Author(s): POMARAT, Z., DUFOUR, J.E., PASSIEUX, J.C., WATIER, B., Institution: LAAS-CNRS, Country: FRANCE, Abstract-ID: 1499

INTRODUCTION:
The scrum is a critical phase of rugby union, involving high-intensity collective force production and strongly influencing match outcomes [1]. The horizontal pushing force generated by the pack is a key determinant of scrum performance [2] and is commonly assessed using instrumented scrum machines. Although widely used for training purposes, these machines may not accurately replicate the mechanical demands of live scrummaging against an opposing pack. The aim of this study was therefore to quantify individual player contributions to the total horizontal pushing force of the pack during live and machine scrummaging.
METHODS:
Eight elite male rugby forwards (age: 20 ± 1 years; height: 192 ± 7 cm; body mass: 112.6 ± 16.2 kg) performed three live scrummaging trials against another pack and three machine-based scrummaging trials. All players wore instrumented insoles (Loadsol Pro®, Novel, Germany; 100 Hz). Individual horizontal pushing forces were estimated using either a pre-trained generic or a personalized machine learning model [3]. During live scrummaging, the ball was introduced for the instrumented pack, whereas no ball was introduced during machine trials. Each trial followed a standardized “crouch–bind–set” engagement sequence, and data were analysed from the “set” command to 4 s post-engagement. Player contributions to the total pack force were calculated during the engagement (from “set” to 0.5 s after peak impact force) and sustained (remainder of the trial) phases.
RESULTS:
During the engagement phase, the front row contributed on average 45 % of the total pack force during live scrummaging (LS) and 46.6 % during machine scrummaging (MS). Contributions from the second and third rows were 25.3 % and 29.7 % (LS) compared with 25.5 % and 28 % (MS) respectively. During the sustained phase, front-row contribution increased in both conditions (50.2 % in live and 56.1 % in machine scrummaging), while contributions from the second and third rows decreased. In addition to row-level differences, the repartition of force among individual players varied between conditions and phases. Notably, during engagement, the tight-head prop contributed most to the front-row force in live scrummaging, whereas the contributions of the two props were equivalent during machine scrummaging. During the sustained phase, the hooker’s contribution was higher in machine scrummaging than in live scrummaging, likely due to ball handling during live trials.
CONCLUSION:
Both the scrummaging condition (live vs machine) and the phase of the scrum (engagement vs sustained) influenced force-sharing strategies within the pack. These findings underline the differences between live and machine-based scrummaging and provide practical insights for coaches seeking to adapt scrum machine training in order to better reflect the mechanical demands of live scrummaging.

References
[1] Scott et al., J. Sci. Med. Sport, 2023
[2] Martin & Beckham, BMC Sports Sci Med Rehabil, 2020
[2] Pomarat et al., J. Biomech., 2025