The history of biomechanics explains how people learned to investigate living movement through mechanics: first by relating anatomy to levers and balance, then by recording motion, measuring forces and constructing models that connect the two. Its central achievement is a clearer distinction between what a body does and what evidence can explain that behaviour.
The gold standard is to understand what each method made knowable and where its conclusions stopped. A sequence of photographs can reveal positions without measuring muscle force. A force plate can record an external interaction without directly measuring every load inside a joint. This guide follows selected historical landmarks, with original worked examples that make those distinctions useful. The numerical cases are fictional teaching models, not measurements from historical experiments or assessments of an individual’s movement.
Use The Gold Standard of Biomechanics for the modern subject. This historical root belongs to the Gold Standard master and the Education parent: the aim is understanding that transfers when an unfamiliar image, graph or claim appears.
Choose your biomechanics history route
Chapters 1–3: Understand mechanical models
Chapters 4–6: Read movement records
Chapters 7–9: Connect images and forces
Chapters 10–12: Interpret models and conventions
Chapter index
Understand mechanical models · Chapters 1–3
1. Ask how the evidence changed, not only who came first
2. Read Borelli’s lever reasoning as a model with assumptions
3. Learn why moment arms matter more than an attractive diagram
Read movement records · Chapters 4–6
4. See what sequential photography added to movement study
5. Compare Muybridge and Marey without collapsing their methods
Connect images and forces · Chapters 7–9
7. Recover the depth that a photograph leaves uncertain
8. Add external force without pretending to see inside the body
9. Follow impulse from a force trace to a change in velocity
Interpret models and conventions · Chapters 10–12
10. Explain what inverse dynamics actually estimates
11. Treat computer models as explicit hypotheses
12. Understand the role of institutions, standards and shared definitions
Test transfer and continue learning · Chapters 13–14
CHAPTER 1 OF 14 · Understand mechanical models
1. Ask how the evidence changed, not only who came first
A history of biomechanics can be organised around people, institutions, instruments or questions. The most useful introductory account connects all four without pretending that one date created the entire field. Observing movement, describing anatomy and calculating mechanical relationships are related activities, but they are not identical achievements.
Start with three questions. What moved? What forces or moments were involved? What properties of the living system influenced the response? The first concerns motion. The second concerns mechanical interactions. The third introduces anatomy, materials, control and biological variation. An experiment may answer one question well while leaving the others open.
Distinguish a description from an explanation
Suppose a drawing shows an arm in two positions. It describes a change in configuration. Add timestamps and a scale, and it can support estimates of movement over time. Add external loads and a justified mechanical model, and it may support further inferences about the forces required. Each extra claim needs extra evidence or an explicit assumption.
This sequence is not a claim that history proceeded neatly through three universal stages. Different methods developed alongside one another, and older methods remain useful. A carefully labelled drawing can still explain a lever more clearly than an elaborate animation. The question is whether the representation supports the particular inference being made.
When reading a historical account, write four short notes: the biological question, the available measurement, the mechanical reasoning and the remaining uncertainty. If the account supplies only a famous name and a celebrated result, look for the missing connection. What did the investigator actually see or record? How did that evidence rule out an alternative?
This habit also improves credit. A discovery can depend on specimen preparation, photography, instrument construction, mathematics and interpretation. Recognising those contributions does not require inventing undocumented stories. It requires avoiding the assumption that the person named in a short summary performed every part of the work alone.
CHAPTER 2 OF 14 · Understand mechanical models
2. Read Borelli’s lever reasoning as a model with assumptions
Giovanni Borelli’s De motu animalium is a major landmark in applying mechanics to animal bodies. The University of Oklahoma’s historical collection record dates the work to 1680–81 and describes its analysis of bones and muscles through levers and simple machines. That record supports a specific claim about a work and its approach; it does not require declaring that nobody previously reasoned mechanically about living things.
The enduring intellectual move is to replace a vague account of effort with a relationship between force and geometry. A force acting farther from a pivot can produce the same turning effect as a larger force acting closer to it. The relevant distance is the perpendicular moment arm, not automatically the length of the entire bone.
