Wednesday, August 23, 2006

The Laws of rugby

Rugby is or should be a simple game yet it has 150 pages of laws not including variations and regulations. One would think that with that much detailed information it would be a straightforward matter to know if play is valid or not. The laws are written in english and the meaning of the language is clear yet many referees allow play contrary to the wording of the laws and no-one complains.
Consider the lineout. (law 19)
A lineout player must not lift a team mate. Yet jumpers can rise as much as 1.5m off the ground (higher than an olympic high jumper). The fiction is that the catcher jumps and the team mate supports him once he is in the air.
A player must not jump or support a player before the ball has left the throwers hands (i)
So why do we see players up waiting for the ball? It takes longer to jump to the height of the ball than it does to throw it to the front of the line. This law is not physically possible to comply with!
A player must not pre grip below the waist. Most pre gripping is on the shorts below the buttocks yet no penalties.
Why not dump 19 (g-k) and allow lifting above the knees?
19.5 The thrower must not step into the field of play when the ball is thrown; so why does no one care?
Tackle ruck and maul
law 15 definition
A tackle is when a ball carrier is held and brought to ground. The person who put him there is only a tackler if he too goes to ground! Just to make 15.4 meaningful because the tackler must release, get up or move away before playing the ball.
An opponent on his feet is not supported by the ground or any player on the ground. What is bridging? Why do so many kneel on the tackled player to reach the ball?
15.5 (b,c,d) a tackled player may release the ball by passing or placing the ball. 15.7(a) No player may prevent a tackled player from passing the ball, yet almost always an attempt to take the ball while the grounded player is trying to get rid of it is rewarded by the call holding on,
because 15.5(e), 15.6(b) any player on his feet may take the ball from the tackled player.
That is you can't stop the player laying the ball down, but you can take it from him before he can lay it down.
Offside at a ruck is the hindmost foot on your own side. A player not joining the ruck must retire behind the off-side line. What are those players on the side in possession who stand to the side of the ruck half way up and never get penalised while the defending side gets caught whenever the cross the line by a few centimetres? It is blatant off-side and obstruction at the same time.
What about the forward pass? nuff said
How can we simplify things? many of the laws are meaningless but we all know how to play, if badly, so why not reduce the laws to principles and leave it to the referee. Once you get on the field he/she makes the rules anyway.

The Lineout

The Lineout
The lineout in rugby union is unique. While other ball sports have a method of returning the ball to play after it has crossed the side line to leave the field of play and this is often by a throw (or hit) from the sideline, only rugby has a closely regulated method where several players compete for possession. In many other games the ball may be propelled in any direction into the field of play, aimed directly for a fellow player and opponents usually compete on a one-to-one basis. For example, the soccer or basketball throw in or the hockey hit.
In rugby, if the ball is not thrown in immediately, before a lineout has formed, the ball must be propelled at right angles to the sideline and must travel at least 5m before it is touched by any player. The players contesting the lineout must include a minimum of two players from each team and a maximum number decided by the throwing team. The defending team may have no more players than the throwing team. The two lines of players must stand half a metre from the centre line of the lineout and may not close that gap until the ball is thrown. The ball must be thrown down the centre line. Apart from one player defending the thrower from within 5m of the touch line, and one receiver who, initially at least, stands out of the line, all other players must be at least 10m back from their side of the lineout.
According to the rules of the game, a player may not jump or be supported by a colleague until the ball is thrown, the thrower may not baulk the throw, a player may not be lifted, a player may not be supported before he has jumped and there may be no pre-gripping below the waist. In reality all these rules are ignored and, if the breach is not blatant or no disadvantage occurs to either team, no penalty is imposed.
Where the ball is thrown, how this intention is conveyed to the players and concealed from the opposition and what is subsequently done with the ball are tactical decisions that affect the success of the lineout.
All of the above rules and tactics determine how the ball must be thrown and caught, the physics and biomechanics of throwing and jumping determine whether this occurs successfully.

