The whole game — when it comes to exercise, the main driver of muscle growth is mechanical tension, and there’s just one thing. Mechanical tension is simply a pulling force on your muscle fibers. (Other non-exercise pathways exist — anabolic steroids let you sit on a couch and put on muscle — but that’s not what we’re doing here.)

When you perform a movement, your brain sends an electrical signal to a muscle to switch it on, and it starts contracting. The fibers sense that pulling force — likely via a little structure called titin — and through mechanotransduction the mechanical signal is translated into the chemical, downstream processes that actually build muscle. That’s the entire event we’re trying to create.

So the goal of a good training approach is blunt: maximize tension while minimizing muscle damage. Which brings us to two things people thought caused growth but don’t.

Here is the whole thing, start to finish. Follow it in order — each link depends on the one before it.

Because a fiber must be activated to experience significant tension during strength training, growing all of your muscle means turning all of it on. But the fibers don’t share the workload — or the upside — equally.

The boogie monster — fatigue is a temporary, reversible reduction in exercise performance caused by a previous bout of exercise. If you normally bench 255, run a 400m sprint and then try immediately — no chance (unless you’re an extremely conditioned athlete). Rest 30 minutes and you might hit it. The performance didn’t vanish; it was temporarily blocked.

What fatigue does to growth is the important part: at a high level it gets in the way of your ability to recruit motor units. If you can’t turn fibers on, they can’t experience tension, so they can’t grow. That’s why stacking sets has steep diminishing returns — as fatigue builds, later sets carry fewer and fewer stimulating reps.

The stimulus side of this industry is well understood by now. The fatigue side isn’t, and it’s what actually dictates how much training is productive before I’m shooting myself in the foot. Managing fatigue is where advanced programming lives.

A calorie is a unit of energy, nothing more. Your weight tracks one equation: energy in minus energy out. Eat more than you burn and you're in a surplus; burn more than you eat and you're in a deficit. Intake is the simple half of that equation and the half you actually control. "Energy out" is the complicated half, and it breaks into four pieces (your TDEE, total daily energy expenditure):

Once you've found roughly where maintenance sits (a TDEE calculator gets you a ballpark; weigh in daily for a few weeks and adjust by 200 to 300 calories based on the trend), you point it at a goal:

Calories decide how much weight moves. Macros decide the quality of that movement, meaning the ratio of muscle to fat you gain or lose, and protein is the lever that matters most. It's necessary to build muscle and to hold onto it while you're cutting, it keeps you full, and it speeds recovery. Set it by goal:

That leaves the carbs-versus-fats argument everyone loves to have. Once protein and the fat floor are set, whatever calories remain can go mostly to carbs, mostly to fat, or split evenly, and the outcome barely budges. The evidence for that is stronger than most people realize, so it's worth laying out properly.

Fiber is the most underrated nutrient in the fitness space, and it isn't close. It's a carbohydrate your body can't fully break down, so it passes through largely intact and you absorb little of its energy. It comes in two forms that do different jobs. Soluble fiber (oats, apples, flax, psyllium) forms a gel that slows digestion and improves LDL cholesterol and blood sugar control. Insoluble fiber (whole wheat, vegetables, nuts, seeds) keeps you regular, feeds your gut bacteria, and appears to be the main driver behind lower colon cancer risk.

The health evidence is genuinely striking. Reynolds 2019 pooled 185 studies across more than 135 million person-years and found the highest fiber consumers had a 15 to 30% lower risk of dying from anything. Park 2011 tracked 31,000 deaths over nine years and saw a 22% drop in all-cause mortality plus a 56% drop in death from cardiovascular, respiratory, and infectious disease. For cancer, the dose-response is clean: every 10 g of added fiber lowers colorectal cancer risk by about 10%. Type 2 diabetes and inflammation move in the same direction. An umbrella review graded the mortality and CVD findings as "convincing evidence," the top tier in this field.

It's also a quietly powerful fat-loss tool, through appetite. Fiber slows how fast food leaves your stomach and stimulates GLP-1, the fullness hormone that the newer weight-loss drugs target. The practical payoff showed up in a 2001 finding: every 14 g increase in fiber cut spontaneous calorie intake by about 10%. Go from almost none to 28 g and people ate roughly 20% less without being told to. That can be the whole difference between a surplus and a deficit.

Two supplements clear the evidence bar decisively. For years I ranked creatine at the top, and it's still elite. After going through the full body of fish oil research, fish oil has passed it as my single top pick. That's an update to the ranking as the evidence came in, not a contradiction, and fish oil doesn't replace creatine; if I could only keep one, I'd keep the fish oil.

Micronutrients are the vitamins and minerals you need in small amounts, and if you eat mostly whole foods while hitting your calories and protein, they largely take care of themselves. Five deficiencies show up most often in lifters and the general population, and each has a fix on your plate before it needs a pill:

Get a vitamin and mineral panel from your doctor, fix gaps with whole foods first and targeted supplements second, and a general multivitamin is a reasonable insurance policy if your diet isn't perfect. This section stays short on purpose: there's no dedicated micronutrient deep-dive behind it yet, so I'm keeping it to what's well supported rather than padding it out.

You can’t prescribe what you can’t define — volume simply refers to how much of something, and we need to say how much of what. The obvious metric, volume load (sets × reps × weight), sounds scientific but does not correlate well with growth.

Proof: take a 100-lb bench that’s a 12-rep max. Do three sets of 4 (rack it, rest, repeat) and you get little to no growth. Do a single set of 12 to failure and you get significant growth. Same volume load, wildly different results, and the literature agrees with this consistently. Why? Because not all reps are created equal. In a 10-rep max, the first five reps do next to nothing, and the next five reps do a lot.

So researchers moved off volume load and count what actually matters: the number of hard sets, per muscle group, per week, taken close to failure. That’s the metric to use.

The Schoenfeld 2017 meta-analysis pooled 15 studies (each ≥6 weeks, comparing different training volumes, using ≥65% of one-rep max) and bucketed weekly sets per muscle group into low (1–4), medium (5–9), and high (10+). Three findings:

Why the curve bends — the reason more sets stop paying off is fatigue: it erodes motor-unit recruitment, so later sets carry fewer of the stimulating reps that actually drive growth. That mechanism gets its own deep-dive — here the takeaway is the shape of the curve. What to do with it (how many sets, how close to failure, how to arrange them) is the job of the Training Blueprint.

You have to get close — sets must be taken somewhere close to failure to get a hypertrophic outcome, because that’s where the stimulating reps live. Leave 5–6 reps in the gas tank and you walk away with almost no benefit.

But you do not have to grind to failure on every set. Roughly one rep in reserve (1 RIR) tends to be the sweet spot for managing stimulus and fatigue. Close enough to buy the stimulating reps, not so deep that fatigue wrecks your next set. The reframe I give clients: this is calculated and precise, not suffering for its own sake. You’ve already done the vast majority of the set. Give me ten more seconds of effort. Give me four more reps and you will walk away with way better results.

The definition matters — fatigue is not a feeling. It is an objective, measurable drop in performance caused by a previous bout of exercise. That distinction is not pedantic. You can have powerful fatigue mechanisms grinding away inside a muscle while a stopwatch reads zero fatigue, because a potentiation effect is quietly cancelling them out — walk into the gym motivated, recruit a few more motor units, and the local fatigue never shows up on the bar.

So the useful move is to stop treating fatigue as one blurry thing. A movement is really a chain of steps — the brain pulls a stored motor program off the shelf, cranks up the command to recruit fibers, fires the signal down the nerves, and inside the muscle an electrical signal is converted into a chemical one (calcium) that finally lets actin and myosin pull. Every fatigue mechanism is just an interference at one specific link in that chain. Name the link, and the mechanism stops being mysterious.

Ignore the textbook zoo of mechanisms. For a lifter, local fatigue sorts into two opposite camps, plus one global effect coming from the brain. Knowing which is which tells you whether a mechanism is stealing your gains or handing them to you.

If calcium accumulation runs both your growth ceiling and your recovery bill, then the whole game is controlling calcium. Exactly five variables do that — three that apply to any set, plus two that only appear when a fiber is stretched under load.

Pick up a 25 lb dumbbell and let it hang by your side — it’s little-to-no challenge to the biceps. Curl the elbow and it gets significantly harder. Unless this is a magic dumbbell, the weight didn’t change. So what did? In one word, torque.

Torque is rotational force, and you already understand it. Every joint — elbow, shoulder, hip, knee, ankle — is simply a rotational axis, and every movement you perform — even motion that looks perfectly straight — is rotation about your joints. Each time you lift, you apply torque to your joints externally, and your muscles generate torque internally in response. That’s how we fundamentally train muscles.

The equation is Torque = linear force × moment arm. Don’t do math with the numbers — care about the trends. It’s a lot of drawing. You find torque by drawing two things:

The weight supplies the demand (external torque). Your muscles are the supply (internal torque) — if this external torque is the demand, think of the internal torque your body generates as the supply. Both obey force × moment arm, but now the force is generated by the muscle and the moment arm is the perpendicular distance from its line of pull to the joint center.

Two identical lads — same size, same training. One maxes a 40 lb curl, the other a 60. The only difference: the biceps tendon inserts 4 cm from the elbow on the weak one versus 6 cm on the strong one. Those 2 cm — a 50% torque advantage, a 50% strength advantage without gaining a single gram of muscle. You can’t change where your muscle inserts; it’s the illustration that matters.

Error can enter at every stage. Walk a study from top to bottom and check each one:

Every pec fiber inserts on the same spot — your upper arm bone (the humerus). It’s the origins that fan out. That one fact drives almost everything about how you train the chest.

Here’s the payoff: the up (clavicular) and down (costal) pulls cancel each other out, so the net force of flat work is simply “across.” That’s why flat pressing and flat flys train the entire chest — you might not need as much variety as you think. Angles let you specialize, but for 99% of people it’s presses and flys, meat and potatoes.

One more fact that pays off later — a bigger ribcage pushes the pec’s line of pull further from the shoulder, giving it a longer internal moment arm: more leverage, and a chest that grows better relative to the delts. Thin-ribcage lifters get less chest leverage — which is exactly why the ribcage flare tool in the spotlight below matters.

Three complementary, non-redundant patterns cover the entire chest. Don’t stack near-identical movements (flat barbell → flat dumbbell → flat machine is the same function three times — “stepping over a dollar to pick up a penny”). And you don’t have to do all three the same day — spread the presses across your week.

All three regions share the same insertion — the deltoid tuberosity of the humerus (your upper arm bone). It’s the origins that differ, and that one fact drives everything. The deltoid is like a shell wrapping around the shoulder — the most mobile joint in the body. Fun fact: by mass, the delts are the largest muscle group in the upper body.

