My aim here is to investigate the role of global quantifiers—quantifiers ranging over the entire universe of sets—in the formalization of Zermelo-Fraenkel set theory. The use of such quantifiers in the formulas substituted into axiom schemata introduces, at least prima facie, a strong element of impredicativity into the thapry. The axiom schema of replacement provides an example of this. For each instance of that schema enlarges the very domain over which its own global quantifiers vary. The fundamental question at issue is this: How does the employment of these global quantifiers, and the choice of logical principles governing their use, affect the strengths of the axiom schemata in which they occur?

I shall attack this question by comparing three quite different formalizations of the intuitive principles which constitute the Zermelo-Fraenkel system. The first of these, local Zermelo-Fraenkel set theory (LZF), is formalized without using global quantifiers. The second, global Zermelo-Fraenkel set theory (GZF), is the extension of the local theory obtained by introducing global quantifiers subject to intuitionistic logical laws, and taking the axiom schema of strong collection (Schema XII, §2) as an additional assumption of the theory. The third system is the conventional formalization of Zermelo-Fraenkel as a classical, first order theory. The local theory, LZF, is already very strong, indeed strong enough to formalize any naturally occurring mathematical argument. I have argued (in [3]) that it is the natural formalization of naive set theory. My intention, therefore, is to use it as a standard against which to measure the strength of each of the other two systems.

A fundamental problem in the theory of abelian groups is to determine the structure of Ext(A, Z) for arbitrary abelian groups A. This problem was raised by L. Fuchs in 1958, and since then has been the center of considerable activity and progress.

We briefly summarize the present state of this problem. It is a well-known fact that

where tA denotes the torsion subgroup of A. Thus the structure problem for Ext(A, Z) breakdown to the two distinct cases, torsion and torsion free groups. For a torsion group T,

which is compact and reduced, and its structure is known explicitly [12].

For torsion free A, Ext(A, Z) is divisible; hence it has a unique representation

Thus Ext(A, Z) is characterized by countably many cardinal numbers, which we denote as follows: ν0(A) is the rank of the torsion free part of Ext(A, Z), and νp(A) are the ranks of the p-primary parts of Ext(A, Z), Extp(A, Z).

If A is free it is an elementary fact that Ext(A, Z) = 0. The second named author has shown [16] that in the presence of V = L the converse is also true. For countable torsion free, nonfree A, C. Jensen [13] has shown that νp(A) is either finite or and νp(A) ≤ ν0(A). Therefore, the case for uncountable, nonfree, torsion free groups A remains to be studied.

We give here alternative definitions for the notions that S. Shelah has introduced in recent papers: the dimensional order property and the depth of a theory. We will also give a proof that the depth of a countable theory, when denned, is an ordinal recursive in T.

In this paper, we prove a result that J. Y. Girard has conjectured for the last few years. That is,

This result is part of a study of ordinal notations. For the proof, we use some concepts introduced in Parts I, II and III of this article (see [GV1], [GV2] and [V]); in particular, in Part III (see [V]) we defined the category GAR of gardens (of type Ω), and we have shown that there is an isomorphism between GAR and the category DIL Ω of dilators which send Ω into Ω. To describe this isomorphism, we defined a functor SYN from GAR to DIL Ω and a functor DEC from DIL Ω to GAR which are inverse to each other. The functor SYN is constructed by induction on the height x of the garden Jx, iterating, at each step of cofinality Ω, the operation UN (unification of variables; see [G1, Chapter 3]). Here, we define the functor S⊿ from GAR to DIL Ω: this construction is very close to that of SYN, but we iterate, at each step of cofinality Ω, the operation ⊿ (identification of variables; see [G, Chapter 3]). Technically, the definition of S⊿ is easier and more natural than that of SYN (as ⊿ is easier than UN) but is not inversible (as ⊿); for example, if we consider the (regular) garden GεΩ+1 introduced in [GV2],S⊿(GεΩ+1) is exactly the dilator N = ⋃0<n<ωNn, with N1 = 2 + Id, and , but we are incapable of expressing SYN(GεΩ+1).

