Cross-sectional data on soft tissue morphometry of the growing hand and fingers of dextral individuals 5–65 years old

Authors


Correspondence

Terry M. Mayhew, School of Biomedical Sciences, Queen's Medical Centre, E Floor, University of Nottingham, Nottingham NG7 2UH, UK. E: terry.mayhew@nottingham.ac.uk

Abstract

This study examines both hands of right-handed (dextral) subjects 5–65 years old in order to define the separate growth trajectories of digit lengths (2D–5D) and hand widths; to assess how 2D : 4D and other digit ratios also vary with age; and to test whether lengths are influenced by gender dimorphism and lateral (right/left) asymmetry. Calliper measurements were made from hand photocopies. Growth patterns were analysed by linear regression and correlation, main and interaction effects of age and gender were resolved by analysis of variance, and lateral asymmetries were identified by paired tests. All digits, and hand width, grew in a biphasic pattern in both hands, and inflection points between phases showed gender dimorphism. In the early fast-growing phase, male digits grew over a longer period than those in females, before switching to a slower growth phase during which gender dimorphism became more exaggerated. In right hands, age differences in digit ratios were confined to 2D : 4D and, except for 4D : 5D, females tended to show larger ratios than males. In left hands, all ratios (except 3D : 5D) varied with age and gender influenced only 2D : 4D, 2D : 5D and 3D : 5D. Again, ratios were greater in females. In females, 2D was longer in the right hand of older subjects, whilst 3D, 4D and 5D tended to be shorter in the right hand of younger subjects. No asymmetries were seen in 2D, 3D or 4D in males, but 5D tended to be shorter on the right in the group 9–12 years old. Finally, hand width tended to be greater in females on the right at 9–65 years old, and in males on the right at 18–23 years old. A further novel finding was that certain relationships (inflection points, correlation coefficients and gender differences in digit lengths) seemed to follow gradients running from 2D to 5D. It is tempting to speculate that these are manifestations of the antero-posterior gradients established by signalling events that control digit development and patterning in utero.

Introduction

The buds of limbs appear as paddle-like outgrowths during week 5 in utero, and the forelimb buds appear slightly earlier than those of the hindlimbs. At the bud tip, a thickened apical ectodermal ridge (AER) acts as a promoter of the underlying mesenchyme, inducing it to proliferate and differentiate. At 6 weeks, the distal part of the forelimb bud flattens to form the hand plate which, by 7 weeks, displays four radial grooves separating what will become the five digits (Langman, 1981).

Factors that control digit development and patterning are the subject of intense interest (for reviews, see Tickle, 2006; Hu & He, 2008; Stricker & Mundlos, 2011). Apart from the AER, there are other specialised regions of the limb bud such as the zone of polarizing activity (ZPA) and non-ridge ectoderm. Signals from the ZPA cause the hand plate mesenchyme to proliferate, and sonic hedgehog acts with this to fix digit identity in the so-called antero-posterior (1D, thumb to 5D, little finger) direction. Fibroblast growth factor signalling from the AER, together with Wnt signalling from ectoderm, stimulates distal limb bud expansion. Other signalling pathways involving Wnt, β-catenin, homeobox genes and bone morphogenetic protein assist in regulating digit condensation, elongation and phalanx number.

Once the basic anatomy of the hand and digits is established, other genetic and environmental factors sculpt the differences observed over time (pre- and postnatal age effects), between right and left hands (lateral asymmetry) and between males and females (gender dimorphism). For example, the length ratio of digits 2D (index) and 4D (ring) tends to be greater in females and in right hands (Manning et al. 1998; Williams et al. 2000; McFadden & Shubel, 2002; Hönekopp & Watson, 2010). These differences have been attributed, in part, to prenatal exposure to testosterone and oestradiol, and the influence of homeobox and androgen receptor genes (Mortlock & Innis, 1997; Manning et al. 1998, 2002, 2003; Lutchmaya et al. 2004; Hönekopp et al. 2007). Smaller 2D : 4D ratios are associated with high foetal testosterone : oestradiol levels and greater ratios with high oestradiol : testosterone levels (Lutchmaya et al. 2004). Recently, experimental evidence from mice was adduced to support the notion that the 2D : 4D ratio is, indeed, an indicator of prenatal exposure to sex hormones (Zheng & Cohn, 2011). It was found that androgen and oestrogen receptor activities were greater in 4D than 2D, and led to differential growth of 4D in females and males. Moreover, inactivation of the androgen receptor led to an increased 2D : 4D ratio, whilst inactivation of oestrogen receptor α produced a diminished ratio. More recently, the finding that 2D : 4D ratios are determined during a narrow time window of foetal digit development by the balance of testosterone and oestrogen signalling has been discussed in an historical context (Manning, 2011).

