SundayRider said:
So its not the final climbs and TTs where riders lose many many minutes that have more of an impact.
Not on the average speed of the winner, no. On the placings between riders, yes of course the final climbs and TTs count. On the average speed of the winner, from year-to-year, the pace set by the bunch has a far greater effect.
Take the 2012 Tour, from the top to the bottom we have;
1. Bradley WIGGINS, Sky, in 87:34:47 (87.5797 hours)
153. Jimmy ENGOULVENT, Saur-Sojasun, at 3:57:36 (91.5397 hours)
Now the race was, 3,496.9km meaning that while Wiggins' average speed was 39.93km/h, Engoulvent had an average of 38.20km/h. If you look upthread you can see that Engoulvent raced it faster than the winners 1983 to 1989. The vast majority of the Tour is raced as a peloton - the speed set by the pacemaking teams on the flat has a much bigger effect on the overall time of the winner than times up climbs.
To look at another example, let's take the stage 9 ITT;
1. Bradley WIGGINS, Sky, in 51:24
178. Jimmy ENGOULVENT, Saur-Sojasun, at 11:10
Neatly, Wiggins and Engoulvent bookend the stage as they did the overall. The difference between the two on this stage is, as we can see, just 11:10 - over the 3,496.9km. Had Wiggins finished at the same time as poor Jimmy, his average speed would have been 39.84km/h, just 0.085km/h slower than it was. While this would have been enough to lose the Tour to Froome, it does not constitute a major difference in the average speed.