dyslexics, who have been described as being late maturers, differ dermatoglyphically from a sample of normal controls (Sorenson Jamison, 1988, 1990; Meier, 1990, and work in progress). The Geschwind Hypothesis (Geschwind and Behan, 1982; Geschwind and Galaburda, 1985) identifies a n excess of prenatal testosterone to be the central causative factor involved in the development of dyslexia, a s well a s of a number of other central nervous system disorders. The fact that one of the effects of prenatal testosterone is the modification of developmental rate (McEwen, 1981; MacLusky and Naftolin, 1981; Bardin and Catterall, 1981; Geschwind and Galaburda, 1985) ties the two lines of investigation together. If dyslexics and late maturers demonstrate significantly different dermatoglyphic variation than controls, and if testosterone has been related to both the development of dyslexia and to slowed maturation, then the dermatoglyphic investigation of testosterone is a logical next step. It is not possible to test directly for the differential dermatoglyphic effect of prenatal testosterone in humans because of the ethical problems associated with such a n endeavor. However, indirect support for the importance of testosterone to both the variation in dermatoglyphic traits and to the development of dyslexia might be gained if a relationship between dermatoglyphic variation and secondary testosterone, which is more easily and ethically measured, could be demonstrated. Testosterone secretion is not constant throughout the life cycle (Lowrey, 1986; Tanner, 1990). The male fetus secretes increasing amounts of testosterone from the time of Leydig cell differentiation, a t approximately eight weeks, until midgestation (George et al., 1981). From that point, the testosterone level declines gradually until by a few months following birth it has reached the low level characterizing childhood. At puberty, testosterone secretion once again increases and orchestrates the development of secondary sexual characteristics. Testosterone levels do not remain constant during adulthood either: numerous studies have documented a n age-related decline in testosterone starting in early adulthood (at least by the early thirties) in both serum (Deslypere and Vermeulen, 1984; Lewis et al., 1976; Pirke and Doerr, 1973; Purifoy et al., 1981) and saliva (Read and Walker, 1984; Campbell e t al., 1991). While the relationship between prenatal and secondary testosterone levels is not presently known, evidence of a genetic influence on both pubertal and adult levels of testosterone and other androgens has been reported (Akamine et al., 1980; Meikle e t al., 1982, 1988; Rotter et al., 1985). These studies do suggest a n association between a n individual’s prenatal and secondary testosterone levels. It would seem reasonable to assume that the genetic effect on testosterone level would be greatest at the earliest ages (when environmental effects have had less chance to exert themselves) and would then decrease over time. Since environmental influences on testosterone can only obscure a n association between prenatal and postpubertal testosterone levels, a demonstrated relationship between secondary testosterone level and dermatoglyphic patterns may be taken as indirect evidence for a n association between prenatal testosterone level and developing dermatoglyphic features. The focus of the present study was the investigation of the relationship between secondary testosterone level and dermato- DERMATOGLYPHICS AND TESTOSTERONE glyphic asymmetry. Dermatoglyphic asymmetry variables were chosen for a number of reasons. First, such asymmetry has been repeatedly indicated a s being related to developmental variation (e.g., Jantz and Webb, 1980; Livshits and Kobliansky, 1991). A second reason for the selection of asymmetry was that a central prediction of the Geschwind Hypothesis is that a n excess of prenatal testosterone is expected to result in greater morphological asymmetry. Additionally, one of the most significant findings from the dyslexic study was that dyslexics exhibited greater bimanual asymmetry than the controls. MATERIALS AND METHODS The sample for this study included 39 healthy Caucasian male volunteers, between the ages of 18 and 45.Digital prints were taken by C.S.J. with Hollister Footprinters and analyzed by R.J.M., and palm prints were taken with graphite and tape (Robertson-Neufeld and Murray, 1978) and analyzed by C.S.J., using a modified version of the Penrose-Loesch topological technique (Loesch, 1983; Sorenson Jamison, 1988). Saliva specimens, for the testosterone analysis, were taken by R.J.M. on a second visit, in order to ensure comparability of intervening variables. To control for the well documented diurnal variation in salivary testosterone (Walker et al., 1980a; Campbell e t al., 1991; Magrini et al., 1986; Dabbs, 19901, all subjects provided saliva specimens at the same time of day (between 10 and 11in the morning). To avoid any potential contamination of the saliva sample from food a n d o r bleeding gums, subjects were requested not to have eaten or brushed their teeth before arriving. Furthermore, since acute rises in testosterone levels may result from both exercise (Cumming et al., 1986) and sexual activity (Fox et al., 1972), subjects were asked to refrain from these activities in the previous 24 hours. Prior to saliva collection, subjects were given a stick of gum (Carefree Sugarless) to stimulate saliva flow, which, according to Ellison (1988), has no effect on accuracy of testosterone measurement. Saliva was collected into a test tube containing sodium azide, an antibacterial agent. 