Executive Summary

An evaluation of the effect of the experimental law variations (ELVs) on match activities and match injuries resulting in player replacement was undertaken. Seventeen ELVs were trialled in the 2008 Air New Zealand Cup (ANZC) competition. Data were obtained from all ANZC matches from 2000 to 2008 (522 matches in all; 77 were played under the ELVs). Looking at match information from 2000 to 2008 enabled the effect of the ELVs to be distinguished from any underlying trends in match activities that may have been occurring. Information about injuries was obtained for the 2003 to 2005 seasons (pre-ELVs) and compared to the 2008 season.

Overall, the ELVs had little impact on points scoring, although the rate of penalty goals kicked per match showed a moderate decrease. There was a small increase in the proportion of tries scored from tap penalties and free kicks, and a small decrease in the proportion scored from lineout possession.

The ELVs were associated with a moderate decrease in lineouts per match. The proportions of full lineouts and quick throw-ins both showed increases associated with the ELVs, and the proportion of short lineouts in the ANZC decreased in 2008. The percentage of lineouts won by the team throwing the ball in did not change appreciably.

An ELV relating to the scrum engagement was trialled in the ANZC in 2008. The ELV changed the “Crouch”-“Touch”-“Pause”-“Engage” sequence by eliminating the verbal command “Pause”. Scrum numbers decreased slightly as a result of the ELV. The number of collapsed scrums per match was lower in 2007 (under C-T-P-E) and 2008 (under C-T-E) than high levels observed in 2005 and 2006, but the number remained higher than had been the case in 2000 through 2004.

There was a moderate increase in the number of kicks during play under the ELVs, up to 64 per match in 2008 from an average of 49 per match from 2000 to 2007. There was a substantial decrease in the number of mauls per match from 2000 to 2008; the ELVs were associated with a further reduction in maul numbers, which fell to 10 per match in 2008. There was a small increase in the number of rucks (post-tackle breakdowns) per match associated with the introduction of the ELVs. There was an average of 160 rucks per match in ANZC rugby in 2008 – ball retention at ruck time varied little from 2000 to 2008, and was between 91% and 94% over the entire period. There were an average of 15 free-kicks and 11 penalties awarded per match in 2008; the overall number of sanctions awarded increased by a small amount from 2007, but was similar to that seen from 2001 to 2006.

There was a moderate increase in the number of tackle injury replacements per 1000 player hours in 2008 when compared to the rate from 2003 to 2005. Although the increase in risk to each player per match is low (their chance of being replaced due to a tackle injury has increased from about 1.3% per match to about 3.2% per match), over the entire playing population such an increase would probably carry noticeable increases in costs to playing squads, and, for coaches, have an impact on player availability.

Other aspects of the sport (scrums, rucks, tackles, collisions) showed trivial to small increases in injury replacement rates. Given the low underlying rates, the changes did not represent any real concerns from a risk management perspective. The rate of injury replacements from mauls was so low that no clear effect could be determined.

The current trials of the experimental law variations have been conducted in part to examine what the actual, versus the intended, effects of the law changes has been. From an administrative and risk management perspective, such an approach is laudable, because the analyses associated with the trials permit decisions to be made with at least some degree of evidence about what the likely effects of the laws on match activities and player safety will be. Even so, it would be prudent to continue evaluating their effects for some time, as further changes may emerge in the future as coaches and tacticians adapt their strategies and tactics to both the way that match officials are administering the laws and the strategies and tactics other teams are adopting.

Introduction

Rugby Union (rugby) has a long history of changes to the laws of the sport ¹³. The patterns of match activity have also changed substantially over time ^{6, 8}. Explanations for the changes in match activities have attributed them in part to the law changes, in part to the changes in application of the existing laws, and in part to developments in team strategies and tactics ^{6, 8, 13}.

