I’ve been cogitatin’ ’bout this since I first heard about the shootings.
The shootings are a terrible crime. However, I think another crime is being
committed by the aggrandizement of the so-called ballistic fingerprinting.
Ballistic fingerPOINTING is more correct.

As I’ve stated, I’ve been involved in research involving modern machining
techniques. One strong area of research we did is called surface metrology
(metrology means “the study of measurement”). There are a LOT of problems
with trying to measure the surface charactoristics of “standard” bullets and
comparing them to crime scene bullets (or fragments).

To start with, the bullets are not nice neat flat planes. Very few are
simple cylinders. Typically they are a compound (or simple) curved surface.
When fired through a rifled barrel, a portion of the bullet will be engraved
by the rifling (this area is called the bearing band). This area will
usually be engraved with a good representation of the spacing and shape of
the lands and grooves, as well as the twist angle of the rifling, but not
always. Bullet-to-bore size relationship will affect the quality of the
engraving, as will bullet surface hardness. Indeed, an extremely soft lead
bullet can have “smeared” engraving where the surface pulled back, rather
than engage the rifling (this is a well known phenomena with reloaders who
cast pure lead pistol bullets).

Bullet shape also affects the engraving. For example, with the .223 Rem.
that is suspected as having been used in the recent killings, bullets of 40,
45, 55, and 69 grains (all common loads in this caliber) would all have
different engraving marks. They would be similar in many respects, but
because of the difference in the shape of the cross section of the bullet,
the bearing band area will be different for each.

Making a computer algorithm that represents the measurement of a 45 grain
bullet fired from a gun will in all likelyhood match the results of the same
gun firing a 69 grain bullet.

I don’t know exactly how they’re generating the data that represents the
measurements of the bullets. It HAS to be a non-contact method. A contact
method such as dragging a probe across the bullet, or even doing sequential
point measurements, will leave marks on the surface, permenantly changing
the surface. Another problem is the resolution of the measurement. The
finer the granularity (the smaller the feature that can be measured), the
more data there is per bullet/gun measured. In one sense, this is good
because you can make a more accurate match between two specimens, but the
down side is that it takes proportionatly longer to make matches with larger
data files. There’s also a certain amount of noise, or unusable data that
will occur in each measurement.

These are problems in dealing with whole bullets in “new” condition. How
the system deals with partial bullets, bullet fragments, or
expanded/distorted bullets is also unknown. The less material there is to
measure, the less precise the match will be.

Has this system ever been analysed statisticly to see the amount of hits
that were false positives and false negatives? If we’re dealing with a
serious crime fighting tool, I want to be darned sure of the reliability.

And lastly, one of my main objections to creating a database of bullets from
new weapons is that it is one more step towards creating a registry of gun
owners (although from some of the stories I’ve heard, there may already be a
good one in place).