In addition to molecular and genetic analyses, many morphometric studies have been conducted on the absolute and relative lengths and girths of individual digits. Their aims have been diverse, and included testing for age, gender and lateral differences (Garn et al. 1975; Manning et al. 2004, 2007; McIntyre et al. 2005, 2006; Malas et al. 2006; Gillam et al. 2008; Manning & Fink, 2008; Manning, 2010), predicting stature for forensic reasons (Habib & Kamal, 2010) and correlating with manual dexterity, hand preference, reproductive success, the menstrual cycle, sexual preference, athletic ability, ethnicity, susceptibility to diseases, disabilities and injuries (Manning & Bundred, 2000; Williams et al. 2000; Manning & Taylor, 2001; Brown et al. 2002a,b; Buck et al. 2003; Fink et al. 2004; Manning et al. 2004, 2007; McIntyre et al. 2006; Trivers et al. 2006; Mayhew et al. 2007; Gillam et al. 2008; Manning & Fink, 2008; Hohendorff et al. 2010; Barut et al. 2011; Schulter et al. 2012), and molecular data (Lutchmaya et al. 2004; Zheng & Cohn, 2011). Some studies have focussed on a single hand or been susceptible to experimental confounders (such as those just identified) that affect comparisons of relative and absolute digit lengths.

Understandably, most investigators have relied on cross-sectional data because these are easier to generate for wider age spans than longitudinal studies. The present cross-sectional study extends our earlier investigations, which focussed solely on digits 2D and 4D (Mayhew et al. 2007; Gillam et al. 2008), and has three main aims: first, to determine the separate growth trajectories of digit lengths (2D–5D) and hand widths over a span of 60 years; second, to examine how 2D : 4D and other digit ratios also vary with age; and third, to test whether lengths are influenced by gender dimorphism and lateral asymmetry.

Materials and methods

Subjects were recruited from different educational sites in Nottinghamshire and Leicestershire as part of an earlier study on 2D : 4D ratios (Gillam et al. 2008). The age ranges were 5–18 years (school pupils), 18–23 years (medical undergraduates) and 24–65 years (teachers). The initial study was approved by the University of Nottingham Medical School Ethics Committee, the Director of Education of Nottinghamshire County Council, and the Head Teachers and Boards of Governors of relevant schools. Parents of all children under 16 years old consented by means of a reply slip attached to a letter that explained the purpose of the study. Subjects over 16 years old decided whether or not to participate based on written and oral explanations of the study and an opportunity to seek further clarification.

Individuals were classed as dextral (right-handed) or sinistral (left-handed) on the basis of volunteered information about hand preference or empirical evidence obtained by watching them write. About 90% were dextral, and present analyses were confined to these individuals because of limitations of sample sizes for left-handers. Moreover, individuals of ambiguous or ambivalent hand preference were excluded.

Subjects were divided into five different age groups for the purpose of statistical analysis. Because of exclusions (see also below), a potential sample of 409 subjects was reduced to one of n = 212. The group ages, sample sizes and genders are summarised in Table 1.

Table 1. Mean age (SEM) and sample size per group arranged by gender
Age group (range, years old)FemalesMales
Mean ageSample sizeMean ageSample size
Group 1 (5–8)6.48 (0.22)236.35 (0.19)23
Group 2 (9–12)10.14 (0.16)2910.12 (0.14)34
Group 3 (14–18)15.81 (0.19)2715.70 (0.21)20
Group 4 (18–23)20.29 (0.28)1720.43 (0.42)14
Group 5 (24–65)39.80 (2.94)1044.80 (2.82)15
Total 106 106

Hand photocopying

For internal consistency, photocopies were made according to a strict protocol described previously (Mayhew et al. 2007; Gillam et al. 2008). With this protocol, inter-individual variation in absolute and relative finger lengths was low. In the present investigation, observed coefficients of variation (CV = standard deviation/group mean) for 2D–5D finger lengths and hand widths were in the range 4–9% (see below).

Photocopies were required to display clear hand images with visible relevant skin creases and finger tips. Subjects were asked to remove rings or jewellery that would compromise measurement. Attempts were made also to eliminate other potential confounders, such as damage to bone through disease or trauma.