The IRB Experimental Law Variations (ELVs) arose as the result of a review of the Laws of the Game in 2004. The ELVs are a set of modifications to the laws of rugby union commissioned by the Laws Project Group of the IRB. A full background to the ELVs can be downloaded from the following link:

IRB Guide to Experimental Law Variations, August 2008

The objectives of the ELVs, as explained by the IRB to member unions were that:

• The game is easier to play, referee and understand
• Simpler to understand by spectators
• More enjoyable for all
• Players determine the outcome of the game
• Contest and continuity maintaining the integrity of the game
• Rugby as a sport for all shapes and sizes
• Players participating at their skill level
• A balance between participant enjoyment and spectator entertainment

Evaluation of the extent to which the ELVs have achieved the goals listed above is outside the scope of this report. The purpose of the current report is to describe changes to match activities and on-field injury risks attributable to the ELVs, and where possible, to place the sizes of the changes in context with respect to their importance. Hopefully, this will help administrators, coaches, and others with an interest in the outcomes of the ELVs to assess the extent to which the ELVs have fulfilled the subjective criteria listed.

A number of statistical reports of changes in match activities following the introduction of the ELVs have already been provided to the IRB. Most of these have focussed on the differences between matches played in the year prior to the introduction of the ELVs and matches played in the season in which the ELVs were introduced. Changes in match activities observed from one season to the next can help provide a gauge as to the effect of the ELVs, but they are hampered by the fact that inferences drawn about changes fail to account for any underlying trends that may have been occurring in match activities over a longer period. We sought to overcome this issue by tracking changes in match activities in Air New Zealand Cup (ANZC) rugby from 2000 to 2008. The methods used to obtain and analyse the information reported are provided in a technical note at the end of the document.

Summary of the Experimental Law Variations in the 2008 ANZC

There were 17 ELVs trialled in the 2008 ANZC. The ELVs are listed below:

Results

Section 1. Points Scoring

From 2000 to 2008 there was an average of 50 points scored per match in Air New Zealand Cup
(ANZC) matches. The average difference in score between the winning and losing team was 16;
33 points were scored on average by the winning team and 17 by the losing team. Given the
match to match variability in scores, the changes from year to year are trivial to small. The
introduction of the ELVs was associated with a trivial decrease (-6%; 90% CI ± 8%) in scoring.

Table 1. Average points scored per match by year*

*Scores are presented as the mean ± one standard deviation

Tries scored

On average, there have been between five and seven tries per match in Air New Zealand Cup (ANZC) fixtures over the period 2000 to 2008. Any impact the ELVs may have had on rate of try scoring has been minimal. Tries did, however, make up a slightly higher percentage of all points scored in 2008 than they had in previous years.

Table 2. Average tries scored per match by year

Possession source and try scoring

The percentage of tries scored from various sources of possession has varied over the period. A team’s own possession at set piece (scrum and lineout) has typically been the platform for 40% to 50% of all tries. The percentage from lineouts trended upward until 2005 and decreased thereafter. The decrease accelerated in 2008 under the ELVs, with tries resulting from own-team ball at lineout dropping under 20% for the first time. The percentage of tries resulting from ownteam scrum ball has ranged from 16% to 23%, but has not shown a consistent trend up or down. The most noticeable increase in percentage of tries from a possession source following the introduction of the ELVs has from tap penalties and free kicks – up to 15% in 2008 from an average of 7% from 2000 to 2007.

Table 3. Percentage of tries from possession source by year

Origin of tries

Table 4. Percentage of tries by where possession was obtained

Possession was obtained between the attacking 22 and the goal line on 38% of try-scoring occasions, with a further 34% coming from possession obtained between the halfway line and the attacking 22. Possession was obtained in the defending 22 on only 4% of try-scoring occasions. There was little variation from year to year in the percentage of tries scored by region of the field.

Timing of tries

The greatest proportion of tries (9.5%) was scored in the last five minutes of matches, and the least (3.2%) in the first five minutes.

Figure 1. Tries by time period in ANZC Rugby – 2000 to 2008

The pattern in 2008 under the ELVs showed the same pattern of try scoring by time period as the previous years.

