1 The Blue and Brown Books, (Basil Blackwell, 1958), pp. 17–18.

2 Philosophical Investigations, (Basil Blackwell, 1953), Part I, sections 66-67. Hereafter the Investigations will be abreviated as Phil. lnv. and numbered references will refer to sections of Part I.

3 The doctrine is not entirely new but has its roots in the scholastic notion of analogy, which may well have been Wittgenstein's source. Alternatively, Pitcher, George The Philosophy of Wittgenstein (Prentice-Hall, 1964), p. 218Google Scholar, suggests he may have got the idea from James, William’ Varieties of Religious Experience, (Longman, 1911), pp. 26–8,Google Scholar where it is used in connection with the terms ‘religion’, ‘religious sentiment’ and ‘government’. The earliest modern treatment of such a notion that I known of is by Mill, J. S. A System of Logic (London, 1843)Google Scholar, Bk. I, Chapt. 8, §8; and the first use of the term ‘family likeness’ is, I think, by J. F. Moulton in a lecture to the Royal Institution in 1877. Cf. l. Bragg, and Porter, G. (eds.) Royal Institution Library of Science: Physical Sciences,(Elsevier, 1970), vol. II, p. 529.Google Scholar

4 Wittgenstein does sometimes take account of this, for example when he says: ‘What is common to all … and what makes them into … [They) are related to one another in many different ways. And it is because of this relationship, or these relationships, that we call them all .. .’ (Phil. Inv. 65; first and third italics mine).

5 ‘Family Resemblance Predicates’, American Philosophical Quarterly, 1965, pp. 238–244.

6 (d) might actually be counter-intuitive in certain cases. For example, we might expect that a very complex game would have a larger bracketed sub-set than a very simple game. At any rate it is not at all clear that all games will have even approximately the same numbers of marks of game-hood.

7 ‘Universals and Family Resemblances’, Proceeding of the Aristotelian Society, 1960-61, pp. 207–22. Reprinted in Pitcher, George (ed.) Wittgenstein: The ‘Philosophical Investigations’, (Macmillan, 1968), pp. 186–204.Google Scholar Subsequent references will be to the latter.

8 Zettel, ed. Anscombe, G.E.M. and Wright, G. H. von (Basil Blackwell, 1967)Google Scholar.

9 Remarks on the Foundations of Mathematics, ed. Wright, G. H. von Rhees, R. and Anscombe, G. E. M. (Basil Blackwell, 1956).Google Scholar

10 ‘Family Resemblance’, Philosophical Quarterly, 1967, p. 64.

11 The third interpretation is contextually implausible; if this was what Wittgenstein wanted to say he would surely not have used ‘because of’. Pompa admits in parenthesis ‘this would be rather a strange use’ (ibid.).

12 Mill, discusses these views in A System of Logic,(London, 1843)Google Scholar Bk. I, Chapt. 8, 8, and dismisses them as being of only peripheral interest to the philosopher.

13 Assuming, as Bambrough claims, his theory is a family resemblance theory and solves the problem of universals, then it is inconsistent. For a family resemblance solution to the problem of universals requires that for each family resemblance term ‘F’, all and only the things which are F have a family resemblance in common. Yet the major thesis of Bambrough's theory is that they have nothing in common apart from being F.

14 In fact Wittgenstein was trying to use the example of a game, through the notion of family resemblance, to elucidate the concept of a language-that is, he sought, through his analysis of ‘game’, to provide an analysis of ‘language’ by analogy. But this analogy breaks down for different reasons suggested by Rhees, Rush Discussions of Wittgenstein, (Routledge and Kegan Paul, 1970), especially pp. 71–2,Google Scholar 74; where a fundamental disanalogy between the explication of ‘game’ and the explication of ‘language’ is persuasively argued.

15 There may be a terminological problem here as the phrase ‘family resemblance predicate’ may be used both to signify those predicates the members of whose reference classes simply have the requisite criss-crossing and overlapping relationships; and to signify those predicates for which such relationships not only exist but are sufficient to draw the boundary of the predicate. In the first sense the boundaries of the predicate are drawn by some other means. A literal family resemblance employs this first use, for the members of the family have the criss-crossing and overlapping similarities but the boundaries of the group are drawn by the blood relationship between members. With this use I have no quarrel, but with the other sense I take issue in what follows. Where there may be a possibility of confusion over the two uses I shall refer to the second as a ‘fully fledged family resemblance predicate.’

16 ‘ “Something Common”,’ journal of Philosophy, 1962, p. 829.

17 ‘Are games all “amusing” [unterhaltend]? Compare chess with noughts and crosses.’ (Phil. lnv., 66).

18 There are exceptions to this, of course. Consider the form of disjunctive essentialism just mentioned; or the definition of ‘elector’ in a fancy franchise.

19 Compare the definition in the Concise Oxford Dictionary (p. 501) ‘Contest played according to rules and decided by skill, strength or luck’. As ‘played’ conveys the element of pleasureableness this definition is quite close to my own: at least the principle involved is the same. If common language is enshrined in dictionaries then my definition of ‘game’ is closer to it than Wittgenstein's. Cp. also the legal definition given by Widgery, Lord in the case Wilson, Adcock V.(1967) quoted in Saunders, J. B. (ed.) Words and Phrases Legally Defined, (2nd edn., Butterworth, 1969) vol. ii, pp. 310–311.Google Scholar20 I am grateful to Mr. Roger Montague and Mr. Michael Scholar, of Leicester University, and to members of the Philosophy Department of the Australian National University for the care with which they have commented on earlier versions of this paper.