The throw

There are a number of variations where the ball is passed underarm to the player at the front of the lineout or thrown quickly to that player before he has jumped, but in general the thrower uses an overarm throw to a player who has jumped and been lifted/supported as high as possible. The thrown ball thus be considered as a projectile whose relative release height is the difference between the heights of the thrower’s hands and the catcher’s hands. For this type of throw the distances fall into one of three groups: Front (about 6m from the thrower), middle (8 – 10m) and back (10 – 12m). Any throw over the back of the lineout does not normally include a jump and is not included in this analysis. The thrower may change the angle of release and the speed of release, but the height of release is relatively constant. This analysis will attempt to determine how much choice the thrower has over these variables, and how precisely they must be chosen to give an accurate throw.
In initial calculations the drag on the ball is assumed to be zero and the ball behaves as an ideal projectile. In actual practice the drag at the speeds used is as much as 25% of the weight of the ball, decreasing with the square of the speed of the ball. To make the calculations, a spreadsheet was prepared using the kinematic equations for constant acceleration (assuming no drag etc.), and allowing for the initial conditions to be set for each set of circumstances. The results were made conditional on the ball reaching the catcher at the peak of the jump yet clearing an opponent at the same height in front of the catcher.

Front of lineout.
For a throw to reach a standing player (no relative release height) the ball must be thrown at at least 7.7m/s, any less and it will drop below the player’s hands. To reach a lifted jumper (relative release height of -2.0m) the throw must be at greater than 9m/s, at which speed the release angle is 54o, i.e. a lob. For a flat throw (release angle less than 30o) the speed needs to be at least 12m/s for a lifted jumper. (note that a “flat” throw is not physically possible as the ball will always accelerate down as a result of gravity).
Let us consider a throw to a lifted jumper at the front:
Horizontal distance 6.0m, relative release height -2.0m, speed of release 10m/s and angle of release 40o.
For these values the horizontal velocity component (v.cos40) is 7.7m/s and the vertical component (v.sin40) is 6.4m/s.
The time to reach the jumper is time = distance/horizontal speed = 0.78s
In a series of full power jumps by elite junior rugby players during practice lineouts at a range of distances the time for the jump was 1.1s from initiating the jump to the peak height or 0.9s from the time that the feet left the ground. On this basis it would not be possible to complete a legal jump in these conditions. With at least 9m/s and an angle greater than 54o the time of flight is equal to or greater than the time to jump, but the throw would be a high lob which would give the opposition time to react and disrupt the possession.
At 10m/s and 40o the height of the ball at 6.0m from the thrower is 4.0m. An error of 3o in the angle of release, either too large or too small means that the ball is 30cm higher or lower than intended showing the accuracy demanded of the thrower. Similarly, 0.5m/s error in speed of release makes a difference of 30cm at the distance of 6.0m.

Middle of the lineout
Using the middle of the line out as 10m from the thrower (we can use any distance and adjust accordingly) we find that the minimum usable throw is 10.9m/s at an angle of 50o. With these figures the ball actually reaches its highest point as it crosses the front of the lineout and drops towards the jumper from then on, remaining above any jumper closer to the thrower. Only if the target jumper misses the ball does it become available to opponents behind who are not able to out-jump the throwing side. A 5% error in throwing speed in this case will leave the ball within jumping height up to 80cm in front of the target, or will clear the target by 80cm and will not reach catching height until 1.0m behind the target. The same error in release angle will allow the ball to be caught up to 40cm in front of the target or anywhere behind him, suggesting that at this distance at least, the speed with which the ball is thrown is more important than the angle of release for relatively small errors.

