Professional Documents
Culture Documents
Are 2D:4D Finger-Length Ratios Related To Sexual Orientation? Yes For Men, No For Women
Are 2D:4D Finger-Length Ratios Related To Sexual Orientation? Yes For Men, No For Women
The ratio of index and ring finger lengths (2D:4D) is thought to be a marker of prenatal androgen
exposure. In a sample of over 2,000 participants, men had significantly lower 2D:4D ratios than women
(d ⫽ .36 and .23 for right and left hands, respectively), and these results were consistent across ethnic
groups. Heterosexual men had significantly lower (more male typical) 2D:4D ratios than gay men (d ⫽
.32 and .31 for right and left hands, respectively), and these results tended to be consistent across ethnic
groups. Heterosexual and lesbian women showed no significant differences in 2D:4D ratios, after
ethnicity was taken into account. The current findings add to evidence that prenatal hormonal factors may
be linked to men’s sexual orientation.
Neurohormonal theories of human sexual orientation have fo- they propose that androgens likely have organizational effects on
cused on possible prenatal effects of sex steroids, particularly sexually dimorphic structures in subcortical regions of the devel-
androgens, on the developing brain (Byne & Parsons, 1993; Ellis oping brain that are responsible for sexual behaviors (e.g., regions
& Ames, 1987; Meyer-Bahlburg, 1984). These theories draw sup- of the hypothalamus; see Byne et al., 2001; LeVay, 1991).
port from animal experimentation showing that prenatal and peri- Because experimental studies on the behavioral effects of pre-
natal sex hormones influence male- and female-typical sexual natal hormones cannot be conducted on humans for obvious eth-
behaviors in a number of mammalian species, including nonhuman ical reasons, researchers have tended instead to rely on correla-
primates (Adkins-Regan, 1988; Breedlove, 1994; Cooke, Heg- tional evidence from individuals who have experienced atypical
strom, Villeneuve, & Breedlove, 1998; Gorski, 1985; Vasey, 1995; levels of hormones or who have been exposed to drugs or to
Wallen, 1996). Typically, animal experiments have demonstrated environments that influence hormones (Collaer & Hines, 1995;
that early exposure to androgens tends to “masculinize” and “de- Lippa, 2002a). The most complete and methodologically sophis-
feminize” sexual behaviors and, presumably, also the brain struc- ticated data come from studies of females with congenital adrenal
tures that underlie these behaviors. Drawing on animal models, hyperplasia (CAH; see Berenbaum, 2002). Because of an enzy-
neurohormonal theories of human sexual orientation propose that matic defect, CAH females’ adrenal glands produce atypically
high prenatal androgen exposure during critical periods of devel- high levels of androgens prenatally, resulting in varying degrees of
opment may be associated with heterosexuality in men and homo- physical and behavioral masculinization. Consistent with the neu-
sexuality in women, whereas low androgen exposure may be rohormonal theory of sexual orientation, CAH women report ele-
associated with homosexuality in men and heterosexuality in vated levels of homosexual fantasy and behavior compared with
women (Bailey, 1995; Ellis & Ames, 1987). non-CAH controls (Berenbaum, 2002; Dittmann, Kappes, &
The period when androgens surge during male fetal develop- Kappes, 1992; Zucker et al., 1996).
ment and presumably trigger male-typical brain development oc- Other conditions that produce atypical prenatal hormone levels
curs from the 7th to 24th weeks of fetal development, with andro- or atypical hormone action include androgen insensitivity syn-
gens peaking at approximately the 18th week (i.e., the middle of drome (which occurs when individuals lack androgen receptors to
the second trimester; Wilson, 1999). Although there may be dif- utilize testosterone), 5␣-reductase deficiency (a genetic enzyme
ferent neural critical periods for various kinds of sex-typed behav- deficiency that prevents testosterone from being converted to
iors (e.g., sex-typed play, aggression, sexual behaviors), neurohor- dihydrotestosterone and that leads genetic males to appear female
monal theories of sexual orientation generally focus on effects or genitally ambiguous at birth), and Turner syndrome (which
from the 2nd to 5th months of gestation (Ellis & Ames, 1987), and occurs when affected individuals have just a single X chromosome
instead of a pair of sex chromosomes; for a review, see Collaer &
Hines, 1995). Because studies of individuals exposed to atypical
I thank the following research assistants who devoted many hours to prenatal hormone levels are correlational in nature, their findings
collecting data, entering data, and measuring thousands of fingers: Michael are necessarily ambiguous, with biological and social effects con-
Cassens, Shelby Davis, Andria Gong, Alicia Gregory, Francine Kirouac, founded. For example, CAH girls are sometimes physically mas-
Ann Litvin, Jonathan Schnaible, and Cynthia Tipps. Special thanks go to
culinized at birth and subject to “corrective” surgery. As a result,
Orange County Parents, Families and Friends of Lesbians and Gays
they may be treated differently from nonaffected girls by their
(PFLAG) for allowing this research project to share PFLAG booths at the
Long Beach, California and Orange County, California gay pride festivals. caretakers. Androgen-insensitive XY individuals appear female
Correspondence concerning this article should be addressed to Richard physically, and thus it is not clear whether their female-typical
A. Lippa, Department of Psychology, California State University, Fuller- behaviors are due to rearing or to lack of early androgen effects.
ton, California 92834. E-mail: [email protected] Because reductase-deficient XY individuals who are reared as girls
179
180 LIPPA
sometimes look genitally ambiguous, their ability to adopt mas- sex hormones, particularly androgens (Manning, 2002; Williams et
culine roles after puberty may reflect socialization differences as al., 2000). Brown, Hines, Fane, and Breedlove (2001) have col-
well as the organization effects of prenatal testosterone resulting lected preliminary data showing that both CAH males and females
from their XY genotype. In addition, Turner syndrome females tend to have smaller (more male-typical) 2D:4D ratios than non-
experience direct effects of their atypical genotype as well as a CAH controls, and this provides new evidence that 2D:4D ratios
lack of sex hormones because of absent or nonfunctioning ovaries. are associated with prenatal androgen levels. Manning (2002)
For both practical and methodological reasons, researchers have reviewed research showing that 2D:4D ratios are correlated with a
not been able to assess prenatal androgen levels directly in humans variety of physiological, behavioral, and psychological traits that
and then trace their relation to later sex-typed social and sexual are plausibly influenced by prenatal sex hormones.
behaviors. However, recent studies have pursued a promising Several recent studies have investigated possible links between
alternate strategy to investigate possible associations between pre- 2D:4D ratios and sexual orientation. In a widely reported study,
natal hormones and adult sex-typed behaviors. By assessing par- Williams et al. (2000) assessed 2D:4D ratios and sexual orientation
ticipants on physiological and behavioral markers that are thought in 720 adults attending a street fair in the San Francisco Bay area.