187 Salivary testosterone was utilized because the methodology offers many advantages over plasma or urinary assays. This method is noninvasive and nondisruptive, the specimen is easy to collect and to store, and furthermore, i t has been demonstrated to reflect free, biologically active plasma testosterone levels (Ellison, 1988; Lipson and Ellison, 1989; Wang et al., 1981; Schurmeyer and Nieschlag, 1982). Testosterone levels in the saliva were determined by BCC with radioimmunoassay, using a specific antibody (No. 250, supplied by Gordon Niswender), as described in Ellison et al. (1989). All samples were assayed in a single assay with a n interassay variability of 9.0%. Salivary testosterone values ranged from 242 pmoVL to 701 pmoVL, within the normal range reported for salivary testosterone of adult males (Read and Walker, 1982; Walker e t al., 1980b; Schurmeyer and Nieschlag, 1982; Campbell et al., 1991; for instance, Ellison et al., 1989 reported a normal range for males in western population to vary between 100 and 1,000 pmol/L). Dermatoglyphic variables and statistical analyses All statistical analyses were run using SPSSPC-t, version 4.01 (NorusidSPSS Inc., 1990). Bimanual asymmetry variables were computed (left minus right) from the following left and right original dermatoglyphic variables: digital radial count, digital ulnar count, palmar a-b count, digital pattern intensity, palmar pattern intensity, total digital ridge count, and combined digital and palmar pattern intensity. The names and definitions of the asymmetry variables examined in the present study are listed in Table 1. These variables, as well as age and TABLE I . List of dermatoglyphic asymmetry variables used in the anaLyses Variable name Definition ~ RCASYM UCASYM ABASYM DPIASYM PPIASYM TDRCASYM COMPASYM Digital radial count Digital ulnar count Palmar a-b count Digital pattern intensity Palmar pattern intensity Total digital ridge count Combined digital and palmar pattern intensity C. SORENSON JAMISON ET AL 188 TABLE 2. Descriptiue statistics for original dermatoglyphic variables and their Pearson correlations with testosterone and age as well as partial correlation with testosterone leuel (after the effect of age has been removed) Variable name Mean S.D. r (Testos) r (Age) Left radial digital count Right radial digital count Left ulnar digital count Right ulnar digital count Left palmar a-b count Right palmar a-b count Left digital pattern intensity Right digital pattern intensity Left palmar pattern intensity Right palmar pattern intensity Left total digital ridge count Right total digital ridge count Left complex pattern intensity Right comDlex Dattern intensitv 61.10 61.15 18.59 23.26 38.33 37.21 6.05 6.51 1.62 1.59 76.69 84.41 7.67 8.10 26.74 24.13 18.72 23.42 3.97 4.74 2.04 1.82 0.91 0.75 41.25 42.58 2.34 2.11 p.218 p.229 -.122 -.3572 p.027 p.227 -.122 -.327' .3892 -.098 -.I97 - .32V ,045 -.3162 ,283 ,231 ,094 .3642 .093 ,233 ,140 .314' p.278 -.090 ,226 .3312 .014 ,238 __ Partial r - ,090 p.133 -.087 p.215 .024 -.130 p.060 - .206 .300 p.167 p.098 p.194 ,060 p.234 N = 39. 'Significant at the .05level stature, were first analyzed with salivary testosterone level using Pearson Product Moment correlation. Regression analyses were then run to examine the predictive effect of salivary testosterone level on dermatoglyphic asymmetry. Thirdly, Analysis of Covariance using the W O V A program was used to determine whether a significant amount of variance in the dermatoglyphic asymmetry variables could be explained by high and low testosterone group membership. RESULTS Descriptive statistics for the original dermatoglyphic variables, from which the asymmetry variables were computed, are presented in Table 2 . On the right side of the table are also found the correlations of the TABLE 3 Pearson product moment correlation coefficients and associated probability (two-tail) between salivary testosterone levels and dermatoglyphic asymmetry uariables, age, and stature Variable RCASYM UCASYM ABASYM DPIASYM PPUSYM TDRCASYM COMPASYM AGE STATURE Entire sample' r P -.0345 ,3984 .2500 ,2722 ,3754 ,3280 .4154 -.5061 .1571 = r P ,835 .I502 ,420 ,012 ,125 ,094 .019 .042 .009 ,001 ,340 ,2307 ,3975 ,2827 ,3967 -.1778 3830 ,212 ,027 ,123 .027 ,339 ,033 ' Entire sample, N = 39. Sample 25 years or younger, N Age LE 25 yrs' 31. ::;;: ::;: dermatoglyphic variables with testosterone and age, and the partial correlation of the variable with testosterone after controlling for the effect of age. The results of the correlation analysis between the dermatoglyphic asymmetry variables and testosterone can be seen in Table 3. It is immediately apparent that while many of the asymmetry variables demonstrate a positive relationship with testosterone level, age is highly significantly correlated with testosterone as well. Two possibilities presented themselves as ways to deal with this relationship. One alternative would be to control for the