A substantial part of what is known about this material is a question of technique rather than interpretation. A curse tablet is a soft metal sheet scratched with a stylus in a cursive hand, folded, and then buried or thrown into water for between fifteen hundred and two thousand years. The metal corrodes from both faces. The scratches that carry the text are shallower than the corrosion is deep.
Conservation of the Uley and Bath groups involved unfolding sheets that had become brittle, cleaning surfaces without removing the inscribed layer, and reading letters that survive as differences in surface texture rather than as visible marks. Raking light, casts and radiography have all been used. Some tablets have been read only in part, and some readings have been revised.
That has two consequences for anybody using this material. The first is that the corpus is not fixed: what a tablet says can change with re-examination, and the published figure for how many tablets a site produced is not the same as the number that can be read. The second is that the illegible majority is not neutral. If short texts corrode to nothing more readily than long ones, or if a particular hand or metal thickness survives better, then the readable sample is not a random sample of what was deposited, and every generalisation drawn from it inherits that bias. This is why the atlas records the count of tablets recovered separately from the count that can be read.