Phases and players per try

First phase ball refers to possession obtained directly from a start of play (such as a kick-off or set piece) or from a turnover. Second and higher phase ball is dependent upon how many rucks/mauls have taken place since possession was gained. Approximately 80% of tries are scored in four phases or fewer. The introduction of the ELVs appears to have had little effect on the number of phases per try.

Table 5. Percentage of tries by number of phases per year

A similar pattern was apparent when the number of players who handled the ball in the possession leading to the try is examined. In 42% of tries between one and five players handled the ball; in 30% of cases six to ten players handled the ball. Once again, the introduction of the ELVs had little effect.

Penalty goals and drop goals

Over the nine years, 69% of penalties were successful. The average number of penalty goal attempts per match has been trending downward over time. Taking the linear trend of a 1.9% decrease per year into account, the introduction of the ELVs was associated with a 30% decrease. The introduction of the ELVS was associated with a moderate decrease (-36%; 90% CI ± 9%) on penalty against an essentially static (-0.2% per year) yearly trend.

Table 6. Average penalty goals and drop goals per match by year

The distribution of penalties is shown below. The first graphic represents the number of penalty attempts in five metre blocks around the field (the two red dots at the top of the picture represent the goal posts); the second represents the percentage of successful attempts from each area.

Figure 2. Penalty attempts by field position in ANZC Rugby – 2000 to 2008

Figure 3. Penalty percentage by field position in ANZC Rugby – 2000 to 2008

The average distance for penalty shots was 31 metres (± 12 metres). This showed no real change over the period, and did not appear to be affected by the introduction of the ELVs. The percentage of successful penalty shots ranged from a low of 63% in 2000 to a high of 76% in 2004. In 2008 68% of penalty shots were successful.

Drop goals do not feature prominently in ANZC rugby – there is on average one successful drop goal every 10 matches. The numbers per year are too low to make examining trends over the period worthwhile.

Conversions

An attempt is made to convert almost every try in the ANZC. Table 7 shows the typical number of conversions and percentage success in matches per year.

Table 7. Average conversions per match by year

The angle of the kick is directly related to the probability of success – kickers struggle with conversion attempts near the sideline. This is illustrated graphically in the pictures below.

Figure 4. Conversion attempts by field position in ANZC Rugby – 2000 to 2008

Figure 5. Conversion percentage by field position in ANZC Rugby – 2000 to 2008

Section 2. Match Times and Player Participation Times

Elapsed time and ball in play time

Although total match time has remained at 80 minutes throughout the period, the total running time has increased. The running time exceeds 80 minutes mainly because of injury stoppages, substitutions and references to the Television Match Official. The ball-in-play time had shown a slight decrease (-4.5%) over the period, which was reversed under the introduction of the ELVs (+8.1%) – the average ball-in-play time in 2008 was 3 minutes and 6 seconds longer than in 2007. It appears that the running time for professional matches has continued to trend upward (at least in ANZC rugby).

Table 8. Match, ball in play and player participation time

Player participation and substitutions

Player participation time, which dropped rapidly following the introduction of non-injury substitutes, increased slightly from 2000 to 2004, and has decreased by an average of three minutes per player per match since. The number of substitutes used per match is now almost invariably the full complement of seven from each team. In only three matches since 2000 were seven from each team not used – one match in 2000, and two in 2002.

Section 3. Set Piece

Lineouts

Since ‘supporting’ of the lineout jumper was permitted in the late 1990’s, the de facto result has been lifting in the lineout, with ‘supporting players’ effectively gripping jumpers around the thighs and lifting them as they jump. Thus the final two ELVs related to the lineout (listed above) have been changes in wording of the law, rather than a substantial change in practice.

The number of lineouts per match had remained essentially static at approximately 30 ± 5.2 from 2000 to 2007. The introduction of the ELVs was associated with a moderate decrease (-12.5%; ± 4.1%) in the number of lineouts per match, down to 26 ± 5.5 in 2008.

