The town site that we can best examine for our present purpose is Calleva or Silchester, ten miles south of Reading, which has been completely excavated with care and thoroughness.  Here a few fairly complete inscriptions on stone have been discovered, and many fragments of others, which prove that the public language of the town was Latin.[1] The speech of ordinary conversation is equally well attested by smaller inscribed objects, and the evidence is remarkable, since it plainly refers to the lower class of Callevans.  When a weary brick-maker scrawls SATIS with his finger on a tile, or some prouder spirit writes CLEMENTINVS FECIT TVBVL(um) (Clementinus made this box-tile), when a bit of Samian is marked FVR—­presumably as a warning from the servants of one house to those of the next—­or a rude brick shows the word PVELLAM—­probably part of an amatory sentence otherwise lost—­or another brick gives a Roman date, the ’sixth day before the Calends of October’, we may be sure that the lower classes of Calleva used Latin alike at their work and in their more frivolous moments (Figs. 2, 3, 4).  When we find a tile scratched over with cursive lettering—­possibly part of a writing lesson—­which ends with a tag from the Aeneid, we recognize that not even Vergil was out of place here.[2] The Silchester examples are so numerous and remarkable that they admit of no other interpretation.[3]

[Footnote 1:  For these and for the following graffiti see my account in the Victoria History of Hampshire, i. 275, 282-4.  For the ‘Clementinus’ tile (discovered since) see Archaeologia, lviii. 30.  Silchester lies in a stoneless country, so that stone inscriptions would naturally be few and would easily be used up for later building.  Moreover, its cemeteries have not yet been explored, and only one tombstone has come accidentally to light.]

[Footnote 2:  Sir E.M.  Thompson, Greek and Latin Palaeography (1894), p. 211, first suggested this explanation; Eph. ix. 1293.]

[Footnote 3:  To call them—­as did a kindly Belgian critic of this paper in its first published form—­’un nombre de faits trop peu considerable’ is really to misstate the case.]

[Illustration:  FIG. 2. ... puellam.]

[Illustration:  FIG. 3. Fecit tubul(um) Clementinus.]

[Illustration:  FIG. 4. vi K(alendas) Oct(obres)....]

[Illustration:  FIGS. 2, 3, 4.  GRAFFITI ON TILES FROM SILCHESTER. (P. 25.)]

[Illustration:  FIG. 5.  GRAFFITO ON A TILE FOUND AT SILCHESTER (P. 25). Pertacus perfidus campester Lucilianus Campanus conticuere omnes. (Probably a writing lesson.)]