AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 140:376–380 (2009) Brief Communication: Familial Resemblance in Digit Ratio (2D:4D) Martin Voracek* and Stefan G. Dressler Department of Basic Psychological Research, School of Psychology, University of Vienna, Vienna A-1010, Austria KEY WORDS digit ratio (2D:4D); heritability; family study; sex differences; prenatal testosterone ABSTRACT Familial resemblance in the secondto-fourth digit ratio (2D:4D), a proxy for prenatal androgen action, was studied in 1,260 individuals from 235 Austrian families. In agreement with ﬁndings from twin studies of 2D:4D, heritability estimates based on parent– child and full-sib dyad similarity indicated substantial genetic contributions to trait expression (57% for right hand, 48% for left hand 2D:4D). Because twin studies have found nonadditive genetic as well as shared environmental effects on 2D:4D to be negligible or nil, these family-based estimates in all likelihood reﬂect the narrow-sense (additive genetic) heritability of the trait. Directional (right-minus-left) asymmetry in 2D:4D was only weakly heritable (6%). The pattern of same-sex and different-sex parent–child and full-sib correlations yielded no evidence for X-linked inheritance. This is surprising, considering evidence for associations of male 2D:4D with sensitivity to testosterone (functional variants of the X-linked androgen receptor gene). 2D:4D was particularly strongly heritable through male lines (father–son and brother–brother correlations), thus raising the possibility that Y-linked genes (such as the sex-determining region SRY) might inﬂuence 2D:4D expression. Am J Phys Anthropol 140:376–380, 2009. V 2009 Wiley-Liss, Inc. Men on average display a lower second-to-fourth digit ratio (2D:4D) than women (Manning, 2002). This anatomical sex difference, rediscovered by Manning et al. (1998), generalizes to the digit ratios of a variety of vertebrate non-human species (reviews: Voracek, 2006; Lombardo and Thorpe, 2008). Individual and sex differences in human 2D:4D emerge prenatally (Malas et al., 2006; Galis et al., in press) and are fairly stable during postnatal growth (Trivers et al., 2006). 2D:4D studies have recently become increasingly popular in human biology, medicine, psychology, and the intersections of these ﬁelds (McIntyre, 2006; Voracek and Loibl, in press). Speciﬁcally, 2D:4D has been proposed as an indirect marker for prenatal androgen action and the organizing (permanent) masculinization effects of this on the brain, behavior, and physique (Manning, 2002). If this prenatal-androgen-marker hypothesis is correct, studying within-sex associations of 2D:4D with target variables would enable straightforward (noninvasive and thus easily implemented) tests of the biological bases of traits and phenotypes conceivably partly inﬂuenced by early sex-hormonal effects. There is now various circumstantial, but persuasive, evidence in favor of this hypothesis (review: Voracek et al., 2007a). More direct evidence includes the ﬁnding that the testosterone-to-estradiol ratio in the amniotic ﬂuid is a negative correlate of children’s 2D:4D (Lutchmaya et al., 2004). In a similar vein, lower (masculinized) 2D:4D in men is associated with higher sensitivity to testosterone, through the inﬂuence of a functional polymorphism in the androgen receptor gene (Manning et al., 2003). Because this gene resides on the X chromosome, one expectation is that 2D:4D should show signs of sex-inﬂuenced (X-linked) inheritance. This has indeed been suggested in a forerunner study of modern 2D:4D research (Phelps, 1952), which conjectured an X-linked modifying factor (dominant in men, recessive in women) for a short index ﬁnger (2D). However, another classic family study (Ramesh and Murty, 1977) found no evidence for X-linked inheritance. In hindsight, these two early accounts are open to several points of criticism, e.g., both studies did not actually analyze 2D:4D, but rather distal ﬁnger-extent difference, and the locale of the second