Table 9. Lineouts per match and average metres gained

A small number of lineouts do not result in either team obtaining immediate possession from which to attack (e.g. when there is an infringement at the lineout) – hence the lineout possessions per match are slightly lower than the number of lineouts per match in the table above. As noted in the section on points scoring, lineouts are an important means of obtaining possession that leads to try scoring – between 23% and 28% of tries per year have been scored from possession originally obtained at the lineout. Teams can typically expect to gain about 15 metres of territory when they obtain lineout ball (either from their own ball or by stealing opposition ball), but as can be seen from the large standard deviations around the metres gained in the table, there is a great deal of variability in how much territory they make from one lineout to the next.

Verusco have coded extra information about the structure of the lineout since 2005. The proportions of quick lineouts and full lineouts both increased in 2008 under the ELVs, at the expense of shortened lineouts, which dropped from an average of 37% of all lineouts from 2005 to 2007, down to 11% in 2008.

Figure 6. Lineout types by year – ANZC Rugby 2005 to 2008

The percentage of lineouts won by the team throwing ranged from 78% to 83% per year for both full and shortened lineouts. The percentage of quick throw-ins won by the team throwing in is higher – it ranged from 91% in 2005 up to 97% in 2008. Overall the percentage of own lineouts/throw-ins won was 82% in 2005, and has been 79% from 2006 to 2008.

Scrums

There were three ELVs related to the scrum in ANZC rugby in 2008. These were the Introduction of an offside line 5 metres behind the hindmost feet of the scrum, the identification of scrum half offside lines and a modification of the “Crouch”-“Touch”-“Pause”-“Engage” scrum engagement sequence that saw the call “Pause” replaced by a natural pause without a call. The number of scrums set per match from 2000 to 2008 is shown below.

Table 10. Scrum information

There was a small decrease (-4.8%; ± 4.4%) in the number of scrums set associated with the introduction of the ELVs when the linear trend over the period was accounted for (-3.4%; ±4.7%). There was also a small decrease (-20%; 8.1%) in the number of reset scrums per match following the introduction of the C-T-P-E sequence in 2007 when compared with the number per match from 2000 to 2006. There was a further 13% decrease in scrum resets per match in 2008. Although the off-side line at the scrum has moved backward by 5m under the ELVs, there has been no apparent increase in the average metres gained from scrum possession to end of the first phase (up until a ruck, maul, change of possession or stoppage in play). The team feeding the scrum generally wins possession – the rate of scrums won ranged from 94% to 97%, and did not change appreciably with the introduction of either the C-T-P-E or the C-T-E sequence.

The average number of collapsed scrums throughout the entire period (2000 to 2008) was 4.1 ± 2.5. The linear trend over the period showed a moderate increase (52%; ± 20%) – as can be seen from graph below there was a large percentage increase (43%) in scrum collapses in 2005 compared to the previous years – although in terms of numbers this represents an increase from 3.5 to 5 collapses per match. There was no specific law change to which this change can be attributed. Anecdotally, changes in scrum set-up and engagement techniques involving teams moving into a ‘semi-crouch’ prior to engagement occurred in 2005. Changes in jersey technology, which have had the effect of tightening the fabric and perhaps making it more difficult for frontrow players to get a grip on the opponent, may also have played a role; other (unknown) factors may have also contributed to the observed increase. Following the introduction of the C-T-P-E sequence in 2007 the number of collapses per match decreased in comparison to 2005 and 2006, but remained higher than it had been from 2000 through to 2004. There was a decrease of 17% (90% CI ± 10.2%) in collapsed scrum numbers attributable to the introduction of the C-TP-E sequence, and a further 4% decrease associated with the ELV trial in 2008. As can be seen in the graph below, given the variability in scrum collapse numbers from match to match this represents no real change. Note that not all collapses result in scrum resets – on some occasions the referee permits play to continue after the scrum collapses.

