Long Femurs and Squat Form: Debunking the "Long Femur" Myth

by Dr. Brent Brookbush, DPT, MS, CPT, HMS, SPC, IMT

Does Femur Length Really Affect Your Squat? (The "Long Femur" Myth)

It has become popular to blame poor squat form on "long femurs" and to search for "long femur squat" fixes, despite there being no reliable way to measure femur length outside of medical imaging. The claim is that a long thigh bone, relative to the torso or the shin, forces changes in squat form, including excessive forward trunk lean, reduced depth, and a rounding of the lower back at the bottom (lumbopelvic flexion, or "butt wink"). When a lifter assumes the cause is structural, a bone length that cannot be changed, rather than assuming a mobility and strength issue, which can be changed, they often adopt squat modifications that are less than optimal and may create new problems. Skeletal proportions can influence movement, so the claim sounds reasonable. However, this issue is also a myth, and the research below completely dismantles it.

Following a comprehensive systematic review (below) and a geometric model, this article demonstrates:

1. Femur length cannot be measured accurately outside of medical imaging.
2. Human proportions vary far less than the myth assumes: 95% of adults of the same height have femurs within about 4 cm (less than 2 inches) of each other.
3. Even the longest femurs require only a few degrees of additional ankle dorsiflexion to squat well.
4. That amount of dorsiflexion is less than a person can gain in a few weeks of training.
5. Ankle dorsiflexion, which can be measured and trained, predicts squat form more strongly than any bone-length measurement.
6. Correcting ankle dorsiflexion improves not only the squat but every movement pattern that requires dorsiflexion.
7. Common squat recommendations for long femurs may increase the risk of pain and dysfunction, and should be used only as a temporary bridge, if at all.
8. Research-supported sample program for improving dorsiflexion within 5-8 weeks (self-administered and manual therapy interventions)

How Do You Measure Femur Length? (And Why You Can't)

Most of the time, people don't measure "long femurs" at all. People estimate it by eye: a lifter compares their legs to others', notices their legs look long or their hips sit high, and, when their squat doesn't look or feel the way they expect, concludes their femurs must be long. Even careful attempts at femur-length assessment fail, let alone these crude observations. Research demonstrates femur length cannot be assessed accurately by eye or by feel, only by medical imaging.

Individuals who try to assess femur length directly generally use a tape measure. They find a proxy for the top of the femur (the greater trochanter, which cannot be reached directly and sits under soft tissue), use the lateral epicondyle of the knee as the bottom, and measure between them. This method depends on locating these landmarks by feel, and that is where this assessment fails. Research demonstrated that even expert examiners could not locate the top of the femur accurately. When 30 experienced hip surgeons were asked to palpate the greater trochanter and their marks were checked against ultrasound imaging, the average miss was 2.4 cm (0.9 inches), only 1 of 30 landed within 1.0 cm (0.4 inches), and the marks were consistently placed about 2.2 cm (0.9 inches) too high (Christoforetti et al., Section 1.3). A separate study found that the greater trochanter and the outside of the knee were 2 of the 4 least consistently located landmarks out of 12 tested, and that accuracy was worse if the individual assessed was overweight (Moriguchi et al., Section 1.3).

The size of that error is the problem. The full range of femur length among adults of the same height was about 3.7 cm (1.5 inches), from the longest-femured to the shortest-femured individual (Aitken, Section 1.2). The disagreement between examiners locating a single landmark was about 4.8 cm (1.9 inches), and a tape measure requires locating two. That is larger than the entire range the measurement is trying to detect (Christoforetti et al., Section 1.3). So, a tape measure can find "a length" of the femur, but long, average, and short femurs would fall within the margin of error.

The takeaway is that "do you have long femurs?" is a question a lifter cannot answer accurately, and neither can a trainer or a physical therapist without imaging. Fortunately, as the next sections demonstrate, the answer does not matter as much as the myth assumes, and the factor that does matter, ankle dorsiflexion, can be both measured and changed.

Caption: Relative femur length is a very consistent anatomical proportion that can be normally distributed.

Human Proportions Vary Far Less Than You Think

The "long femur" explanation assumes that limb proportions differ widely from person to person. Large measurement studies demonstrated that they do not.

When the femur is measured directly, from the bone, it is 26.5% to 27.0% of a person's standing height, and this held across a worldwide sample of more than 13,000 people, a European cadaver series, and a North American radiographic series (Feldesman et al., Section 1.1; Hauser et al., Section 1.1; Paley et al., Section 1.1). Males and females differed by no more than half a percent. Variation between individuals was small: the standard deviation was about 0.4% to 0.6% of height, meaning 95% of adults of the same height had femurs within about 4 cm (1.6 inches) of each other (Hauser et al., Section 1.1; Feldesman, Section 1.1).

The femur relative to the tibia was just as consistent. Across 1,455 limbs measured on imaging, the femur was 1.26 to 1.28 times the length of the tibia, and no ratio outside a narrow band was observed even in the largest study (Aitken, Section 1.2; Strecker et al., Section 1.2). To make that concrete, consider 2 people of the same height and the same leg length, one with average proportions and one with the most extreme femur-to-tibia ratio measured in over a thousand limbs. Their femurs differed by about 3.7 cm (1.5 inches), and their tibias differed by the same amount in the opposite direction (the longer femur sat on a shorter tibia) (Aitken, Section 1.2).

The part of the body that varied most between 2 lifters of the same height was not the femur; it was the torso, which varied about twice as much as the femur-to-tibia ratio (Aitken, Section 1.2; Strecker et al., Section 1.2; McKean & Burkett, Section 1.2; Fuglsang et al., Section 1.2; Berglund et al., Section 1.2). Yet torso length did not correlate with squat depth or trunk position, because the body accommodates it with small changes in hip and trunk angle (Berglund et al., Section 2.1; Fuglsang et al., Section 2.1). This is the important part: the segment that varied the most had no measurable effect on the squat. If the largest proportional difference in the body is absorbed by a few degrees of joint motion, the smaller difference in femur length is not the structural barrier the myth claims it is.

The takeaway is the one the myth gets backward. If you have long femurs, you are most likely just a taller person whose bones have scaled with your height, not someone with unusual proportions. And whichever segment is longest in your body, the research demonstrated that the difference is small and the body accommodates it with a few degrees of motion at the hip, knee, and ankle. Proportion is not the reason a squat breaks down. The next section calculates exactly how little proportion actually demands of the one joint that matters, the ankle.

How Much Extra Dorsiflexion Does a Long Femur Require?

So how much dorsiflexion does a long femur actually demand? We can answer this with geometry. This calculation is a model, not a study, so it states its assumptions first and its limits at the end.

At the bottom of a squat, the knee moves forward, and the tibia leans forward from vertical as it pivots at the ankle. That forward lean of the tibia is the ankle's dorsiflexion, measured as the angle of the tibia from vertical. The further forward the knee travels, the more the tibia leans, and the more dorsiflexion the ankle requires. We can treat the tibia and a vertical line rising from the ankle as 2 sides of a right triangle, and use geometry to determine how much dorsiflexion is needed to accommodate femurs of different lengths. The triangle has 3 corners: the ankle, the knee, and a point in the air directly above the ankle, level with the knee.

• The tibia is the long side (the hypotenuse), running from the ankle up to the knee.
• The vertical side (y) runs from the ankle straight up to the point level with the knee.
• The horizontal side (x) runs from the top of that vertical side across to the knee. This is how far the knee has traveled forward of the ankle.
• The dorsiflexion angle is the angle at the ankle, between the tibia and the vertical side (y).

In a right triangle, if we know the long side (the tibia) and the horizontal side (how far the knee traveled forward), we can calculate the angle. The tool for this is the arcsine function, which returns an angle when you divide the side opposite the angle (x) by the hypotenuse (the tibia). On a calculator, it is the "sin⁻¹" button. The formula is: dorsiflexion angle = arcsine (knee-forward distance ÷ tibia length)

Caption: Geometry shows that just a few degrees of dorsiflexion are enough to accommodate even the largest femur-length differences.

To keep from falling over during a squat, the load (center of mass) must stay over the foot (base of support). If a lifter has a longer femur, the knee must travel farther forward, which requires more dorsiflexion. If a long femur is present but additional dorsiflexion is not available, the hips are forced farther back, which requires more forward trunk lean to keep the center of mass over the base of support.

The research (Demers et al., Section 2.1 below) demonstrated that extra femur length was not accommodated by more hip flexion. Torso length varied about twice as much as the femur (Aitken et al., Section 1.2; Strecker et al., Section 1.2; McKean & Burkett, Section 1.2; Fuglsang et al., Section 1.2; Berglund et al., Section 1.2), but it did not correlate with squat depth and was accommodated by forward lean at the hip (Berglund et al., Section 2.1; Fuglsang et al., Section 2.1). So neither forward lean nor torso length is likely to change the dorsiflexion the squat requires. This model also assumes a narrow stance (feet hip-width and parallel); Demers et al. (Section 2.1) demonstrated that a wider stance reduced the dorsiflexion required and reduced the correlation between segment lengths and squat form, though a wide stance may itself be correlated with dysfunction (discussed later). The model therefore assumes the entire extra femur length is absorbed by the knee traveling farther forward. This is a conservative estimate, because it removes 3 mechanisms that would otherwise reduce the demand: additional trunk lean, a wider stance, and the fact that even without those, less than the full extra femur length would translate into forward knee travel (the goal being to balance the center of mass over the foot, not to push it as far forward as possible). In short, this is a 2-dimensional model of a squat from the side, at a narrow parallel stance, under the worst-case assumption that all extra femur length is added to forward knee travel with no additional trunk lean; it assumes both individuals squat to the same depth with the same trunk angle. The calculation below is therefore the maximum dorsiflexion that could be required, and the true requirement for a longer femur is likely significantly less.

We chose the worst case deliberately. The recommendation that follows depends on whether the dorsiflexion implied by the model can be achieved through training or manual therapy, and we wanted that recommendation to hold even under the least favorable assumptions. In short, we did not assume the smallest possible demand and then claim it was achievable only through cutting-edge programming available with a Brookbush Institute membership. We assumed the greatest possible demand and recommended only what peer-reviewed, published research has demonstrated as an average improvement.

Every value in the geometric model came from Section 1.2 of the systematic review below: the femur-to-tibia data (Aitken et al., 1,077 limbs). The average femur-to-tibia ratio was 1.28, with a standard deviation of 0.04. The average femur was 50.0 cm, and the average tibia was 39.0 cm, for a total leg length of 89.0 cm. We compared 2 individuals of the same height, and therefore the same leg length (89.0 cm). One had average proportions (a femur-to-tibia ratio of 1.28), and the other had a long femur. Because their total leg length was the same, a longer femur meant a shorter tibia by the same amount. This keeps the comparison fair: the "long-femur" person did not have longer legs, only a longer femur within the same-length leg.

The steps below calculate the extra dorsiflexion required at any point in the population, using a femur 2 standard deviations above average as the worked example. The table that follows applies the same steps across the full range.

1. Determine the ratio at the chosen number of standard deviations above the mean. The femur-to-tibia ratio is a plain ratio with no units (a femur 1.28 times as long as the tibia gives a ratio of 1.28), with a standard deviation of 0.04. For example, a ratio 2 standard deviations above the mean is 1.36, found by adding 2 standard deviations to the mean: 1.28 + 0.04 + 0.04 = 1.36. Fewer than 1 in 40 adults have a ratio this high.
2. Determine the bone lengths of the 2 individuals. Based on averages from the research study by Aitken et al., the average leg length is 89.0 cm (femur + tibia). For the average person (ratio 1.28), the tibia = 89.0 ÷ (1 + 1.28) = 39.0 cm, and the femur = 89.0 − 39.0 = 50.0 cm. For the long-femur person (ratio 1.36), the tibia = 89.0 ÷ (1 + 1.36) = 37.7 cm, and the femur = 89.0 − 37.7 = 51.3 cm. The long femur was 1.3 cm longer, and its tibia was 1.3 cm shorter.
3. Find how far the average person's knee traveled forward of the ankle. We first need a starting dorsiflexion angle, which sets how far the knee is already forward before we add any extra femur length. We used 30° from vertical at the bottom of a squat, a value reported in the research below (for example, Fuglsang et al. measured a shank angle equivalent to about 31° from vertical at parallel depth). The knee-forward distance (side x) is the tibia length times the sine of that angle: x = 39.0 × sine(30°) = 39.0 × 0.5 = 19.5 cm.
4. Add the extra length of the long femur. For example, the long-femur person's knee would travel that same 19.5 cm plus the extra 1.3 cm of femur length, for a total of 20.8 cm. But this distance was covered by a shorter tibia (37.7 cm).
5. Find the long-femur person's dorsiflexion angle. Using the formula dorsiflexion angle = arcsine (knee-forward distance ÷ tibia length): arcsine (20.8 ÷ 37.7) = arcsine (0.552) = 33.5°.

The average person needed 30° of dorsiflexion. The long-femur person needed 33.5°. The difference was about 3.5° of extra dorsiflexion, for a proportion reached by fewer than 1 in 40 adults, and calculated under the assumptions that produce the largest possible number.

A femur-to-tibia ratio at or below +2 standard deviations accounts for 97.7% of adults, and every ratio in that range required 5° or less of extra dorsiflexion. The most extreme femur reported in the available research (a ratio of 1.39, the highest in 1,077 limbs, about 1 in 300) required 4–6°, and a ratio 3 standard deviations above the mean (about 1 in 740, or 0.13% of adults) required 4–7°.

Most importantly, that entire range, out to the third standard deviation and 99.9% of the population, is smaller than the increase in dorsiflexion that research demonstrated an integrated, multi-week program can produce. In Section 3.2 below, an 8-week program (Sohrabi et al.) and a 5-week program (Georgoulas et al.), both combining soft tissue release or joint mobilization with stretching and strengthening, increased dorsiflexion by 5 – 7°. Both programs included self-administered mobilization, and the improvements were sustained or grew during a 3-month follow-up (in Georgoulas et al., the advantage of adding mobilization over exercise alone grew from 3.5° at 5 weeks to 5.2° at 3 months). In other words, the extra dorsiflexion demanded by virtually any femur length can be achieved within 5 to 8 weeks. An example routine is provided below.

Ankle Dorsiflexion, Not Femur Length, Predicts How You Squat

The trigonometry above establishes that a longer femur can be accommodated by only a few degrees of additional dorsiflexion. However, research demonstrates a stronger reason for attempting to improve dorsiflexion: ankle dorsiflexion predicts squat form, and femur length does not. First, during loaded squats, trunk position was not correlated with any segment ratio (femur-to-tibia, torso-to-femur, or thigh-to-shank), but ankle dorsiflexion ROM was correlated (Fuglsang et al., Section 2.1; Berglund et al., Section 2.1). Second, when researchers placed a wedge under the foot that forced the ankle into dorsiflexion at the start of the squat, using up 7.5° of the dorsiflexion available during the movement, peak knee flexion dropped by 15.6°, the center of mass shifted backward, quadriceps activity decreased, and the knee collapsed inward (medial knee displacement increased sharply); hip motion did not change (Macrum et al., Section 3.1). In other words, using up 7.5° of available dorsiflexion produced the exact pattern blamed on longer femurs, and the participants were healthy individuals with normal ankle range, so the change in their squat came from the restriction alone, not from their anatomy. An additional study demonstrated that individuals assessed with limited dorsiflexion ROM during a weight-bearing lunge test also squatted with 6° to 8° less dorsiflexion and 12° to 15° less knee flexion when compared to individuals with normal dorsiflexion (Dill et al., Section 3.1). In summary, these studies demonstrate that dorsiflexion ROM is correlated with squat form and segment ratios are not, and restrictions in dorsiflexion are correlated with the same changes in squat form often attributed to long femurs.

This does not mean the ankle is the only thing worth assessing. A full movement assessment can reveal other limitations, at the hip, knee, or spine, that also change a squat, and those should be addressed when found. But when researchers compared ankle dorsiflexion against a range of other mobility and strength measures, dorsiflexion was still the strongest single predictor of how deep a person squatted, ahead of hip mobility and hip and ankle strength (Kim et al., Section 3.1). So a full assessment is the ideal; in its absence, dorsiflexion is the most defensible first target.

Ankle Dorsiflexion Can Be Measured and Improved

Femur length cannot be reliably measured in a fitness facility or clinic, and it cannot be changed (at least without surgery). Ankle dorsiflexion can be both reliably measured and changed, and in the absence of a movement assessment, it is the target with the highest likelihood of improving squat form and depth. (Note that movement assessment may imply that other joint restrictions are present, and they should be the focus of an intervention plan. Again, no reliable movement assessment or clinical test measures segment length.)

Dorsiflexion can be measured with the weight-bearing lunge test, the same assessment that separated the groups in the previously discussed research. The lifter places one foot a set distance from a wall and drives the knee forward over the toes, keeping the heel down, until the knee touches the wall or the heel begins to lift. The distance from the wall (ideally 4-5 inches, 10-12 cm) or the angle of the shin (ideally 40+°) is the score. The test is reliable (in the research, examiners had near-perfect agreement) and requires no special equipment (Dill et al., Section 3.1). Alternatively, active-assisted ROM measured with goniometry in supine is a great test for tracking improvement over time (see the video here: Dorsiflexion Goniometry ).

