In the case of the Bogen & some other PA xformers, this is the black 'common' and the brown wire, which corresponds here to a 'straight' 'Z' of appx 100-200^. Close enough & it'll work fine just like that. Simple! We then take our choice of output from the assortment of the Bogen's taps or, perhaps, using the 8^ winding to drive a small speaker with a loud local. Aside from the DC resistance 'Z' matching VR/C device and variable comp cap output coupling (both straight off Ben Tongue's site), the apparent complication of the STM/UtM ckts comes from my decision to configure the entire assembly as an series autoformer. I do this to encorporate the full length of all windings into the working ckt to leave no 'dangling' elements, which is my esthetic design instinct. I also percieve a slight subjective efficiency advantage to the autoformer config and designed it in as a 'flourish', if you will. Since the passive reciever dicipline is a matter of 'inches into yards', every little 'freebie' should be taken, IMO. Of course the tricky pitfall to autoformer-ing is phasing all of the individual winding elements to 'add up', rather than buck each other. Another way to look at the total STM ckt is to see it as a way to 'unify' two separate & very different xformers into one. What we want is to be able to have an economical & highly efficient xformer for xtal sets with a very high input 'Z' and an adjustable range of lower output 'Z' options to match a variety of headsets. The input 'Z' of the T3/AM-20 is 100k^ and the Bogen T725 40k^ total across the tapped long winding. Putting them in series in the fore-mentioned 'autoformer' config gives a total input 'Z' of 140k^. Great for reduced 'Q' loading and perfect for many good fixed detector diodes, except the two windings are not on one core, so the effective coupling to the headset load of both xformers doesn't occur unless we find a way to 'unify' the two separate xformers. We can try doing this by linking them together via coupling two reasonably closely 'Z' matched & correctly phased windings (one must be isolated) on each xformer, which should effectively tightly couple the separate xformers. This appears to work as postulated with the STM ckt, giving this series-connected autoformer'ed link-intercoupled STM ckt an efficiency advantage as well as higher 'Z' input advantage over the 'straight' method of simply coupling the output of the T3/AM-20 detector xformer to taps on the separate Bogen T725 headset xformer. This is the "flourish" I mentioned previously. 

      Unconventional but elegant, it's more of an 'intuitive' concept than an 'engineered' solution.  It just works.

Steve Bringhurst