The palmar surfaces of both hands were placed on the photocopier with the three middle digits (2D–4D) straight and adducted, and 1D and 5D slightly abducted (Peters et al. 2002). Elbow joints were extended and 3D kept in line with the midpoint of the radius and ulna to avoid abduction or adduction at the wrist. Visual checks were made to ensure that 2D–4D of each hand did not overlap. Photocopies showing adducted thumbs or loss of medial and/or lateral borders of the hand were excluded.

Primary measures

The primary measures were hand widths and the lengths of digits 2D–5D on left and right hands taken using digital callipers accurate to 0.01 mm. All measurements were made by the same person and all individuals were anonymous. Measuring errors introduced by use of callipers have negligible impacts on the observed CVs between individuals (Mayhew et al. 2007). The lengths of digits 2D–5D were measured from the midpoint of each proximal crease to the finger tip. Hand width was measured between the medial and lateral borders of the palm of the hand where they are met by the distal and proximal transverse creases, respectively. This line roughly corresponds to that running between the most medial part of the fifth metacarpophalangeal joint and the lateralmost part of the second metacarpophalangeal joint (Malas et al. 2006).

Derived quantities

These took the form of digit ratios and indices of lateral asymmetry. Although 10 digit ratios are available for analysis on each hand, difficulties in measuring the first digit (1D, thumb) reduce this to six. These correspond to the following (expressed as mm mm−1): 2D : 3D; 2D : 4D; 2D : 5D; 3D : 4D; 3D : 5D; and 4D : 5D.

Right–left asymmetries were examined by calculating R/L ratios for each pair of digits and each pair of hand widths. For symmetry, the expected value is R/L = 1. A ratio R/L > 1 signifies a larger right side, and a ratio < 1 a larger left side. Of course, asymmetry can also be analysed taking RL lengths. In this case, symmetry is expected to yield RL = 0.

Study comparisons and statistical analyses

Growth patterns of finger lengths and hand width were analysed by linear regression (Sokal & Rohlf, 1981) as previous studies on 2D and 4D lengths had indicated biphasic linear growth (Gillam et al. 2008). However, as a precaution, the validity of this type of regression was confirmed by observing regression plots for all primary measures. Length (L, in mm) was taken as the dependent and age (A, in years) as the independent variable. Each age group was subjected to two regression analyses, the first covering the period 5–12 years old, and the second 14–65 years old. In the regression equation L = SA + I, the factor S represents the slope of the line and I is the intercept on the L-axis. Values of S and I were calculated together with their corresponding standard errors of the means (SEMs). By solving both regression equations for a common value of L, the point at which the two lines intersected (the inflection point) was calculated. This represents the age at which the initial growth trajectory (5–12 years old) shifts to another trajectory (14–65 years old).

The strengths of associations between digit lengths were examined by means of Pearson's correlation coefficients (Sokal & Rohlf, 1981). Separate analyses were performed for 2D–5D in males and females and in their left and right hands.

The main effects of age and gender were resolved by two-way analyses of variance (anova), which generate age × gender interaction terms (Sokal & Rohlf, 1981). Finally, lateral asymmetries were analysed from anova means and SEMs derived for right/left ratios of the primary measures (e.g. 2D right/2D left). For symmetry (the null hypothesis), the ratio is equal to 1 and mean/SEM provides a t-value.

Null hypotheses were rejected at a probability level of P < 0.05. All calculations, descriptive and inferential statistics were undertaken using unistat v5.5 software (Unistat, London, UK).

Results

Findings are summarised in Tables 2-8. The potential effects of having relatively small sample sizes in the two oldest groups (18–23 and 24–65 years old) were mitigated by the small inter-individual variations (CV values) observed for all variables.

