-----Original Message-----
>John Pollard wrote:
>
>"Johnson, Tracy" wrote:
>
>> Target Practice would actually leave portions of the disc
>> intact and therefore readable through forensic processes.
>>
>> Even writing ones and zeroes three times is not effective with
>> the proper technology to recover the data.  Although it is usually
>> sufficient for government purposes, and is the usual guideline.
>> There is a procedure to recover erased data from disks that uses
>> latent states of magnetism from of the original write on
>> magnetic media.
>
>I have read, in more than one place I believe, that a minimum of seven
>overwrites will foil the above recovery process.

I did a little checking around, and even asked my dear old engineer friend
(the one with the alleged Hysteresis) and found the following:

I don't remember any particular name associated with the effect. The fundamentals go something like this.

BACKGROUND:

As you know the magnetic coating used on hard disks, floppy disks or tapes is a "soft" magnetic material. Meaning the polarity can be easily changed. The magnetic coating is made up microscopic magnetic particles that act like very little bar magnets held together in with some kind of bonding agent usually a plastic material of some kind.

To "write" information on the magnetic material (coating) a external magnetic field is generated to align the particles with the field. This is done with the "heads" which are just very small electromagnets (a piece of Iron or Steel with wires wrapped around it through which electrical current is passed). With audio tapes the strength of the magnetic field is changed linearly with data storage the strength is changed in steps. The strength of the magnetic field in the coating is proportional to the percent of the magnetic particles that are aligned in a particular direction.

With digital data storage you don't try to make 100% of the particles align in "1" or "0" polarity. There is a whole list of reasons I'm aware of and probably a much bigger list of ones I'm not aware of as to why.  All you need is enough to detect the difference.  So, for arguments sake lets shoot for 90%.

Answer:

Every time you write a bit to the disk you align 90% of the particles in that direction. What happens when you write new data to the disk over old data, Assuming a 0% alignment start to start with is.

        was     is      strength
        0       0       -99%
        0       1       +89%
        1       0       -89%
        1       1       +99%

For the purpose of reading the new data only + or - matter but if you can detect the difference between +/- 99/89 as well you can recover both the new and the old data.

Well that covers it. This is a simplistic example. In the real world you get into noise issues, coding algorithms and error correction stuff.

Tracy Johnson
MSI Schaevitz Sensors 

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *