Tethered Swim Force Test: Validity, Setup, and Interpretation

Ask ten swim coaches how they test power and eight will hand you a vertical jump mat, a medicine ball throw, or a bench pull one-rep max. Every one of those numbers describes an athlete standing on solid ground. None of them describe what happens when that same athlete's hand locks onto the water and pulls against something with mass behind it. Yeater et al. (1981) ran the first rigorous version of that missing test — a swimmer tethered to a fixed rope and load cell — and found the force recorded on the line correlated with free-swim sprint velocity at r = 0.85 to 0.93 across crawl, breast, and back. No dry-land test since has matched that relationship as consistently. This piece covers why the tethered setup produces a more valid propulsion number than anything measured on land, how to rig it so the reading actually means something, and where the method still falls short of race-day truth.

The core validity argument for tethered testing is mechanical, not statistical. Because the cable is fixed to the wall and effectively inextensible, the swimmer's centre of mass does not translate forward during the effort. With no net forward velocity, there's no meaningful acceleration of body mass to subtract from the force budget and no velocity-dependent form drag building up stroke after stroke the way it does in free swimming. What the load cell reads is close to the swimmer's raw propulsive output at that instant — a signal a vertical jump or a bench pull simply can't produce, because neither loads the specific joint angles, hand pitch, and catch timing of a swimming stroke against water resistance.

Dry-land power tests still have a place in training, but their correlation with swim velocity is modest and inconsistent across studies. A swimmer who adds 10 kg to their bench pull hasn't necessarily improved the specific pulling pattern and timing that convert strength into swim speed. The tethered test measures that conversion directly, in the medium where it actually happens.

Four components determine whether a tethered rig produces a trustworthy number, and getting any one wrong quietly corrupts the data with no obvious warning sign.

• Tether material: a non-stretch line — braided stainless steel cable or a low-stretch cord such as Dyneema, 3 to 5 mm diameter. Nylon rope with more than 1 to 2% elongation under load acts like a soft spring: it smooths the force-time curve, delays the rise to peak, and can understate true peak force by 15% or more.
• Load cell: a 0 to 500 N in-line strain-gauge transducer is sufficient for age-group and most female swimmers; elite male sprinters can exceed 250 N at peak, so a 0 to 1000 N cell gives headroom without sacrificing resolution. Sample at 50 Hz minimum, 100 Hz if possible, so individual stroke peaks are resolved rather than smeared across a couple of data points.
• Attachment point: a belt around the hips, near the swimmer's centre of mass — the standard since early studies found ankle attachment altered kick mechanics enough to change the force signal for reasons unrelated to fitness.
• Anchor geometry: mount the pulley or eye-bolt at water-surface height so the cable stays within roughly 5 degrees of horizontal through the full stroke. An anchor set too high or low introduces a vertical force component that a load cell reading only axial tension cannot separate from genuine propulsion.

Calibrate before every session by hanging known weights (5, 10, 15, 20 kg) on the line and checking the reading against expected force. Log water temperature too; a noticeably cold pool stiffens limb tissue slightly and is worth noting if numbers look unusually low.

1. Warm-up: 600 to 800 m of progressive swimming, finishing with 4 x 25 m building to race pace.
2. Familiarisation: at least two submaximal tethered trials of 10 to 15 seconds, three minutes apart. Amaro et al. (2014) found reliability in age-group swimmers jumped from an ICC of roughly 0.68 to above 0.90 once a single familiarisation trial preceded the test — swimmers who've never felt what it's like to pull hard and not move forward tend to shorten their stroke and panic-kick through the first real attempt.
3. Trial structure: a 10-second maximal trial captures peak force and rate of force development; a separate 30-second trial captures mean force and the fatigue index. Leave at least 8 minutes of rest between them.
4. Start: the swimmer floats into position with the line already under light tension, not slack, and starts on a standardised cue. Exclude the first two stroke cycles from analysis — they carry the transient spike of building tension before a steady rhythm settles in.
5. Filtering: pass the raw signal through a fourth-order low-pass Butterworth filter, cutoff around 10 Hz, before extracting peak, mean, and decay values.
6. Repeat sessions: keep tether length, anchor height, belt position, and warm-up identical every time. A rig moved even 30 cm changes the cable angle enough to shift the numbers independent of any real change in the swimmer.

