Ask which exercise is best for a muscle and you'll get a confident answer. Ask what range that exercise is performed through, where in the range it's hardest, and how the answer changes for someone with different limb proportions, and the confidence usually evaporates.
That second cluster of questions determines most of the outcome. The first is close to unanswerable without them.
The name of an exercise specifies almost nothing
"Squat" describes a category. Within it, depth, stance, bar position, torso angle and individual proportions vary enormously — and those variations change which muscles are loaded, at what lengths, and how much. A half-depth high-bar squat and a full-depth one are not the same stimulus with different amounts. They're different stimuli.
Same for everything else. A pulldown to the collarbone and one stopped at the forehead. A press through full range and one cut at three-quarters. The label is identical; the mechanical event is not.
So when someone says an exercise didn't work for them, the far more likely explanation is that the version they performed didn't load the muscle where it mattered — not that the movement pattern is defective.
What the range-of-motion research actually says
Two findings are reasonably robust, and one is genuinely interesting.
Full range generally beats partial range. Across a range of muscles and populations, training through a fuller range tends to produce equal or greater hypertrophy than a shortened one — often at lighter absolute loads, which is worth sitting with. Less weight through more range beats more weight through less.
Range influences where growth happens. Hypertrophy is not uniform along a muscle. Different ranges preferentially develop different regions — a fact that quietly dismantles the idea that a muscle is a single unit receiving a single dose.
The length at which you train appears to matter more than the total range. This is the interesting one. A growing body of work suggests that training in the lengthened portion of the range is disproportionately effective — and that partial reps performed at long muscle lengths can match or exceed full-range training for hypertrophy, while partials at short lengths reliably underperform.
That's still an actively developing area and worth holding loosely rather than treating as settled. But the direction is consistent enough to act on: the stretched position is doing a lot of the work, and it's the position people most commonly cut short.
Proposed mechanisms include greater passive tension contributing to the mechanical signal, higher tension per active fibre at long lengths, and possibly the addition of sarcomeres in series. Mechanism is unsettled; the practical implication is not.
Resistance profiles: where the exercise is actually hard
The second half of this is that resistance isn't constant through a range, and the pattern differs by implement.
With free weights, resistance comes from gravity acting straight down, so the demand on a muscle depends on the moment arm — the perpendicular distance from the joint to the line of force. That distance changes continuously as you move. It's greatest when the limb segment is horizontal and shrinks to nearly nothing when it's vertical.
Which explains a lot of otherwise puzzling things:
- A dumbbell curl is brutally hard at 90° and nearly free at the top, because the moment arm collapses as the forearm approaches vertical.
- A dumbbell lateral raise is hardest at the top and trivial at the bottom — the stretched position, where you might most want tension, has almost none.
- Lying and incline variations exist largely to move peak resistance to a different part of the range. An incline curl loads the stretched position that a standing curl doesn't.
Cables and many machines break the gravity dependence, delivering a much flatter profile, and well-designed machines use cams to shape resistance against the joint's natural strength curve. This is the substantive argument for using them alongside free weights — not that they're easier, but that they can load positions free weights structurally cannot.
Your leverages change the answer
The last piece is individual, and it's why exercise rankings fail as universal advice.
Segment lengths — femur relative to torso, forearm relative to upper arm — change the joint angles required to perform a movement, and therefore change which muscles are emphasised. A long-femur, short-torso lifter squatting to depth must lean further forward, shifting demand toward the hips and back. Same exercise, same depth, materially different distribution.
Tendon insertion points vary too, altering internal moment arms and where in the range a muscle produces force most effectively. Muscle attachment variation influences which portion of a range feels strong.
None of this is an excuse for anything. It's a reason that a fixed list of best exercises can't be right for everyone, and that the sensible unit of individualisation is which positions get loaded rather than which exercise names appear in the programme.
What to actually do
Take everything through your full available range. The default position, deviated from only for a specific reason. Where load and range conflict, range usually wins.
Make sure something loads the stretch. For each muscle, at least one exercise where meaningful tension exists in the lengthened position. This is often the gap in an otherwise sensible programme — plenty of peak-contraction work, almost nothing at length.
Check your actual range on video. Perceived range and real range diverge, and the divergence grows with fatigue, exactly as the previous piece describes. Most people's range is shorter than they believe, particularly under a heavy load.
Use the implement that loads the position you want. If the movement's hard part is at the wrong end of the range, change the implement or the angle rather than the exercise. An incline, a cable, a different machine — small change, different mechanical event.
Don't buy load with range. Adding weight while shortening the movement is one of the most common ways progress stalls invisibly. The logbook improves; the stimulus doesn't. Progress the number only when the range is held constant — otherwise you're not comparing like with like.
Choose variations that fit your proportions. If a movement can't be loaded well given your build, a variation almost always exists that trains the same muscle through a better range for you. That's individualisation on the variable that matters, not on the exercise name.
The point
Exercise selection is a proxy — a shorthand for a bundle of underlying mechanical properties. It's a useful shorthand, and it's the wrong level of abstraction to make decisions at. The properties themselves are what produce the adaptation, and they can be varied enormously within any given exercise name.
Which is why two lifters can run the same programme and get different results, and why the answer usually isn't the programme. It's that the reps were different, in a way the spreadsheet has no column for.