Dorsiflexion can also improve, although not every method is effective or equally effective. The research drew a clear line between what worked and what did not. A single session of any one technique, soft tissue work or stretching alone, produced a change too small to distinguish from measurement error (Rowlett et al., Section 3.2). A 12-week program of stretching and eccentric calf loading, without soft tissue release or joint mobilization, did not increase dorsiflexion at all (Lagas et al., Section 3.2). The methods that worked shared a feature: they combined several interventions over several weeks. An 8-week program that combined soft tissue release, joint mobilization, stretching, and strengthening increased dorsiflexion by about 7° (Sohrabi et al., Section 3.2), and a 5-week program that added joint mobilization to eccentric loading and stretching increased it by about 5° more than the same program without mobilization (Georgoulas et al., Section 3.2). In both, the improvements were maintained or continued to increase when measured again during a 3-month follow-up. (Sample program below)

Those numbers close the argument that the trigonometry section opened. The most extreme femur lengths in the population only required an additional 3-6° of dorsiflexion ROM (Section 1.4). An integrated, multi-week program increased dorsiflexion by 5° to 7° and at least maintained that improvement (Sohrabi et al., Section 3.2; Georgoulas et al., Section 3.2). In other words, the entire range of femur proportions ever documented, out to the rarest 1-in-300 outlier, demanded less ankle mobility (3-6°) than a person can add in about 5 to 8 weeks of the right training (5-7°). The trait the myth treats as a fixed structural barrier is smaller than what a lifter can change in less than 2 months.

Fixing Dorsiflexion Improves More Than Your Squat

Another reason to correct dorsiflexion rather than work around a suspected long femur: the correction transfers to other movements, and squat-form modifications do not.

A squat-form modification changes only the squat. Elevating the heels, widening the stance, or holding the load anteriorly reduces the dorsiflexion one exercise demands, but it does not change the presence of limited dorsiflexion, loss of hip internal rotation, gluteus medius weakness, pes planus, right-to-left asymmetry, knee valgus, and so on. As the research reviewed above suggests, in the absence of other assessment findings, addressing dorsiflexion is the most likely method for improving squat form. Any other movement that requires the shin to travel forward over the foot, such as a lunge, a step-down, or even walking, running, and jumping, still requires optimal dorsiflexion for optimal motion, so the same limitation reappears. Restoring dorsiflexion, by contrast, increases the range available to every one of those movements at once. This is the difference between addressing the quality of motion with corrective interventions and modifying an exercise: corrective interventions improve every pattern involving the same joints, rather than aiding compensation during one lift.

This also resolves the risk of a wrong guess. Because femur length cannot be reliably measured (discussed above), a lifter who attributes their squat to a long femur is guessing. If the guess is wrong, and the true limitation is the ankle, then correcting dorsiflexion addresses the actual problem. If the guess is right, and a long femur does demand a few additional degrees, the research has demonstrated that more than enough dorsiflexion can be attained (Section 1.4; Section 3.2). Correcting dorsiflexion is therefore the appropriate first target whether or not someone has a relatively long femur.

Caption: Dorsiflexion mobility has a stronger correlation with squat depth than femur length.

When Should You Modify the Squat? (Crutches, Not Corrections)

The modifications that reduce a squat's dorsiflexion demand are useful, but only as a temporary bridge while dorsiflexion is being corrected, not as a permanent answer to a suspected long femur. Each reduces the range of motion required by the ankle and other joints during squats, so each carries a cost if it becomes the lifter's permanent strategy. Ranked from most to least useful:

Anterior load first. Holding the load in front of the body (a goblet squat, front squat, or a counterbalance held at arm's length) shifts the center of mass forward, which allows the hips to sit back and the torso to stay more upright at the same depth, reducing the dorsiflexion the squat demands. It is the preferred first option for a reason beyond mechanics: bracing against an anterior load requires more from the trunk, so the modification that allows time to attain adequate dorsiflexion at the ankle may also aid in core muscle development, rather than simply hiding the limitation.

Heel elevation second. Raising the heels (a wedge, plate, or lifting shoe) pre-positions the ankle in dorsiflexion, so it takes less additional range to reach depth. It is effective and low-risk in the short term, and raising the heel can be iteratively reduced as dorsiflexion improves. A lifter who always squats on a wedge and does not attempt to improve dorsiflexion or reduce the wedge height never uses or strengthens the dorsiflexion required for optimal motion, so the limitation persists in every unassisted movement.

A modest widening of stance, third and with limits. A modest widening of stance, third and with limits. Widening the stance reduces the dorsiflexion a squat requires (Demers et al., Section 2.1). A widening toward shoulder width is a reasonable adjustment. A very wide, heavily turned-out stance is not, because it trades the dorsiflexion demand for other positions that have been correlated with dysfunction. Although the supporting studies are beyond the scope of this review, several additional courses have reviewed research correlating knees bowing out (varus), feet turning out, and collapse of the arch (pes planus), with impairment, pain, and dysfunction, just as reduced ankle dorsiflexion is itself associated with pain and dysfunction. These risks are probabilistic. A lifter may remain pain-free for years with a wide, turned-out stance, but that does not mean the increased risk is absent. It seems irrational to knowingly accept a higher risk when an effective, relatively simple, and easily accessible (self-managed) option for reducing it exists. For more on assessing signs of dysfunction and the research correlating those signs with risk of pain, dysfunction, and injury, start with our courses on the Overhead Squat Assessment .

The common thread is that every modification reduces what the squat asks of the ankle without changing what the ankle can do. That is exactly why they are bridges or crutches. The can be used while additional dorsiflexion is acquired, to allow the lifter to keep training productively. But, if used long-term, they limitation remains for squats, and all other movement patterns that require optimal dorsiflexion. The goal is to need them progressively less; for the 8 weeks or less, it will likely take an integrated strategy to attain the necessary dorsiflexion, and then the modifications should not be needed at all.

Sample Programs

Self-administered Corrective Exercise Program for Excessive Forward Lean

• Load: Light to Moderate (60 - 85% of 1-RM), focus on form and ROM for all activation exercises
• Reps/set: 12 - 20 reps-to-failure/set
• Sets/exercise (circuits): 1-2 sets
• Frequency: Frequency is likely more important than intensity. 1 set/2x day may be ideal.
• Rest between exercises: Circuit training, 1 min rest between exercises
• Training Time: 15 - 20 minutes

1. Release: Self-administered release of the gastrocnemius and soleus
2. Release: Self-administered release of the TFL
3. Mobilization: Self-administered ankle mobilization
4. Lengthening: Calf and fibularis static stretching
5. Lengthening: Kneeling hip flexor stretch
6. Isolated Activation: Tibialis anterior activation (unequipped)
7. Isolated Activation: Quick glute activation  (progress by increasing band resistance)
8. Subsystem Integration: Squat to Row

Manual Therapy/Clinical Intervention (Licensed Manual Therapists Only) for Excessive Forward Lean

• Manual Release
  • Static manual release of the gastrocnemius and soleus
  • Static manual release of the fibularis muscles
  • Static manual release of the extensor hallucis longus (EHL) and extensor digitorum longus (EDL)
  • Static manual release of the flexor hallucis longus (FHL) and flexor digitorum longus (FDL)
  • Static manual release of the tensor fascia latae
  • Static manual release of the gluteus minimus
• Mobilization or Manipulation
  • Ankle Mobilization
  • Mobilization of the proximal tibiofibular joint
  • Mobilization of the distal tibiofibular joint
  • Talonavicular and Cuneonavicular Joint Manipulation
  • Cuboid manipulation
• Manual Lengthening
  • Manual static stretch of the calf
  • Manual static stretch of the hip flexors
• Instrument Assisted Soft Tissue Mobilization
  • Plantar Fascia IASTM
  • Crural Fascia IASTM
• Activation (same acute variables as above)
  • Tibialis anterior activation with band
  • Quick glute activation  (progress by increasing band resistance)
  • Squat to Row
• Taping:
  • Tibialis anterior activation taping

Caption: Calf release is part of an optimal program for improving dorsiflexion.

Systematic Review

Systematic Research Review Summary of Findings

Summary of Research Findings: Femur length varies less between adults than commonly assumed, cannot be measured accurately without imaging, and predicts squat form less than the modifiable factor of ankle dorsiflexion ROM. All available research showed that femur length was 26.5% to 27.0% of height when measured directly, was the same in males and females, and varied between individuals with a standard deviation of about 0.4% to 0.6% of height. Additionally, femur length was 1.26 to 1.28 times the length of the tibia, with a standard deviation of 0.04 to 0.05. Last, the torso-to-leg ratio varied about twice as much as the femur-to-tibia ratio (with a standard deviation of 4% to 8% of its mean, when compared to 3% to 4%). For example, for adults of the same height (5'9", 175 cm), 95% have femurs within 4 cm or less (less than 2 inches) of each other, with the largest observed difference in any study being 3.7 cm (1.5 inches); the tibia would exhibit the opposite change in length with a range from 34.8 to 38.5 cm (13.7 to 15.2 inches), and torso lengths may differ by up to 13 cm (5 inches).

Identification of individuals with long femurs may not be possible clinically, as a research study demonstrated that palpation of the greater trochanter by expert examiners missed by 24 mm on average (verified by imaging), with a consistent bias of 22 mm upward, and the range between examiners (48 mm) exceeded the difference in femur length that would account for the longest-femured and shortest-femured adults of similar height.

Note that the changes attributed to longer femurs have been correlated with a reduction in dorsiflexion. Experimentally reducing available dorsiflexion by 7.5° reduced peak knee flexion by 15.6°, shifted the center of mass backward, and increased medial knee displacement, with no change at the hip. Additionally, individuals with limited dorsiflexion on the weight-bearing lunge test squatted with 12 - 15° less knee flexion and 6 - 8° less dorsiflexion than individuals with normal ROM. Note that longer femurs correlated with fewer squat repetitions in males only; however, this finding could not be separated from body mass, as the study did not investigate relative femur length (only absolute femur length).

Last, a single technique in a single session, or stretching and eccentric calf loading over several weeks, may be insufficient: a single session of soft tissue mobilization or stretching increased dorsiflexion by less than the test's measurement error, and a 12-week program of stretching and eccentric loading did not increase dorsiflexion. However, a multi-week program that combined release, joint mobilization, stretching, and strengthening increased dorsiflexion by 5 - 7°, with significant carry-over at follow-up (3 months in the study).

• Section 1: Femur Length Relative to Height: Femur length was 26.5% to 27.0% of standing height in adults when the bone was measured directly, and the value was the same in males and females, in samples from Europe, North America, and a worldwide aggregate of 13,149 people. Tape measurements from the greater trochanter to the lateral femoral epicondyle reported 23.3% to 24.1% in 4 of 4 studies (the lower number was explained by the tape measure method not measuring from the true ends of the bone). The study-to-study range of means was less than 1% of the height for both methods. The individual standard deviation of the ratio was approximately 0.4% to 0.6%, so 95% of adults had a femur between 25.5% and 27.9% of their height, a difference in femur length of 4 cm (less than 2 inches) in a 175 cm individual (approximately 5'9"). Taller adults had slightly longer legs for their height, and the extra length was in the tibia rather than the femur. The mean was well established; the individual variability rested on 2 sources totaling 113 people and should be stated with that caveat.
• Section 2: Relative Segment Length and Squat Kinematics: Ankle dorsiflexion ROM had a stronger correlation with knee, ankle, and trunk motion during squats than femur (thigh) length, tibia (shank) length, or their ratio. In loaded squats, no segment ratio correlated with trunk angle or lumbopelvic flexion. Dorsiflexion ROM correlated with both, explaining 45% of the difference between lifters in trunk angle and 21 - 27% of the difference in lumbopelvic flexion. In unloaded squats, a relatively longer thigh correlated with more knee flexion and more dorsiflexion at narrow and medium stances (explaining 12 - 23% of the difference between individuals), but did not correlate with hip flexion (forward lean) at any stance, and did not correlate with any joint angle with a wide stance (200% of pelvic width). With a wide stance, less dorsiflexion ROM was required. A relatively longer tibia, not a longer femur, correlated with knee abduction and internal rotation at depth. Taller males squatted with more forward lean and taller females squatted more upright with less knee flexion in the 1 study that examined height by sex, with effect sizes not reported. Last, longer femurs correlated with fewer squat repetitions in males but not in females, but the effect could not be separated from body mass because the lifters in these studies with longer femurs were also the heavier lifters (implying a difference in absolute femur length and lifter size and not just relative length).
• Section 3: Ankle Dorsiflexion Range of Motion and Squat Mechanics: Restricting ankle dorsiflexion range of motion (ROM) changed squat mechanics, and dorsiflexion ROM increased most following multi-week, integrated programs that included joint mobilization (supervised, self-administered, or both). A wedge that reduced available dorsiflexion ROM by 7.5° reduced peak knee flexion by 15.6°, shifted the center of mass backward, reduced quadriceps activity, and increased medial knee displacement, with no change in hip ROM. This experiment verified the effects of reduced dorsiflexion on squat mechanics. Additionally, individuals with limited dorsiflexion on the weight-bearing lunge test squatted with 12 - 15° less knee flexion and 6 - 8° less dorsiflexion when compared to individuals with normal ROM. Dorsiflexion ROM by itself accounted for 21 - 38% of the difference in squat depth between individuals, more than femur proportions accounted for in any joint angle investigated in Part 2 of this review. A single session of soft tissue mobilization or stretching increased dorsiflexion significantly when compared to a control group, but by less than the test's measurement error; however, a single session of a 4-phase inhibit-lengthen-activate-integrate sequence, in a separate study, increased dorsiflexion by an amount equal to measurement error. Further, programs of 5 to 8 weeks that combined soft tissue release or joint mobilization with stretching and strengthening increased dorsiflexion by 5 - 7°, and the increase was maintained or grew at 3-month follow-up. Note that a 12-week program of stretching and eccentric calf loading, without soft tissue release or joint mobilization, did not increase dorsiflexion

Part 1: How Much Do Human Proportions Vary?

1.1 Femur Length Relative to Height

• Summary Statement: Femur length was 26.5% to 27.0% of standing height in adults when the bone was measured directly, and the value was the same in males and females, in samples from Europe, North America, and a worldwide aggregate of 13,149 people. Tape measurements from the greater trochanter to the lateral femoral epicondyle reported 23.3% to 24.1% in 4 of 4 studies (the lower number was explained by the tape measure method not measuring from the true ends of the bone). The study-to-study range of means was less than 1% of the height for both methods. The individual standard deviation of the ratio was approximately 0.4% to 0.6%, so 95% of adults had a femur between 25.5% and 27.9% of their height, a difference in femur length of 4 cm (less than 2 inches) in a 175 cm individual (approximately 5'9"). Taller adults had slightly longer legs for their height, and the extra length was in the tibia rather than the femur. The mean was well established; the individual variability rested on 2 sources totaling 113 people and should be stated with that caveat.

Femur length averages 26.5% to 27.0% of standing height when the bone is measured directly

Four studies measured femur length directly (skeletal, cadaver, or radiographic) alongside stature: 1 aggregate analysis of 13,149 individuals in 51 population samples, 1 radiographic analysis of 4 child growth studies, 1 cadaver series of 91 adults, and 1 radiographic series of 145 adults.