Let M be a term of the type free λ-calculus and let be a set of occurrences of redexes in M. A reduction sequence from M which first contracts a member of and afterwards only residuals of is called a development (of M with respect to ). The finite developments theorem says that developments are always finite.

There are several proofs of this theorem in the literature. A plausible strategy is to define some kind of measure for pairs (M, ), which—if M′ results from M by contracting a redex occurrence in and ′ is the set of residuals of in M′— decreases in passing from (M, ) to (M′, ′). This procedure is followed as a matter of fact in the proofs in Hyland [4] and in Barendregt [1] (both are covered in Klop [5]). If, as in the latter proof, the natural numbers are used as measures, then the measure of (M, ) will actually denote an upper bound of the number of reduction steps in a development of M with respect to .

In the present proof we straightforwardly define for each pair (M, ) a natural number, which can easily be seen to indicate the exact number of reduction steps in a development of maximal length of M with respect to .

Intuitionistic Zermelo-Fraenkel set theory, which we call ZFI, was introduced by Friedman and Myhill in [3] in 1970. The idea was to have a theory with the same axioms as ordinary classical ZF, but with Heyting's predicate calculus HPC as the underlying logic. Since some classically equivalent statements are intuitionistically inequivalent, however, it was not always obvious which form of a classical axiom to take. For example, the usual formulation of the axiom of foundation had to be replaced with a principle of transfinite induction on the membership relation in order to avoid having excluded middle provable and thus making the logic classical. In [3] the replacement axiom is taken in its most common classical form:

In the presence of the separation axiom,

this is equivalent to the axiom

It is this schema Rep that we shall call the replacement axiom.

Friedman and Myhill were able to show in [3] that ZFI has a number of desirable “constructive” properties, including the existence property for both numbers and sets. They were not able to determine, however, whether ZFI is proof-theoretically as strong as ZF. This is still open.

Three years later, in [2], Friedman introduced a theory ZF− which is like ZFI except that the replacement axiom is changed to the classically equivalent collection axiom:

He showed that ZF− is proof-theoretically of the same strength as ZF, and he asked whether ZF− is equivalent to ZFI.

One of the most fundamental results of higher-type recursion theory was first proved by Thomas Grilliot [3]. He showed that the functional embodying number quantification, 2E, is computable from a type-2 object F if and only if F has an effective discontinuity. For the background to this work we recommend Normann's book [9] or his summary [10], and for an example of its many applications Normann-Wainer [11].

The purpose of §1 of this paper is to give an effective proof of the “only if” part of this result: i.e. a construction of a discontinuity of F from the information 2E = {e}F. This contrasts with Grilliot's argument which is topological and nonconstructive, and shows that “Grilliot's trick” for computing 2E from a discontinuity of F is essentially the only method.

We then apply the method to two more general situations in which it gives new results. In §2 we treat type-3 objects in which 3E is computable (assuming AC and CH), and obtain a strengthening of the result of Hartley [5, §5]. In §3 we consider the superjump and characterise it as the least among a class of functionals with a particular sort of family of discontinuities.

In this paper it will be shown that HEO and HROE are isomorphic with respect to extensional equality. This answers a question of Troelstra [T, 2.4.12, p. 128]. The main problem is to extend effective operations to a larger domain. This will be achieved by a modification of the proof of the continuity of effective operations. Following a suggestion of A. S. Troelstra, similar results were obtained for ECF(U) and ICFE(U), where U is any universe of functions closed under “recursive in”.

Whenever many-sorted theories are discussed in logic texts (e.g., [3, pp. 483–485]), it is fashionable to observe that every many-sorted theory can be effectively replaced by an equally powerful one-sorted theory *. The theory * contains for each sort σ of a unary predicate symbol Sσ used to indicate that an individual is of sort σ; the nonlogical axioms of * are the “translations” of those of together with axioms asserting that there is at least one individual of each sort and that all function symbols behave properly with respect to sorts. This observation suggests that perhaps many-sorted theories are no more useful than one-sorted theories. That this is not always the case has been pointed out previously [1, p. 13]. The content of this note is that can be interpretable in without * being interpretable in *.