Because the origins differ, the fibers run in different directions and pull the upper arm in different directions. That’s also why internal and external rotation change which fibers you hit: raise the arm out to the side with a neutral or internally rotated arm and the side delt is on top of the joint, best suited to lift; go into a lot of external rotation and the front delt is now best positioned. Degrees of rotation during a movement change the target.

One more — delts are commonly thought of as three heads, but some studies identify up to seven distinct heads. A fun fact, but it doesn’t change how we train them practically: three regions, three go-to movements.

Three variables set your programming: training status, current physique, and goals. Beginners keep it simple — a chest press, an overhead press, and a row or pulldown will grow the delts just fine; add a reverse fly only if it doesn’t come at the expense of compounds. Never swap a rowing movement for a rear-delt movement — that’s stepping over a dollar to pick up a penny.

Intermediate and advanced lifters are more individual, but the broad truth is most don’t need more front-delt work — allocate more hard sets to side and rear delts. Sometimes it’s about rebalancing: a lifter who benched three times a week for years may pull the overhead press out entirely and hammer laterals and rear flies. If you’re gung-ho on growing, spread ~5 to 10 sets over the week, hit the muscle about twice a week, over at least two sessions (say 3 + 3). Start on the low end; only add volume if you stop progressing. It’s highly individual. That’s where coaching comes into play.

Three heads make the “triceps,” and all three share basically the same insertion on the ulna (your forearm bone). A muscle creates a bone-to-bone pull — contracting drags its origin and insertion closer. Two of the heads have one job; the third is a special case.

There are two ways to stimulate the triceps: pressing (multi-joint — the elbow and shoulder both move) and pure elbow extension (upper arm stays still, only the elbow moves). Which head gets emphasized comes down to where your shoulder is:

Three complementary, non-redundant movements cover all three heads — and they’re a great starting point for most people, not strict must-dos in a fixed order, so get creative. Your medial and lateral heads already pick up stimulus from your normal chest and shoulder pressing, which is exactly why — if you only prioritize one triceps movement — it should be the arm-at-side extension for the long head, the part pressing neglects.

The biceps are made of two muscles — the short head and the long head (hence “bi”). Both share the same insertion on the radial tuberosity, a little notch in your radius, so they both pull on the same forearm bone. They only differ at the origin: the long head comes up and wraps over the top of your humerus, while the short head goes straight up to the coracoid process of the scapula. Those origins sit very, very close together.

Which is exactly why the classic “curl this way for the long head, that way for the short head” advice falls apart. My verdict — for all intents and purposes, these are either going to grow together or they’re going to shrivel up into nothing together. There is no evidence you can preferentially activate or grow one head over the other.

Two other muscles contribute significantly to elbow flexion — the brachialis (which lies underneath the biceps) and the brachioradialis (the hunk of forearm that pushes up against the biceps). That’s four muscles total across the complex. Your brain chooses which of them to recruit for you, subconsciously, based on leverages and physics. Your brain is going to choose for you because it’s smarter than you. Screaming “brachioradialis” while you curl does nothing.

The strategy is to understand what the brain does automatically, then pick an exercise that takes advantage of it.

The foundational law — A muscle’s main function is to manage torque at a given joint. Take that one with you, store it somewhere in your brain and keep it for the rest of your life. Torque is rotational force: linear force times the moment arm. The longer the moment arm, the more torque for the same exact force.

These three aren’t redundant — each covers a different range and grip, and together they train the whole complex. None are strictly “must-do” (similar alternatives exist), but progressively overloading these over time is a complete elbow-flexor program.

If you want to intelligently train a muscle — you need to understand its anatomy. Muscles create a bone-to-bone pull: they attach on one bone, attach on another, and when they contract they slide over one another and pull one end toward the other. The lat is a big, fan-shaped muscle with many origins — all along the spine, along the hip, some of the lower 3–4 ribs, and for about a third of people it even attaches directly to the scapula. All of those converge to one insertion on the humerus, your upper arm bone. That’s the fact everything hinges on: the lat cares where your upper arm is — it doesn’t attach to your hand.

One more expectation-setter: the tendon is the middleman between muscle belly and bone, and everyone has different insertion points and tendon lengths. This is genetically preset and cannot be changed. You can’t grow tendons like you grow muscle fibers. No matter how thick or wide you build the belly, you can’t make the lat sit lower than its insertion point. So don’t hold your set of lat genetics up to somebody else’s and be like, “I want to look like that. How do I train my lower lats?” That’s not how it’s going to work.

Why form is king here — the triceps are the only elbow extensor, so decent form plus elbow extension grows triceps. The lats have no such luck: rear delts, teres, biceps, and traps all share their functions. Slap heavy load on a bar and just pull, and the dominant helpers take over while the lats barely work. So I’m more of a stickler for form on lat-biased movements relative to many other movements.

Two movements are great for most people; three may optimize it for someone super gung-ho on a lat-focused program. The two core picks are a lat-biased row plus a medium-wide grip pulldown — the row covers extension, the pulldown covers adduction, so together they train all the functions. Add an optional pullover (or a flap, or another pulldown) for the third slot by preference.

Sequencing — do the row first. It’s more of a mid-to-shortened-position exercise, while the pulldown is lengthened-range and generates more fatigue; leading with the shortened-position movement lets the pulldown be more effective. Reverse the order and the row suffers.

All muscles do fundamentally is create bone-to-bone pull — they attach on one bone, attach to another, and pull the two closer together. The quads are four muscles that all share one insertion point: they insert onto the kneecap via the same tendon and pull the tibia toward the femur. It’s the origins that separate them — and that one fact drives everything.

The three vasti originate along the femur, so they’re mono-articular — they cross only the knee and do one thing: knee extension. The rectus femoris originates from the pelvis, crosses both the hip and the knee, and can do both knee extension and hip flexion. That has big implications. The knee is a hinge joint like the elbow — it folds up over the thigh, and all four heads create one line of pull: knee extension, straight up. The knee cannot be pulled out to the side by muscular contraction.

The two-joint problem — in a squat you’re doing knee extension plus hip extension on the way up. The rectus femoris wants to do hip flexion — the opposite — so the CNS turns it down. Your brain is like, “Oh no, no, you’re no good for the job. Let’s turn that down.” The result: the rectus femoris gets no meaningful growth from squatting or leg pressing. The fix is to hold one joint fixed and train the other — keep the hip fixed and train just the knee. That’s the leg extension. (The hamstrings have the same problem in reverse, which is why they don’t grow well from squats either.)

Two physiology facts — the quads are a big muscle group with a large voluntary activation deficit: they’re hard for the brain to fully recruit, so stability and setup are critical and they may recover a bit faster. And they operate on the descending limb of the length-tension relationship — lots of passive tension — so they respond exceptionally well to being loaded in a stretched position. Outcome data and anecdote agree.

People butcher the hack squat all the time — usually via bad setup, short range of motion, or bouncing. The formula is simple.

Two movement patterns, done well and overloaded over time. Pick one compound and get brutally strong at it — don’t program-hop. Pair it with a leg extension to cover the shortened range no squat reaches. Choose whatever squatting pattern is working for you and feels best, and just get brutally strong at that and your quads will grow. If you do both in one session, put the leg extension first — it’s less fatiguing and warms up the knees.

Muscles create bone-to-bone pull: when fibers slide over each other they pull one bone closer to another. The hamstrings are four muscles, and the key split is how many joints each one crosses.

Three of the four heads originate from the ischial tuberosity — the “sit bone” you feel digging your hips into a chair — and cross both the hip and the knee, making them bi-articular. They do hip extension (pulling your hips from tilted to upright) and knee flexion. The biceps femoris short head is the oddball: it originates on the femur, crosses only the knee, and can only flex the knee. So the fully lengthened hamstring is a straight knee plus a flexed hip; fully shortened is a fully flexed knee plus an extended, upright hip.

The two-joint problem — in any squat or leg-press concentric you do hip extension and knee extension at the same time. The hamstrings can do the hip extension, but they also want to do knee flexion — and a muscle can only contract, it can’t fire at the hip but not the knee. So the CNS turns them way down. Squats and leg presses are not hamstring exercises — a study had subjects squat and found virtually no hamstring growth, while the quads and glutes did grow. The fix is to hold one joint fixed and train the other: knee flexion with a fixed hip (a leg curl), or hip extension with a fixed knee (a hinge).

Two physiology facts — the hamstrings operate into the descending limb of the length-tension relationship, so there’s significant passive tension and strong stretch-mediated hypertrophy — load them in a lengthened position (head-to-head, the seated curl crushes the lying curl). But they’re an outlier with very high voluntary activation: by Henneman’s size principle, driving activation up recruits the largest, most glycolytic, damage-prone type II fibers — and lengthened-range work is already more fatiguing. So manage your volume; the hamstrings may recover a bit slower. One anatomy tiebreaker: only three heads act on the hip, but all four act on the knee — so if forced to choose, lean toward knee flexion.

The hip hinge (my favorite is the stiff-legged deadlift) is the harder movement to self-manage — you can’t see your own back — so the execution rules matter even more.

Scale it to how far you want to go. The one-movement minimalist does the seated curl well. The optimal minimalist adds a hinge so you train both knee flexion and hip extension across the week. Full optimization brings in all three patterns. One tip, my two cents and not a hard rule: put hip extension on one day and the seated curl on the other, so you cover the full contractile range each day; slot the lying curl onto whichever day (or both). You can do one set of seated leg curls twice a week, and if you do them well you’ll get way better results than trying to optimize your program and not doing them correctly.

Every muscle does the same fundamental thing: it attaches on one bone, attaches on another, and pulls those two bones closer together — or resists them moving apart. For the glutes, the two bones are the pelvis and the femur (hip bone and thigh bone). One end of the glute max originates all along the pelvis; the other end inserts down by the femur.

Historically, that insertion was thought to attach largely onto the IT band — the tissue running down the leg with a knee-stabilizing role. The 2025 study overturned that: 75% of glute max fibers attach directly to the femur. That single fact lets us narrow the entire training conversation down to four hip functions: extension, abduction, adduction, and rotation — not the knee.

The glute max is a regionally complex muscle with divergent architecture — a broad spread of origins and insertions — so which fibers work depends on where your hip is. Some regions can abduct the thigh while others adduct it. The same overall muscle being able to pull in opposite directions is quite weird.

Standing straight up (anatomical position) — the glute max is largely a hip extensor. Hip extension can look like two things: driving your leg behind your body, or — if slightly hinged — pushing your hips forward to bring the torso upright. Both are hip extension. It also assists external rotation of the thigh, but it’s a poor abductor here.

Move into hip flexion — and the fibers reorient. The upper glute max fibers gain a lot of leverage to abduct, which is exactly why a seated hip abduction done leaning forward becomes a big upper-glute-max exercise. The glute medius does the opposite: it’s a strong abductor standing tall but loses leverage the more hip-flexed you get — the mirror image of the glute max’s abduction function.