to be linked to prenatal androgen exposure, these studies investi- These researchers found a significant difference between lesbians’
gate whether markers are associated with sex-typed behaviors and and heterosexual women’s right-hand 2D:4D ratios, with lesbians
traits in adults. If physical markers that are thought to covary with showing lower (more male-typical) ratios. Although gay men and
prenatal androgen exposure also covary with adult sex-typed be- heterosexual men did not show a significant difference in 2D:4D
haviors, then this is seen as evidence that prenatal androgen levels ratios, there was a nonsignificant tendency for gay men to have
are linked to adult sex-typed behaviors. Among recently studied higher (more female-typical) ratios in the left hand. Brown, Finn,
putative markers of early androgen exposure are dermatoglyphics Cooke, and Breedlove (2002) conducted a follow-up study on the
(finger print ridge patterns), handedness, autoacoustic emissions relationship between 2D:4D ratios and women’s sexual orienta-
(click-induced or spontaneous faint sounds generated by the inner tion, again assessing participants at a Bay area street fair. They
ear), waist-to-hip ratios, and index-to-ring-finger (2D:4D) finger- found lesbian– heterosexual differences in 2D:4D ratios for
length ratios. Recent research has found associations between all “butch” but not for “femme” lesbians (with butch lesbians showing
of these markers and human sexual orientation (Hall, 2000; lower 2D:4D ratios), and this finding hints that prenatal androgens
Lalumière, Blanchard, & Zucker, 2000; Manning, 2002; McFad- may be linked to sexual orientation only for certain subgroups of
den, 2002; McFadden & Pasanen, 1998; Mustanski, Bailey, & women.
Kaspar, 2002; Singh, Vidaurri, Zambarano, & Dabbs, 1999; Wil- In a British study of 88 gay men and 88 age-matched hetero-
liams et al., 2000). However, the results have been complex, sexual men, Robinson and Manning (2000) reported that gay men
fragmentary, and at times inconsistent. had lower (more male-typical) left-hand 2D:4D ratios than hetero-
The study reported here focused on possible associations be- sexual men, and they inferred from this that gay men may be
tween 2D:4D finger-length ratios and sexual orientation in men exposed to higher prenatal testosterone levels than heterosexual
and women. The 2D:4D ratio is an easily measured physical men. This finding contradicts perhaps the most straightforward
characteristic that is sexually dimorphic and thought to be linked to version of the neurohormonal theory of sexual orientation—that
prenatal androgen levels. Physical anthropologists have long noted high prenatal androgen levels masculinize sexual behavior and low
that men are more likely than women to have ring fingers rela- levels feminize sexual behavior. Complicating Robinson and Man-
tively longer than their index fingers (i.e., low 2D:4D ratios), ning’s results, however, was the additional finding that exclusively
whereas women are more likely than men to have index fingers homosexual men had higher (more female-typical) 2D:4D ratios
relatively longer than their ring fingers (i.e., high 2D:4D ratios; than nonexclusively homosexual and bisexual men.
Baker, 1888; for a review, see Peters, Mackenzie, & Bryden, Manning (2002) noted that some theories argue that unusually
2002). Manning (2002) reported that sex differences in 2D:4D high prenatal androgen exposure in males can lead to homosexu-
ratios tend to be small to moderate in magnitude (d ⫽ .2–.5), ality (Geschwind & Galaburda, 1985a, 1985b), whereas others
depending on the population studied. have argued just the opposite (Ellis & Ames, 1987). Complicating
Although adult ranges of 2D:4D ratios are present early in fetal matters further, some theories argue that stress-induced testoster-
development (Garn, Burdi, Babler, & Stinson, 1975; Phelps, one surges in one period of fetal development may be followed by
1952), it is not clear when sex differences in 2D:4D ratios first abnormal declines in later periods (e.g., James, 1989), and this may
appear. Robinson and Manning (2000) speculated that 2D:4D produce seemingly paradoxical effects—that is, characteristics as-
ratios are likely fixed by the 14th week of gestation, which is sociated with high testosterone levels in one period (e.g., low
within the critical period when testosterone has organizational 2D:4D ratios) could then sometimes become associated with char-
effects on developing fetuses. After birth, 2D:4D ratios and sex acteristics that are associated with low testosterone levels in an-
differences in 2D:4D ratios appear to be stable over time, and they other critical period (e.g., male homosexuality). Finally, some
seem not to be affected by postnatal variations in hormone levels, theorists have proposed that the effects of prenatal testosterone
including the large variations that occur at puberty (Manning, levels may be nonmonotonic (McFadden, 2002). In the case of
2002). This again suggests that 2D:4D ratios are fixed at an early male sexual orientation, this would imply that up to a certain point,
age, probably prenatally. increased levels of testosterone are masculinizing, but after that
Most physical sex differences that occur early in life (such as point increased levels are feminizing.
genital differences) are thought to result from the action of sex Robinson and Manning’s (2000) study of heterosexual and
hormones, and researchers have therefore inferred that sex differ- homosexual men assessed a relatively ethnically homogeneous
ences in 2D:4D ratios also likely result from the early effects of population (mostly White British and Scottish men), whereas the
FINGER-LENGTH RATIOS 181
Williams et al. (2000) and Brown et al. (2002) studies likely did to be age dependent. Analyses that contrasted younger gay and lesbian
not (because they used convenience samples in ethnically diverse participants (those 30 years old or younger) with heterosexual participants
California). The ethnic background of participants is important yielded similar results to analyses that contrasted all gay and lesbian
because recent research shows that 2D:4D ratios vary substantially participants with heterosexual participants. Thus, participants of all ages
were included in most analyses to maximize statistical power.
across ethnic and nationality groups (Manning, 2002). Ethnic
Ethnic characteristics of participants. Participants were asked to re-
differences in 2D:4D ratios can be much larger than sex or sexual
port their ethnicity (White, Hispanic, Asian, African American, Middle
orientation differences, and therefore ethnic variations have the Eastern American, Native American, Multiracial, or other). Table 1 pre-
potential to generate spurious findings in studies of sexual orien- sents the ethnic breakdown of heterosexual men, heterosexual women,
tation and 2D:4D ratios if the ethnic composition of heterosexual homosexual men, and homosexual women. As Table 1 shows, homosexual
samples differs from that of homosexual samples. samples included a relatively higher percentage of White participants than
Two strategies exist for dealing with the methodological prob- heterosexual samples, whereas heterosexual samples included a relatively
lem posed by large ethnic variations in 2D:4D ratios: (a) Research- higher percentage of Hispanic and Asian participants than homosexual
ers can assess ethnically homogeneous populations of heterosexual samples.
and homosexual participants, or (b) researchers can assess suffi-
ciently large samples to ensure that ethnic variations tend to Photocopying Hands and Measuring Finger Lengths
balance out and are therefore less likely to generate spurious
homosexual– heterosexual differences. The current study used Participants’ hands were photocopied with a Canon PC400 portable
both of these strategies. Because participants were assessed on copy machine. Participants were asked to place their hands palm down on
the glass surface of the photocopier with their fingers held together and the
ethnicity, the current analyses could examine the consistency of
tips of their right and left index fingers and right and left thumbs touching.
heterosexual– homosexual differences across ethnic groups, and
There were a small number of participants who could have only one hand
because large numbers of heterosexual and homosexual partici- photocopied because of injury or disability. Also, a small number of
pants were assessed, participants’ ethnicity and national origins participants whose hands were photocopied had unusable questionnaire
were likely to be quite varied. By investigating the relationship data because of missing questionnaire responses. Photocopies of partici-
between sexual orientation and 2D:4D ratios in large, ethnically pants’ hands were stapled to their completed questionnaires. Subsequently,
diverse groups of heterosexual and homosexual men and women, each questionnaire and hand photocopy pair were labeled with matched
the current study hoped to provide new and methodologically identification numbers and then separated. Researchers who measured
cleaner evidence on possible relationships between 2D:4D ratios finger lengths from photocopies were blind to the gender and sexual
and sexual orientation in both men and women. orientation of participants.