Table 2. Linear regression analyses of growth in length (mm) with age (years) for right and left hands of dextral females 5–12 and 14–65 years old. At the inflection point, linear growth changes from an early to a late phase
LengthPhase, years oldRight slope (SEM)Right intercept (SEM)Inflection, years oldLeft slope (SEM)Left intercept (SEM)Inflection, years old
2D5–122.175 (0.229)37.87 (2.008)15.032.055 (0.244)38.44 (2.140)15.32
14–65−0.039 (0.056)71.15 (1.329)−0.057 (0.057)70.77 (1.347)
3D5–122.380 (0.249)42.60 (2.180)14.652.175 (0.250)44.43 (2.193)14.99
14–65−0.081 (0.058)78.65 (1.382)−0.069 (0.060)78.08 (1.426)
4D5–122.133 (0.222)40.31 (1.947)14.671.919 (0.230)42.37 (2.020)15.03
14–65−0.065 (0.053)72.54 (1.261)−0.058 (0.055)72.09 (1.299)
5D5–121.824 (0.232)32.02 (2.028)14.321.688 (0.232)34.11 (2.030)14.23
14–650.012 (0.053)57.97 (1.249)0.009 (0.047)58.01 (1.124)
Width5–122.144 (0.230)49.98 (2.013)13.941.998 (0.237)50.66 (2.078)13.75
14–650.017 (0.048)79.62 (1.141)0.042 (0.044)77.54 (1.052)
Table 3. Linear regression analyses of growth in length (mm) with age (years) for right and left hands of dextral males 5–12 and 14–65 years old. At the inflection point, linear growth changes from an early to a late phase
LengthPhase, years oldRight slope (SEM)Right intercept (SEM)Inflection, years oldLeft slope (SEM)Left intercept (SEM)Inflection, years old
2D5–122.061 (0.197)39.00 (1.743)16.742.091 (0.205)38.45 (1.811)16.73
14–650.065 (0.048)72.41 (1.410)0.083 (0.046)72.05 (1.364)
3D5–122.131 (0.227)44.98 (2.007)17.362.175 (0.225)44.53 (1.990)17.30
14–650.087 (0.048)80.46 (1.400)0.079 (0.049)80.80 (1.446)
4D5–122.020 (0.211)41.93 (1.867)17.031.975 (0.210)42.33 (1.854)17.55
14–650.108 (0.049)74.49 (1.440)0.093 (0.050)74.84 (1.486)
5D5–121.628 (0.192)34.52 (1.698)17.781.690 (0.207)34.66 (1.831)17.21
14–650.065 (0.048)62.32 (1.405)0.073 (0.051)62.47 (1.500)
Width5–121.878 (0.292)54.70 (2.581)18.281.877 (0.271)54.26 (2.396)17.52
14–650.158 (0.058)86.14 (1.704)0.163 (0.045)84.30 (1.318)
Table 4. Results of two-way anova applied to linear dimensions (mm) in the right hands of dextral individuals. Values are group means (CV%). For each age group and variable, the change from female to male is expressed as a percentage
GroupsRight, 2DChangeRight, 3DChangeRight, 4DChangeRight, 5DChangeRight, widthChange %
  1. Significant effects: A, age; G, gender; A × G, age × gender interaction.

1, female51.5 (6.6)1.257.6 (6.2)1.053.7 (5.6)1.943.3 (6.3)3.763.5 (5.2)3.9
1, male52.1 (7.7)58.4 (7.1)54.7 (7.9)44.9 (8.7)66.0 (5.9)
2, female60.3 (5.9)−0.867.0 (6.2)−0.662.3 (5.8)0.251.0 (7.3)0.072.0 (5.3)2.9
2, male59.8 (5.3)66.6 (5.5)62.4 (5.2)51.0 (5.6)74.1 (6.4)
3, female70.1 (5.4)6.676.9 (5.3)7.571.3 (5.2)8.357.9 (6.6)10.079.7 (4.1)9.2
3, male74.7 (7.2)82.7 (5.7)77.2 (6.2)63.7 (7.0)87.0 (4.4)
4, female71.1 (6.0)1.078.0 (5.7)4.171.5 (4.8)5.358.6 (6.9)7.580.1 (4.5)14.6
4, male71.8 (5.3)81.2 (5.5)75.3 (6.9)63.0 (8.5)91.8 (8.2)
5, female69.4 (5.9)8.675.2 (5.2)12.069.9 (6.7)13.458.6 (5.2)11.680.6 (4.5)15.5
5, male75.4 (5.4)84.2 (5.8)79.3 (5.7)65.4 (6.7)93.1 (5.5)
EffectsA, G, A × G A, G, A × G A, G, A × G A, G, A × G A, G, A × G 
Table 5. Results of two-way anova applied to linear dimensions (mm) in the left hands of dextral individuals. Values are group means (CV%). For each age group and variable, the change from female to male is expressed as a percentage
GroupsLeft, 2DChangeLeft, 3DChangeLeft, 4DChangeLeft, 5DChangeLeft, widthChange %
  1. Significant effects: A, age; G, gender; A × G, age × gender interaction.

