Does an isometric hold only strengthen that one angle?
Knee holds at 65 degrees for 4 weeks gave +12% there, +5 to 11% at nearby angles and nothing 30 degrees deeper. Untrained men, one leg; train near the angle you need.
Lanza, Balshaw and Folland asked whether the claimed joint-angle specificity of isometric training is real once a control group and a proper group x angle test are used.
Thirteen young men trained the knee extensors of one leg at 65 degrees of knee flexion (0 = straight; 65 is mid-range, near the angle of peak torque) for 14 sessions over 4 weeks; nine controls kept their habits. Each session combined explosive contractions and maximal 3-s contractions.
Strength rose 12% at the training angle, 11% at 50 degrees, 7% at 80 degrees and 5% at 35 degrees, with no gain at 95 degrees and no change in controls. The gain was larger after training than in controls at some angles than at others (group x angle P = 0.029), which the authors call the highest level of evidence for angle specificity.
Muscle activation (EMG) rose at three of the same angles, but showed no angle specificity of its own, so a neural cause is suggested but not shown. Adding explosive contractions to the maximal ones did not improve early rapid force (first 100 ms), which the authors read as possible interference.
Protocol
Participants: 22 recreationally active men, no lower-body injury and no systematic lower-body resistance training for at least 12 months; randomised to training (n=13; 22 ± 3 y, 73 ± 7 kg) or control (n=9; 23 ± 3 y, 75 ± 8 kg). A priori power analysis asked for n=16 (effect size 0.8).
Kind of isometric: overcoming (pushing against a fixed strap on a rigid dynamometer), unilateral dominant-leg knee extension, hip at 70 degrees, knee at 65 degrees (measured 62 ± 2 degrees under load).
Session: warm-up 50%, 75%, 90% contractions; then explosive-contraction training (3 sets x 10, "as fast and hard as possible" to at least 80% of maximal voluntary torque for about 1 s, 5-10 s rest between (the manuscript gives both), rate-of-torque biofeedback); then sustained-contraction training (3 sets x 6 maximal contractions of 3 s, "push as hard and fast as possible", 20-30 s between (the manuscript gives both), target line = best torque so far). 2 min between sets; about 25 min per session.
Frequency and duration: 3 sessions a week for 2 weeks, then 4 a week for 2 weeks (14 sessions, 4 weeks).
Tests: two duplicate sessions before and two after; maximal and explosive voluntary contractions and evoked twitches and octets at 35, 50, 65, 80 and 95 degrees; quadriceps surface EMG at six sites normalised to the angle-specific M-wave.
Key Results
| Angle (0 = straight knee) | Training group, maximal torque change | Control | Training group, EMG at maximal torque |
|---|---|---|---|
| 35 degrees (shorter quadriceps) | +5% ± 2% (ES ≤ 0.37) | no change | +8% ± 3% |
| 50 degrees | +11% ± 2% (ES ≥ 0.98) | no change | +19% ± 5% |
| 65 degrees (training angle) | +12% ± 2% (ES ≥ 0.98) | no change | +18% ± 5% |
| 80 degrees | +7% ± 2% (ES ≤ 0.37) | no change | not significant |
| 95 degrees (longer quadriceps) | no significant change | no change | not significant |
| Test | Result |
|---|---|
| Training group, time x angle (within group) | P < 0.001; 65 > 35, 80, 95 (P ≤ 0.025); 50 > 35 and 80 (P ≤ 0.004); 80 > 95 (P = 0.009) |
| Training vs control, main effect of group | P = 0.011 |
| Training vs control, group x angle | P = 0.029; 65 vs 95 (P ≤ 0.010, ES ≥ 0.91) and 50 vs 95 (P ≤ 0.007) |
| EMG, group x angle | P = 0.587 (no angle specificity of activation) |
| Early explosive torque (T50, T100) | no change in either group (P ≥ 0.123) |
| Late explosive torque (T150) | rose only at 65 degrees (P = 0.019, ES 0.38; about 7%); no group difference |
| Evoked octet peak torque | +~4.3% at 65 and 50 degrees in the training group; no group difference |
| Reliability (controls, CVw) | maximal torque 3.9%, T150 7.8%, T100 9.6%, T50 24.4%, EMG at maximal torque 9.8% |
Mechanisms
Measured: EMG at maximal torque rose 8-19% at the training angle and the two straighter angles, three of the four angles where strength rose, with no change in controls; but neither within- nor between-group tests showed EMG changing more at one angle than another.