The lengths of index (2D) and ring (4D) fingers were measured in each
hand by at least two independent raters (I served as one rater and student
Method research assistants served as second raters). Using a clear plastic rule, raters
measured finger lengths to the nearest half millimeter from the basal
Participants (lowest) finger crease to the fingertip along the medial line bisecting the
finger. Some fingers were measured by two student research assistants, and
Participants were 849 men and 1,235 women. Most heterosexual partic-
in these cases measured lengths were averaged to form one “research
ipants were college students and staff at California State University,
assistant” measurement.
Fullerton. Most gay and lesbian participants were volunteers solicited at
gay pride festivals in Long Beach, California and Orange County,
California.1 The Reliability of Finger Lengths and 2D:4D
Assessing sexual orientation. Participants completed an anonymous
questionnaire that included a demographic cover sheet, a section that asked To assess the reliability of finger-length measures, the two measures of
about sexual orientation, and a number of personality scales. Sexual finger lengths were correlated. These correlations were .94 for right 2D, .96
orientation was assessed by asking participants to respond “true” or “false” for right 4D, .93 for left 2D, and .96 for left 4D (n ⫽ 2,073–2,077, all ps ⬍
to whether they currently used each of the following labels to describe
themselves: heterosexual (“straight”), gay, lesbian, and bisexual.
1
Men and women were classified as gay or lesbian, respectively, if they One reviewer of this article noted that college students (who made up
checked “true” to using these labels to describe themselves. Participants the huge majority of heterosexual participants in the current study) and gay
were classified as heterosexual if they checked “true” to using the label pride festival attendees (who made up the majority of gay and lesbian
heterosexual to describe themselves and if they also checked “false” to participants) do not constitute representative heterosexual and homosexual
using the label gay (for men) or lesbian (for women) or if they did not samples. The reviewer specifically proposed that “it seems plausible that
respond to these items. Because not all participants responded to sexual reasons for choosing to openly celebrate one’s sexual orientation are
orientation questions and because some participants (e.g., those who la- related to personality factors that, in turn, stem from hormones.” All
beled themselves only as bisexual) were not categorized as heterosexual, participants in the current study were assessed on several measures of
gay, or lesbian, the number of participants in analyses that contrast sexual masculinity and femininity, and college student participants were also
orientation groups is somewhat smaller than the total sample of 2,084 assessed on Big Five personality traits. In general, relations between
participants. 2D:4D finger-length ratios and personality traits were weak and nonsig-
Final analyses included 351 heterosexual men (mean age ⫽ 23.8 years, nificant. Thus, there was little evidence that 2D:4D ratios and, by impli-
median age ⫽ 20.0 years, SD ⫽ 10.69), 461 gay men (mean age ⫽ 40.0 cation, prenatal androgen levels were linked to broad personality dimen-
years, median age ⫽ 36.0 years, SD ⫽ 11.18), 707 heterosexual women sions. These findings do not rule out the possibility that gay men and
(mean age ⫽ 22.9 years, median age ⫽ 19.0 years, SD ⫽ 8.89), and 472 lesbians who participate in gay pride festivals differ in personality from
lesbian women (mean age ⫽ 35.7 years, median age ⫽ 35.0 years, those who do not. However, if attendees and nonattendees differ on the
SD ⫽ 9.98). Although gay and lesbian participants were on average older broad personality traits we assessed, these personality traits seem not to be
than heterosexual participants, none of the key findings that follow proved much associated with 2D:4D ratios.
182 LIPPA
Table 1 lowing sections typically begin with two-tailed t tests and present
Ethnic Composition of Heterosexual Male, Heterosexual group contrasts in terms of effect sizes (Cohen’s d statistic).
Female, Homosexual Male, and Homosexual Female Samples Because multiple measures were obtained from each participant
(Percentages of Whole Sample That Reported Ethnicity) (four finger lengths and right-hand and left-hand 2D:4D ratios),
repeated-measures analyses of variance (ANOVAs) are often pre-
Heterosexual Homosexual sented as secondary analyses to provide more complete analyses of
Ethnicity Male Female Male Female patterns in the data. Finally, because 2D:4D ratios and finger
lengths covaried with ethnicity, ethnic groups are treated as
White 45 42 69 69 between-subjects factors in some analyses. Analyses of ethnicity
Hispanic 22 22 15 15 generally contrasted the three largest broad ethnic groups in the
Asian 17 20 4 2
African American 3 3 3 2
study: Whites, Hispanics, and Asians.
Middle Eastern 2 1 1 1
Native American 0 0 1 2
Multiracial 5 5 4 5 Sex Differences in 2D:4D Ratios and Finger Lengths
Other 6 6 3 4
Table 2 presents sex differences in 2D:4D ratios for all partic-
ipants, White participants, Hispanic participants, and Asian par-
.001). Interrater correlations for computed 2D:4D ratios were .73 ( p ⬍ ticipants.2 In all cases, men had significantly lower 2D:4D ratios
.001) for the left hand (␣ for mean rating ⫽ .84) and .73 for the right hand than women.
( p ⬍ .001; ␣ for mean rating ⫽ .84). For all groups, sex differences in 2D:4D were larger for the right
Because over 8,000 fingers were measured two times, there were likely hand than for the left hand. To determine whether this right-hand-
a small number of measurement, data transcription, and data entry errors. versus-left-hand difference was significant, a 2 ⫻ 2 (sex ⫻ hand)
To identify possible errors, the difference between the two measures of repeated-measures ANOVA was conducted on 2D:4D ratios, with
each finger was computed, and when this difference was greater than or sex of participant serving as a between-subjects factor and right-
equal to 3 mm, I remeasured that finger. The new measurement then
hand versus left-hand a within-subjects factor. Supporting the
replaced the old measurements. This process identified a small number (a
small fraction of 1%) of gross measurement and data entry errors. Because
observation that 2D:4D sex differences were larger in the right
of this data correction process, the reliability statistics given in the previous hand than in the left hand, the ANOVA yielded a significant Sex ⫻
paragraph are lower bound estimates. In all subsequent analyses of finger Hand interaction, F(1, 2067) ⫽ 10.00, p ⫽ .002, 2 ⫽ .005. The
lengths, the mean of the measures from the two independent raters (or the ANOVA also yielded a significant main effect for sex, F(1,
replacement measures for double-checked fingers) serve as the final mea- 2067) ⫽ 50.99, p ⫽ .000, 2 ⫽ .024, and for hand, with right-hand
sure of finger lengths, and right-hand and left-hand 2D:4D ratios were 2D:4D ratios lower than left-hand 2D:4D ratios (respective Ms ⫽
computed from these final finger-length measures. .951 and .963); F(1, 2067) ⫽ 344.31, p ⫽ .000, 2 ⫽ .143.