Back of the lineout
The minimum initial conditions for the ball to be thrown 15m are a release speed of 12.95m/s and a minimum release angle of 49o. In these circumstances an error of 5o will make the ball available to opposing jumpers within 30cm in front of the target. A 5% error in speed will make the ball accessible to jumpers up to 1.4m in front of the target or will clear the target by 1.0m and will not be catchable by a jumper until 1.5m behind the target or by a standing player until about 3m behind the target. Again the speed of release has a much smaller tolerance than the angle of release.
For each distance there is a range of combinations of angle and speed which will allow the ball to reach the target at the correct height. A lower angle and higher speed will reduce the time for opponents to react, but will make it easier for them to reach the ball if they do react. A greater angle will make the ball clear the opposition, but will require a greater speed than optimal making the ball travel high and take longer in flight allowing the opponents to challenge the target jumper. The optimal throw will clear the opposition at the smallest possible angle and least possible flight time.
As well as accuracy in release angle it is necessary have a horizontal angle that ensures that the ball remains between the two lines of players otherwise the referee will penalise the offending team. This requires a margin of error of less than 3o for a throw to the middle of the line.
A lineout thrower, usually the hooker, must, therefore, be capable of throwing the ball with an accuracy of less than 3o in both the horizontal and vertical axes while varying the vertical angle by up to 20o between throws to different points in the lineout. He must also be able to judge the speed of the ball to within a few cm/s while varying the speed from 8 to 14m/s.
All these calculations assume that the jumper will be precisely as stated, but in reality positions in the line vary as players move to confuse the opposition so that the actual jump is unopposed. This results in the actual position of the jumper being somewhat indeterminate at the time of the throw so that the thrower must allow for a less than optimal jump by dropping the ball lower than indicated by the calculations above. The competition then becomes not one of precision, but one of timing and communication.
The thrower
As we have seen accuracy of the throw depends on the speed and angle of release both in the horizontal and vertical planes. These are determined by the velocity of the hands at release so that these must be moving at right angles to the sideline and along the desired initial trajectory. Using the principle of kinetic linkage (large muscles followed by small muscles) the sequence of events in reverse order is hand motion determined by arm, shoulder, back and if necessary legs to give the required speed. To disguise the destination of the throw, the thrower will normally use the same sequence but with more or less force depending on the length of the throw. In most current teams the thrower is the hooker, but could be any other player and some women’s teams, where strength is an issue, have used a prop to throw especially to the back of the line. Most throwers adopt a two handed throw with one hand behind the ball for power and the other on the side for control. In this format the hands tend to remain on the line of the throw for longer. A few throwers have retained the older ‘bowling’ action using one hand only and a side-on stance.
When a step is taken to increase the force by involving the legs and hips, rotation of the hips and torso results and the thrower must ensure that the upper body is square on to the lineout when the ball is thrown. Small errors of timing in this case can result in the throw being off line. Stepping with the left foot tends to rotate the thrower towards his/her own side while stepping with the right foot has the opposite effect. The main aim in all cases is consistency of action and, therefore, of throw. Mass repetition of each different length of throw is necessary if accuracy is to be achieved.
It has become common for the receiver to move around before the throw is taken to confuse the opposition It is assumed that the thrower can actually aim at a precise point in space into which the jumper will move at the appropriate time. The thrower may find it difficult to avoid focussing on some stationary object in the field of view and thus selecting the throw parameters incorrectly.

