This, then, is the second trouble — the unskilful composition of
many of the epigrams. Errors of transmission constitute a third
difficulty. For even inscriptions are not absolutely authentic. Be-
tween us and the author of an epigram on an Attic stele there do
not stand, it is true, a dozen blundering copyists; but there does
stand the stonecutter. Now the Greek stonecutter was a very dread-
ful fellow. He chipped recklessly ahead; if he left out a letter or
cut a wrong one he seldom tried to correct it; he transposed the
words ; he misread his copy, or deliberately tinkered it. Thus he
cut SET1MQ for o-' (n. 48), AEPETHS for St dpe-ri/s; (n.
56), irapoi ixTepLcrav for tTapoi Kreptcrav (n. 183); by inserting a redun-
dant re in
£vvov 'AOavoSiapov re ko.1 'AaioTToSuipov toSe pepyov
(n. xcv) he made an already faulty verse atrocious. A luculent case is
in n. 58a RM. The copy had HAE0ANEN, which was meant to be
178' tOavcv, but the graver took it as tjSe. Oavdv, without reading the
context; so he undertook to improve the spelling of the infinitive
(it was just at that time when EI was beginning to replace E in such
words), and cut HAE0ANEIN. On a still worse muddle, in n. 96,
see the foot-note on p. 128.
A singular case is that of the paean of Isyllus (n. xcvn4). The
shape in which it stands on the stone cannot be exactly the shape in
which it was composed. The aberrations will be pointed out on
P- 192 : the original in three cases is certain, in another doubtful.
Now the noteworthy thing is that these do not look like stonecutters'
blunders. Not only is a certain semblance of sense everywhere pre-
served, but — what is more remarkable — the Ionic metre is nowhere
°n pp. 117, 118. A similar masculine in -as or -tjt could be put in the vocative, as
EuOi'a oiiK, 3S (cp. 65), or the genitive (see examples on pp. 116—118). Or an un-
contracted form could be used, as A-np.oip6uy, S6; or, on the other hand, contrac-
tion or synizesis be resorted to: nvSiW, 26: compare the examples on p. 104.
Sometimes an archaic form helped out; so Ee!<:wpaT7js appears as =.eifoKpaTijs,
768 a pref.; and Aaixaaay6pas as AaficuraaytSpcis, 234. A more desperate case
like 'SrpaTela (.205) induced neglect of position; see p. 79. Finally, the name
•s not unfrequently forced in, with absolute violation of natural quantity: NtKtas,
AeavoSupov; see p. 75. A more circumspect poet chose another metre — iambic
trimeter, or some combination: on this see p. 44. The device of dividing a proper
name between two verses (Simonides, frag. 131 Bgk.) is not found in our inscrip- '
lions, but occurs in the later epigram, Kaibel n. 805 a add. = CIG. 5974.