effects of age statistically while continuing to use the entire sample, the other is to restrict the analysis to a homogeneous age range. Examination of the frequencies for age (see Fig. la) demonstrates that the first break in the range appears at age 25, and this was the break we decided to use in our analyses. A scatterplot showing the distribution of testosterone level by age as well as grouped testosterone level frequencies for the entire sample and the selected restricted age sample are also included in Figure 1 (b, c, and d, Eespectively). The reduced variability apparent in testosterone level with age was Fig. 1. Freauencies of age - a n d testosterone. a: Age for entire sample. b: Scatterplot of testosterone by age. c: Testosterone levels for entire sample. d Testosterone I levels for restricted sample. I - 10 100-200 rFREQUENCY 12 10 301-400 401-500 501-600 TESTOSTERONE PMOL/L 201-300 601-701 Y n 2- 4- 8- 6- 100-200 FREQUENCY 301-400 401-500 501-600 TESTOSTERONE PMOL/L 201-300 601-701 (AGE LE 25, N = 3 1 ) 40 d. FREQUENCIES OF TESTOSTERONE LEVEL 30 . (ENTIRE SAMPLE, N = 3 9 ) 20 . . .. ' c . FREQUENCIES OF TESTOSTERONE LEVELS i . .* . . LEVEL AGE 10 200 400 600 SALlVAFN TESTOSERONE (ENTIRE SAMPLE, N = 3 9 ) b. SCATTERPLOT OF AGE AND TESTOSTERONE AGE 11119 202 122 23 24 25 2827 2829 30 3132 3334 3536373839404 142 4344 45 a. FREQUENCIES OF AGES (ENTIRE SAMPLE, N=39) 50 C. SORENSON JAMISON ET AL 190 TABLE 4a. Results of multiple regression analyses predicting dermatoglyphic asymmetry variables from stature and saliuarv testosterone levels Dermatoglyphic variable RCASYM UCASYM ABASYM DPIASYM PPIASYM TDRCASYM COMPASYM t Stature P B t Testosterone P B -0.72 1.54 -0.16 -0.65 -0.18 0.93 -0.53 ,480 ,136 ,875 ,522 355 ,362 .601 p.215 ,716 -.018 - .024 p.007 ,501 p.030 1.02 0.62 2.31 1.42 2.19 1.11 2.33 ,315 ,539 ,029 ,168 .037 ,277 ,028 ,015 ,027 ,013 ,003 ,004 ,029 ,006 ~~ - Equation R2 ~ .04 .13 .15* .07 .162 .ll ,172 ‘Age 525, N = 31, none of the multiple regression equations were significant a t the .05 level. “The hivariate regression, without controlling for stature, is significant at the .05level. TABLE 46. Results of multiple regression analyses predicting dermatoglyphic asymmetry variables from age and salivarv testosterone levels Dermatoglyphic variable ___ Testosterone Age P -~ t B t D B 1.39 -1.94 -0.46 -0.63 0.21 -0.78 -0.32 ,173 .061 .650 .535 .836 -439 ,750 ,426 p.851 -.057 -.025 .007 -.425 p.018 0.52 1.38 1.11 1.16 2.20 1.42 2.20 ,606 .175 275 .256 .034 ,166 ,034 ,006 .024 .005 ,002 .003 .031 ,005 Equation R’ ~ RCASYM UCASYM ABASYM DPIASYM PPIASYM TDRCASYM COMPASYM .05 24’ -07 -08 .153 .12 ,182.3 ‘Entire sample, N = 39. “he multiple regression equation IS significant at the .05 level. 3The hivariate regression, without controlling for age, is significant a t the .05level. viewed a s offering further support for the decisions to concentrate only on the homogeneous lower age group. However, it must be emphasized that since the number of individuals in our sample over the age of 25 is so small (eight), no valid conclusions can be drawn concerning this age group from the population. There are valid arguments to both methods of dealing with the age effect. Loss of information is always to be avoided, if possible; however, we felt that the problem of this potential loss was overridden by a n additional problem of environmental noise that is presumably introduced with increasing age. Genetic influence on testosterone level was assumed to decrease subsequent to the prenatal period and to become increasingly weaker with advancing age. Therefore, we chose to restrict our analyses to a relatively homogeneous age sample between 18 and 25 years. However, in order to allow readers to form their own judgments about our data, we do include the results of both types of analyses (that is, entire sample versus restricted sample) for comparison purposes. The correlation was run a second time, after restricting the sample to those individuals 25 years old or younger (right side of Table 3). Several of the dermatoglyphic variables continued to be correlated with testosterone in this sample, but age was no longer significantly related to testosterone. However, at this point, stature became important (P = ,033). On the basis of these results the decision was made to examine the effect of stature, controlling for this variable in the restricted sample. The multiple regression results for the age-restricted sample are presented in Table 4a. The effect of stature was not of even borderline significance on any of the variables (column 21, while the effect of testosterone was significant for three variables, ABASYM, PPIASYM, and COMPASYM, even after controlling for the effect of stature Fig. 2. Scattergrams illustrating significant bivariate relationships between testosterone level and asymmetry variables (N = 31, age 25). a: ABASYM. b: PPIASYM. c: COMPASYM. a. TESTOSTERONE AND a COUNT ASYMMETRY 10 -b - b Count bymnwtry I 1 b. TESTOSTERONE AND PALMAR PATTERN INTENSITY ASYMMETRY a 2 Palmer Pattern Inhnolty Aoynnnotry 9.- + --Q- c. TESTOSTERONE AND COMBINED PATTERN INTENSITY ASYMMETRY C o m b l d PotMn lnhnrltv AIYrnnntrv C. SORENSON JAMISON ET AL. 192 TABLE 5. Frequency distrlbution of saliuaq testosterone ; = 311 leuel (in pmollL; age ~ 2 . 