• Feldesman et al. compiled published femur length and stature data for 13,149 individuals (gender distribution and ages not verified) from 51 population samples. Outcome measures included maximum femur length, stature, and the femur-to-stature ratio for each sample and for all samples combined. As characterized by the abstract, the findings demonstrated a mean femur-to-stature ratio of 26.74%, statistically similar ratios for males and females, and a narrow range of variation across the 51 population means. An analysis of variance across 3 ethnic groupings was statistically significant, with post hoc testing attributing the difference to 1 group. Note that the reported range describes the spread of population means and not the spread of individuals. Note that the individual-level standard deviation is not reported in the abstract; individual variability is drawn from Hauser et al. (2005) and the Tupman data in Feldesman (1992) (??).
• Feldesman, M. R. compiled radiographic femur lengths and standing statures for males and females (age: 8 to 18 years) from 4 longitudinal child growth studies: Maresh (1970; 140 children, age-class means from age 10), Anderson et al. (1963; 50 males and 50 females, age-class means), Anderson et al. (1964; 67 males and 67 females, age-class means, 51 overlapping with the 1963 sample), and Tupman (1962; individual data for 11 males and 11 females measured at 2 time points approximately 1 year apart). Femur lengths were corrected for radiographic magnification (factors of 0.954 to 0.968 for teleoroentgenography, 0.985 for scanography, none for orthoroentgenography) and for landmark differences (1.005 for bicondylar length, 1.018 for length to the lateral condyle) before ratios were calculated as femur length divided by stature times 100. Outcome measures included the femur-to-stature ratio by age class and gender for each study and for the pooled sample (77 age-class data points; 39 male, 38 female), comparisons with the adult ratio of 26.74%, and stature predictions for 1 autopsy specimen and 3 Tupman males using the adolescent ratio, the adult ratio, and the Trotter and Gleser regression equations. The findings demonstrated pooled means of 27.19 ± 0.47% for males, 27.05 ± 0.29% for females, and 27.12 ± 0.40% combined (standard deviations of age-class means, not of individuals). Males and females were statistically similar in the pooled sample (p = 0.14), and both differed significantly from the adult ratio (p less than 0.001). The ratio differed significantly across age classes (p less than 0.001), and post hoc comparisons attributed the difference to the 8- to 11-year-old classes; from age 12 to 18, age had no significant effect (p = 0.28). In a 2-way analysis of variance on the age 12 to 18 data, gender had a significant effect (p less than 0.001), and the ratio averaged 27.44% for males and 27.16% for females. The ratio increased from age 8 to a peak at age 12 in females (27.35%) and age 14 in males (27.56%), then decreased toward the adult value at age 18 (26.95% females; 27.24% males). In the Tupman individual data, ratios ranged from 25.95% to 27.86% at the first measurement and 26.01% to 28.08% at the second; the 2 measurements were statistically similar within individuals (p = 0.05), and within-individual means (26.48% males; 27.09% females) were statistically similar to the adult ratio (p = 0.32 and 0.09). In the 3 Tupman males, the adolescent ratio under-predicted stature by 4.0 cm on average, the adult ratio under-predicted by 2.1 cm, and the Trotter and Gleser equations over-predicted by 6.1 cm. Note that BI computed a mean of 26.81% and a standard deviation of 0.62 percentage points from the 22 first-measurement Tupman values; this is the only individual-level distribution of the ratio in the Feldesman publications supplied. Note that the paper restates 3 results of Feldesman et al. (1990): the adult ratio of 26.74% was derived from more than 13,000 individuals in 51 skeletal samples, male and female ratios were statistically similar, and an analysis of variance across 3 ethnic groupings was significant (0.01 less than p less than 0.05) with the difference attributed to 1 group (??).
• Hauser et al. investigated 91 cadavers from the contemporary Polish population, including 71 males (age: 19 to 87 years; body length 157.5 to 192.7 cm) and 20 females (age: 28 to 74 years; body length 155.7 to 168.0 cm). Body length was measured in a supine position with steel squares and a tape measure, and both femora were dissected and measured for greatest length (femoral head to distal condyles) on an osteometric board to 0.1 mm. Outcome measures included body length, left and right femur length, the correlation between femur length and body length, regression equations for body length from femur length, and the standard deviation from the regression line (reconstruction error). The findings demonstrated a mean body length of 174.9 ± 7.9 cm for males and 162.3 ± 3.8 cm for females, and a mean femur length of 47.2 ± 2.5 cm for males and 43.0 ± 1.3 cm for females. Femur length correlated with body length at r = 0.923 for males, r = 0.892 for females, and r = 0.950 for all cadavers. The reconstruction error was 3.0 cm for males, 1.8 cm for females, and 2.8 cm for all cadavers. Note that BI computed femur-to-body-length ratios of 27.0% for males and 26.5% for females from the reported means, and individual standard deviations of 0.56 and 0.36 percentage points from the reported means, standard deviations, and correlations. Note that supine cadaver body length was measured, not living standing stature (??).
• Paley et al. investigated 145 patients (109 males, 36 females) (age: mean 29.4 years) evaluated for stature-lengthening surgery because of height dysphoria, with no lower extremity deformity, arthritis, or limb length discrepancy of more than 10 mm. All participants underwent standing full-body biplanar (EOS) radiography. Outcome measures included foot height, tibial length (ankle joint line to knee joint line), femoral length (knee joint line to top of the femoral head), upper body length, total height, the regression of femoral length on foot height plus tibial length, and the regression of upper body length on tibial plus femoral length. The findings demonstrated a mean femoral length of 43.1 ± 2.7 cm, tibial length of 34.5 ± 2.4 cm, and height of 162.4 ± 7.4 cm. Femoral length correlated with foot height plus tibial length at r = 0.82. Upper body length correlated with tibial plus femoral length at r = 0.41 for all participants, r = 0.57 for females, and r = 0.20 for males, with a standard error of the estimate of 32.8 mm. Note that BI computed a femur-to-height ratio of 26.5% and a femur-to-tibia ratio of 1.25 from the reported means. Note that the sample was selected for perceived short stature and averaged 162 cm in height, which limits generalization to the adult population (??).

Conclusion: Femur length was 26.5% to 27.0% of standing height in adults. The value was consistent in a worldwide aggregate sample, a European cadaver series, and a North American radiographic series. Males and females did not differ by more than 0.5%. The ratio was slightly higher (27.2% to 27.4%) during adolescence and settled to the adult value by age 18.

• Feldesman, M. R., Kleckner, J. G., & Lundy, J. K. (1990). Femur/stature ratio and estimates of stature in mid- and late-Pleistocene fossil hominids. American Journal of Physical Anthropology, 83(3), 359–372.
• Feldesman, M. R. (1992). Femur/stature ratio and estimates of stature in children. American Journal of Physical Anthropology, 87(4), 447–459.
• Hauser, R., Smoliński, J., & Gos, T. (2005). The estimation of stature on the basis of measurements of the femur. Forensic Science International, 147(2–3), 185–190.
• Paley, D., Sutaria, S., Pinsky, D., Roberts, D., & Robbins, C. (2024). Is human height based on a Lucas sequence relationship between the foot height, tibial length, femur length and upper body length? Journal of Anatomy, 244(5), 861–872.

Tape-measured femur length averages 23.3% to 24.1% of height, and the value depends on the landmarks used

Five studies measured femur or thigh length with a tape measure or motion-capture sensors alongside standing height, in 113 students and 194 trained lifters. None was designed to report the ratio; the Brookbush Institute computed it from the reported means. Four of these studies were designed to test squat mechanics or repetition performance and are annotated in full here because their femur and height data are the only surface-measured proportion data in the set.

• Shella and Parinduri investigated 113 medical students (55 males, 58 females) (age: 21 to 23 years) from Universitas Muhammadiyah Sumatera Utara in Medan, Indonesia (exclusion criteria not reported). Standing height was measured with a microtoise, and femur length was measured with a tape measure from the greater trochanter to the lateral femoral condyle. Outcome measures included height, femur length, the Pearson correlation between femur length and height, and linear regression equations for height from femur length. The findings demonstrated a mean height of 161.9 ± 7.4 cm and a mean femur length of 38.1 ± 3.8 cm. Femur length correlated significantly with height at r = 0.46 for all participants, r = 0.53 for males, and r = 0.38 for females. Note that BI computed a femur-to-height ratio of 23.6% and a coefficient of variation for femur length of 10% from the reported means and standard deviations. Note that the values the paper labels as standard error of the estimate (0.164 to 0.272) are standard errors of the regression coefficients, not of the height estimate (??).
• Falch et al. compared 19 resistance-trained males (age: 24.3 ± 3.5 years; height 182 ± 7.3 cm; mass 87.1 ± 13.3 kg; 4.2 ± 2.5 years of training; squat 1-RM 146 ± 34.9 kg) and 17 resistance-trained females (age: 22.1 ± 3.0 years; height 166.1 ± 3.7 cm; mass 65.5 ± 5.6 kg; 4.7 ± 2.3 years of training; squat 1-RM 88.6 ± 17.3 kg) older than 18 years with at least 12 months of resistance training experience. Height, upper arm, lower arm, thigh (greater trochanter to distal lateral femoral condyle), and shank (distal lateral femoral condyle to lateral malleolus) lengths were measured 3 times to 0.1 cm with a tape measure, and body composition was measured by bioelectrical impedance. All participants performed a squat and bench press 1-RM test, followed by a reps-to-failure set with 60% of 1-RM loads in each exercise, during 1 session. Outcome measures included 1-RM strength, reps-to-failure at 60% of 1-RM, relative strength (1-RM divided by body mass), Pearson correlations between anthropometric variables and each strength measure for all participants and by sex, and Z-tests for differences in correlation coefficients between sexes. The findings demonstrated that males had significantly longer thighs (42.4 ± 2.4 versus 38.9 ± 3.2 cm) and shanks (43.9 ± 2.6 versus 39.7 ± 1.7 cm) than females, and statistically similar squat reps-to-failure (18.4 ± 3.2 versus 21.4 ± 5.8; p = 0.07). For all participants, squat 1-RM correlated positively with height (r = 0.66), thigh length (r = 0.46), and shank length (r = 0.64), and squat reps-to-failure correlated negatively with height (r = −0.41) and stance-width-to-height ratio (r = −0.41), and did not correlate significantly with thigh length (r = −0.18) or shank length (r = −0.26). In males, squat reps-to-failure correlated negatively with thigh length (r = −0.67; p less than 0.05); in females, squat reps-to-failure did not correlate with thigh length (r = 0.06), and the coefficients differed significantly between sexes. Note that BI computed thigh-to-height ratios of 23.3% for males and 23.4% for females from the reported means (??).
• Lee et al. investigated 50 resistance-trained right-leg-dominant males (age: 23.9 ± 4.5 years; height 175.7 ± 4.8 cm; mass 74.0 ± 7.8 kg; squat 1-RM 109.8 ± 16.1 kg) with at least 1 year of back squat training at moderate or higher intensity at least 3 sessions/week, and no history of lower extremity musculoskeletal injury or postural asymmetry within the past year. Thigh length (greater trochanter to lateral femoral epicondyle) and shank length (lateral femoral epicondyle to lateral malleolus) were measured 3 times by 1 examiner with a tape measure after marking landmarks. All participants performed 3 deep back squats (posterior thigh contacting the calf; knee flexion of at least 120°) with approximately 50% of 1-RM loads, feet at 100% to 120% of pelvic width with 15° to 30° of external rotation, a metronome-controlled tempo (seconds/phase not reported), on 2 force plates with 8-camera motion capture, during 1 session. Outcome measures included intra-rater reliability (ICC) of thigh and shank length, the shank-to-thigh length ratio, peak knee flexion, abduction/adduction, and rotation angles, peak anterior and medial knee displacement, peak knee moments in 3 planes and peak knee shear and compressive forces (normalized to body mass plus barbell load), and Pearson correlations and simple regressions between the length variables and the knee variables. The findings demonstrated a mean thigh length of 42.4 ± 2.5 cm (ICC 0.996), a mean shank length of 39.6 ± 2.1 cm (ICC 0.993), and a mean shank-to-thigh ratio of 93.7 ± 7.1%. Peak knee flexion averaged 130.0 ± 8.4°, peak anterior knee displacement averaged 17.3 ± 3.4 cm, and peak knee extension moment averaged 0.72 ± 0.28 N·m/kg. Thigh length correlated positively with peak anterior knee displacement (r = 0.537; R² = 0.289; p less than 0.001) and peak knee extension moment (r = 0.387; R² = 0.151; p = 0.005), and did not correlate with peak knee flexion. Shank length correlated negatively with peak knee adduction angle (r = −0.555; R² = 0.308; p less than 0.001), indicating more abduction with longer shanks, and positively with peak knee internal rotation angle (r = 0.327; R² = 0.107; p = 0.020). The shank-to-thigh ratio correlated negatively with peak knee adduction angle (r = −0.385; R² = 0.149; p = 0.006), negatively with peak knee extension moment (r = −0.390; R² = 0.152; p = 0.005), and positively with peak knee internal rotation angle (r = 0.426; R² = 0.181; p = 0.002). Remaining knee kinematic and kinetic variables did not correlate significantly with thigh length, shank length, or the shank-to-thigh ratio. Note that BI computed a thigh-to-height ratio of 24.1% from the reported means. Note that the intra-rater ICC reflects re-measurement between marked landmarks and does not test the accuracy of landmark placement over the bone (??).
• McKean and Burkett compared 16 males (age: 24.3 ± 5.1 years; height 179.4 ± 6.8 cm; mass 83.2 ± 12.2 kg) and 12 females (age: 24.2 ± 6.5 years; height 167.7 ± 5.3 cm; mass 62.0 ± 7.8 kg) with at least 12 months of regular squat training and no history of musculoskeletal injury. All participants performed 4 sets of 8 back squats during 1 session, in random order, including narrow stance (anterior superior iliac spine width) and wide stance (2 times anterior superior iliac spine width), each with body weight and with body weight plus 50% of body mass as barbell load, with a moderate (2 min) rest between sets, and 3-dimensional magnetic tracking sensors over the lower limbs and torso, from which thigh, shank, and torso lengths were also derived. Outcome measures included maximum hip, knee, shank, lumbar, and sacrum angles and their timing within the squat cycle, during the eccentric and concentric phases, and Spearman correlations of each angle with anterior superior iliac spine width, height, thigh length, shank length, torso length, and segment ratios (reported as the number of significant correlations, out of a possible 8 per pair, at p less than 0.05). The findings demonstrated that males had significantly longer thighs (42.0 ± 2.9 versus 39.0 ± 2.6 cm), shanks (40.6 ± 2.3 versus 38.3 ± 1.9 cm), and torsos (82.6 ± 5.0 versus 77.8 ± 3.8 cm), and a significantly higher femur-to-tibia ratio (1.04 ± 0.08 versus 1.02 ± 0.08) than females, and statistically similar torso-to-leg ratios (1.00 ± 0.08 versus 1.01 ± 0.05). Maximum knee angle, maximum sacrum angle, and maximum lumbar flexion angle differed significantly between males and females across stance, load, and phase (p less than 0.01). In males, maximum hip angle correlated negatively with height (3 of 8 eccentric and 2 of 8 concentric comparisons) and positively with torso length (2 and 2), and maximum lumbar angle correlated positively with shank length (3 and 3). In females, maximum hip angle correlated positively with height (2 eccentric and 3 concentric comparisons), maximum knee angle correlated positively with anterior superior iliac spine width (2 and 2), height (4 and 4), thigh length (4 and 4), and torso length (3 and 3), and maximum shank angle correlated negatively with height (2 and 2). Note that BI computed thigh-to-height ratios of 23.4% for males and 23.3% for females from the reported means. Note that the segment lengths were derived from sensor positions rather than palpated landmarks, and the resulting femur-to-tibia ratios (1.02 to 1.04) are not comparable to radiographic ratios. Note that results were reported as correlation counts rather than coefficients or effect sizes (??).
• Cooke et al. investigated 58 well-trained males and females (43 males, 15 females) (age: 23 ± 3 years; training age 5.5 ± 2.5 years; height 172.8 ± 8.8 cm; mass 80.7 ± 16.3 kg; body fat 11.0 ± 3.5%; squat 1-RM 142.2 ± 49.7 kg) who had performed the back squat at least 3 sessions/week for at least 2 years, with a squat 1-RM of at least 1.5 times body mass for males and at least body mass for females, and no contraindication to exercise. Height, body mass, 3-site skinfold body fat, and femur length were measured according to International Society for the Advancement of Kinanthropometry standards, with femur length measured with a tape measure from the marked trochanterion to the marked tibiale laterale on the right leg (mean of 2 measurements; intra-investigator technical error of measurement 1.37%). All participants then performed a squat 1-RM test, 2 single reps at each of 30%, 40%, 50%, 60%, 70%, 80%, and 90% of 1-RM loads with a very long (5 min) rest between reps, a 10 min rest, and 1 set of squats with reps-to-failure with 70% of 1-RM loads, during 1 session. Outcome measures included reps performed at 70% of 1-RM, Pearson correlations between reps performed and body mass, body fat, femur length, age, sex, 1-RM, and relative 1-RM, a forced-entry multivariate regression with relative importance analysis, and an independent t-test comparing reps performed by the 10 participants with the longest femurs and the 10 with the shortest femurs. The findings demonstrated a mean of 14 ± 4 reps (range 6 to 26) at 70% of 1-RM and a mean femur length of 47.1 ± 2.6 cm (47.3 ± 2.6 cm for males; 46.6 ± 2.6 cm for females). Reps performed correlated inversely with body mass (r = −0.352; p = 0.003), body fat (r = −0.278; p = 0.014), and femur length (r = −0.265; p = 0.019), and did not correlate significantly with age (r = −0.169), sex (r = 0.25), 1-RM (r = −0.146), or relative 1-RM (r = 0.089). The multivariate model explained 20.0% of the variance in reps performed (R² = 0.200); body mass exhibited a trend toward significance (p = 0.057) and contributed 43.87% of the explained variance, femur length contributed 16.45% and was statistically non-significant (p = 0.793), and no other variable was significant. The 10 participants with the longest femurs (51.0 ± 1.3 cm; 7 males, 3 females) performed significantly fewer reps (14 ± 3) than the 10 with the shortest femurs (43.3 ± 1.2 cm; 6 males, 4 females) (19 ± 6; p = 0.027). Note that femur length alone explained 9.5% of the variance in reps performed (R² = 0.0947). Note that BI computed a femur-to-height ratio of 27.3% from the reported means; this value exceeds other tape-measure studies because the tibiale laterale landmark lies on the tibia, approximately 4 cm distal to the lateral femoral epicondyle used in those studies (??).

Conclusion: Tape-measured thigh length was 23.3% to 24.1% of height in 4 of the 5 studies. These 4 studies all measured from the greater trochanter to the lateral femoral epicondyle (23.3%, 23.4%, 23.6%, and 24.1%). The 1 study that used a tibial landmark, rather than the lateral femoral epicondyle, reported 27.3%. The studies in Paragraph 1 used direct bone measurement and reported femur length as 26.5% to 27.0% of height. Surface measurement therefore reported a femur length that was 3% to 4% shorter than that reported by studies using direct bone measurement, and the difference was explained by the tape not measuring the true end-to-end length of the femur. The range of means in the 4 surface-measured studies (0.8%) was similar to the range of means in the bone-measured studies (0.5%). Both methods produced a mean that varied by less than 1% of height from study to study.

• Shella, R., & Parinduri, A. G. (2025). Determination of height based on estimated femur length in Medan City. Buletin Farmatera, 10(3), 239–247.
• Falch, H. N., Haugen, M. E., Larsen, S., & van den Tillaar, R. (2023). Association of strength performance in bench press and squat with anthropometric variables between resistance-trained males and females. Journal of Functional Morphology and Kinesiology, 8(1), 19.
• Lee, J., Kwon, M., & Park, J. (2025). Influence of thigh and shank lengths and ratios on kinematic and kinetic characteristics of the knee joint during barbell back squat. Applied Sciences, 15(17), 9448.
• McKean, M., & Burkett, B. J. (2012). Does segment length influence the hip, knee and ankle coordination during the squat movement? Journal of Fitness Research, 1(1), 23–30.
• Cooke, D. M., Haischer, M. H., Carzoli, J. P., Bazyler, C. D., Johnson, T. K., Varieur, R., Zoeller, R. F., Whitehurst, M., & Zourdos, M. C. (2019). Body mass and femur length are inversely related to repetitions performed in the back squat in well-trained lifters. Journal of Strength and Conditioning Research, 33(3), 890–895.

