The Don’s ‘Missing’ Runs: A Question of Averages


Written for the Sydney Morning Herald, The Age and the West Australian. Original Text.


By Charles Davis

30th July 2008




It’s the most famous statistic in Australian sports, familiar to most Australians who love cricket, and to a great many who do not. Don Bradman’s 99.94 runs per dismissal in Test cricket is entrenched in Australian folklore, and the story of The Don’s final innings, out for a duck when just four runs would have taken his average to 100, has been told countless times.


Just four runs, spread over a career of almost 20 years and more than 197 hours at the crease! It’s worth asking, just how sure can we be that the number is correct? As it happens, recent research has shown that it could be wrong: a tantalising clue to four missing runs has been found. It is all to do with an activity usually taken for granted, the scoring.


In June, when Ricky Ponting scored his 10,000th Test run, few would have given a thought to just how much meticulous recording has gone into producing that figure. It comes from scorers obsessive about detail; their methods, including continuous cross-checking, almost always work well, especially in the computer age. It would be fair to expect that an official scorebook would be an absolutely watertight guide to the events of a Test match.  Yet sometimes it is not. Some years ago I embarked on a project to examine old Test matches scorebooks closely, to uncover previously hidden stats like balls faced. Over the years, it came as a great surprise to find that apparent errors and anomalies arise quite regularly.


These problems are almost never obvious. You won’t find a scorebook that says Bradman scored x when the accepted score is y. The issue is more subtle: a batsman’s batting strokes may not add up to his score, or more commonly, there are mismatches between the strokes attributed to batsman and the strokes marked against bowlers.


There are surviving scorebooks for many Tests of the past. All but seven of Don Bradman’s Tests are represented, preserved in libraries at Lord’s, Cricket NSW, and elsewhere. Earlier Test scorebooks also survive, as far back as 1880. Problems in the detail begin with that 1880 Test at the Oval, which has three “rogue” overs, and crop up in about two-thirds of the scores that I have examined closely in Tests before the 1960s. Some interesting examples:


-             In the famous Sydney Test of 1903/04, Victor Trumper scored 187 not out, but his scoring strokes only add up to 185. Monty Noble’s 133 only adds up to 128.

-            Charlie Macartney’s famous 345 against Notts in 1921 was probably 343.

-            There is probably an error in Wally Hammond’s mammoth 336 not out in New Zealand, when he broke the Don’s record score of 334. Fortunately for Hammond, the error appears to be in his favour; he may have scored 337 not out.

-            Sid Barnes and Don Bradman both scored 234 at the SCG in 1946/47. The story goes that Barnes threw his wicket away so he could have the same score as the Don, but in one surviving scorebook his strokes add up to only 233!


This is mostly  the stuff of trivia, although occasionally something of greater import arises. In a Karachi Test in 1972/73, Mushtaq Mohammad suffered the rare indignity of being run out for 99. He would probably be intrigued to hear that a re-scoring  of the official scorebook gives a score of 103.


Unfortunately, most of these anomalies are inconclusive. If something in a scorebook does not compute, this does not mean that the accepted score must be wrong. In the Barnes case, for example, there is good reason (from the bowling details) to think he really did score 234. In 1957/58, Garry Sobers’ 365 not out against Pakistan broke Len Hutton’s record Test score by just one run, but in the scorebook, there is a boundary out of place. Close inspection suggests that Sobers’ score should stand; it does create uncertainty, however.


So it is with Bradman. In the scorebook of the epic eight-day Melbourne Test of 1928/29, won by Australia by five wickets, there is a ‘problem’ boundary in the final stages, when Bradman was batting with Jack Ryder. (I found this when re-scoring the entire Test, ball by ball, to recreate the exact sequence of events). The relevant sections of Bill Ferguson’s original score are illustrated: there are four runs, attributed to Ryder, that are in the wrong place in both the batting section of the score and in the bowling section (Maurice Tate’s 35th over). There is no doubt that a recording error of some kind has occurred. So where do these runs belong? Perhaps Ryder scored them at some other point of the innings, perhaps they weren’t scored at all (in which case Australia, technically, did not win the match). More importantly, perhaps they were scored by Bradman.


There are a number of possible resolutions, complicated by the presence of an unmarked leg bye. At least one resolution involves transferring the boundary to Bradman. If so, a ‘holy Grail’ of statisticians, four more runs to Bradman, has been found, and the ‘perfect’ average of 100.00 achieved!


Is it really possible? Well yes it is, but unfortunately it is unlikely. One newspaper mentions some confusion between bye calls and boundary calls late in the match, but not an extra boundary to Bradman, and other possibilities, giving the runs to Ryder earlier in his innings, seem rather more likely.


Most of Bradman’s scorebooks have not been checked at this level of detail. It is painstaking work. However, the chances of finding other anomalies, based on experience with many other scores, seem high. It’s worth remembering, of course, that errors could easily cut both ways: Bradman could lose runs as easily as gain runs this way.


Ultimately, that iconic average of 99.94 will probably stand. Wisden is against the retrospective alteration of scores (“that way madness lies”) and I tend to agree. I do think, however, that problems with  scores from the pre-computer age may create uncertainties of a few parts in a thousand. For most statistics, this is no more than historical footnote. The Bradman average is an exception: if it really is 99.94 ‘plus or minus’, there will always be that tantalising possibility of the magic 100.


Score excerpts were reproduced with permission of Cricket NSW.


[Figure Caption: the scorebook of the final Test of 1928/29. Arrows point out hits for four, in the batting and bowling sections, that are inconsistent with each other and with match descriptions. So who actually hit these four runs? Ryder or Bradman?]


Charles Davis

July 2008