*edit*
Just checked the bit loading vs dB on the HLOG. It seems that (dependent upon frequency) the cut off for getting a "bit" per tone seems to be in the region -65 (higher tone) to -68dB (lower tone).
I would concur with those suggested values.
FWIW, my line is around 1100m (confirmed exact route, physically measured with a trundle wheel & allowing for down one pole and any 'slack' within joint chambers etc.).
I initially thought it was shorter than that, but I am now aware that it doubles back on itself in 2 places, adding approximately 200m to its 'real' length.
The 1100m is also confirmed by a visiting engineer's JDSU readout (see attached).