Distributions of 2D:4D ratios were visually inspected, and several outlier To examine ethnic differences in 2D:4D ratios, a repeated-
values (greater than 3.5 standard deviations from the mean) were elimi- measures 2 ⫻ 3 ⫻ 2 (sex ⫻ ethnicity ⫻ hand) ANOVA was
nated. The presence of these values did not substantially alter any of the
conducted on 2D:4D ratios. The three ethnic groups examined
results that follow. However, their elimination slightly reduced the vari-
ance and increased the normality of distributions. Visual inspection showed
were Whites, Hispanics, and Asians. Because there were other
that 2D:4D distributions closely approximated superimposed normal dis- ethnic groups represented in the total sample, the current analysis
tributions. Over all participants, skewness and kurtosis statistics were .13 was conducted on a subset of the total sample. Consistent with
and .04 for left-hand 2D:4D ratios and .05 and .05 for right-hand 2D:4D previous analyses, this ANOVA showed a sex difference in 2D:4D
ratios. For men, skewness and kurtosis statistics were .11 and .09 for the ratios, F(1, 1768) ⫽ 47.38, p ⫽ .000, 2 ⫽ .026. There was also
left hand and ⫺.01 and .03 for the right hand, and for women these a significant main effect for ethnicity, F(2, 1768) ⫽ 71.67, p ⫽
statistics were .12 and .00 for the left hand and .10 and .02 for the right .000, 2 ⫽ .075, but no significant interaction between sex and
hand. ethnicity, F(2, 1768) ⫽ 1.05, ns. Mean 2D:4D ratios for Whites,
Hispanics, and Asians were .957, .940, and .943 for the right hand
Finger Lengths, 2D:4D, and Height and .970, .950, and .953 for the left hand (respective ns were 1,137
Self-reported height (recorded in inches) was collected from 299 men for Whites, 395 for Hispanics, and 242 for Asians).
(129 heterosexual men and 170 gay men) and from 414 women (253 It is noteworthy that the effect size (2) for ethnicity was
heterosexual women and 161 lesbians). Not surprisingly, height was mod- considerably larger than the effect size for sex. This has implica-
erately correlated with finger lengths. For men, height correlated .56 with tions for subsequent analyses that contrast heterosexual and ho-
right 2D, .55 with right 4D, .55 with left 2D, and .53 with left 4D (all ps ⬍ mosexual participants, for if sexual orientation groups differ in
.001). For women, the corresponding correlations were .50, .46, .50, and their ethnic composition (as they did in the current study), then
.47 (all ps ⬍ .001). For men, height correlated .03 with right 2D:4D and .01 ethnic differences in 2D:4D ratios could lead to spurious results
with left 2D:4D (both ns). For women, height correlated .11 both with right
and with left 2D:4D ( ps ⬍ .05).
2
The current sample included only small numbers of African American
Results participants: 11 heterosexual men, 12 gay men, 21 heterosexual women,
and 9 lesbian women. Neither sex differences nor heterosexual–
The central analyses that follow contrast men with women and homosexual differences in 2D:4D ratios were significant for African Amer-
heterosexual with homosexual participants on 2D:4D ratios and on ican participants. However, because of the small sample sizes, these results
finger lengths. To convey results as simply as possible, the fol- should be viewed as unreliable.
FINGER-LENGTH RATIOS 183
.348. It also showed a significant Sex ⫻ Digit interaction, F(1, RH ⫺.32*** .937 .031 .948 .035
LH ⫺.31*** .952 .033 .962 .034
Whiteb
RH ⫺.37*** .941 .030 .953 .036
Table 3 LH ⫺.44*** .955 .032 .969 .032
Sex Differences in Absolute Finger Lengths Hispanicc
RH ⫺.24†† .929 .035 .937 .030
Mean finger length LH ⫺.21† .941 .036 .948 .034
(in millimeters) Asiand
RH ⫺.22 .931 .028 .937 .030
Finger Effect size Men Women LH .17 .945 .028 .940 .033
Age ⱕ 30 yearse
Right 4D 1.51*** 77.46 (4.71) 70.69 (4.29) RH ⫺.29** .937 .031 .946 .030
Left 4D 1.51*** 77.59 (4.72) 70.88 (4.27) LH ⫺.30** .951 .033 .961 .033
Right 2D 1.26*** 73.04 (4.63) 67.54 (4.20)
Left 2D 1.34*** 74.24 (4.53) 68.39 (4.24) Note. Negative effect sizes (Cohen’s d) indicate that heterosexual men
have lower 2D:4D ratios than gay men, and positive effect sizes indicate
Note. Positive effect sizes (Cohen’s d) indicate that men have longer the reverse. Significance levels are based on two-tailed t tests comparing
fingers than women. Significance levels are based on two-tailed t tests heterosexual and gay men’s 2D:4D ratios.
a
comparing men’s and women’s mean finger lengths. Data in parentheses 349 heterosexual, 459 – 460 gay. b 155–156 heterosexual, 317–318
are standard deviations. Men: n ⫽ 846 – 848; women: n ⫽ 1,229. 4D ⫽ gay. c 78 heterosexual, 71 gay. d 59 heterosexual, 19 gay. e 306 het-
ring finger; 2D ⫽ index finger. erosexual, 135–136 gay.
*** p ⬍ .001. † p ⫽ .20. †† p ⫽ .15. ** p ⬍ .01. *** p ⬍ .001.
184 LIPPA
tends to be consistent across groups: Heterosexual men had sig- show a significant interaction between sexual orientation and digit,
nificantly lower (more male-typical) 2D:4D ratios than gay men F(1, 806) ⫽ 20.73, p ⫽ .000, 2 ⫽ .025, consistent with the pattern
did. The only group for which this pattern did not hold was Asian shown in Table 5.
men. However, because of the small number of gay Asian men As noted earlier, self-reported height was obtained from 128
(n ⫽ 19), the difference between heterosexual Asian and gay Asian heterosexual men and 169 gay men. The previous ANOVA was
men’s 2D:4D ratios should be regarded as statistically unreliable. conducted on just these participants, with height treated as a
To compare sexual orientation and ethnicity effects, a 2 ⫻ 3 ⫻ 2 covariate. Despite the reduced sample size and the statistical
(sexual orientation ⫻ ethnicity ⫻ hand) repeated-measures control for height, this ANOVA once again showed a significant
ANOVA was conducted on men’s 2D:4D ratios. Sexual orienta- interaction between sexual orientation and digit, F(1, 294) ⫽ 5.58,
tion and ethnicity were between-subjects factors, and hand (right p ⫽ .019, 2 ⫽ .019.