The Jump

There are two parts to the basic lineout jump, the initiation by the jumper and the lift and support by the lifters.(The term “lifter” is used throughout despite the fact that lifting is against law 19.9 (g), since the jumper rises above the height which he/she could attain without assistance and is thus lifted). In a well timed jump, the acceleration of the jumper comes from his/her leg drive and the rise is then maintained at a near constant speed by the lifters. We can isolate these parts initially and look at each as a separate event.
The jump can be considered as a standing jump which is an area that has been the subject of some research as it is used by biomechanists as a good indication of the power of the athlete. The jump is initiated by flexing the knees hips and ankles either in a static position or squat, or as the result of a small step which is called a countermovement jump. The countermovement jump makes use of the stretch - shorten cycle which stretches the muscles in the relevant groups before they shorten in the jump. The prestretching of the muscle enables the muscle to exert a greater force during the shortening or concentric phase. It would appear that the transfer from stretch to shorten must occur within less than 100ms to be effective. An extreme form of the countermovement jump is the drop jump where a downward jump initiates the shortening phase. The countermovement or drop jump has been shown to give a greater vertical height than the squat jump where the athlete is equally familiar with the techniques. The depth of countermovement is not a major consideration.
Jumpers can initiate the countermovement by sinking slightly on the spot, by stepping forward to the take-off or taking a small forward jump. The last technique, as a form of drop jump using both legs would seem to be the most effective, subject to effective training of the jumper.
To reach a maximum height the jumper then extends all three joints while the feet are in contact with the ground. Since the height risen corresponds to the increase in gravitational potential energy, the jumper must gain the maximum possible kinetic energy in the take-off phase. This in turn corresponds to the maximum possible work done by the muscles and hence the greatest distance moved and force exerted. While a deeper squat might increase the distance moved, it will decrease the initial force and may result in the energy of the countermovement being dissipated rather than being conserved in the elastic stretch of the tendons and muscles.
In a vertical jump from a stationary start, an athlete might expect to rise 40 – 50cm in good conditions. With an additional reach of up to 50cm by raising the arms, a player might be able to catch a ball as much as one metre above standing height. A lock who is normally one of the taller players might expect to catch a ball at 3m in an unassisted jump. With assistance the jumper can reach about 1.8m above the height of the lifter with outstretched arms. This will be approximately 2.2m (1.8m height plus 40cm arm reach) giving a total height reached of about 4.0m.
To maximise the height of the lift, the primary lifter who stands behind the jumper should bind as low as possible within the laws (i.e. not below the shorts) which will tend to be most practicably below the buttocks. In optimum conditions the primary lifter will be tall and will end up erect and on his/her toes at full stretch and will have ended up below and just behind the centre of mass of the jumper. The secondary lifter in front of the jumper provides a lesser lifting force acting more to control and support the lift. The front lifter binds on the thighs of the jumper (commonly now on strapping above the knee designed to give grip) and thus does not need to be as tall as the primary lifter to reach the same height in the lift.
Quantitative calculations based on sample jumpers.
The centre of mass of a standing player is about 1.2m above the ground and this is raised to 1.7m in an unassisted jump and 2.8m in an assisted jump. In preparation for the jump he will sink about 20 – 30cm, either in countermovement to tension the muscles or in the preparatory squat to increase the distance over which the force is applied, so that the total rise in the jump alone is about 75cm.
Thus the work done by the jumper in rising out of the squat (assuming a mass of 100kg) is:
In addition the jumper gains a take-off speed which is typically about
2.5 ms-1 representing a kinetic energy of:313J
or a total amount of work done in the jump of 1050J
Some of this energy will come from conservation of elastic energy in the muscles and tendons during the countermovement. The work of the rise must be done while the feet are in contact with the ground which lasts for about 0.3s.
The power generated by the jumper is thus: 3500W
Taking only the rise above the ground gives work done of 700J, a take-off speed of 2.5m/s and a power generated during the spring phase of 2300W.
If we consider the total energy to be generated over the whole movement, both down and spring, the time to generate the 700J is extended by a factor of about twice and the power generation is about 1200W
In the assisted jump the centre of mass rises 1.6m making the work done: 1570J

or a total of 1900J including kinetic energy
In this case the lift lasts nearer to 0.6s from the lifters making contact with the jumper, or 0.9s from the start of the jump, to the top of the lift if completed at the speed of take-off.

Power in total = 2100W (approximately the same as the power of the jumper during the spring phase)

The lifting portion provides the remaining energy in reaching the high point so that the work done by the lifters is: 1570 – 300 = 1270J which is done in 0.6s giving a net power for the lifters of : 2100W (shared between the lifters).

We can calculate the approximate forces generated during the jump by relating the energy gained to the work done. In jumping, the ankles extend so that the total distance moved while in contact with the ground is extended from 25cm to 30 cm. Usin W=F.d we get a force of 3500N