study was within an endogamous group, which may well have affected inheritance patterns. Also, the second study found no evidence for sex differences in distal ﬁnger-extent difference, which is known to be sexually differentiated, thereby casting doubt over study conclusions. More generally, family studies cannot disentangle genetic from environmental effects. That is, family resemblance on a trait results from genetic or environmental similarity or a mixture of both. However, there are now four twin studies of 2D:4D in humans (United Kingdom: Paul et al., 2006; Austria: Voracek and Dressler, 2007; USA: Gobrogge et al., 2008; Australia: Medland and Loehlin, 2008). Despite great differences in methodology and in size, recruitment, provenance, and characteristics of samples, their key ﬁndings converge on the following important points: additive genetic factors contribute substantially to 2D:4D variation, i.e., the trait is highly heritable (across studies: h2 5 50–80%); nonshared environmental factors are also signiﬁcant, but smaller (20–50%); whereas shared environmental C 2009 V WILEY-LISS, INC. C *Correspondence to: Martin Voracek, Department of Basic Psychological Research, School of Psychology, University of Vienna, Liebiggasse 5, Rm 03-46, Vienna A-1010, Austria. E-mail: email@example.com Received 17 November 2008; accepted 17 April 2009 DOI 10.1002/ajpa.21105 Published online 15 June 2009 in Wiley InterScience (www.interscience.wiley.com). 377 DIGIT RATIO (2D:4D) FAMILY STUDY factors are not signiﬁcant and much smaller or nil (0–20%). Substantial heritability of 2D:4D (h2 5 60%) was also indicated in an additional small-sample study of monozygotic lesbian twins (Hall and Love, 2003). One study (Medland and Loehlin, 2008) explicitly tested for nonadditive genetic factors, i.e., dominance and epistasis effects (allele–allele and gene–gene interactions), and found these to be nil. Of note, 2D:4D may well be similarly highly heritable in other species, as indicated by two studies of zebra ﬁnches (h2 5 70–80%; Forstmeier, 2005; Forstmeier et al., 2008). The twin study ﬁndings summarized here have two important consequences that simplify the interpretation of family studies of 2D:4D. First, because shared environmental effects apparently are negligible or nil, familial resemblance in 2D:4D will reﬂect genetic similarity (conversely, familial dissimilarity will reﬂect nonshared environmental effects). Second, because nonadditive genetic effects also are nil, h2 estimates derived from family data will reﬂect narrow-sense heritability (additive genetic effects only) rather than broad-sense heritability (additive plus nonadditive genetic effects). Given these speciﬁc knowns, family studies appear a viable, informative approach to further elucidate inheritance patterns of 2D:4D. Family studies can also help to determine whether 2D:4D is a sex-linked trait or not. Twin study ﬁndings have been inconsistent regarding this question: although the largest study (Medland and Loehlin, 2008) found no evidence for it, a smaller one (Gobrogge et al., 2008) observed stronger shared environmental effects for male than for female twins. Because male relatives share their Y chromosome, this could reﬂect Y-linked genetic contributions to 2D:4D variation (such as via the sexdetermining region SRY), because in the classic twin study design such effects are falsely attributed to shared environmental factors among males. To recap, given the suggestive ﬁndings of Phelps (1952) and associations of 2D:4D with variants of the X-linked androgen receptor gene, sex-linked inheritance of 2D:4D is expected, but the ﬁndings from twin studies on this matter are contradictory. Given the other (speciﬁc and convergent) ﬁndings of 2D:4D twin studies discussed earlier, the family study design appears informative in this context. So far, three studies have reported family data for 2D:4D, all of them concluding that family resemblance for 2D:4D is substantial. These ﬁndings were based on 41 father–child and 64 