The Osteogenic Window
Your skeleton is not hardware. It is not the calcium scaffolding you learned about in secondary school biology, fixed in place the moment you stopped growing. Bone is living tissue. It has a blood supply, a nervous system connection, a metabolic rate. It is being disassembled and rebuilt, continuously, at rates that would alarm you if you could watch the process under a microscope. Right now, as you read this sentence, osteoclasts are dissolving old bone matrix while osteoblasts are depositing new mineral into the vacated spaces. By the time you finish this chapter, roughly one million osteoclasts will have completed their work somewhere in your body, and one million osteoblasts will have begun theirs.
The ratio between those two activities determines whether you are gaining bone, maintaining bone, or losing it. And that ratio is governed, more than by any other single factor, by mechanical load.
This is not a metaphor. It is a physical transduction mechanism. When bone tissue is compressed or bent or twisted, it deforms slightly—the technical term is strain—and that deformation generates electrical signals. Piezoelectric signals, to be precise: bone crystals produce a voltage when stressed, in the same way that quartz does in a watch. The osteocytes embedded throughout your bone matrix sense these signals and respond by upregulating bone formation at exactly the sites where the strain is highest. Take the load away, and the signal stops. The osteoblasts slow their work. The osteoclasts continue theirs. The net balance shifts toward resorption.
This principle was first formally articulated in 1892 by the German anatomist Julius Wolff, who observed that the trabecular architecture of cancellous bone—the spongy inner network visible in cross-sectional images—is oriented precisely along the lines of maximal mechanical stress, as if the bone had calculated the optimal structural solution to the loads placed upon it. Wolff’s insight, now known as Wolff’s Law, was correct in its basic form. He was wrong about some of the mechanisms—piezoelectric signaling and osteocyte mechanotransduction were unknown to nineteenth-century anatomy—but the core observation holds. Bone builds where it is stressed. It resorbs where it is not.
The consequential question, the one that exercise physiologists have spent decades trying to answer with precision, is this: how much stress is enough?
The Threshold
In 1989, two Swedish researchers named Jan Nilsson and Alf Thorstensson published a paper in Acta Physiologica Scandinavica that ought to be required reading for every person who has ever laced up a pair of running shoes, bought a bicycle, or stepped onto an elliptical trainer. The paper’s title was unremarkable—“Ground Reaction Forces at Different Speeds of Human Walking and Running”—and it generated nothing like the popular attention of, say, a study on resveratrol or intermittent fasting. But what Nilsson and Thorstensson measured, and what their data revealed, maps the entire landscape of human locomotion in a way that fundamentally changes how you should think about exercise.
Ground reaction force is simply the force the floor pushes back against your foot with every step. By Newton’s third law, it equals the force your foot is pushing down with—which is, in turn, a function of your body weight plus any momentum effects from the way you land. A force plate embedded in a laboratory floor measures this continuously as a subject walks or runs across it, producing a characteristic curve for each stride. Nilsson and Thorstensson had subjects walk and run at systematically varied speeds, from a slow amble to near-maximal effort, and they measured ground reaction forces throughout.
The results were unambiguous. Walking at comfortable speeds produces peak vertical ground reaction forces of approximately 1.0 to 1.5 times body weight. The curve is smooth, double-humped—one peak at initial contact, one at push-off—and the forces ramp up and down gradually. There is no sudden shock. Running is entirely different. Even at slow jogging speeds, peak vertical forces jump to approximately 2.0 times body weight. At faster running speeds, they reach 2.5 to 2.9 times body weight. And critically, running produces what biomechanists call an impact transient: a sharp spike in force at heel contact that is largely absent in walking gait, a spike that loads the skeleton at a rate orders of magnitude faster than the smooth walking curve. It is the difference between a gradual squeeze and a hammer blow.
Unloaded walking GRF: 1.0–1.5× body weight. Running GRF: 2.0–2.9× BW with sharp impact transient. Rucking at 25–30% BW: ~1.6–1.8× total system mass—above the osteogenic threshold, below the injury threshold.
