“Use leg drive,” “keep the upper back planted” and “arch harder” are usually intended to organize the lifter against the floor, bench and bar. Yet an arch can contain different amounts of thoracic extension, lumbar extension, pelvic rotation and scapular retraction. The scientific record begins by asking which configuration changes the task, and which part of the force-transfer chain was actually measured.
The path here runs from base of support to bar, using regional spinal language, a contact audit and a firm external-versus-internal measurement boundary. The complete athlete–bench interface and Launch Pad evidence progression sits in the athlete-to-bench interface article; the shoulder-mechanics article contains the detailed shoulder force-model record.
Configuration record · name what changed
“Arch” Is Not a Regional Spinal Measurement
A bench-press arch usually describes elevated thoracic geometry and reduced bar travel, but two visually similar arches can distribute motion differently. A complete description should distinguish thoracic position, lumbar position, pelvic orientation, shoulder-girdle setup, head and buttocks contact, foot placement and the distance between the bar and torso.
The 2026 IPF rulebook defines the competition boundary: during the lift, the lifter’s head, shoulders and buttocks must remain in contact with the bench surface and the feet must remain flat on the platform or blocks.[1] Legality is not a biomechanical optimum and is not a medical judgment.
Practitioner context: Tate and Simmons on the pressing base
Practitioner systems treat these contacts as one setup problem. Dave Tate’s EFS Bench Manual coordinates foot position, upper-back and lat tension, shoulder-blade position and pressure through the upper back as a repeatable pressing base.[2] Louie Simmons’ Westside manual likewise varies foot position, arch, grip, upper-back and lat strategy in service of the lift.[3] These are practitioner models: they generate hypotheses about contact and force exchange, but they do not measure regional spinal motion or internal tissue load.
System map · follow every boundary
How the Feet, Bench, and Bar Share the Task
The chain is not a one-way pipe. Forces and motion are constrained simultaneously by the floor, bench and bar, and the lifter must maintain equilibrium. “Force transfer” should therefore be operationalized: less torso drift, a different pressure path, a change in horizontal bar force, a higher external load, a different velocity, or a more repeatable setup.
Operational definition · choose an endpoint
What would “better force transfer” look like in a record?
Less movement at the support
Torso drift and contact repeatability.
A different external strategy
Pressure distribution and three-dimensional force direction.
A different test result
Load, bar velocity and repeatability under defined conditions.
Leg drive: plausible pathways and the measurements needed
Leg drive is especially easy to over-interpret because the coaching action is visible but the transfer pathway is not. Foot force may help maintain the chosen torso position, alter pressure against the bench, change horizontal body drift or contribute to a more consistent touch point. A barbell study that measures only upper-extremity EMG cannot determine which of those pathways occurred. A complete test requires synchronized force and kinematic records at more than one boundary.
Horizontal forces matter. Mausehund and colleagues showed that omitting lateral barbell forces can materially alter calculated shoulder and elbow net moments during bench pressing.[4] A vertical-only bar trace cannot reconstruct the complete upper-extremity force strategy.
Arch record · compare protocols, not silhouettes
Three Studies, Three Protocols, No Universal Effect
Protocol matters · two 1RM comparisons
A visible arch does not guarantee a larger lift
Read each result within its protocol. A guided bar and a free bar create different comparison conditions.
1RM in the Smith-machine comparison
SD 17.9 kg
SD 18.4 kg
kg · bars show means; SD printed separately
1RM difference: arched minus flat
95% CI 0.0 to 8.4 kg
kg · point = reported difference; line = 95% CI
García-Ramos · Smith machine
Natural lumbar arch and moderate scapular retraction compared with a pronounced arch.
Bartolomei · free barbell
Flat and arched 1RM tests; subsequent sets used 50%, 70% and 90% of each technique’s own 1RM.