1, female51.4 (6.5)1.058.1 (6.2)0.554.4 (6.1)0.744.4 (6.2)1.863.1 (5.2)4.0
1, male51.9 (8.0)58.4 (7.5)54.8 (7.1)45.2 (8.4)65.6 (5.5)
2, female59.6 (6.8)0.066.8 (6.0)0.062.1 (5.9)0.351.7 (6.9)0.471.3 (5.2)3.2
2, male59.5 (6.0)66.5 (5.6)62.3 (5.6)51.9 (5.9)73.6 (6.0)
3, female69.7 (5.8)6.676.3 (5.5)8.771.0 (5.5)9.257.6 (6.2)10.978.2 (4.1)10.4
3, male74.3 (6.9)82.9 (5.8)77.5 (6.5)63.9 (8.3)86.3 (4.2)
4, female69.9 (6.1)3.677.9 (5.7)4.571.3 (5.3)5.358.9 (4.9)7.578.1 (3.9)12.7
4, male72.4 (5.6)81.4 (6.1)75.1 (7.3)63.3 (8.1)88.0 (5.3)
5, female68.5 (5.8)10.775.0 (5.3)12.169.7 (6.1)13.258.6 (5.9)12.679.7 (4.2)15.4
5, male75.8 (5.4)84.1 (5.7)78.9 (5.3)66.0 (6.8)92.0 (5.4)
EffectsA, G, A × G A, G, A × G A, G, A × G A, G, A × G A, G, A × G 
Table 6. Results of two-way anova applied to digit ratios (mm mm−1) in the right hands of dextral individuals. Values are group means (CV%)
GroupsRight, 2D : 3DRight, 2D : 4DRight, 2D : 5DRight, 3D : 4DRight, 3D : 5DRight, 4D : 5D
  1. Significant effects: A, age; G, gender; A × G, age × gender interaction.

1, female0.894 (2.5)0.960 (3.6)1.190 (4.4)1.074 (2.2)1.333 (5.1)1.241 (4.4)
1, male0.892 (4.0)0.953 (4.9)1.163 (5.6)1.068 (2.3)1.304 (4.2)1.221 (3.9)
2, female0.900 (3.4)0.968 (2.4)1.185 (4.2)1.076 (2.6)1.318 (4.8)1.224 (3.8)
2, male0.899 (2.8)0.960 (3.1)1.175 (4.1)1.068 (2.5)1.308 (4.4)1.225 (4.4)
3, female0.913 (2.6)0.984 (3.0)1.215 (4.8)1.078 (2.3)1.331 (4.1)1.235 (3.8)
3, male0.903 (2.8)0.967 (3.5)1.175 (4.9)1.072 (2.6)1.301 (4.4)1.214 (3.7)
4, female0.912 (3.3)0.993 (3.4)1.214 (4.7)1.090 (2.6)1.333 (4.3)1.223 (3.2)
4, male0.885 (2.7)0.955 (4.7)1.145 (7.7)1.079 (3.5)1.293 (6.1)1.198 (5.0)
5, female0.923 (1.4)0.994 (3.4)1.186 (3.6)1.076 (3.1)1.284 (3.1)1.194 (3.5)
5, male0.896 (2.8)0.952 (2.5)1.156 (4.3)1.063 (2.0)1.290 (5.2)1.214 (4.4)
EffectsA?, GA, GGGGNone
Table 7. Results of two-way anova applied to digit ratios (mm mm−1) in the left hands of dextral individuals. Values are group means (CV%)
GroupsLeft, 2D : 3DLeft, 2D : 4DLeft, 2D : 5DLeft, 3D : 4DLeft, 3D : 5DLeft, 4D : 5D
  1. Significant effects: A, age; G, gender; A × G, age × gender interaction.

1, female0.885 (2.6)0.944 (2.6)1.157 (5.2)1.067 (2.1)1.309 (5.6)1.226 (4.6)
1, male0.889 (3.2)0.946 (4.3)1.151 (7.2)1.065 (2.4)1.294 (5.4)1.216 (4.4)
2, female0.891 (2.8)0.959 (2.7)1.153 (3.8)1.076 (2.1)1.294 (4.1)1.203 (3.5)
2, male0.895 (2.8)0.955 (4.0)1.148 (4.2)1.067 (2.1)1.284 (4.1)1.203 (4.2)
3, female0.914 (3.0)0.983 (3.2)1.212 (4.8)1.076 (2.8)1.328 (5.0)1.234 (4.3)
3, male0.895 (2.7)0.959 (3.3)1.165 (4.4)1.071 (2.8)1.302 (4.8)1.216 (3.5)
4, female0.898 (2.4)0.981 (4.1)1.186 (3.9)1.093 (2.9)1.322 (3.3)1.210 (3.5)
4, male0.890 (3.5)0.966 (4.8)1.148 (6.4)1.085 (2.8)1.289 (4.3)1.188 (3.9)
5, female0.913 (2.2)0.983 (3.2)1.168 (2.6)1.077 (1.7)1.281 (2.9)1.189 (3.3)
5, male0.902 (1.7)0.961 (2.7)1.151 (4.4)1.065 (2.3)1.276 (5.3)1.199 (5.8)
EffectsAA, GA,GAGA
Table 8. Results of paired tests applied to right/left digit lengths and hand widths (mm mm−1) in dextral individuals. Values are group means (SEM)
Groups2D R/L3D R/L4D R/L5D R/LWidth R/L
  1. a

