Xen. Memor. 1, 7, 3

[Constantinus Philo]

Δηλον ότι κυβερναν κατασταθεις ο μη επισταμενος απολεσειεν αν ους ηκιστα βουλοιτο. In this sentence there is an assimilation of the relative to the optative of the principal clause. Therefore this must be a conditional relative clause, because assimilation takes place only in cond rel clauses. However it is difficult to see it as conditional, or maybe it's a general conditional clause. In any case I cannot see how ους can be resolved into ει Τίνας , or can it?

[Hylander]

It’s optative because it’s part of the potential optative idea, same as απολέσειεν.

[Hylander]

You are looking at Smyth 2186a:

2186. Assimilation to the Optative.—When an optative of the principal clause refers to future time (potential optative and optative of wish), the subordinate clause takes the optative by assimilation in the following cases.

a. Conditional relative clauses (regularly): πῶς γὰρ ἄν (1832) τις, ἅ γε μὴ ἐπίσταιτο, ταῦτα σοφὸς εἴη; for how could any one be wise in that which he does not know? X. M. 4.6.7, τίς μισεῖν δύναιτ᾽ ἄν ὑφ᾽ οὗ εἰδείη καλός τε καὶ ἀγαθὸς νομιζόμενος; who could hate one by whom he knew that he was regarded as both beautiful and good? X. S. 8. 17, ““ἔρδοι τις ἣν ἕκαστος εἰδείη τέχνην” would that every man would practise the craft that he understood” Ar. Vesp. 1431, τίς ἂν . . . μόλοι (1832), ““ὅστις διαγγείλειε τἀμ᾽ εἴσω κακά” would that some one would come to report within my tale of woe” E. Hel. 435.

N. 1.—If the relative has a definite antecedent, assimilation does not take place; but not all relative clauses with an indefinite antecedent are assimilated. Cp. ““ὥσπερ ἂν ὑμῶν ἕκαστος αἰσχυνθείη τὴν τάξιν λιπεῖν ἣν ἂν ταχθῇ ἐν τῷ πολέμῳ” as each one of you would be ashamed to leave the post to which he may be appointed in war” Aes. 3.7.

I question whether Smyth's formulation of rules on this point, and maybe the term "assimilation" too, obscure to some extent what's really going on when the verb in a subordinate clause is "assimilated" to the optative of the principal clause.

To my mind, when the antecedent of the relative clause is indefinite, as in the sentence from Xenophon, the relative clause is likely to be part of the potential, and therefore unreal, optative idea of the main clause, and therefore belongs in the optative along with the verb of the principal clause.

Conversely, when the relative clause refers to a specific, real antecedent, the relative clause is typically not part of the potential optative idea of the principal clause and belongs in the indicative.

In the example in Smyth's Note 1, the conditional/general nature of the relative clause would be lost if the subjunctive were to be recast as an optative (without ἂν). The idea, I think, is that every one of the jurors would be ashamed to desert any post whatsoever to which he had been assigned, reinforcing the idea that only a shameless coward (Demosthenes) would desert his post.

[mwh]

I too have reservations about Smyth’s formulations and his terminology, which to my mind make things unduly complicated and messy. There are subordinate clauses with optative, and subordinate clauses with subjunctive+ἄν (and of course subordinate clauses with indicative, and with plain subjunctive, and with opt.+ἄν)—just as there are main clauses with a variety of constructions (essentially indic., subj.., opt.,, indic.+ἄν, opt.+ἄν). Each construction has its own semantic function. It makes no difference to the syntax whether a given subordinate clause is a relative clause or a conditional clause or a temporal clause or whatever else; it’s just the introductory word (ὅς etc.[relative], εἰ [conditional}, ὅτε [temporal], ἐπεί [temporal/causal], etc.) that determines that.

In the case of πῶς γὰρ ἄν τις, ἅ γε μὴ ἐπίσταιτο, ταῦτα σοφὸς εἴη, for example, we have opt.+ἄν (“potential” opt.) in the main clause and opt. in the subordinate clause, a perfectly regular pattern. Same again with Δηλον ότι κυβερναν κατασταθεις ο μη επισταμενος απολεσειεν αν ους ηκιστα βουλοιτο and with τίς μισεῖν δύναιτ᾽ ἄν ὑφ᾽ οὗ εἰδείη καλός τε καὶ ἀγαθὸς νομιζόμενος; and with τίς ἂν . . . μόλοι ὅστις διαγγείλειε τἀμ᾽ εἴσω κακά; So too with ἔρδοι τις ἣν ἕκαστος εἰδείη τέχνην except that there the main verb is not potential but true optative (no ἄν).

In the case of ὥσπερ ἂν ὑμῶν ἕκαστος αἰσχυνθείη τὴν τάξιν λιπεῖν ἣν ἂν ταχθῇ ἐν τῷ πολέμῳ (Smyth’s N.1) we have another potential optative in the (quasi-main) ὥσπερ clause but subj.+ ἄν in the subordinate clause, serving its invariable function. (ἣν ἂν ταχθῇ “that he’s assigned [whatever that may be]”.)

[Constantinus Philo]

If anyone can read Kuhner can he tell us what he says on the point. I think German grammarians could be more accurate.

[Constantinus Philo]

There is also Crito 45b μη δυσχερές σόι γενεσθω ότι ουκ αν έχεις ο τι χρωο σαυτω where χρωο is explained as replacing dubitative subjunctive in a secondary séquence. This is not a secondary séquence however because both subjunctive γενεσθω and optative εχοις we would count as primary tenses. Therefore this is another example of assimilation to optative in a rel. clause.

[mwh]

ουκ αν έχoις ο τι χρωο σαυτω. (You mistakenly typed έχεις. And γενεσθω is not subjunctive.)
This is just one more example of opt. in a subordinate clause depending on opt.+αν in the leading clause, just like the one you started with, απολεσειεν αν ους ηκιστα βουλοιτο, and the examples in Smyth. You can call it assimilation if you want, or attraction, or quasi-secondary sequence, or nothing at all. It’s perfectly regular.
(I haven’t checked Kühner, but that will not be more “accurate” than Smyth, which is highly derivative.)
The only difference is that in the Crito the reader will understand the optative as representing a deliberative subjunctive, but that difference is erased in the sentences as they stand.