Individual femur-to-height ratio varies with a standard deviation of about 0.6 percentage points, and tall people have proportionally longer legs

Three studies provide individual-level variability in leg-segment-to-height ratio: 1 cadaver series with correlations from which the ratio's standard deviation can be derived, 1 individual-level adolescent radiographic sample, and 1 anthropometric series of 231 males divided by height. The first 2 are annotated in full in Paragraph 1; their variability data are restated here.

• Hauser et al. (annotated in full in Paragraph 1) reported means, standard deviations, and correlations for femur length and body length in 71 male and 20 female cadavers. Note that BI derived individual standard deviations for the femur-to-body-length ratio of 0.56 percentage points (males) and 0.36 percentage points (females) from the reported values, using the standard approximation for the standard deviation of a ratio of correlated variables (??).
• Feldesman, M. R. (annotated in full in Paragraph 1) reported individual femur-to-stature ratios for 22 adolescents in the Tupman sample, ranging from 25.95% to 27.86% at the first measurement. Note that BI computed a mean of 26.81% and a standard deviation of 0.62 percentage points from those values (??).
• Duyar and Pelin investigated 231 Turkish males (age: 18.0 to 34.3 years), including a study group of 121 and a control group of 110 recruited from a university population. Standing height and tibia length (medial condyle to medial malleolus) were measured with a Martin anthropometer. The study group was divided at the 15th and 85th height percentiles into short, medium, and tall groups. Outcome measures included height, tibia length, the tibia-to-height ratio by height group, the correlation between tibia length and height, general and height-group-specific regression equations, and the prediction error of each equation in the control group. The findings demonstrated a mean tibia length of 39.0 cm and a mean height of 174.6 cm for all participants. Tibia length correlated with height at r = 0.90. The tibia-to-height ratio was lowest for the short group (21.6%), higher for the medium group (22.3%), and highest for the tall group (23.1%), and the differences between groups were statistically significant (p less than 0.0001). The standard deviation of the ratio within each height group was approximately 0.7 percentage points. The general regression equation over-predicted the height of short males and under-predicted the height of tall males by approximately 3.5 cm, and the height-group-specific equations reduced this error. Note that BI computed the overall tibia-to-height ratio of 22.3% from the reported means (??).

Conclusion: The individual standard deviation of the femur-to-height ratio was approximately 0.4% to 0.6%, based on 2 sources totaling 113 people. At a mean of 26.7%, 95% of adults fell between about 25.5% and 27.9%, a spread of 4 cm in femur length in a 175 cm person. The tibia-to-height ratio showed a similar within-group standard deviation (0.7%) and increased by 1.5% from short to tall adults, implying that taller people had slightly longer legs for their height, with the extra length in the tibia. The variability estimate rested on fewer people than the mean estimate and should be stated with that caveat.

• Hauser, R., Smoliński, J., & Gos, T. (2005). The estimation of stature on the basis of measurements of the femur. Forensic Science International, 147(2–3), 185–190.
• Feldesman, M. R. (1992). Femur/stature ratio and estimates of stature in children. American Journal of Physical Anthropology, 87(4), 447–459.
• Duyar, I., & Pelin, C. (2003). Body height estimation based on tibia length in different stature groups. American Journal of Physical Anthropology, 122(1), 23–27.

1.2 Femur Length Relative to Tibia

Summary Statement: The femur was 1.26 to 1.28 times the length of the tibia in adults, with a standard deviation of 0.04 to 0.05, in 2 imaging studies totaling 1,455 limbs, and the largest study found no ratio outside 1.16 to 1.39 in 1,077 limbs. For 2 individuals of the same height (5'9", 175 cm) and the same leg length at the extreme ends of that observed range, the femurs differed by about 3.7 cm (1.5 inches) and the tibias by the same amount in the opposite direction, so the person with the "long femurs" had a femur of 48.4 cm (19.1 inches) with a tibia of 34.8 cm (13.7 inches), and the person with the short femurs was likely to have a femur of 44.7 cm (17.6 inches) and a tibia of 38.5 cm (15.2 inches). The ratio between the femur and tibia decreased slightly as the tibia lengthened, so taller adults got more of their leg length from the tibia than from the femur. Torso length was a larger source of individual difference than femur length: torso length correlated with leg length at only r = 0.41 (a correlation of 1.0 would mean leg length predicted torso length exactly, and 0.41 meant it explained about 17% of the difference between people; in males r = 0.20, about 4%), and the torso-to-leg ratio varied about twice as much from person to person as the femur-to-tibia ratio (4% to 8% of its mean against 3% to 4%), which, at the same height and leg length, implied torsos that differed by up to 13 cm (5 inches). If proportions explained why 2 lifters of the same height looked different at the bottom of a squat, the length of the torso is more likely the reason than the length of the femur.

The femur-to-tibia ratio averages 1.26 to 1.28 with a standard deviation of 0.04 to 0.05 on imaging

Three imaging studies report femoral and tibial length in the same limbs: 1 standing radiographic series of 1,077 limbs, 1 CT series of 378 limbs, and 1 standing radiographic series of 145 patients (annotated in full in 1.1 Paragraph 1).

• Aitken, S. A. investigated 1,077 lower limbs from 753 adults (459 males, age: 16 to 82 years; 294 females, age: 15 to 83 years; mean age approximately 51 years) retrieved from a hospital picture archiving system, with no history of lower limb trauma, tumor, congenital disorder, moderate-to-severe knee osteoarthritis, or open growth plates. All limbs were measured on standing full-length digital radiographs by 3 observers for reliability testing. Outcome measures included lower limb length, femoral length (top of the femoral head to the center of the medial femoral condyle), tibial length (center of the medial tibial plateau to the center of the tibial plafond), the femorotibial ratio, side-to-side length differences in 324 paired limbs, the correlation between tibial length and the femorotibial ratio, and inter-observer reliability (ICC). The findings demonstrated a mean lower limb length of 89.0 cm (range 70.2 to 103.9 cm), a mean femoral length of 50.0 cm (standard deviation 3.6 cm; range 39.3 to 58.4 cm), and a mean tibial length of 39.0 cm (standard deviation 3.0 cm; range 30.8 to 46.5 cm), all normally distributed. The femorotibial ratio averaged 1.28 (standard deviation 0.04; range 1.16 to 1.39) and was normally distributed. The femorotibial ratio was significantly higher for limbs with tibiae shorter than 39 cm (1.29) than for limbs with longer tibiae (1.27), and tibial length correlated inversely with the femorotibial ratio at r = −0.35 (p less than 0.001). The median side-to-side difference was 0.4 cm for lower limb length (maximum 1.8 cm), 0.3 cm for femoral length (maximum 1.2 cm), and 0.3 cm for tibial length (maximum 1.1 cm). Inter-observer ICC was 0.99 for femoral length and 0.98 for tibial length. Note that the study reported dispersion as 2 standard deviations (7.2 cm femur, 6.0 cm tibia, 0.08 ratio); BI halved these values to report 1 standard deviation (??).
• Strecker et al. investigated 355 adults (231 males, age: 18 to 78 years, mean 32.3 years; 124 females, age: 16 to 73 years, mean 35.8 years) from the University of Ulm, Germany, comprising healthy volunteers and the uninjured limbs of patients with a unilateral femoral or tibial fracture, with no history of trauma, infection, tumor, or congenital disorder in the measured limb. All limbs were measured for length and torsion on computed tomography by the Ulm method, with landmarks at the center of the femoral head, the center of the greater trochanter, the dorsal tangent of the femoral condyles, the tibial head, and the ankle joint line. Outcome measures included femoral length (511 femora), tibial length (513 tibiae), lower limb length (378 limbs), the femoral-to-tibial length ratio (378 limbs), side-to-side length differences in paired limbs, and femoral, tibial, and whole-limb torsion. The findings demonstrated a mean femoral length of 46.3 cm (standard deviation 3.2 cm; range 37.2 to 54.1 cm), a mean tibial length of 37.0 cm (standard deviation 2.8 cm; range 29.2 to 43.7 cm), and a mean lower limb length of 83.2 cm (standard deviation 5.7 cm; range 67.9 to 96.7 cm), all normally distributed with statistically similar values for left and right sides. The femoral-to-tibial length ratio averaged 1.26 (standard deviation 0.05). The side-to-side length difference at the 99th percentile was 1.2 cm for femora and 1.0 cm for tibiae. Femoral torsion averaged 24.1° internal, tibial torsion averaged 34.9° external, and femoral torsion did not correlate with tibial torsion. Note that the study reported dispersion as 2 standard deviations; BI halved these values to report 1 standard deviation. Note that femoral length measured from the center of the femoral head is approximately 4 cm shorter than femoral length measured from the top of the femoral head (Aitken, 2021), which accounts for the difference in absolute means between the 2 studies (??).
• Paley et al. (annotated in full in 1.1 Paragraph 1) reported a mean femoral length of 43.1 ± 2.7 cm and tibial length of 34.5 ± 2.4 cm on standing biplanar radiographs in 145 patients selected for perceived short stature. Note that BI computed a femur-to-tibia ratio of 1.25 from the reported means (??).

Conclusion: The femur was 1.26 to 1.28 times the length of the tibia, with a standard deviation of 0.04 to 0.05, in 2 imaging studies totaling 1,455 limbs. In that range, 68% of adults fell between 1.24 and 1.32, 95% between 1.20 and 1.36, and 99.7% between 1.16 and 1.40; the largest study observed no ratio outside 1.16 to 1.39 in 1,077 limbs. This implied that two individuals of the same height (5'9", 175 cm), at the most extreme ends of the range observed in the largest study, would have a femur of 44.7 cm (17.6 inches) and 48.4 cm (19.1 inches), and a tibia of 38.5 cm (15.2 inches) and 34.8 cm (13.7 inches). The example used the femur-to-height estimate from Section 1.1 (26.7% of height, 46.7 cm at 175 cm), the mean ratio of 1.28 to set the tibia (36.5 cm), and held total leg length constant at 83.2 cm. The longest-femured adult in 1,077 limbs therefore had a femur about 3.7 cm (1.5 inches) longer, and a tibia about 3.7 cm shorter, than the shortest-femured adult of the same height. The ratio decreased slightly as the tibia lengthened, implying that long-legged adults (likely taller) received more of their leg length from the tibia than from the femur.

• Aitken, S. A. (2021). Normative values for femoral length, tibial length, and the femorotibial ratio in adults using standing full-length radiography. Osteology, 1(2), 86–91.
• Strecker, W., Keppler, P., Gebhard, F., & Kinzl, L. (1997). Length and torsion of the lower limb. Journal of Bone and Joint Surgery (British Volume), 79(6), 1019–1023.
• Paley, D., Sutaria, S., Pinsky, D., Roberts, D., & Robbins, C. (2024). Is human height based on a Lucas sequence relationship between the foot height, tibial length, femur length and upper body length? Journal of Anatomy, 244(5), 861–872.

Leg-to-torso proportion varies more than femur-to-tibia proportion

Four studies report a torso or upper-body length alongside leg segment lengths: 1 radiographic series (2 studies previously annotated), 1 motion-capture series (annotated in full in 1.1 Paragraph 2), and 2 squat-mechanics studies (annotated in full in Section 2.1).

• Paley et al. (annotated in full in 1.1 Paragraph 1) reported that upper body length correlated with tibial plus femoral length at r = 0.41 for all 145 participants, r = 0.57 for females, and r = 0.20 for males, with a standard error of the estimate of 32.8 mm, against r = 0.82 for femoral length predicted from foot height plus tibial length (??).
• McKean and Burkett (annotated in full in 1.1 Paragraph 2) reported a torso-to-leg ratio of 1.00 ± 0.08 in males and 1.01 ± 0.05 in females from motion-capture-derived segment lengths (??).
• Fuglsang et al. investigated 11 recreationally trained males (age: 22.9 ± 1.8 years; range 19 to 26 years; height 1.80 ± 0.04 m; mass 78.4 ± 4.7 kg) from Aarhus University with an average of 4.1 years (range 1.5 to 8) of heavy resistance training experience including the back squat, at least 2 training sessions/week, and no injury in the 6 months before testing; 3 additional recruits were excluded for inability to reach parallel depth under the test conditions. Reflective markers were placed bilaterally on the greater trochanter, lateral femoral epicondyle, lateral malleolus, and fifth metatarsal head, and on the center of the bar, and segment lengths were calculated from marker positions in the upright squat position (shank: lateral epicondyle to lateral malleolus; thigh: greater trochanter to lateral epicondyle; trunk: greater trochanters to bar). All participants performed a self-selected warm-up, 2 warm-up sets with depth checked by motion capture, and 3 continuous high-bar back squats to at least parallel (hip markers below knee markers) with approximately 75% of self-estimated 1-RM loads (102.7 ± 22 kg; estimated 1-RM 136.8 ± 24.4 kg), barefoot, feet pointing straight ahead at acromion-width stance, during 1 session with 8-camera motion capture. Participants then performed a weight-bearing lunge test on each ankle (3 reps, held 3 s, holding the squat rack for support) with motion capture recording maximal dorsiflexion as the anterior angle between the shank and horizontal. Outcome measures included shank length, thigh length, trunk length, shank-to-thigh, trunk-to-shank, and trunk-to-thigh ratios, shank angle in the lunge test, shank angle and trunk angle (anterior angle between the hip-to-bar line and horizontal) at the parallel position during the descent, the dorsiflexion deficit (lunge shank angle minus parallel-squat shank angle), and 4 regression models: lunge shank angle predicting parallel-squat shank angle (Model 1), lunge shank angle predicting trunk angle (Model 2), the 3 segment ratios predicting trunk angle (Model 3), and lunge shank angle plus the 3 ratios predicting trunk angle (Model 4). The findings demonstrated a mean shank length of 44.0 ± 1.9 cm, thigh length of 41.5 ± 1.9 cm, trunk length of 59.1 ± 2.3 cm, shank-to-thigh ratio of 1.063 ± 0.055, trunk-to-shank ratio of 1.344 ± 0.07, and trunk-to-thigh ratio of 1.427 ± 0.089. The shank angle averaged 47.8 ± 5.7° in the lunge test and 59.2 ± 4.8° at parallel, a deficit of 11.4 ± 4.4° that was independent of maximal range of motion, and trunk angle at parallel averaged 56.9 ± 5.4°. Lunge shank angle significantly predicted parallel-squat shank angle (R² = 0.41; adjusted R² = 0.35; p = 0.034). Lunge shank angle significantly and negatively predicted trunk angle (R² = 0.45; adjusted R² = 0.39; B = −0.64; p = 0.024), indicating that participants with a lower shank angle in the lunge test (more dorsiflexion) squatted with a higher trunk angle (more upright) at parallel. The 3 segment ratios together did not significantly predict trunk angle (R² = 0.37; adjusted R² = 0.10; p = 0.334), and no individual ratio reached significance (p = 0.139 to 0.152). The combined model of lunge shank angle plus the 3 ratios did not reach significance (R² = 0.60; adjusted R² = 0.33; p = 0.183), and within it no predictor was significant (p = 0.113 to 0.267). Post hoc statistical power was 0.886. Note that the single knee marker on the lateral epicondyle lengthened the measured shank relative to the thigh, which is why the shank-to-thigh ratio exceeds 1.0 and is not comparable to radiographic femur-to-tibia ratios. Note that the between-participant range of segment ratios was narrow (standard deviations of 0.055 to 0.089), which limits the power of the ratio models. Note that participants squatted with 11.4° less dorsiflexion than they demonstrated in the lunge test regardless of how much range they had (??).
• Berglund et al. investigated 18 competitive powerlifters and Olympic weightlifters (11 males, 7 females; 12 powerlifters, 6 weightlifters) (age: 24.9 ± 4.4 years; height 170.4 ± 7.5 cm; mass 79.2 ± 11.6 kg; 7.2 ± 4.5 years of training; squat 1-RM 135.3 ± 41.8 kg) from clubs in Umeå, Sweden, with at least 2 years of resistance training experience, training to compete or competing, and no current injury preventing full training (8 participants reported a current or previous low back or lower extremity pain or injury). All participants performed 3 back squats to competition depth with 70% of 1-RM loads during 1 session, with habitual stance, wrist wraps and shoes of choice, no belt, and inertial movement sensors over L2 and S2. All participants then underwent tape-measure anthropometry (femur: greater trochanter to lateral femoral condyle; tibia: medial tibial condyle to medial malleolus; upper body: seat to acromion in sitting), goniometric passive hip range of motion, a weight-bearing lunge test of ankle dorsiflexion (angle and toe-to-wall distance), and a 4-test lumbopelvic movement control battery, administered by 2 physical therapists. Outcome measures included lumbopelvic flexion (angular displacement between L2 and S2 from standing to the bottom position), squat depth, stance width, femur-to-tibia ratio, upper-body-to-femur ratio, passive hip flexion with the knee bent and straight, hip abduction, hip internal and external rotation, ankle dorsiflexion angle and distance for each side and both sides combined, movement control battery score (0 to 4), and univariate linear regression of lumbopelvic flexion on each independent variable. The findings demonstrated a mean lumbopelvic flexion of 18.4 ± 4.7° (range 9.3 to 27.9°) and a mean squat depth of 100.6 ± 13.4° of hip flexion. Lumbopelvic flexion did not correlate with squat depth (r = 0.29; p = 0.24). The femur-to-tibia ratio averaged 1.18 ± 0.06 (1.15 ± 0.04 for males; 1.22 ± 0.06 for females) and the upper-body-to-femur ratio averaged 1.35 ± 0.06. Passive hip flexion with the knee bent averaged 99.8 ± 11.3° (right) and 99.0 ± 6.7° (left), hip flexion with the knee straight averaged 75.2 ± 15.2° and 72.6 ± 17.3°, hip external rotation averaged 51.1 ± 9.3° and 53.6 ± 5.6°, hip internal rotation averaged 43.2 ± 10.1° and 41.7 ± 6.7°, and weight-bearing lunge dorsiflexion averaged 38.7 ± 8.1° (10.5 ± 3.6 cm) and 39.6 ± 8.7° (10.8 ± 3.4 cm). The movement control battery score averaged 2.72 ± 1.07 of 4 tests passed. Lumbopelvic flexion was statistically unrelated to the femur-to-tibia ratio (p = 0.75), the upper-body-to-femur ratio (p = 0.39), hip flexion with the knee bent or straight (adjusted R² −0.06 to 0.04), hip abduction, hip external rotation (adjusted R² −0.06 to −0.05), hip internal rotation (adjusted R² 0.08 to 0.11; p = 0.10 to 0.14), and movement control battery score (p = 0.68). Lumbopelvic flexion increased significantly with ankle dorsiflexion angle on the left (adjusted R² 0.21; p = 0.03), dorsiflexion distance on the right (adjusted R² 0.21; p = 0.03), dorsiflexion distance on the left (adjusted R² 0.27; p = 0.02), and combined dorsiflexion distance (adjusted R² 0.25; p = 0.02). Note that the direction of the dorsiflexion relationship was the reverse of the authors' stated hypothesis. Note that the femur-to-tibia ratio standard deviation of 0.06 in this sample is comparable to the population standard deviation reported on imaging (Aitken, 2021; Strecker et al., 1997), so the range of proportions tested was representative rather than restricted. Note that lumbopelvic flexion and trunk lean are separate outcomes; this finding does not conflict with the inverse relationship between dorsiflexion range and trunk lean reported by Fuglsang et al. (2017) (??).

