[MV=PY]

Here is an interesting blog post on the subject. While Wright is attempting to make the case that, for all practical purposes, overwriting data once is sufficient to obfuscate older data on magnetic media, it all begins with an analog signal of variations in magnetic flux strength from which "bits" are extracted. This is very different from truly discrete "bits" as found in solid state memory.

Old "signals" (writes) remain on the magnetic media. This has nothing to do with logical organizations of binary data such as sectors. It's at a lower level. It's all about residual magnetic flux.

Many of the comments at the bottom of the post are informative. I believe the last comment to be correct: given a strong enough reason, forensics can be effective.

The company I mentioned has many levels of recovery services, including the very low level analog stuff.

I hope this helps explain where I was coming from. ;-)

[coon2000]

Your link agrees with my position, even the last comment. I briefly touched this in the two posts above yours.