Just found this topic by chance and can talk about Tokyo splits: The results were presented in New Studies in Athletics 1992, volume 7, n°1, and presented by JAAF/IAF Biomechanical Project Team. Some extracts:
“Methods - The analysis of the men’s 100 metres was carried out on the basis of recordings obtained from 12 video cameras (60 fields/sec). All cameras were synchronized with the flash of the starter’s gun. Ten cameras were located at 10m intervals along the home straight of the stadium. From these elapsed time, interval times and average speed was determined. A further two cameras were located at each en of the straight, recording the athletes ‘head on’ and from behind. From these the touchdown time for each stride, and thus stride frequency, was determinded”.
On the table presenting findings, it is mentioned that error allowance is +/- 0.02 seconds.
That means that the last 60m intevals by lewis
0.84 0.85 0.84 0.83 0.87 0.86
can also be viewed as
0.85 0.84 0.84 0.84 0.86 0.86
or even
0.83 0.85 0.83 0.82 0.88 0.85
There is an other problem with Rome’87 analysis.
Preliminary report gives numbers as:
1.86
2.87 (1.01)
3.80 (0.93)
4.66 (0.86)
5.55 (0.89)
6.38 (0.83)
7.21 (0.83)
8.11 (0.90)
8.98 (0.87)
9.83 (0.85)
But the Scientific Report published by IAF in their first and second editions gives splits as:
1.84
2.86 (1.02)
3.80 (0.94)
4.67 (0.87)
5.53 (0.86)
6.38 (0.85)
7.23 (0.85)
8.10 (0.87)
8.96 (0.86)
9.83 (0.87)
Here again the difference between the 2 analysis which were published by Charles University is 0.02.
Explanation: in the report we can read:
“Informations and findings from the video recordings were published as FAST INFORMATION REPORTS and were available to the press and to the athletes and coaches in the Athlete Village in Rome. The recordings from each round were shown the following day as part of a video presentation given to the athletes in the Village.
[…]
The location of the high speed cameras enabled 3-D analysis of the athletes […] The films were also used for comparing with the material obtained from the video recordings in the preparation of this report”
So first data was given to press and athletes during the competition, and more precise analysis were published in the final report.
However, we must take in account that 0.01sec is important for us, but sometimes difficult to see on film recordings. 0.01sec at sprinter’s speeds is worth 10cm. And when people ask for 0.001 accuracy for intermediate times, they don’t understand that they ask for a 1cm accuracy, which depends a lot of athlete’s body position in his stride cycle, more exactly on the arm position, as arm movements lead to shoulder rotations, which makes 0.001 accuracy very … inaccurate.
Same problem with photofinishes, technology allow to search times even to 0.0001, but the question is when does the athletes crosses the line. The IAAF rule is so vague (torso without neck and arms) that there are often discussion on where the torso starts and the arm finish, and from which part of the shoulder the time is taken. I know plenty of cases of debates for chosing official time at 0.01 because the sprinter was leaning in a torsion movement, the arm pointing ahead parallel to the track, and shoulder passing before neck and torso himself.