If a function symbol f in a many-sorted theory takes as its arguments n terms a1,…, an of sorts σ1, …, σn, respectively and produces a term fa1 … an of sort σ, then f is said to be of type (σ1, …, σn;σ). Likewise, a predicate symbol is of type (σ1, …, σn) if it takes as its arguments n terms of sorts σ1, …, σn. We assume that for each sort σ there is a predicate symbol = σ of type (σ, σ).

One long-range objective of logic is to find models of arithmetic with noteworthy properties, perhaps properties that imply some long-standing number theoretic conjectures. In areas of mathematics such as algebra or set theory, new models are often made by extending old models, that is, by adjoining new elements to already existing models. Usually the extension retains most of the characteristics of the old model with at least one exception that makes the new model interesting. However, such a scheme is difficult in the area of arithmetic. Many interesting properties of the fine structure of arithmetic are diophantine and hence unchangeable in extensions. For instance, one cannot change a prime number into a composite one by adjoining new elements.

One could possibly get around this diophantine difficulty in one of two ways. One way is to change the usual language of addition and multiplication to an equivalent language that does not transmit so much information to extensions. For instance, multiplication is definable from the squaring function, as one sees from the identity 2xy = (x + y)2 − x2 − y2, and the squaring function in turn is definable either from the unary square predicate (as one sees from the fact that n = m2 if n and n + 2m + 1 are successive squares) or from the divisor relation (as one sees from the fact that n = m2 if n is the smallest number such that m divides n and m + 1 divides n + m). Either of these two alternatives to multiplication might make for interesting extensions.

Let ℜ = (R, + R,…) be the ordered field of real numbers. It will be shown that the -theory of ℜ is decidable, where denotes the Malitz quantifier of order n in the ℵ1-interpretation.

In this article propositions will be identified with a certain kind of changes of belief. The intended interpretation is that a proposition is characterised by the change it would induce if added to a state of belief. Propositions will thus be defined as functions from states of belief to states of belief. A set of postulates concerning the properties and existence of propositions will be formulated. A proposition will be said to be a tautology iff it is the identity function on states of belief. The main result is that the logic determined by the set of postulates is intuitionistic propositional logic.

The basic epistemic concept is that of a belief model, which is defined as a pair 〈, 〉, where is a nonempty set and is a class of functions from to . The elements in will be called states of belief and they will be denoted K, K′,…. A discussion of the epistemological interpretation of the states of belief can be found in Gärdenfors [2]. Here, no assumptions about the structure of the elements in will be made.

The elements in will be called propositions, and A, B, C, … will be used as variables over . Functions from states of belief to states of belief can be characterised as epistemic inputs. The intended interpretation of the functions in is that they correspond to changes of belief where the new evidence is accepted as “certain” or “known” in the resulting state of belief. This means that not all functions defined on can properly be called propositions.

Craig's theorem is a result about the cardinality of a proper basis for the theory of combinators. Its proof given in [3] was shown to be incomplete by André Chauvin [2]. By using a different approach, we give a very short proof of this theorem. We use the notation of [1].

Definition 1. A combinator Q is proper if there exists a natural number n such that for arbitrary variables x1,…,xn we have the following contraction rule:

where C is a pure combination of the variables x1,…,xn. Q is to be understood as an abstract symbol, not as a combination of S and K's. Therefore Q comes with a contraction rule.

Definition 2. A set (Q1,…, Qm} of combinators is a basis for combinatory logic if for every finite set {x1,…, xk} of variables and every pure combination C of these variables, there exists a pure combination Q of Q1,…, Qm such that Qx1 … xk ↠ C.

Craig's Theorem. Every basis for combinatory logic containing only proper combinators contains at least two elements.