Authors' interpretation: increased neuromuscular activation probably underlies the strength gains, and the lack of angle-specific EMG is likely a measurement-noise problem (EMG varied far more than torque).
Authors' interpretation: the 4% rise in evoked octet torque suggests a small morphological change, probably early hypertrophy, already after 4 weeks; muscle size was not measured.
Authors' interpretation: explosive and maximal contractions in the same session may interfere, so the early rapid-force gain seen after explosive training alone (REF-1161) did not appear. Untested here; the authors call for replication.
Practical Application
Does an isometric hold only strengthen that one angle? Knee holds at 65 degrees for 4 weeks gave +12% there, +5 to 11% at nearby angles and nothing 30 degrees deeper. Untrained men, one leg; train near the angle you need.
Carry-over reached a much straighter knee (+5% at 35 degrees) but not a more bent one (95 degrees), so expect gains to spread more toward the shorter-muscle side.
To reach the bent-knee range (as in a deep squat or the top of a pedal stroke for a hip-flexed rider), hold there or train through the range. The authors' data show no gain 30 degrees deeper than the training angle; that cycling and squatting angles correspond is the base's inference.
Do not expect early rapid force from mixing explosive and maximal holds in one session. Here the combination raised maximal and late force only. If rapid force matters, keep explosive contractions separate or alone (REF-1161); the separation advice is the base's reading of the authors' interference hypothesis.
Limitations
Small, short and narrow: 13 trained, 9 controls, 4 weeks, young untrained men, one muscle group, one training angle. The control group was smaller than the power analysis assumed for each group.
The test range was ±30 degrees around the training angle, not the whole range of motion.
Testing five angles per session reduced reproducibility compared with the lab's single-angle work; the authors averaged duplicate sessions to compensate. Early explosive torque (T50) was too noisy (CVw 24.4%) to detect change.
Muscle size, architecture and tendon were not measured, so neural vs morphological causes cannot be separated.
No funding statement in the submitted manuscript; the first author was supported by the CAPES Foundation, Ministry of Education of Brazil (affiliation). No conflict-of-interest statement in the manuscript version read.
How this entry was checked and changed
Why this level
parallel RCT, n=22 (13 training, 9 control); the first trial to test angle specificity with a group x angle interaction against a control. Small control group, short intervention, young untrained men only; read from the authors' final submitted manuscript (full text)
Full text (checked 2026-10-16)
RT group (n=13; age, 22 ± 3 years) and control group (CON; n=9; age, 23 ± 3 years); power analysis target n= 16 (full text).
14 training sessions in 4 weeks (x3/week for 2 weeks and then x4/week for 2 weeks) at a 65° knee-joint angle (full text).
ECT: three sets of ten repetitions, "as fast and hard as possible" up to ≥80% of maximum voluntary torque for ~1 s; SCT: three sets of six repetitions, MVC for 3 s with 20 s rest between contractions (full text).
MVT increases: 65°, +12% ± 2%; 80°, +7% ± 2%; 50°, +11% ± 2%; 35°, +5% ± 2% (full text).
Between-group main effect of group P= 0.011 and group x angle interaction P=0.029 (full text).
Normalized EMG at MVT increased at 35° (+8% ± 3%), 50° (+19% ± 5%) and 65° (+18% ± 5%); group x angle P= 0.587 (full text).
T150 increased only at 65° (Bonferroni P= 0.019; ES= 0.38); Octet PT increased by ~4.3% at 65° and 50° (full text).
Within-participant CV: MVT, 3.9%; T150, 7.8%; T100, 9.6%; T50, 24.4% (full text).
Actual knee angles during MVCs: 33º ± 2º, 50º ± 1º, 62º ± 2º, 78º ± 2º, 91º ± 2º (full text).