mother– child dyads (Marshall, 2000), 62 parent–child dyads (with the children affected by autism-spectrum disorders; Manning et al., 2001), and 88 mother–child dyads (Manning, 2002, pp. 12–13). A 2D:4D family study utilizing a larger sample from the general population is unavailable. Therefore, this study set out to look for signs of X-linked or Y-linked inheritance of 2D:4D in a larger family sample. MATERIALS AND METHODS Participants were from 235 Austrian families (556 males, 704 females). Because there is evidence for significant variation of 2D:4D across macro-ethnic groups (Manning et al., 2007), White Caucasians were eligible for study participation. Several research assistants recruited families from the general population opportunistically, using snowball-sampling techniques (i.e., per- sonal contacts, referral, and word-of-mouth advertizing). Study participation was mainly solicited in the capital city of Vienna, but also covered the surrounding Eastern Austrian areas. Data were collected outside academia or speciﬁc university settings, so families representing a variety of urban and rural living backgrounds as well as diverse educational levels and occupations were included. Ages in the sample ranged from 2 to 97 years (M 5 37.8, SD 5 19.8, lower and upper quartiles: 21 and 52 years). Spousal data from this sample have been used to test for assortative mating on 2D:4D (Voracek et al., 2007a). Palmar-view photocopies of subjects’ right and left hands were presented in random order to three experienced, mutually blind investigators, who measured ﬁnger lengths from the ﬂexion crease proximal-most to the palm to the ﬁngertip with digital vernier calipers accurate to 0.01 mm. Finger-length measurements were averaged prior to calculating digit ratios. Interobserver measurement repeatabilities (assessed with intraclass correlations coefﬁcients, ICC; two-way mixed-effects model with absolute-agreement deﬁnition; Voracek et al., 2007b) were found to be in good order (all P \ 0.001): ICCs were 0.996 (right-hand 2D), 0.997 (right-hand 4D), 0.995 (left-hand 2D), 0.997 (left-hand 4D), 0.913 (righthand 2D:4D), 0.905 (left-hand 2D:4D), and 0.769 (righthand minus left-hand 2D:4D or DR-L). Following standard practice of family studies (Ramesh and Murty, 1977), heritability estimates were obtained by doubling the Pearson correlation coefﬁcients (r) calculated to quantify parent–child and full-sib resemblance. For calculating the correlations, the full dyadic information extractable from the speciﬁc family constellation was utilized, e.g., a ﬁve-member family comprised of the couple, one son, and two daughters contributed six data points to the parent–child analysis, three data points each to the father–child, mother–child, and full-sib analyses, two each to the father–daughter, mother–daughter, and brother–sister analyses, and one each to the father– son, mother–son, and sister–sister analyses. The resulting heritability estimates are unaffected by potential measurement-method effects on 2D:4D (such as measuring ﬁngers from photocopies rather than directly; see Manning et al., 2005), since r is invariant under linear transformations (e.g., through measurement-speciﬁc bias) of the variables from which it is computed. RESULTS Males had lower right-hand digit ratios (R2D:4D) than females (males: M 5 0.957, SD 5 0.035; females: M 5 0.977, SD 5 0.035; t1258 5 29.75, two-tailed P \ 0.001, d 5 20.57), lower left-hand digit ratios (L2D:4D; males: M 5 0.962, SD 5 0.035; females: M 5 0.976, SD 5 0.034; t1258 5 27.09, P \ 0.001, d 5 20.41), and also lower DR-L (males: M 5 20.0048, SD 5 0.0298; females: M 5 0.0007, SD 5 0.0297; t1258 5 23.24, P 5 0.001, d 5 20.18). For both sexes, R2D:4D and L2D:4D were comparably strongly positively correlated (r 5 0.646 and 0.629 for males and females; both P \ 0.001). Consistent with the literature (Manning, 2002), sex differences in 2D:4D were of medium size, the effect was more pronounced for the right hand than for the left, and 2D:4D of both hands corresponded