Here's the dope on double bias peak:
I have not seen this in EMC heads (have not seen those on a 90 yet) but a lot of heads do this. Some heads will exhibit this problem only a couple dB over the first peak (at 30ips w 10K) while others have that peak about 4dB of overbias above the first.
You want to bias - according to John French at JRF magnetics - at 1-1/4 to 1-1/2 dB 'over' using 10KHz at 30ips.... this would be roughly 2-1/2 to 3dB over using 10KHz at 15ips.
I am not a tape head guy but I am told that the "hardness" of the core material in the head (in terms of permeability) is part of the equation, with harder material moving that second peak lower in terms of bias current and closer to the first primary peak.
You want to overbias relative to the first primary peak - not the second, so make sure you crank bias level way down and well below the primary bias peak first and then add bias until you are over.
Once your 10KHz sensitivity stops dropping with added bias (above the first peak) and starts rising into the second peak your distortion will climb like a bat out of hell.
hope that helps