5 N Testosterone level 242 270 281 292 293 304 306 318 337 350 355 369 424 426 427 435 444 447 476 482 484 520 521 532 536 550 603 621 657 691 701 (column 5). Scattergrams for the three significant bivariate relationships (that is, regressions without controlling for stature) are illustrated in Figure 2, with the regression lines indicated. Table 4b includes the results from the multiple regression analyses on the entire sample, controlling for the effect of age rather than stature. To examine the effect of high versus low testosterone level as a categorical variable, subjects were divided into testosterone groups on the basis of a frequency distribution (presented in Table 5). Testosterone values ranged from a low of 242 pmoWL to a high of 701 pmoWL, with a mean of 441.74 pmoWL. There was a level, above and below 400 pmoWL, which provided a reasonable sectioning point between two clear clusters. Although this division resulted in two groups of unequal size, it was utilized because of the obvious gap between the values of 369 pmoWL and 424 pmol/L. The low testosterone group (LE 369 pmol/L) included 12 subjects; the high testosterone group (GE 424 pmoWL) was composed of 19 individuals. Using the SPSSPC + program MANOVA, analyses of covariance were run with stature as a covariate. The results can be seen in Table 6a. The effect of stature (presented under the regression columns) was only of borderline significance for the dermatoglyphic variable UCASYM (2' = ,0771,and of no significance for any of the others. Testosterone was not significantly related to UCASYM. However, for the remaining six analyses, it can be seen that five of the variables are significantly affected by testosterone. While ABASYM was not significant (as it had been in the regression analysis), RCASYM, DPIASYM, and TDRCASYM demonstrate significance along with the continuing importance of PPIASYM and COMPASYM. Again, the analyses were run a second time on the entire sample, with age rather than stature as the covariate, and the results are presented in Table 6b. The means and standard deviations for the high and low testosterone groups are presented in Table 7a. One very striking finding that is apparent from this table is that in the low testosterone group, all of the dermatoglyphic asymmetry variable means are negative and all of those in the high testosterone group are positive. Since the asymmetry variables were calculated left minus right, Table 7a indicates that for the low testosterone groups, the right side value was higher than that of the left, while for the high testosterone groups, the left side value was higher. Figure 3 graphically illustrates the differences between the two groups for each of the dermatoglyphic asymmetry variables. Table 7b includes the descriptive statistics for the entire sample divided into low and high testosterone groups on the same basis (LE 369 and GE 424 pmoWL). The positive and negative signs are the same as in Table 7a with the single exception of ABASYM in the low testosterone group. DISCUSSION These results lend strong support to our hypothesis that testosterone level has a significant effect upon dermatoglyphic asymmetry. This is in agreement with a recent general review of the subject of asymmetry and developmental shift by Livshits and Kobyliansky (1991). Of the seven variables studied, only one (UCASYM) failed to demonstrate a clearly significant relationship to salivary testosterone in either the regression or analyses of covariance. This variable, it may be remembered, was also the only one to have a borderline significant relationship to stature. The regression analyses examined testosterone level as a continuous variable. Three of the seven dermatoglyphic variables demonstrated a significant relationship (P < .05) to testosterone level. Two of these variables, ABASYM and PPIASYM, are the 193 DERMATOGLYPHICS AND TESTOSTERONE TABLE 6a. Results of analysis of couariance predicting dermatoglyphic variables from high and low testosterone groups, with stature as a couariate Dermatoglyphic Regression2 RCASYM UCASYM ABASYM DPIASYM PPIASYM TDRCASYM COMPASYM Constant F P 0.35 3.37 0.36 0.14 0.22 1.81 0.01 .560 .077 ,555 ,708 ,640 .189 .935 variable 'Age 525. Total N = 31, high N = 19, low N 'Regression statistics refer to the covariate. = - -. - Testosterone - F P F P 0.30 3.50 0.30 0.09 0.23 1.97 0.02 ,590 ,072 .591 ,773 ,635 ,171 5.42 1.68 2.73 7.54 8.79 6.21 15.11 ,027 ,206 ,110 ,010 ,006 ,019 ,001 .886 12 TABLE 6b. Results of analysis of couariance predicting dermatoglyphic variables from high and low testosterone groups, with age as a couariate Regression2 Dermatoglyphic variable RCASYM UCASYM ABASYM DPIASYM PPIASYM TDRCASYM COMPASYM 'Total N = 39, high N = Constant F P F 3.59 5.47 0.58 0.38 0.00 0 80 0 21 ,066 3.51 3.02 1.28 0.00 0.00 0 16 0 00 21, low N = ,025 ,453 .540 ,989 378 648 .. .. Testosterone ._ .