Figure 7. Scrum collapses per match by year – ANZC rugby 2000 to 2008

Phase play, kicking and tackles

Table 11. Phase play, kicks in play, passes and tackles

The introduction of the ELVs in ANZC rugby, was associated with a small (5.7%; ± 2.1%) increase in the number of rucks per match given the essentially static underlying trend of an increase of 0.3% per year. As can be seen from the table, however, there has been substantial year-to-year variation in ruck numbers. Ball retention at ruck time varied little from year to year, but was slightly lower under the ELVs than it had been in any year since 2001.

Mauls

An ELV under Law 17 changed the options available to teams for defending a maul. In 2008, teams were able to defend a maul by pulling it down and the existing reference to the head and shoulders not being lower than the hips was removed. The number of mauls per match has demonstrated a large decrease over the period (-50%; ± 2.9%). The introduction of the ELVs was associated with a moderate decrease in mauls beyond that which was already occurring (-32%; ±4.8%).

Passes

As was the case with the rucks, there is considerable variation in the number of passes – both from match to match and from year to year. In the 2000 and 2003 series the number of passes per match was higher than that observed under the ELVs in 2008; for the remaining series (2001-2002 and 2004-2007) the number of passes per match was lower. In terms of trend, the number of passes per match has decreased by a moderate amount (10.3%; 90% CI ± 1.5%) from 2000 to 2008. This trend was reversed under the ELVs, which were associated with a small to moderate increase (9.4%; 90% CI ± 1.7%) in the average number of passes per match.

Kicks during play

The average number of kicks in play per match from 2000 to 2007 was 49 – the effect of the ELVs was to increase this to 64 kicks per match. In terms of the trends, from 2000 to 2008 there was a small (9.7%; ± 3.4%) decrease in the average number of kicks per match; there was a moderate increase (37%; 90% CI ± 4.5%) in kicks in play associated with the introduction of the ELVS in ANZC rugby. The number of kicks per match has decreased over a longer time period – it was around 66 per match in Bledisloe Cup rugby in 1995.

Tackles

The ELVs were not associated with any noticeable change in the number of tackles per match. Although the average number of tackles per match varied from a low of 312 in 2001 to a high of 361 in 2003, over the series the increases and decreases have balanced out, meaning that the number of tackles per match in 2008 was essentially identical to that in 2000.

Section 4. Law infringements

Until the 2008 season, Verusco did not distinguish free kicks from penalties in their coding template, so we are unable to assess the change in the relative numbers of penalties and free kicks over the period. We can, however, examine the change in the total number of penalties and free kicks (grouped together) associated with the introduction of the ELVs. The number of penalties and free kicks awarded showed a small increase (9.3%; 90% CI ± 5.4%) in 2008, against a background trend over the period of a trivial decrease of -2.8% (90% CI ± 5.2%).

Table 12. Penalties and free kicks

In 2008, there was an average of 15 ± 3.6 free kicks per match, and 11 ± 3.2 penalties. Of the penalties, 22% were offences elevated from the sanction of a free kick to a penalty by the referee. These were most often repeated infringements at the breakdown.

Yellow cards and red cards

There were 10 red cards awarded over the 2000 to 2008 ANZC competitions, two of which were in one match in 2000. The number of yellow cards awarded showed an increase of 114% (90% CI ± 105%) from 2000 to 2008. The effect of the ELVs on the rate of yellow cards awarded given the underlying trend was unclear.

Table 13. Yellow Cards awarded

Section 5. Injuries

Table 14. Injury replacements per 1000 hours by cause*

*Mean ± standard deviation

Tackle injury replacements

Tackle replacements per 1000 player hours were 150% higher in 2008 than for the pre-ELVs seasons. Given the match to match variability in rates this represents a moderate increase in replacement rate, and is an aspect of the ELVs that warrants careful evaluation in terms of overall player welfare. Having said that, there were two years (2006 and 2007) in which tackle injury data were not collected, so the possibility that factors other than the ELVs may have played a part in bringing about the observed change cannot be ruled out. Although the increase in risk to each player per match is low (they would probably not notice a difference – their chance of being replaced due to a tackle injury has increased from about 1.3% per match to about 3.2% per match), over the entire playing population such an increase would probably carry noticeable increases in costs to playing squads, and, for coaches, have an impact on player availability. The event preceding tackle injuries was also examined. The tackle injury replacement rates following rucks, open play and scrums showed small increases in 2008 compared with the grouped rates from 2003 to 2005.

