In my post a few weeks ago about Tendulkar’s 52nd Test hundred I included a bit about Sir Donald Bradman being the best batsman of all time. There are considerable difficulties comparing batsmen from different eras, but fortunately the Don was so good that there is little need for any subjective weighting of eras. Unfortunately it’s not so clear cut with respect to bowlers. The primary statistic for a batsman is his average, and whilst more recent ones do tend to be higher they are still readily comparable. The primary statistics for bowlers are wickets taken and bowling average though and those are both problematic when comparing bowlers from different eras. Bowling average has the same problem as batting average, plus it compares wickets to the frugality or otherwise of a bowler rather than to the total amount played (as batting average does). Total number of wickets taken, usually the primary stat, is faulty because many more Tests are played now than in years past. Murali leads the chart with 795 Test wickets, (like Bill Frindall, I exclude the ICC Super Test) but he played in almost five times the number of Tests as Sydney Barnes. A more seldom seen stat, but more theoretically useful in this case, is bowler’s strike rate, a measure of how many wickets a bowler took per deliveries bowled. (Technically it’s the other way around I know, but my phrasing still measures the same thing.) It’s still not perfect, however. It favours bowlers in the past who had the benefit of uncovered pitches and matches would end sooner than they do now.

Earlier this week I used a wickets taken per match to demonstrate Sri Lanka’s bowling ineptitude. After some thought, I concluded that it would probably be the best way to compare different bowlers from different eras as well. Obviously it corrects for total matches played, but since the total number of wickets that can be taken in a match has never changed it ought not to favour any particular era over any other. Like any average, it will require a minimum number of matches played. I opted to use 15, which would be small enough not to unfairly exclude bowlers who played in the very early days of Test cricket, but high enough to not include those who have played for less than a few years in modern times. In my analysis I actually use 30 innings bowled, so as to exclude batsmen who have played dozens of matches, but only occasionally bowled. This is the most subjective aspect to the analysis and a case could probably be made that 15 is a bit low and 20 would be a better number. That would leave out George Lohmann, the man with the lowest bowling average of all time, and Fred ‘The Demon’ Spofforth, however, so I prefer 15 as the cutoff.

Since Statsguru does not (to the best of my knowledge) have a feature to display wickets per match and I do not wish to perform the analysis by hand for every bowler in history to have bowled at least 30 innings I decided to look at the top 35 bowlers in each of the three traditional stats above. The Statsguru screenshots for these are shown below. (Click to make them large enough to actually read.)

After accounting for overlap between the lists I was left with 68 bowlers from all eras. I should point out that this is not an exhaustive list of the top 68 bowlers in terms of wickets per match, only the 68 which seemed most likely to have a high ratio. After checking to see if I had missed anyone with a particularly high ratio, I added Clarrie Grimmet to the list. He was not amongst the top 35 in any of the original categories, but still took an average of 5.84 wickets per match. Ultimately I found 12 bowlers who had a career wickets per match ratio over five and they are presented in the table below.

By this analysis Sydney Barnes is the greatest bowler to have ever played the game, having averaged exactly seven wickets per match over the course of his career. It’s probably not the first name to leap to most minds, but nor is it an unreasonable choice. He still holds the record for most wickets in a single series, having taken 49 South African wickets in four matches in 1913-14 and his 17-159 at the Wanderers in that series is second only to Laker’s 19-90 in terms of match analyses. Even by the traditional metrics 189 wickets at 16.43 apiece is very, very good. It’s also not a novel choice. As recently as 2009 David Frith in Cricinfo suggested that Barnes was ‘[p]robably the greatest bowler who ever measured out a run-up’.

I don’t think there could be many complaints about the other eleven on the list either. The only one of whom I had not previously heard was Charlie Blythe, though Bobby Peel (the cricketer not the Prime Minister) is best remembered for allegedly urinating on the pitch during a Roses match and being banned from playing for Yorkshire by Lord Hawke. (By all accounts he was a very good bowler though.) After those two there are all quite famous names from almost every era. The fact that Murali is third on the list and the only contemporary bowler to average better than six wickets per match is not surprising, given the number of wickets he took in his career. (And it goes a long way to explaining why Sri Lanka have struggled so badly recently.) The most notable omissions from that list are the great West Indian fast bowlers. They did not miss out by much it should be pointed out; all bar Courtney Walsh averaged better than four wickets per match. I suspect the reason for this is that the Windian bowlers tended to share the wickets around. With so many greats in the same team and only twenty wickets available none were able to stand out as much as they undoubtedly would have had they played separately.

I would like to be able to compile a longer list, but I cannot be sure of the accuracy as I go farther down. Without going through every single bowler to have ever bowled in more than 30 innings it is not possible to ensure that a longer list would actually be complete. Hopefully at some point wickets per match will be considered a proper statistic and be easily available on Statsguru.