Conclusion: Torso length was only moderately related to leg length. On imaging, upper body length correlated with femur plus tibia length at r = 0.41 in all participants and r = 0.20 in males, while femur length correlated with foot plus tibia length at r = 0.82 in the same people. The torso-to-leg ratio varied by 4% to 8% of its mean from person to person (8% and 5% in McKean, 6% and 5% in Fuglsang, 4% in Berglund), and the femur-to-tibia ratio varied by 3% to 4% of its mean (3% in Aitken, 4% in Strecker, from Paragraph 1). A person's torso length relative to their legs was therefore a larger source of individual difference than their femur length relative to their tibia, by roughly 2 to 1. In actual length, for 2 individuals of the same height (5'9", 175 cm) and the same leg length (83 cm), the 95% range of torso-to-leg ratio implied a torso about 13 cm (5 inches) longer in one than the other, against about 4 cm (1.5 inches) of femur for the femur-to-tibia ratio in Paragraph 1. This proportion, not femur length, was the anthropometric variable that varied most among lifters.

• Paley, D., Sutaria, S., Pinsky, D., Roberts, D., & Robbins, C. (2024). Is human height based on a Lucas sequence relationship between the foot height, tibial length, femur length and upper body length? Journal of Anatomy, 244(5), 861–872.
• McKean, M., & Burkett, B. J. (2012). Does segment length influence the hip, knee and ankle coordination during the squat movement? Journal of Fitness Research, 1(1), 23–30.
• Fuglsang, E. I., Telling, A. S., & Sørensen, H. (2017). Effect of ankle mobility and segment ratios on trunk lean in the barbell back squat. Journal of Strength and Conditioning Research, 31(11), 3024–3033.
• Berglund, L., Öhberg, F., Strömbäck, E., & Papacosta, D. (2024). Are anthropometric measures, range of motion, or movement control tests associated with lumbopelvic flexion during barbell back squats? International Journal of Sports Physical Therapy, 19(9), 1097–1107.
• Aitken, S. A. (2021). Normative values for femoral length, tibial length, and the femorotibial ratio in adults using standing full-length radiography. Osteology, 1(2), 86–91.
• Strecker, W., Keppler, P., Gebhard, F., & Kinzl, L. (1997). Length and torsion of the lower limb. Journal of Bone and Joint Surgery (British Volume), 79(6), 1019–1023.

1.3 Reliability and Accuracy of Surface Landmark Palpation

Summary Statement (Same as conclusion): Expert examiners mislocated the tip of the greater trochanter by 24 mm on average, only 1 of 30 landed within 10 mm of the ultrasound-confirmed point, the average mark landed 22 mm above and 8 mm in front of the true tip (a consistent bias, not random scatter), and the 95% range between examiners along the length of the femur was 48 mm (1.9 inches). In a second study, the greater trochanter and lateral femoral epicondyle, the 2 landmarks that define a tape-measured femur, were 2 of the 4 least consistently palpated landmarks of 12 tested, and consistency was worse in overweight participants. Re-measuring between marked stickers was highly repeatable (ICC 0.996 and 0.993, where 1.0 means identical measurements every time), and repeat measurement by 1 examiner varied by about 0.6 cm (1.37%), but repeatability between stickers said nothing about whether the stickers sat on the bone. Radiographic measurement was both repeatable and accurate (ICC 0.99 for the femur and 0.98 for the tibia between observers). In actual length, the 22 mm upward bias alone lengthened a tape-measured femur by about 2 cm, which on a 46.7 cm femur (the 175 cm individual from Section 1.1) was a 4% error, and it changed the femur-to-tibia ratio by 0.05 on a 39 cm tibia, equal to the entire population standard deviation of 0.04 to 0.05 from Section 1.2. The 48 mm range between examiners was larger than the 3.7 cm (1.5 inch) difference in femur length between the longest-femured and shortest-femured adults of the same height in Section 1.2. A tape measure could rank a group of lifters, and 2 tape protocols agreed on the group mean, but it could not classify an individual lifter as having long femurs, because the systematic error alone equaled 1 standard deviation of the trait and the random error between examiners exceeded the trait's full range at a given height.

Two studies tested examiner reliability and accuracy during palpation of landmarks used to measure femur length on the skin: 1 photogrammetric reliability study of 12 landmarks in 30 males, and 1 ultrasound-referenced accuracy study of 5 landmarks palpated by 30 expert hip surgeons. Three further studies (annotated in full previously) reported repeatability of tape or bone measurement and were restated for comparison.

• Moriguchi et al. investigated 30 healthy male university students (age: 18 to 30 years), including 15 with normal body mass index (age: 24.0 ± 3.3 years; height 172 ± 7 cm; mass 66.9 ± 7.6 kg) and 15 with overweight body mass index (age: 23.8 ± 3.1 years; height 176 ± 6 cm; mass 88.7 ± 9.5 kg), with no history of hypersensitivity or intolerance to manual palpation. Two trained physiotherapists each palpated and marked 12 anatomical landmarks (including the greater trochanter and lateral femoral epicondyle) with a fluorescent pen visible only under ultraviolet light, so that neither rater could see the other's marks, in standing, with landmark positions recorded by photogrammetry. Intra-rater comparisons were made between 2 palpation sessions by each rater, and inter-rater comparisons were made between raters. Outcome measures included intra-rater and inter-rater palpation discrepancy (mm) for each landmark, the resulting discrepancy in 8 postural angles, and a 10,000-iteration simulation of the angular variation produced by the observed palpation discrepancies. The findings demonstrated that inter-rater palpation discrepancy was larger than intra-rater discrepancy for all landmarks. Palpation discrepancy was largest for the greater trochanter, lateral femoral epicondyle, anterior superior iliac spine, and seventh cervical vertebra (values reported graphically on a scale reaching 40 mm; exact means not tabulated). Palpation discrepancy was significantly larger for the overweight group than for the normal-weight group. Palpation discrepancy at the anterior superior iliac spine and greater trochanter explained 46% to 55% of the variance in trunk flexion, hip flexion, and pelvic inclination angles (adjusted R² 0.46 to 0.55). Simulated angular variation from maximum palpation discrepancy remained within 5° for intra-rater and 10° for inter-rater comparisons for plantar flexion, knee flexion, and trunk inclination angles (??).
• Christoforetti et al. investigated 30 expert hip arthroscopists (instructors for the Arthroscopy Association of North America), each of whom palpated and labelled 5 surface landmarks on the same awake healthy adult male volunteer in a supine position. An expert musculoskeletal ultrasonographer then placed reference labels for the same landmarks under ultrasound guidance, and fixed-mount photographs of examiner and ultrasound labels were overlaid for measurement. Outcome measures included the distance from each examiner's label to the ultrasound label for the anterior superior iliac spine, tip of the greater trochanter, and 3 additional landmarks; the proportion of examiners within 10 mm of the ultrasound label; the mean direction and distance of the group's error (95% confidence interval); and the inter-observer 95% prediction interval. The findings demonstrated a mean error of 24 mm for the tip of the greater trochanter and 31 mm for the anterior superior iliac spine. One of 30 examiners (3%) placed the greater trochanter label within 10 mm of the ultrasound label. The group placed the greater trochanter label 22 mm superior and 8 mm anterior to the ultrasound label (95% confidence interval of the mean), and the 95% prediction interval for the greater trochanter spanned 33 mm anterior-to-posterior by 48 mm superior-to-inferior. The greater trochanter exhibited the smallest error of the 5 landmarks. Note that a 22 mm superior error at the proximal landmark would lengthen a tape-measured femur by approximately 5% of a 47 cm femur (??).
• Lee et al. (annotated in full in 1.1 Paragraph 2) reported intra-rater ICC of 0.996 for thigh length and 0.993 for shank length when 1 examiner re-measured 3 times between marked landmarks. Note that this reflects re-measurement between stickers and does not test whether the stickers sit on the bone landmarks (??).
• Cooke et al. (annotated in full in 1.1 Paragraph 2) reported an intra-investigator technical error of measurement of 1.37% (approximately 0.6 cm) for tape-measured femur length in 20 individuals measured on 2 visits (??).
• Aitken, S. A. (annotated in full in 1.2 Paragraph 1) reported inter-observer ICC of 0.99 for femoral length and 0.98 for tibial length on standing radiographs (??).

Conclusion: Expert examiners mislocated the tip of the greater trochanter by 24 mm on average, only 1 of 30 landed within 10 mm of the ultrasound-confirmed point, the average mark landed 22 mm above and 8 mm in front of the true tip (a consistent bias, not random scatter), and the 95% range between examiners along the length of the femur was 48 mm (1.9 inches). In a second study, the greater trochanter and lateral femoral epicondyle, the 2 landmarks that define a tape-measured femur, were 2 of the 4 least consistently palpated landmarks of 12 tested, and consistency was worse in overweight participants. Re-measuring between marked stickers was highly repeatable (ICC 0.996 and 0.993, where 1.0 means identical measurements every time), and repeat measurement by 1 examiner varied by about 0.6 cm (1.37%), but repeatability between stickers said nothing about whether the stickers sat on the bone. Radiographic measurement was both repeatable and accurate (ICC 0.99 for the femur and 0.98 for the tibia between observers). In actual length, the 22 mm upward bias alone lengthened a tape-measured femur by about 2 cm, which on a 46.7 cm femur (the 175 cm individual from Section 1.1) was a 4% error, and it changed the femur-to-tibia ratio by 0.05 on a 39 cm tibia, equal to the entire population standard deviation of 0.04 to 0.05 from Section 1.2. The 48 mm range between examiners was larger than the 3.7 cm (1.5 inch) difference in femur length between the longest-femured and shortest-femured adults of the same height in Section 1.2. A tape measure could rank a group of lifters, and 2 tape protocols agreed on the group mean, but it could not classify an individual lifter as having long femurs, because the systematic error alone equaled 1 standard deviation of the trait and the random error between examiners exceeded the trait's full range at a given height.

• Moriguchi, C. S., Sato, T. O., & Gil Coury, H. J. C. (2009). Reliability of intra- and inter-rater palpation discrepancy and estimation of its effects on joint angle measurements. Manual Therapy, 14(3), 299–305.
• Christoforetti, J. J., DeLong, J., Hanypsiak, B. T., Suri, M., Domb, B. G., Snibbe, J. C., & Gerhardt, M. B. (2017). Precision and accuracy of identification of anatomical surface landmarks by 30 expert hip arthroscopists. American Journal of Orthopedics, 46(1), E65–E70.
• Lee, J., Kwon, M., & Park, J. (2025). Influence of thigh and shank lengths and ratios on kinematic and kinetic characteristics of the knee joint during barbell back squat. Applied Sciences, 15(17), 9448.
• Cooke, D. M., Haischer, M. H., Carzoli, J. P., Bazyler, C. D., Johnson, T. K., Varieur, R., Zoeller, R. F., Whitehurst, M., & Zourdos, M. C. (2019). Body mass and femur length are inversely related to repetitions performed in the back squat in well-trained lifters. Journal of Strength and Conditioning Research, 33(3), 890–895.
• Aitken, S. A. (2021). Normative values for femoral length, tibial length, and the femorotibial ratio in adults using standing full-length radiography. Osteology, 1(2), 86–91.

Part 2: Segment Length, Ankle Mobility, and Squat Mechanics

2.1 Relative Segment Length and Squat Kinematics

Summary Statement: Ankle dorsiflexion ROM had a stronger correlation with knee, ankle, and trunk motion during squats than femur (thigh) length, tibia (shank) length, or their ratio. In loaded squats, no segment ratio correlated with trunk angle or lumbopelvic flexion. Dorsiflexion ROM correlated with both, explaining 45% of the difference between lifters in trunk angle and 21 - 27% of the difference in lumbopelvic flexion. In unloaded squats, a relatively longer thigh correlated with more knee flexion and more dorsiflexion at narrow and medium stances (explaining 12 - 23% of the difference between individuals), but did not correlate with hip flexion (forward lean) at any stance, and did not correlate with any joint angle with a wide stance (200% of pelvic width). With a wide stance, less dorsiflexion ROM was required. A relatively longer tibia, not a longer femur, correlated with knee abduction and internal rotation at depth. Taller males squatted with more forward lean and taller females squatted more upright with less knee flexion in the 1 study that examined height by sex, with effect sizes not reported. Last, longer femurs correlated with fewer squat repetitions in males but not in females, but the effect could not be separated from body mass because the lifters in these studies with longer femurs were also the heavier lifters (implying a difference in absolute femur length and lifter size and not just relative length).

Paragraph 1: Segment ratios did not predict trunk lean or lumbopelvic flexion in loaded squats; ankle dorsiflexion range did

Two studies tested segment ratios and ankle dorsiflexion range of motion as predictors of trunk position during loaded back squats in trained lifters: 1 study of 11 recreationally trained males at 75% of 1-RM, and 1 study of 18 competitive powerlifters and weightlifters at 70% of 1-RM. Both studies are annotated in full in 1.2 Paragraph 2 and are repeated here because this paragraph addresses their primary outcomes.