Original balance calculation
Imagine an ideal horizontal lever. A downward load of 30 newtons acts 0.30 metres from a pivot. Its moment magnitude is 30 × 0.30 = 9 newton-metres. A supporting force acts with a perpendicular moment arm of 0.03 metres on the appropriate side of the balance. To oppose the load’s moment, it must have magnitude 9 ÷ 0.03 = 300 newtons.
The large supporting force is not an arithmetic mistake. A short moment arm requires more force to produce the same moment. This example explains why an external load does not directly reveal the magnitude of an internal force.
The model omits the lever’s own weight, multiple muscles, changing geometry, acceleration and tissue deformation. Those omissions make the calculation readable, but they also define its limits. It is not a prescription for lifting and not a reconstruction of a particular person’s arm.
A useful historical comparison asks which features an early mechanical model captured and which required later evidence. It is possible to respect a model’s insight without treating every anatomical detail as settled. A simple lever explains an important relationship; a living joint contains more than a simple lever.
CHAPTER 3 OF 14 · Understand mechanical models
3. Learn why moment arms matter more than an attractive diagram
A diagram can show two forces with the same magnitude while hiding a difference in their turning effect. The moment about a chosen point depends on the force and its perpendicular distance from that point. Change the line of action and the moment changes, even if the force magnitude remains fixed.
This is one reason mechanical reasoning brought something new to anatomical description. Naming a muscle or locating a bone does not by itself specify the mechanical effect of an interaction. Geometry connects the anatomical arrangement with a quantitative prediction.
Worked example: change one assumption
Keep the 30-newton external load from Chapter 2, but move it from 0.30 metres to 0.20 metres from the pivot. The external moment becomes 6 newton-metres. With the same 0.03-metre supporting moment arm, the required balancing force becomes 200 newtons. The load’s weight did not change; its mechanical demand in this model did.
Alternatively, keep the external moment at 9 newton-metres but increase the supporting moment arm to 0.045 metres. The supporting force again becomes 200 newtons. Two different geometric changes produce the same numerical force estimate. The answer alone does not identify which change occurred.
The repair is to label the model. State the pivot, force direction, perpendicular distance and whether acceleration is neglected. A line drawn between two anatomical landmarks may not be the required moment arm. An apparent angle in a photograph may not be the corresponding three-dimensional angle.
For a useful paper exercise, draw a door viewed from above. Apply a force near the hinge and then farther from it. Keep the force perpendicular to the door. Explain why the turning effect changes before inserting numbers. Then change the force direction and identify the perpendicular distance again. This is a geometry exercise, not a recommendation to test loads on a body.
The historical lesson is that a good mechanical diagram exposes assumptions. A decorative diagram merely makes a conclusion look plausible. Always ask which distances and directions the calculation actually uses.
CHAPTER 4 OF 14 · Read movement records
4. See what sequential photography added to movement study
Eadweard Muybridge’s Animal Locomotion work made phases of movement available for repeated inspection. The University of Pennsylvania archives document the university’s role in the project, while the Library of Congress catalogue identifies the published collection. The 1887 publication is a landmark in the recording and comparison of consecutive movement phases.
The important change was not simply that a picture became sharper. A sequence let viewers compare positions that ordinary observation could merge into one impression. It could challenge an assumed ordering of events or show a configuration too brief to inspect comfortably with the unaided eye.
What a sequence can and cannot establish
Imagine five images of a moving foot. If their temporal order is known, they can show the order of visible positions. If their time intervals are also known, they can support an estimate of movement rate. If the spatial scale is known and the movement stays in the calibrated plane, distances may be estimated. Remove any of these conditions and the corresponding claim weakens.
The images alone do not directly reveal muscle tension, joint contact force or energy consumption. Those are different quantities. A convincing photograph can answer an observation question without answering a causal question about all the forces involved.
Historical pictures also contain choices. Camera position, background, task instructions, selected subjects and selected frames influence what appears in the record. A published plate is evidence, but it is not an unmediated view of every possible movement under every condition.
To study a sequence responsibly, describe only what is visible first. Then list the metadata needed for a stronger claim: time interval, scale, viewpoint and task. Finally separate measured features from inferred ones. This exercise turns an iconic historical image into a lesson in evidence rather than treating it only as an illustration beside a date.