strongly. Replicating Manning et al. (2007) and in contrast to the null ﬁnding of Putz et al. (2004), a small, but signiﬁcant sex effect was also seen for DR-L. American Journal of Physical Anthropology 378 M. VORACEK AND S.G. DRESSLER DISCUSSION TABLE 1. Familial resemblance in 2D:4D Pearson r Relationship type (dyad) n (dyads) R2D:4D L2D:4D DR-L Parent–child Father–child (Fa-Ch) Father–son (Fa-So) Father–daughter (Fa-Da) Mother–child (Mo-Ch) Mother–son (Mo-So) Mother–daughter (Mo-Da) Full sibs Brother–brother (Br-Br) Sister–sister (Si-Si) Brother–sister (Br-Si) 1145 499 224 275 0.292***a 0.312***a 0.363***a 0.276*** 0.232***b 0.222***b 0.305***b,c 0.131*b} 0.027 0.042 0.079 0.015 646 0.299***a 0.245***b a b 291 355 0.328*** 0.293*** 549 104 0.245*** 0.267***b 0.408***} 0.543***c 177 268 0.246** 0.148*} 0.194** 0.299***b,c 0.217**b 0.158*b} 0.019 0.035 0.003 0.017 0.195* 20.048 0.014 Same-column comparisons: aR2D:4D column entries signiﬁcantly different (P \ 0.05) to lowest (brother–sister) correlation. b L2D:4D entries different to highest (brother–brother) correlation. c L2D:4D entries different to lowest (father–daughter) correlation; all other differences not signiﬁcant. Same-row comparisons: } Not signiﬁcantly different from correlation for DR-L. * P \ 0.05, ** P \ 0.01, *** P \ 0.001 (two-tailed). For R2D:4D and L2D:4D, familial resemblance was signiﬁcant for all investigated types of parent–child and full-sib dyads (Table 1). Correlations were numerically (but never signiﬁcantly) larger for R2D:4D than for L2D:4D in 7 of 11 comparisons, and signiﬁcantly smaller for DR-L than for R2D:4D or L2D:4D in 18 of 22 comparisons. X-linked inheritance would be indicated through rSi-Si [ rBr-Br [ rBr-Si and rFa-Da 5 rMo-So [ rMo-Da [ rFa-So, with the theoretically expected values being 0.75 [ 0.50 [ 0.35 and 0.71 5 0.71 [ 0.50 [ 0 under additive sex-linked genes (Ramesh and Murty, 1977). Of note, the current data did not at all conform with this pattern (tested with difference of correlation tests; Steiger, 1980). For L2D:4D, the succession of correlation coefﬁcients was rBr-Br [ [rSi-Si rBr-Si], i.e., the latter two correlations were commensurable, and both of them signiﬁcantly smaller than rBr-Br (for R2D:4D, rBr-Br rSi-Si rBr-Si). Further (using the same notation), rFa-So rMo-So rMo-Da rFa-Da (for R2D:4D), but [rFa-So rMo-Da] [ [rMo-So rFa-Da] for L2D:4D. Among parent– child dyads, the strongest resemblance was between fathers and sons, and for full-sib dyads, among brothers. For DR-L, only brother–brother resemblance was signiﬁcant. The heritability estimates (weighted average of doubled correlation coefﬁcients across all types of dyads) were 57.4%, 47.7%, and 5.8% for R2D:4D, L2D:4D, and DR-L. Supplemental analyses (details omitted) indicated the following: ﬁrst, controls for age or nonparametric analysis (rank-order correlation) did not materially alter these results. And second, controls for maternal (versus paternal) digit ratio (in analysis of covariance models) did not differentially (or signiﬁcantly) alter the magnitude of the sex difference observed in digit ratios across brother-sister dyads, which might imply that sexchromosomal loci do not substantially contribute to sex differences in digit ratios. American Journal of Physical Anthropology The heritability estimates from this 2D:4D family study were very similar to the result of a classic account (Ramesh and Murty, 1977: weighted average h2 5 57%). They also tallied to h2 values calculated from parent– child dyads found in the literature (Marshall, 2000: h2 5 41%; Manning et al., 2001: h2 5 58%; Manning, 2002: h2 5 69%). Familial resemblance in 2D:4D was substantial (concordant with twin study ﬁndings), slightly higher for R2D:4D than for L2D:4D, but largely absent for DR-L. The slightly higher heritability estimate for R2D:4D, as compared with L2D:4D, is consistent with evidence that L2D:4D seems to be more sensitive to environmental factors than R2D:4D (Flegr et al., 2008). DR-L, the directional asymmetry