This is net force of approximately three and one half times the body weight and is in addition to the support of the body so that the total force exerted by the muscles of the leg will be approximately four and one half times the body weight.
The lifters on the other hand do not accelerate the jumper so that the net force exerted between them is approximately the body weight of the jumper or 980N with the rear lifter supporting a larger fraction than the front lifter.
A number of lineouts by elite male and elite female players were studied using 50 field per second video and videopoint analysis software to gain approximate values for speed and height. (in live practice and games, accurate calibration is not possible). The data showed remarkably little difference between male and female jumps with the speed of the lift varying between 2.0 ms-1 and 2.7ms-1 both within a forward pack and between packs, the highest speed being for an international female lock with international props. The two factors that most affected the speed of the lift appeared to be speed of jumper (without assistance), and the timing of the lift. In most jumps the lifters were able to maintain the speed of the jump by applying their force at the time when the jumper left the ground. If the lift was delayed until the jumper was off the ground, the speed began to drop and the lift was slower or the lifters had to apply a greater force to accelerate the jumper and thus expend more energy.
While the male lifters were substantially heavier than their female counterparts, and could thus be expected to exert much greater lifting forces, the male jumpers were heavier than their female counterparts by a greater amount, explaining the similarity of their jumping patterns.
( a coaches perspective can be obtained at

Thursday, July 27, 2006

More forward pass.
There has been debate over several years as to what actually constitutes a forward pass under the rules and as interpreted by the referees. There appear to be two main camps 1) those who say the ball must not travel forward (towards the line) and 2) those who say it must not be thrown (ie aimed) forward.
The law allows for both interpretations depending on whether we focus on the meaning of "throw or pass", or we focus on the definition of "forward".
It may seem as an argument in semantics, but definition (1) would preclude the flat attacking backline that has become popular with many teams unless the players stood very close to one another and pass at high speed. It would, however, make for ease of interpretation as it defines forward with respect to a fixed reference, the gound and its markings.
Some extreme examples have been given of the results this interpretation gives eg if a player running at high speed throws the ball straight back over his head, but slower than his running speed it travels forward, but can it really be called a forward pass?
Definition (2) can also be interpreted relatively easily when both players continue to run and there are no field markings to distract the judgement. In this case the line of advantage remains the ball and all that is required is that the line of advantage remains behind the passer. The difficulty arises when the passer is stopped eg by a tackle and the ball then clearly travels ahead of the passer. The travel of the ball is no different, but the perception has changed and it is difficult to judge how the ball would have travelled relative to the passer. Even if there is no tackle, but the pass is made close to a ground marking there can be a change in perception. In an extreme case consider where the passer imparts a backward component of 1m/s back while running at 8m/s forward, the ball now has a net forward velocity of 7m/s. The receiver is 15m to the side of the passer and the ball is given a sideways velocity of 15m/s (55km/h) so that the pass takes 1.0s to reach the receiver and thus travels forward 7m. Is it a forward pass? No says (2). But what if the pass was made 3m short of the goal line? the receiver would be 4m in goal before catching the ball. Would any referee allow that try?
It would seem there is no simple answer that would still allow play to flow freely without producing blatant anomolies within the spirit of the law. The current practice of allowing some inertial forward travel which does not allow the attacking team to gain an unfair advantage is probably the best we can hope for. A pass to someone in front is always going to be forward and a pass that travels back will always be legitimate. Between those is anyone's guess, but the referee's decision.

Wednesday, July 12, 2006

The forward pass

Last week I presented a workshop at conference and I promised to give a few details of an analysis of the forward pass that I did there. Here it is.

In rugby, the ball is moved by passing from player to player. When the ball is passed it must not travel toward the opposition dead-ball line or it is deemed to be a forward pass and a scrum may be awarded to the opposition (IRB Laws of the game 2005). By analysing video of the IRB seven-a-side tournament in Wellington in 2006 it was possible to see that almost half the passes between players running in open play actually travelled forward yet none of these was penalised. Similar analysis and observation of 15-a-side rugby show that a substantial proportion of passes between running players is similarly forward as defined by the rules. Below is an analysis of one such pass to show why this might have happened.
Using the field markings as reference points the player positions were noted for each frame of video at 1/24th second intervals and speeds calculated accordingly.
At the instant the pass was made the players were running parallel to one another at approximately 8m/s. They were 6m apart laterally and the receiver was 1.0m behind the passer. Between the pass being made and the receiver catching the ball took 0.30s.
The receiver remained 1.0m behind the passer when he caught the ball.
Using these figures we can see that between the release of the ball and the catch, the receiver travelled forward a distance:
d= 8.0 x 0.30 = 2.4m

The passer was initially only 1.0m ahead so that the ball must have travelled 1.4m forward.

