Once and for all (and sorry for the triple-post): the OP picture clearly shows the match time of Santoro-Clement being 6:35.
Why does pretty much any other source has it at 6:33?
You shouldn't be surprised if I didn't add some matches around 4 hours. Usually I count match time from all Grand Slam
matches with solid scorelines on official websites, and later on ATP website there is different match time. I don't know
why it happens often, the same with number of aces
An example of the case at the AO 2004 Safin's time
Safin's Total Time at 2004 Australian Open - 21 hours 6 minutes (1266 minutes)
Round IBM* (& ATP site) ITF** Difference
1st RD (vs. Vahaly) 2:26 2:28 +2 min
2nd RD (vs. Nieminen) 2:39 2:42 +3 min
3rd RD (vs. Martin) 3:25 3:25
4th RD (vs. Blake) 3:08 3:08
QF (vs. Roddick) 3:23 3:25 +2 min
SF (vs. Agassi) 3:42 3:43 +1 min
F (vs. Federer) 2:15 2:15
------------------- ------ ------ -------
Total’s 20:58 21:06 +8 min
* - Time according to 2004 Australian Open official site
** - Official time according to umpire’s scorecard (via ITF Media Notes)
But also on the ATP official site in the Safin's career highlight section i found this info - "2004 - Began
season as runner-up at Australian Open (l. to Federer)...Played a total of 21 hrs. and six mins. and his 30
sets played was most in a Grand Slam in Open Era, tying Harold Solomon's mark set at Roland Garros in 1976".
So it's a bit weird that in one site we have two variants of infos. They get the time stats via IBM, while
21h 06m - via ITF post-tournament info.
I think the main difference is that the umpire on his device and the IBM-staff (who started supporting grand slams in 90's) on the tennis court timer, start the time differently. The chair umpire starts the time when calls "time" right after warm-up, while IBM may start the stopwatch after umpire says "play" right before the 1st serve, or maybe something else.
The official time in any case is determined by a umpire’s scorecard, not by on-court timer