CHAPTER 5 OF 14 · Read movement records
5. Compare Muybridge and Marey without collapsing their methods
Étienne-Jules Marey used photography to analyse movement in ways that included successive exposures on a single plate. The Science Museum Group’s record of a fixed-plate chronophotographic camera describes an instrument from about 1882 and the use of visible markings to make movement easier to analyse. Its 1890 camera record documents another part of this instrument history.
The distinction between separate images and multiple phases recorded together matters. Each arrangement makes some comparisons easy and others harder. Overlaid positions can reveal a trajectory clearly, but overlapping features can become difficult to separate. Separate frames can preserve an individual configuration clearly, while requiring the reader to compare across images.
A modern teaching comparison
Imagine recording a point at five evenly spaced times. On a single coordinate sheet, mark all five positions. On five separate sheets, mark one position per sheet. Both records contain the same supplied coordinates, yet they invite different reading habits. The first emphasises the path; the second emphasises each moment.
Now remove the timestamps. Neither display alone guarantees that equal spacing between points means constant speed. Unequal exposure intervals could produce the same spatial pattern. A presentation choice cannot replace missing timing information.
This is a useful way to understand historical instruments: ask what the recording arrangement preserves and what it obscures. Do not assume that every sequence was produced with the same apparatus, exposure method or timing accuracy. Read the collection record and, where available, the original methodological description.
The broader contribution of these photographic approaches was to make movement into a record that could be revisited, compared and measured. Their limitations remained consequential. Two-dimensional recording, visibility of landmarks and timing all affect the transition from “I can see the pattern” to “I can defend this numerical estimate.”
CHAPTER 6 OF 14 · Read movement records
6. Calculate speed only after establishing time and scale
Motion analysis requires a spatial reference and a temporal reference. Pixels alone are not metres, and frame numbers alone are not seconds. Historical and modern records both need those connections before a physical speed can be calculated. A smoothly playing video can still have an incorrect or unknown time scale.
Worked example: four intervals, not five
Suppose a point appears at positions 0, 0.10, 0.20, 0.30 and 0.40 metres in five frames captured at 20 frames per second. The interval between neighbouring frames is 0.05 seconds. From the first frame to the fifth there are four intervals, so elapsed time is 0.20 seconds. Average speed over that displacement is 0.40 ÷ 0.20 = 2 metres per second.
A common mistake divides by five frame intervals and obtains 1.6 metres per second. Counting the observations is not the same as counting the gaps between them. Draw five marks on paper and count the four spaces to make the error visible.
Now suppose the same coordinates were recorded at 10 frames per second. The elapsed time becomes 0.40 seconds and the average speed becomes 1 metre per second. The spatial sequence is unchanged. The time calibration changes the physical interpretation.
These examples assume known frame intervals, correct position calibration and motion in the relevant plane. If the object moves towards or away from the camera, a simple image-plane distance does not necessarily equal the full distance travelled. If the trajectory bends between frames, endpoint displacement and path length also differ.
A repair checklist can be written as a sentence: establish the scale, establish the time intervals, define the measured point and distinguish displacement from distance. Only then calculate speed. This is why the history of movement recording is also a history of measurement conventions, not merely a succession of better cameras.
CHAPTER 7 OF 14 · Connect images and forces
7. Recover the depth that a photograph leaves uncertain
A photograph projects a three-dimensional arrangement into two dimensions. Some information is lost. Two different spatial configurations can produce similar projected positions. This limitation matters when reading historical plates, assessing a modern movement video or interpreting an apparent joint angle in a social-media image.
The existing spatial-angle specialist guide demonstrates why even known segment lengths may not resolve every depth ambiguity. Its examples concern labelled geometric segments rather than clinical assessments, which is the appropriate distinction here too.
Original coordinate example
Consider two points whose image-plane coordinates differ by 3 units horizontally and 4 units vertically. Their projected separation is 5 units because 3² + 4² = 25. If their depth coordinates are equal, the spatial separation is also 5. If they differ by 12 units in depth, the spatial separation becomes 13, because 3² + 4² + 12² = 169.
The same two-dimensional separation is compatible with very different spatial distances. The missing depth is not recovered by enlarging the picture. Additional viewpoints, calibration or justified constraints are needed. Without them, the answer must remain conditional.
This does not mean all two-dimensional analysis is invalid. If the question concerns movement approximately confined to a known plane, a properly arranged two-dimensional record may be useful. The limitation becomes serious when the interpretation quietly assumes that out-of-plane movement is absent without checking it.