- P F P .069 .091 ,265 ,997 ,972 693 984 3.26 2.66 1.04 5.00 9.56 5 76 13 75 ,080 ,112 ,315 ,032 ,004 022 001 18. a Regression statistics refer to the covariate only two palmar traits included in the study; the third significant variable, COMPASYM, is a combination of the palmar and digital pattern intensity. Examination of the unstandardized Betas from the regression analyses (Table 4)indicated that they were all positive (including those that were not significant), suggesting that for every dermatoglyphic variable (but significantly so for the palmar variables), a higher testosterone level is predictive of greater dermatoglyphic asymmetry. The importance of salivary testosterone when it is considered as a categorical variable is even more clearly demonstrated: the high and low testosterone groups have significantly different ( P < .05) dermatoglyphic asymmetry values for five of the seven variables investigated. Furthermore, the directional effects of the differences are striking: individuals with high testosterone levels have higher values, for every dermatoglyphic asymmetry variable, on the left hand, while individuals with low testosterone levels have higher values on the right hand. These analyses emphasize the rela- tionship of digital variable asymmetry to testosterone level, in contrast to the regression analyses which emphasized the palmar variables. The asymmetry of the radial ridge count, total ridge count, and digital pattern intensity demonstrated significance along with the combined pattern intensity and the palmar pattern intensity. Thus, while both palmar and digital variable asymmetries are significantly affected by testosterone levels, the effect of the hormone is demonstrated in different ways for the palms and digits. There is some overlap in significant variables, but perhaps the differences between the dermatoglyphic regions are more interesting, particularly in relation to the ridge count traits. Only palmar ridge counts were found to be significantly affected by testosterone in the regression analyses, while only digital ridge counts (with sole exception of the ulnar count) were found to be significantly affected in the analyses of covariance. Perhaps for the digital traits, which develop at a different rate than the palmar traits (Could, 19481,the influence of testosterone is exerted a s a threshold effect, 194 C. SORENSON JAMISON ET AL. TABLE 7a. Means and standard deviations for dermatoglyphic asymmetry variables i n low and high testosterone groups, used in analyses of covariance (age S25, N = 31) ance indicated higher values to be on the left hand of the high testosterone group while the opposite was true of the low testosterone group may be a reflection of the greater inLow testosterone High testosterone fluence of testosterone upon the left side of group group the body. Additional support for this suppoMean SD Dermatoglyphic Mean SD (N = 19) variable (N = 12) sition may be found in the fact that the dys~. lexic study demonstrated there to be more 1.684 8.453 9.166 RCASYM -5.650 1.053 15.936 13.192 UCASYM -6.750 left hand variables which differentiated sig2.211 4.131 -0,167 2.918 ABASYM nificantly between dyslexics and controls DPIASYM -1,000 0.739 0.053 1.177 1.071 1.155 0.579 PPIASYM -0.667 than right hand variables (Sorenson Jami2.737 15.779 16.223 TDRCASYM -12.000 son, 1988). 0.632 1.535 -1.667 1.664 COMPASYM Hence, one interpretation of the findings of the present study is to suggest that the relationship between salivary testosterone TABLE 76. Means and standard deviations for and dermatoglyphic asymmetry is a reflecdermatoglyphic asymmetry variables i n low and high testosterone groups, used i n analyses of covariance tion of the much earlier effect of prenatal (entire s a m d e . N = 39) testosterone upon developing dermatoLow testosterone High testosterone glyphic variation. The present research pergroup group mits no precise prediction of the relationship Mean SD Mean SD Dermatoglyphic between levels of prenatal and secondary (N = 18) (N = 21) variable - .. testosterone. However, prenatal testoster1.619 8.322 -2,000 10.655 RCASYM one levels in the male fetus appear to be 0.529 15.276 13.132 UCASYM -10.722 ABASYM 0.222 3.422 1.905 4.146 largely the result of fetal production, which 0.000 1.140 -1,000 1.237 DPIASYM reaches adult male levels (Winter, 1987). 1.030 0.984 0.524 PPIASYM -0.556 While the fetus is exposed to testosterone 2.143 15.519 16.613 TDRCASYM -12.772 0.524 1.504 1.617 COMPASYM -1.556 from maternal sources as well, maternal levels of “free testosterone” are so much lower than those produced fetally (Rivarola e t al., while for the palmar traits there is a more 1968)that any normal variation in maternal directly linear effect of the hormone upon levels (James, 1987a,b) would appear negligible when compared to potential variation ridge formation. Interpreting these results is somewhat in fetal testosterone. Thus prenatal tesdifficult. The Geschwind Hypothesis (Gesch- tosterone exposure and adult levels in norwind and Galaburda, 1985) predicts that an mal men represent testicular production a t excess of prenatal testosterone results in the different points in the life cycle. We hypothesize that in spite of the considslowed maturation of the left side of the erable intra-individual variation in the sebrain, with concomitant enhancement of the right side (and thus, abilities associated creted amounts of both prenatal and secondwith left side dominance, such a s reading, ary testosterone (due to developmental and are impaired, while those associated with cyclical variables), there is still a within-inright side dominance, such as music and dividual relationship between the average mathematics, are augmented). Since prena- circulating levels of the two hormones. The tal testosterone is hypothesized to disrupt dermatoglyphic data presented in this study the characteristic asymmetry of the brain, it support this hypothesis. would seem to be a reasonable extension of the Geschwind Hypothesis to suggest that it might also disrupt developing bilateral symmetry for those additional morphological deFig. 