    Significant lateral asymmetry (i.e. ratio is significantly different from 1). Significant effects: A, age; G, gender; A × G, age × gender.

1, female1.002 (0.006)0.992 (0.003)a0.986 (0.005)a0.974 (0.007)a1.007 (0.004)
1, male1.004 (0.005)1.001 (0.005)0.998 (0.005)0.993 (0.006)1.006 (0.004)
2, female1.013 (0.005)a1.003 (0.005)1.003 (0.003)0.986 (0.006)a1.010 (0.004)a
2, male1.007 (0.004)1.002 (0.004)1.002 (0.004)0.984 (0.006)a1.006 (0.004)
3, female1.006 (0.004)1.007 (0.003)a1.005 (0.004)1.005 (0.005)1.019 (0.003)a
3, male1.006 (0.004)0.997 (0.003)0.997 (0.004)0.998 (0.006)1.008 (0.005)
4, female1.017 (0.005)a1.002 (0.004)1.004 (0.006)0.994 (0.008)1.025 (0.004)a
4, male0.991 (0.005)0.997 (0.005)1.003 (0.006)0.996 (0.009)1.043 (0.017)a
5, female1.014 (0.003)a1.002 (0.004)1.003 (0.005)1.000 (0.012)1.011 (0.004)a
5, male0.995 (0.003)1.002 (0.005)1.004 (0.005)0.991 (0.008)1.013 (0.006)
1–5, female1.010 (0.002)a1.002 (0.002)1.000 (0.002)0.991 (0.003)a1.014 (0.002)a
1–5, male1.002 (0.002)1.000 (0.002)1.000 (0.002)0.991 (0.003)a1.012 (0.003)a
EffectsG, A × GNoneA? AA

Linear regressions and growth patterns

The growth characteristics of digits 2D–5D and hand width obtained by linear regression analysis are provided separately for females and males in Tables 2 and 3. All measured lengths increased in two phases. In the early phase, slopes and intercepts of regression lines did not differ significantly between digits or between genders. However, intercepts tended to differ between digits and to be larger in males in the later growth phase. There were further gender differences in inflection points. In females, primary measures reached their inflection points sooner (13.7–15.3 vs. 16.7–18.3 years old). This implies that early growth in males proceeds for longer before the switch to the second growth phase. Finally, the inflection point age discrepancy between females and males appeared to vary with digit type from 2D (1.4 year left, 1.7 year right) to 5D (3.0 year left, 3.5 year right). These findings are suggestive of antero-posterior gradients. In the case of hand width, the gender discrepancy was 3.8–4.3 years.

Correlation coefficients and digit lengths

When analysing separate Pearson correlation coefficients for digits in the right and left hands of males and females, some consistent patterns emerged. Whilst all six pairings of digit lengths produced significant strong positive correlations (P < 0.001 in every case), the strongest correlations were seen between adjacent digits in the 2D–4D set (0.976–0.987), but correlations declined with digit separation. For example, correlations between 2D and 3D yielded correlation coefficients of 0.976 (females, right hand), 0.977 (females, left hand), 0.981 (males, right hand) and 0.984 (males, left hand). In contrast, corresponding coefficients for 2D and 4D were 0.971, 0.968, 0.972 and 0.970, and those for 2D and 5D were 0.943, 0.937, 0.945 and 0.949. These suggest that the strength of length relationships declines along the antero-posterior axis. The weakest adjacent pair correlations were found for 4D and 5D (0.958, 0.942, 0.964 and 0.963, respectively).

anova and effects of age and gender

The effects of age and gender on absolute and relative lengths were examined by two-way anova (Fig. 1; Tables 4-7). In the case of absolute lengths, both hands showed significant main effects of age and gender, together with significant first-order interactions between age and gender (P < 0.001 in every case). Digit lengths and hand widths tended to be longer in males, the effect becoming more exaggerated as age increased (Tables 4, 5). This is consistent with faster growth in males once the inflection point age had been passed. The greatest differences were those in hand widths. CVs for absolute lengths were small regardless of age group, gender or laterality (range 3.9–8.7%).