Read the complete three-study comparison and inference boundaries
| Study | Participants + contrast | Measured result | Inference boundary |
|---|---|---|---|
| García-Ramos et al. 2021 | 11 competitive male powerlifters; natural arch/moderate scapular retraction versus pronounced arch. | 1RM: 115.9 ± 17.9 versus 115.7 ± 18.4 kg (p = .942; ES = .01); no velocity differences. | Highly specific familiar lifters and protocol; a pronounced arch did not automatically improve performance.[5] |
| Cudlip et al. 2022 | 20 experienced men; powerlifting arch versus standardized technique at 25%, 50% and 75% of self-reported 1RM. | Arch increased latissimus mean/peak EMG; standardized technique produced approximately 8% larger integrated shoulder moments. | EMG and modeled/net moments are not spinal tissue load, pain or injury risk.[6] |
| Bartolomei et al. 2024 | 15 highly trained power athletes; flat versus arched technique at 50%, 70% and 90% of respective 1RM. | Arched 1RM was 4.2 kg higher (95% CI 0.0–8.4; ES .22). Flat produced greater displacement and slightly higher mean/peak velocity; power did not differ. | Small performance effect in a defined cohort; not a universal benefit or medical-safety result.[7] |
Bartolomei et al. · displacement
The travel difference depends on the load
Extra bar displacement with the flat technique
95% CI 34.0 to 61.0 mm
95% CI 24.0 to 55.0 mm
95% CI 6.0 to 50.0 mm
mm · point = reported difference; line = 95% CI
Pinto and Dickerson add a geometric mechanism: in 20 men tested at 25%, 50% and 75% 1RM, an arched technique reduced vertical bar displacement by about 11% and the bar-to-glenohumeral-joint moment arm by about 20%.[8] That explains how external mechanical advantage may change. It does not measure spinal tissue stress.
Together, the studies reject two simplistic positions. An arch is not mechanically meaningless, and more arch is not automatically better. Its effect depends on how the alternative condition is defined, the lifter’s proficiency, the load and the endpoint.
Measurement firewall · external is not internal
External Performance Does Not Reveal Internal Tissue Load
AMM Research · Visual 15 · contact and measurement
Visible posture does not resolve internal loading
Hands ↔ bar
Measure vertical and mediolateral force components alongside bar position and time.
Body ↔ bench
Record contact, pressure or reaction force with a defined method. Appearance alone supplies no force magnitude.
Feet ↔ floor
Record foot placement and floor reactions. A leg-drive cue is not itself a force measurement.
External record
Contacts + motion + forces
Synchronized measurements describe the external task.
Internal estimate
Regional inputs + model
Internal loading estimates depend on model assumptions and measured inputs.
Clinical record
Separate outcomes
Pain, injury and long-term tolerance require their own evidence.
Contact points · bar displacement · bar velocity · external load · ground/bench/bar forces · surface EMG · whole-body and segment kinematics · performance reliability.
Segmental spinal loading · passive-tissue strain · internal muscle force · pain mechanism · injury probability · long-term clinical outcome.
Read the limitations of inverse dynamics, models and surface EMG
Inverse dynamics and musculoskeletal models can estimate joint moments or selected reaction forces, but those outputs depend on inputs and assumptions. Surface EMG records electrical activity from sampled muscles; it is not a force-transfer meter. A faster bar can indicate a different external task outcome without specifying how internal loading changed.
The right claim follows the endpoint. If torso drift decreases, say stability or repeatability improved. If horizontal bar force changes, say the external force strategy changed. If 1RM increases, say performance improved in that test. Do not promote any of those results into a spinal-safety claim without the required evidence.
Field audit · make setup reproducible
What Coaches Can Record Before Drawing Conclusions
Feet and floor
Record foot position, heel contact, block use and slipping. A coaching cue cannot compensate for an unreported base.
Buttocks contact
Record whether and where contact is maintained, plus any visible shift during unrack and press.
Thoracic and lumbar setup
Use consistent landmarks or video views. “Big arch” is not a repeatable regional measurement.
Contact and drift
Record scapular cue, surface position and movement relative to pad markings or indexed settings.
Path and force direction
Track touch point, vertical travel, velocity and—when researching mechanics—horizontal force.
Performance and tolerance
Record load, repetitions, exertion and symptoms during and after. Repeatability is an outcome, not a visual impression.
For coaching, this audit makes setups comparable. For research, it defines covariates and independent variables. For persistent pain, neurological symptoms, postoperative restrictions or return-to-sport decisions, it does not replace clinical assessment.
AMM boundary · design history versus measured evidence
Support Geometry Can Change the External Task—Mechanism Still Requires Measurement
AMM’s earlier Inventing The Launch Pad article framed shoulder clearance, lumbar support, leg drive and stability as one equipment-design problem.[9] That page documents the origin of the question. It does not independently prove the proposed mechanism, and its broad wording is narrowed here to measurable contact, force and kinematic hypotheses.
