Label loss, which is total loss
Updated
Every other failure on this list degrades a specimen. This one deletes it.
The only unrecoverable failure
A cracked specimen is a cracked specimen. A faded one is a faded one. Both are still what they were and still documented.
A specimen that has lost its data is no longer a specimen. It cannot be restored by any amount of expertise, because the information was never in the stone.
Which makes this the failure to design against hardest, and it is the one that gets the least attention because it is not chemistry.
How it happens
Cleaning. The specimen goes to the sink, the label stays on the bench, and three specimens come back to two labels. This is the single commonest cause.
Rehousing, when a collection is moved between boxes in an afternoon.
Shows, where specimens are picked up and put back in the wrong place.
Inheritance, when somebody tidying discards "the bits of paper".
And loose labels in a box that is turned over in transit.
The acid case
The one that is chemistry after all. A decaying pyrite specimen produces sulfuric acid that attacks paper, and acetic acid vapor from wood or PVC embrittles it.
So the specimens most likely to destroy their own labels are exactly the ones whose labels you most need, because they are also the ones that will be unrecognisable once the decay advances.
Photograph the labels of every sulfide specimen now. It costs an hour for a whole collection.
Preventing it
A catalog number written on the specimen itself, which is the whole point of numbering.
A catalog kept somewhere other than the collection room.
Photographs of labels, especially old ones.
And one procedural rule that removes most of the risk: the label and the specimen move together, or neither moves. At the sink, that means the label goes into a labeled tray the specimen came from, not onto the bench.