Proof. Let {Q} be a singleton basis for combinatory logic, and let us show that we cannot have combinatory completeness. This is an easy consequence of the next two lemmas.

Lemma 1. Q is a projection. That is, Qx1 … xn ↠ xj, for some j.

Proof. Let I be a proper combination of Q such that Ix ↠ x for a variable x, and let M be a term such that Ix ↠ M → x and M → x is a nontrivial contraction.

When dealing with axiomatic theories from a recursion-theoretic point of view, the notion of r.e. preordering naturally arises. We agree that an r.e. preorder is a pair = 〈P, ≤P〉 such that P is an r.e. subset of the set of natural numbers (denoted by ω), ≤P is a preordering on P and the set {〈;x, y〉: x ≤Py} is r.e.. Indeed, if is an axiomatic theory, the provable implication of yields a preordering on the class of (Gödel numbers of) formulas of .

Of course, if ≤P is a preordering on P, then it yields an equivalence relation ~P on P, by simply letting x ~Py iff x ≤Py and y ≤Px. Hence, in the case of P = ω, any preordering yields an equivalence relation on ω and consequently a numeration in the sense of [4]. It is also clear that any equivalence relation on ω (hence any numeration) can be regarded as a preordering on ω. In view of this connection, we sometimes apply to the theory of preorders some of the concepts from the theory of numerations (see also Eršov [6]).

Our main concern will be in applications of these concepts to logic, in particular as regards sufficiently strong axiomatic theories (essentially the ones in which recursive functions are representable). From this point of view it seems to be of some interest to study some remarkable prelattices and Boolean prealgebras which arise from such theories. It turns out that these structures enjoy some rather surprising lattice-theoretic and universal recursion-theoretic properties.

After making our main definitions in §1, we examine universal recursion-theoretic properties of some r.e. prelattices in §2.

As is well known, NF is a first-order theory whose language coincides with that of ZF. The nonlogical axioms of the theory are: Extensionality. (x)(y)[(z)(z ∈ x ↔ z ∈ y) → x = y].

Comprehension. (Ex)(y)(y ∈ x ↔ ψ) for every stratified ψ in which x does not occur free (a formula of NF is said to be stratified if it can be turned into a formula of the simple theory of types by adding type indices (natural numbers ≥ 0) to its variables).

Before stating our result, a few preliminaries are in order. Let T be the simple theory of types. If ψ is a formula of T, we denote by ψ+ the formula obtained from ψ by raising all type indices by 1. T* is the result of adding to T every axiom of the form ψ ↔ ψ+. A formula of T is n-stratified (n > 0) if it does not contain any type index ≥ n. A formula of NF is n-stratified if it can be turned into an n-stratified formula of T by adding type indices to its variables. (In practice, we shall allow ourselves to confuse an n-stratified formula of T with the corresponding n-stratified formula of NF). For n > 0, Tn (resp. ) is the subtheory of T (resp. T*) containing only n-stratified formulae. For n > 0, NFn is the subtheory of NF generated by those axioms of NF which are n-stratified. Let = 〈M0, M1,…,=, ∈〉 be a model of T.

The Principle of Dependent Choice is shown to be equivalent to: the Baire Category Theorem for Čech-complete spaces (or for complete metric spaces); the existence theorem for generic sets of forcing conditions; and a proof-theoretic principle that abstracts the “Henkin method” of proving deductive completeness of logical systems. The Rasiowa-Sikorski Lemma is shown to be equivalent to the conjunction of the Ultrafilter Theorem and the Baire Category Theorem for compact Hausdorff spaces.

Interpretability in reflexive theories, especially in PA, has been studied in many papers; see e.g. [3], [6], [7], [10], [11], [15], [26]. It has been shown that reflexive theories exhibit many nice properties, e.g. (1) if T, S are recursively enumerable reflexive, then T is interpretable in S iff every Π1 sentence provable in T is provable in S; and (2) if S is reflexive, T is recursively enumerable and locally interpretable in S (i.e. every finite part of T is interpretable in S), then T is globally interpretable in S (Orey's theorem, cf. [3]).