FIGURE: Ground Reaction Force Comparison
Walking (unloaded): Smooth double-peak curve, 1.0–1.5× BW. No impact transient.
Running: Sharp impact spike at heel strike (the "Hammer Blow"), followed by propulsive peak. 2.0–2.9× BW. Impact transient drives 50–75% annual injury rate.
Rucking (25–30% BW): Same smooth double-peak as walking, amplitude raised to ~1.6–1.8× total system mass. Osteogenic stimulus preserved. Impact transient absent.
The smooth wave vs. the hammer blow. Same cardiovascular stimulus. Fundamentally different mechanical cost.
Between these two regimes lies a gap. Walking tops out around 1.5 times body weight. Running begins at 2.0 times body weight. There is a mechanical no-man’s land, a range of forces that walking simply cannot reach at normal speed with an unloaded body.
That gap is precisely where the osteogenic threshold sits.
The exact number varies somewhat in the literature—bone biology is not physics, and individual variation is substantial—but the preponderance of evidence places the threshold for meaningful osteogenic stimulation somewhere around 1.4 to 1.7 times body weight in the axial skeleton, depending on loading rate and repetition. Below that, the mechanical signal is insufficient to tip the osteoblast-osteoclast balance toward net formation. Unloaded walking at 5 kilometres per hour, pleasant as it is for cardiovascular health and mental clarity, generally does not reach this threshold. It produces bone maintenance at best—and in people with already-compromised bone density, not even that.
Running, at its lower end, crosses the threshold. But running also crosses something else: the injury boundary. Ground reaction forces of 2.0 to 2.9 times body weight, combined with the impact transient and the repetitive nature of running gait, produce substantial injury rates in recreational runners—estimates range from 37 to 56 percent annually depending on how broadly injury is defined and which populations are surveyed. Stress fractures. Tibial stress reactions. Patellofemoral pain syndrome. Plantar fasciitis. Iliotibial band syndrome. These are the characteristic injuries of running gait, the predictable consequence of repeatedly loading a biological structure at or near its damage threshold. (It should be noted, as Chapter One discusses, that recreational running at moderate volumes is also chondroprotective—runners have lower rates of knee osteoarthritis than sedentary controls. The injury question and the joint-health question are not the same question.)
There is, embedded in the Nilsson and Thorstensson data and confirmed by subsequent modeling work, a solution to this problem. It is not complicated. It is, in retrospect, obvious.
Add weight to a walking body.
The Arithmetic of Load
When you add a backpack to your frame, several things happen simultaneously in the biomechanical system. Your total system mass increases. Your center of gravity shifts slightly. Your gait kinematics adapt—you take shorter strides, slightly increase cadence, flex forward a few degrees at the trunk. And your ground reaction forces rise, because the floor is now supporting more weight with every step.
The relationship is not perfectly linear. The body’s neuromuscular system actively attenuates impact as loads increase, employing strategies—greater knee flexion at contact, adjusted ankle stiffness—that partially buffer the added force. But the net effect is clear and consistent. Loads of 20 to 30 percent of body weight, carried in a properly fitted backpack at a brisk walking pace, produce peak vertical ground reaction forces in the range of 1.4 to 1.7 times body weight.
At the lower end of rucking loads—say, 15 percent of body weight at normal walking speed—you are still below the osteogenic threshold. At the upper end—35 to 40 percent, or heavy military loads—you begin approaching the stress levels that cause injury at scale over repeated exposures. The sweet spot, mechanically speaking, is the 25 to 30 percent body weight range at a pace at or above 1.5 metres per second, approximately 5.4 kilometres per hour.
A study by Xu and colleagues published in 2016 in the Journal of Biomechanical Engineering used an integrated musculoskeletal and finite-element model, built from a female subject’s actual anatomy, to trace the path of these forces through the tibia and lower extremity joints during loaded walking at various percentages of body weight. The results were instructive and slightly alarming. Carrying 30 percent of body weight increased knee joint reaction forces by 26.2 percent compared to unloaded walking. Ankle joint reaction forces increased by 16.4 percent. Hip joint reaction forces increased by 19.0 percent. And the model identified regions of high cumulative tibial stress—the anterior crest and medial surface—that correspond exactly to the anatomical sites where military recruits develop stress fractures when loads and distances increase too rapidly.
