The Generalisation Trap: Deconstructing the 1955 Cadaver Heuristic

By Forwod Team, April 16, 2026

The accuracy of any digital physics engine is fundamentally dependent on the quality of its baseline inputs.

A computational model can process calculations at immense scale, and its formulas mapping Newtonian gravity and fluid dynamics can be mathematically flawless. But if the biological parameters fed into those formulas are inaccurate, the resulting power outputs will always be metrologically invalid.

When tracking human kinematics, software must calculate the exact physical distance a load (or a body) travels. To calculate the mechanical Work of an athlete performing a strict pull-up or a heavy back squat, the engine must determine the specific lever lengths of their limbs and the precise location of their biological Centre of Mass (CoM).

Because most fitness applications operate using only a user's total height and body mass, algorithms are forced to rely on anthropometric heuristics—standardised mathematical tables that estimate limb mass and joint fulcrums as a percentage of total body weight.

When we audited the foundational sports science literature to build the Forwod Calculation Engine, we uncovered a systemic failure at the absolute baseline of modern commercial biomechanics.

The 1955 Baseline

The historical benchmark for human Body Segment Parameters (BSPs)—the data dictating segmental mass, geometric joint centres, and centre of gravity—was largely established by biomechanist Wilfred T. Dempster in 1955.

Dempster’s research was conducted by physically dissecting exactly eight elderly, Caucasian male cadavers (with an average age of 68.5). The original purpose of this study was not to measure elite athletic performance; it was commissioned by the military to map spatial requirements for aircraft pilot ejection seats. This dataset was later adapted into widely used digitised kinematic models by Plagenhoef, Evans, and Abdelnour in 1983.

While advanced clinical laboratories and elite motion-capture software have since migrated to updated academic models (such as Zatsiorsky-Seluyanov’s 1983 gamma-ray scans or de Leva’s 1996 clinical adjustments), the commercial fitness software industry still overwhelmingly relies on the foundational derivatives of Dempster and Plagenhoef.

The Mathematical Anomaly & Phantom Joules

The fundamental danger of relying on legacy data is that the original arithmetic rarely survives modern scrutiny.

A rigorous, programmatic audit of the widely cited Plagenhoef (1983) legacy arrays reveals severe mathematical anomalies that outright corrupt physics calculations. In various heavily cited revisions of this data designed to improve the accuracy of the trunk and head, researchers isolated the thorax, abdomen, and pelvis into separate volumetric spheres. Because these spheres technically overlap at the geometric joint centres, if an algorithm unknowingly iterates over these baseline metrics, the sum of the human body mathematically accounts for exactly $108.14\%$ of total body mass.

An algorithm that inadvertently assumes a human being possesses $108\%$ mass results in a fundamentally corrupted mechanical work calculation.

If a $100\text{ kg}$ athlete is processed by this legacy heuristic, the software actively treats them as weighing $108\text{ kg}$. It artificially inflates their Joules and Work Capacity on every single repetition. To prevent their platforms from violating the laws of thermodynamics, engineers are forced to write arbitrary, hardcoded normalisation loops simply to shrink the cadaver data back down to $100\%$, completely destroying any claim to true metrological accuracy.

Linear Scaling in a Non-Linear Reality

Furthermore, even if a platform utilises the "modern" 1996 de Leva tables, they are still falling into the Generalisation Trap. All of these tables rely on static, linear scaling. They assume that an elite $100\text{ kg}$ strength athlete possesses the exact same limb-mass distribution and tissue density ratios as a $70\text{ kg}$ amateur, simply scaled up by a uniform multiplier.

In reality, modern physiological adaptation, extreme lower-body muscular hypertrophy, and differing adipose tissue densities actively shift an athlete's true biological centre of mass. We cannot measure modern, multimodal athletes using static, table-based anatomical assumptions.

The Forwod Override

At Forwod, identifying the "Generalisation Trap" forced us to architect a fundamentally different approach to digital anthropometry.

We engineered our calculation pipeline to mathematically reject static legacy dependencies. Rather than blindly applying historical tables to modern telemetry, our engine utilises strict algorithmic bounding and mathematical overrides to ensure that an athlete's physical mass and segment mechanics obey the laws of physics before a single Joule is calculated.

By treating anthropometry as a dynamic, clinically constrained variable rather than a naive lookup table, the Forwod engine actively quarantines these historical anomalies. This ensures that the mechanical Work and Power data remains pristine, verifiable, and free of phantom Joules.

Selected Clinical Context & Further Reading

• Dempster, W. T. (1955). Space requirements of the seated operator... Wright Air Development Center.
• Plagenhoef, S., Evans, F. G., & Abdelnour, T. (1983). Anatomical data for analyzing human motion. Research Quarterly for Exercise and Sport.
• de Leva, P. (1996). Adjustments to Zatsiorsky-Seluyanov's segment inertia parameters. Journal of Biomechanics.