Introduction
The first speakers that I purchased was a pair of Mordaunt Short MS3.2s back in the early 1990's (1990 or 1991).  That would make them almost 30 years old at the time of writing this. Appearances-wise they looked like they survived relatively intact, with primarily what looks like UV damage to their plastic front baffles.  Impedance and frequency response measurements howere showed something different. These speakers are currently used as surround speakers for my home theatre setup, where their performance isn't really critical, but I still decided to have a go at them to see if I could get them working a bit better without significant expenditure.

Impedance Tests
I did a full impedance test of the speakers, and both of them suggested that the boxes were leaky.  See that minimum point between the two impedance peaks at bass frequencies? That should be closer, a lot closer to the minimum impedance measured for the speaker (around 7 Ohms), but it isn't.  I ran a test tone through both speakers at Fb, the resonance frequency (the minimum point between the two impedance peaks), and the result was some pretty audible leaking around and above the tweeters. Definitely something that needed to be addressed, and that it was present in both speakers suggested that it was some sort of design flaw. The issue doesn't appear to be audible when playing back music on the speakers however, except under very specific circumstances.

Frequency Response Tests
The results of the frequency response tests suggested that something was wrong with the tweeter for one of the speakers, as the response at high frequencies was considerably lower compared to the other.  To fix this would require replacing the tweeter.

Actions
I opted to try refinishing the panels and replacing the tweeters. I took apart the speakers and removed the front section of the speaker baffles to find that the rear sections were starting to detach themselves from the rest of the box, so that needed addressing as well, which I did by running a bead of glue around the edge and clamping the box until the glue dried.  As for the tweeters, I opted to replace both at the same time. Luckily it turns out that the MS3.2s use a tweeter that was very popular at the time and for which replacements are still available, the Audax TW6.  Parts Express was selling replacement tweeters for the Advent Baby III, and these turned out to be the same Audax TW6 tweeters mounted in custom baffles, so I purchased two of them for  about US$32 in total - a bit expensive, but at least no extensive modifications would need to be made to the MS3.2s to fit the new tweeters.  I also opted to upgrade the crossovers from the original 6dB/octave crossovers (which produced the non-flat response curves seen above because of too much overlap between 5~8KHz ) with 12dB/octave crossovers. I designed them in XSim using the working tweeter, with the hope than its impedance and frequency response would not be that much different to the replacements that I was purchasing.  In any case, I had enough spare crossover parts lying around that I was sure I'd be able to make some minor adjustments if I found them necessary. During the testing of the new crossovers, I found that one of the Positec devices for the tweeters was also measuring a bit off (its impedance was double that of its twin in the other speaker), so I opted to leave them out of the rebuild - it's not like I will ever be playing those speakers at a level to have them be activated anyway.  Oh, and that leak that I thought was likely to be a design flaw?  It turned out to be cutouts around the custom tweeter baffle - filling them with hot glue fixed the leaks.

Results
The rebuild of the MS3.2 speakers was quite successful.  The speakers have a slightly upward-tilting response (a consequence of removing the "positic" devices from the tweeter circuit in the crossover), but they are quite listenable, and the response above 4kHz is a lot flatter and there is a much closer match between the response of the two speakers. The acoustic crossover point seems to be around 3.3kHz and the response below and above this point is smoother than what the original crossover achieved.  One tweeter unfortunately demonstrates a bit more distortion than the other, but I can't hear the impact of the distortion, so I'm not going to worry about it.