vs. left) was a repeated-measures factor. The three main ethnic 2D:4D ratios in heterosexual and lesbian women. Table 6
groups examined were Whites, Hispanics, and Asians. This presents heterosexual–lesbian differences in women’s 2D:4D ra-
ANOVA showed significant main effects for sexual orientation, tios for the right and left hands. Although there was one significant
F(1, 693) ⫽ 4.79, p ⫽ .029, 2 ⫽ .007, and for ethnicity, F(2, difference and one marginally significant difference for all women
693) ⫽ 19.08, p ⫽ .000, 2 ⫽ .052. It is noteworthy that the combined, these differences were not consistent across groups.
magnitude of the ethnicity effect was much greater than the mag- Furthermore, these differences were opposite in direction to those
nitude of the sexual orientation effect. found in previous studies—that is, in the current data lesbian
Absolute finger lengths of heterosexual and gay men. The women had higher (more female-typical) ratios than heterosexual
previous analyses showed the heterosexual men had lower 2D:4D women. It is possible that the differing ethnic composition of the
ratios than gay men, but they do not indicate whether these heterosexual and lesbian samples may have contributed to these
differences were due to heterosexual– gay differences in the length significant differences in 2D:4D ratios. If so, then these differences
of 2D, the length of 4D, or the length of both fingers. To answer may represent spurious findings.
this question, analyses were conducted that compared the absolute To take ethnicity effects into account, a 2 ⫻ 3 ⫻ 2 (sexual
finger lengths of heterosexual men and gay men. Table 5 presents orientation ⫻ ethnicity ⫻ hand) repeated-measures ANOVA was
mean 2D and 4D lengths for heterosexual men and gay men, and conducted on women’s 2D:4D ratios. Sexual orientation and eth-
it also presents effect sizes for heterosexual– gay differences. The nicity were between-subjects factors, and hand (right vs. left) was
data in Table 5 show significant heterosexual– gay differences for a repeated-measures factor. The three main ethnic groups exam-
4D but not for 2D. The pattern of results suggests that the signif- ined were Whites, Hispanics, and Asians. There proved to be no
icant heterosexual– gay difference in 4D length was not due simply main effect for sexual orientation, F(1, 987) ⫽ 0.04, p ⫽ .84, but
to an overall height or stature difference between heterosexual and there was a significant main effect for ethnicity, F(2,
gay men, which presumably would affect all digit lengths. 987) ⫽ 31.60, p ⫽ .000, 2 ⫽ .06.
To test whether there was in fact a stronger link between sexual There were very few Asian lesbians in this analysis (n ⫽ 11),
orientation and 4D than between sexual orientation and 2D, a 2 ⫻ and it is possible that this small cell size distorted the results of the
2 ⫻ 2 (sexual orientation ⫻ digit ⫻ hand) repeated-measures ANOVA. To test this possibility, the ANOVA was conducted
ANOVA was conducted on men’s finger lengths. Sexual orienta- again, this time with ethnicity defined in terms of just two groups
tion was a between-subjects factor, and digit (2D vs. 4D) and hand (White, Hispanic) rather than three groups (White, Hispanic,
(right vs. left) were within-subjects factors. This analysis did not Asian). Despite the change, the results were much the same. There
show an overall difference in finger length between gay men and was no main effect for sexual orientation, F(1, 836) ⫽ 0.37, p ⫽
heterosexual men, F(1, 806) ⫽ 1.93, p ⫽ .17. However, it did .54, but there was a significant main effect for ethnicity, F(1,
837) ⫽ 60.01, p ⫽ .000, 2 ⫽ .067.
Absolute finger lengths of heterosexual and lesbian women.
Table 5 Table 7 presents absolute 2D and 4D lengths for lesbians and
Differences in Absolute Finger Lengths of Heterosexual and heterosexual women and effect sizes for lesbian– heterosexual
Gay Men differences. Unlike the corresponding table for men, there were no
significant differences for women.
Mean finger length
(in millimeters)
Sex Differences in 2D:4D Ratios Revisited
Finger Effect size Heterosexual Gay
Table 1 presents findings on sex differences in 2D:4D ratios.
Right 4D 0.16* 77.91 (4.61) 77.15 (4.71) However, the current study sampled large numbers of gay men,
Left 4D 0.20** 78.13 (4.56) 77.21 (4.77) who proved to have higher 2D:4D ratios than heterosexual men.
Right 2D ⫺0.02 72.99 (4.40) 73.07 (4.75) Because of the large number of gay men in the current sample,
Left 2D 0.02 74.28 (4.30) 74.19 (4.70)
Table 1 may overestimate population means for men’s 2D:4D
Note. Positive effect sizes (Cohen’s d) indicate that heterosexual men ratios, and as a result, the effect sizes for sex differences presented
have longer fingers than gay men, and negative effect sizes indicate the in Table 1 may underestimate population sex differences in 2D:4D.
reverse. Significance levels are based on two-tailed t tests comparing To explore this possibility, effect sizes for sex differences in
heterosexual and gay men’s mean finger lengths. Data in parentheses are
standard deviations. Heterosexual men: n ⫽ 349 –350; gay men: n ⫽
2D:4D ratios were computed for just heterosexual participants.
460 – 461. 4D ⫽ ring finger; 2D ⫽ index finger. These effect sizes were in fact larger than those presented in
* p ⬍ .05. ** p ⬍ .01. Table 1 (d ⫽ ⫺.49 for the right hand and d ⫽ ⫺.37 for the left
FINGER-LENGTH RATIOS 185
hand, both ps ⫽ .000). The corresponding effect sizes for sex Table 7
differences in homosexual participants’ 2D:4D ratios were Differences in Absolute Finger Lengths of Heterosexual and
smaller: ⫺.35 ( p ⫽ .000) for the right hand and ⫺.19 for the left Lesbian Women
hand ( p ⫽ .004). Representative population samples would, of
course, include many more heterosexual than homosexual individ- Mean finger length
(in millimeters)
uals, and therefore sex differences in 2D:4D ratios in representa-
tive samples would likely be closer to the computed sex differ- Finger Effect size Heterosexual Lesbian
ences for heterosexual participants.3
Right 4D 0.06 70.81 (4.12) 70.55 (4.57)
Left 4D 0.00 70.90 (4.09) 70.88 (4.54)
Discussion Right 2D ⫺0.06 67.47 (4.17) 67.70 (4.27)
Left 2D ⫺0.06 68.31 (4.16) 68.58 (4.41)
The current study found that heterosexual men had lower (more
male typical) 2D:4D ratios than gay men, and these differences Note. Positive effect sizes (Cohen’s d) indicate that heterosexual women
appeared to result from differences in 4D lengths rather than have longer fingers than lesbians, and negative effect sizes indicate the
differences in 2D lengths. After controlling for ethnicity, there was reverse. All differences are nonsignificant on the basis of two-tailed t tests
comparing heterosexual women’s and lesbians’ mean finger lengths. Data
no evidence that heterosexual women differed from lesbian
in parentheses are standard deviations. Heterosexual women: n ⫽ 705;
women in 2D:4D ratios, nor did heterosexual and lesbian women lesbians: n ⫽ 468. 4D ⫽ ring finger; 2D ⫽ index finger.
differ in their absolute finger lengths. The current study improved
on previous research by assessing large, ethnically diverse samples
and by comparing the consistency of findings across ethnic groups.
