Necessary Conditions for a Just War

I'm off to climb a mountain today (Mt. Chocorua, in southeast New Hampshire), so I only have time to post something brief, a small question about language. Today I'll pose the question, and tomorrow I'll give a possible solution.

In Cicero's discussion of ius ad bellum in De Officiis, he refers to the sacred 'fetial' code of Rome, and says the following:

Ex quo intellegi potest nullum bellum esse inustum, nisi quod aut rebus repetitis geratur aut denuntiatum ante sit et indictum (I. 36).

That is, he mentions three conditions, but he groups them thus: R or (D and I).

The problem is, there is independent evidence that all of these conditions were necessary conditions, in the fetial code, for a war's being just. Also, in his own De Republica, Cicero writes as if he believes that as well (I quote from memory, but the sentence is more or less the following):

...nullum bellum iustum habetur, nisi rebus repetitis, nisi denuntiatum, nisi indictum.

If we rule out that Cicero changed his mind about the interpretation of the code, then it may look as though he was just sloppy in De Officiis (after all, it was written in ten days), and that he wrote 'or' when he meant 'and.'

This is the view taken by Miriam Griffin, the editor of the Cambridge Text. The translation (by Margaret Atkins) reads with literal correctness as follows: "From this we can grasp that no war is just unless it is waged after a formal demand for restoration, or unless it has been formally announced and declared beforehand." But Griffin explains in her note:

The old Roman practice was for the priesthood of the fetiales to deliver an ultimatum to the enemy demanding compensation for his alleged oppression. If no satisfaction was forthcoming, a threat of war was announced and war was then formally declared by the Roman assembly. C.'s 'or' here is inexact: he means all three conditions to apply (cf. Rep. III.23 and 25).

Yet I wonder if we need to regard Cicero's language as inexact. Nisi quod means 'unless', which can be understood, equivalently, as 'if not'. But then suppose that the negation, the 'not' , gets distributed across the alternative, aut ... aut.... Then, by DeMorgan's law (that not-P or not-Q is equivalent to not (P and Q)), it becomes equivalent to the negation of a conjunction.

That is, not-R or not-(D and I)) is equivalent to not: (R and D and I), which is what we wanted. We simply need to distribute the negation, understood from nisi, across the alternation.

I posed this solution to two Latinists at the Mayweek seminar. The one said: "That sounds right to me. The scope of negation is notoriously flexible in ordinary language, in Latin as in English." (He had in mind such sentences, familiar to logicians, as "Everything that glitters is not gold", which means, precisely, "Not everything that glitters is gold.") The other said to me: "Nope. Won't work. Some things you just can't do. That just isn't allowed in Latin."

I cannot decide this myself, and I would not wish to have to weigh these authorities against each other. I'll suppose then, for the time being, that this solution won't work, and I'll propose an alternative solution tomorrow.

(Yes, I know, this is not the most serious difficulty in the world. But it has been bothering me, and, as I said, I don't have time today for anything more.)