The medius’ real job — stabilizing the pelvis, not just abducting the thigh. During anything on one leg — a split squat, a lunge — a force tries to tip your pelvis to one side, and the medius holds it level. This is no token “stabilizer” role: biomechanical analyses show it can have to produce forces of 2 to 3 times your body weight just during unweighted single-leg exercise. If we train all the different functions of the glutes, we can be pretty certain we’ll get good all-around development.

These four were chosen very carefully on biomechanical grounds. Three of them most directly challenge hip extension — they differ in where the movement is hardest (the resistance profile) and which non-glute muscles assist. Two are contracted / shortened-range (hip thrust + abduction), two are lengthened-range (RDL + Bulgarian split squat). If you expanded to a “Fantastic Five,” the add-on would be a glute kickback variation — but the Core Four are the first four that should make it into any glute program.

Assuming all four movements are in your routine, split the work over two training days. Two picks are shortened-range and two are lengthened-range — so on day 1 choose one of each, and on day 2 choose one of each. Within any session, run the contracted movement before the lengthened one. Keep volume tight, and never slam glute workouts back to back — give them at least two rest days so they actually have time to recover and grow. Expanding to a Fantastic Five? Add a glute kickback.

Fatigue is the reason volume has diminishing returns — it erodes motor unit recruitment, so later sets carry fewer stimulating reps. A brutal second set after 30 seconds’ rest can land at only ~50–60% of the first set’s recruitment. That second set was brutal, difficult, and extremely painful, but it wasn’t necessarily effective. Manage fatigue and you get less volume with better outcomes.

I lean low-volume for three reasons. It preserves set quality: pile on 7 exercises × 5 sets and by the fourth you’re tapped out and your execution sucks. It makes adherence skyrocket, because shorter sessions mean people actually show up, and consistency is king. And it protects joints for the long haul: they only have so many articulations under load, so if you can get 80–90% of the benefit at much lower volume, do that. I want my clients in the gym for decades, not six months. More is not better. Better is better.

You’ve likely heard that progressive overload is the key driver of muscle growth. It’s not really a driver at all. It’s an indicator that growth has likely occurred — not the cause, the effect. The actual cause is mechanical tension, specifically when operating near maximum levels of effort.

Double progression is, in my opinion, the most reliable strategy for knowing when to increase the weight, how often to do it, and actually implementing progressive overload. You progress two variables — reps or weight — inside a fixed rep range at a fixed intensity.

You just rinse and repeat over time until your muscles are so big that all of the men compliment you. Shocker. Run the whole thing for 12–16 weeks minimum; if you’re still progressing, stay on it — I’ve stuck with programs for the better part of a year.

Always start with who — and what their goals are. This plan was built for one of my training teams: mostly young, healthy males without many joint issues, whose goal is to prioritize upper-body growth. Legs still get trained — I won’t have any twig people in my training groups — but the torso is the priority for this block.

Physiology, mechanisms, research, optimality — none of it matters if people don’t actually stick to the plan. Programming for a group is a shotgun approach to get everybody to attend as much as possible. A perfect plan nobody follows is worthless. Adherence beats optimality.

Day selection is data-driven. Tracking adherence across roughly 30 people, one pattern makes attendance skyrocket: Monday, Tuesday, Thursday, Friday. For whatever reason, people like this best. It also happens to force a useful arrangement — two back-to-back days, a rest day, then two more back-to-back days.

Frequency is very important — the literature is clear on this. Hitting a body part only once per week can grow you, but it isn’t the most efficient use of your training. Split the volume across the week so you’re recovered in time to hit the muscle again. Training chest only on Mondays, like a lot of people do, is probably not the best way.

A Pull / Push / Lower / Upper arrangement hits most body parts twice per week and lets volume be spaced for recovery. Two back-to-back training days, a rest, then two more — no muscle gets hammered on consecutive days.

A great starting and ending point for most people training a muscle twice a week: about 5–10 (6–10) good, hard working sets per muscle group per week. Some can go higher; some get great results below that. In this plan, chest lands around 6 hard sets a week — roughly three on Tuesday plus two or three more on Friday.

Start low, add only if needed — there’s a really good argument to start on the low end of volume and only add more if you’re not progressing. The common mistake is to just throw sets at a wall and see what sticks. For reference: my own training runs no more than 6–8 hard working sets per muscle per week, and I’m still growing after training a long time.

Leg day runs against the common “squat first, then accessories” habit. Start with isolation / shortened-position work, then move to compound / lengthened-position work. Leg extension goes first, then a set of seated hamstring curl, then the leg press, then the RDL.

Each priority muscle is trained with movements that hit different functions or positions, so no two exercises are just doing the same set twice.

Exercise selection favors machine-based, easy-access “jump on, jump off” movements that need little setup — so seven exercises can be finished in an hour or less. The first two exercises take about half the total time (more warmup and setup); the rest breeze by because you’re already warmed up.

Don’t jump ship too early — run a plan for at least a couple/few months. Early gains come fast, then level off — that’s normal strength adaptation, not failure. Just because you’re not making 20- or 30-pound jumps doesn’t mean it’s not effective; it’s actually a really good spot to be at. Jumping ship too quick is one of the biggest mistakes people make.

My own process: start with the science and mechanisms as guardrails (not hard rules), design the program, test it on myself for 12–16 weeks tracking progressive-overload data and body-part circumferences, then roll it out to the ~30-person teams (I’m my own guinea pig first, and they’re round two), and track results with analytics software. Switch programs every 2–4 months (3–4 is the sweet spot), longer if it’s working and priorities haven’t changed. This is a first draft — I’ll revisit and adjust order based on client feedback and measured progress.

Lagging body parts come down to genetics plus training history and gym behavior. The genetics you can’t change, but the behavior is where the leverage is. After figuring out my quad and my chest training, they became my favorite muscles to train, and you might find the very same.

The mechanism behind all of it — fatigue, defined in exercise science as a temporary and reversible reduction in exercise performance caused by a previous bout of exercise. You accrue more and more of it as a workout goes on, and it gets in the way of maximum performance and maximum motor-unit recruitment on your working sets. To maximally stimulate growth, you want maximum performance, and not all sets are created equal.

On a five-exercise day, exercise #1 is far more effective at stimulating growth than #5 — accumulated fatigue lowers performance and motor-unit recruitment on the later sets. So the fix is simple: put lagging body parts earlier in the workout, when you’re fresh.

My own example — on a push day with chest, shoulders, quads, and triceps, my chest and quads lagged behind my shoulders and triceps. So I’d put the chest and quad work toward the beginning, and the shoulder and triceps work toward the end.

Muscle growth happens during recovery, after the gym — so how you space the days matters. Never train the same muscle twice per week on back-to-back days; spread them out. With twice per week, space them relatively evenly — e.g. Monday & Thursday or Tuesday & Friday. Give at least 48–72 hours between sessions for that muscle. More volume per day means more recovery time needed; less volume, less time.

Redundancy = using multiple exercises that train the same exact function of the muscle — same range of motion and same resistance profile — which is effectively just doing more sets of the same exercise. Avoid it. Choose complementary exercises that challenge the muscle differently.

Let progressive overload be your North Star. Rested, properly warmed-up, intensely pushed sets mean you probably don’t need as much as you think. A lower-volume, high-quality, high-intensity approach beats throwing sets at the wall and trying to see what sticks. If you’re consistently getting stronger — adding weight or reps — keep the volume where it is and grind. Add more only if progress stalls.

The more advanced you are as a lifter, the less volume you need in order to continue growing. It sounds backwards, so I’ll make the case two ways: one is common-sense meathead wisdom, and number two is deep physiology and biochemistry.

Meathead wisdom — advanced trainees are more efficient at using, turning on, and targeting muscle fibers than beginners. So if they more efficiently produce large tension on the muscle, why on earth would they need more sets than someone still carrying all those handicaps? It makes no sense.

You have roughly a 50/50 split of type I and type II fibers (with some nuance). They behave very differently, and that difference is the whole story.

Fatigue is a temporary and reversible reduction in exercise performance caused by a previous bout of exercise. It’s an outcome — a measurable drop in performance — with many possible underlying mechanisms. What it fundamentally does: it prevents you from recruiting and turning on as many of your muscle fibers as you can. And it de-recruits from the top down — the fibers you lose access to under fatigue are the largest type II, the mirror image of recruitment order: largest turn off first.

A bench-press example — Set 1: 225 lb for 12 reps, barely getting the 12th at failure. Keep piling on sets and by set 6 you’re lucky to get 3 or 4 reps. Did your muscles shrink or atrophy between set 1 and set 6? No — fatiguing mechanisms are at play, and you fundamentally can’t access as many fibers to assist. The ones you lost are the largest type II.

Why do the big type II fibers fatigue and get damaged so much more easily? It comes down to how they handle calcium ions.

You can design the perfect theoretical plan for someone, but if they don’t follow it, it’s the worst plan for them. The technical side of fitness isn’t the most valuable skill — the most valuable skill is designing life systems so people make the right decisions automatically. Success in any goal comes down to consistently making the right decisions — everything you’ve ever achieved was a result of the decisions leading up to that point.

You’re in the gym only about one hour, four to five times a week — all the other hours are spent making food decisions, which is why nutrition dominates. Architect your environment so good choices are the path of least resistance and bad choices carry friction.

Everyone preaches “everything in moderation,” and yes, you can fit nearly any food into a well-rounded diet. But if moderation isn’t your strong suit — you can’t stop at two Oreos — it may be better to eliminate a food entirely, at least for a period. There’s a time and a place for balance and moderation, and there’s also a time and a place for brute force.

Expand the time frame — people judge balance over a single day (good breakfast and lunch, so there’s time for a naughty dinner) or a week (six clean days, so cheat day seven). That’s short-sighted. Better: commit fully to a yearlong phase that’s unbalanced by most people’s standards. You’ll emerge after that year a different person, with lifelong systems. Then return to balance. It takes so much more effort to reach a fitness goal than to maintain it. It’s just like a plane taking off: all the energy goes into getting the bird off the ground, and once you’re cruising it takes far less to stay there.

The behavior half gets you to the gym. The science half decides whether that hour builds muscle — and here the most common trap is chasing the wrong signal. Barry’s Bootcamp, F45, Orange Theory: they’re objectively difficult, but difficulty isn’t the point. They’re the wrong tool for the job. You know what else is brutal? Trying to drill a hole with a screwdriver. Bad tool for the job.

There is one driver of muscle growth: mechanical tension, a pulling force on the muscle fiber. A fiber only experiences real tension when the brain turns it on (motor-unit recruitment). Maximal growth needs two things at once: very high recruitment and slow contraction velocities, which happens only when you’re close to failure. And fatigue is the enemy of exactly that: it lowers recruitment, so when you’re excessively fatigued you can’t turn on those muscle fibers no matter how hard you try.

