RELATIVE* AND INTERRO&ATIVES. 69
such words, like ocrrts, are regularly used in indirect
questions ; even 6's is occasionally found (and that
as early as in Plat. Rep. viii. p. 559&) in a sentence
where oVris or even rt's might have been expected :
but the step from vague instances like these to the
use in a direct question seems a hard one to take.
If defensible at all, it must be explained from the
influence of Latin, in which the relative pron. is
always in form nearly akin to the interrogative,
and may itself in certain cases (of which this is not
one) be used interrogatively.
"Oo-ri? is, in the N. T., as a rule confined to two of
its relative usages the indefinite one, almost exactly
expressed by the Latin quicumque or the English
" whosoever," and a less strictly definable one, cor-
responding to the Latin qui with the subj., and
capable of various more or less adequate translations
in English according to the context or the exact
shade of meaning " which,"* " such as," " such that,"
or " seeing that he. ..." Only in Acts ix. 6 (true
text) OTI is used in an indirect question, in the wide
grammatical sense of the term.
Of the cases where rt? is used and ocms might
have been, the most defensible cases, those where the
usurpation of the functions of the rel. is least, are
those where it follows e^etv or rather OVK e^ctv Matt,
xv. 32= Mark viii. 1, 2; cf. vi. 36. These are not
more harsh than e.g., Mark ix. 6, ov yap ?;6Vi TL
\a\rjcrrj
: and in fact this phrase has unquestionable
classical precedents, Soph. (E. C. 317; Xen. Hell. I.
* In the archaic use of "which," according to which it is
not exclusively neut., it differs from "who" almost exactly
as 6'<ms from 6s.