§§163-164

Someone might, instead of writing out the cursive letter that is horizontally across from the printed letter in a table, write out the cursive letter below (consistently). In this case we'd say that they had derived their version from the table (in a similar way to the case of writing out the letter horizontally across).
Someone might not stick to a single way of transcribing but alter it according to a simple rule. There is no clear boundary between this case and a random one.
Wittgenstein asks - "does this mean that the word 'derive' really has no meaning?" - I assume his answer would be 'no'.
He goes on to say in §164 that there is a family of cases of deriving and similarly for reading.
His example - of the person not sticking to a single way of transcribing - makes it sound as though 'derive' is vague - that there are some cases between clear-cut cases of deriving and randomly moving from one side of the table to the other where is is not clear whether what is going on counts as deriving something from an original. Does it give us reason to think that 'derive' is a family resemblance concept?