Even setting fatigue aside, effective growth needs three things these classes lack — and they double as the blueprint for what to do instead.

Maximum strength comes from six real adaptations — three central-nervous-system (CNS) adaptations that change the signal, and three peripheral adaptations that change the tissue itself. Everything else you see in textbooks is minor. Here’s the whole map before I open each one up.

After these six there are four more mechanisms you’ll find in textbooks and online that just aren’t important in the context of strength training. The point isn’t to memorize the six. It’s that each one gets its own adaptation → stimulus → implication pass, and the framework is what keeps them from blurring together.

Now the tissue itself. These change what a given signal is actually able to do.

The textbook story is that beginner strength gains are “neural first, hypertrophy later” — as if the nervous-system piece mainly means better motor unit recruitment. The better read is narrower: beginner strength gains are part force capacity, part task skill. If the test is the exact machine someone practiced, the skill piece gets counted as “strength.”

Each adaptation arrives on its own clock — and knowing the clock changes how you read a beginner’s progress. Fast adaptations can mask slow ones; a complex lift can delay recruitment gains for weeks.

These four fill textbooks and social feeds. My gripe: people cite them because they don’t know the history of exercise science — they’re reading decades-old textbooks instead of recent reviews. Here’s why each is a dead end for maximum strength.

Peer-reviewed sources for the concepts above — supporting research, not necessarily the exact studies behind each concept above:

Before the mechanism — the word “passive” gets used two completely different ways, often in the same paragraph, and it breaks people. You have to keep the two meanings apart in your head.

The trap: you can have passive mechanical tension inside a fully activated muscle fiber. The two meanings collide, and that collision is the whole reason this topic confuses everyone. Exercise science has a genuinely horrible terminology problem — we use these words in all kinds of sloppy ways, nobody is going to fix it now, and we're stuck with it. So when this trips you up, understand the fault is the vocabulary, not you.

Titin is the molecule responsible for producing most of the passive tension inside a non-activated fiber. I take this from one of the most famous experiments in the field: suspend a muscle fiber in an apparatus to hold it tight, then strip out its components one at a time and watch when the tension changes. Pull out the endomysium and collagen — the sheath around the fiber — and tension barely changes, which surprises almost everyone. Remove titin from the middle and the fiber disintegrates instantly, which tells you titin was holding the whole thing together. (You can dig up studies giving the endomysium a role in certain situations, but the majority of passive tension comes from titin.)

Titin runs in three segments. The technical names are opaque, so here's how I picture it — a rope tied to a stiff elastic band by a connector:

This structure explains the shape of the passive length-tension curve. Put a relaxed person's limb in a dynamometer and wind it into maximum range of motion until they squeak at the edge of discomfort. The machine plots joint torque against joint angle: torque sits at roughly zero for most of the range, then jumps up sharply at end-range. That's not muscle activation — the subject is fully relaxed with no stretch reflex. It's purely the passive tissue stretch, called passive resistive torque. Stretch an isolated single fiber and measure force vs. length and you get the exact same graph shape — the length-tension relationship matches the joint-angle-torque relationship. Why that shape? Pull the rope-plus-band assembly: first the compliance segment unravels (flat, near-zero force), then once it hits its maximum length only the stiff segment can stretch further — and force kicks upward. The model reproduces the real curve exactly, which is how I know we're dealing with the right structure.

The instant you activate the fiber, the bridge piece locks onto the nearest actin. That's the move that makes everything click. Now the rope is out of the picture — the only way left to stretch the fiber is to pull the stiff segment directly. So the entire curve shifts: the upward kick in passive force that used to appear only at end-range now happens at a much shorter muscle length.

This is why the overall length-tension relationship is really two curves amalgamated — an active one and a passive one — and why an isometric contraction and an eccentric contraction don't share the same shape. Textbooks rarely draw it; they just say it's different. You can see it live in a dynamometer: wind the arm back, tell the subject to resist at some point, and the passive-force kick appears immediately from that point onward — far earlier in the range than the end-ROM you'd need if the fiber were relaxed.

And that is why strength training produces passive tension — and a passive-tension adaptation — even though gym exercises never get anywhere near maximum range of motion. Sit relaxed on a bench with your knee at 90° and there's no passive tension in your lower-body muscles. Do a squat at that exact same joint angle and passive tension is present — purely because the fibers are activated, which latched titin onto actin and dragged the whole curve over. That paradox clicking into place was one of the moments that genuinely reframed how I think about training.

How much force does passive tension actually add? String a single fiber in an apparatus, stimulate it electrically, record its isometric force. Then — while it's still activated — stretch it. Force jumps upward dramatically, because you've stacked passive (eccentric) tension on top of the active cross-bridge tension. The ratio of eccentric to isometric force tells you how much extra passive force exists relative to the active force. In human single fibers that ratio is ~1.8–1.9 — meaning the passive titin-based force is almost identical to the active actin-myosin force. Call it a 1:1 ratio. It's fairly stable across fibers because every myofibril carries actins, myosins and titins together.

Now apply that 1:1 ratio to a real set. Take the last, maximal rep of an 8RM biceps curl. In a small, trained person the biceps can reach very close to 100% motor-unit recruitment (call it 98–99%). The moment you start lowering the weight, perceived effort tanks — and a big drop in effort signals a big drop in recruitment. Here's the arithmetic: if 100% activation on the way up used only active tension, then on the way down — where you have double the fiber force available (active + roughly-equal passive) — you only need 50% activation to control the same weight. Recruitment falls from 100% → 50% across the transition. EMG backs this up: concentric phases show roughly double the activation of eccentric phases at the same load.

Where it shows up, and when it stops — in humans, sarcomerogenesis reads out as fascicle-length increase, and it appears after (a) eccentric training, (b) strength training in stretched positions, and (c) static stretching. Despite all the complaints you'll read online, human distraction case studies measured sarcomerogenesis directly decades ago, and modern eccentric-training studies show fascicle-length increases that correspond closely to it — length increases really are a good readout. In a fixed-range-of-motion strength program the gain runs for about 3, 4 or 5 months (maybe a bit longer) and then plateaus: as sarcomeres are added in series, the passive tension at each joint angle keeps dropping, until the standardized ROM simply can't generate enough tension to trigger any more (~3–6 months). Static stretching is the exception — as long as flexibility keeps improving you keep reaching longer muscle lengths to compensate, so it could in principle continue almost indefinitely, into full Jean-Claude Van Damme territory. Nobody has run a stretch study long enough to confirm that.

A sanity check — I lean on a sense-check heuristic: does the adaptation you're triggering actually help you cope with the stimulus you just experienced? A stretch stimulus → more sarcomeres in series → yes, that helps you reach longer, more flexible positions. Adding myofibrils in parallel would do nothing for flexibility. I don't treat it as a perfect way to make predictions — it's a very good way to check that a hypothesis actually makes sense.

Muscles differ enormously in how susceptible they are to stretch-mediated hypertrophy, because in some muscles the sarcomeres stretch a lot for a given joint movement and in others they barely stretch at all. Three architectural factors decide it:

The biceps vs. triceps contrast makes it concrete. The biceps brachii has a big moment arm and fusiform fibers; going from contracted to fully straight moves an enormous distance, so its sarcomeres stretch across roughly double the excursion of the triceps for the same joint movement — it's highly susceptible. The triceps brachii is the archetype of a resistant muscle: it's wrapped around the bone with a small moment arm, has many sarcomeres in series, and barely moves — its operating sarcomere length stays tiny throughout the range, so there's almost no stretch to trigger anything. (On pennation: a fusiform muscle's angle can technically be ~zero, but in reality it bulges out away from the bone, so the fiber curves and tilts and stops behaving like a perfectly straight line — the effect is smaller than in a truly pennate muscle, but it exists.) And a gripe I keep coming back to applies throughout: muscles are non-uniform, so where you take the measurement changes the number completely — a single-point muscle measurement often tells you nothing at all.

The active length-tension relationship has named regions — the plateau (all myosin heads lined up with actin, maximal tension), the ascending limb (sarcomere shortens, myosins drop off one end toward zero force), and the descending limb (sarcomere lengthens, myosins drop off the other end). The word “limbs” has nothing to do with arms and legs — they're just names for points on the graph. On the descending (long) side, passive titin tension compensates so total force never falls to zero; on the short side there's no such compensation, which is why you get active insufficiency at the contracted end.

A muscle adds or subtracts sarcomeres in series so its plateau lines up with the joint angle where it's most often activated — visible as a shift in the angle of peak torque. The glutes are most active in extended hip, so their plateau sits right there and sarcomeres are shed until it sits perfectly where you want it. The gastrocnemius is really only activated in its most stretched position, giving it a huge long ascending limb and the ability to reach active insufficiency when contracted. The triceps — the sole elbow extensor, active across the whole range — has architecture weird enough that it sits on the plateau for essentially the entire range of motion.

Two consequences worth carrying into the gym. First, vastus medialis vs. lateralis: normally quads grow by their size (lateralis is biggest, so it grows most), but the medialis has a longer descending limb, so in stretched positions it soaks up more passive tension and the pattern reverses — stretch-position quad work biases growth toward the medialis. Second, a curve that shifts toward longer lengths can hide strength gains at short angles. In a landmark old study untrained people trained the leg press and their leg-press strength rose ~40–50% — but their knee-extension strength at a 30° knee angle didn't increase at all. Later modeling showed the angle of peak torque had shifted toward longer muscle lengths; as the curve tilted and moved up, strength at the short 30° angle stayed put. The shift negated the gain at that specific angle.

Athletes doing maximal-effort eccentrics — Nordic curls, for instance — post enormous strength gains in the trained exercise. In a dynamometer you see crazy gains in maximal eccentric strength and much smaller gains in isometric and concentric modes: a contraction-mode-specific strength gain bigger than just about anything else in sports science. And almost all of it is neural — improved coordination and improved motor-unit recruitment, not new muscle.

That should be impossible: central motor command is generated in the brain and isn't specific to what you do locally, so how can recruitment gains be mode-specific? The resolution is that, at the start, the coordination demand of the eccentric is astronomically higher than its concentric equivalent. Two lines of evidence: on fMRI, a concentric contraction shows just a couple of bright spots while the eccentric lights the brain up like a Christmas tree; and at matched sub-maximal activation, effort should be identical if coordination demand were equal — but the eccentric feels far harder. That high coordination demand drives high effort perception, which suppresses recruitment in the eccentric mode — an extra voluntary activation deficit on top of the concentric one. Training then fixes the coordination demand and simultaneously closes the recruitment deficit. So it was never a gain in central motor command — it's one coordination improvement producing two effects.

