A Die Crack Is a Network, Not a Line

Somebody shows you a photo of a raised line on a coin and asks if it is a die crack. You cannot
tell them. Not because you are not good enough, but because one line photographed once does not carry
enough information to answer.

What answers it is the rest of the coin.

What a die does when it fails

A working die is a piece of hardened steel taking tons of pressure thousands of times. It does not
develop one tidy crack and stop. It fatigues, usually starting at the rim or at a stress point in the
design, and once it starts the pattern spreads.

That is why late-state coins from a failing die show a web. Lines that meet each other.
Lines that branch. Lines that run straight through a device instead of going round it. And where the
steel finally gives up, a piece falls out of the die face and the coin gets a raised blob of metal
filling the hole – a die break, sometimes called a cud when it reaches the rim.

Any single one of those is ambiguous. All of them together are not.

The two things that make a line convincing

It meets another line. A scratch does not intersect another scratch except by
coincidence, and coincidences are countable. Two crossings might be luck. Five crossings in one area
is a network, and a network is what a failing die makes.

It ignores the design. This is the strong one. The devices on a coin are raised
metal, so damage after the strike interacts with them – a scratch rides over a letter, skips, changes
depth, or stops at the edge of it. A crack in the die does not care where the design is, because the
die is one piece of steel and the break runs through it. A raised line that travels from the field,
straight through a letter, and out into the field on the other side at the same strength is behaving
like something that was in the die.

Why mapping beats looking

You cannot hold a network in your head. I have tried. You look at a line, then another, then you
go back to the first one and you are no longer certain it was where you thought.

Worse, you cannot communicate it. “The crack near the middle” means nothing to anyone else, and it
will mean nothing to you in six months when you are comparing this coin against another one.

So map it properly: trace each raised line, mark where they cross, mark the breaks separately, and
give every feature an address somebody else can check. Not “near the middle” – something like “column
six, device band, running out into the lower field”.

Do that and you have a document. Two coins from the same die will produce matching maps, and two
coins that merely both have marks will not.

Address the coin, not the photograph

One technical point that matters more than it sounds. Tie your grid to the coin, not to
the picture. If the addresses are based on where things fall in the frame, then a shot at a different
zoom renumbers everything and your map does not travel.

Anchored to the coin’s own outline, the same feature gets the same address in every photograph you
ever take of it. That is the only reason the addresses are worth writing down.

On a close-up with no whole coin visible, a grid can only be laid over the frame – which is still
useful within that one photograph, but it will not match anything else, and any map that does this
should say so rather than letting you believe otherwise.

Then test the lines with light

Everything above organises what you believe you saw. None of it measures anything.

The measurement is the lamp. A raised crack is bright on the side facing the light and shadowed
away from it; a scratch is bright on the side away from the light, because the lamp reaches
the far wall of the groove. Shoot the same framing with the lamp on opposite sides and watch which
way each line reverses.

Opposite sides genuinely matters. I ran a set where the lamp had only moved about 47 degrees
between frames and the test refused to decide – correctly, because 47 degrees is not the other side,
it is the same side twice. Ninety degrees at the minimum. A hundred and eighty is better.

What a finished map is worth

Two counts, mostly: how many intersections, and how many lines cross a design boundary rather than
staying in one area. Those two numbers are the argument, and they are the ones a photograph alone
never gives you.

Plus the thing that matters in a year – a record you can put beside the next coin. Die states
progress, cracks grow, and a coin from the same die later in its life should show your map plus more.
That is a check you can only run if you wrote the first one down.

CoinScope Pro has a die crack mapper that does the bookkeeping: you trace the lines, it numbers the
coin into columns and bands, finds every crossing for you, tests each line against a lamp that moved,
and writes the map out in words and pictures so it can be compared against the next one.

See the die crack mapper — measure it, see the working, and get a refusal when the evidence will not carry the question.

Not the tool for your problem? GenM Software makes others.