The Xu model is a cautionary tale for those who would ruck with 50-kilogram loads and call it exercise. But it is also a validation for the modest end of the loading range. Joint reaction forces at 30 percent body weight are elevated meaningfully above unloaded walking—enough to constitute a mechanical stimulus—while remaining well below the forces generated by running. The strain is there. The damage threshold is not crossed.
The geometry of the loading spectrum, mapped by Nilsson and Thorstensson in 1989 and refined by three decades of subsequent biomechanical research, looks like this when you lay it out: unloaded walking at the bottom, insufficient for osteogenesis; running above it, sufficient but injurious at scale; and loaded walking at 25 to 30 percent body weight occupying the mechanically optimal band between them. Enough force to signal the osteoblasts. Not enough to threaten the tissue.
There is a further consideration that the raw force numbers alone do not capture: the role of loading rate. The speed at which force builds on the skeleton with each footfall—measured in units of body weight per second—is distinct from peak force magnitude, and it matters independently. Running’s impact transient loads bone not only harder but faster than walking gait. The rate-dependent response of osteocytes means that a rapid, sharp load signal is processed differently than a slow, smooth one. Highly dynamic loading—jumping, for instance—can produce osteogenic signals at lower peak forces than slow loading at higher magnitudes, because the rate of change drives mechanosensor activation. Walking with a backpack preserves the smooth loading rate of walking gait while elevating the magnitude. This combination—elevated peak force, unchanged loading rate—may represent a particularly efficient osteogenic signal: strong enough to matter, gradual enough to avoid damage.
Speed matters too. Castro and colleagues demonstrated that gait cadence is a significant modulator of ground reaction force during loaded walking: faster walking with a backpack produces substantially higher impact peaks than the same loaded walking at a slow shuffle. At a cadence of 120 steps per minute—roughly a purposeful 5 to 6 kilometre-per-hour pace—loaded walking approaches its upper limit of vertical ground reaction force generation while remaining in a walking gait. This is not a coincidence with the pace recommendations that emerge from military load carriage doctrine and recreational rucking practice. Brisk but sustainable. Not a saunter.
This is the osteogenic window. Brisk pace, appropriate load, adequate duration—the biology responds to all three simultaneously.
The Crisis It Addresses
There is a figure in the Nilsson and Thorstensson data that the fitness industry has collectively declined to take seriously: the zone below 1.5 times body weight, where unloaded walking operates. The industry sells walking as exercise. It is exercise—it produces cardiovascular adaptation, metabolic improvement, mood benefit, and a dozen other documented health effects. But the industry does not tell you, and in many cases does not know, that the exercise it is selling you does not speak bone’s language at the force levels most walkers generate. Your heart hears the message. Your lungs hear it. Your mitochondria respond. Your osteoblasts do not. They are waiting for a stronger signal, and the gentle footfalls of a 5-kilometre morning stroll are not loud enough.
This matters because the people who most need to hear the bone-building signal are precisely the people who have been told that walking is their exercise prescription. Older women. Postmenopausal women. Women who have been warned off running by their knees or their orthopaedist or their own painful experience with shin splints or stress fractures. They walk. They walk diligently. They accumulate their ten thousand steps. And their bone density continues its slow, relentless decline, because the signal is below threshold.
The solution is not to run. The solution is to carry something.
Somewhere in your body, if you are a woman past forty, the clocks are running in different directions. Estrogen has been declining since your early-to-mid thirties—gradually at first, then, around perimenopause, with gathering speed. And estrogen, among its many functions, is the primary brake on osteoclast activity. It promotes bone formation, inhibits bone resorption, and protects the trabecular microarchitecture that gives bone its compressive strength without excessive weight. When estrogen declines, the brake releases.