Figure 8. Tackle replacements by positional groups

The rate of tackle replacements for players grouped into positions of forwards, inside backs, midfield backs and outside backs is shown in the chart above. All of the positional groups showed a small increase in tackle replacement rate in 2008 compared with the rate over the 2003 to 2005 period. The primary thing to draw from the chart, however, is how much variability there is in replacement rate from match to match – this is illustrated by the error bars, which show one standard deviation in the positive direction. The percentage increase for each group is large, but the per player rates for each match are low – once again this is something that would benefit from a watching brief at administrative level, but the change in rate is unlikely to be noticed by individual players.

Ruck, scrum, collision and maul injury replacements

There was a small increase in the rate of player replacements associated with ruck injuries, and trivial increases in the rates associated with scrums and collisions. There were so few maul injuries that any change was unclear; regardless of whether there was a small increase or a small decrease the overall rate of maul injury replacements was negligible – it equates to one maul related injury replacement for each 78 matches played.

Discussion

As noted in the introduction, rugby has a long history of changes both to the laws and to the relative numbers of activities that comprise a match. The characteristics of particular activities, such as scrums, lineouts and rucks have also changed markedly over time. One factor that has been noted elsewhere is the impact of the unintended consequences of law changes to activities in rugby ^{7, 8}. Rugby demonstrates some of the characteristics of a dynamic system ¹⁰, and as such a change made to one area of the sport can result in substantial changes to other aspects of the sport. An example of the law of unintended consequences in rugby relates to changes made to the off-side laws at the scrum in the early 1960’s and subsequent development of power scrummaging followed by an increased rate of scrum-related spinal injuries documented in the 1970’s and 1980’s ^{8, 11}. A more recent example involved changes to maul laws (the use it or lose it rule) that, along with an increased emphasis on defence following the introduction of professionalism, resulted in a rapid increase in the number of tackles and rucks (although technically many of the post-tackle situations were not rucks as defined in the laws of the game) in the mid 1990s.

The current trials of the experimental law variations have been conducted in part to examine what the actual, versus the intended, effects of the law changes has been. From an administrative and risk management perspective ^{1, 2}, such an approach is laudable, because the analyses associated with the trials permit decisions to be made with at least some degree of evidence about what the likely effects of the laws on match activities and player safety will be. Even so, it would be prudent to continue evaluating their effects for some time, as further changes may emerge in the future as coaches and tacticians adapt their strategies and tactics to both the way that match officials are administering the laws and the strategies and tactics other teams are adopting.

A couple of changes reported above are worth highlighting in that they appear to be unintended effects of the set of ELVs trialled in the Air New Zealand Cup. The first is that the number of kicks in play has increased. Part of this increase is related to the change regarding teams being able to put the ball back into their own 22 and then kick directly into touch. The extra kicks are, however, an option teams have chosen to take - it not mandated that they kick the ball. The choice to kick may be a reflection of teams seeking field position, and thus kicking to a point outside the opposition 22. The opposition, who no longer have the option of passing back inside the 22 and kicking directly to touch (and are looking to move into opposition territory in order to generate pressure and launch attacking plays) often repeat the previous play, and kick the ball back.

The second is that there has been a moderate increase in the rate of tackle injury replacements. The increase in rate from open play is likely to be associated with players returning kicks, and that from scrums is probably associated with the increased velocity that the defensive line has generated by the time tackles are made due to the change in the position of the offside line. Because no tackle injury data were available from the 2006 and 2007, the possibility that the injury rate has trended up due to changes in those two years and has acted either in concert with, or separately from, the ELVs once they were introduced in 2008 cannot be ruled out.