• Fuglsang et al. investigated 11 recreationally trained males (age: 22.9 ± 1.8 years; range 19 to 26 years; height 1.80 ± 0.04 m; mass 78.4 ± 4.7 kg) from Aarhus University with an average of 4.1 years (range 1.5 to 8) of heavy resistance training experience including the back squat, at least 2 training sessions/week, and no injury in the 6 months before testing; 3 additional recruits were excluded for inability to reach parallel depth under the test conditions. Reflective markers were placed bilaterally on the greater trochanter, lateral femoral epicondyle, lateral malleolus, and fifth metatarsal head, and on the center of the bar, and segment lengths were calculated from marker positions in the upright squat position (shank: lateral epicondyle to lateral malleolus; thigh: greater trochanter to lateral epicondyle; trunk: greater trochanters to bar). All participants performed a self-selected warm-up, 2 warm-up sets with depth checked by motion capture, and 3 continuous high-bar back squats to at least parallel (hip markers below knee markers) with approximately 75% of self-estimated 1-RM loads (102.7 ± 22 kg; estimated 1-RM 136.8 ± 24.4 kg), barefoot, feet pointing straight ahead at acromion-width stance, during 1 session with 8-camera motion capture. Participants then performed a weight-bearing lunge test on each ankle (3 reps, held 3 s, holding the squat rack for support) with motion capture recording maximal dorsiflexion as the anterior angle between the shank and horizontal. Outcome measures included shank length, thigh length, trunk length, shank-to-thigh, trunk-to-shank, and trunk-to-thigh ratios, shank angle in the lunge test, shank angle and trunk angle (anterior angle between the hip-to-bar line and horizontal) at the parallel position during the descent, the dorsiflexion deficit (lunge shank angle minus parallel-squat shank angle), and 4 regression models: lunge shank angle predicting parallel-squat shank angle (Model 1), lunge shank angle predicting trunk angle (Model 2), the 3 segment ratios predicting trunk angle (Model 3), and lunge shank angle plus the 3 ratios predicting trunk angle (Model 4). The findings demonstrated a mean shank length of 44.0 ± 1.9 cm, thigh length of 41.5 ± 1.9 cm, trunk length of 59.1 ± 2.3 cm, shank-to-thigh ratio of 1.063 ± 0.055, trunk-to-shank ratio of 1.344 ± 0.07, and trunk-to-thigh ratio of 1.427 ± 0.089. The shank angle averaged 47.8 ± 5.7° in the lunge test and 59.2 ± 4.8° at parallel, a deficit of 11.4 ± 4.4° that was independent of maximal range of motion, and trunk angle at parallel averaged 56.9 ± 5.4°. Lunge shank angle significantly predicted parallel-squat shank angle (R² = 0.41; adjusted R² = 0.35; p = 0.034). Lunge shank angle significantly and negatively predicted trunk angle (R² = 0.45; adjusted R² = 0.39; B = −0.64; p = 0.024), indicating that participants with a lower shank angle in the lunge test (more dorsiflexion) squatted with a higher trunk angle (more upright) at parallel. The 3 segment ratios together did not significantly predict trunk angle (R² = 0.37; adjusted R² = 0.10; p = 0.334), and no individual ratio reached significance (p = 0.139 to 0.152). The combined model of lunge shank angle plus the 3 ratios did not reach significance (R² = 0.60; adjusted R² = 0.33; p = 0.183), and within it no predictor was significant (p = 0.113 to 0.267). Post hoc statistical power was 0.886. Note that the single knee marker on the lateral epicondyle lengthened the measured shank relative to the thigh, which explained why the shank-to-thigh ratio exceeded 1.0 and was not comparable to radiographic femur-to-tibia ratios. Note that the between-participant range of segment ratios was narrow (standard deviations of 0.055 to 0.089), which limited the power of the ratio models. Note that participants squatted with 11.4° less dorsiflexion than they demonstrated in the lunge test regardless of how much range they had (??).
• Berglund et al. investigated 18 competitive powerlifters and Olympic weightlifters (11 males, 7 females; 12 powerlifters, 6 weightlifters) (age: 24.9 ± 4.4 years; height 170.4 ± 7.5 cm; mass 79.2 ± 11.6 kg; 7.2 ± 4.5 years of training; squat 1-RM 135.3 ± 41.8 kg) from clubs in Umeå, Sweden, with at least 2 years of resistance training experience, training to compete or competing, and no current injury preventing full training (8 participants reported a current or previous low back or lower extremity pain or injury). All participants performed 3 back squats to competition depth with 70% of 1-RM loads during 1 session, with habitual stance, wrist wraps and shoes of choice, no belt, and inertial movement sensors over L2 and S2. All participants then underwent tape-measure anthropometry (femur: greater trochanter to lateral femoral condyle; tibia: medial tibial condyle to medial malleolus; upper body: seat to acromion in sitting), goniometric passive hip range of motion, a weight-bearing lunge test of ankle dorsiflexion (angle and toe-to-wall distance), and a 4-test lumbopelvic movement control battery, administered by 2 physical therapists. Outcome measures included lumbopelvic flexion (angular displacement between L2 and S2 from standing to the bottom position), squat depth, stance width, femur-to-tibia ratio, upper-body-to-femur ratio, passive hip flexion with the knee bent and straight, hip abduction, hip internal and external rotation, ankle dorsiflexion angle and distance for each side and both sides combined, movement control battery score (0 to 4), and univariate linear regression of lumbopelvic flexion on each independent variable. The findings demonstrated a mean lumbopelvic flexion of 18.4 ± 4.7° (range 9.3 to 27.9°) and a mean squat depth of 100.6 ± 13.4° of hip flexion. Lumbopelvic flexion did not correlate with squat depth (r = 0.29; p = 0.24). The femur-to-tibia ratio averaged 1.18 ± 0.06 (1.15 ± 0.04 for males; 1.22 ± 0.06 for females) and the upper-body-to-femur ratio averaged 1.35 ± 0.06. Passive hip flexion with the knee bent averaged 99.8 ± 11.3° (right) and 99.0 ± 6.7° (left), hip flexion with the knee straight averaged 75.2 ± 15.2° and 72.6 ± 17.3°, hip external rotation averaged 51.1 ± 9.3° and 53.6 ± 5.6°, hip internal rotation averaged 43.2 ± 10.1° and 41.7 ± 6.7°, and weight-bearing lunge dorsiflexion averaged 38.7 ± 8.1° (10.5 ± 3.6 cm) and 39.6 ± 8.7° (10.8 ± 3.4 cm). The movement control battery score averaged 2.72 ± 1.07 of 4 tests passed. Lumbopelvic flexion was statistically unrelated to the femur-to-tibia ratio (p = 0.75), the upper-body-to-femur ratio (p = 0.39), hip flexion with the knee bent or straight (adjusted R² −0.06 to 0.04), hip abduction, hip external rotation (adjusted R² −0.06 to −0.05), hip internal rotation (adjusted R² 0.08 to 0.11; p = 0.10 to 0.14), and movement control battery score (p = 0.68). Lumbopelvic flexion increased significantly with ankle dorsiflexion angle on the left (adjusted R² 0.21; p = 0.03), dorsiflexion distance on the right (adjusted R² 0.21; p = 0.03), dorsiflexion distance on the left (adjusted R² 0.27; p = 0.02), and combined dorsiflexion distance (adjusted R² 0.25; p = 0.02). Note that the direction of the dorsiflexion relationship was the reverse of the authors' stated hypothesis. Note that the femur-to-tibia ratio standard deviation of 0.06 in this sample was comparable to the population standard deviation reported on imaging (Aitken, 2021; Strecker et al., 1997), so the range of proportions tested was representative rather than restricted. Note that lumbopelvic flexion and trunk lean were separate outcomes; this finding did not conflict with the inverse relationship between dorsiflexion range and trunk lean reported by Fuglsang et al. (2017) (??).

Conclusion: In the 2 studies that tested segment ratios against dorsiflexion range as predictors of trunk position in a loaded squat, no segment ratio reached significance in either study (p = 0.334 for the 3 ratios combined in Fuglsang; p = 0.75 for femur-to-tibia and p = 0.39 for upper-body-to-femur in Berglund), and dorsiflexion range reached significance in both. More dorsiflexion range went with a more upright trunk at parallel (R² = 0.45, meaning ankle range explained 45% of the difference in trunk angle between lifters) and with more lumbopelvic flexion at competition depth (adjusted R² 0.21 to 0.27, or 21% to 27%). Trunk angle (the whole trunk relative to horizontal) and lumbopelvic flexion (the angle change between L2 and S2) were different measurements, so a more upright trunk in one study and more lumbopelvic flexion in the other were not contradictory findings; dorsiflexion range was a significant predictor of both. The femur-to-tibia ratio in the Berglund sample (standard deviation 0.06) varied at least as much as the population ratio in Section 1.2 (standard deviation 0.04 to 0.05), so the null result was not explained by a sample of lifters who all had the same proportions. In actual length, the Berglund sample spanned roughly the same 3.7 cm (1.5 inch) femur difference at a given height as the extremes in Section 1.2, and that difference predicted none of the trunk behavior measured.

• Fuglsang, E. I., Telling, A. S., & Sørensen, H. (2017). Effect of ankle mobility and segment ratios on trunk lean in the barbell back squat. Journal of Strength and Conditioning Research, 31(11), 3024–3033.
• Berglund, L., Öhberg, F., Strömbäck, E., & Papacosta, D. (2024). Are anthropometric measures, range of motion, or movement control tests associated with lumbopelvic flexion during barbell back squats? International Journal of Sports Physical Therapy, 19(9), 1097–1107.

Paragraph 2: A relatively longer thigh was absorbed by more knee flexion and ankle dorsiflexion, not by more hip flexion, and a relatively longer shank went with more knee abduction and internal rotation

Two studies correlated thigh-to-shank or shank-to-thigh ratio with joint kinematics during squats: 1 study of 32 healthy adults performing unloaded squats at 3 stance widths, and 1 study of 50 resistance-trained males performing deep squats at 50% of 1-RM (annotated in full in 1.1 Paragraph 2).

• Demers et al. investigated 32 healthy males and females (20 males, 12 females) (age: 24 ± 2 years) who could perform an unloaded back squat to at least parallel with feet at 100% of pelvic width, with no leg length difference of more than 2.5 cm and no self-reported condition affecting lower limb movement, strength, or balance. Trunk length (acromion to greater trochanter), thigh length (greater trochanter to lateral femoral epicondyle), and shank length (lateral femoral epicondyle to lateral malleolus) were measured with a tape measure on the dominant leg. All participants performed unloaded back squats at narrow (100% of pelvic width), medium (150%), and wide (200%) stances, in random order, during 1 session, with 3 reps/stance, arms crossed over the shoulders, self-selected toe-out angle, and electromagnetic motion capture sensors over the sacrum and the thigh, shank, and foot of the dominant leg. Outcome measures included the trunk-to-thigh length ratio, the thigh-to-shank length ratio, and hip flexion, knee flexion, and ankle dorsiflexion angles at each 10° of thigh inclination from 10° to 80° during the eccentric phase, and Pearson correlations between each joint angle and each length ratio at each stance. The findings demonstrated a mean trunk-to-thigh ratio of 1.37 ± 0.10 and a mean thigh-to-shank ratio of 0.98 ± 0.04. Ankle dorsiflexion was significantly lower for wider stances at all thigh inclinations, with effect sizes increasing with squat depth (partial η² 0.148 to 0.697). Hip flexion was significantly lower for the wide stance only at 20° and 30° of thigh inclination (partial η² 0.128 to 0.184) and statistically similar across stances from 40° to 60°; hip flexion differed across all stances at 70° and 80° (partial η² 0.533 to 0.656). Knee flexion was significantly lower for the narrow stance only at 10° and 20° of thigh inclination (partial η² 0.108 to 0.165) and statistically similar across stances from 30° to 80°. Hip flexion did not correlate with either length ratio at any stance (r = −0.271 to 0.159). Knee flexion correlated positively with the thigh-to-shank ratio at all thigh inclinations for the narrow stance (r = 0.369 to 0.454) and the medium stance (r = 0.368 to 0.480), and correlations were statistically non-significant for the wide stance (r = 0.243 to 0.336). Ankle dorsiflexion correlated positively with the thigh-to-shank ratio for the narrow stance (r = 0.352 to 0.451) and the medium stance (r = 0.354 to 0.432), and correlations were statistically non-significant for the wide stance (r = 0.133 to 0.218). Knee flexion correlated negatively with the trunk-to-thigh ratio at 30° to 60° for the narrow stance (r = −0.374 to −0.402), at 50° for the medium stance (r = −0.350), and at 40° and 50° for the wide stance (r = −0.352 to −0.360). Ankle dorsiflexion correlated negatively with the trunk-to-thigh ratio only for the medium stance at 10° to 50° (r = −0.354 to −0.453). Note that the significant correlations explained 12% to 23% of the variance in knee flexion and ankle dorsiflexion (??).
• Lee et al. (annotated in full in 1.1 Paragraph 2) reported, in 50 resistance-trained males performing deep back squats at 50% of 1-RM, that thigh length correlated positively with peak anterior knee displacement (r = 0.537; R² = 0.289) and peak knee extension moment (r = 0.387; R² = 0.151) and did not correlate with peak knee flexion, and that a higher shank-to-thigh ratio (93.7 ± 7.1%) correlated with more knee abduction (r = −0.385 with peak adduction angle), more knee internal rotation (r = 0.426), and a lower knee extension moment (r = −0.390), with R² values of 0.149 to 0.181 (??).

Conclusion: In these studies, "thigh" was the segment measured from the greater trochanter to the lateral femoral epicondyle (the femur), and "shank" was the segment measured from the lateral femoral epicondyle to the lateral malleolus (the tibia); the terms differ from Section 1 only because the studies measured on the skin rather than on the bone. A relatively longer thigh correlated with more knee flexion (r = 0.37 to 0.48) and more ankle dorsiflexion (r = 0.35 to 0.45) at narrow and medium stances, and did not correlate with hip flexion at any stance (r = −0.27 to 0.16, none significant). At a wide stance (200% of pelvic width), the correlations were non-significant. Thigh-to-shank ratio explained 12% to 23% of the difference between people in knee flexion and dorsiflexion (Demers et al.), and absolute thigh length explained 29% of the difference in anterior knee displacement (Lee et al.); no other joint variable reached those values. A relatively longer shank, not a longer thigh, correlated with knee abduction and internal rotation at depth (r = −0.39 and 0.43; 15% to 18% of the difference between lifters). In actual length, the thigh-to-shank ratio standard deviation of 0.04 in the Demers sample corresponded to about 1.6 cm (0.6 inches) of thigh on a 39 cm shank. Two findings from Demers were used in the dorsiflexion-demand calculation later in this review: additional thigh length was accommodated at the knee and ankle rather than at the hip, and abducting the hip reduced the amount of that length projected in the sagittal plane.

• Demers, E., Pendenza, J., Radevich, V., & Preuss, R. (2018). The effect of stance width and anthropometrics on joint range of motion in the lower extremities during a back squat. International Journal of Exercise Science, 11(1), 764–775.
• Lee, J., Kwon, M., & Park, J. (2025). Influence of thigh and shank lengths and ratios on kinematic and kinetic characteristics of the knee joint during barbell back squat. Applied Sciences, 15(17), 9448.

Paragraph 3: Taller males leaned further; taller females flexed the knee less

One study correlated height and segment lengths with squat joint angles separately by sex in 28 trained lifters (annotated in full in 1.1 Paragraph 2).

• McKean and Burkett (annotated in full in 1.1 Paragraph 2) reported, in 16 males (height 179.4 ± 6.8 cm) and 12 females (height 167.7 ± 5.3 cm) with at least 12 months of squat training, that maximum hip angle correlated negatively with height in males (3 of 8 eccentric and 2 of 8 concentric comparisons), indicating more trunk lean in taller males, and correlated positively with height in females (2 and 3), indicating less trunk lean in taller females; that maximum knee angle correlated positively with height (4 and 4), thigh length (4 and 4), and torso length (3 and 3) in females, indicating less knee flexion in taller females; and that maximum lumbar angle correlated positively with shank length in males (3 and 3). Results were reported as counts of significant Spearman correlations out of 8 comparisons (2 stances by 2 loads by 2 phases) rather than as coefficients (??).

Conclusion: In males, height correlated negatively with maximum hip angle in 5 of 16 comparisons (3 eccentric, 2 concentric), indicating more forward trunk lean in taller males, and shank length correlated positively with maximum lumbar angle in 6 of 16 comparisons. In females, height correlated positively with maximum hip angle in 5 of 16 comparisons and with maximum knee angle in 8 of 16, indicating less trunk lean and less knee flexion in taller females, and maximum knee angle also correlated positively with thigh length (8 of 16) and torso length (6 of 16). Height was therefore associated with squat strategy in both sexes, in opposite directions. In actual terms, the male sample was 179.4 ± 6.8 cm tall, so roughly 95% of the sample fell within a 27 cm (10.6 inch) height range, and the only sagittal-plane effect of height in males was the smaller hip angle. Because the results were reported as counts of significant Spearman correlations rather than coefficients or effect sizes, the magnitude of every association was unknown, and this was the only study in the review that examined sex-specific relationships between height and squat kinematics.

• McKean, M., & Burkett, B. J. (2012). Does segment length influence the hip, knee and ankle coordination during the squat movement? Journal of Fitness Research, 1(1), 23–30.

Paragraph 4: Absolute Segment Length and Repetition Performance

Paragraph 1: Longer femurs went with fewer squat repetitions in males, and the effect was weak and confounded with body mass

Two studies correlated absolute femur or thigh length with squat repetitions to failure in resistance-trained lifters: 1 study of 58 lifters at 70% of 1-RM, and 1 study of 36 lifters at 60% of 1-RM.

• Cooke et al. (annotated in full in 1.1 Paragraph 2) reported, in 58 well-trained lifters, that reps at 70% of 1-RM correlated inversely with femur length (r = −0.265; R² = 0.09; p = 0.019), body mass (r = −0.352), and body fat (r = −0.278); that no variable was significant in the multivariate model (R² = 0.200) and femur length contributed 16.45% of the explained variance against 43.87% for body mass; and that the 10 longest femurs produced 14 ± 3 reps against 19 ± 6 for the 10 shortest (p = 0.027) (??).
• Falch et al. (annotated in full in 1.1 Paragraph 2) reported, in 19 males and 17 females, that squat reps-to-failure at 60% of 1-RM correlated negatively with height (r = −0.41) and not with thigh length (r = −0.18) across all participants; that thigh length correlated negatively with reps in males (r = −0.67) and not in females (r = 0.06), with the coefficients differing significantly; and that squat 1-RM correlated positively with height (r = 0.66), thigh length (r = 0.46), and shank length (r = 0.64) (??).

Conclusion: Absolute femur length correlated with fewer repetitions at 70% of 1-RM (r = −0.265) and explained 9% of the difference between lifters, but was not significant once body mass was added to the model (p = 0.793), and the study's authors reported that the 2 variables could not be isolated because heavier lifters had longer femurs. This likely also explains why, in both studies, longer femurs correlated with a higher 1-RM (r = 0.46 for thigh length in Falch et al.). In the second study, thigh length correlated with fewer repetitions at 60% of 1-RM in males (r = −0.67), but did not correlate in females (r = 0.06) or in the pooled sample (r = −0.18). In actual numbers, the 10 lifters with the longest femurs (51.0 cm, 20.1 inches) performed 14 ± 3 repetitions and the 10 with the shortest femurs (43.3 cm, 17.0 inches) performed 19 ± 6. The difference in femur length between these 2 groups was 7.7 cm (3.0 inches). That difference was twice the 3.7 cm range in femur length between individuals of the same height (Section 1.2), so the long-femur group was likely taller and heavier, not just differently proportioned. In summary, absolute lever length was a small and inconsistent factor in repetition performance, present in males and absent in females, and could not be separated from body mass in the data available.