The numbers are not abstract. In the first three to twelve months after the final menstrual period, bone mineral density at the lumbar spine drops by approximately 0.9 percent. At the femoral neck—the narrowest point of the upper femur, where most osteoporotic hip fractures occur—it drops by approximately 0.7 percent. These are annual figures, and they compound. The average rate of bone loss in the first five to ten years after menopause runs at 2 to 4 percent per year. Over the initial six years post-menopause, a woman can lose approximately 15 percent of her total bone mass. Not 15 percent of a marginal reserve. Fifteen percent of the structural foundation of her skeleton.
The clinical consequences of this trajectory are not subtle. Approximately one in three women over fifty will sustain an osteoporotic fracture. Hip fractures—the endpoint of severe femoral neck bone loss combined with a fall—kill roughly one in four patients within a year of the event. Of those who survive, 40 percent cannot walk independently a year later, and 60 percent still require assistance with activities of daily life. The person who fractures a hip at seventy-five is not simply recovering from a broken bone. She is fighting for her functional independence, often unsuccessfully.
There are approximately 10 million Americans with osteoporosis and 44 million with bone mineral density low enough to classify as osteopenia—the pre-osteoporotic state that most clinicians treat as a warning rather than a crisis. They are wrong to be so sanguine. Osteopenia is where the clock is actually running. By the time osteoporosis is diagnosed, structural damage to trabecular microarchitecture has often been sustained that cannot be fully reversed by any intervention, pharmaceutical or mechanical. The window for prevention—the window that corresponds to the osteogenic window in exercise terms—is the decade before and the decade after menopause.
A large cross-sectional study examining osteoporosis prevalence across female age groups found that rates at the lumbar spine were 3 percent in the 30-to-39 age group, 3.4 percent in the 40-to-49 age group, 14.3 percent in the 50-to-59 age group, 18.6 percent in the 60-to-69 age group, and 36.4 percent in women over seventy. This is not a gradual slope. It is a cliff that begins in the decade of menopause and steepens with every subsequent decade.
Male osteoporosis prevalence, by comparison, remains comparatively stable across the same age ranges. This is not because men’s bones are fundamentally superior. It is because men have no menopause—no estrogen withdrawal event—to initiate the accelerated resorption cascade. Women are playing a different game, under different rules, against a more aggressive opponent.
If you are a woman reading this at forty, the timer started years ago. If you are reading at fifty, the critical window is open and will not stay open. If you are reading at sixty, the investment you make now in bone density is the investment your seventy-year-old self will draw against when she reaches for the top shelf, misses a step, or navigates an icy pavement. Bone you build in the next five years is bone you may not be able to rebuild in the five years after that.
The Evidence
In 2022, a systematic review and meta-analysis by Sánchez-Trigo, Rittweger, and Sañudo, published in Osteoporosis International, asked a precise question: does non-supervised weight-bearing exercise improve bone mineral density in adult women, and if so, by how much? The analysis drew on data from 668 women across multiple studies and reported results as standardised mean differences—a statistical metric that allows comparison across studies measuring the same outcome in different units.
The findings for the overall population were encouraging: weight-bearing exercise produced a standardised mean difference of 0.40 for lumbar spine bone mineral density and 0.51 for femoral neck bone mineral density. These are meaningful effect sizes in a clinical population. But the sub-group analysis—the portion of the results that every woman over forty should read carefully—was more striking.
In women who already had osteopenia or osteoporosis, the effect sizes rose substantially. Lumbar spine bone mineral density improved with a standardised mean difference of 0.73. Femoral neck bone mineral density improved with a standardised mean difference of 0.85. These are large effects by the standards of exercise intervention research—effects that are, in the femoral neck, comparable in magnitude to those produced by bisphosphonate therapy, the first-line pharmaceutical treatment for osteoporosis.
The implication embedded in that sub-group finding is important and counterintuitive. Women with the greatest bone loss—the population for whom bone health is most urgent—are also the population that responds most vigorously to weight-bearing mechanical stimulus. This is consistent with the biology: bone that has undergone architectural degradation retains more mechanosensitive capacity than bone that has already adapted to chronic loading at the same level. The osteoblast system, in other words, responds most aggressively when the deficit is greatest. Bone that needs help responds to help.
