Friday, 26 April 2013

2013 Londom Marathon Elite demonstrate folly of a Positive Split

Positive splits in Marathon and Ultra distances are not only typical for masses but also very common in the Elite runners as well.  Does this mean that Positive Splits are best or does this just illustrate how hard run Even or Negative Splits when trying to run the perfect race? 
In Stuart Mills recently blog post "A Tiny Bit More on Pacing, and Trail Running Sussex Develops" he analysis the results for recent 2013 London Marathon both the Elite and masses and with it shows just how few people achieve a Negative Spit.  Stuart also goes further and suggests that this is proof that a positive split is a good thing and that a negative split just leads to a sub-par performance.  Is this suggestion sound?  

The fact that this years Elite marathon were at marathon pace at the half way point but the race winner ended being the slowest in 7 years suggest that Positive Split is actually detrimental.  To explore this further I'll look at the splits and finishing times the last four years of the London Marathon that I could find record of, and the also for comparison look at the Marathon world records and the London Marathon Winners Personal Best.

Last four years of London Marathon Winners

Tsegaye Kebede 2010 London Marathon 02:05:19 01:03:07 01:02:12 -1.45%
Emmanuel Mutai 2011 London Marathon 02:04:40 01:02:45 01:01:55 -1.33%
Wilson Kipsang 2012 London Marathon 02:04:44 01:02:12 01:02:32 0.54%
Tsegaye Kebede 2013 London Marathon 02:06:04 01:01:34 01:04:30 4.76%

What clearly stands out is this years race Positive Split of 4.76% was huge compared with the previous 3 years, and the finishing time was the worst.  The course record of 2:04:40 was set with a -1.33% Negative Split, while Wilson Kipsang's time of 2:04:44 was with a modest 0.54% Positive Split.  This data suggests that this years large Positive Split was significant factor in the slowest winning time at the London Marathon in 7 years.  A quote from the  winner says it all : "It was too quick," said Kebede of the early pace.

Personal Bests of London Marathon Winners and World Records.

Paula Radcliffe 2003 Womens's World Record 02:15:17 01:08:02 01:07:15 -1.15%
Haile Gebreslassie 2008 Men's World Record 02:03:59 01:02:03 01:01:56 -0.19%
Emmanuel Mutai 2011 London Marathon, PB 02:04:40 01:02:45 01:01:55 -1.33%
Tsegaye Kebede 2012 Chicargo Marathon, PB 02:04:38 01:02:53 01:01:45 -1.80%
Wilson Kipsang 2011 Frankfurt Marathon, PB 02:03:42 01:01:40 01:02:02 0.59%
Patrick Makau 2011 Men's World Record 02:03:38 01:01:43 01:01:55 0.32%

Here the range of are from -1.33% Negative Split to a 0.59% Positive Split, with the average a -0.59% Negative Split  Even the largest Positive Split still was only a slowing of 22 seconds over second half marathon distance, which really is incredibly even split.

These are the very best performances of the very best of  marathoners, they are the  absolutely pinnacle of human distance running and clearly demonstrate that a Negative Split  will more often deliver a personal best performance than a Positive SplitAnything more than running a very modest Positive Split is detrimental to performance as the London 2012 result show.

These figures really show the importantance of starting out at a pace that you feel is sustainable for the whole race:  "Start out steady and Finish Strong" wins the day.


  1. I heard about an analysis of lap times of all the 5k and 10k world records, which found that in almost all of them, the two fastest laps are the first and last. (Unfortunately can't find a reference ...)

  2. I think this was it

  3. Thanks for the link Keith. The observation about the first and last km's being the fastest suggests over excitement in the first km leading to a fast start, followed by easing off a little to make sure one gets to the end in decent shape, then finally laying everything out on track. While I doubt this is physiologically ideal pacing it probably is an expression of psychological aspects of racing and the imperfect physiologically mechanisms humans have for appraising appropriate pace.

    Curiously Haile Gebreslassie 2008 World record showed similar pacing trait - starting off a bit too fast, slow down in the middle, then finishing stronger, resulting overall with a very small Negative Split of -0.19%. The Science of Sport blog has a good post analysing this race.

  4. Robert
    I am sure you are correct about the benefit of negative split in marathon and HM. Maybe the mechanism is related to the harmful effect of lactate on muscle efficiency. One want to minimise the build up of lactate in the early stages to preserve efficiency. A similar argument might apply to any cause of muscle damage: the given amount of damage occurring early in the race would have a cumulatively greater effect on performance than the same amount of damage occurring later. It might also be that the putative ‘central governor’ is more tolerant of damage late in the event, and therefore ‘permits’ an increase in speed.

    However, I think different principles come into play in shorter races. Perhaps alactic metabolism makes a greater contribution to both the first and last lap, so the most efficient strategy is to utilise alactic metabolism to the full. However alactic metabolism will have a trivial effect in long races.

    1. Thanks for your thoughts. Do you have any references for the link between lactate and reduced muscle efficiency?

      I also believe that 'central governor' model might be a useful way of understanding how the body balances the accumulation of damage vs race performance. I wonder if RPE might be useful way of gauging this damage factor that might influence how strongly we can finish in a race.

      I hadn't thought about the influence of alactic metabolism on initial and finishing performance so it's great you've mentioned it. I believe I witnessed "alactic metabolism" in action this weekend - I was pipped to post in a final sprint at this weekend Balfron 10k. I just sprinted to entertain the crowd I thought I held off my pursuer, finding this last bit of energy seemingly from nowhere, yet he was able to dig deeper in the last few meters to surge past. This final sprint happened despite our legs swimming in lactate and otherwise feeling fully maxed out.

  5. Robert,
    Here is showing acidity reduced peak power of rat muscle
    Knuth, S.T., H. Dave, J.R. Peters and R.H. Fitts. (2006). Low cell pH depresses peak power in rat skeletal muscle fibers at both 30 degC and 15 degC: implications for muscle fatigue. Journal of Physiology,575(3): 887-899.
    The topic has been controversial because the results depend on temperature. There is a good disucussion of the issue by Maglishgo

    Acidity can also inhibit development of mitochondria:
    Bishop et al (Medicine & Science in Sports & Exercise:May 2008 - Volume 40 - Issue 5 - p S33, doi: 10.1249/01.mss.0000321589.97446.a8) compared the effect of 10×2 min cycle intervals at 80% VO2max intensity on two separate occasions: once after consumption of ammonium chloride, which causes an artificial acidosis and once after the consumption of calcium carbonate which does not cause acidosis. On both occasions there was a marked increase in translation of the DNA that codes for a protein that regulates the generation of mitochondria. However, the increase was less marked in the presence of the acidosis induced by ammonium chloride. Thus even in the presence of an artificially induced acidosis the production of mitochondria is likely to be increased by vigorous exercise, but mitochondrial production is likely to be greater during vigorous exercise in the absence of acidosis. Overall this study suggests that intervals at 80% VO2max (which would be likely to be upper aerobic for many athletes) promote marked increase in mitochondrial production, while at anaerobic levels mitochondrial production would be expected but might be less marked.