Hi sven_und_jen,
I did some telepathy experiments, the outcome was you get much noise and very rarely good hits, BUT these hits will only be absolute convincing if you had set up a proper experimental situation, which makes information leakages impossible and provide clear evaluation.
For example, take your number experiment. No chosen pool of numbers, say 0-999 ? Evaluation difficult. If you had chosen to the range 0-999 and the sender thinks of 71, and the receiver gets 70, now is this a simply a "missed" or is it quite good because near to 71?

I would recommend just a relaxed state, Focus 10 is probably very good. Regarding the targets, pictures would be much better, easier to permit. There should be a pool of easy to differentiate pictures (of course unknown to the receiver), the best would be provided by a third person, so that the sender would only know the target picture.
However, you may have a look at how professionals are doing this kind of stuff, Joe McMoneagle for example.

Spooky