It is worth being specific about what these effect sizes mean in practice, because standardised mean differences are not intuitive. A standardised mean difference of 0.85 at the femoral neck means that the average woman in the exercise group moved roughly 0.85 standard deviations in the direction of better bone density compared to the control group. In a clinical context, where one standard deviation in femoral neck bone mineral density corresponds to a T-score change of 1.0, and where each standard deviation reduction in T-score roughly doubles fracture risk, an effect size of 0.85 represents a clinically substantial shift. It is not a complete solution to postmenopausal bone loss. It does not replace the estrogen that menopause removes. But it meaningfully offsets the trajectory—and in a population for whom pharmaceutical options carry side effect profiles that many women reasonably decline, a safe, sustainable, side-effect-free intervention with effect sizes of this magnitude deserves to be a first-line recommendation, not an afterthought.
The direct evidence for weighted walking specifically—as opposed to weight-bearing exercise in its various forms—is anchored by a landmark study from Christine Snow and colleagues published in 2000 in the Journal of Gerontology. Snow’s team followed postmenopausal women over five years, comparing a group who performed a weighted vest and jumping program three times per week with a control group who did not. At the end of five years, the exercise group had maintained femoral neck bone mineral density. The control group had lost bone. Not a dramatic loss, but a statistically significant, clinically meaningful one—the kind of cumulative loss that, compounded over a decade, becomes the structural deficit that precedes a fracture.
The Snow study used jumping in addition to the weighted vest, which introduces a force component above what walking with a loaded pack produces. But the underlying principle is identical: external load applied to a walking and standing skeleton, creating mechanical strain above the remodeling threshold, at sufficient frequency and duration to sustain osteoblast activity. The study also demonstrated something that is easy to overlook in the enthusiasm for acute exercise effects: five years of sustained mechanical loading preserved bone that would otherwise have been lost. The intervention was not about growing dramatically denser bones. It was about not losing the bones you have.
A note on transferability is warranted here. The Snow RCT, the Sánchez-Trigo meta-analysis, and the Beavers INVEST trial all use weighted vests, not backpacks. Vests and backpacks are not biomechanically identical. A backpack places load posteriorly, shifting the body’s center of mass rearward and inducing a compensatory forward trunk lean of five to fifteen degrees at loads of 20 to 40 percent of body weight. A vest distributes mass circumferentially, keeping the center of mass close to its unloaded position and minimising postural compensation. Both systems increase the gravitational load borne through the skeleton during stance phase — the osteogenic stimulus at the femur, tibia, and lumbar spine is present with either — but the loading rate profiles differ. Backpacks produce higher impact transients at heel strike due to the pendular dynamics of posteriorly placed mass, while vest-borne loads are more rigidly coupled to the torso and show damped transients. The US Army Load Carriage Decision Aid required a separate metabolic equation for vest-borne loads because backpack equations systematically mismatch vest carriage energy costs (Looney et al., 2024). No head-to-head RCT has directly compared vest and backpack carriage for bone mineral density outcomes; the literatures are complementary but not interchangeable. When vest-based BMD findings are applied to rucking, the mechanistic transfer is plausible but the direction of any difference is uncertain: backpack loading may produce a modestly greater osteogenic stimulus at equivalent loads, but at the cost of a slightly higher injury risk profile. Both directions remain untested.
The Beavers Negative
In 2025, Beavers and colleagues published a result in JAMA Network Open that anyone citing weighted loading for bone health is obligated to address honestly (Beavers et al., 2025).
The study was a twelve-month randomised controlled trial of 150 older adults (mean age approximately 67 years) with obesity undergoing intentional weight loss. The intervention group wore weighted vests during daily activities. The primary outcome was bone mineral density at the hip and spine. The result: the weighted vest did not prevent bone loss during weight loss. The intervention group lost bone at essentially the same rate as the control group.
This is the largest and most rigorous RCT of weighted loading for bone preservation in older adults. It directly contradicts the expectation that external load, by itself, protects bone.
