c.mpr[: there is no doubt about the reading of c even though most of the vertical has disappeared. There is some smudging between c and m and either com or cum is possible. We prefer the former and can envisage compr[ehende-/ram; this is perhaps unlikely to mean "understand" without some further qualification (see OLD, s.v.11), but it might well mean that the author had detained or arrested someone de hac re (see OLD, s.v.5).
cras: the reading is not easy but if the pieces are moved flush most of the difficulty is removed by the recognition that the writing is fouled by the hook of a long descender from r in line 1.
Lugu[ualio: the initial letter is crossed by the top-stroke of the preceding s and the last surviving letter is not easy, but it has probably been fouled by a descender from the previous line. For the locative place-name cf. 250.i.9 and see Adams (1994). There is no difficulty in supposing that the sender might be intending to speak to a prefect of a unit at Luguvalium or perhaps someone in a higher position.
It is possible that tibi was written in the lost section at the right.
The surviving tops of letters are consistent with [pra]ef(ecto); further to the right there are exiguous traces which must belong to c[o]h. Below this line there is an apparent trace of ink at the bottom right of the small piece which, if it is to be taken seriously, might be from the end of the name of the sender, written with an upward slant.