For historical reading, ask whether the investigator’s conclusion required full spatial reconstruction. A claim about visible contact order may need less information than a claim about an exact three-dimensional joint configuration. Judge the method against the question it was meant to answer.
A good caption therefore identifies the view and the quantity being shown. “Projected angle in this view” is more defensible than an unqualified “joint angle” when depth has not been resolved. Clear language preserves the information the image really contains.
CHAPTER 8 OF 14 · Connect images and forces
8. Add external force without pretending to see inside the body
Movement records describe positions and their changes. Force measurements add evidence about interactions. A force plate, for example, can measure aspects of the interaction between a subject and the supporting surface. It does not directly report the force of each muscle or the contact load at every joint.
The distinction follows from the boundary of the measured system. An external force acts across that boundary. Internal forces act between parts inside it. A whole-body balance can constrain their combined effects without uniquely determining how individual tissues share the load.
Worked example: support and acceleration
Use a fictional body of mass 60 kilograms and take gravitational acceleration as 9.81 metres per second squared. Its weight magnitude is 588.6 newtons. If the measured upward support force is 708.6 newtons and no other vertical external forces matter, the net upward force is 120 newtons. The centre-of-mass acceleration is 120 ÷ 60 = 2 metres per second squared upward.
The calculation requires subtracting weight. Dividing the full support force by mass would give 11.81 metres per second squared, which is not the net acceleration in this model. The support force and the resultant force are different quantities.
This example assumes a consistent vertical direction, an appropriate mass and a correctly measured force. It does not identify the forces at the knee, ankle or hip. Those require additional modelling and information. Nor does one instant of acceleration determine the entire movement trajectory without initial conditions and a time history.
The historical advance is a broader ability to connect recorded movement with measured interaction, not a magical removal of inference. To read a force graph carefully, label the sensor output, the reference direction and the system boundary. Ask whether the plotted quantity is raw force, net force or a normalised value. Similar-looking curves can represent different physical quantities.
CHAPTER 9 OF 14 · Connect images and forces
9. Follow impulse from a force trace to a change in velocity
A force acting over time changes momentum. In a simple one-dimensional case, the integral of net force over time equals mass multiplied by the change in velocity. This creates a bridge between a force record and a movement prediction. It also makes baseline errors consequential because an error can accumulate during integration.
The existing force-baseline and jump-height specialist article explores that issue in depth. Here the purpose is to understand why recording a time history can reveal something a single peak value cannot.
Worked example: equal peaks, unequal impulses
Imagine a constant net upward force of 120 newtons acting for 0.20 seconds on the 60-kilogram model. The impulse is 24 newton-seconds. The velocity change is 24 ÷ 60 = 0.40 metres per second. If the same force acts for 0.10 seconds, the impulse is 12 newton-seconds and the velocity change is 0.20 metres per second.
The peak force is identical in both cases. The duration changes the result. This is why “the higher peak wins” is not a general explanation of movement performance. The relevant quantity depends on the question, and sometimes the area under the curve matters more than its tallest point.
Now suppose a constant uncorrected error of 10 newtons is added to the force record over 0.20 seconds. It adds 2 newton-seconds of spurious impulse. Dividing by 60 gives an error of approximately 0.033 metres per second in the inferred velocity change. The error is small at each instant but systematic across the interval.
These are arithmetic demonstrations, not advice to perform a jump test. The assumptions include fixed mass, a correct initial velocity where absolute velocity is required and consistent force conventions. A defensible historical or modern analysis states the integration limits and the baseline treatment. Without them, an impressive curve is not yet a complete calculation.
CHAPTER 10 OF 14 · Interpret models and conventions
10. Explain what inverse dynamics actually estimates
Inverse dynamics starts with a model and observed motion, together with relevant external forces, and solves for forces or moments consistent with that model’s equations. The OpenSim documentation explains the distinction between net inter-segmental force and bone-on-bone joint force. These should not be treated as interchangeable outputs.
The idea is easier to understand by contrasting two directions of reasoning. In a forward problem, supplied forces and initial conditions help predict motion. In an inverse problem, supplied motion and other known quantities help estimate the required net mechanical effects. Both directions depend on the model, and neither turns an unmeasured quantity into a direct observation.