In this paper we want to study such statements for nonreflexive theories, especially for finitely axiomatizable theories (which are never reflexive). These theories behave differently, although they may be quite close to reflexive theories, as e.g. GB to ZF. An important fact is that in such theories one can define proper cuts. By a cut we mean a formula with one free variable which defines a nonempty initial segment of natural numbers closed under the successor function. The importance of cuts for interpretations in GB was realized already by Vopěnka and Hájek in [30]. Pioneering work was done by Solovay in [24]. There he developed the method of “shortening of cuts”. Using this method it is possible to replace any cut by a cut which is contained in it and has some desirable additional properties; in particular it can be closed under + and ·. This introduces ambiguity in the concept of arithmetic in theories which admit proper cuts, namely, which cut (closed under + and ·) should be called the arithmetic of the theory? Cuts played the crucial role also in [20].

We define, in the spirit of Fenstad [2], a higher type computation theory, and show that countable recursion over the continuous functionals forms such a theory. We also discuss Hyland's proposal from [4] for a scheme with which to supplement S1–S9, and show that this augmented set of schemes fails to generate countable recursion. We make another proposal to which the methods of this section do not apply.

In this paper we investigate the first order metatheory of the classical propositional logic. In the first section we prove that the first order metatheory of the classical propositional logic is undecidable. Thus as a mathematical object even the simplest of logics is, from a logical standpoint, quite complex. In fact it is of the same complexity as true first order number theory.

This result answers negatively a question of J. F. A. K. van Benthem (see [van Benthem and Doets 1983]) as to whether the interpolation theorem in some sense completes the metatheory of the calculus. Let us begin by motivating the question that we answer. In [van Benthem and Doets 1983] it is claimed that a folklore prejudice has it that interpolation was the final elementary property of first order logic to be discovered. Even though other properties of the propositional calculus have been discovered since Craig's orginal paper [Craig 1957] (see for example [Reznikoff 1965]) there is a lot of evidence for the fundamental nature of the property. In abstract model theory for example one finds that very few logics have the interpolation property. There are two well-known open problems in this area. These are

1. Is there a logic satisfying the full compactness theorem as well as the interpolation theorem that is not equivalent to first order logic even for finite models?

2. Is there a logic stronger than L(Q), the logic with the quantifierthere exist uncountably many, that is countably compact and has the interpolation property?

Nielsen [7] has proved that every subgroup of a free group of finite rank is free. The theorem was later strengthened by Schreier [8] by eliminating the finiteness restriction on the rank. Several proofs of this theorem (known as the Nielsen-Schreier theorem, henceforth denoted by NS) have appeared since Schreier's 1927 paper (see [1] and [2]). All proofs of NS use the axiom of choice (AC) and it is natural to ask whether NS is equivalent to AC. Läuchli has given a partial answer to this question by proving [6] that the negation of NS is consistent with ZFA (Zermelo-Fraenkel set theory weakened to permit the existence of atoms). By the Jech-Sochor embedding theorem (see [3] and [4]) ZFA can be replaced by ZF. Some form of AC, therefore, is needed to prove NS. The main purpose of this paper is to give a further answer to this question.

In §2 we prove that NS implies ACffin (the axiom of choice for sets of finite sets). In §3 we show that a strengthened version of NS implies AC and in §4 we give a partial list of open problems.

Let y be a set; ∣y∣ denotes the cardinal number of y and (y) is the power set of y. If p is a permutation of y and t ∈ y, the p-orbit of t is the set {pn(t): n is an integer}. If

we call p a cyclic permutation of y. If f is a function with domain y and x ⊆ y, f″x denotes the set {f(t):t ∈ x}. If A is a subset of a group (G, °) (sometimes (G, °) will be denoted by G) then A−1 = {x−1:x ∈ A} and [A] denotes the subgroup of G generated by A.