3. Summaries of statistical relationships bevelopments that are occurring when prenatween high a n d low testosterone groups (N = 31, tal testosterone is circulating. A listing of age < 25). a: Means. b: Sta nda rd deviations. c: 95% such features would include dermatoglyphic Confidence levels of low group. d 95%Confidence levels traits. The fact that the analyses of covari- of high group. -25 -15 -10 -20 -6 0 5 .- -12 -10 -8 -8 -4 - RwgyM UC48YM ABABYM - I - LO = T Gfp H u n I ,,I I EB ni Qrp uAn ASYMMETRY VARIABLES I DPIASYM PPUB/MTDRUBYUCOMPMYM - I 0. mHIatp8D - - I - I L T T T d. 95% CONFIDENCE INTERVALS OF ASYMMETRY VARIABLES IN HIGH TESTOSTERONE GROUP mLoQtpS0 M Y MMETRY VARIABLES b. STANDARD DEVIATIONS OF VARIABLES OF HI AND LO TESTOSTERONE GROUPS 5- n 5 10 I5 -2 *O 0 I 2 4, a. MEANS OF ASYMMETRY VARIABLES OF HI AND LO TESTOSTERONE GROUPS 196 C. SORENSON JAMISON ET AL The precise nature of the relationship between levels of prenatal testosterone and secondary testosterone cannot yet be defined. However, our results certainly can be viewed as adding support to such a relationship, especially when combined with the dermatoglyphic findings of early and late maturers (Meier et al., 1987) and dyslexia (Sorenson Jamison, 1988, 1990; Meier, 1990, and work in progress). There is some supportive additional research emphasizing the proposed relationship between prenatal and secondary hormone levels. In a recent study with sheep, the authors reported that the timing of postnatal neuroendocrine maturation is a function of prenatal androgen level (Wood et al., 1991). Additionally, a n investigation of human lateralization (Tan, 1990) led to the conclusion that cerebral lateralization is a function of both prenatal and secondary testosterone levels. Although the likelihood of direct evidence of the relationship between human prenatal and secondary testosterone being demonstrated in the near future is not great (because of the associated ethical problems), additional indirect evidence might be gained from testing salivary testosterone levels from individuals identified by the Geschwind Hypothesis as having been subjected to a n excess of prenatal testosterone. Such groups might include, for instance, dyslexics, stutterers, left-handers, individuals suffering from autoimmune disorders or migraine headaches, individuals who are nearsighted, and individuals gifted in rightbrain abilities such a s mathematics, music, or art. A direct effect of prenatal testosterone level upon dermatoglyphic traits cannot be definitively stated as yet. But again, the results of this study offer considerable support for the hypothesized effect. As has been explained more fully in earlier publications (e.g., Sorenson Jamison, 1988, 1990), a mechanism by which prenatal testosterone could exert a n effect on dermatoglyphic traits might exist in the fact that the hormone serves as a stimulus for both Nerve Growth Factor and Epidermal Growth Factor. Furthermore, testosterone is circulating within fetuses of both sexes during the time period of dermatoglyphic formation (Wilson et al., 1981; Zaaijer and Price, 19711, and additional amounts are of placental and maternal origin (Geschwind and Galaburda, 1985). The earlier research with dyslexics does not definitely associate their significant dermatoglyphic differences from controls with levels of prenatal testosterone, but the present results clearly demonstrate the significant dermatoglyphic effect of that hormone, albeit a measure of secondary rather than prenatal testosterone. The direction of future research appears to be obvious: to see if a definite relationship between levels of prenatal testosterone and dermatoglyphic variation can be determined. Since this type of research on humans is unethical a s well as impractical, the next best approach would be to utilize already available, cage-reared, nonhuman primates. Such a study is presently underway using dermatoglyphic data collected from macaques a t the University of Wisconsin Regional Primate Research Center (Meier et al., 1992).Monkeys whose mothers were differentially injected with testosterone a t various points during pregnancy serve as subjects and are compared to a suitable control group of like-aged macaques from the Regional Primate Center. ACKNOWLEDGMENTS The authors gratefully acknowledge the editorial and statistical help of Paul L. Jamison. We also very much appreciate the timeconsuming contribution made by our volunteer subjects and the use of equipment and supplies from the Reproductive Ecology Lab, Anthropology Department, Harvard University, Peter ellison, Director. The suggestions of three anonymous reviewers were extremely valuable. This project was supported by grants received from the Biomedical Sciences Program, Indiana University, and The National Science Foundation. LITERATURE CITED Achs R, Harper RG, and Siege1 M (1966) Unusual dermatoglyphic findings associated with rubella embryopathy. N Eng J Med 274:148-150. Akamine Y, Kato K, and Ibayashi H (1980) Studies on changes in the concentration of serum adrenal androgens in pubertal twins. Acta EndocrinoI93:356-364. Babler W J (1987) Prenatal development of dermato- DERMATOGLYPHICS AND TESTOSTERONE glyphic patterns: Associations with epidermal ridge, volar pad and bone morphology. Collegium Antropologicum 11t297-304. Bardin CW and Catterall CF (1981) Testosterone: Amajor determinant of extragenital sexual dimorphism. Science 212r1285-1294. Blangero J (1988) The selective neutrality of dermatoglyphic variation. Int. J . Anthropol. 