Because both adaptations are exercise-specific, the impressive gains are not transferable to sport — which flips the usual programming logic on its head:

Teams historically used block-based eccentric programming for no real reason — the move I favor is the smallest possible dose kept in continuously. One more payoff: once recruitment is high enough to actually train the faster-twitch fibers, eccentrics reduce muscle-strain injury risk through three adaptations — sarcomerogenesis (fibers can be stretched to a longer length before hitting trouble), more lateral force transmission (extra costameres), and more titin (the molecules proliferate, adding passive force, so you aren't pulled to dangerous lengths as easily). For scale on the stakes: rodent muscle tears at roughly 150–200% of resting length, and stripping lateral force transmission makes it tear much shorter.

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies referenced above:

Note: this is an animal, eccentric-biased (downhill-running) model of serial-sarcomere addition — a supporting demonstration that eccentric loading drives sarcomerogenesis, not the exact human study behind the concepts above.

The definition is load-bearing — I treat fatigue as the objective external measurement: you do a bout of work, and I measure a reduction in your exercise performance as a result. That reduction is the fatigue. This isn’t pedantry to me. Fatigue mechanisms are the internal processes; measured fatigue is the external result — and the two can come apart completely.

A potentiation effect can hide a mechanism entirely. You do set one, then someone walks into the gym and fires you up — set two recruits more motor units, cancels the local fatigue, and my stopwatch reads zero fatigue even though recruitment is genuinely higher (with its own adaptive consequences). Run it the other way and you get the opposite: motor-unit recruitment drops, but a local potentiation effect bumps up muscle-fibre force, and again you land at zero measured fatigue — this time hiding real local muscular fatigue underneath. Identical “zero fatigue,” two totally different adaptation outputs. So everything below is about mechanisms, always tied back to that external measurement. The way I think about it: all fatigue is ever doing is causing problems or interferences with your normal way of doing a movement.

My scaffold here is horribly oversimplified, but it does a great job. A movement is three steps in the central nervous system followed by four phases inside the muscle. Learn the seven and every fatigue mechanism gets an address — it’s simply a disruption at one of these steps. Where the classic movement chain becomes a fatigue chain, I run each mechanism through the same three questions from my adaptation framework: what is the mechanism, which adaptation’s stimulus does it hit (and how), and how do I exacerbate or ameliorate it.

A precision note I insist on: the literature’s term “excitation–contraction coupling” is imprecise. We’re not coupling excitation with a contraction — we’re coupling it with a chemical signal at an electrochemical junction. Contraction is two steps further down the chain (calcium transmission through the cytoplasm, then the cross-bridge cycle itself). Exercise science is genuinely bad at terminology, and this is a perfect example of it.

“Central nervous system fatigue” is shorthand for three different things. When someone throws that phrase around, I make them clarify which they mean — because only one of the three is truly global; the other two are specific to the exact exercise and muscles being used.

There are three local calcium-related sites — the cell membrane, the electrochemical junction, and the cytoplasm — and three calcium-driven damage mechanisms that live among them. All of them do the same thing: stop the calcium signal, kill the cross-bridge, end the tension, end the growth for that fibre.

Five levers drive the circuit breaker — three on any set (proximity to failure, rep range, volume) plus two that only appear under stretch. Stretching a fibre opens stretch-activated ion channels — their own channel system, running from the cytoplasm to the membrane parallel to the T-tubules — letting extracellular calcium in on top of the store-operated pool. The stretch-side levers are stretch position (resistance profile or range of motion) and eccentric intensity. Static stretch alone starts reducing mechanical tension after about 45–60 seconds; eccentric contractions pile stretch-channel calcium on top of everything else, so the mechanism develops far faster than in a concentric.

Two applications fall straight out of this. First, eccentric hypertrophy plateaus fast — after roughly 3 sets of maximum-effort eccentrics the volume dose-response flattens; don’t over-volume them. Second, exercise order: put stretch-position exercises later, not earlier. Do a stretchy lat pulldown before a row and you’ve switched off a whole bunch of muscle fibres the row wanted; the reverse order avoids the problem. As for when the breaker trips within a set — this is my honest ballpark, not backed by direct data — it’s the last rep on a heavy 5RM (short set, less time for calcium to build), the last ~3 reps on a light 15RM, and you see far more of it in low-force isometrics than high-force ones.

Metabolite relativity is two players — phosphates and acidosis — both working by building up on one side of a concentration gradient. Neither stops a cross-bridge; they freeze the cycle in place, preserving tension while dramatically slowing shortening speed. That’s a disaster for athletic speed but perfectly fine — even useful — for a bodybuilder.

The pump is osmotic, not just pooling — squeezing the muscle shuts the veins, so during the concentric nothing leaves; on the relaxation phase the channels open and blood plus metabolites exit (which is exactly why metabolite fatigue clears fast). But while blood is trapped and acidosis builds, it creates an osmotic gradient that pulls water out of the blood, across the cell membrane, and into the fibres to equalize. The fibres swell — and now the liquid has to be re-extracted from inside the cells, which takes time.

That’s the tell that a pump is more than blood pooling: if it were only blood sitting in arteries and capillaries, it would squirt out the instant you relaxed and be gone. Instead, water is inside the fibres and has to be drawn back out — so the pump persists for about 15 minutes before it’s completely gone. (Vascularization, by the way, is a pull process, not a push one: capillarization tracks fibre hypertrophy almost perfectly — the fibre determines what happens to it externally, not the reverse. You can’t grow capillaries to feed growth.)

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies referenced above:

There are exactly five fatigue mechanisms that can follow you out of the gym. Two are carried straight over from the workout itself; three are born after you finish. What makes the whole thing tractable — and this is the part most people never put together — is that all five trace back to one source.

There’s also a conditional sixth I’ll flag: coordination-related fatigue, which only shows up if you then go and perform a specific exercise using those same muscle groups. A revealing special case: if you train a totally different muscle group later, the only mechanism still reaching you is the supraspinal (CNS) one, because it’s the only one that circulates through the bloodstream.

A note on measurement — those three calcium mechanisms (desensitization, membrane damage, junction damage) get read in vivo as a group through a single proxy: low-frequency fatigue. A lot of people assume it’s measuring EC-coupling failure specifically. I don’t think that’s valid — it flags the calcium family as a whole, not one individual member of it. From a practical standpoint it doesn’t matter, because every calcium-related mechanism does the same job: it stops cross-bridge formation and drops mechanical tension. Whether you’ve got all three running or just one, the outcome is identical. It only becomes a problem if you try to stimulate hypertrophy before it recovers.

For decades the story was that eccentric “tearing forces” physically rip the myofibrils apart. I think that’s wrong, and here’s the competing model I run with: damage is biochemical — a continuation of the same calcium cascade, where calpain keeps degrading the myofibrils (and the intermediate-filament system, and the sarcomeric discs — big structures, really) after the workout is over. Four separate bodies of literature settle it for me.

The clean-up crew — once calcium has stayed high for a couple of hours and calpain has done its work, the protease phase hands off to inflammation. First the proteases nibble away at the edges of the fibrils, making them slightly less functional. Then neutrophils arrive and signal for macrophages — I picture them like Pac-Man, coming in and attacking anything that’s been damaged. The macrophages remove the damaged myofibrils wholesale, which sharply cuts the fiber’s force-producing capacity. That’s the timing of the second dip.

This is also why anti-inflammatories are a double-edged sword. Ice or a pharmaceutical can stop the secondary damage and speed strength recovery — but in a training context the likely cost is blunted hypertrophy: you leave damaged myofibrils in place, and they don’t get stimulated as effectively in subsequent workouts. Faster recovery, less growth. That’s the trade I want people to understand before they reach for the ice.

Where the tiredness comes from — repairing muscle damage produces interleukins inside the muscle, some of them identical to the ones lactate produces during the workout. They leak into the bloodstream, get renamed cytokines, circulate, and the brain detects them as feelings of tiredness — supraspinal CNS fatigue. The key part: they stay there as long as the damage is being repaired, so the CNS fatigue matches the function loss almost perfectly. A workout needing 2–3 days to recover gives you 2–3 days of CNS fatigue; one needing 10 days gives you 9–10 days of it.

Two literatures look like they contradict each other on whether fatigue stacks from workout to workout. Reconciling them is one of my favorite examples, because it tells you exactly why high-volume blocks stop working.

The overreaching plateau — keep piling on volume and fatigue drives strength down to about a 50% loss and then just holds there. At that floor, the only fibers still producing any stimulus are the ~20% slow-twitch fibers plus the low-threshold motor-unit fibers at the halfway point. This is non-functional overreaching: a high-volume block where you make no gains in the outcome you’re actually chasing, because the fibers you want to grow can’t be switched on or loaded. You get maybe 1–2 weeks of accumulating fatigue while still stimulating some growth — after that, nothing is getting into the fibers you care about.

This is exactly why I’ve been so vocal against a meta-analysis claiming strength plateaus at high volumes while hypertrophy keeps rising. I think it’s impossible: you can’t invoke cumulative fatigue to explain the strength plateau and then turn around and claim hypertrophy climbs in that same cumulative-fatigue state. If fatigue is huge enough to suppress strength, you can’t trigger the hypertrophy adaptation there either. The likely explanation for the “extra hypertrophy” is muscle swelling — two facts nobody disputes: more damaging exercise causes more swelling, and we can’t tell swelling apart from hypertrophy.

Recovery rules of thumb (moderate loads to failure, well-supported by data): you recover from 3 sets in ~48 hours and 5 sets in ~72 hours; beyond that you’re into once-a-week territory and it stops mattering much. Change any variable — all eccentric, all 1-RIR, all heavier — and the number of studies drops to one or two: I know the direction, but I can’t give you the magnitude. Individuals also vary enormously: someone with high voluntary activation and ~80% fast-twitch gets knocked sideways by every workout, while someone with low activation and ~80% slow-twitch can walk through anything you throw at them and come out the other side barely noticing.

Glycogen has no fatigue mechanism of its own. What it does is turbocharge two mechanisms you already have — and understanding why kills one of the most persistent myths in the field.

The real mechanism — the brain can detect rapid drops in glucose, in muscle glycogen, or even in carbohydrate still sitting in the digestive system, and it manufactures a sensation of fatigue and tiredness in response. My favorite proof is carbohydrate mouth-rinsing: feel wrecked, swill glucose around your mouth and spit it out, and almost instantly the brain registers that more fuel is on the way — dialing back the tiredness and giving you access to a few more higher-threshold motor units. Nothing was absorbed. It’s signaling, not fuel.

Both of these are mechanisms we really don’t want, so walking into a workout glycogen-depleted is probably a bad idea.

Replenishment meets damage — textbooks say glycogen refills within 24–36 hours after aerobic exercise if you eat sensibly — almost certainly true, but not with muscle damage. You can’t get glucose across the cell membrane of a damaged fiber. After eccentric workouts, damaged fibers can still show glycogen depletion 12 days later. That creates a vicious loop in overreaching: you can’t replenish glycogen in already-damaged fibers, every new workout damages more fibers, and large chunks of muscle end up unable to refill — which then turbocharges both supraspinal fatigue and EC-coupling failure, worsening the overreaching.