Context matters. The Beavers participants were wearing vests passively during daily activities—standing, walking around the house, doing errands—not performing structured rucking sessions at Zone 2 intensity for forty-five to sixty minutes. The concurrent intentional weight loss created a caloric deficit that may have overridden any osteogenic signal from the vest. And passive vest wearing produces a static load increase without the dynamic, cyclical loading pattern that walking gait generates at each step—the very pattern that Wolff’s Law and the mechanotransduction literature identify as the key stimulus.
These caveats are real, but they should not be used to dismiss the finding. The Beavers study establishes that simply adding weight to the body is not sufficient to protect bone. The activity performed while carrying the weight—its intensity, duration, and biomechanical characteristics—appears to matter as much as or more than the weight itself. This is consistent with the mechanobiology: osteocytes respond to dynamic strain and fluid flow, not static compression. A vest worn while sitting at a desk delivers static load. A pack carried during a brisk walk delivers rhythmic, dynamic loading at every step. Whether this distinction is sufficient to produce meaningfully different bone outcomes remains a hypothesis, not a demonstrated fact.
Every decade you fail to protect bone mineral density is a decade you cannot recover. This is not pessimism. It is physiology.
Why Peak Bone Mass Is the Number Nobody Talks About
Before the menopause conversation, there is a prior conversation that receives almost no public attention. It concerns peak bone mass—the maximum bone mineral density and structural integrity your skeleton will ever achieve, typically reached sometime in your late twenties to early thirties depending on skeletal site. Osteoporosis, in the simplest mechanistic framing, results from one of two failures: either you did not build enough bone during the accumulation phase, or you lost too much of it during the depletion phase. Most of the public conversation focuses on the depletion phase—menopause, aging, medication side effects. The accumulation phase is rarely discussed, and it is where a substantial fraction of lifetime fracture risk is determined before any decline has begun.
The concept of peak bone mass matters because it establishes the baseline from which postmenopausal losses are calculated. A woman who reaches menopause with a femoral neck T-score of +1.0—indicating bone mineral density one standard deviation above the young adult mean—has a substantially different fracture trajectory than a woman who reaches menopause with a T-score of -0.5. Both women will lose bone during the menopausal transition at roughly the same rate. But the woman who started higher will take longer to cross the clinical threshold for osteoporosis, and she may never cross it before other causes of mortality intervene. The woman who started low may be in fracture territory before she turns sixty.
Physical activity during adolescence and young adulthood is a critical determinant of peak bone mass. The mechanosensitivity of bone is greatest during growth, when osteoblast activity is at its lifetime peak and the skeleton is actively responding to every new mechanical stimulus with construction rather than mere maintenance. Weight-bearing activities that generate substantial ground reaction forces—sports, loaded carries, jumping, resistance training—build skeletal reserve during this window that cannot be fully replicated later. This is the first argument for beginning loaded walking as early as is appropriate, and for encouraging the young women in your life to carry weights.
But even for those who are reading this at forty-five or fifty-five, having not heard this argument during the relevant developmental window, the peak bone mass conversation matters for a different reason: it establishes the urgency of beginning now. The bone you build in your forties is not equivalent to the bone you would have built in your twenties, but it is bone you will have in your seventies. The bone you fail to protect in your fifties is bone your hip will not have when it meets the pavement at seventy-two.
Wolff Understood, Not Memorised
It is worth pausing on Julius Wolff, not as a textbook entry but as a scientist trying to solve a problem. In 1892, Wolff was working before X-rays, before the electron microscope, before any of the molecular biology that would eventually explain why his observations were correct. He was looking at thin sections of bone under a light microscope, at the patterns of trabeculae—the delicate internal struts of spongy bone—and noticing something that demanded explanation.
The trabeculae were not random. They were organised. They ran in predictable directions, along predictable axes, and when Wolff compared these axes to the mechanical loads the bone experienced during normal function—loads that engineers of his era could calculate using the mathematics of structural mechanics—the correspondence was striking. The bone’s internal architecture looked like the solution to an engineering optimisation problem. Minimal material, maximum structural efficiency, struts placed exactly where they needed to be to resist the forces the bone would encounter in life.