3:289-299. Campbell BC, Colfax D, and Ellison PT (1991) Age-related changes in pulsatile release of salivary testosterone among adult males. Paper presented a t the 60th Annual Meeting of the American Association of Physical Anthropologists, Milwaukee, Wisconsin. Cumming DC, Brunsting LA 111, Strich G, Ries A, and Rebar RW (1986) Reproductive hormone increases in response to acute exercise in men. Med. Sci. Sports Exerc. 18:369-373. Cummins H (1929) The topographic history of the volar pads (walking pads; tastballen) in the human embryo. Contributions to Embryology, No. 113.Carnegie Institute of Washington 20r103-106. Cummins H and Midlo C (1961) Finger Prints, Palms and Soles. New York: Dover. Dabbs JM (1990) Salivary testosterone measurements: Reliability across hours, days, and weeks. Physiol. Behav. 48t83-86. Dar H and Jaffe M (1983) Dermatoglyphic and palmarcrease alterations as indicators of early intra-uterine insult in mental retardation. Dev. Med. Child Neurol. 25t53-59. Deslypere J P and Vermeulen A (1984) Leydig cell function in normal men: Effect of age, lifestyle, residence, diet and activity. J. Clin. Endocrinol. Metab. 59:955962. Ellison PT (1988) Human salivary steroids: Methodologies, considerations and applications in physical anthropology. Yearbook Phys. Anthropol. 31:115-142. Ellison PT, Lipson SF, and Meredith MD (1989) Salivary testosterone levels in males from the Ituri Forest of Zaire. Am. J. Hum. Biol. 1.2-24. Fox CA, Ismail AAA, Love DN, Kirkham KN, and Loraine JA (1972) Studies on the relationship between plasma testosterone levels and human sexual activity. J. Endocrinol. 5 2 5 - 5 8 . George FW, Griffin J E , Leshin M, and Wilson JD (1981) Endocrine control of sexual differentiation in the human. In MJ Novy and JA Resko (eds.): Fetal Endocrinology. New York: Academic Press, pp. 341-357. Geschwind N and Behan P (1982) Left-handedness: Association with immune disease, migraine, and developmental learning disorder. Proc. Nat. Acad. Sci. USA 79t5097-5100. Geschwind N and Galaburda AM (1985) Cerebral lateralization. Biological mechanisms, association and pathology: A hypothesis and a program for research. 1-111. Arch. Neurol. 43:428459,521-552,634-654. Gould ES (1948) A topographic study of the differentiation of the dermatoglyphics in the human fetus. Doctoral dissertation, Tulane University, New Orleans. Hale AR (1949) Breadth of epidermal ridges in the human fetus and its relation to the growth of the hand and foot. Anat. Rec. 105:763-776. 197 Hale AR (1952) Morphogenesis of volar skin in the human fetus. Am. J. Anat. 91t147-173. Holt SB (1956) Genetics of dermal ridges: Parent-child correlations for finger ridge count. Ann. Hum. Genet. 20:270-281. James WH (1987a) The human sex ratio. Part 1: A review of the literature. Hum. Biol. 59t721-752. James WH (198713) The human sex ratio. Part 2: A hypothesis and a program for research. Hum. Biol. 59,873-900. Jantz RL and Webb RS (1980)Dermatoglyphic asymmetry as a measure of canalization. Ann. Hum. Biol. 7:489-493. Lewis JG, Ghanadian R, and Chisolm CD (1976) Serum 5-alpha-dihydrotestosteroneand testosterone change with age in man (sic). Acta Endocrinol. (Copenh) 82:444-448. Lipson SF and Ellison PT (1989) Development of prototypes for the application of salivary steroid analyses to field conditions. Am. J. Hum. Biol. 1t249-255. Livshits G and Kobyliansky E (1987) Dermatoglyphic traits as possible markers of developmental processes in humans. Am. J. Med. Genet. 26t111-122. Livshits G and Kobyliansky E (1991)Fluctuating asymmetry as a possible measure of developmetal homeostasis in humans: A review. Hum. Biol. 63r441-466. Loesch D (1983) Quantitative Dermatoglyphics: Classification, Genetics, and Pathology. Oxford: Oxford University Press. Loesch DZ (1986) Bivariate and multivariate analysis of the skin ridge pattern intensities. Am. J. Phys. Anthropol. 69t287-298. Loesch D, Lisiewicz H, and Czyzewska J (1978) Genetical distance and dermatoglyphic characters. 111. Dermatoglyphic distances within twin pairs, between left and right sides, and between normals and 21-trisomics. Hum. Hered. 28t456-467. Lowrey GH (1986) Growth and Development of Children. 8th ed. Chicago: Yearbook Publishers. MacLusky N J and Naftolin F (1981) Sexual differentiation of the central nervous system. Science 211;12941303. Magrini G, Chiodoni G, Rey F, and Felber J P (1986) Further evidence for the usefulness of the salivary testosterone radioimmunassay in the assessment of adrogenicity in man and basal and stimulated conditions. Worm. Res. 23t65-73. McEwen BS (1981) Neural gonadal steroid actions. Science 2llt1303-1311. Meier R J (1990) Dermatoglyphics and the Geschwind Hypothesis. 