Figure 1.

Age and gender differences in digit length (mm) illustrated for 5D in dextral males and females, and right and left hands. Digit length increases with age from group 1 (5–8 years old) to group 5 (24–65 years old). There is also gender dimorphism, and the graph demonstrates that gender differences increase with age (i.e. there is an interaction effect of age × gender). As with digits 2D–4D and hand width, growth is biphasic, with a faster early phase superseded by a slower late phase. The switch between these phases occurs at the inflection point and is gender-dependent.

When gender differences were expressed as a percentage change from female values, all age groups displayed antero-posterior gradients from 2D to 5D, with the smallest changes affecting 2D and the largest 5D (Tables 4, 5).

In the right hand, significant age differences in relative lengths (Table 6) were confined to the 2D : 4D ratio (P < 0.001). The apparent age effect on 2D : 3D ratio just failed to attain significance. Except for 4D : 5D, females tended to show larger digit ratios than males (P < 0.05 at least). There were no significant age × gender interaction effects.

On the left, age affected all ratios (P < 0.05 at least) except 3D : 5D (Table 7). Gender affected only 2D : 4D, 2D : 5D and 3D : 5D ratios (P < 0.05 at least). Again, ratios were greater in females and no significant interaction terms were detected.

Interestingly, inter-individual variations in ratios were even smaller than those in absolute lengths (CV range 1.4–7.7%).

Lateral asymmetries

Indices of asymmetry, R/L, were found inconsistently across age groups (Table 8). In females, 2D was significantly longer in the right hand in older subjects (age groups 2, 4 and 5). In contrast, 3D, 4D and 5D tended to be shorter in the right hand in group 1, but 3D was longer in group 3 and 5D was shorter in group 2. No asymmetries were seen in 2D, 3D or 4D in males. However, male 5D also tended to be shorter on the right in group 2. Finally, hand width tended to be greater in females on the right in groups 2–5, and in males on the right in group 4.

Two-way anova confirmed the presence of gender and interaction effects on 2D asymmetry, and age effects on 5D and hand width asymmetries. The apparent age effect on 4D asymmetry just failed to reach significance. No significant main or interaction terms were detected for 3D or 4D.

Discussion

Using subjects ranging in age from 5 to 65 years, photocopier images of digits 2D–5D and hand widths were measured in order to examine their growth trajectories, and test for main and interaction effects of age and gender and for lateral symmetry. All digits, and hand width, grew in a biphasic pattern in both hands, but the inflection point between phases was gender dependent. In the early fast-growing phase, male digits grew for a longer period than those in females, before switching to a slower growth phase during which gender dimorphism became more exaggerated. The greatest changes involved hand widths, and inter-individual variation was low enough (4–9%) to indicate tight regulation.

In the right hand, age differences in digit ratios were confined to 2D : 4D, but the possibility of an effect on 2D : 3D cannot be excluded without further analysis of larger samples. Except for 4D : 5D, females tended to show larger ratios than males. In the left hand, all ratios (except 3D : 5D) varied with age. Gender influenced only 2D : 4D, 2D : 5D and 3D : 5D. Again, ratios were greater in females. Inter-individual variations in ratios were less than those in absolute lengths (1–8%). This might be merely a consequence of the strong positive correlations between digit lengths, but it could also reflect tight regulation by molecular and genetic factors in utero.

Lateral asymmetries were detected. In females, 2D was longer in the right hand in older subjects, whilst 3D, 4D and 5D tended to be shorter in younger right hands. No asymmetries were seen in 2D, 3D or 4D in males but, again, 5D tended to be shorter on the right in 9–12 year olds. Finally, hand width tended to be greater in females on the right at 9–65 years old, and in males on the right at 18–23 years old.

A novel and intriguing feature of the present study was the detection of certain relationships indicative of gradients running from 2D to 5D. Specifically, there was a gender-related discrepancy in inflection points that seemed to change from 1.4–1.7 years for 2D to 3.0–3.5 years for 5D. In addition, the strengths of relationships between digit lengths, expressed as correlation coefficients, declined along the 2D–5D axis. The further apart digit pairs, the weaker their correlations. Finally, when gender differences in digit lengths were compared, gradients from 2D to 5D were noted, with the smallest differences in 2D and the largest in 5D. It is tempting to speculate that these morphometric manifestations are signatures of the antero-posterior gradients established by signalling factors that regulate digit development and patterning in utero (Tickle, 2006; Hu & He, 2008; Stricker & Mundlos, 2011).