in digit ratios, is considered as an additional pointer to prenatal androgen exposure (Manning, 2002, pp. 21–22) and has especially proven useful in digit ratio studies about sports performance (reviews: Voracek et al., 2006; Bescós et al., 2009). Being another complex phenotype, it may well be affected by additional factors beyond those that inﬂuence digit ratios. Therefore, it is perhaps not surprising that results for DR-L diverged. Alternatively or additionally, the lower repeatability of DR-L may have reduced the strength of its dyad correlations and, as a consequence, its calculated heritability. This latter argument is deducible as follows (see Voracek et al., 2007b): higher random measurement error (i.e., ‘‘noise’’) leads to lower repeatabilities. Random measurement errors mutually cancel out, when calculating averages of measurements; but they do not cancel out, when calculating differences; and they multiply, when calculating ratios. Hence measurement repeatabilities for 2D:4D (a ratio) are lower than for ﬁnger lengths, and still lower for DR-L (the difference of two ratios). On the other side, assortative mating on a trait will inﬂate family-based heritability estimates by the factor 1 1 rAM, where rAM is the spousal correlation on that trait. There still is a paucity of studies on assortative mating on 2D:4D. The effect seems rather small (Manning, 2002, p. 50: rAM 5 0.15 or less; Voracek et al., 2007a: about 0.20 or less), and it is not known whether it is partly due to social homogamy (i.e., milieu similarity, leading to geographic ‘‘background correlations’’). Assuming rAM 5 0.15, this decreases the current heritability estimates (57.4%, 47.7%, and 5.8% for R2D:4D, L2D:4D, and DR-L, respectively) to 49.9%, 41.5%, and 5.0% (assuming rAM 5 0.20, to 47.8%, 39.8%, and 4.8%). At any rate, owing to the lack of strength of the assortative-mating effect, the impact of these downward corrections is modest and thus largely preserves the substantial levels of heritability observed for R2D:4D and L2D:4D. Although the minimum effective sample size for the types of dyads investigated here (N 5 104 for the brother–brother correlations; see Table 1) was larger than the effective sample sizes of the three prior family studies of 2D:4D (see Introduction), it is also true that sample sizes available for analysis of the different types of dyads varied greatly. This is an inherent characteristic of family study designs and could have impacted results. On the other side, although 2D:4D varies signiﬁcantly across macro-ethnic groups (Manning et al., 2007), the existing evidence supports the idea that genetic differences between populations rather than environmental factors contribute signiﬁcantly to observed population differences in 2D:4D (Loehlin et al., 2006). Since the present study’s family sample was DIGIT RATIO (2D:4D) FAMILY STUDY ethnically homogeneous, population differences in 2D:4D could not have inﬂuenced the ﬁndings. Heritability estimates from combinations of full-sib dyads were not larger than those from combinations of parent–child dyads. This implies absence of dominance deviation effects (i.e., no contributions of recessive genes), absence of epistasis effects (i.e., no interloci interactions), and absence of shared environmental effects (in the case of 2D:4D, evidently prenatal ones, i.e., no maternal or womb environment inﬂuences). These conclusions also concur with the twin study ﬁndings. Replicating Ramesh and Murty (1977), evaluation of the rank order of h2 estimates for same-sex and different-sex parent–child and full–sib dyads yielded no evidence for X-linked trait inheritance. This suggests absence of major genes for 2D:4D variation on this sex chromosome. By implication, effects of variants of the X-linked androgen receptor gene on 2D:4D presumably are small, and, in turn, sensitivity to testosterone may matter less than testosterone exposure levels. However, familiality patterns of 2D:4D did not match those of circulating testosterone. Although