Isn't there additional phase change from using multiple caps?
Everything else being equal, I would think it would be better to use several smaller caps in parallel. The smaller caps would likely have lower stray inductance, and lower esr (equivalent series resistance), both of which would then be further reduced by the parallel connections with the other capacitors (parallel connection reduces inductance and resistance, while it increases capacitance).
The only difference in phase effects that I can envision would correspond to the difference in stray inductance, which, although probably minimal, would be in the direction of favoring the parallel combination.
It's hard to say without having detailed information on esr, stray inductance, etc., and it may not make a significant difference either way, but my instinct would be go with whichever approach (within the same make/model series) minimizes the use of capacitors which are physically largest.
In other words, if the 50 and/or 33 are significantly larger than the 24, go with the five 24's. If not, and if the 12 is significantly smaller than the others, then perhaps go with the staggered values. My thinking is that a physically larger capacitor is likely to have significantly greater stray inductance.
I'd suggest to use several of the same Capacitors from the same batch and match values as closely as you can. Probably the most important will be to choose something fairly musical and yet reassonably priced and long lasting such as Ansar Supersound polypropylene. Ceramic would be terrible. I would be concerned about mix and match in the high signal level path - caps change shape slightly and "ring" with high voltages applied across them - it might be better to have a closely matched set or you may lose clarity.
I have heard proponents of both theories. There are some who swear by "cascading" caps such as 10uf/7uF/4uF/2uF/1uF to make 24uF. There is another camp that says that the caps should all be of the same size when paralleling them in order to avoid time-smearing the sound, such as 6 X 4 uF to get 24uF.
In my Infinity RSIIb crossovers I'm using the latter approach.
An excellent source for high quality caps, and Michael will also match your caps to within 1%: (http://www.percyaudio.com/Catalog.pdf) I've used DynamiCaps in Maggies with great results. auriCAPs are also an excellent cap: (http://www.partsconnexion.com/t/catalog/CapacitorsFilm.html) Whatever caps you use, be certain to observe proper lead(outer foil)orientation: (http://www.audience-av.com/capacitors/a_applications.php) Another service that Percy offers is marking the outer foil leads on the caps he sells. Keep in mind that when you first replace the caps, and fire the system back up; it probably won't sound great. It takes a while, with operational voltages, for the dielectric in a high quality cap to "form".
Thanks for all the responses. I think I will be going with 5 Clarity SA 24uf caps in parallel for a combined total of 120uf. I will hold off on a bypass cap because there seeme to be two camps on this. I will try it without bypass for a while, then get some, and see which way I prefer.
Thanks to all... John