Age effects on absolute and relative digit lengths

In an earlier cross-sectional study on digits 2D and 4D from 4 to 60 years old, inflection points for dextral females were 14–15 years old, but those for dextral males were about 17 years old (Gillam et al. 2008). Though based on smaller sample sizes, present findings match our earlier estimates. Using data from the BBC Internet Study, Manning (2010) noticed that 2D and 4D lengths in right hands showed more exaggerated gender differences after about 14 years old. A longitudinal analysis of absolute and relative finger lengths in the left hands of male and female subjects between 1 and 17 years old (McIntyre et al. 2005) found that growth patterns of individual phalanges conformed to inflection points at about 13 years old when the proximal phalanges in females were significantly longer than those in males.

Hohendorff et al. (2010) conducted a cross-sectional investigation on subjects 3–10 years old. Finger lengths and girths increased with age, but were not affected by gender. The current study covered a larger age span, and allowed the exploration of age and gender effects and their interactions. The age-related changes in hand widths are partly attributable to changes in finger and bone girth (Malas et al. 2006).

A longitudinal study of left hands (McIntyre et al. 2005) revealed that, by 5 years old, females had higher 3D : 4D ratios and, by 9 years old, higher 2D : 4D ratios as well. Hand preferences in their samples were not identified. By contrast, the present results failed to reveal gender differences in 3D : 4D in left hands of dextral individuals. In a longitudinal study on Jamaican subjects between 7 and 17 years old, 2D : 4D ratios increased slightly with age, but the effect was more marked in the left hand (Trivers et al. 2006). Except for the greater impact in the left hand, the present cross-sectional data are also consistent with these findings, although the earlier study was based on pooled results from males and females.

Gender dimorphism and lateral asymmetry

Differences in finger lengths and finger tip extents have been reported previously (Peters et al. 2002). At about 19–20 years old, longer right and left 2D were detected in Canadian undergraduates. A recent meta-analysis has confirmed that most studies detect a larger 2D : 4D gender difference in the right hand, with greater ratios in females (Hönekopp & Watson, 2010). Present findings support this. Gender differences appear to be similar in direction in different ethnic groups (McIntyre et al. 2006). Apart from 2D : 4D, gender dimorphism has been recorded also for 2D : 3D, 2D : 5D, 3D : 4D and 3D : 5D in one or both hands (McFadden & Shubel, 2002; Manning et al. 2003). The present data confirm the impact of gender on these ratios, at least in the right hand, but also on 3D : 5D in both hands. In contrast, no gender effects on 4D : 5D were detected for either hand. Trivers et al. (2006) found significant gender differences for 2D : 3D (left hand) and 3D : 5D (right hand), with higher values in females.

Whilst some have argued against the value of 2D : 4D as a marker of asymmetry and other traits (Putz et al. 2004), the BBC Internet Study did find asymmetry in males and females in relation to 2D : 4D ratios (Manning & Fink, 2008). The current study has revealed significant age, gender and interaction effects involving asymmetry with respect to 2D and 5D lengths and hand width. Therefore, findings need to be interpreted with caution, taking into account subject age and gender. Habib & Kamal (2010) found no evidence of lateral asymmetry in phalangeal lengths (2D–5D) in Egyptian male or female University students aged 18–25 years. This corresponds to current age group 4, for which larger right hand 2D lengths were found in females but not males.

The balance of these and earlier studies suggests that the lateral differences represent directional rather than fluctuating asymmetries. The former are characterised by a distribution that is not centred around 0 (using RL measures) or 1 (using R/L measures), whereas the latter is characterised by small random deviations from perfect symmetry (RL = 0 or R/L = 1).

Concluding remarks

Bringing the morphometric focus back to growth in absolute digit lengths (2D–5D) has given greater clarity to our understanding of how digit ratios come to alter with age, gender and laterality. It has revealed also that aspects of absolute and relative sizes appear to follow a gradient from thumb to little finger that corresponds to the antero-posterior axis identified by molecular and genetic studies of the developing hand. These findings from cross-sectional data warrant further testing by longitudinal studies of digit lengths and ratios.

Acknowledgements

The author is grateful to the staff, pupils and parents associated with the following schools: Arnbrook Primary & Nursery, Carlton-Le-Willows, Heymanns and Oakham. I thank also Lucy Gillam and Rosie McDonald, who photocopied hands of students as part of a 2004–2005 BMedSci Honours project supervised by Prof. Fran Ebling and myself at the University of Nottingham.