This is an investigation into the fundamental connections between the referential use of language and our rich human experience. All types of experience — perceptual, practical, scientific, literary, esthetic, ludic, ... — are tightly unified into one total experience by the structure of reference to real or possible items. Singular reference is essential for locating ourselves in our own corner of the world. General reference, by means of quantifiers, is our main tool in ascertaining the accessible patterns of the world. Both are primitive and mutually irreducible. (Often this has been denied.) The unity of total experience is constructed through the biographical unity of a person, and the sociological unity of the communications across a community. This unity of experience is wrought out by an underlying unitary system of reference. We need, therefore, a comprehensive theory of individuation, existence, predication, and truth. One such a theory is Guise Theory.

This essay examines the role of non-existent objects in "epistemological ontology" — the study of the entities that make thinking possible. An earlier revision of Meinong's Theory of Objects is reviewed, Meinong's notions of Quasisein and Außersein are discussed, and a theory of Meinongian objects as "combinatorially possible" entities is presented.

Following Bolzano, I suggest that there are two types of entity: those that are states of other things and those that are not. The second type includes, not only substances, in the traditional sense, but also such abstract objects as numbers, attributes and propositions. It is argued that the theory of states, when combined with an intentional account of negative attributes, will yield a theory of negative entities and of events.

Four questions are raised about the semantics of Quantified Modal Logic (QML). Does QML admit possible objects, i.e. possibilia? Is it plausible to admit them? Can sense be made of such objects? Is QML committed to the existence of possibilia?The conclusions are that QML, generalized as in Kripke, would seem to accommodate possibilia, but they are rejected on philosophical and semantical grounds. Things must be encounterable, directly nameable and a part of the actual order before they may plausibly enter into the identity relation. QML is not committed to possibiha in that the range of variables may be restricted to actual objects.Support of the conclusions requires some discussion of substitution puzzles; also, the semantical distinction between proper names which are directly referring, and descriptions even where the latter are "rigid designators".Views of W.V. Quine, B. Russell, K. Donnellan, D. Kaplan as well as S. Kripke are invoked or evaluated in conjunction with these issues.

The proposed synthesis is set within general object-theory. The underlying idea of the synthesis is that the alternative worlds semantics — arrived at in pursuit of a universal semantics (a general semantics for all languages, including relevant ones) and, connectedly, as part of a comprehensive object-theory — be applied also in fundamental physics, most importantly to the matter of the origin, history, and physical features of the cosmos, but as well, again connectedly, elsewhere, in particular in the interpretation of quantum theory. The universal semantics is a many worlds — a many nonexistent worlds — theory. The point of applying such an interpretation in cosmology also is explained by way of examples, concerning the understanding of the contingency of existence and the improbability of present arrangements. A resolution of the basic question, 'Why does anything at all exist?' is sketched, leading to the further question why the fundamental constants of physics have the particular surprisingly sensitive values they appear to have. Chauvinistic answers through anthropic principles are critically rejected, in favour of resolution by way of world selection.

The God of the Biblical and patristic tradition, though perhaps incomplete, possesses properties including those that involve genidentity or C-connections with us. Thus God's existence is at least possible. Using a modified version of Parson's elaboration of Meinong's theory of objects, we find that God exists if we do. But we also find that much else exists if we do; rather too much for confident belief.

The view that a possible world is an existing abstract object implies that all nonexistent possible individuals have a principle of individuation in terms of existing objects, properties, and relations. However, some individuals of this kind are totally out of this world both in the subjective sense that nobody in this world can pick them out, and in the ontological sense that they would neither be created by assembling or arranging existing bits of matter nor otherwise be generated by existing items. The only acceptable principle of individuation for such nonexistent possibles is that they are individuated by their unexemplified haecceities.

An attempt is made in the present essay to accommodate various senses of the notion of existence and ofthat of non-existence within the framework of logic. With this aim in view a system of Lesniewski's Ontology, referred to as System S, is outlined. Equipped with appropriate definitions and illustrated with a selection of theses it offers a logical theory of existence and non-existence. The usefulness of the theory is then tested by interpreting in its terms some of the principal notions and assertions of Meinong's ontology. A few brief comments on the notion of 'possible object' and on 'semantics' of fiction conclude the essay.

Two connected themes have been at the core of the old perplexity regarding thinking and speaking about non-existent objects. One involves a question of reference. Can we refer to non-existent objects without, thereby, recognizing, in some sense, non-existent entities as objects of reference? The other involves a question about existence. Is existence a property representable by a predicate in a logically adequate symbohsm? It is argued (1) that existence is not to be construed as an attribute represented by a predicate, (2) that nonnaming names introduce problems, not solutions to problems, (3) that purported properties such as self-identical are specious, and (4) that the Russell property is also seen to be specious by our consideration of predication.