• Cooke, D. M., Haischer, M. H., Carzoli, J. P., Bazyler, C. D., Johnson, T. K., Varieur, R., Zoeller, R. F., Whitehurst, M., & Zourdos, M. C. (2019). Body mass and femur length are inversely related to repetitions performed in the back squat in well-trained lifters. Journal of Strength and Conditioning Research, 33(3), 890–895.
• Falch, H. N., Haugen, M. E., Larsen, S., & van den Tillaar, R. (2023). Association of strength performance in bench press and squat with anthropometric variables between resistance-trained males and females. Journal of Functional Morphology and Kinesiology, 8(1), 19.

Section 3: Ankle Dorsiflexion Range of Motion and Squat Mechanics

• Summary Statement: Restricting ankle dorsiflexion range of motion (ROM) changed squat mechanics, and dorsiflexion ROM increased most following multi-week, integrated programs that included joint mobilization (supervised, self-administered, or both). A wedge that reduced available dorsiflexion ROM by 7.5° reduced peak knee flexion by 15.6°, shifted the center of mass backward, reduced quadriceps activity, and increased medial knee displacement, with no change in hip ROM. This experiment verified the effects of reduced dorsiflexion on squat mechanics. Additionally, individuals with limited dorsiflexion on the weight-bearing lunge test squatted with 12 - 15° less knee flexion and 6 - 8° less dorsiflexion when compared to individuals with normal ROM. Dorsiflexion ROM by itself accounted for 21 - 38% of the difference in squat depth between individuals, more than femur proportions accounted for in any joint angle investigated in Part 2 of this review. A single session of soft tissue mobilization or stretching increased dorsiflexion significantly when compared to a control group, but by less than the test's measurement error; however, a single session of a 4-phase inhibit-lengthen-activate-integrate sequence, in a separate study, increased dorsiflexion by an amount equal to measurement error. Further, programs of 5 to 8 weeks that combined soft tissue release or joint mobilization with stretching and strengthening increased dorsiflexion by 5 - 7°, and the increase was maintained or grew at 3-month follow-up. Note that a 12-week program of stretching and eccentric calf loading, without soft tissue release or joint mobilization, did not increase dorsiflexion

3.1 Dorsiflexion Restriction and Squat Kinematics

Paragraph 1: Restricting dorsiflexion by 7.5° cost 15° of knee flexion and produced medial knee displacement; lifters with limited weight-bearing lunge dorsiflexion squatted with 12° to 15° less knee flexion

Two studies tested the effect of reduced dorsiflexion on squat kinematics: 1 within-subject experiment restricting dorsiflexion with a wedge in 30 healthy adults, and 1 between-group comparison of 40 adults classified as normal or limited by 2 dorsiflexion tests.

• Macrum et al. investigated 30 healthy recreationally active males and females (15 males, 15 females) (age: 18 to 30 years; height 173.5 ± 12.1 cm; mass 72.0 ± 16.4 kg) who engaged in at least 30 min of physical activity at least 3 days/week, with no history of lower extremity injury. All participants performed 7 double-leg squats to self-selected depth during a no-wedge protocol (feet flat) and a wedge protocol (standing on a 12° forefoot-elevating wedge running the length of the foot, placing the ankle in approximately 12° of dorsiflexion at the start position), in counterbalanced order, during 1 session, with 3-dimensional motion capture and surface EMG of the vastus lateralis, vastus medialis oblique, lateral gastrocnemius, and soleus (normalized to MVIC). Outcome measures included peak hip flexion, abduction, and rotation angles; peak knee flexion, varus/valgus, and rotation angles; peak ankle dorsiflexion angle; hip, knee, and ankle sagittal, frontal, and rotational displacement; medial knee displacement; anterior-posterior center of mass displacement; and mean EMG activity of each muscle during the eccentric phase. The findings demonstrated that the wedge protocol resulted in a starting ankle position 9.5° more dorsiflexed than the no-wedge protocol (−8.8 ± 3.0° versus 0.7 ± 2.8°). Peak knee flexion was significantly lower during the wedge protocol (81.1 ± 19.2°) than the no-wedge protocol (96.7 ± 17.3°; effect size 0.81), and knee sagittal displacement was significantly lower (85.1 ± 18.5° versus 100.2 ± 16.8°; effect size 0.82). Peak knee valgus was significantly higher during the wedge protocol (4.1 ± 3.3° versus 3.4 ± 3.2°; effect size 0.21), and medial knee displacement was significantly higher (effect size 2.92); knee frontal plane and rotational displacement were statistically similar between protocols. Peak ankle dorsiflexion was significantly higher during the wedge protocol (30.0 ± 6.3° versus 28.0 ± 6.4°; effect size 0.31), and ankle dorsiflexion displacement was significantly lower (21.3 ± 5.9° versus 28.8 ± 6.0°; effect size 1.25). Peak hip angles and hip displacements were statistically similar between protocols in all planes. Center of mass displacement was significantly more posterior during the wedge protocol (effect size 0.71). Vastus lateralis EMG activity (55.1 ± 18.6% versus 62.4 ± 21.8% of MVIC; effect size 0.33) and vastus medialis oblique EMG activity (61.2 ± 21.1% versus 66.3 ± 26.1%; effect size 0.20) were significantly lower during the wedge protocol, soleus EMG activity was significantly higher (23.4 ± 2.4% versus 21.8 ± 2.6%; effect size 0.64), and lateral gastrocnemius EMG activity was statistically similar. Note that the wedge reduced available dorsiflexion excursion by 7.5° and served as a mechanical proxy for a dorsiflexion restriction; participants did not have restricted dorsiflexion range of motion (??).
• Dill et al. compared 40 physically active males and females (20 males, 20 females) (age: 19.5 to 20.7 years by group; height 171 to 173 cm) recruited from a university population, with no history of lower extremity injury within the past 6 months, lower extremity surgery, or neurological disorder. All participants were classified twice into normal and limited dorsiflexion groups (20 per group each time) by a non-weight-bearing passive dorsiflexion test with the knee extended (normal at least 15°, mean 16.78 ± 2.16°; limited 5° or less, mean 1.63 ± 2.58°) and by the weight-bearing lunge test split at the sample median (normal at least 44°, mean 50.84 ± 5.16°; limited 43° or less, mean 38.91 ± 3.48°), with intra-rater ICC of 0.888 for the non-weight-bearing test and 0.953 for the lunge test. All participants performed overhead squats to at least 60° of knee flexion with heels down, single-leg squats, and jump landings from a 30 cm box, with 5 trials each, during 1 session, with electromagnetic motion capture of the dominant leg. Outcome measures included knee flexion, knee valgus, knee varus, knee internal rotation, knee external rotation, and ankle dorsiflexion displacement (peak minus start) during the eccentric phase of each task, and peak knee flexion, compared between normal and limited groups for each classification method by 1-way ANOVA. The findings demonstrated that 12 of 40 participants (30%) changed groups between the 2 classification methods. When classified by the non-weight-bearing test, all knee and ankle displacements were statistically similar between normal and limited groups during all 3 tasks. When classified by the weight-bearing lunge test, the normal group exhibited significantly more knee flexion displacement during the overhead squat (103.94 ± 11.79° versus 89.00 ± 14.62°; mean difference 14.94°; effect size 1.31) and the single-leg squat (76.36 ± 11.02° versus 63.97 ± 10.55°; mean difference 12.39°; effect size 1.51), significantly more ankle dorsiflexion displacement during the overhead squat (32.03 ± 6.36° versus 24.14 ± 4.27°; mean difference 7.89°; effect size 1.49) and the single-leg squat (30.67 ± 5.67° versus 24.23 ± 4.49°; effect size 1.86), significantly more peak knee flexion during both squats (mean differences 15.26° and 11.62°), and significantly more knee varus displacement during the single-leg squat (mean difference 5.50°; p = 0.048). Knee valgus displacement (overhead squat 1.72 ± 3.60° versus 1.93 ± 2.37°; effect size 0.07) and knee rotation displacement were statistically similar between lunge-test groups during both squats. All displacements were statistically similar between groups during jump landings under both classification methods (??).

Conclusion: A wedge that reduced available dorsiflexion by 7.5° reduced peak knee flexion by 15.6° (96.7° to 81.1°), shifted the center of mass backward, reduced quadriceps activity by 5% to 7% of MVIC, and increased medial knee displacement (effect size 2.92). Hip motion did not change in any plane. Participants classified as limited by the weight-bearing lunge test (38.9° versus 50.8° for the normal group) squatted with 12° to 15° less knee flexion and 6° to 8° less dorsiflexion than participants classified as normal. Knee valgus angle did not differ between lunge-test groups (effect size 0.07). In summary, a loss of 7.5° of dorsiflexion resulted in nearly twice the reduction in knee flexion during a squat, and the weight-bearing lunge test predicted squat mechanics.

• Macrum, E., Bell, D. R., Boling, M., Lewek, M., & Padua, D. (2012). Effect of limiting ankle-dorsiflexion range of motion on lower extremity kinematics and muscle-activation patterns during a squat. Journal of Sport Rehabilitation, 21(2), 144–150.
• Dill, K. E., Begalle, R. L., Frank, B. S., Zinder, S. M., & Padua, D. A. (2014). Altered knee and ankle kinematics during squatting in those with limited weight-bearing-lunge ankle-dorsiflexion range of motion. Journal of Athletic Training, 49

Paragraph 2: Dorsiflexion range explained 32% to 44% of the variance in squat depth

One study correlated hip and ankle range of motion and strength with squat depth in 101 healthy adults.

• Kim et al. investigated 101 healthy males and females, including 64 males (age: 25.7 ± 5.9 years; height 174.3 ± 5.9 cm; mass 67.6 ± 7.7 kg) and 37 females (age: 22.0 ± 2.2 years; height 159.4 ± 24.5 cm; mass 54.0 ± 6.3 kg), with no lower extremity pain or neurological signs, no lower extremity injury within the previous 6 months, and no history of lower extremity surgery. Passive hip flexion, hip internal rotation, and hip external rotation, and ankle dorsiflexion with the knee extended and with the knee flexed to 90° (prone), were measured 3 times with a goniometer on the dominant leg, and hip flexor and ankle dorsiflexor strength were measured with a hand-held dynamometer (intra-rater ICC 0.978 and 0.988; normalized to body mass). All participants then performed 3 squats as deep as possible with heels on the floor, holding the bottom position for 5 seconds, with sagittal-plane video, during 1 session. Outcome measures included hip flexion, internal rotation, and external rotation range of motion, ankle dorsiflexion range of motion with the knee extended and flexed, hip flexor and ankle dorsiflexor strength, squat depth (vertical distance from the hip to the floor at the bottom position, normalized to leg length; lower values indicating deeper squats), Pearson correlations, and stepwise multiple regression of squat depth on the range of motion and strength variables, analyzed separately by sex. The findings demonstrated mean ankle dorsiflexion of 8.2 ± 5.3° (knee extended) and 16.3 ± 6.2° (knee flexed) in males, and 11.5 ± 5.6° and 20.8 ± 6.7° in females, and mean hip flexion of 113.9 ± 9.8° in males and 120.3 ± 7.7° in females. In males, squat depth correlated significantly with hip flexion, hip internal rotation, and ankle dorsiflexion with the knee extended and flexed (r = −0.239 to −0.623; negative values indicating deeper squats with more range), and ankle dorsiflexion with the knee flexed and hip flexion range of motion together explained 43.5% of the variance in squat depth (R² = 0.435; p less than 0.001). In females, squat depth correlated significantly with ankle dorsiflexion with the knee extended and flexed (r = −0.460 to −0.487) and with ankle dorsiflexor strength (r = 0.279), and ankle dorsiflexion with the knee extended and dorsiflexor strength together explained 32.4% of the variance in squat depth (R² = 0.324; p less than 0.001). Hip external rotation range of motion and hip flexor strength did not correlate significantly with squat depth in either sex. Note that the study did not measure segment lengths (??).

Conclusion: Ankle dorsiflexion ROM was the strongest single predictor of squat depth in both sexes. By itself, dorsiflexion ROM correlated with squat depth at r = −0.62 in males and r = −0.46 to −0.49 in females (negative because depth was measured as hip height, so a lower number is a deeper squat), which corresponds to 38% of the variation in depth between males and 21 - 24% between females. In a stepwise regression, dorsiflexion with the knee flexed combined with hip flexion ROM accounted for 43.5% of the variation in males, and dorsiflexion with the knee extended combined with dorsiflexor strength accounted for 32.4% in females. Hip rotation ROM and hip flexor strength did not predict depth in either sex. Segment lengths were not measured, so this study could not compare mobility against proportion directly. In summary, dorsiflexion ROM by itself accounted for 21 - 38% of the variation in squat depth, which was more than femur proportion accounted for in any joint angle in Part 2 (12 - 23% for thigh-to-shank ratio; 29% for absolute thigh length and anterior knee travel).

• Kim, S. H., Kwon, O. Y., Park, K. N., Jeon, I. C., & Weon, J. H. (2015). Lower extremity strength and the range of motion in relation to squat depth. Journal of Human Kinetics, 45, 59–69.

3.2 Interventions to Increase Dorsiflexion Range of Motion

Paragraph 1: A single session increased weight-bearing dorsiflexion by 1.6° to 3.9°, at or below the minimal detectable change

Two studies tested a single-session intervention: 1 randomized trial of 60 healthy adults comparing instrument-assisted soft tissue mobilization, stretching, and control, and 1 uncontrolled pilot of 8 adults with medial knee displacement receiving a 4-phase corrective sequence.

• Rowlett et al. compared 60 healthy males and females (age: 25.8 ± 6.7 years; 63% female; height 176.7 cm; mass 72.4 ± 17.4 kg) recruited from a university population, aged 18 to 65 years, with no acute knee, ankle, or foot pathology, no positive response on a health history questionnaire or Physical Activity Readiness Questionnaire, and no contraindication to stretching or instrument-assisted soft tissue mobilization. Participants were randomized by opaque envelope to an instrument-assisted soft tissue mobilization group (n = 20), a stretching group (n = 20), or a control group (n = 20), with the assessor blinded to allocation. All participants performed a 2 min step-tap warm-up on a 12-inch box before pre-intervention measurement. The soft tissue mobilization group received 2 min of instrument-assisted mobilization (Edge Mobility Tool) along the full length of the gastrocnemius-soleus complex with alternating proximal and distal strokes parallel to the muscle fibers; the stretching group performed 3 × 30 s wall stretches with the knee extended followed by 3 × 30 s with the knee flexed, to a light pull; the control group rested for 3 min. Outcome measures included dorsiflexion range of motion measured 3 times before and after intervention by the weight-bearing lunge test (digital inclinometer at the tibial tuberosity), Modified Root Position 1 (non-weight-bearing, knee extended, goniometer), and Modified Root Position 2 (non-weight-bearing, knee flexed to 90°). The findings demonstrated statistically similar baseline range of motion across groups. Weight-bearing lunge dorsiflexion increased in the mobilization group (40.2 ± 8.3° to 42.0 ± 8.7°; mean change 1.81 ± 1.70°) and the stretching group (41.2 ± 5.9° to 42.8 ± 7.0°; mean change 1.60 ± 3.52°), and decreased in the control group (40.6 ± 6.2° to 40.1 ± 6.4°; mean change −0.54 ± 2.33°), with a significant group-by-time interaction (p = 0.011); both intervention groups differed significantly from control (mobilization p = 0.018; stretching p = 0.034) and were statistically similar to each other (p = 0.965). Modified Root Position 1 (knee extended) increased by 2.35°, 2.33°, and 1.98° in the mobilization, stretching, and control groups, with no significant interaction (p = 0.943) and no between-group differences. Modified Root Position 2 (knee flexed) increased by 3.43 ± 4.13°, 3.15 ± 3.83°, and 0.37 ± 3.88°, with a significant interaction (p = 0.031); the mobilization group differed significantly from control (p = 0.045), the stretching group exhibited a trend toward significance versus control in the Tukey test (p = 0.075) and a significant difference in pairwise comparison (mean difference 2.78°; p = 0.030), and the 2 intervention groups were statistically similar (p = 0.972). Note that the weight-bearing lunge gains of 1.6° to 1.8° fell below the minimal detectable change of 3.7° to 3.8° cited for the digital inclinometer method, and the knee-flexed gains of 3.2° to 3.4° fell below the cited 6.5° to 7.9°. Note that participants had normal baseline dorsiflexion, and that the paper's conclusion stated a difference between mobilization and stretching in the knee-flexed position that the reported statistics did not support (p = 0.972) (??).
• Ban and Yang investigated 8 physically active young adults (6 females, 2 males) (age: 21.3 ± 4.7 years; range 18 to 32 years; height 163.2 ± 7.9 cm; mass 63.7 ± 16.8 kg) who displayed medial knee displacement during 5 double-leg overhead squats (feet shoulder-width, toes forward, arms overhead) that was corrected when 5 further squats were performed on 2-inch heel lifts, who participated in at least 30 min of physical activity at least 3 days/week, and who had no history of lower extremity surgery, lower extremity injury within the previous 6 months, or neurological disorder; a power analysis indicated 8 participants for 80% power on straight-knee dorsiflexion. All participants completed 1 session consisting of a 5 min treadmill warm-up, pre-intervention assessment, a bilateral corrective sequence of approximately 20 min following the National Academy of Sports Medicine inhibit-lengthen-activate-integrate model (vibrating foam roller at 32 Hz on the gastrocnemius and soleus for 2 min each with 30 s holds on painful areas; weight-bearing static stretching of the gastrocnemius with the knee straight and the soleus with the knee bent for 3 × 30 s each; posterior tibialis strengthening with an elastic band and heel raises with the toes internally rotated for 2 × 10 each; single-leg squats for 2 × 10), and immediate post-intervention assessment. Outcome measures included passive dorsiflexion range of motion with the knee straight and with the knee flexed to 30° (goniometer, subtalar neutral, first point of restriction; intra-rater ICC 0.84 to 0.97; standard error of measurement 0.75° to 0.98°), weight-bearing lunge dorsiflexion (inclinometer 15 cm below the tibial tuberosity), and maximum medial knee displacement of the knee joint center relative to standing during 5 overhead squats (8-camera motion capture), all measured on the leg displaying greater medial knee displacement, with paired t-tests and Hedges' g effect sizes. The findings demonstrated that all participants completed the session without adverse effects and none rated it difficult or intense. Straight-knee dorsiflexion increased from 2.7 ± 2.8° to 6.9 ± 4.3° (change 4.3 ± 2.0°; 95% CI 2.6 to 5.9; p = 0.0005; g = 1.18), bent-knee dorsiflexion increased from 10.0 ± 5.2° to 13.2 ± 2.9° (change 3.2 ± 2.9°; 95% CI 0.8 to 5.6; p = 0.02; g = 0.56), and weight-bearing lunge dorsiflexion increased from 39.7 ± 6.8° to 43.5 ± 6.5° (change 3.9 ± 3.4°; 95% CI 1.1 to 6.7; p = 0.01; g = 0.58). Medial knee displacement during the overhead squat decreased from 1.7 ± 1.7 cm to 1.2 ± 1.2 cm (change −0.6 ± 0.8 cm; 95% CI −1.3 to 0.0; p = 0.02; g = 0.56). Note that the study had no control group, and the 95% confidence interval for the change in medial knee displacement reached zero. Note that the weight-bearing lunge gain of 3.9° sat at the minimal detectable change of 3.7° to 3.8° cited by Rowlett et al. (2019) for the inclinometer method. Note that the intervention sequence was the same 4-phase model that underlies the Brookbush Institute integrated approach (??).