The fact that heterosexual– homosexual differences in men’s are concordant and discordant on sexual orientation. Such siblings
2D:4D ratios tended to be consistent across hands, across ethnic would necessarily be matched on ethnicity (and on other sources of
groups, and in age-matched groups increases the likelihood that genetic variation as well).
these findings are reliable. Conversely, the fact that the one sig- In the current study, heterosexual– homosexual differences for
nificant heterosexual–lesbian difference observed in women’s men tended to be smaller in magnitude than sex differences in
2D:4D ratios was not consistent across ethnic groups suggests that 2D:4D ratios, and both heterosexual– homosexual differences and
it was a spurious result. sex differences were considerably smaller in magnitude than eth-
The current study used a relatively coarse system for classifying nic differences. If both heterosexual– homosexual differences and
participants’ ethnicity, and future studies could improve on this by sex differences in 2D:4D ratios result from androgen effects, then
assessing participants’ ethnicity and national origins in more fine- it makes sense that heterosexual– homosexual differences should
grained ways. Another promising method to control for ethnicity be smaller than sex differences. In a sense, male–female differ-
effects would be to assess 2D:4D ratios in same-sex siblings who ences serve as the gold standard of hypothesized androgen effects
on 2D:4D, for it is certain that most males and females experience
very large mean differences in prenatal exposure to androgens. If,
Table 6 as suggested by neurohormonal theories of sexual orientation,
Heterosexual–Lesbian Differences in Right-Hand (RH) and Left- adult sexual orientation is sometimes influenced by variations in
Hand (LH) Index-to-Ring-Finger (2D:4D) Ratios for Various prenatal androgen levels at critical stages of development, then
Groups of Women these variations would be superimposed on large mean sex differ-
ences in prenatal androgen levels.
Heterosexual Lesbian Whenever possible, future studies on 2D:4D ratios and sexual
orientation should compare effect sizes for heterosexual–
Participant group Effect size M SD M SD
homosexual differences with effect sizes for sex differences. If the
Alla first is larger than the second, then this increases the likelihood that
RH ⫺0.22*** .953 .032 .960 .034 heterosexual– homosexual differences may be influenced by meth-
LH ⫺0.11† .964 .035 .968 .037 odological artifacts such as ethnic differences in heterosexual and
Whiteb homosexual samples. It seems likely that 2D:4D ratios are influ-
RH ⫺0.07 .961 .031 .964 .034
LH ⫺0.01 .973 .033 .974 .035 enced by many factors in addition to androgen levels, because sex
Hispanicc differences in 2D:4D ratios are relatively small in comparison to
RH ⫺0.21 .943 .031 .949 .028 other common physical sex differences (e.g., male–female height
LH 0.18 .955 .031 .948 .041 differences or sex differences in absolute finger length).
Asiand
RH 0.08 .950 .032 .947 .026
The current study is the first to show consistent heterosexual–
LH ⫺0.02 .959 .035 .960 .049 homosexual differences in 2D:4D ratios in both hands. The con-
sistency of the sexual orientation effect for men, both across hands
Note. Negative effect sizes (Cohen’s d) indicate that heterosexual women and across ethnic groups, may be due to the large samples obtained
have lower 2D:4D ratios than lesbians, and positive effect sizes indicate the in the current research. Given the likelihood that 2D:4D ratios
reverse. Significance levels are based on two-tailed t tests comparing
heterosexual women’s and lesbians’ mean 2D:4D ratios.
a
704 –705 heterosexual, 467– 468 lesbian. b 292 heterosexual, 325 3
lesbian. c 155–156 heterosexual, 71 lesbian. d 142 heterosexual, 11 les- By implication, one explanation for why Asians were the group with
bian. the largest sex differences in 2D:4D (see Table 2) is that there were
† p ⫽ .068. *** p ⬍ .001. relatively few gay Asian men in the current sample.
186 LIPPA
constitute a very “noisy” indicator of prenatal androgen exposure individual has (Blanchard, 1997). Mathematical analyses of rele-
and that only a fraction of the variance in 2D:4D ratios is linked to vant data have suggested that with each additional older brother, a
androgen effects, it is probably necessary to sample very large male’s odds of being homosexual increase by approximately 33%,
numbers of participants for reliable sexual orientation differences and about one gay man in seven is estimated to be homosexual
in 2D:4D ratios to emerge. By the same token, the null results for because of the older brothers effect (Cantor, Blanchard, Paterson,
women in the current study are also likely to be reliable, given the & Bogaert, 2002). The most promising theoretical explanation for
large numbers of heterosexual and lesbian women studied. the older brothers effect is that, with each succeeding male fetus,
The current study found that sex differences in 2D:4D were a mother is increasingly likely to develop an immunological reac-
stronger for the right hand than for the left hand, and this finding tion against Y-linked antigens in male fetal cells, and this immu-
has been reported by others as well (Brown et al., 2002; Manning, nological reaction affects the sexual differentiation of the fetal
Scott, Wilson, & Lewis-Jones, 1998; Williams et al., 2000). Brown male brain (Blanchard, 2001). This example provides a concrete
et al. (2002) observed that such findings “suggest that the right demonstration that (a) a given theory may account for only some
hand finger ratios are more sensitive to prenatal androgen than are cases of homosexuality (or heterosexuality), and (b) a given phys-
those on the left” (p. 126). Some theories of brain lateralization iological process may affect sexual orientation in one sex but not
have speculated that the two halves of the brain grow at different in the other.