A question I get constantly: if the popular explanation for creatine leans on the debunked energy-crisis idea, what’s the real mechanism? Here’s how I understand it — with the honest caveat that nutrition isn’t my primary lane, so weigh it accordingly:

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies I reference above:

My approach to exercise selection never changes. I don’t start with the exercise. I start with the adaptation I’m chasing. Once I know the adaptation, I already know its stimulus — that work is done. So the only real question left is the third one: do any features of an exercise change how big that stimulus is? Answer that, and you’ve answered exercise selection.

The adaptation — expand the amount of motor-unit recruitment you can generate, i.e. the number of motor units you can actually get access to. The stimulus — you have to hit a genuinely high level of recruitment in that muscle, in the exercise itself. A big central motor command has to reach it. Recruitment climbs in an orderly way — smallest motor units first, largest last, the size principle running underneath everything — and the CNS steers that command toward the units best placed to do the job. So the audit question becomes: what features of an exercise change the size of the command arriving at the target muscle?

I picture central motor command like a single spotlight of fixed brightness. Flood it across a whole stage — a big compound movement with a lot of muscle involved — and no single spot is bright. Narrow the beam onto one performer — a small, isolated movement — and that spot is blazing. That’s the entire game with recruitment. Three features move that beam:

The muscle-mass lever is the strongest and the most counter-intuitive, so hold onto it. Line up a single movement pattern by how much mass it involves and you get a clean spectrum, largest to smallest: squats → leg presses → two-leg knee extensions → single-leg knee extensions. Recruitment per quad is highest at the far right, on the least impressive-looking exercise in the list.

The cleanest proof this is real is the bilateral force deficit: each limb recruits more when it works on its own than when both limbs work together. Split the effort across two legs and neither leg gets the full beam. So whenever the goal is to maximise recruitment for this adaptation, the advice is almost the opposite of what a gym would tell you — keep coordination low, keep stability high, and dramatically reduce the amount of muscle mass in the exercise. A single-leg knee extension can out-recruit a squat, per muscle.

Recruitment tells you how much. Two physiological rules tell you where. Get these two straight and you stop guessing which exercise builds which muscle.

Priority — Rule B outranks Rule A, but it runs on Rule A. Matching supplies the leverage information; the two-joint rule uses that information to decide whether a bi-articular muscle gets switched on or benched.

The study that proves it — In the early 2000s a study sat people in a dynamometer and measured their maximal hip-extension torque and maximal knee-extension torque (strength here just means joint torque, which is leverage × muscle force). Then it set a target of 50% of maximum torque and recorded muscle activation three ways: an isolated 50%-max hip extension, an isolated 50%-max knee extension, and a simultaneous contraction doing both at 50% at once — mechanically almost identical to the two isolated efforts added together.

If neuromechanical matching were the whole story, stacking the two isolated activation maps should reproduce the simultaneous one exactly. It didn’t — the maps came out completely different. In the combined contraction the two-joint muscles dropped out: hamstring activation fell away, rectus femoris fell away, while single-joint quadriceps activation climbed. The adductor magnus wasn’t measured, but glute activation didn’t move at all — so the extra hip-extension torque had to be produced by something, and the only candidate left is the adductor magnus stepping in. Modelling work on the squat reached the same verdict from the other direction: at the loads we actually squat, the hamstrings can’t be meaningfully recruited — if they fired hard they’d bend the knee back under you and you’d never stand the weight up.

So what actually decides which muscle has the best leverage in a given position? Five features. Three apply to every muscle; two extra apply only when you’re dealing with a two-joint muscle. Learn them as dials — and know that they stack.

Stacking is the whole point — you combine features to pin a stimulus onto one muscle. Two of my own worked examples:

You can’t talk exercise selection without EMG, because it’s the number everyone waves around. EMG measures the electrical noise at the muscle’s surface. People read a bigger EMG number as “more activation” and therefore “more growth.” It is neither.

The finding that should end that habit: when resistance exercise is taken to task failure, muscle-fibre activation is the same regardless of the load or how fast you move. Light and heavy, slow and fast — all reach the same fibre activation at failure. So EMG-amplitude differences between those conditions aren’t reporting activation, and they’re certainly not reporting growth.

Where the “EMG is invalid” noise came from — people used the tool for a job it was never built for: comparing different states of fatigue. Take a light-load set to failure and a heavy-load set to failure. Both hit failure, but they arrive with very different amounts of fatigue in the muscle, and fatigue changes the signal on the membrane. Different EMG, identical full activation — the amplitude gap is a fatigue artefact, nothing to do with the growth stimulus. Reaching failure in two situations does not mean the two involved equal fatigue.

Where EMG is genuinely useful — in a minimally fatigued muscle, to compare the relative involvement of different muscles in a given exercise. That’s real value — but only if you control two things.

Bottom line — EMG amplitude is a comparison tool for relative involvement in a fresh muscle. It is not a growth meter, and it is never a fibre-activation meter. Treat it that way and it’s useful; treat it as a scoreboard for “which exercise activates more” and it’ll lie to you every time.

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies referenced above:

Around 2011–2012 the field settled into a tidy three-part model: hypertrophy is driven by mechanical tension, muscle damage, and metabolic stress working together. It’s clean, it’s easy to teach, and it’s on the first slide of basically every course. My problem with it is simple — it treats three ideas with wildly different track records as if they were equals. So before I tell you which one I run with, let me lay out all three exactly as the field presents them.

Two of these three are ghosts. They keep getting cited, they keep showing up on the slide, but the mechanism underneath them is gone. My job here is to show you the autopsy on both — and why the one that’s left standing was very nearly lost by accident.

You can’t evaluate these ideas without knowing how each one was born, because in every case the birth explains the flaw. Start at the single-fiber level. Hypertrophy, at its root, is the addition of myofibrils — the contractile strands make up about 90% of the fiber. You add myofibrils two ways: an existing one splits into two smaller strands that then get topped up with fresh actin and myosin, or a brand-new one is built from scratch (de novo myofibrillogenesis). For roughly 40–50 years splitting was assumed to be the only route, and I still think it’s the more likely one — the de novo evidence is recent and I’m watching where it lands.

There are actually two different damage hypotheses running around, both traced to that same 1970s coincidence, and it’s worth separating them. The public-domain version is the gym-poster line: muscles get broken down and built back bigger. That’s the original claim, and it fell over almost immediately because, mechanically, it just doesn’t work. The scientists’ version is quieter and more defensible: damage isn’t what causes growth, but it might be a possible stimulus for it. That softer version is the one that got folded into the three-part model. Both go back to the same origin; they just backpedaled to different distances.

The field made its real mistake in method: when people wanted to defend the damage idea, they went looking for justifications — reasons it might be true — instead of trying to invalidate it. That’s backwards. And the justifications they found were thin. There were four:

Now flip it around — stop justifying and start trying to break it. This is where it dies. Go to the contusion and laceration models — animal work where you deliberately bruise or cut a muscle. You produce enormous damage. And you get mTOR signaling and elevated myofibrillar protein synthesis rates — all the biochemical fingerprints that look exactly like the run-up to hypertrophy. But you get zero muscle growth. None. The markers show up and the muscle doesn’t get bigger. Damage markers are not hypertrophy; they’re a similar type of process that turns out to be genuinely distinct.

Why? Because the pathway is identical to exercise, and that’s the whole point. In a workout, calcium enters the fiber through the sarcoplasmic reticulum or through stretch-activated channels. In a contusion or laceration, you breach the cell membrane and calcium pours in through the hole. From that moment on, everything downstream is the same cascade — same calcium, same signaling, same markers. There is no difference after the calcium is in. Which tells you the decisive thing: muscle damage is a biochemical process, not a mechanical one. It isn’t tearing forces ripping fibers apart. It’s calcium chemistry. And if bruising a muscle produces every damage marker with no growth, then damage markers can’t be your evidence for growth. The hypothesis is done.

Where it stands now — the strongest nail in the coffin came from tracking damage and growth across a whole program: at the start of a program damage is high and hypertrophy is low; by the end hypertrophy is high and damage is low. They move in opposite directions. That put a lid on it for a while. Two pillars — eccentrics and stretched positions — have been quietly dropped because passive tension explains them. Two — inflammation and satellite cells — still get cited. And I’ll make a prediction: someone new to the field will resurrect “damage causes growth” in five to ten years, because this idea behaves like a weed. You pull it up, and a season later it’s pushing back through the same crack. Every field does this — a new cohort would rather re-run the old experiment than sit down and read what the existing data already says.

Metabolic stress deserves more respect than damage, and I’ll tell you exactly why. Its birth, in 1991–1992, is the honest exception in this whole field. Two studies did something nobody had done before and — as far as I can tell — nobody has really done since: they explicitly stated a hypertrophy model, then tested it. The model they started with was the whole-muscle version of mechanical tension — the idea that the force–time integral of the muscle during training should predict how much it grows. That’s the same model being floated as fresh today. These people had it thirty years ago and actually put it on the stand.

Then it got messy — a pile of unrelated ideas got swept under the metabolic-stress umbrella after the fact, and none of them hold. A single early-2000s study suggested light loads grow slightly more slow-twitch fiber and heavy loads slightly more fast-twitch — later work found no such difference, so it was just an unlucky one-off observation. Sarcoplasmic hypertrophy was floated as load-dependent, then shown not to differ between heavy and light. The bodybuilder anecdote — that lifters started using lighter loads and shorter rest to chase the “pump” — ignores that humans lifted heavy objects to get bigger since the ancient Greeks; the light-load, short-rest style only appeared after anabolic steroids arrived in the 1960s–70s. Don’t assume the modern crowd out-thought the Greeks. And the worst one — the blood-flow-restriction claim that metabolites directly increase motor-unit recruitment. Recruitment is under voluntary control. You cannot dial it up with a chemical; the idea creates circular nonsense, and yet it sailed through peer review.

Under all that clutter there are only four mechanisms actually worth testing — the ones with a real proposed mechanism. Here’s each one on the table.

The BFR knockout — the cleanest long-term evidence against both hypoxia and swelling comes from blood-flow-restriction studies that trained sets to failure at different cuff pressures. Crank the cuff and you fail sooner with more hypoxia; ease it and you fail later with less. Long-term result: the actual occlusion pressure of the cuff has zero effect on hypertrophy. Absolutely zero. As long as your load is at least ~30% of 1RM — enough to occlude the vessels in the concentric phase anyway — you get the normal stimulating-rep result and the added hypoxia buys you nothing. That’s the test failing in the field, not just the tube.