There was concern expressed within the New Zealand Rugby Union from a player safety perspective regarding the permitting of teams to collapse mauls. A documented case of a player in a sitting position sustaining paraplegia after players in a maul collapsed on top of him was reported in the British Journal of Sports Medicine in 2004 ¹². Even though injuries in mauls are extremely rare, it may be prudent to develop coaching materials that demonstrate to players how to collapse mauls ‘safely’ i.e. how to avoid being trapped in a sitting position after pulling a maul down. Even though the rate of mauls per match has decreased more quickly than the rate of injury replacements per match, (meaning that risk of injury per maul has increased) the rate of injury per maul is so low that it represents a minimal risk.

Overall, the introduction of the ELVs into the ANZC has been associated with small to moderate changes across the activities that typically comprise a match. The changes have not been so dramatic that the shape of the game has been completely altered. The ELVs trialled represent a subset of those originally proposed, and it would have been interesting to examine the effect of the ELVs concerning the off-side line at the tackle, which was not trialled in the ANZC, on the shape of the game.

The overall patterns of injuries seen are similar to those in other studies examining injury risk in rugby ^{3, 9}. On the evidence of the ANZC, players are unlikely to notice any real change in injury rates associated with the introduction of the ELVs, although team medical staff and the administrators who bear the costs of injury may see some increase in reported injuries (at the level of the group). If injuries in rugby are viewed as a transfer of energy that exceeds the ability of the body structures to maintain their integrity (i.e. their structure and function) ⁴, then any changes to the sport that result in higher average energies when players collide with each other will be likely to result in higher injury rates overall. Aspects of play under the ELVs that may provide opportunities for more frequent high energy collisions are when players are tackled having run the ball back in open play following kicks, and when the backlines meet following set scrums now that an extra five metres has been added to the gap between them.

Acknowledgements

Thanks to the IRB and the NZRU for project funding, and to Charlotte Wilson of the NZRU for assistance with the preparation and formatting of the report.

The work undertaken by Verusco Technologies in coding the matches is gratefully acknowledged.

References

1. Fuller CW. Catastrophic injury in rugby union: is the level of risk acceptable? Sports Med. 2008; 38(12):975-986.
2. Fuller CW. Managing the risk of injury in sport. Clin J Sport Med. May 2007; 17(3):182-187.
3. Fuller CW, Brooks JH, Cancea RJ, et al. Contact events in rugby union and their propensity to cause injury. Br J Sports Med. May 18 2007; 41:862-867.
4. Fuller CW, Molloy MG, Bagate C, et al. Consensus statement on injury definitions and data collection procedures for studies of injuries in rugby union. Br J Sports Med. May 2007; 41(5):328-331.
5. Hopkins WG, Marshall SW, Batterham AM, et al. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc. Jan 2009; 41(1):3-13.
6. IRB. Changes in the playing of international rugby over a 20 year period. 2005.
7. Merton RK. The unintended consequences of purposive social action. American Sociological Review. 1936; 1:894–904.
8. Quarrie KL, Hopkins WG. Changes in player characteristics and match activities in Bledisloe Cup rugby union from 1972 to 2004. J Sports Sci. Jun 2007; 25(8):895-903.
9. Quarrie KL, Hopkins WG. Tackle injuries in professional rugby union. Am J Sports Med. Sep 2008; 36(9):1705-1716.
10. Reed D, Hughes M. An exploration of team sport as a dynamical system. International Journal of Performance Analysis in Sport. 2006; 6(2):114-125.
11. Stewart JJ. Rugby. Developments in the Field of Play. Palmerston North: Massey University; 1997.
12. Walsh AJ, Shine S, McManus F. Paraplegia secondary to fracture-subluxation of the thoracic spine sustained playing rugby union football. Br J Sports Med. Dec 2004; 38(6):e32.
13. Williams J, Hughes M, O'Donoghue P. The effect of rule changes on match and ball in play time in rugby union. International Journal of Performance Analysis in Sport. 2005; 5(3):1-11.