The passer was clearly aiming the ball back to the receiver who receives it from in front so how does it travel forward?
The answer lies in Newton’s first law of motion. An object in motion remains in motion unless acted on by an outside force. As the ball is released it is moving forward with the passer at 8.0m/s. He, by throwing backward, gave it an additional 3.3m/s backward so that the net forward speed is now 4.7m/s. The speed across the field is 20 m/s. By vector addition we can show the situation as revealed by the camera or a stationary observer above the field.
The referee and linesman must keep up with the play as much as possible and, therefore relative to either of them, moving at the same speed as the players, the ball appears to travel backward – no infringement. (Rugby World 2005)
Let us now reconsider this pass to see how it could have been made legally. At the limit the ball must not travel forwards so that we shall assume that it travels straight across the field. The passer is travelling forward at 8.0m/s and thus must give the ball a rearward component of 8.0m/s to render it stationary. To reach the receiver before he is in front of the release point the ball must cover the 6.0m of their separation in the time it takes him to travel 1.0m. i.e. 0.125s. The ball must travel across the field at 48 m/s. Combining these results by Pythagoras’ for this one pass we get: 175 km/h for the pass
Clearly the passer will be unable to achieve this and the receiver would have difficulty surviving it.
Using the figures earlier can we determine how the players should be positioned to give a valid pass. We decided that the receiver travelled 2.4m during the pass so it would seem that he should run 2.4m back from the line of the passer. This will not actually work as the passer must aim much farther back. Using the need to have a rearward component of 8.0m/s and a speed of pass of 20m/s(as used above) we get 23 degrees back from the transverse line.
The transverse component of the velocity is therefore 18.33 m/s
At this speed the pass will take 0.327s in which time the receiver will travel 2.62m. He must be at least this far behind to receive a legitimate pass.
The farther the players are apart, the longer it will take for the pass and, therefore, the farther back the receiver will need to be to receive the pass before crossing the advantage line. Alternatively the receiver will need to slow down to avoid crossing the line and thus will lose the advantage of speed built up in the move.
In a recent test match a centre passed to the winger who carried on to score the try. There was no debate amongst either set of supporters that it was a forward pass despite the fact that it was released before the 22 and received after the 22 with a forward separation of 2.2m. Careful study of the video shows that the players had a lateral separation of 13.2m and longitudinal separation of 5.0m at the time the ball was released. The pass took 0.9s and the receiver was travelling at 8.0ms-1. The passer was slowing down at the time of the pass and his speed was approximately 5.5m/s. In this case, therefore, the pass was released at 15.0m/s in a direction which was ahead of the instantaneous position of the receiver (to allow for their speed differential). In actual fact the receiver still had to slow slightly to take the pass. To make the pass legitimate in this case it would have required the ball to be aimed at the receiver at 22.3m/s. This is an increase of about 50% on the actual pass, but it would have been flatter and take only 0.6s rather than 0.9s giving less risk. The key to this pass being possible legitimately is that the receiver was initially 5m behind the passer.

Thursday, June 29, 2006

The mechanics of the scrum in Rugby

In all forms of scrum, loose or set, the forwards contest possession and territory by using strength with technique to move forward or hold ground while moving the ball toward the rear of the pack or to move the opposition back to make it difficult for them to control the ball. The mechanics of pushing and resisting in each form of scrum is similar and good technique in one is transferable to the others. The set scrum, being the most formal, is the most predictable and thus simplest to analyse physically and biomechanically.