twin studies evidence that testosterone levels are more heritable among men than women (h2 5 66% vs. 41%), corresponding parent– offspring correlations (including those of fathers and sons) are nil, with the exception of a moderate mother– daughter resemblance (Harris et al., 1998). Furthermore, although the within-sex variation in testosterone levels is several times larger in males than in females, both prenatally (Finegan et al., 1989) and postnatally (Vermeersch et al., 2008), there is no such sex difference in the variance of 2D:4D (see Results). Intriguingly, the current data suggest particularly strong heritability of 2D:4D through male lines, as both R2D:4D and L2D:4D were most similar for male–male dyads (replicating Marshall, 2000), and the only signiﬁcant family correlation for DR-L was for brothers. These ﬁndings raise the possibility that Y-linked genes (e.g., the sex-determining region SRY, as suggested by Gobrogge et al., 2008) might also inﬂuence digit ratio expression. 2D:4D comparisons of sex-chromosome aberrations with normal karyotypes (47,XXY and 47,XYY men with 46,XY men; 45,XO and 47,XXX women with 46,XX women; and corresponding animal models), halfsib designs, whole-genome scans, and studying the genetics of individual differences in sex-hormone action will be ﬁtting methodological approaches to investigate these matters further. The exact genetics of 2D:4D is not yet known, remains to be pinpointed, and in all likelihood is complex. Ultimately, it would be of great interest to undertake genome scans for 2D:4D, as has been recently done to elucidate the genetics of ﬁnger-ridge counts (Medland et al., 2007). The exact nature of the nonshared environmental factors inﬂuencing the expression of 2D:4D also remains to be elucidated. Because sex and individual differences in 2D:4D emerge prenatally and appear rather stable postnatally, the nonshared environmental inﬂuences must be searched in the prenatal environment. In addition, 2D:4D research might also beneﬁt from adopting within-family study designs (e.g., comparisons of related individuals discordant for a target trait; for a successful application, see Hall and Love, 2003), to control for the apparently substantial genetic effects on 2D:4D which in studies of unrelated individuals could obfuscate or attenuate real associations of 2D:4D with target traits (Forstmeier et al., 2008). 379 LITERATURE CITED Bescós R, Esteve M, Porta J, Mateu M, Irurtia A, Voracek M. 2009. Prenatal programming of sporting success: associations of digit ratio (2D:4D), a putative marker for prenatal androgen action, with world rankings in female fencers. J Sports Sci 27:625–632. Finegan JA, Bartleman B, Wong PY. 1989. A window for the study of prenatal sex hormone inﬂuences on postnatal development. J Genet Psychol 150:101–112. Flegr J, Lindová J, Pivonková V, Havlı́ček J. 2008. Latent toxoplasmosis and salivary testosterone concentration: important confounding factors in second to fourth digit ratio studies. Am J Phys Anthropol 137:479–484. Forstmeier W. 2005. Quantitative genetics and behavioural correlates of digit ratio in the zebra ﬁnch. Proc R Soc B 272: 2641–2649. Forstmeier W, Rochester J, Millam JR. 2008. Digit ratio unaffected by estradiol treatment of zebra ﬁnch nestlings. Gen Comp Endocrinol 156:379–384. Galis F, Ten Broek CMA, Van Dongen S, Wijnaendts LCD. Sexual dimorphism in the prenatal digit ratio (2D:4D). Arch Sex Behav (in press). Gobrogge KL, Breedlove SM, Klump KL. 2008. Genetic and environmental inﬂuences on 2D:4D ﬁnger length ratios: a study of monozygotic and dizygotic male and female twins. Arch Sex Behav 37:112–118. Hall LS, Love CT. 2003. Finger-length ratios in female monozygotic twins discordant for sexual orientation. Arch Sex Behav 32:23–28. Harris JA, Vernon PA, Boomsma DI. 1998. The heritability of testosterone: a study of Dutch adolescent twins and their parents. Behav Genet 28:165–171. Loehlin JC, McFadden D, Medland SE, Martin NG. 2006. Population differences in ﬁnger-length ratios: ethnicity or latitude? Arch Sex Behav 35:739–742. Lombardo MP, Thorpe PA. 2008. Digit ratios in green anolis lizards (Anolis carolinensis). Anat Rec 291:433–440. Lutchmaya S, Baron-Cohen S, Raggatt P, Knickmeyer R, Manning JT. 2004. 