Worked example: a net moment does not identify every muscle force
Suppose a simplified joint requires a net moment of 10 newton-metres. One hypothetical arrangement produces 10 newton-metres in one direction with no opposing moment. Another produces 15 in that direction and 5 in the opposite direction. Both give the same net value of 10.
If a calculation reports only the net moment, it cannot choose between these arrangements. The second may involve greater opposing internal forces, despite the identical net turning effect. Additional physiological assumptions or measurements are required to estimate how individual structures contribute.
This is a basic example of non-uniqueness, not a claim that real muscles can be assigned arbitrary forces without constraints. Anatomical paths, activation, force-generating capacity and passive tissues restrict the possible solutions. The important point is that one net output does not by itself contain all those details.
In historical perspective, computational tools expanded the scale and complexity of calculations that could be handled. They did not abolish the need to define a model or test its assumptions. A colourful animation should therefore be read alongside the inputs, parameter choices and validation, rather than as a direct view inside a moving body.
CHAPTER 11 OF 14 · Interpret models and conventions
11. Treat computer models as explicit hypotheses
A computer model can make assumptions more inspectable because its equations, parameters and input data can be recorded. It can also make weak assumptions less visible when a polished animation distracts from them. The scientific value lies in what can be checked, not in the visual realism of the output.
OpenSim’s workflow documentation distinguishes inverse dynamics from static optimisation, which introduces criteria to estimate individual muscle forces. That distinction illustrates why successive stages of a workflow should not be merged into one claim that “the computer measured the muscles.”
A sensitivity exercise
Return to the ideal lever that must balance 9 newton-metres. With a moment arm of 0.030 metres, the estimated force is 300 newtons. With 0.033 metres, it becomes approximately 272.7 newtons. A 10 percent increase in the moment arm produces about a 9.1 percent decrease in the force estimate, because the relationship is inverse.
This exercise does not establish the uncertainty of a real model. It shows how to investigate sensitivity: change one parameter within a justified range and observe the effect on the conclusion. A model may be robust for one question and sensitive for another. Predicting the direction of a change can be easier than estimating its exact magnitude.
Validation should be matched to the intended use. Reproducing the data used to tune a model is useful, but it is not the same as predicting an independent observation. A model that describes one movement well may not retain the same accuracy under a different task or for a different population.
For a practical reading exercise, find three labels in any simulation report: measured input, assumed parameter and predicted output. Then ask which independent evidence tests the output. If those categories cannot be separated, the report may be difficult to interpret even if its animation looks convincing. The correct response is to clarify the chain, not to assume that all modelling is unreliable.
CHAPTER 12 OF 14 · Interpret models and conventions
12. Understand the role of institutions, standards and shared definitions
The International Society of Biomechanics records its founding at Penn State University on 30 August 1973. Its official account describes a field spanning biological systems and multiple disciplines. This is an institutional milestone, not the beginning of every biomechanical idea or experiment.
Institutions help researchers compare work by supporting meetings, publications and shared terminology. Comparison becomes difficult when two laboratories use the same word for different quantities or different coordinate conventions for the same quantity. A numerical result needs enough context to be translated between methods.
Worked example: opposite signs can describe the same event
Suppose Laboratory A defines upward as positive and reports an upward force of +100 newtons. Laboratory B defines downward as positive and reports the same upward force as −100 newtons. The numbers disagree in sign but agree physically. Before judging a conflict, translate both into the same coordinate convention.
Now suppose A reports force in newtons and B reports force divided by body weight. A value of 1.2 in B’s system is dimensionless, not 1.2 newtons. For a teaching body weight of 500 newtons, that normalised value corresponds to 600 newtons. Units and normalisation must be restored before comparison.
Standards and recommendations do not make all methods equivalent. They help expose where methods differ. A shared convention can reduce avoidable disagreement while leaving substantive questions about measurement error, sample selection and biological interpretation.
Historical accounts benefit from the same discipline. Label a publication date as a publication date, an instrument date as an instrument date and a society’s foundation as an organisational event. Avoid compressing all three into a single timeline labelled “discoveries.” The field grows through practical coordination as well as through individual intellectual breakthroughs, and both deserve a clear place in the story.
