Rugby Scrum Push Force Measurement: A Scrum-Machine Load-Cell Protocol for Peak vs Sustained Force

Ask a scrum coach what the machine read today and you'll usually get one number: the peak on the sled's display, and a nod that the pack is hitting hard. That number gets treated as the whole story. It isn't. A pack can slam a machine with real force for a third of a second, then bleed pressure for the next four seconds while the sled drifts backward under a steady opposition shove. On the scoreboard that pack loses a five-metre scrum penalty despite an engagement that looked ferocious on video.

The gap between how hard a pack hits and how hard it holds is what a single peak-force reading cannot separate. Two packs can post the same headline number and be completely different animals: one an explosive hitter that fades, the other a grinder that never gives an inch. Telling them apart means logging the full force-time curve, not just the maximum, then doing the arithmetic that splits engagement spike from sustained hold. The protocol below covers the load-cell setup, the sequence to run with a forward pack, and the ratio that turns two numbers into one training signal.

Quarrie and Wilson (2000), in the Journal of Sports Sciences, instrumented a scrummaging machine with load cells and measured force production across schoolboy, club, and representative-level packs engaging under game-realistic conditions. The headline finding wasn't simply that better packs push harder; it was that scrum force has two distinct phases: a short, high-magnitude spike at engagement, then a lower, steadier force during the sustained push. Mean peak engagement force ran well above mean sustained force, with representative packs peaking in the region of 15-17kN before settling closer to half that figure. The authors flagged a limitation worth carrying forward: force against an instrumented sled doesn't fully replicate force against a live, reactive front row, so machine values are best read as a standardized comparison tool, not a stand-in for a contested match scrum.

That two-phase structure is also why the sport changed its engagement laws. Preatoni, Stokes, England, and Trewartha (2013), in the British Journal of Sports Medicine, compared crouch-touch-pause-engage against the newer crouch-bind-set technique on an instrumented machine, specifically to separate the two phases, because scrum injury risk sits overwhelmingly in the compressive spike at first contact, not the sustained shove that follows. A protocol reporting only one number can't tell a coach whether a pack's risk sits in an overloaded engagement, a weak hold, or both.

The test needs a scrum machine instrumented with load cells capable of logging a continuous force-time curve, not a machine that only shows a peak digit on a dial.

Sampling rate matters more here than in most field force tests, since the engagement spike lasts a fraction of a second. A logger sampling below roughly 100Hz smooths over the true peak and understates it; 200Hz or above captures the spike shape cleanly enough to separate it from the sustained phase that follows.

1. Warm-up (15 minutes): General movement prep, contact-specific mobility, then two submaximal engagements to rehearse the bind without scoring.
2. Familiarization: Two moderate-intensity trials to standardize foot spacing, bind depth, hip height, and back angle.
3. Maximal trials: Run the full crouch-bind-set sequence, engage on the set call at maximal effort, hold a sustained push for 4-5 seconds, then release under control. Run 3 trials with 3-5 minutes of full recovery between each; an eight-man drive is near-maximal effort, and short rest inflates fatigue into the data.
4. Capture the curve: Log continuous force from 0.5 seconds before the set call through at least 5 seconds after engagement. Don't rely on a single peak digit; the full trace is what the math below needs.
5. Valid trial criteria: A trial only counts if the full pack binds and engages together, no player pre-loads ahead of the call, and the scrum doesn't collapse before the sustained window completes.
6. Scoring: Use the trial with the highest valid peak force and read its matching sustained force from that same trial's curve.

Total session time, warm-up included, runs 20-25 minutes for three scored trials. A truncated two-trial version works for weekly monitoring once a baseline is on file.

Two values come from every valid trial. Peak force (F_peak) is the highest instantaneous force in the engagement window, typically within 0.3-0.5 seconds of the set call. Sustained force (F_sus) is the mean force across a defined hold window after the spike settles, commonly 1.5 to 4.5 seconds post-engagement, chosen to exclude both the spike and any late-trial decay as the pack tires near release.

From those two numbers: PSR = F_peak / F_sus. A PSR near 1.0 means a pack holds almost as hard as it hits. A high PSR means the engagement number looks impressive while the grinding contest that decides scrum penalties and five-metre put-ins is comparatively weak.

Worked example: Pack A engages at F_peak = 15.8kN and holds a sustained mean of 8.4kN, a PSR of about 1.88. Pack B engages lower at 12.1kN but holds 9.6kN, a PSR of about 1.26. Pack A wins any test reporting only peak force. Pack B grinds out the scrum penalty in the 70th minute, because it converts more of its engagement into force it can actually sustain against live opposition.

Quarrie and Wilson (2000) reported that playing level changed the sustained phase more than the engagement spike. Representative packs didn't simply hit harder than schoolboy packs by a proportional margin; the larger separation showed up in the sustained push, where trained, heavier packs held a meaningfully higher fraction of peak force through the hold. That's the biomechanical signature of a PSR closer to 1.0, and the data pointed to sustained-phase strength, not raw engagement power, as the better marker of playing level.

Preatoni, Stokes, England, and Trewartha (2013) isolated what happens when the engagement sequence changes. Comparing crouch-touch-pause-engage against the newer crouch-bind-set sequence, they reported the newer technique meaningfully reduced peak compressive forces at engagement, on the order of a 25% drop at several contact points, while sustained push force was largely preserved. In PSR terms, the law change pulled packs toward a lower ratio without weakening the part of the scrum that actually produces a penalty or a stable put-in. Their stated limitation carries into any field application: the comparisons ran on instrumented machines under controlled conditions, and force against a machine doesn't map one-to-one onto a live, uncooperative front row, so absolute kN figures are level- and machine-specific rather than universal targets.

These bands come from the peak-to-sustained relationship in the research above and should be read against a pack's own history first, since absolute force varies by machine calibration, playing level, and body mass.

Read the band alongside the absolute sustained force, not instead of it. A pack at a PSR of 1.2 with a low sustained number isn't automatically in good shape; it may simply be under-hitting on engagement rather than over-performing on the hold. The ratio flags where the imbalance sits, but the absolute F_sus number is still what tells you whether the pack can win a scrum penalty against a comparable opponent.

A pack above 2.0 with a fast-fading sustained trace usually responds to work that targets holding position under load rather than more explosive engagement drills: extended isometric holds against the machine at 80-90% of peak force for 5-8 seconds, sled pushes emphasizing continuous drive over distance rather than acceleration, and coaching on hip height and back angle, since a pack that loses spinal alignment mid-scrum bleeds sustained force even with adequate individual strength. Retest every 3-4 weeks; sustained-phase strength moves slower than engagement power, and weekly testing mostly captures noise rather than real adaptation.

A pack already in the balanced band with strong absolute sustained force is closer to a conditioning and technique-maintenance problem than a strength problem. The more useful question shifts to individual contribution: which forward's drive holds up across a full match-length session and whose effort curve drops off, exactly the layer a team-level machine reading can't show and individual monitoring during scrum sessions can.