2nd to 4th digit ratios, fetal testosterone and estradiol. Early Hum Dev 77:23–28. Malas MA, Dogan S, Evcil EH, Desdicioglu K. 2006. Fetal development of the hand, digits and digit ratio (2D:4D). Early Hum Dev 82:469–475. Manning JT. 2002. Digit ratio: a pointer to fertility, behavior, and health. New Brunswick (NJ): Rutgers University Press. Manning JT, Baron-Cohen S, Wheelwright S, Sanders G. 2001. The 2nd to 4th digit ratio and autism. Dev Med Child Neurol 43:160–164. Manning JT, Bundred PE, Newton DJ, Flanagan BF. 2003. The second to fourth digit ratio and variation in the androgen receptor gene. Evol Hum Behav 24:399–405. Manning JT, Churchill AJG, Peters M. 2007. The effects of sex, ethnicity, and sexual orientation on self-measured digit ratio (2D:4D). Arch Sex Behav 36:223–233. Manning JT, Fink B, Neave N, Caswell N. 2005. Photocopies yield lower digit ratios (2D:4D) than direct ﬁnger measurements. Arch Sex Behav 34:329–333. Manning JT, Scutt D, Wilson J, Lewis-Jones DI. 1998. The ratio of 2nd to 4th digit length: a predictor of sperm numbers and concentrations of testosterone, luteinizing hormone and oestrogen. Hum Reprod 13:3000–3004. Marshall D. 2000. A study estimating the heritability of 2nd to 4th digit ratio in humans. Unpublished Master’s thesis, University of Liverpool. McIntyre MH. 2006. The use of digit ratios as markers for perinatal androgen action. Reprod Biol Endocrinol 4:10. Medland SE, Loesch DZ, Mdzewski B, Zhu G, Montgomery GW, Martin NG. 2007. Linkage analysis of a model quantitative trait in humans: ﬁnger ridge count shows signiﬁcant multivariate linkage to 5q14.1. PloS Genet 3:1736–1744. Medland SM, Loehlin JC. 2008. Multivariate genetic analyses of the 2D:4D ratio: examining the effects of hand and measurement technique in data from 757 twin families. Twin Res Hum Gen 11:335–341. American Journal of Physical Anthropology 380 M. VORACEK AND S.G. DRESSLER Paul SN, Kato BS, Cherkas LF, Andrew T, Spector TD. 2006. Heritability of the second to fourth digit ratio (2d:4d): a twin study. Twin Res Hum Genet 9:215–219. Phelps VR. 1952. Relative index ﬁnger length as a sex-inﬂuenced trait in man. Am J Hum Genet 4:72–89. Putz DA, Gaulin SJC, Sporter RJ, McBurney DH. 2004. Sex hormones and ﬁnger length: what does 2D:4D indicate? Evol Hum Behav 25:182–199. Ramesh A, Murty JS. 1977. Variation and inheritance of relative length of index ﬁnger in man. Ann Hum Biol 4:479– 484. Steiger JH. 1980. Tests for comparing elements of a correlation matrix. Psychol Bull 87:245–251. Trivers R, Manning JT, Jacobson A. 2006. A longitudinal study of digit ratio (2D:4D) and other ﬁnger ratios in Jamaican children. Horm Behav 49:150–156. Vermeersch H, T’Sjoen G, Kaufman JM, Vincke J. 2008. 2D:4D, sex steroid hormones and human psychological sex differences. Horm Behav 54:340–346. American Journal of Physical Anthropology Voracek M. 2006. Of mice and men–cross-species digit ratio (2D:4D) research: comment on Bailey. Wahlsten, and Hurd (2005). Genes Brain Behav 5:299–300. Voracek M, Dressler SG. 2007. Digit ratio (2D:4D) in twins: heritability estimates and evidence for a masculinized trait expression in women from opposite-sex pairs. Psychol Rep 100:115–126. Voracek M, Dressler SG, Manning JT. 2007a. Evidence for assortative mating on digit ratio (2D:4D), a biomarker for prenatal androgen exposure. J Biosoc Sci 39:599–612. Voracek M, Loibl LM. Scientometric analysis and bibliography of digit ratio (2D:4D) research, 1998–2008. Psychol Rep (in press). Voracek M, Manning JT, Dressler SG. 2007b. Repeatability and interobserver error of digit ratio (2D:4D) measurements made by experts. Am J Hum Biol 19:142–146. Voracek M, Reimer B, Ertl C, Dressler SG. 2006. Digit ratio (2D:4D), lateral preferences, and performance in fencing. Percept Mot Skills 103:427–446.