Sometimes you have to wonder.
Consider the Teac FD235HF (most common Teac HD drive that I can think of). It can step 80 tracks at 135 tracks/inch or a track spacing of 0.1875mm.
The stated recording width is
here stated as 0.115mm after guard erase (you need this to separate the tracks and avoid inter-track noise). So the total size of the guard band is 0.0725 mm.
Now, let's go to the granddaddy of 720K 3.5" drives, the Sony 0A-32D, a full-height drive with a 26-pin interface.
OEM manual here Note that the track recording width is given as 0.125mm with a guard band width of 0.063mm.
In short the recorded track width is
slightly wider (0.0095mm or 9.5µM). I suspect that if I run across a spec sheet for the FD-235F (720K), the recorded width will match that of the FD-235HF.
So what's the difference? (I've gone over this multiple times and have documentation from the 3M National Media Laboratory to back me up. Namely, coating thickness, particle size and less importantly, coercivity. The same situation applies to the ED 2.88M recording method, save that a different formulation is used to obtain coercivity in the area of 1200 œ and a particle and coating size that works with perpendicular recording. Track width remains the same.
I can see where the difference between 48 and 96 tpi 5.25" drives gets conflated with the 3.5" drive picture, but they're not the same, aside from the magnetic coercivity of the two media.
If someone can offer hard evidence otherwise, I'd be delighted to review it.