What Wolff inferred—what we now know to be mechanistically accurate through the work of twentieth-century researchers who identified osteocytes as the load-sensing cells and described the transduction pathways in molecular detail—is that bone is a self-optimising structure. It does not grow uniformly in response to load. It grows at the specific locations where strain exceeds the remodeling threshold, and it resorbs at locations where strain falls below the maintenance threshold. The result, over years of habitual loading, is a skeletal architecture matched to the habitual demands placed upon it.
This is why athletes who bear their weight differently develop different bone architectures. Tennis players have asymmetrically denser cortical bone in their dominant arm—not because they were born with asymmetric arms, but because decades of repeated loading drove asymmetric osteogenesis. Long-distance runners develop denser tibias at the locations of maximal compressive stress. Gymnasts who begin training in childhood have bone mineral densities at the lumbar spine that exceed age-matched norms by margins that persist into adulthood.
And it is why astronauts lose bone at catastrophic rates during spaceflight. In microgravity, the mechanical loading that normally maintains the osteoblast-osteoclast balance disappears almost entirely. Without gravity pressing the skeleton into the ground with every step, without the habitual loads that human bone evolved to experience, the remodeling equation rapidly shifts toward net resorption. Astronauts returning from six-month stays on the International Space Station have lost 1 to 2 percent of femoral neck bone mineral density per month in some cases—rates that compress decades of postmenopausal decline into months. Countermeasures exist—resistance training protocols, vibration platforms, pharmacological agents—but none fully substitute for the gravitational loading the skeleton was built to receive. Space medicine has, inadvertently, run the most rigorous experiment possible in Wolff’s Law: remove the mechanical signal entirely, and bone loss begins within weeks.
The converse experiment is run every time a person with a sedentary lifestyle begins a loaded exercise program. The signal arrives. The osteoblasts respond. The architectural remodeling begins. It is slower in adults than in children, and it is slower in older adults than in younger ones—the mechanosensitivity of the osteocyte network diminishes with age, though it never disappears entirely—but the direction of effect is reliable. Wolff’s Law does not stop operating at fifty. It operates more slowly, and the magnitude of response is reduced, but the mechanism remains intact.
And it is why sedentary individuals—people whose bodies bear their own weight and almost nothing else, who walk on flat surfaces in cushioned shoes and sit in chairs that support most of their trunk—have bones that are architecturally adapted to minimal load. Not bad bones, necessarily. Appropriate bones. Bones that have received the signal to be exactly as dense as they need to be for a life of walking to the car and back.
If that life then encounters a fall—a patch of ice, a kerb misjudged in low light, a stumble on a hiking trail—those appropriately minimal bones are asked to absorb a force they were never trained to absorb. The femoral neck, which has been losing trabecular struts at 2 to 4 percent per year since menopause, which never received a decade of mechanical signal above the osteogenic threshold, meets an impact force of 4 to 8 times body weight as the person falls sideways onto their hip.
And it breaks.
This is not bad luck. It is not a random event. It is the terminal consequence of a decades-long failure to provide the skeleton with the mechanical signal it requires to maintain itself. The biology is deterministic. The intervention window was open for years. It simply was not used.
The Goldilocks Zone and What It Demands
The mechanically optimal band for bone formation is not infinitely wide. It has boundaries, and both boundaries matter.
On the lower boundary: load must be sufficient to exceed the osteogenic threshold. For most adults, this means approximately 1.4 times body weight or greater in peak vertical ground reaction force, achieved through either loaded walking at 25 to 30 percent body weight, or through the higher natural forces of running and jumping. Unloaded walking does not reliably cross this threshold at normal speeds. Cycling does not cross it at all—it is not a weight-bearing exercise. Swimming does not cross it. Neither does rowing, or the elliptical trainer, or the exercise bicycle in the corner of the hotel gym. None of these activities are useless. But none of them speak the language that bone responds to.