11:Digital results of dyslexia and developmental implications. In N Durham and C Plato (eds.): Trends in Dermatoglyphic Research. Boston: Kluwer, pp. 114-122. Meier RJ, Sorenson Goodson C, and Roche EM (1987) Dermatoglyphic development and timing of maturation. Hum. Biol. 59:357-373. Meier RJ, Sorenson Jamison C, and Jamison P (1992) Experimental prenatal testosterone effect on Rhesus (Mucuca muluttul dermatoglyphic development. 61st annual meeting of the American Association of Physical Anthropologists. 198 C . SORENSON JMISON ET AL Meikle AW, Stanish WM, Taylor N, Edwards CQ, and Bishop CT (1982) Familial effects on plasma sex-steroid content in Man (sic): Testosterone, estradiol and sex-hormone-bindingglobulin. Metabolism 31 :6-9. Meikle AW, Stringham JD, Bishop DT, and West DW (1988) Quantitating genetic and nongenetic factors influencing androgen production and clearance rates in men. J . Clin. Endocrinol. Metab. 67rlO4-109. Mulvihill JJ and Smith DW (1969) The genesis of dermatoglyphics. J . Pediatr. 75t579-589. Pirke KN and Doerr P (1973) Age related changes and interrelationships between plasma testosterone, oestradiol, and testosterone-binding globulin in normal adult males. Acta Endocrinol. 74t792-800. Purifoy FE, Koopmans LH, and Mayes DM (1981) Age differences in serum androgen levels in normal adult males. Hum. Biol. 53:499-511. Qazi QH, Masakawa A, McGann B, and Woods J (1980) Dermatoglyphic abnormalities in the Fetal Alcohol Syndrome. Teratology 21:157-160. Read GF and Walker RF (1982) Variation of salivary testosterone with age in men. In GF Read, D RiadFahmy, RF Walker, and K Griffiths (eds.):Immunoassays of Steroids in Saliva. Cardiff: Alpha Omega, pp. 215-21 8. Read GF and Walker RF (1984) Variation of salivary testosterone with age in men. In GF Read (ed.): Ninth Tenovus Workshop: Immunoassay of Steroids in Saliva. Cardiff: Alpha Omega, pp. 215-220. Reed T, Uchida IA, Norton JA, and Christian J C (1977) Sources of information discriminating MZ and DZ twins by dermatoglyphic patterns. Acta Genet. Med. Gemellol. 26:83-86. Reed T, Evans MM, Norton J A Jr, and Christian J C (1979)Maternal effects on fingertip dermatoglyphics. Am. J. Hum. Genet. 31:315-323. Rivarola MA, Forest MG, and Migeon CJ (1968) Testosterone, androstenedione, and dehydrepiandrosterone in plasma during pregnancy and at delivery: Concentration and protein binding. J. Clin. Endocrinol. Metab. 28:34-40. Robertson-Neufeld L and Murray J (1978) An inkless method of recording dermal ridges. In J Mavalwala (ed.):Dermatoglyphics:An International Perspective. The Hague: Mouton, pp. 3-9. Rosa P (1985) Associations between dermatoglyphic variation, topography, and climate in Kenya. Am. J. Phys. Anthropol. 68r395-408. Rose LI, Gabbe SG, Teicholz LE, Villee DB, and Williams GH (1972) Dermatoglyphics associated with fetal wastage. N. Engl. J. Med. 287:451-452. Rotter JI, Wong L, Lifrak ET, and Parker LN (1985) A genetic component to the variation of dehydroepiandrosterone sulfate. Metabolism 34:731-736. Schurmeyer T and Nieschlag E (1982) Salivary and serum testosterone under physiological conditions. In GF Read, D Riad-Fahmy, RF Walker, and K Griffiths (eds.): Immunoassays of Steroids in Saliva. Cardiff Alpha Omega, pp. 203-209. Sorenson Jamison C (1988) Palmar dermatoglyphics of dyslexia. Am. J. Phys. Anthropol. 76:505-513. Sorenson Jamison C (1990) Dermatoglyphics and the Geschwind Hypothesis. I. Theoretical background and the palmar results of Dyslexia. In N Durham and C Plato feds.): Trends in Dermatoglyphic Research. Boston: Kluwer, pp. 99-113. Tan U (1990) Relation of testosterone and hand preference in right-handed young adults to sex and familial sinistrality. Int. J. Neurosci. 53:157-165. Tanner JM (1990) Foetus into Man (sic): Physical Growth from Conception to Maturity. Revised and Enlarged Edition. Cambridge: Harvard. Walker RF, Riad-Fahmy D, and Read GF (1980a) Assessment of testicular function by the radioimmunoassay of testosterone in saliva. Int. J . Androl. 3334-40. Walker RF, Wilson DW, Read GF, and Riad-Fahmy D (1980b) Assessment of testicular function by the radioimmunoassay of testosterone in saliva. Int. J. Androl. 3:105-112. WangC, Plymate S, NieschlagE, andPaulsenCA(1981) Salivary testosterone in men: Further evidence of a direct correlation with free serum testosterone. J. Clin. Endocrinol. Metab. 53:1021-1024. Wilson JD, George FW, and Griffin J E (1981) The hormonal control of sexual development. Science 21k1278-1284. Winter JSD (1987) Sexual differentiation. In P Felig, JD Baxter, AE Broadus, and LA Frohman (eds.):Endocrinology and Metabolism. 2nd ed. New York: McGrawHill, pp. 18-54. Wood RI, Ebling FJ, I'Anson H, Bucholtz DC, Yellon SM, and Foster DL (1991) Prenatal androgens time neuroendocrine sexual maturation. Endocrinology 128r2457-2468. Wright HT, Parker CE, and Mavalwala J (1972) Unusual dermatoglyphic findings associated with Cytomegliac Inclusion Disease of infancy. Cal. Med. 116:14-20. Zaaijer JJP and Price D (1971) Early secretion of androgen hormones by human fetal gonads and adrenal glands in organ culture and possible implications for sex differentiation. In M Hamburgh and WJW Barrington (eds.): Hormones in Development. New York: Appleton-Century-Crofts,pp. 537-546.