The reliability and validity picture for tethered swimming is stronger than most swim-specific tests, though the studies behind it are smaller than strength-and-conditioning researchers would like.

• Yeater et al. (1981) established the foundational relationship: tethered force correlated with competitive sprint velocity at r = 0.85 to 0.93 across three strokes. The limitation is the era — equipment and filtering have improved substantially since, so these coefficients should be read as a floor, not a current benchmark.
• Kjendlie and Thorsvald (2006) reported excellent test-retest reliability for mean tethered force, with ICCs in the 0.95 to 0.99 range across sessions in trained swimmers. The limitation of tightly controlled lab studies like this one is sample size — a handful of well-trained swimmers doesn't guarantee the same consistency with a less standardised club rig.
• Morouço et al. (2011) measured tethered force across all four competitive strokes in national-level swimmers and found mean force correlated with 25 m free-swim velocity at r = 0.78 to 0.93 depending on stroke, strongest in front crawl. The authors' own flagged limitation: the sample was restricted to national-level swimmers, so the relationship may not hold as cleanly at recreational level.
• Amaro et al. (2014) found reliability in age-group swimmers depended heavily on familiarisation — moderate without a practice trial, excellent with one. The limitation is that youth force output is also confounded by growth and maturation between sessions, which adults don't have to account for.
• Morouço et al. (2015) partitioned tethered force into arm-only, leg-only, and full-stroke front crawl and found arms contributed roughly 70 to 75% of total force versus 25 to 30% from the legs. The limitation: isolating arms or legs changes body position and buoyancy enough that the isolated numbers don't simply sum to the full-stroke result.

A tethered trial produces a raw force-time curve, and five derived numbers turn that curve into something a coach can act on.

Published ranges for full-stroke front crawl give a rough sense of where a given swimmer sits, though rig geometry differs enough between labs that these should be read as context rather than a strict pass-fail standard:

Most bad tethered data doesn't come from a broken load cell — it comes from small setup choices nobody flags as a problem until the numbers stop making sense session to session.

• Cable angle drift: an anchor even 20 degrees off horizontal reads roughly 6% low on true propulsive force, since the cell only measures axial tension, not the true horizontal component. Recheck anchor height whenever pool depth or lane assignment changes.
• Elastic tether material: nylon rope stretching 3 to 5% under a 150 N load rounds off the rise to peak force, understating peak and rate of force development even though mean force over a longer trial looks roughly normal — exactly why this error goes unnoticed for a long time.
• Skipping familiarisation: without a practice trial, swimmers shorten stroke length and raise stroke rate because there's no forward glide to time against — that panic-stroke pattern contaminates the data with a technique the swimmer would never use mid-race.
• Changing rig geometry between sessions: a tether shortened by half a metre or an anchor remounted on a different lane line shifts force readings independent of any real change in the athlete.
• Inconsistent verbal encouragement: a self-paced maximal effort without a standardised cue produces meaningfully lower force than an encouraged trial — a well-documented effect in maximal-effort testing generally.

Tethered force is a strong correlate of swim performance, not a direct substitute for it. Because the swimmer's centre of mass never actually travels forward, the water flow around the body doesn't replicate the velocity-dependent form drag and bow-wave interaction that build up during free swimming at race speed. A swimmer can post a strong tethered number and still lose a race to poor stroke timing, breathing pattern, or turns — none of which the tether captures.

Stroke mechanics also shift under tether. Most swimmers raise stroke rate and shorten stroke length once they realise they aren't moving forward, so the test partially measures a modified pattern rather than actual race-day technique. And as Morouço et al. (2015) showed, arm and leg contributions measured in isolation don't cleanly sum to the full-stroke total, so isolated-limb testing is a rough proportional guide, not an exact breakdown.

There's also a real cost barrier: a calibrated load cell, acquisition software, and a properly mounted anchor represent genuine money and setup time next to a stopwatch and a 25 m sprint — why most programmes run the full protocol quarterly and lean on cheaper proxies in between.