Using the training calculator on tibia-stats.com (http://www.tibia-stats.com/index.php?akcja=skillCalc) Also, all of the offline times I mention will not include the time taken to regenerate any offline training if necessary.
For a Knight to get 1% of a melee skill from 100 to 101 would take 1 hour(s) 28 minute(s) 34 seconds of online skilling and 2 hour(s) 57 minute(s) 8 seconds offline.
For a Knight to get 1% of a melee skill from 130 to 131 would take 1 day(s) 1 hour(s) 45 minute(s) 10 seconds of online skilling and 2 day(s) 3 hour(s) 30 minute(s) 20 seconds offline.
In both of these examples offline training takes exactly twice as long as online training.
If we use similar examples for a paladin to calculate distance it is a little different because for distance skill you get approximately twice the benefit for getting a "blood hit" on your target vs getting the hit blocked so you will see a bigger gap here. Again, offline training does not take into account the time taken to recover the offline training time. Also, the rate will change slightly based on what ammo you use. I ran these calculations with spears as they are cheap and you didn't have to worry about killing the Target Dummy.
For a Paladin to get 1% of a distance skill from 100 to 101 would take 30 minute(s) 12 seconds of online skilling and 2 hour(s) 12 minute(s) 50 seconds offline.
For a Paladin to get 1% of a distance skill from 130 to 131 would take 8 hour(s) 46 minute(s) 46 seconds of online skilling and 1 day(s) 14 hour(s) 37 minute(s) 45 seconds offline.
In the first example with a bit of rounding offline training is 4.398x slower and in the second example it is 4.399x slower. So looks like we can assume 4.4x slower for offline training as a paladin.
So in conclusion it looks like offline training a melee skill on a knight is 2x slower than online while offline training distance on a paladin is 4.4x slower than online. I guess this explains why we've seen melee skills close the gap on highscores compared to distance skills in the last few years.