rates and that prenatal androgens can retard brain development and Why did the current study find heterosexual– homosexual dif-
therefore differentially affect the two halves of the brain, depend- ferences in 2D:4D ratios for men but not for women? One possi-
ing on the timing of androgen exposure (Geschwind & Galaburda, bility is that male sexual orientation is more influenced by prenatal
1985a). It is conceivable that androgen effects could also differ for androgens than female sexual orientation is, or that there is more
the right and left sides of other bilateral bodily structures. How- variation, on average, in males’ than in females’ prenatal androgen
ever, such theories remain speculative and unproven. Another levels. However, given other sorts of evidence that prenatal an-
possibility is that the density of androgen receptors varies in drogens play a role in female sexual orientation (e.g., evidence
different parts of the body. If androgens affect the development of from CAH females), this hypothesis seems unlikely to be true in its
the right hand more than the left and if variations in prenatal strongest form—that is, that prenatal androgens influence male
androgen levels also influence sexual orientation, it then seems sexual orientation but do not at all influence female sexual orien-
reasonable to expect that heterosexual– homosexual differences in tation. A second possibility is that the critical period when andro-
2D:4D might be greater in the right hand than in the left hand. This gens influence sexual orientation differs for males and females. If
pattern was not observed in the current data, however. the prenatal period during which 2D:4D ratios are established
Manning (2002) has argued that androgens affect 4D more than coincides with the male but not with the female critical period for
2D length, and this was supported in the current study. Sex androgen influences on sexual orientation, then this could generate
differences proved to be larger for 4D than for 2D, and there were the current pattern of results. One way to investigate this possibil-
heterosexual– homosexual differences in men’s 4D but not 2D ity is to study other physical markers, which may be influenced by
lengths. Although not reported here, ethnic groups sometimes prenatal androgen levels during other periods of fetal development,
showed differences in 2D lengths as well as 4D lengths, and this and see if they are associated with adult sexual orientation. Yet a
provides evidence that the processes underlying ethnic differences third possible explanation is that only some subgroups of homo-
in 2D:4D may not be the same as those underlying sex differences sexual men and women owe their sexual orientation to the effects
and heterosexual– homosexual differences in 2D:4D. More of prenatal androgens. This hypothesis is supported by recent
broadly, this suggests that there are multiple physiological pro- research showing different correlates of sexual orientation among
cesses that affect 2D:4D ratios. butch versus femme lesbians (Brown et al., 2002; Singh et al.,
The current study found that heterosexual men have lower 1999). However, even if only some subgroups of lesbians owe
(more male-typical) 2D:4D ratios than gay men, whereas Robinson their sexual orientation to the prenatal effects of androgens, this
and Manning (2000) reported just the opposite. Additional studies would seemingly weaken but not eliminate heterosexual–lesbian
will be needed to resolve these conflicting results. Nonetheless, it 2D:4D differences in very large samples of women.
is worth noting again that the very large sample sizes in the current The finding that male but not female sexual orientation was
study and the stability of the current findings for men across hands associated with 2D:4D ratios is consistent with recent theoretical
and across ethnic groups increase the likelihood that the current speculations that various aspects of male sexuality are relatively
findings are reliable. Furthermore, the current findings are consis- more driven by biological factors (e.g., genes, prenatal hormones,
tent with a number of other studies showing that gay men are adult hormone levels), whereas aspects of female sexuality are
intermediate between heterosexual men and heterosexual women relatively more molded by cultural, social, and situational factors
on a number of sex-linked behaviors, traits, and physical charac- (Baumeister, 2000; Baumeister & Tice, 2001; Peplau, 2001). This
teristics (e.g., Bailey, Gaulin, Agyei, & Gladue, 1994; Gladue & is not to say that male sexual orientation is impervious to social
Bailey, 1995; Kimura, 1999; LeVay, 1991; Lippa, 2000, 2002b). influences or that female sexual orientation is free of biological
The current data add to evidence that for men at least, prenatal influences but rather that the relative balance of biological and
androgen levels may be linked to adult sexual orientation. How- social influences may differ, on average, for males and females. In
ever, even if neurohormonal theories of sexual orientation have regard to this hypothesis, future research can investigate whether
some validity, they may prove to provide only a partial explanation other putative markers of prenatal androgens show sex differences
for the development of human sexual orientation. Illustrating this in their relation to sexual orientation (e.g., see Bogaert, Friesen, &
point is recent research showing that male but not female sexual Klentrou, 2002, for evidence that there are heterosexual–
orientation is associated with the number of older brothers an homosexual differences in the age of puberty for males but not for
FINGER-LENGTH RATIOS 187
females, and Lalumière et al., 2000, for evidence that the relation Breedlove, S. M. (1994). Sexual differentiation of the human nervous
between handedness and sexual orientation may differ for men and system. Annual Review of Psychology, 45, 389 – 418.
women). Brown, W. M., Finn, C. J., Cooke, B. M., & Breedlove, S. M. (2002).
The current data provide at least partial support for the hypoth- Differences in finger length between self-identified “butch” and
“femme” lesbians. Archives of Sexual Behavior, 31, 123–128.
esis that prenatal androgen levels are associated with adult sexual
Brown, W. M., Hines, M., Fane, B., & Breedlove, S. M. (2001). Mascu-
orientation. Even if neurohormonal theories of sexual orientation
linized finger length ratios in humans with congenital adrenal hyperpla-
are true in general outline, though, the linkage between prenatal sia (CAH) [Abstract]. Hormones and Behavior, 39, 325.
hormones and adult sexual orientation is likely to be a complex Byne, W., & Parsons, B. (1993). Human sexual orientation: The biological
one, moderated by many factors (Berenbaum, 2002; McFadden, theories reappraised. Archives of General Psychiatry, 50, 228 –239.
2002). For example, the effects of prenatal androgens may depend Byne, W., Tobet, S., Mattiace, L. A., Lasco, M. S., Kemether, E., Edgar,
critically on the timing of androgen surges and declines. Androgen M. A., et al. (2001). The interstitial nuclei of the human anterior
effects may depend not only on absolute androgen levels but also hypothalamus: An investigation of variation within sex, sexual orienta-
on variations in individuals’ ability to utilize androgens (e.g., tion, and HIV status. Hormones and Behavior, 40, 86 –92.
variations in the nature, location, or quantity of androgen receptors Cantor, J. M., Blanchard, R., Paterson, A. D., & Bogaert, A. F. (2002).
in various tissues). As noted earlier, some androgen effects may be How many gay men owe their sexual orientation to fraternal birth order?
nonlinear, with increasing androgen levels masculinizing struc- Archives of Sexual Behavior, 31, 63–71.
Collaer, M. L., & Hines, M. (1995). Human behavioral sex differences: A
tures up to a point, but then feminizing structures after some
role for gonadal hormones during early development. Psychological
“optimal” level is reached. Finally, androgen effects are likely to
Bulletin, 118, 55–107.
be moderated by the presence of other hormones and chemical Cooke, B., Hegstrom, C. D., Villeneuve, L. S., & Breedlove, S. M. (1998).
factors. Sexual differentiation in the vertebrate brain: Principles and mecha-
The existence of all these potentially complicating factors nisms. Frontiers in Neuroendocrinology, 19, 323–362.
makes the current findings all the more remarkable—that for men, Dittmann, R. W., Kappes, M. E., & Kappes, M. H. (1992). Sexual behavior
at least, there was in fact a consistent and significant relationship in adolescent and adult females with congenital adrenal hyperplasia.
between a possible marker of prenatal androgen levels and adult Psychoneuroendocrinology, 17, 153–170.
sexual orientation. Additional large-scale studies will be necessary Ellis, L., & Ames, M. A. (1987). Neurohormonal functioning and sexual
to establish whether the current findings replicate in other popu- orientation: A theory of homosexuality– heterosexuality. Psychological
lations. If they do, then the question of whether 2D:4D ratios are Bulletin, 101, 233–258.
related to sexual orientation may prove to have a two-part answer: Garn, S. M., Burdi, A. R., Babler, W. J., & Stinson, S. (1975). Early
prenatal attainment of adult metacarpal–phalangeal rankings and pro-
yes for men but no for women.
portions. American Journal of Physical Anthropology, 43, 327–332.