And the study everyone quotes — the long-term ones showing that people who accumulate more metabolites, or whose muscles swell more, after their first workout are the same people who grow most. Sounds like proof. It isn’t. It’s a fast-twitch-fiber study in disguise: people with more fast-twitch fibers accumulate more lactate, swell more, and grow more. The fiber type causes all three. It’s a correlation between different people, not evidence that swelling or metabolites cause anything. Stated out loud it lands like a card trick — say the sentence, everyone goes “wow” — but there’s nothing in the hand.

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies I reference above:

Two hooks, not new physics — to be honest, specificity and individuality don’t add a single new mechanism to anything I’ve already taught. They’re not more physiology. Think of them as the two coat-hooks I hang every mechanism on — the rep ranges, the rest periods, the proximity to failure, the fatigue math, all of it. Specificity decides what you’re training for; individuality decides how it has to be shaped for the person in front of you. Drop everything you’ve learned into those two, and you have a program.

Reasoning backwards from the finish line — at the top level of sport, specificity is a long chain you work backwards. Take a rower chasing a faster 2k. You don’t start in the gym — you start at the finish line. Analyze what actually separates the quick crews and it comes down to average power output per stroke. Now you have a defined ability to build. Break that down biomechanically and the engine is the leg drive — hip and knee extension producing the power. Power itself sits on the force–velocity relationship — it’s strength and speed together. Only now, after travelling all the way from the racecourse down to a specific power generator, do you build the strength-training exercises that target it. It’s an incredibly complex process, but at the highest level of sport it’s exactly what we should be doing.

Bodybuilding specificity is the opposite of complex. There’s really only one question: which muscle do you want to grow? Divide the body into regions and segments, then attack it segment by segment. The one genuine specificity decision you make is generic vs. specialization — do you train everything, or do you prioritize a few areas and put the rest on maintenance? A female physique goal is a classic specialization case: you’re not chasing every muscle, you’re building certain body parts and resting others on maintenance. Once that call is made, the rest is mechanical — I already know which exercises target which muscles. Everything left over is just rest period, rep range and proximity to failure, and I set those from the hypertrophy and fatigue material I’ve already built.

Once specificity has named the target, I tailor the program to the person — and I do it in a strict order of importance. These are the five dials, from the one I turn first to the one I barely touch.

Every set is not worth the same — the first proper dose-response profile for training volume came out of a Schoenfeld meta-analysis back around 2018, and there’s been more analysis since. The profiles differ in detail, but they share the one feature that matters: they are non-linear. You do not get the same stimulus from every set in a workout. Set one is worth a lot; by set five or six, the return has shrunk hard. That’s not a mystery — it’s exactly what you’d expect, because fatigue builds across the workout, leaving less and less capacity to generate a real stimulus set to set. On top of the diminishing returns, the work of Stuart Phillips points to a plateau somewhere between 6 and 8 sets in a single workout for a muscle — beyond that, you’re not stimulating meaningful extra growth in that muscle that session.

The right per-session volume isn’t “as much as you can survive.” It’s the amount that still lets you train that muscle multiple times a week — because weekly volume, spread across recoverable sessions, is what actually accumulates. And that turns the whole question into recoverability math. For moderate loads taken to or near failure, the recovery figures are roughly 3 sets → 48 hours and 5 sets → 72 hours. Run those forward into a week and the caps write themselves.

Where the numbers bend — those figures come from studies using failure, moderate loads and stretch-position exercises — the most damaging combination there is. Move toward lower-damage work and the math loosens. A set of hip thrusts at a 5RM with a rep left in reserve is not the same beast as the squats tested at an 8–10RM in the literature — so a person doing the former might genuinely get away with 5 sets three times a week. There isn’t much of a trick to get around the underlying limit; the only real lever is chiselling away the variables that cause excess damage. As a ballpark for the volume dose-response, though, the 3 and the 5 are probably pretty good numbers to build around.

What about a cap on total sets across the whole session, all muscles combined? This is one of the most common questions I get, and I’ll be straight: I have no data for it — there’s genuinely nothing to show you. From personal experience I land somewhere around 10–12 sets per session, but I sit at the very fast-twitch end of the spectrum, so I wouldn’t hand that number to someone with an endurance background and expect it to hold. Treat it as my anecdote, not a rule.

One thing the recoverability math does settle for me: it keeps pulling me toward full-body training over upper/lower splits. On an upper/lower you almost can’t max out the per-muscle volumes without leaving a problem for the following day — the frequencies just don’t line up. Full body every other day, or four-times-a-week full body, gets powerful precisely because the movement frequency is so high. That’s where I keep coming back to.

Fatigue, run in reverse — potentiation is a temporary increase in exercise performance caused by a previous bout of exercise. That last part is the whole discriminator: it has to be exercise that caused it. If a training partner cheering you on lifts your performance, that’s real, but it isn’t potentiation, because exercise didn’t do it. Inside a workout, potentiation splits into two mechanisms that got tangled together for years.

Motivation isn’t a soft add-on — it physically changes the stimulus. Under the effort model, anything that raises your maximum tolerable perception of effort raises the central motor command you can generate, which raises motor-unit recruitment. And that pays twice: bigger hypertrophy signal now, plus a higher recruitment ceiling for next session too. There are two separate routes into that effect, and the good news is they add together.

And they’re additive — stand next to that bar-speed screen and start calling out the numbers — someone sees 1.19, 1.18, 1.17 and you push them for a round 1.2 — and they hit it. That’s the internal pathway (their own reading of the numbers) plus the external pathway (your voice) stacking into a total higher than either produces alone. They’re two genuinely separate things that sum. For a lifter that means the best case is simple: keep the logbook and have someone in your corner pushing you past last week.

I’ll flag this upfront: focus of attention is nowhere near as useful as motivation. I mostly cover it to clean up internet misconceptions. Two options for where you point your conscious mind during a set:

External focus — conscious mind on the output: moving the bar from A to B, while the subconscious runs the actual movement.

Internal focus — conscious mind on the way your body is moving. The mind–muscle connection is a subclass of internal focus, not the whole of it — telling a powerlifter “keep your elbows tucked” is an internal cue but it isn’t a mind–muscle connection. Mind–muscle is just a more extreme form of internal focus, where you deliberately squeeze the target.

External focus produces three physiological effects — and the third is the one that gets S&C coaches shouting at me:

Flip all three for internal focus: worse coordination, higher antagonist coactivation, worse performance — but more recruitment concentrated in the target muscle. Add a mind–muscle connection and you create an agonist–antagonist squeeze that layers two different motor commands: “do the movement” gives one recruitment order, “squeeze the muscle” gives another, and the combination lights up a different set of motor units — which may even drive different regional hypertrophy than an external focus would. That’s the one genuinely interesting thing internal focus offers. But it comes with a catch almost nobody sees.

Strip all of it down and here’s how these principles cash out on the gym floor.

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies I reference above:

Before I can talk frequency for any adaptation — neural, hypertrophy, tendon, whatever — I build it out of the same three moving parts. Get these three right and you can model literally any adaptation you can name. Frequency is just the question of how these three pieces overlap in time.

A word on tendons — tendons get stiffer and bigger, and their collagen synthesis rate tracks muscle protein synthesis fairly closely, lasting roughly 48 to 72 hours. So there’s probably no meaningful difference between tendon and muscle adaptation timescales — we just have far fewer studies to say much more. Sarcomere genesis is the same story: we know a static stretch triggers it, but nobody has run a single stretching workout and watched the clock, so I can’t give you its exact stimulus duration. I’d bet it’s very similar.

A muscle fiber isn’t a balloon that stays inflated until something pops it. It’s more like a dial spring-loaded to return to zero — the moment your hand comes off it, it starts drifting back down on its own. Atrophy isn’t something being done to you. It’s just what happens in the absence of a stimulus. Growth and shrinkage are the same process pointed in opposite directions: deliver a stimulus and the dial climbs; stop, and it falls back to default. That’s it.

The cleanest proof comes from immobilization studies — the ones everybody wants to throw out because they seem too extreme to apply to training. Physiologically they aren’t different from de-training at all; the only difference is that a greater number of fibers are involved. The objection people raise is that immobilized muscle shows elevated inflammation, and they insist that inflammation is actively catabolic — a wrecking crew that moves in and guts the fiber from the inside. So test it: biopsy the immobilized muscle and measure both synthesis (MPS) and breakdown (MPB).

What actually happens — synthesis rates tank down to nothing, and breakdown rates actually slightly reduce. There is no wrecking crew. There is no argument that inflammation is actively catabolic. You atrophy for exactly one reason: you stopped supplying a hypertrophy stimulus. I think people reject this simply because it doesn’t sound nice — a previously unrecruited fiber sits there thin, with almost no contractile protein, and only starts building when you finally switch it on. Stop switching it on and it goes right back to that thin default. That’s the whole story.

If atrophy is the default, the real programming question is: how fast does the dial fall? Three separate lines of data answer it, and they all point the same direction — much faster than the internet believes.

The curve isn’t a straight line — and the shape has a nasty implication. The more of something you have, the easier it is to shed a big chunk of it fast. So the fastest rate of loss happens immediately at the end of the stimulus period. The instant your ~48-hour window closes and you drop into dissipation, that exact moment is when you’re bleeding muscle mass fastest. Recovery isn’t a grace period; it’s the steepest part of the slide.

Tendons play by their own rules — and they’re biphasic. Coming off training into normal daily activity, tendon stiffness takes about twice as long as muscle mass to drift back — if muscle takes a month, the tendon takes two; it really drags its feet. But going the other way, from activity into immobilization, tendon stiffness evaporates at twice the speed of muscle. Practically: a holiday or a training break, don’t give the tendon a second thought. But an injury that immobilizes a limb for two weeks — now I’m worried about the tendon, because two weeks is plenty for a massive stiffness change. Do not let someone charge straight back into something explosive on day one, especially a strong person, because the tendon injury risk is real.

So what do you actually do with stretch-position exercises? Because sarcomere genesis reverses so fast, you can’t space them out — you still need them about twice a week, same as everything else. You just take the recovery edge off with heavier loads and a rep or two in reserve rather than grinding them to failure.

Now stitch it together. The stimulus lasts ~24 hours. Atrophy is the default and it’s fast. Put those two facts side by side and frequency stops being a preference — it becomes the highest-leverage variable you’ve got. The proof is in the maintenance data: a couple of studies show that just 3-4 sets once a week is enough to hold your size exactly where a higher-frequency program left it. Sit with that. If a single weekly session maintains size, then you must be losing size at the back end of every week and clawing it back at the start of the next. Maintenance isn’t a special signal — it’s a break-even between building and bleeding.

You cannot argue this is some “maintenance stimulus” that tells the fiber to hold steady all week — that would require the fiber to know no further workout is coming. It doesn’t know anything. The same number of sets spread across multiple workouts grows muscle robustly; the same sets crammed into one workout merely maintains. The difference is entirely frequency.