CHAPTER 13 OF 14 · Test transfer and continue learning
13. Transfer the reasoning to a new movement claim
A fictional report states that a new shoe reduces knee loading because a side-view video shows a smaller apparent knee angle and a force plate shows a lower peak vertical force. The report gives no three-dimensional reconstruction, no joint-load model and no force-time comparison. What can a careful reader conclude?
The observations may establish a change in a projected angle and a measured external peak under the tested conditions. They do not automatically establish a reduction in every internal knee load. The apparent angle may depend on viewpoint. The force peak omits duration. Internal load sharing may differ even when the external result looks smaller.
Build the alternative before accepting the explanation
For an original force example, let condition A produce a net force of 100 newtons for 0.20 seconds and condition B produce 80 newtons for 0.30 seconds. Their impulses are 20 and 24 newton-seconds respectively. B has the lower peak but the greater impulse. The example is deliberately rectangular so the calculation is transparent; real curves require integration rather than assuming this shape.
For an internal-moment example, let both conditions have a net joint moment of 10 newton-metres. One could involve 12 opposing 2, while another involves 20 opposing 10. The same net moment does not establish identical internal contributions. These invented alternatives show why the report’s preferred causal story is not determined by the stated observations alone.
The next step is to define the target claim. Does “loading” mean a measured external force, a net moment, an estimated contact force or accumulated exposure? Each calls for different evidence. A general claim of reduced injury risk would require evidence beyond the mechanical examples supplied here.
The transferable answer is precise: describe the observed change, name the uncertainty and identify the additional measurement or model needed. This avoids two errors—accepting an unsupported conclusion and dismissing useful observations simply because they do not answer every question. The history of biomechanics is valuable because it teaches where each new method strengthens that chain.
CHAPTER 14 OF 14 · Test transfer and continue learning
14. Choose a learning route and interrogate the sources
Does biomechanics begin with Borelli?
Borelli’s work is a landmark in mechanical analysis of animal bodies. Calling one person the sole beginning of a field can hide earlier observation, anatomy and mechanics. This article uses the documented 1680–81 work as a concrete starting point for a particular analytical approach, not as a claim that all previous thinking was absent.
Did motion photography measure forces?
It recorded movement configurations. With appropriate timing, scale and geometric assumptions, images can support kinematic estimates. Inferring forces requires additional information and modelling. A visible movement and its mechanical causes are connected, but they are not the same kind of evidence.
Is a modern three-dimensional model always better?
It may answer questions that a simpler method cannot, but complexity alone is not a guarantee. The important tests are whether the model represents the relevant geometry, whether inputs are sufficiently reliable and whether the conclusion has been validated for the intended question. A simple model can be preferable when it captures the necessary relationship transparently.
Can these examples tell me how to change my exercise technique?
No. They teach interpretation of measurements and models. They do not assess an individual’s anatomy, symptoms, capacity or training needs. Their practical value is in helping readers ask clearer questions about evidence rather than copying a numerical target or a photographed pose.
How can I study the subject without a laboratory?
Begin with paper models and supplied datasets. Balance moments about a pivot, calculate displacement over known intervals and compare peak force with impulse. Label assumptions before calculating. Change one assumption and explain what changes in the result. Then read an actual study and identify which quantities were measured and which were inferred.
A useful sequence is to work Chapters 2–3 for levers, Chapters 4–7 for recorded motion, Chapters 8–9 for force histories and Chapters 10–11 for models. Finish with the changed-context case in Chapter 13. A learner who can explain why a conclusion is limited has understood more than a learner who merely remembers the names of instruments.
Which historical records should I read first?
Start with the University of Oklahoma’s Borelli record, the Penn archives’ Muybridge collection context, and the Science Museum Group’s Marey camera record. Compare the purpose of each source: a book record, an archival account and an instrument catalogue preserve different evidence. Use the ISB’s institutional account for its founding milestone, and the OpenSim inverse-dynamics explanation for the modern computational distinction.
These are selected landmarks, not a complete global history. A fuller account would need additional research into many laboratories, clinical traditions, animal studies, engineering practices and participants whose work is not represented by this short route. Keeping that boundary explicit is more honest than treating a few famous examples as the whole field.
Continue with History of Biophysics for related measurement questions, the Biomechanics root for the modern discipline and the How X Works hub for explanations across subjects. Return to the Gold Standard master when choosing another learning route.