Kidwell et al. · 2026 · Table 1
The external outcomes, side by side
- 10 trained men
- 70% 1RM
- 5 repetitions
- Randomized crossover
Vertical bar displacement
SD 7 cm
SD 4 cm
cm · bars show means; SD printed separately
Mean bar velocity
SD 0.09 m/s
SD 0.09 m/s
m/s · bars show means; SD printed separately
Peak bar velocity
SD 0.11 m/s
SD 0.09 m/s
m/s · bars show means; SD printed separately
Concentric power
SD 103 W
SD 93 W
W · bars show means; SD printed separately
Read the original numerical findings in full
In the published Launch Pad randomized crossover, 10 resistance-trained men performed five repetitions at 70% of established 1RM. The support condition increased mean vertical bar displacement from 38 ± 7 to 44 ± 4 cm (p = .005), mean velocity from 0.40 ± .09 to 0.47 ± .09 m/s and peak velocity from 0.57 ± .11 to 0.66 ± .09 m/s (both p < .001). Concentric power was not significantly different: 249 ± 103 versus 271 ± 93 W (p = .071).[10]
Those are acute external and EMG outcomes from a specific thoracic-support intervention. The study did not measure ground reaction force, bench reaction force, lumbar loading, segmental motion or an injury outcome. It therefore shows that the support condition was not mechanically neutral for the measured task—not how force was transmitted internally.
The next research step should synchronize floor, bench and three-dimensional bar forces with regional kinematics, contact mapping and performance. That design can test whether a configuration changes drift, force direction or efficiency. Longitudinal and clinical endpoints would still be needed for adaptation or safety claims.
Research path · return to the evidence spineRevisit the Athlete-to-Bench Interface for the complete direct Launch Pad study progressionReferences
- International Powerlifting Federation. (2026). Technical Rules Book, effective March 1, 2026, version 3. Official rulebook. ↩
- Tate D. (2008). EFS Bench Manual. elitefts, pp. 8, 12. Practitioner source supplied for this review. Related official elitefts article. ↩
- Simmons L. (2009). Westside Barbell Bench Press Manual. Westside Barbell, pp. 24–28. Practitioner source supplied for this review. Official manual page. ↩
- Mausehund L, Werkhausen A, Bartsch J, Krosshaug T. (2022). Understanding bench press biomechanics—The necessity of measuring lateral barbell forces. Journal of Strength and Conditioning Research, 36(10), 2685–2695. doi:10.1519/JSC.0000000000003948. ↩
- García-Ramos A, Pérez-Castilla A, Villar Macias FJ, Latorre-Román PÁ, Párraga JA, García-Pinillos F. (2021). Differences in the one-repetition maximum and load–velocity profile between the flat and arched bench press in competitive powerlifters. Sports Biomechanics, 20(3), 261–273. doi:10.1080/14763141.2018.1544662. ↩
- Cudlip AC, Maciukiewicz JM, Pinto BL, Dickerson CR. (2022). Upper extremity muscle activity and joint loading changes between the standard and powerlifting bench press techniques. Journal of Sports Sciences, 40(9), 1055–1063. doi:10.1080/02640414.2022.2046937. ↩
- Bartolomei S, Caroli E, Coloretti V, Rosaci G, Cortesi M, Coratella G. (2024). Flat-back vs. arched-back bench press: examining the different techniques performed by power athletes. Journal of Strength and Conditioning Research, 38(7), 1200–1205. doi:10.1519/JSC.0000000000004778. ↩
- Pinto BL, Dickerson CR. (2021). Vertical and horizontal barbell kinematics indicate differences in mechanical advantage between using an arched or flat back posture in the barbell bench press exercise. International Journal of Sports Science & Coaching, 16(3), 756–762. doi:10.1177/1747954120982954. ↩
- Advanced Muscle Mechanics. Inventing The Launch Pad. Historical design article, reviewed August 2026. Original AMM article. ↩
- Kidwell JA, Yamamoto T, Hetherton KJ, Truneh N, Bright JJ, Blatney AE, et al. (2026). Acute effects of thoracic-spinal elevation via a novel bench press pad on sEMG and barbell kinetics in resistance-trained males. International Journal of Exercise Science, 19(1), 1003. doi:10.70252/IJES2026103. ↩