Conclusion: One session of a single technique, either 2 min of instrument-assisted soft tissue mobilization or 3 min of calf stretching, increased weight-bearing lunge dorsiflexion by 1.6° to 1.8° in healthy adults with a starting range of 40° to 41°, and the 2 techniques produced statistically similar results (p = 0.965). Both increases were smaller than 3.7° to 3.8°, the smallest change this test can distinguish from measurement error (minimal detectable change). One session of a 20 min inhibit-lengthen-activate-integrate sequence increased weight-bearing lunge dorsiflexion by 3.9° in 8 adults who demonstrated medial knee displacement during an overhead squat, an increase equal to the minimal detectable change, and reduced medial knee displacement by 0.6 cm (0.2 inches); the study had no control group. In summary, a single session of 1 technique increased dorsiflexion by less than measurement error, and a single session of the 4-phase sequence increased dorsiflexion by an amount equal to measurement error.

• Rowlett, C. A., Hanney, W. J., Pabian, P. S., Holland, J. E., Rothschild, C. E., & Kolber, M. J. (2019). Efficacy of instrument-assisted soft tissue mobilization in comparison to gastrocnemius-soleus stretching for dorsiflexion range of motion: A randomized controlled trial. Journal of Bodywork and Movement Therapies, 23(2), 233–240.
• Ban, R., & Yang, F. (2022). Preliminary study on acute effects of an intervention to increase dorsiflexion range of motion in reducing medial knee displacement. Clinical Biomechanics, 95, 105637.

Paragraph 2: Multi-week programs that included joint mobilization increased dorsiflexion by 5° to 7° and held or grew the gain at follow-up; 12 weeks of stretching and eccentric loading alone did not increase dorsiflexion

Three studies tested multi-week programs in participants with restricted dorsiflexion: 1 randomized trial of an 8-week integrated program in 30 female athletes, 1 randomized trial comparing eccentric and stretching exercise with and without talocrural mobilization over 5 weeks in 38 basketball athletes, and 1 controlled trial of 12 weeks of stretching and eccentric loading in 47 adolescent soccer players.

• Sohrabi et al. investigated 30 female track and field athletes (age: 15 to 25 years) from athletic clubs in Hamedan, Iran, with weight-bearing lunge dorsiflexion of 34° or less, training at least 3 sessions/week of 60 min, and a body mass index of 20 to 25 kg/m², with no pain, lower extremity surgery, lower extremity injury within the previous 6 months, or contraindication to instrument-assisted soft tissue mobilization. Participants were randomized to a training group (n = 15) or a control group (n = 15) using sealed opaque envelopes, with assessors blinded to allocation. The training group performed 24 supervised 30 min sessions (3 sessions/week for 8 weeks) delivered by certified physical therapists, consisting of a 15 min lower extremity warm-up; instrument-assisted (Graston) soft tissue mobilization of the gastrocnemius, Achilles tendon, and plantar fascia for 5 min (1 min scanning, 4 min on identified adhesions; instrument angle and pressure increased from session 7, and a 30 s foam roller self-release added after session 12); gastrosoleus stretching combined with strap-assisted talocrural mobilization into dorsiflexion in a standing lunge position; and heel-toe raise strengthening progressed from double-leg to single-leg, straight-knee to bent-knee, and floor to step, with sets, reps, and resistance individualized and progressed. The control group received no intervention. Outcome measures included weight-bearing lunge dorsiflexion range of motion (trigonometric method; ICC 0.97), ankle joint position sense (absolute repositioning error at 20° of plantar flexion), Y-Balance composite score (normalized to leg length), and surface EMG of the tibialis anterior, peroneus longus, medial gastrocnemius, and soleus (normalized to MVIC) during the descending and ascending phases of a metronome-paced overhead squat to 80° of knee flexion, all measured before and after the 8 weeks, plus minimal detectable change and minimal clinically important difference for each measure. The findings demonstrated a significant group-by-time interaction for dorsiflexion range of motion (p = 0.001; partial η² 0.714), with the training group increasing by approximately 7° and the control group statistically unchanged. Repositioning error decreased by approximately 3° in the training group (interaction p = 0.005; partial η² 0.253), and Y-Balance composite score increased by 7 cm (interaction p = 0.001; partial η² 0.423). During the descending phase of the overhead squat, tibialis anterior EMG activity decreased by approximately 15% MVIC (p = 0.005; partial η² 0.250), soleus activity decreased by approximately 12% MVIC (p = 0.008; partial η² 0.223), and medial gastrocnemius activity decreased significantly (p = 0.048; partial η² 0.132) in the training group; during the ascending phase, tibialis anterior activity decreased by approximately 9% MVIC (p = 0.003; partial η² 0.272) and soleus activity by approximately 14% MVIC (p = 0.001; partial η² 0.415). Peroneus longus activity was statistically similar between groups in both phases. All 15 training participants exceeded both the minimal detectable change and the minimal clinically important difference for dorsiflexion range of motion, compared with 3 of 15 controls; 10 versus 4 exceeded both thresholds for proprioception, and 7 versus 0 for balance. Note that the 7° gain was more than twice the additional dorsiflexion demand calculated for a femur-to-tibia ratio 2 standard deviations above the mean (approximately 3°). Note that the program included each component of the Brookbush Institute integrated approach (release, mobilize, lengthen, activate), and that the sample was female and adolescent-to-young-adult, which limited generalization (??).
• Georgoulas et al. compared 38 basketball athletes (males and females; age: 21.26 ± 2.52 years; range 18 to 25 years) recruited from the International Hellenic University and clubs in Thessaloniki, Greece, with a unilateral restriction in ankle dorsiflexion (weight-bearing lunge side-to-side difference of at least 2 cm), no ankle pain, and at least 1 month of regular training, and with no history of lower extremity surgery, lower extremity injury within the previous 6 months, neurological, vestibular, or balance disorder, or connective tissue or systemic disease affecting the ankle. Participants were randomized, stratified by sex, to an exercise-only group (n = 19) or an exercise plus mobilization group (n = 19), with assessors blinded to allocation. Both groups completed 2 supervised sessions/week for 5 weeks (approximately 25 min/session) delivered by the same manual therapy certified physiotherapist, consisting of eccentric heel lowering off a step through full available range with the knee extended (gastrocnemius) and slightly flexed (soleus) for 3 sets of 15 reps at a slow (3 to 4 s) eccentric tempo with a moderate (60 s) rest between sets, step-lunge dorsiflexion drills, and static gastrocnemius and soleus stretching in a lunge for 3 × 30 s. The mobilization group additionally received posterior talar glide mobilization with active dorsiflexion (Mulligan mobilization-with-movement; grade III to IV; 3 sets of 10 reps) in non-weight-bearing and in a weight-bearing lunge with a stabilization belt, and performed daily home self-mobilization with a belt in a weight-bearing lunge for 3 × 30 to 60 s. Both groups continued regular team training. Outcome measures included active ankle dorsiflexion range of motion (digital goniometer, seated, knee approximately 20° flexed), weight-bearing lunge test toe-to-wall distance, countermovement jump height (Optojump), single hop and triple hop for distance, 6 m timed hop, reactive strength index, fatigue index (percentage decline across repeated single-leg hops), and maximal isometric dorsiflexor and plantar flexor strength (hand-held dynamometer), measured at baseline, 5 weeks, and 3-month follow-up. The findings demonstrated that no participant was lost to follow-up and no adverse events occurred. Both groups significantly increased dorsiflexion range of motion and weight-bearing lunge distance (p less than 0.001), and the mobilization group increased significantly more (interaction p less than 0.001 for both): the between-group difference in dorsiflexion range of motion was 3.52° (95% CI 2.37 to 4.66) at 5 weeks and 5.17° (95% CI 4.24 to 6.11) at 3 months, and the between-group difference in lunge distance was 1.39 cm (95% CI 1.12 to 1.65) at 5 weeks and 1.34 cm (95% CI 1.12 to 1.56) at 3 months. Single hop distance improved significantly in the mobilization group only (interaction p = 0.015; between-group p = 0.006 at 5 weeks and p = 0.007 at 3 months). Triple hop distance improved in both groups and significantly more in the mobilization group (interaction p less than 0.001; between-group difference 7.20 cm at 5 weeks and 12.89 cm at 3 months). Countermovement jump height improved in both groups and significantly more in the mobilization group (interaction p less than 0.001; between-group difference 1.17 cm at 5 weeks and 1.80 cm at 3 months). The 6 m timed hop, reactive strength index, and fatigue index improved over time in both groups with no group-by-time interaction (p = 0.198, 0.453, and 0.438). Maximal isometric dorsiflexor strength did not change (p = 0.154), and plantar flexor strength showed a small main effect of time (p less than 0.001) with no interaction (p = 0.276). Note that the mobilization group's total weight-bearing lunge improvement exceeded 3 cm against a reported minimal detectable change of 1.5 to 2.0 cm, while the exercise-only group's improvement remained near that threshold. Note that the between-group dorsiflexion difference widened from 5 weeks to 3 months with no further treatment, and that jump and hop advantages widened over the same interval. Note that the mobilization group also performed daily home self-mobilization, so the comparison was exercise alone versus exercise plus supervised and home mobilization, not mobilization dose alone (??).
• Lagas et al. compared male adolescent soccer players (age: 14 to 21 years) from the Under-16, Under-17, and Under-19 squads of 2 Dutch professional premier division clubs, free of musculoskeletal injury at baseline, in a prospective controlled trial with group assignment by club. Of 107 players screened, 47 (44%) had decreased ankle dorsiflexion, defined as a weight-bearing dorsiflexion lunge test of 10 cm or less, soleus flexibility of 34° or less, or gastrocnemius flexibility of 34° or less, and only these players were analyzed (intervention club: 15, 14, and 23 players by criterion; control club: 32, 17, and 34). Analyzed players were statistically similar between groups at baseline for age (16.3 ± 1.2 years), height (1.76 m), mass (65.6 ± 8.1 versus 68.0 ± 10.5 kg), body mass index, and weekly training hours (9.4 versus 9.0). Both groups continued regular training 4 sessions/week of approximately 2 h. The intervention group additionally performed, 3 sessions/week for 12 weeks after training, lunge-position static stretches for the soleus (knee bent) and gastrocnemius (knee straight) for 3 × 30 s each, and eccentric heel lowering off an elevation for the soleus (knee slightly flexed) and gastrocnemius (knee extended), progressed by approximately 20%/week from 2 sets of 4 reps to 3 sets of 15 reps with a moderate (1 min) rest between sets, on the restricted leg only; group instruction was provided in week 1 and individual correction throughout, and compliance was recorded by direct observation. Outcome measures included the weight-bearing dorsiflexion lunge test (toe-to-wall distance in cm), soleus flexibility (tibial inclination in degrees, knee bent), and gastrocnemius flexibility (tibial inclination in degrees, knee straight) at baseline and 12 weeks; intra- and inter-observer reliability and minimal detectable change of each test; compliance; and Achilles tendon injuries. The findings demonstrated that the lunge test was statistically unchanged in the intervention group (7.1 ± 1.8 to 7.4 ± 2.4 cm; p = 0.381) and increased significantly in the control group (6.1 ± 2.1 to 8.2 ± 2.9 cm; p less than 0.001), and the between-group difference in change favored the control group (95% CI −2.7 to −0.5 cm; p = 0.005). Soleus flexibility was statistically unchanged in the intervention group (31.0 ± 1.7° to 32.5 ± 3.3°; p = 0.075) and increased significantly in the control group (28.3 ± 3.4° to 33.6 ± 4.7°; p less than 0.001), with the intervention significantly associated with less improvement in multivariate regression (p = 0.009); baseline soleus flexibility differed between groups (p = 0.011). Gastrocnemius flexibility was statistically unchanged in the intervention group (29.8 ± 3.0° to 31.0 ± 3.5°; p = 0.188) and increased significantly in the control group (28.3 ± 4.4° to 31.2 ± 5.6°; p = 0.004), with no significant group effect (p = 0.221). Mean compliance was 64% to 69% of prescribed sessions (approximately 2 sessions/week), and compliance did not correlate with change in lunge test (r = −0.31; p = 0.275) or soleus flexibility (r = −0.41; p = 0.163) and correlated positively with change in gastrocnemius flexibility (r = 0.47; p = 0.022). Reliability was excellent for all 3 tests (intra-observer ICC 0.94 to 0.98; inter-observer ICC 0.98 to 0.99), with minimal detectable change of 1.5 cm (lunge), 4.7° (soleus), and 4.9° (gastrocnemius). No Achilles tendon injuries occurred in either group. Note that the control group's 2.1 cm lunge test gain exceeded the minimal detectable change while the soleus and gastrocnemius changes in both groups fell within it. Note that groups were assigned by club rather than randomized, that the control group began the season with lower soleus flexibility, and that the intervention contained no soft tissue release or joint mobilization component (??).

Conclusion: An 8-week program of instrument-assisted soft tissue mobilization, strap-assisted talocrural mobilization with stretching, and progressive calf strengthening increased weight-bearing lunge dorsiflexion by approximately 7° in female athletes with a starting range of 34° or less. All 15 trained participants exceeded the minimal detectable change (the smallest change the test can distinguish from measurement error), compared with 3 of 15 controls. A 5-week program of eccentric calf loading and stretching increased dorsiflexion in basketball athletes with a restriction on 1 side, and adding talocrural mobilization to that program increased dorsiflexion by a further 3.5° at 5 weeks and 5.2° at 3 months. The difference between the 2 groups increased after the program ended. A 12-week program of stretching and eccentric calf loading, without soft tissue release or joint mobilization, did not increase weight-bearing lunge distance in adolescent soccer players (7.1 to 7.4 cm; p = 0.381). The regular-training control group increased by 2.1 cm (p < 0.001), and the increase was significantly larger than in the intervention group (p = 0.005). The 5° to 7° increase in the 2 programs that included joint mobilization was 1.5 to 2 times the 3.7° to 3.8° minimal detectable change for the lunge test. In summary, the 2 programs that included joint mobilization increased dorsiflexion by 5° to 7° in 5 to 8 weeks and maintained the increase at follow-up, while the 1 program based on stretching and calf loading alone did not increase dorsiflexion in 12 weeks.

• Sohrabi, T., Saki, F., Ramezani, F., & Tahayori, B. (2024). Comprehensive corrective exercise program improves ankle function in female athletes with limited weight-bearing ankle dorsiflexion: A randomized controlled trial. PLOS ONE, 19(10), e0312152.
• Georgoulas, V., Kallistratos, I., Apostolou, T., Kasimis, K., Lytras, D., & Iakovidis, P. (2026). Effects of adding posterior ankle joint mobilization to eccentric training on ankle range of motion and athletic performance in basketball athletes with restricted ankle dorsiflexion: A randomized controlled trial. Journal of Functional Morphology and Kinesiology, 11(1), 92.
• Lagas, I. F., Meuffels, D. E., Visser, E., Groot, F. P., Reijman, M., Verhaar, J. A. N., & de Vos, R. J. (2021). Effects of eccentric exercises on improving ankle dorsiflexion in soccer players. BMC Musculoskeletal Disorders, 22(1), 485.