I don't know anymore. Some of my teachers cheated and made the gaps between the top ranked people very small (1%) anyway. It meant that in the exam the bottom 2/3rds of the class, say, would get their external marks and would likely be roughly in the same distribution as internal marks. The top third however, would be more widely distributed from each other than they were internally, so the internal gaps would be partially mapped onto the external marks for the top third, pulling up the top ranked people closer to the mark of the people who came first. Roughly speaking, of course. I hate statistics. The reason they could only do it to the top third is that if they did it to the entire class, nothing would happen since the gap distribution would be directly proportional to that of the external marks.
If I'm wrong in any way feel free to correct me. It's been about a year since I've thought about it very hard.
I suppose if you think about it, if nobody has improved on their internal marks or if everybody has improved by roughly the same ratio, and assuming teachers don't fiddle with internal marks, then gaps between marks mean nothing at all.