Most reasoning about matters of equality, distributive justice and universalizability follows the same pattern in which a statement about a certain form of equality is succeeded by some kind of exceptive clause (or adverbial clause of open condition), usually starting with unless or except. For example, the claim that moral principles are universalizable must be understood in the following way --it has been said: if one person ought to do x, everyone else ought to do x 'unless there are relevant differences' between the two persons and/or between their situations. This principle 'that no person should be treated differently from any or all other persons unless there is some general and relevant reason which justifies such a difference in treatment' has been called "a fundamental principle of morality, if not of rationality itself". Another theorist has expressed the same idea by stating that it is 'part of the concept of morality that no person is simply excluded from moral consideration'. Being excluded altogether is distinguished here from being considered differently. But also in the latter case 'the difference made must be justified by some morally relevant ground of distinction'. What it all amounts to is that people are 'to be treated equally, except when unequal treatment can be justified'. A very special justification which has been mentioned in this context is that the unequal treatment may promote greater equality in the long run. Where this is indeed the case, it is a certain kind of inequality which serves equality.

What most, if not all, of those philosophizing on equality, justice and related issues seem to maintain, superficially speaking, is that people or cases are (relevantly) equal, unless (relevantly) different. Purely truth-conditionally (relevantly) equal, unless (relevantly) different and (relevantly) different, unless (relevantly) equal are tautologies which signify the same: nothing of interest. But we are not just dealing with two (parts of) propositions which are true under all circumstances; we are dealing with people (ethical theorists, for instance) who choose to utter the one proposition and not the other. Their utterances (or inscriptions) are moves in a conversational game. Someone saying "(relevantly) equal, unless relevantly different" does say something of interest, because 'e does not just choose any tautology, but 'e draws a distinction between this one and all others for a reason. ( Compare someone saying "a promise is a promise" at a particular place and time.)

It is a maxim of linguistic pragmatics that a person should be relevant every time 'e contributes something to a conversation. The maxim is part of a general principle governing people's conversation which is called "the cooperation principle". It is on the basis of this principle and maxim that the equal, unless different pattern is selected, rather than a different, unless equal pattern. The only sensible interpretation of the equal, unless pattern is, then, that people or things have to be treated (relevantly) alike, unless there can be shown to exist a (relevant) difference which justifies treating them in a different way. Almost all writers on ethical issues like equality and justice explicitly speak of departures from equality and of differences having to be justified (for example, by stating that it is unequal treatment which requires justification). For all of them the burden of proof rests with the one who wants to make a difference to prove that such a difference is relevant. This means quite something else than different, unless .., when the burden of proof would rest with someone who wants to treat cases or people alike. That person would have to prove that no distinction on the basis of whatever factor were relevant, and particularly, that a distinction or set of distinctions drawn by 'er opponent were irrelevant (assuming that the opponents do agree on the focus of relevancy itself ).

That unequal treatment may promote greater equality in the long run is only one kind of justification. Other justifications which have been mentioned rest, for example, on considerations of beneficence or utility, and on people's differing needs, situations or preferences. With respect to citizenship it has been argued that everyone should be 'treated as capable of citizenship unless very strong evidence of personal incapacity is produced', such as 'infancy and insanity'. (But --it is added-- 'social characteristics' such as 'poverty and race' have not been 'accepted generally as good reasons for exclusion'.) That the burden of proof itself lies with those who want to treat people or things differently cannot be proved; at the most it can be proved that individual theorists believe it to lie there.

The advantage of the egalitarian formula that a difference made has to be justified and may always be disputed, is that it is an action, not an omission or nonaction, which requires a justification. The burden of proof, then, even rests with those who want to treat cases relevantly alike by treating them dissimilarly in relevant respects. (If the egalitarian has thought about the other side to the picture at all.) A serious problem is that no-one has ever provided an exact criterion, or set of criterions, to verify relevance, just as no-one can give a criterion to verify universal nonanalytical statements. (In division 4 of this chapter we will only discuss some criterions for falsifying relevance judgments.) It follows that an egalitarian cannot require someone who makes a distinction to prove in the strict sense that this distinction is relevant, in any way, discriminationally relevant. (We will see that statistical relevance is not sufficient.) An anti-egalitarian might therefore object that it is the egalitarian who has to prove the irrelevance of the distinction. This is the different, unless approach. Such an objection would be unreasonable, however, since it requires 'im to prove the irrelevance of a distinction made by others after it has been made, or which 'e 'imself does not want to make. Moreover, altho it is possible to prove irrelevance in some cases, this does not mean that every instance of irrelevance can even theoretically be proved. In practise the possibilities are even limited to a much greater extent, if present at all.

Whereas actual discriminational relevance can never be irrefutably proved (if this is right), it is a proof of potential relevance which can be demanded in the equal, unless procedure. Statistical relevance is such a proof. The subsequent assumption of actual relevance may then not be contested, by proving the irrelevance of the distinction, for instance. This seems to be a weak point in the whole procedure. Yet, the question is not so much whether it is weak or strong, but whether it is weaker than another spot in the total scheme, namely the choice of focus, which must not be contested either. And presumably, it is not.

A traditionalist approach to the burden-of-proof issue would be to require anyone who deviates from traditional practise or convention to prove that a difference is irrelevant altho 'everyone' makes it, or to prove that it is relevant altho 'no-one' makes it. It is a viewpoint which makes normative thought, and moral reasoning in particular, depend on what the majority in a certain (sub)culture or social environment have 'always' taken to be relevant and irrelevant, right and wrong. It stresses factual morality in a historical or sociological sense, and assigns normative significance to mere belief, coherent or not coherent. Within the framework of normative philosophy, however, only an equal, unless tenet constitutes a systematic code, or for that matter, a different, unless tenet.

To require that it is a material difference of treatment of which the relevance has to be made plausible, represents a kind of idea of equality (and justice) which is nothing else than a relevance principle interpreted in an equal, unless manner, nothing else than some 'principle of nondiscrimination'. It is not to be concerned yet with the content of the focus of relevancy, that is, with the kind of goal, purpose or other directional entity in question. A different sort of notion of equality (and justice) is it which entails that equality itself should be a focus. Thus according to the relevance principle under the equal, unless interpretation one must treat the rich and the poor alike, unless their wealth makes a relevant difference in respect of some goal recognized as a focus. The richness and poorness of people is, then, still taken for granted, however big the difference in wealth may be. And any goal could be a focus of relevancy, whether pertaining to an economic quantity or not. But if the relevance principle is combined with a separate principle of equality determining that the equality of wealth (the incomes and/or assets of people) ought to be a goal in itself, one must not only treat the rich and the poor in a relevantly similar way, but one must also strive for making them equally rich ( or 'equally poor' -- it might be sarcastically commented ). All distinctions made to promote this value now become relevant, for example, to have a rich person pay disproportionately more taxes than a poor person, the difference in wealth thus being lessened as much as possible.

Altho those who speak of "equality" and "(distributive) justice" turn out to use the phrases relevant and morally relevant with the greatest ease, the above example should show how important it is that they clearly state the following two things: (1) whether they recognize only relevance or both relevance and equality (as a focus of relevancy); and (2), where for them the burden of proof lies. To make such a precise statement is conceptually possible, altho it requires a minimum insight into the role of relevancy itself. To obtain this minimum insight we should also look at other disciplines than ethics or normative philosophy, even tho this is going to confirm our expectations.