For the first two decades after it became an organized sport, hockey was an iron-man sport. That is, if you played the game, you played the full 60 minutes. Until the 1910s, substitutions were generally made only in the case on injury, and even then (especially in the earliest years) the other team often dropped a player to stay even, rather than having a new player coming on. Compare this situation to the modern game, where changes are made constantly throughout the game, every 40 seconds or so. Teams dress 18 skaters a game, despite only five being allowed on the ice at any one time. It's a drastic change, so I thought it would be interesting to examine the genesis of the change, which did not occur overnight.
Just to illustrate how the game was played before 1910, let's have a look at the complete scoring record for the 1906 Montreal Wanderers, who finished 9-1 on the Eastern Canada Amateur Hockey Association (ECAHA) season and won the Stanley Cup. Note that under the positions, 'R' stands for rover, 'CP' for cover-point and 'P' for point. Cover-point and point were the defensemen positions, who played one in front of the other (cover-point in front), instead of side-by-side.
1906 Montreal Wanderers
Player Pos GP G A Pts PIM
Russell, Ernie RW 6 21 2 23 13
Patrick, Lester R/CP 9 17 3 20 18
Johnson, Ernie LW 10 12 3 15 44
Glass, Frank C 10 10 1 11 12
Blachford, C. RW/R 6 7 2 9 13
Kennedy, Rod CP 7 6 2 8 10
Strachan, Billy P 10 1 2 3 17
Arnold, Josh RW 1 0 1 1 0
Kennedy, Ned C 1 0 0 0 0
Note that the team played 10 games, and there exactly 60 man-games played by the players. Since there were six men on the ice at any one time (excluding goaltenders) due to the rover position, the Wanderers used no substitutes at all this year. Every man who played played every minute of his games, except when he was in the penalty box or injured.
This was standard operating procedure, and didn't change until the second year of the National Hockey Association's (NHA) existence. From 1905-06 (the first season of the ECAHA) through 1909-10 (the first season of the NHA), teams played an average of 6.05 to 6.13 skaters per game, meaning a team played an average of 0.05 to 0.13 substitutes per game. Between the 1909-10 and 1910-11 NHA seasons, two of its franchises (Haileybury and Cobalt) left the association. Seemingly in response to having a surplus of high level talent available due to these defunct teams, the number of substitutes per game shot up to 0.77 per game. The number was very consistent between teams, as well, so it wasn't just one club playing more subs than the others to skew the numbers; the range of man-games over the 16-game schedule was a very tight 106 to 110. The 1911 Stanley Cup champions were the Ottawa Senators, whose scoring stats follow:
1911 Ottawa Senators
Player Pos GP G A Pts PIM
Walsh, Marty C 16 35 5 40 51
Kerr, Albert LW 16 33 3 36 45
Ridpath, Bruce RW 16 23 9 32 54
Darragh, Jack R 16 18 6 24 36
Lake, Fred P 16 5 6 11 12
Shore, Hamby CP 16 7 3 10 53
Currie, Alex LW 4 1 0 1 10
Gaul, Horace C 3 0 0 0 3
Stuart, Bruce RW 3 0 0 0 0
The Senators were able to use the same six starters in all 16 of their games. The difference between this team and the 1906 Wanderers is that here, the substitute players were not only on hand in case on an injury. Though they did not play every game, substitutes were used to give a starter a break, though not very often.
The next season, the average number of skaters per game in the NHA actually dropped to 6.26. Of course, the NHA had dropped the rover position from the game, so the average number of substitutes actually increased from 0.77 to 1.26. This number jumped again in 1912-13 to 1.75, to 2.56 in 1914-15, to 2.77 in 1915-16 and 3.34 in 1915-16. In the latter season, the Montreal Canadiens won their first Stanley Cup championship. Their stats look like this:
1916 Montreal Canadiens
Player Pos GP G A Pts PIM
Pitre, Didier RW 24 24 15 39 42
Lalonde, E. C 24 28 6 34 78
McNamara, H. CP 24 10 7 17 119
Laviolette, J. LW 18 8 3 11 62
Prodger, George P/CP 24 8 3 11 86
Ronan, Skene C 8 6 4 10 14
Corbeau, Bert P 23 7 0 7 134
Poulin, Georges RW 16 5 1 6 43
Arbour, Amos LW 20 5 0 5 6
Berlinguette, L LW 19 2 2 4 19
Fournier, Jack RW 9 1 0 1 4
Here we can see that left wing was being split among three players in Laviolette, Arbour and Berlinguette. Skene Ronan, a gifted scorer in his own right, gave more than a breather to Newsy Lalonde, playing enough to score 10 points in only eight games. The Habs also had a spare defenseman in Goldie Prodgers, who split time at both the starting point and cover-point. This team had a very different character than the Wanderers of just ten years before.
So the eastern teams were rapidly increasing the number of substitutes they employed each game. Here is a graphical representation of the average number of substitutes per game, from 1906 to 1926, covering the ECAHA, ECHA, NHA and NHL:
At this juncture, something of an equilibrium point seems to have been reached, with the average number of substitutes ranging from 2.83 (in the war years, which caused a small shortage of players) to 3.60 between 1915-16 and 1923-24. Standard practice was to employ two substitute forwards and one substitute defenseman. Managers and coaches were experimenting, trying to strike a balance between having your players rested by not having to play every minute, and having your best players on the ice at any one time. Substitute players might have fresh legs, but they're also inferior than your starters.
Although the number of substitutes was on the rise at this time, the role of the substitute was not really changing yet. They were still used to give starters a breather, and did not yet receive anything near equal ice time to the starters. We can see this by looking at the average points-per-game of each team's starting forwards, divided by the average points-per-game of the subs.
Ratio of Starter points per game to Substitute points per game
This indicates that although there were more substitutes in each game, the minutes played by substitutes probably wasn't changing much yet. And it didn't progress in subsequent years; the quotient in the NHL in 1922-23 was 5.8, for example, which implies that substitutes were playing fewer minutes than in the years noted above.
It's also interesting to realize that although the PCHA started behind the NHA in terms of substitutes per game, during the same years that the NHA started to see an increase in the number of substitutes used, the same occurred on the west coast. The PCHA played iron-man hockey in its first year of existence, which is understandable given the relative difficulty of obtaining players to play in British Columbia at the time, when all the best players were from east of Saskatchewan. This is one area of the game that the Patricks cannot claim to be the great innovators in. The eastern teams began regularly using substitutes ahead of the west.
All of this brings us to 1924-25 and 1925-26, the last two years on the first graph above. When the NHL added the Boston Bruins and Montreal Maroons in 1924, the average league substitutes increased by 0.42 to 3.96, the highest it had ever been. The following season, the Pittsburgh Pirates joined the league, and led by coach Odie Cleghorn's new ideas, the average number of substitutes jumped again by almost a full player per game, to 4.93. This was the game's introduction to the modern way of playing the game: with forward lines, rather than starters and substitutes. But that's a subject for another column.