The set scrum has two main phases:
Engagement when the two packs are brought together initially from a prepared stationary position determined and controlled by the referee, and
The sustained or secondary shove which is exerted once the ball has been put in and the packs attempt to exert dominance.
The overall position in preparation for engagement is controlled by the referee and should give no advantage to either side as they should react at the same time to commands that are supposed to ensure that the front rows come together in a stable and safe manner. The packs on the other hand are trying to adopt a body position and combination that will give them the greatest speed and hence impulse on the opposition when they come into contact so that they are in the best position for the next phase of the scrum and/or they place the opposition in a weaker position to respond. The rules as interpreted by the referee place strict controls on the means by which the players compete in the interests of safety. A number of the rules correspond to the best scrummaging practice and promote strong body positions which, if applied by both packs will give stability while permitting fair contesting of advantage.
The force exerted by the entire pack is exerted through the front row to the opposition, and by reaction the front row must experience the full force at the points of impact. It is front row players who are most at risk of injury if the engagement, in particular, is incorrect. Since the impulse, and hence the force at impact, depends on the change in momentum rather than the muscular strength, engagement is the time of greatest risk. Thus emphasis is initially placed on the body position of the front row players prior to and during engagement, and on the movement into engagement.
To exert the force onto the opposition the props must adopt a natural back line, slightly hollow at the lumbar back and substantially straight through the upper back and neck. To do this the players look at the opposition, keeping the eyes up but not actually tilting the head back “looking over the top of your glasses”. Tilting the head forward, i.e. looking down, places the neck in the most vulnerable position for possible dislocation if the back of the head is struck on engagement. Looking down also puts a curve into the back which weakens it by producing shear forces rather than compressive forces from in front and behind.
Considering the props alone (though this applies to the hooker in a defensive scrum), they must adopt joint angles that enable the optimum force to be exerted during the move into engagement and that force should be immediate on reaction while allowing the position to remain stable immediately after contact. According to the literature (O’Shea, 2000, p Milburn, 2000, van Heerden 2002 (tug of war)) the optimum angles are: knee 120o (in range 110 - 150), hip 100 (90 – 120).
Video analysis carried out using a Sony DCR TRV240E at 50Hz analysed using “video point” software was carried out video of an international scrum (international competition champions) on a scrum machine, a national representative under age side (world finalists) on a scrum machine and a high school (international competition champions) in live scrummaging practice and on a scrum machine and a leading women’s club side in competition and in practice.
The general results for the international team were as follows:
The front row adopted a position where the shoulder and hips were in a horizontal line (law requires head and shoulders no lower than the hips). The hips were flexed at 80o, the knee was flexed at 120o (outer knee of prop, inner knee not measurable), the foot was plantar flexed and support was on the ball of the foot. The hip angle is slightly below the optimum range, but will extend with the drive.
With the drive into contact both knee and hip extend until at first contact the angles are 95o and 125o respectively and very near the optimum for maximum force production. The feet at all times remained in contact with the ground. The hip extension led the knee extension by about 20ms - 40ms whereas Mills and Robinson() indicate that, in theory, scrummagers with good technique tend to extend simultaneously (the graphs with their paper suggest delay of similar amount in good scrummagers, but a delay of 120ms in poor scrummagers). (The suggested simultaneity was based on the biomechanical push-throw movement for maximum force production). The initial drive carried the joints through to final angles of 105o and 135o respectively, but the shoulders travelled downward into contact.
After contact the settling in preparation for the secondary sustained drive involved a dropping of the hips to bring the angles to 95o and 110o. At all times during the engagement sequence the top grade, experienced, prop maintained joint angles within the optimum range. To date the direction of travel in the third dimension has not been studied and preparations are being made to examine the scrum from vertically above to determine whether lateral movement is substantial.