The question ''Was 'existence' ever a predicate?" in a way already suggests its own answer, that this is really the wrong question to ask, because 'existence' has always been a predicate. Even those, such as Kant, who supposedly opposed this view, in fact held it. They merely denied that 'existence' is a "normal" first-order predicate. Not only Kant, but also Bolzano, Frege and Russell claimed that it is a second-order predicate. There is substantive disagreement between Kant and Bolzano on the one hand and Frege and Russell on the other over two issues: the former claim that this second-order predicate apphes to no concept analytically and that it can be properly ascribed to a singular concept, whereas the latter deny both of these claims.

Motivations for systems of free logics are reviewed and systems are divided according as they are positive (asserting atomic truths with non-denoting terms) negative (denying all such sentences) or neutral. A positive theory is developed and defended. One of the major considerations in favor of the theory is that it allows (via translation) representation of the other points of view. Finally, the relation between free logic and truth theories is elaborated.

Parmenides didn't mention reference as such, but if he had he would have undoubtedly agreed with the philosophers who nowadays hold what is called "the axiom of existence": that one can only refer to what exists. The sources of possible support for this view are examined and rejected. Primary support for the axiom is given by two sorts of argument; one concerning quantification, the other summarizing a standard Parmenidean puzzle. Weaknesses in both are exposed. Finally, the relations between the axiom and connected claims about predication and truth are probed, and an attempt is made to determine the limits of the mutual support these claims give one another.

Nothing can be said about a nonexistent object, but something can be said about the act of (unsuccessfully) attempting to refer to one or, as in fiction, of pretending to refer to one. Unsuccessful reference, whether by expressions or by speakers, can be explained straightforwardly within the context of the theory of speech acts and communication. As for fiction, there is nothing special semantically, as to either meaning or reference, about its language. And fictional discourse is just a distinctive use of ordinary language: pretended communication and within it, pretended reference. However, discourse about fiction is not pretense but is normal communication, a kind of indirect discourse. To describe the world of a fiction is to state what the fiction says (or implies); and what seems to be reference to a fictional character is really attributing (usually implicitly) a feigned reference by the author.

Russell brought three arguments forward against Meinong's theory of objects. None of them depend upon a misinterpretation of the theory as is often claimed. In particular, only one is based upon a clash between Meinong's theory and Russell's theory of descriptions, and that did not involve Russell's attributing to Meinong his own ontological assumption. The other two arguments were attempts to find internal inconsistencies in Meinong's theory. But neither was sufficient to refute the theory, though they do require some revisions, viz. a trade-off between freedom of assumption and unhmited characterization. Meinong himself worked out the essentials of the required revisions.

Anti-Meinongian philosophers, such as Russell, do not explain what they mean by existence when they deny that there are nonexistent objects — they just sense robustly. I argue that any plausible explanation of what they mean tends to undermine their view and to support the Meinongian view. But why are they so strongly convinced that they are right? I argue that the reason is to be found in the special character of the concept of existence, which has been insufficiently examined by anti-Meinongian as well as by Meinongian philosophers.

Meinong's doctrine of the modal moment and the watering-down of extranuclear properties to surrogate nuclear counterparts was offered in response to Russell's problem of the existent round square. To avoid an infinite regress of successively watered-down factualities, Meinong stipulates that the modal moment itself cannot be watered-down. This limits free assumption, since it means that the idea of the existent-cum-modal-moment round square cannot be entertained in thought. It is possible to eliminate the modal moment and watering-down from Meinongian semantics in favor of a strict enforcement of the distinction between nuclear and extranuclear properties. This provides a simpler, more economical Meinongian object theory, and regains unrestricted free assumption.

This essay argues for the importance of developing theories of nonexistent objects. The grounds are utility and smoothness of logical theory. In the latter case a parallel with the theory of negative and imaginary numbers is exploited. The essay concludes with a counterexample to a general argument against the enterprise of developing theories of nonexistent objects, and outlining the foremost problem an adequate theory of nonexistent objects must solve.

In a recent book, K. Lambert argues that philosophers should adopt Mally's Principle of Independence (the principle that an object can have properties even though it lacks being of any kind) by abandoning a constraint on true predications, namely, that all of the singular terms in a true predication denote objects which have being. The constraint may be abandoned either by supposing there is a true predication in which one of the terms denotes a beingless object (Meinong) or by supposing there is a true predication in which one of the terms denotes nothing at all (free logic). However, Lambert's conclusions can be undermined by showing that the data he produces in support of his position can be explained by either of two recent theories of abstract and nonexistent objects, both of which are couched in languages which conform to the traditional constraint.