Firstly, it's unlikely these caps will be leaking. Though it looks like a 50 year old cap, it appears it is new production (just in an older-style can). I've used 50-60 year old caps far larger than the one in the Ref 3A without any difficulty - in fact, they have a far greater shelf life than all modern day electrolytics.
Bypassing is a fairly complex issue. Why I recommended bypassing here was to replicate the value of the cap already in there. For example, if you have a cap value of say 5.17uF, and you can't find a 5.17uF cap to replace it, then you can use a 4.7uF and a 0.47uF to replicate the value. The general rule with capacitance is that paralleled caps add, and series caps have an inverse relationship. This is the opposite for resistance, in case you're interested.
So to replicate 5.17uF, you take a 4.7uF cap and parallel it with a 0.47uF cap. It is important to get the right value, since the differences you hear may be more due to the changed value (therefore changing crossover point of the tweeter) rather than the different type of capacitor.
Now, modders in the 70s and 80s were obsessed with parallelling caps since, in general, smaller value caps have better performance characteristics (ESR etc). This applied to both power supply and signal capacitors. Therefore, if you say had a 0.47uF coupling capacitor, people would parallel this with a 0.047uF, and possibly even smaller value caps. Supposedly you hear more 'air'. I've had variable results from this technique, and prefer instead to just use a high quality cap suitable for the application at the start.
Hope this helps.
Robert
