but interrogative. In the following example, confusion of the two leads to an improper coordination.
What sums he made can only be conjectured, but must have been enormous.—Macaulay.
In the first sentence, 'what' is an interrogative, in the second, a condensed antecedent-relative, standing for 'the sums that'. It is the sums that were enormous: it is the answer to the question 'What sums did he make?' that can only be conjectured. The mistake is possible only because 'can' and 'must' do not reveal their number: 'can' is singular, 'must' plural.
The differentiation between the two whats and their equivalents is not, indeed, complete: just as the condensed antecedent-relative resembles in form, though not in treatment, the unresolved interrogative, so the interrogative, by resolution into 'the (that) which', not only resembles, but is grammatically identified with, the uncondensed relative and antecedent. The resolution is, no doubt, convenient: it should be noticed, however, that the verbs with which alone it can be employed (verbs that may denote either perception of a fact or other kinds of perception) are precisely those with which ambiguity may result. 'I know the house (that) you mean': it may (antecedent and relative) or may not (resolved interrogative) follow that I have ever seen it. 'We must first discover the scoundrel who did it'; antecedent and relative? then we must secure the scoundrel's person; resolved interrogative? then only information is needed. 'I can give a good guess at the problem that is puzzling you': and the solution?—I know nothing of the solution; I was resolving an interrogative.
This, however, does not affect sentences like the Macaulay one above: for although the resolved or uncondensed forms ('the...which') are grammatically identified, the condensed or unresolved forms ('what') are not.
(iv) The omission of the relative in isolated clauses (as opposed to coordinates) is a question not of correctness