Geschwind, N., & Galaburda, A. M. (1985a). Cerebral lateralization.
Biological mechanisms, associations and pathology: I. A hypothesis and
References
a program for research. Archives of Neurology, 42, 428 – 459.
Adkins-Regan, E. (1988). Sex hormones and sexual orientation in animals. Geschwind, N., & Galaburda, A. M. (1985b). Cerebral lateralization.
Psychobiology, 16, 335–347. Biological mechanisms, associations and pathology: II. A hypothesis
Bailey, J. M. (1995). Biological perspectives on sexual orientation. In A. R. and a program for research. Archives of Neurology, 42, 521–552.
D’Augelli & C. J. Patterson (Eds.), Lesbian, gay, and bisexual identities Gladue, B. A., & Bailey, J. M. (1995). Sexual orientation and aggressive-
over the lifespan (pp. 103–135). New York: Oxford University Press. ness. Psychoneuroendocrinology, 20, 475– 485.
Bailey, J. M., Gaulin, S., Agyei, Y., & Gladue, B. A. (1994). Effects of Gorski, R. A. (1985). Sexual differentiation of the brain: Possible mecha-
gender and sexual orientation on evolutionarily relevant aspects of nisms and implications. Canadian Journal of Physiology and Pharma-
human mating psychology. Journal of Personality and Social Psychol- cology, 63, 577–594.
ogy, 66, 1081–1093. Hall, L. S. (2000). Dermatoglyphic analysis of total ridge count in female
Baker, F. (1888). Anthropological notes on the human hand. American monozygotic twins discordant for sexual orientation. Journal of Sex
Anthropologist, 1, 51–75. Research, 37, 315–320.
Baumeister, R. F. (2000). Gender differences in erotic plasticity: The James, W. H. (1989). Foetal testosterone levels, homosexuality, and hand-
female sex drive as socially flexible and responsive. Psychological edness: A research proposal for jointly testing Geschwind’s and Dör-
Bulletin, 126, 347–374. ner’s hypothesis. Journal of Theoretical Biology, 136, 177–180.
Baumeister, R. F., & Tice, D. M. (2001). The social dimension of sex. Kimura, D. (1999). Sex and cognition. Cambridge, MA: MIT Press.
Boston: Allyn & Bacon. Lalumière, M. L., Blanchard, R., & Zucker, K. J. (2000). Sexual orientation
Berenbaum, S. A. (2002). Prenatal androgens and sexual differentiation of and handedness in men and women: A meta-analysis. Psychological
behavior. In E. A. Eugster & O. H. Pescovitz (Eds.), Developmental Bulletin, 126, 575–592.
endocrinology: From research to clinical practice (pp. 293–311). To- LeVay, S. (1991, August 30). A difference in the hypothalamic structure
towa, NJ: Humana Press. between heterosexual and homosexual men. Science, 253, 1034 –1037.
Blanchard, R. (1997). Birth order and sibling sex ratio in homosexual Lippa, R. A. (2000). Gender-related traits in gay men, lesbian women, and
versus heterosexual males and females. Annual Review of Sex Research, heterosexual men and women: The virtual identity of homosexual–
8, 27– 67. heterosexual diagnosticity and gender diagnosticity. Journal of Person-
Blanchard, R. (2001). Fraternal birth order and the maternal immune ality, 68, 899 –926.
hypothesis of male homosexuality. Hormones and Behavior, 40, 105– Lippa, R. A. (2002a). Gender, nature, and nurture. Mahwah, NJ: Erlbaum.
114. Lippa, R. A. (2002b). Gender-related traits of heterosexual and homosex-
Bogaert, A. F., Friesen, C., & Klentrou, P. (2002). Age of puberty and ual men and women. Archives of Sexual Behavior, 31, 83–98.
sexual orientation in a national probability sample. Archives of Sexual Manning, J. T. (2002). Digit ratio. New Brunswick, NJ: Rutgers University
Behavior, 31, 73– 81. Press.
188 LIPPA
Manning, J. T., Scott, D., Wilson, J., & Lewis-Jones, D. I. (1998). The ratio Robinson, S. J., & Manning, J. T. (2000). The ratio of 2nd and 4th digit
of the 2nd to 4th digit length: A predictor of sperm numbers and length and male homosexuality. Evolution and Human Behavior, 21,
concentrations of testosterone, luteinizing hormone and oestrogen. Hu- 333–345.
man Reproduction, 13, 3000 –3004. Singh, D., Vidaurri, M., Zambarano, R. J., & Dabbs, J. M. (1999). Lesbian
McFadden, D. (2002), Masculinization effects in the auditory system. erotic role identification: Behavioral, morphological, and hormonal cor-
Archives of Sexual Behavior, 31, 99 –111. relates. Journal of Personality and Social Psychology, 76, 1035–1049.
McFadden, D., & Pasanen, E. G. (1998). Comparison of the auditory Vasey, P. L. (1995). Homosexual behavior in primates: A review of
systems of heterosexuals and homosexuals: Click-evoked otoacoustic evidence and theory. International Journal of Primatology, 16, 173–204.
emissions. Proceedings of the National Academy of Sciences, 95, 2709 – Wallen, K. (1996). Nature needs nurture: The interaction of hormonal and
2713. social influences on the development of behavioral sex differences in
Meyer-Bahlburg, H. (1984). Psychoendocrine research on sexual orienta-
rhesus monkeys. Hormones and Behavior, 30, 364 –378.
tion. Current status and future options. In G. J. De Vries, J. P. C. De
Williams, T. J., Pepitone, M. E., Christensen, S. E., Cooke, B. M., Huber-
Bruin, H. M. B. Uylings, & M. A. Corner (Eds.), Progress in brain
man, A. D., Breedlove, N. J., et al. (2000, March 30). Finger-length
research (Vol. 61, pp. 375–398). Amsterdam: Elsevier.
ratios and sexual orientation. Nature, 404, 455– 456.
Mustanski, B. A., Bailey, J. M., & Kaspar, S. (2002). Dermatoglyphics,
handedness, sex, and sexual orientation. Archives of Sexual Behavior, Wilson, J. D. (1999). The role of androgens in male gender role behavior.
31, 113–122. Endocrine Reviews, 20, 726 –737.
Peplau, L. A. (2001). Rethinking women’s sexual orientation: An interdis- Zucker, K. J., Bradley, S. J., Oliver, G., Blake, J., Fleming, S., & Hood, J.
ciplinary relationship-focused approach. Personal Relationships, 8, (1996). Psychosexual development of women with congenital adrenal
1–19. hyperplasia. Hormones and Behavior, 30, 300 –318.
Peters, M., Mackenzie, K., & Bryden, P. (2002). Finger length and distal
extent patterns in humans. American Journal of Physical Anthropology,
117, 209 –217. Received June 19, 2002
Phelps, V. R. (1952). Relative index finger length as a sex-influenced trait Revision received October 22, 2002
in man. American Journal of Human Genetics, 4, 72– 89. Accepted October 31, 2002 䡲