A CONTRAST WITH GOODMAN'S SOLUTION

In offering his own solution, Goodman, like me, allows the possibility (whether logical or physical) that persons exist who are justified in pro­jecting the property grue with respect to emeralds after examining emer­alds only before 2500.

Briefly, Goodman’s solution is based on the idea that the term “green” is much better entrenched than “grue.” The term “green” (as well as other terms true of the same class of things) has been used much more fre­quently than “grue” (or other co-extensive terms) in hypotheses of the form “All As are Bs” that have actually come to be adopted. Goodman’s question is this: When is a hypothesis of the form “All As are Bs” pro- jectible, that is, when is it confirmed by instances consisting of reports that particular As are Bs?

Suppose that two conflicting hypotheses “All As are Bs” and “All As are Cs” are such that all their examined instances are true. But suppose that the term B is much better entrenched than the term C. Then, according to Goodman, the hypothesis “All As are Cs” is not projectible. It receives no confirmation from its instances. Thus, although all the examined instances of the hypothesis “All emeralds are grue” are true, that hypothesis does not receive confirming support from those instances. The reason is that this hypothesis is “overridden” by the conflicting hypothesis “All emeralds are green,” which (up to now) has equal numbers of examined instances but uses the better entrenched term “green” and conflicts with no hypotheses with still better entrenched terms. Under these circumstances examined instances of green emeralds confirm the hypothesis that all emeralds are green, whereas examined instances of grue emeralds fail to confirm the hypothesis that all emeralds are grue.

On this solution it is at least logically, if not physically, possible that persons exist for whom examined instances of grue emeralds confirm the hypothesis that all emeralds are grue, whereas examined instances of green emeralds fail to confirm the hypothesis that all emeralds are green.

Goodman’s solution appeals to entrenchment. Although all the emer­alds examined so far are both green and grue, “green” is a much better en­trenched term. It appears much more frequently than “grue” in hypotheses of the form “All As are Bs” that we have come to accept. This claim I do not want to deny. My question, however, is why this is so. Why have we ac­cepted generalizations of the form “All As are green” or “All green things are B” much more frequently than “All As are grue” and “All grue things are B”? My solution offers an answer. (Goodman simply accepts that this is so.)

The answer is that for us grue is a disjunctive property of types (1) and (2) of section 2, whereas green is not. (For us, conditions (a)-(d) of sec­tion 3 hold.) Accordingly, for us, to generalize from examined instances of grue to hypotheses of the form “All As are grue” and “All grue things are B” (where A and B are not for us disjunctive properties of types (1) and

(2) ), we need to examine As (for “All As are grue”) and grue things (for “All grue things are B”) both before and after 2500. Since for us grue is a disjunctive property of type (2), in order to generalize we need to vary the instances and examine both things that satisfy the condition C of a property of type (2) and things that fail to satisfy C. Since green is not for us a disjunctive property of types (1) and (2), in order to generalize from examined instances of green to hypotheses of the form “All As are green” and “All green things are B” (where A and B are not for us disjunctive properties of types (1) and (2)) we do not need to examine As (for “All As are green”) and green things (for “All green things are B”) both before and after 2500, or ones that satisfy some corresponding condition C and others that fail to.

Accordingly, my solution is not based on the idea of entrenchment, which is really an idea about terms used in generalizations we have come to accept. It is based on the idea that for us, because grue is a disjunctive property of a certain sort, whereas green is not, in order to generalize from examined cases of emeralds that are grue we need to examine em­eralds that satisfy one side of the disjunction and emeralds that satisfy the other.

5.