On the upper boundary: peak forces must remain below the threshold at which cumulative microdamage exceeds the skeleton’s repair capacity, a threshold that is lower in bone that has already been architecturally compromised by age or hormone withdrawal. Running at 2.0 to 2.9 times body weight, with its associated impact transient, exceeds this boundary at scale for many people—particularly women with osteopenia, in whom stress fracture rates during running programs are not trivial. A study published in 2024 in PM&R examining a female runner population with a mean age of forty found that 25.4 percent reported prior stress fractures, and the odds of bone stress injury increased significantly in women who already had osteopenia or osteoporosis. Running is asking bone to both absorb damaging forces and repair itself from them simultaneously. For well-mineralised young bone, the repair wins. For older bone with structural deficits, this is not a reliable bet.
Loaded walking at 25 to 30 percent body weight sits inside these two boundaries. It exceeds the lower threshold—the mechanical signal is present and sufficient—while remaining below the upper threshold at any walking speed a healthy adult would sustain for an hour. The Xu model data confirms this: joint reaction force increases of 19 to 26 percent above unloaded walking are meaningful for osteogenesis without approaching the magnitudes that initiate stress injury cascades. The loading rate—the speed at which force builds on the skeleton with each footfall—remains characteristic of walking rather than running, preserving the gradual, double-peaked force profile that distributes strain across the stance phase rather than concentrating it in an impact transient.
This is the geometry of the solution: not harder, not softer. Appropriately loaded. Appropriately paced. Appropriately sustained.
A pack on your back. A brisk walking pace. Sufficient duration to accumulate the mechanical dose the skeleton requires. Done three to four times per week, maintained across decades rather than abandoned after months.
What Delay Costs
The conversation about bone density is often framed as a crisis to be managed after the fact—a pharmaceutical response to a diagnosis, an exercise program prescribed after a DEXA scan returns a T-score below negative 2.5. This framing is late. It is not wrong—exercise-induced improvements in bone mineral density are achievable even in established osteoporosis, and the Sánchez-Trigo data showing effect sizes of 0.73 to 0.85 at femoral neck and lumbar spine in osteopenic and osteoporotic women confirm this. But improvements in already-compromised bone are incremental. They are maintenance and modest recovery, not reconstruction.
The trabecular microarchitecture—the internal lattice of bone—does not fully regenerate once lost. Pharmaceutical bisphosphonates, which suppress osteoclast activity and slow bone resorption, can modestly increase bone density as measured by DEXA, but they do not rebuild the three-dimensional trabecular network that gives bone its structural integrity. A DEXA scan measures mineral content, not architecture. Two bones can have the same mineral density and radically different fracture resistance, depending on whether the trabeculae are organised in load-bearing orientations or have been partly resorbed and incompletely replaced with thicker but less numerous struts.
The window for building bone that has both high mineral density and optimal architecture is the first four decades of life. The window for preventing catastrophic loss of what was built is the decade before and the decade after menopause. After that, the realistic objective shifts from building to preserving—from construction to damage limitation.
Every year between thirty and fifty that passes without mechanical stimulus above the osteogenic threshold is a year in which the bone bank earns no interest and may in fact accrue fees. Every year after fifty in which the skeleton receives insufficient loading is a year in which the structural reserve your seventy-year-old self will depend on narrows. The compound interest works in reverse. The clock runs one direction.
This is not an argument for panic. It is an argument for immediacy. Not the urgency of crisis, but the urgency of a time-sensitive investment—the kind where the earlier you begin, the greater the return; where the cost of waiting is not merely delay but permanent loss of principal.
The intervention requires a backpack, a load that falls between 25 and 30 percent of your body weight, a pair of shoes, and a pace brisk enough to produce the mechanical stimulus the science specifies. It requires repetition—not once, not occasionally, but regularly, across years, with progressive adjustment of load as strength and capacity develop. It requires you to understand that you are not merely getting cardiovascular exercise or burning calories. You are talking to your skeleton in the only language it reliably hears.
The message is simple. The mechanism is ancient. Bone has been responding to mechanical load since the first vertebrates developed mineralised tissue three hundred and fifty million years ago. The osteogenic window is not a discovery. It is a rediscovery of a biological principle your ancestors exercised every day without knowing it had a name.
The pack is the signal. The walk is the language. The skeleton is listening.
But it will not wait indefinitely for you to start speaking.