The Principle of Double Effect is often presented as a four-part conditional test of the morality of an act that has foreseeable negative effects. For example:
Classical formulations of the principle of double effect require that four conditions be met if the action in question is to be morally permissible:
first, that the action contemplated be in itself either morally good or morally indifferent;
second, that the bad result not be directly intended;
third, that the good result not be a direct causal result of the bad result; and
fourth, that the good result be "proportionate to" the bad result.
Supporters of the principle argue that, in situations of "double effect" where all these conditions are met, the action under consideration is morally permissible despite the bad result.
-- Wm. David Solomon, "Double Effect," The Encyclopedia of Ethics, Lawrence C. Becker, editor [reformatted for this post]
This is, as Wm. David Solomon says, the classical formulation, and it's the one I've always found and used in my own discussions on double effect. But now I'm wondering whether it isn't somewhat fussier than strictly necessary.
Look at the third condition. That is, Solomon observes, "the so-called Pauline principle, 'One should never do evil so that good may come.'" But the Pauline principle is valid apart from any double effect principle, so if the third condition is violated, then the first condition is also violated.
Whether or not it's necessary to call out the second and third conditions explicitly, the fourth condition seems to be the heart of the principle. It says that -- in principle -- evil and good effects are commensurate, and the relation between them determines (all else being morally good) whether a particular act is morally licit.
But it's also important to note what it doesn't say, which is how to measure the relation between the effects. That's fine, because for all the talk of "double-effect" reasoning, the Principle of Double Effect itself doesn't talk about how the reasoning occurs.
And it doesn't have to. No one needs to be told that, if the bad effect "outweighs" the good effect, the act ought not be performed, because no one would perform an act if they thought the bad effect outweighed the good.
So when we talk amongst ourselves about whether this or that act, with foreseeable good and evil effects, is morally licit, we shouldn't talk as though the Principle of Double Effect provides a heuristic for settling that question. We need to judge whether the fourth condition is met by means the Principle doesn't provide; at best, it provides reason to think those means exist.
I'll also point out that, in the first condition, judging the moral nature of the act "in itself" is not limited to judging whether the act is evil in its object. Circumstances other than the effects of an act can make an act immoral, even if its object is not. You must account for those circumstances somewhere, and in the above formulation the place to do that is under the first condition.