Despite the undoubted importance of efficient force delivery in the scrum, there is very limited published material addressing the actual dynamics of force delivery.
Powerful scrummaging is dependent on appropriate body position and limb alignment, not just in the relatively static situation immediately after engagement but throughout the entire contest of the scrum. Much of what passes for best practice in scrum formation reflects a failure to critically examine the actual geometry and mechanics of body position and how these change during the scrum contest.
I believe that an optimal configuration of body position and limb alignment on engagement involves hip and knee angles each set at 90° with both trunk and shank being parallel to the ground. During the scrum, hip and knee joints should move synchronously so that their angles remain equal. The hips may sink slightly relative to the shoulders but trunk and shank should remain parallel.
Modern thinking on scrummaging usually advocates consistency of body shape for all participants regardless of position, with the feet approximately shoulder width apart and toes level. There also seems to be general agreement on the need for the trunk to be horizontal or for the shoulders to be slightly higher than the hips. (Greenwood, 1978; Smith, 2000; NSWRU, 2004; Vickery; O'Shea, 2004: Argentinian Bajada method)
However, when joint angles are discussed there is substantial divergence of opinion on the appropriate angle at the knee joint:
Jim Greenwood, Total Rugby, 1978
More than three decades on Greenwood's book, though overtaken by a succession of Law changes, remains a rugby classic. Its underlying logic is compelling. The figure below summarises his views on body position:
"Thighs approximately vertical. It's obvious that the more acute the angle of the knee the greater the potential range of the drive, but the more strength is required to initiate it. … [Y]ou only have to go into the full-flexed position to realise that a drive from that position is very much slower and more difficult than a drive from a half-squat. Players tend to assume the position in which they feel most capable of a snap drive. On the other hand, the smaller their degree of flexion the smaller the range of drive. For a six-foot player, a flexion of 90° at the knee produces a potential forward movement of about a foot, which allows for a snap drive, and the necessary continuation shove. That is more than enough for all practical purposes, and may well be seen as a maximum."
Greenwood also emphasises pack height:
Smith emphasises body position. "Each player must take up a position by which the force generated by the large muscles of the lower body, the quadriceps and gluteals particularly, can be transmitted effectively and SAFELY through the spine, the shoulders and the neck."
Smith examines the consequences of a prop being experienced enough and strong enough "to alter the height of the scrummage quite legally." and "produce a significant disruption of the opposition scrummage. A prop can thus legally force his opponent to scrummage lower, at a height he finds uncomfortable, and which is mechanically inefficient."
An opponent who is unequal to this pressure will normally react in one of two ways. Firstly, he can move his feet further and further back to relieve the discomfort, as in the figure below:
The figure above depicts the limb configurations of a player packed into a scrum with his hip and knee angles both at 90°. (For the sake of illustration I have assumed that the player is 1850mm tall with trunk, thigh and shank lengths of 650mm, 460mm and 480mm respectively.) In order to compare the 90-90 configuration with that advocated by some of the experts listed above, the figure below shows how the body position of the player would change if he retained the 90° hip angle but increased his knee angle to 110°.
Muscles generate most force in the mid range between full extension and full flexion. From a starting point of 90-90 the leg extensors typically remain operating within that efficient range even when the pack achieves a significant shunt forward. Figure 8 illustrates how joint angles change following a push forward of 300mm. As Greenwood suggests, a "forward movement of about a foot ... may well be seen as a maximum" without repositioning of the feet. As can be seen both joint angles have extended to 138°, but this still leaves the players in a position to continue their forward momentum if necessary. Note that both the trunk and shanks have dropped 6° below the horizontal.