By contrast the underage international side showed less awareness of the optimum body position. In an identical practice situation the prop had a hip angle of 50o in the crouch position with a knee angle of 105o. In the drive into contact and settling the angles became 90o returning to 70o for the hip and 130o returning to 80o for the knee. Some of the variation resulted from movement of the scrum machine immediately after contact and to stepping after contact. The locks stepped back on contact and the scrum became unsettled until a firm base was re-established.
The High school side when observed in live practice showed a lack of stability which to some degree is explained by the opposition. The initial angles were 70o hip, and 140o knee with little or no ankle flexion so that in the crouch position the hip was above the shoulder and the player was unprepared to engage. On the engage command the prop gained little forward motion from extending the hip and was unable to extend the knee. At the instant of contact a step back was taken to bring the hip to 85o and allowing the knee extension to reach 125o which are nearer the ideals, but this delayed the drive and meant that the scrum was able only to resist the impetus of the opposition. The scrum became unstable and wheeled substantially before settling for the put in. The tendency to adjust position during or just after engagement with a subsequent initial instability of the scrum was noted with almost all the high school front rows observed.
Subsequent study of a women’s scrum has shown similar angles to the boys’ scrum except that the initial hip angle was even smaller (55o), but by maintaining ground contact throughout the drive the hip angle was increased to 105o during the initial contact and settling. The knee was again substantially flexed placing the hips above the shoulders in the crouch position. As the prop moved into engagement the hips extended while the knee initially flexed to bring the hips in line with the shoulders so that on contact and thereafter the joint angles were near optimum. In live scrummaging, both the boys and the women, the timing of the initial drive into engagement was noticeably different for each front row allowing one side to dominate the engagement. Subject to further study of elite players, it would seem that the optimum engagement is obtained when the props are on the balls of the feet, i.e. plantar flexed, with both knee and hip within, but at the lower end of, the optimum range. This then gives maximum acceleration on the “engage” command while leaving the front row stable and ready to shove immediately, leaving the feet in ground contact throughout. In less able scrums the tendency is to sit back by flexing at the hips so that time is taken while the hips extend before the drive proper begins.
Attempts to identify predictors for scrummaging performance within normal testing procedures have shown () that the time for a 30m sprint from standing start gives the closest correlation. Short sprints are a good measure of acceleration and tend to measure the ability to exert force in the forward direction while in a forward leaning position. This more nearly approximates the pushing position than say a vertical jump which also employs countermovement. This would seem to reinforce the advantage of preparing for engagement in a position on the verge of forward over balancing. This position also ensures that the initial movement will involve extension of both hips and knees as indicated by Mills.

The forces of engagement and shove have been measured by the use of force plates, strain gauges etc in the pads of scrum machines and thus indicate the impulsive force on the machine in the former case rather than the accelerating force of the players. We have calculated this impulsive force using video analysis and mass measurements and have found satisfactory agreement with the direct measurements of () giving some confidence in extending this analysis to other aspects of the forces exerted in scrums.
The force in engagement is calculated by measuring the speed of the players immediately before engagement and the time taken to bring them to rest (the scrum must by law come to rest before the ball is put in). From the impulse and change in momentum we calculate the force. To give the greatest impact on the opposition, the player should stop in the shortest time. On impact the shoulder girdle flexes allowing the spine and most of the mass to slow more gradually. To counter this, the prop throws the arm forward into the binding position and making the shoulder more rigid so that the mass stops more abruptly. The stability of the shoulder girdle also promotes stability of the neck.

Studies of the forces on front row players using multi axis force plates by Milburn suggested that the more inexperienced the forwards in the scrum, the more likely it was that there would be instability on contact and during the sustained shove. The lateral or shear forces on the hooker at high school level in particular exceeded the body weight by a substantial amount and could act in opposite directions on engagement and shove even in the controlled situation of a scrum machine. On the other hand the lateral forces experienced in an international standard scrum were shown to be negligible at all stages even though the forward impulse and shove in the international scrum almost doubled that of the school scrum.
While the front row transmits and experiences the full force exerted in the shove the other players contribute substantially to the impulse, especially when they act simultaneously rather than sequentially (McClymont & Cron). As the experience of sides decreases there is a tendency for locks and number 8 to hang back from the engagement so that the contact becomes a series of small impulses rather than one large one. In some scrums it has become a technique for the no. 8 to pull back on the locks until the engage call on the assumption that when released they will accelerate at a greater rate as if from a ‘set’ position in sprinting. It appears to us that this will delay the drive and remove the no. 8 as an effective force.
Simple two dimensional video analysis from the side of the scrum did not allow the lateral forces to be estimated but observation of live scrums show that there are substantial twisting moments in both horizontal and vertical planes on first engagement as a result of the inconsistent techniques of the front row players on each side.
(To be continued)