Thoft - The 3.5 was introduced early 1986 and sold approximately 5,000 pairs until replaced by the 3.6 in 1992. Those first woofers failed immediately and were recalled and replaced free of any charge (plus the Hedges album). But heat failure presented constant trouble with all Thiel models, especially the equalized 01, CS3 and 3.5. Slow roll offs use more power and the equalizer worked the woofer far harder than any normal demands would place on it.
Now I'll conjecture a little. I don't know whether that voice coil is underhung - I kinda doubt it since Jim's involvement in finite element analysis only started around the mid 80s when the 3.5 was coming out. Irregardless, underhung voice coils produce 90% less distortion (of some forms), but their short, multi-layer (4 layers!) geometry inherently stores heat. Add to that extra heat the eddy current problems with aluminum voice coil formers which took Jim to using Nomex (which doesn't conduct heat) and you have a stacked deck for heat failures. Your separation is most likely from heat build-up for all these reasons. A re-cone has more advanced adhesives.
In my Thiel Renaissance (admittedly very slow) redevelopment work, thermal management is a serious area of attention. I have measured wire temperatures at over 200°F and they are quite probably considerably higher locally. I'm excited about a solution (which I'll describe at the proper time) which dumps a lot of heat out of the crossover and drivers.
