A[t-]/tonem: this is no doubt the same person whose return is requested in ii.3, hence the restoration of the end of line 1. The long descender which can be seen before this must belong to r (a in this hand has one or two short descenders) and is most easily explained as the end of per.
It is notable that there is no interpunct between misi and tibi.
illos expungas: the end of illos is rather smudged but the accusative can be read and is what we would expect. It must refer to persons mentioned earlier in the letter. expungo means "remove from the list"; in a military context it can be used of deserters (Digest 49.16.15, cf. 226.a.10) or of straightforward discharge (cf. _____________, P.Oxy. IX 1204.19); cf. RMR 47.i.16 note, Gilliam (1986), 113). See O.Bu Njem, p.90 and cf. TLL V.2 1833.
There is a clear interpunct between ali and quam, which means that ali must be taken as alii (the form is cited by OLD at CIL 2.6278.25).
The restoration of the end of the line is problematical. After rogo, mi[h]i is compatible with the traces. Thereafter there may be very slight traces and a descender. If we read eu[n-]/dem we would have to suppose that the descender belongs to a letter from the line above (cf. the long descender of r between e and m at the end of line i.2). The only plausible alternative which occurs to us is ib[i-]/dem in the sense "at once" (OLD, s.v.3), in which case we would have to suppose that we have the descender of the first i but that the second example was shorter.
There is an interpunct between dem and Attonem guaranteeing that the name is Atto rather than Matto; both are attested but Atto is the commoner, see NPEL, citing 11 instances in Gallia Belgica. For the name at Vindolanda see 308 and cf. 320.2 note.
R.E.Birley (1990), 18 restores the name of Vettius Severus (305.back) as the addressee (Vettio] S[euero). The identification of Severus as a prefect at Vindolanda in Period 2 is very uncertain (see above, p.26). The descender which is visible is no doubt r in this hand. It can hardly belong to an initial S of Seuero; if it did we would be sure to see the descender of r further to the right and nothing is visible. Thus Vettio Seue]r[o] would be a more attractive restoration, but this is entirely speculative.
Celonio Iusto: we have considered reading the gentilicium as Velonio (see LE 99) but C is a better reading. Celonius is not cited in LE but there is a dedication from Rovenich (Tolbiacum), now lost, by one Celorius Iustus, perhaps concealing a misreading of this name and referring to the same person (CIL 13.7937, VRR II, 31).