Contact Information
Four Oaks House, 160 Lichfield Rd
Sutton Coldfield, West Midlands
B74 2TZ, UK
+44(0)1676 530 730 | +44(0)121 2273 335
Latest Tweets
Be Part Of The Quintic Community…

Rugby Union Scrummaging
(Machine vs Human)

Q4E Case Study 31 – Rugby Union Scrummaging (Machine vs Human)

Proposed subject useage

Sports Science / Mathematics / Physics
Coaches / Performance Analysis

(A/AS level / Degree Yr 1/2)


With approximately 25 scrums per game it has become a skill where dominance can result in performance advantages (Quarrie et al., 2013). This requires a large forward force, long associated with a “low body position”: flat back parallel to the ground and a knee angle of 115-125° (Hislop, 1982). This is supported by Quarrie and Wilson (2000) who suggest a negative correlation between hip and knee angle, and forward force production in a parallel stance.

Scrum machines are widely used to replicate conditions of a scrum during training. However, there is little literature supporting the scrum machine’s effectiveness in developing an efficient in-game scrummaging technique.

The aim of this study is to examine the differences in angles and velocities of the hip and knee when scrummaging against a machine alone and with a modified scrum of five.



Participants: Data was collected on two amateur rugby union props. Neither participant had history of lower leg surgery.


  • Players were recorded during two scrummaging conditions: alone against a scrummaging machine (Predator, Devon, UK) and with a modified pack size of five in a live scrum.
  • Retro-reflective markers (15mm) were placed on joint centres according to the protocol in Figure 1.
  • Data was captured using a high-speed camera (EX-FH100, Casio, Japan), recording at 120 frames per second (fps).
  • Players followed the current IRB engagement protocol and completed a 3 second maximal effort.

Data Processing/Analysis:

  • Trials were digitised, manually and automatically (Quintic Biomechanics v29, Quintic Consultancy Ltd, UK). , from the “Bind” call (Figure 2, A) for the entirety of the player’s effort.
  • Frame markers were used to identify engagement (B) and the start of the maximal static push (C).
  • Angular and linear analysis were performed using the Quintic Biomechanics Angular and Linear Analysis applications, with reference to the hip and knee (Figure 1).

Figure 2. Three key time events used during digitisation. Digitisation started at the “Bind” (A). Frame markers were placed at impact during engagement (B) and the frame where the player assumed a static pushing position (C). Digitisation continued until the pushing position was interrupted or 3 seconds of maximal pushing was complete.


Figure 3. Linear velocity of the hip during impact of live scrums (n=9) (Blue) and scrum machine scrums (n=9) (Red). Data is displayed as the mean (Bold lines) and Upper and Lower 95% Confidence Intervals (Thin lines). The dashed line at zero represents impact.

Peak mean hip velocity occurred at impact and was greater during the machine trials (2.232 m/s) relative to 5-man trials (0.824 m/s), where the peak occurred prior to impact (Figure 3). There was no difference for mean hip velocity between conditions during the post-engagement static maximal push phase.

Figure 4. Angular velocity of the hip relative to horizontal during impact of live scrums (n=9) (Blue) and scrum machine scrums (n=9) (Red). Data is displayed as the mean (Bold lines) and Upper and Lower 95% Confidence Intervals (Thin lines). The dashed line at zero represents impact.

Average angular velocity of the hip relative to horizontal was greater in the scrum machine trial, peaking at 185.75 °/s, 0.002 s after impact. The live scrum trials peaked at 87.69 °/s, 0.089 s before engagement. There was little difference between the mean angular hip velocities between the scrum conditions during the static post-engagement pushing phase (Figure 4).

Figure 5. Angular Displacement of the knee during the post-impact pushing phase of live scrums (n=9) (Blue) and scrum machine scrums (n=9) (Red). Data is displayed as the mean (Bold lines) and Upper and Lower 95% Confidence Intervals (Thin lines). The dashed line at zero represents the frame where both feet are stationary and the start of the post-impact pushing phase.


Linear Hip Velocity

The hip velocity at impact was more than double (170% greater) in the scrum machine trials (2.232 m/s) compared to the 5-man live scrums (0.824 m/s). This suggests that players place a larger emphasis on impact speed when practising than during a live scrum.

Preatoni et al. (2015) suggested that the “Crouch, Bind, Set” (CBS) condition reduces impact speeds by more than 20% when compared to protocols pre-2013. The reduced impact velocity in live scrums compared to the scrum machine trials during the current study indicates that amateur players ignore the new engagement protocol when practising using a machine.


Angular Hip Velocity

Greater angular velocity was evident at the hip during trials using a scrum machine compared to the modified live-scrums. CBS protocol reduces impact velocity and the distance from set-up to impact (Preatoni et al. 2015). Reduced set-up distance and impact speed would facilitate decreased angular velocity at the hip. This further supports that practice with a scrum machine does not accurately replicate conditions during CBS engagement.

Angular Knee Displacement

Despite little difference between the average knee angles during a live and machine scrum, there is greater variation of knee angle during a live scrum. This suggests that an effective knee angle such as 115-125° (Hislop, 1982) is more easily maintained during scrum machine practice than in a live scrum.

Sayers (2007) suggests a negative training effect from practice using a scrum machine due to differences in lower limb kinematics. The current study supports this difference in lower limb kinematics with increased variation of knee angle during post-engagement pushing and increased linear and angular velocity of the hip during engagement when using a scrum machine.



  • Practising scrums using a scrummaging machine is not representative of live scrums using the current CBS engagement protocol.
  • It is suggested that players may benefit from practice using increased live/modified scrums to more closely replicate the technique used with current rules.



Preatoni, E., Stokes, K.A., England, M.E., & Trewartha, G. (2015). Engagement techniques and playing level impact the biomechanical demands on rugby forwards during machine-based scrummaging. British Journal of Sports Medicine, 49 (8), 520 – 528.

Sayers, M. (2007). Kinematic analysis of high-performance rugby props during scrum training. Science and Football VI: The Proceedings of the Sixth World Congress on Science and Football

Quarrie, K.L., & Wilson, B.D. (2000). Force Production in the Rugby Union scrum. Journal of Sports Science, 18 (4), 237-246.

Quarrie, K.L., Hopkins, W.G., Anthony, M.J., & Gill, N.D. (2013). Positional Demands of International Rugby Union: Evaluation of player actions and movements. Journal of Science and Medicine in Sport, 16 (4), 353-359.

Hislop, B. (1982). New Zealand Rugby Skills and Tactics: Training Programmes. Auckland: Landsdowne Press.

Presented at BIG (Biomechanics Interest Group) 2017 – Portsmouth University

Rugby Union Scrummaging (Machine vs Human) | Quintic Sports