Then why do modern meta-analyses say frequency doesn’t matter? Because the last decade of studies pushed volumes so high that everyone walks out of every session with enormous muscle swelling — and we can’t tell swelling apart from real hypertrophy. Drown the signal in enough water weight and the frequency effect vanishes from the data. But look at the lower-volume work and it’s obvious: three sets three times a week blows nine sets once a week out of the water. An early meta-analysis (around 2016) found exactly that — higher frequency won — before the flood of high-volume studies buried it.

The dose-response curves seal it. Two big meta-analyses disagree on how steep the volume curve is — one is shallow, needing about 6 sets to double the response from a single set; the other is much steeper, doubling in about 3 sets (I think that one’s a touch aggressive, probably muscle swelling pulling the top end up). But here’s what matters: both curves are non-linear and both bend over. The first set of the week is worth far more than the sixth. And frequency is nothing but a machine for buying you more “first sets” of the week — more trips up the steep part of the curve, fewer wasted reps up top where the curve has already flattened.

So the logic runs: set your frequency as high as recovery allows, then let volume-per-session fall out of your recoverability. Here are the numbers I actually build from — all drawn from moderate-load-to-failure, stretch-position data, so treat them as the demanding end of the range. Heavier loads with reps in reserve in a contracted position let you recover faster and would push these lines up, but there’s no data to draw that exact line, so I stay conservative.

The maintenance math inside A-B-A-B — think of each body part as a bucket with the drain left open. Every set pours stimulus in; every day of dissipation drains some out. Call one set one unit of stimulus. By the time two days of dissipation have passed, you’re down to about 0.67 of a unit — you’re still owed a day’s worth of loss. Run the accounting across the week and it comes out clean: inside an A-B-A-B rotation, one set per body part holds it at perfect maintenance, and two sets produces actual progress. If a region only gets trained once a week, you need three sets just to break even. Frequency is what makes one or two sets enough — spread the pours out and you never let the bucket drain far.

Why I don’t bother periodizing loads — all loads do fundamentally the same thing; they just hand you different amounts of fatigue. So I keep good exercises in the program instead of rotating them out to reinsert three months later — either an exercise is a good idea or it’s a bad one. Varying things for the sake of variety doesn’t make physiological sense. And “resistance-profile matching” is really just variety by another name: a matched profile loads more tissue and reaches higher up that non-linear curve per workout, exactly the way three sets of three different exercises beats three sets of one at high volume. Same principle, wearing a different hat.

Everyone has lived this: the first workout of a new movement wrecks you for days, and by the second or third session the same work barely registers. That’s the repeated-bout effect — post-workout fatigue and damage dropping off hard across successive sessions. And it’s a great worked example of how I reason about physiology, because the mechanisms that should explain it don’t add up to the curve we actually see. The gap is the whole story.

Since post-workout fatigue is a clean proxy for how much calcium piled up inside the fiber, I can build the repeated-bout effect out of anything that reduces calcium-related fatigue. There are four candidates:

Here’s the mismatch — add up the first three mechanisms and they predict a smooth taper: a big drop from workout 1 to 2, then a gradual glide down to basically zero over about 2-3 months. But the real data is nothing like smooth. It’s a cliff: massive from workout 1 to 2, small from 2 to 3, tiny from 3 to 4, and essentially gone after that. Way steeper than the mechanisms can account for. That mismatch is the clue that something else is doing the heavy lifting up front.

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies I reference above:

Back in the 1980s, a group of powerlifters working in exercise science noticed something: when they started bolting aerobic work onto their weekly routine, their strength training stalled. The gains they expected didn’t show up. They floated the idea that the two kinds of training were somehow interfering with each other, and that idea ran around the world fast.

What matters is how the research then split onto two completely separate tracks, and how differently the two tracks aged.

The molecular model — here’s the story track two told. An aerobic workout drives local endurance adaptations: more mitochondrial density, more capillarization, more oxidative enzyme activity. A strength workout drives local force adaptations: hypertrophy, sarcomerogenesis, lateral force transmission. The proposed collision point was signaling. Aerobic work switches on AMPK — a cellular energy sensor that kicks off the endurance program. Strength work runs through mTOR, the molecule famous for hypertrophy. The claim was that AMPK’s signals were directly competitive with mTOR’s.

The way I’d picture their argument: two departments in the same company — an endurance department and a growth department — both filing a budget request with the same finance office, and finance can only fund one of them at a time. Train both in a week and each gets shortchanged. Neat story. It just wasn’t true.

While the mechanism-chasers were opening panels, the quiet empirical track finished its meta-analyses — and buried in the results was the single most important finding in this whole area. When you rank the interference effect from largest to smallest, it lines up like this:

Sit with that order for a second, because it’s decisive. If aerobic exercise were interfering by blunting hypertrophy, then muscle size should take the biggest hit and everything downstream of it should follow. Instead size is dead last. The order is upside-down from the original hypothesis. That single ranking blows the hypertrophy basis of the whole thing to pieces — slower strength gains cannot be coming from blunted growth when growth is the least-interfered-with outcome on the list.

If cardio is just fatigue, then the game is picking cardio that delivers the cardiovascular hit while creating the least muscle damage. The five variables that drive calcium-related fatigue in lifting collapse, for aerobic work, into effectively three:

The distinction almost everyone misses — the intensity of muscle activation is not the same thing as the intensity of cardiovascular output. You can hit an identical cardiovascular load through wildly different muscle activation. Deliver it through your arms alone on a hand-crank and you’ve crammed all that work into a small muscle mass — serious soreness tomorrow. Deliver the same cardiovascular output through your legs on a bike — probably fine. Deliver it through all four limbs at once on a cross-trainer and you’d likely just walk it off on the way home. Four limbs sharing the load gives a big kick to the cardiovascular system while spreading the muscular demand so thin you never really get into the fast-twitch fibers. It’s the mirror image of what I do with single-joint, single-arm work when I’m trying to maximize motor-unit recruitment — same dial, turned the opposite way.

Plenty of bodybuilding coaches push assault bikes and spin bikes as the “least interference” cardio, and the evidence does show less interference from cycling than running. So an evidence-based coach picks the bike and stops there. But evidence is only interesting when you have no idea what’s going on. If you actually understand the physiology, you don’t stop at the evidence — you ask why cycling wins, realize it’s the stretch and eccentric component, and then extrapolate: the cross-trainer beats the bike too, for the same reason, only more so. That’s a scientific process reaching a better answer than the evidence alone would hand you — and it’s the opposite of what a lot of coaches recommend.

Any hard cardiovascular effort flips you out of your resting parasympathetic state into a sympathetic one — heart rate up, blood pressure up, oxygen shuttled out to working muscle. Afterward you drift back to parasympathetic in two stages, and the difference between them is the whole story on recovery monitors.

What the window costs you — inside that 48-hour window, your cardiovascular system is already sitting partway up the sympathetic ramp, so there’s less headroom left to climb to maximum — and a maximal cardiovascular effort will come up short. Attempt an all-out cardiovascular time-trial inside the window and you probably won’t hit a personal best; wait one extra day so you’re fully outside it, and — assuming you’ve prepared properly — you’ve got a real shot at beating your previous best. So autonomic state only registers as a fatigue mechanism when there’s an actual cardiovascular limitation.

Where HRV actually earns its keep — it can tell you whether you’re still inside a 48-hour window from a previous cardiovascular workout — i.e. tapering for a race or time-trial. That’s real, but even there I’d hold it lightly. Endurance athletes don’t need to run their cardiovascular system at maximum to train it. Elite swimmers finish an Olympic final and are back in the pool that Monday, every single day until the next Games — ticking along at low intensity still delivers a cardiovascular stimulus, even inside the 48-hour window. For a cardiovascular-dependent competition I’d take the 48 hours beforehand off, absolutely. For a power competition I see no reason to chill out for two days first. HRV is, at best, of minimal importance.

A combat-sports aside — fighters will deliberately spike their heart rate for two or three minutes right before a match — a short, sharp ~80–90% burst. Skip that primer and you gas out embarrassingly early. That’s not a contradiction of the 48-hours-off idea — it’s the best of both worlds. Do a power session, take 48 hours off, then compete: you collect the delayed potentiation from the strength work plus the autonomic benefit of the rested window. Then, minutes before you step on, you hit the immediate potentiation mechanisms — both cardiovascular and local — and you get all three stacked at once.

Open any periodization textbook ever written and you’ll find it leaning on the General Adaptation Syndrome to justify why we periodize. It’s worth knowing where GAS came from and why that justification doesn’t actually hold.

In the 1950s, researchers studying the sympathetic nervous system put animals through genuinely disagreeable conditions — including isolating them so they couldn’t see another animal for something like 23 hours a day. They described a two-part response: an alarm phase (the healthy stress reaction) and, if the stress ground on, a collapse phase (the pathological one). Useful physiology. Then exercise scientists borrowed it to model training as a “stressor,” and the actual stress researchers more or less lost their minds — because that borrowing skips the one requirement that defines a stressor in the first place.

The stress research that IS worth your attention — psychological stress genuinely matters, just not through GAS. Studies on life events make the point cleanly: take patients in for a minor operation — roughly a week until the stitches come out — and have them fill in a one-line questionnaire beforehand about big life events in the last six months or so (divorce, moving house, changing jobs). The higher the stress score, the longer the wound takes to heal. Psychological stress suppresses the same immune and inflammatory machinery that repairs muscle damage. So in a high-stress stretch of life, back off — leave more reps in reserve, drop a set. Sleep works the same way, on a threshold: normal is around 8 hours, and chronically running at ~5 (losing an hour or so a night, week after week) becomes a real stressor. But going from normal sleep to more sleep doesn’t buy you extra recovery — it’s a threshold, like protein, not a dial you keep turning up.

Strip periodization down to its essence and it’s just one thing: distributing the workout contents you’ve already decided you need, across time. The metric that defines any periodization scheme is what I’d call its wavelength — the length of time you go without changing a variable. That single number lets you lay every approach out on one spectrum.

Draw the line at zero periodization and every step away from it just stretches the wavelength longer. As you slide from zero toward block, five physiological factors shift in equal and opposite ways — that’s the part that actually decides how to program. Block wins on some, loses on others, and zero periodization is the exact mirror.

Add it up and I’m not a huge proponent of aggressive periodization. I don’t think it has an enormous amount to offer over just building a program that works and sticking with it. People assume a “periodized” program requires a month heavy, a month moderate, a month light — but if that variation wasn’t part of your ideal workout plan to begin with, why is it suddenly buried in your program now? Periodization isn’t a licence to vary things for the sake of varying them. It’s only ever redistributing contents you already decided you needed.

Peer-reviewed sources for the concepts above (via PubMed) — supporting research, not necessarily the exact studies I reference above: