How can we reconcile the following apparent truths : ‘ Sherlock Holmes doesn ’ t exist ’ and ‘ Sherlock Holmes was created by Conan Doyle ’ ?

The problem with the two statements about Sherlock Holmes is that we want to say that both of them are true but it looks like they cannot both be true at the same time. If it is true that Sherlock does not exist, then we would have to accept that Conan Doyle created something that does not exist. If so, did he really create something? Some possible responses here seem to lead us into deep waters. In particular, those responses admitting that there are ‘different types of truth’ generate a large number of extremely difficult problems which I lack space to discuss here. Other responses consider whether there might be different sorts of existence, and this is the type of response I will be examining. In particular, I will argue that ficta – all fictional entities including characters such as Sherlock Holmes – are non-existence concrete objects. I consider the case of Sherlock Holmes partly because he is traditionally chosen in the literature and partly because he is one of the most well-known fictional characters. In general, I think the argument I make should apply to all fictional characters. A benefit of using the character Holmes is that he is richly and vivaciously specified which would not be true of all characters. Imagine that I make up a story about Barg the dragon, and the entire story is: ‘Barg was a dragon. The End.’ The character Barg is woefully underspecified. It is likely that you will make up properties that Barg has that I, the author have not given him. You will probably think Short, T L 2014 How can we reconcile the following apparent truths: ‘Sherlock Holmes doesn’t exist’ and ‘Sherlock Holmes was created by Conan Doyle’? Opticon1826, (16): 8, pp. 1-9, DOI: http://dx.doi.org/10.5334/opt.bs


Introduction
The problem with the two statements about Sherlock Holmes is that we want to say that both of them are true but it looks like they cannot both be true at the same time.If it is true that Sherlock does not exist, then we would have to accept that Conan Doyle created something that does not exist.If so, did he really create something?Some possible responses here seem to lead us into deep waters.In particular, those responses admitting that there are ' different types of truth' generate a large number of extremely difficult problems which I lack space to discuss here.Other responses consider whether there might be different sorts of existence, and this is the type of response I will be examining.In particular, I will argue that ficta -all fictional entities including characters such as Sherlock Holmes -are non-existence concrete objects.
I consider the case of Sherlock Holmes partly because he is traditionally chosen in the literature and partly because he is one of the most well-known fictional characters.In general, I think the argument I make should apply to all fictional characters.A benefit of using the character Holmes is that he is richly and vivaciously specified which would not be true of all characters.Imagine that I make up a story about Barg the dragon, and the entire story is: 'Barg was a dragon.The End.' The character Barg is woefully underspecified.It is likely that you will make up properties that Barg has that I, the author have not given him.You will probably think Short, T L 2014 How can we reconcile the following apparent truths: 'Sherlock Holmes doesn't exist' and 'Sherlock Holmes was created by Conan Doyle'?Opticon1826, (16): 8, pp.1-9, DOI: http://dx.doi.org/10.5334/opt.bs

RESEARCH ARTICLE
How can we reconcile the following apparent truths: 'Sherlock Holmes doesn't exist' and 'Sherlock Holmes was created by Conan Doyle'?Tim L. Short * that he can fly and breathe fire, for example.What this brings out is that Barg has impossible properties, which provides an initial indication that fictional characters may just be possible sets of properties, rather than sets of possible properties, as I will set out below.The other relevant questions of interest, which I can only raise and not answer here, is what do we say about the properties that you as reader ascribe to Barg? Are you now the author?Are those properties on the same footing as the ones I gave him?
We might also consider in this vein the questions arising from Booth's concept of the 'unreliable narrator' (Booth 1983: 339 et seq).What are we to say about properties ascribed to characters by a narrator who is known to sometimes make false ascriptions?Further, flirting with oxymoron, can we allow that the author may not be authoritative?There are also questions of this type raised by irony in fiction.Currie argues for a pretence theory of irony, wherein fictional characters are pretending to do one thing while actually doing the opposite (Currie 2010).For example, if Holmes says something to Watson -as he frequently does -like 'I am once more amazed by the brilliance of your powers of deduction!' it looks very much like Watson now has the property of did not exhibit brilliant powers of deduction even though the text appears to do the exact opposite, before interpretation by us.I will call all of these sorts of question 'Booth questions'; they will become relevant when we consider the properties characters have.These questions make it plausible that it may not be precisely specified what properties characters have.
Returning to the central question, we may examine the ontologies available to us.Both Parsons (1975Parsons ( , 1979) ) and Meinong (1904) admit non-existent concrete objects to their ontology.These non-existent objects are associated with sets of properties.According to Parsons's view, which I will defend, ficta are associated with sets of properties with which no existent object need be associated.
The ficta are non-existent concrete objects while the sets, as with all sets, are abstracta.I will assume without argument that abstracta like sets exist -they are real -but they are not concrete.I understand the term ' concrete' to refer to all objects that have a determinate location in space-time.I will assume that the division is exhaustive -anything not concrete is abstract.
I will consider some implications of Parsons's view and some objections to it.Since Parsons's view is opposed to Creationism, it will be important to consider arguments for Creationism, since to the extent they are successful they are pro tanto objections to Parsons's view.Creationism in relation to ficta holds that authors create them.(We should note, for the avoidance of doubt, that we are not discussing the type of Creationism that is opposed to evolutionary theories.)Against this, Parsons's view holds that the sets associated with ficta are not brought into existence by an act of creation because they do not have existence in the same way that objects in the real world do.The mode of existence that ficta do haveas with other sets or abstracta in generalis timeless and so authors do not give them that mode of existence.Rather, authors specify an abstract object by listing some of its properties.What Conan Doyle does when he writes about Sherlock Holmes is to specify him -more precisely, to determine some of the properties that Holmes has as a fictional object, which means to determine the properties that are in the set associated with the term 'Holmes'.
One of the properties that Holmes has is that he plays the violin.This means that plays the violin is one of the properties that is a member of the set associated with the term 'Holmes'.Holmes does not play the cello.There is a very similar entity to Holmes -let us call him Cholmes -who shares all of Holmes's properties with the exception that Cholmes does not play the violin and does play the cello.According to Parsons's view of ficta, both are non-existent concrete objects.One -Holmes -was specified by Conan Doyle and the other -Cholmes -was not 1 .In the ontology of Parsons's view, being described by a set of properties is sufficient to be an object, but not sufficient to exist.Thus, some items in the world possess the properties blackness and cat-hood, and so there are existent objects in the world which are black cats.On the other hand, no item in the world possesses the properties of goldenness and mountain-hood.While golden mountains are objects, because that set can be specified, there is no existing golden mountain.Both Cholmes and Holmes also fall into that latter category of being nonexistent objects because there is no object in the world which has all the properties of Holmes or Cholmes.
The sets associated with ficta exist but are not concrete, as are sets in general.This division of the real into the domains of concreta and abstracta together with the observation that we allow of entities in both domains that they exist has a major benefit.The division allows us to reconcile the truth of the two statements in the title.They are both true but in different domains of truth.When we say that 'Sherlock Holmes does not exist', we mean 'Sherlock Holmes is not a concrete object', and this is true in the domain of concreta.There is no existent human with all of the specified properties.When we say 'Sherlock Holmes was created by Conan Doyle', we mean that 'the set associated with 'Sherlock Holmes' was specified by Conan Doyle'; this is also true, but has application in the domain of abstracta 2 .In the form of a set, 'Sherlock Holmes' is as real as other sets are 3 .

Parsons's view is not Creationism
We may establish that Parsons's view is not Creationism by considering the following definition of the latter term.
Creationism: fictional entities ' are created […] by the authors of the novels in which they first appear' (Brock 2010: 338).
By this definition, Parsons's view is not Creationism, since authorial acts do not change ontology and nothing abstract changes its status in terms of existence, lack of existence or mode of existence.A fictional character is a man-made artefact.Parson's view is anti-Creationist, and several objections have been made generally against all anti-Creationist views.I will seek to show though that Parson's view is not only distinct from Creationism but superior to it.For this reason, I challenge several of the objections to anti-Creationism below 4 .
The 'man-made artefact' that Parsons's view can recognise is a term whose association has been introduced by convention; this term then behaves like a name in that it is associated with a set of properties.Thus, Conan Doyle does not create Holmes; he arranges that 'Holmes' is associated with a set and determines some of the properties in that set.If it is possible for authors to introduce characters without properties, then they associate a term with an empty set.

Objections Property objections
It may be that there is some lack of clarity in the specification of properties.This is acknowledged by Parsons and used as a challenge by van Inwagen (1977).The 'Booth questions' I listed above would also become relevant here, because they seem to allow for exactly this lack of clarity.We can address this by understanding the question as whether a set of properties must contain either a property F or its negation not-F as one of its members.It is however not the case that a set of properties must contain F or not-F; a set of properties may be incomplete in this way.The purported violation of the laws of logic draws on the intuition that everything is either F or not F.This may be true for all real objects, but even then, that says nothing about what properties must be in a set of properties.The set of properties {mountainhood, goldenness} is a well-constructed set that does not have either silverness or notsilverness as a member.
Another objection to Parsons's view is that objects have properties that cannot be instantiated together.As Salmon (1998: 293) writes 'the Object […], as in the case of the round square, may even have inconsistent properties.'However, this is to confuse a powerful objection with reference to the actual world with an objection that has no force with reference to the world of sets.It would indeed be an objection to the existence of an object in the real world to note that it has non-compossible properties.We could for example ask whether the round square has corners or not.Since it is round, it does not.Since it is square, it does.The round square would generate many contradictions, were it to exist.However, this is not what the proponent of Parsons's view means.He does not believe that there is a round square to be found in the world in exactly the same way as he does not believe that the golden mountain has the property of existence.What he does hold is that there is a non-existent object associated with the set including the properties of roundness and squareness.Since that set does exist in the way that sets exist and does not exist in the way that concreta exist, Salmon's objection fails.

Indeterminacy of identity objection
Parsons's view is a realist view because the sets associated with ficta are real.Some authors have objected that any realist account of ficta will suffer from a possible situation where it is indeterminate whether two fictional characters are identical.For example, Everett (2005) believes it is indeterminate whether the Faust of Marlowe is identical to that of Goethe.The realist is held to be especially exposed to the problem because indeterminacy of identity cannot apply in the real world.
This objection has no force against Parsons's view, because it is clear when two characters are identical.In the very unlikely circumstance that two authors specified the same set by giving their characters exactly the same set of properties, then they are identical characters.Otherwise, they are not.There is no indeterminacy here at least on the surface.The opponent may sharpen their objection here by asking whether two characters are identical if all of their specified properties are the same but one of the unspecified ones is not.But bringing this challenge would require a coherent exposition of what it means for a character to have a property that is at once specified and unspecified.
A similar response is available to the proponent of Parsons's view if it is urged that there is a difficulty with two characters where within the story it is left indeterminate whether they are identical.Again, they are not identical unless they are given all of the same properties.In the example of Frackworld given by Everett (2005: 629 et seq), it is said that there are some 'striking differences' between two allegedly indeterminately identical characters.If we accept Parsons's view, these differences are sufficient to distinguish the characters.Similarly, there is no indeterminacy about whether a Slynx exists --in circumstances where it is deliberately left indeterminate as to whether the Slynx exists within that story.According to Parsons's view, it does exist, as a set.There may be indeterminacy as to which set has been associated with the term `Slynx' by the author of a story in which it is indeterminate whether the Slynx exists.Such indeterminacy however is not problematic for Parsons's view.The indeterminacy is harmless because an indeterminacy with which a set is associated does not involve indeterminacy of identity.

Creativity denial objection
Creationism is commonly defended by noting the intuition we have that there is a great deal of ' creativity' involved in being the author of a fiction; in being the ' creator' of a vibrant fictional character.The charge is that this creativity is not sufficiently recognised or given sufficient weight on a view in which the set of properties is not created.Parsons's view can avoid this charge though.The view does not in any way devalue the ' creativity' involved in authoring a novel or other work of art.There can be a great deal of skill and talent involved in specifying the properties of a set so that the associated character is arresting or entertaining.The author has indeed created something -an association of a term with a pre-existing set -and has in addition exercised artistic skill in selecting which set it shall be, by virtue of deciding which properties the fictional entity will have.
One objection here may be to note that according to Parsons's view, an author may create a term or use a pre-existing term.This frequently happens; authors often write new stories containing ficta from previously written stories.In Parsons's view, this just means that the term 'Holmes' may alter its association to be with a new set with some new properties.The obvious counter supporting Parsons's view is a 'first use' response, whereby an author first associates a term with a set the first time a character is named.Moreover, the set with which a term is associated changes every time a character is given a property, and not just the first time.This avoids an asymmetry between first and other uses of a term.

Rigidity mismatch objection
Defenders of Creationism have argued that fictional objects cannot be sets because there is a temporal and modal rigidity mismatch between sets and fictions.Lee Walters (2012: 92) argues: [l]iterary fictions […] cannot be identified with any number of concrete instances of the fiction, since no particular instance or instances of a fiction are required for the continued existence of the fiction.This fact also rules out identifying fictions with pluralities [or] sets […] concrete given the temporal and modal rigidity of […] set membership.
The objection notes that set membership conditions are temporally and modally rigid.Temporal rigidity means that the identity of a set supervenes on its members at all times.Modal rigidity means that a set has the same elements in any worlds in which it exists, irrespective of how anything else is or could be.So the two claims amount to the view that in discussing sets, once we have identified a particular set by specifying its members, nothing else affects which set it is.
The mismatch objection then becomes the claim that Sherlock Holmes is not temporally and modally rigid while sets are.If so, then if Conan Doyle had counterfactually specified that Holmes played the cello, then he would still have been talking about the same fictional character.Yet to say this is to beg the question against Parsons's view, since for the proponent of that view, Sherlock Holmes in one sense is a set, and Conan Doyle would in those circumstances have been talking about Cholmes not Holmes.It is likely that those who are mereological essentialists in relation to concrete objects will also be essentialists in relation to ficta.Holmes could not have been Cholmes, and one set could not have been another set.
An objector might say it is counterintuitive to argue that Holmes could not have been Cholmes.This objection is mistaken however.It is certainly true that Conan Doyle could have decided that his character could have been a cello player, and he could have named that character 'Holmes'.What he could not have done though, is change the members of the set originally associated with the term 'Holmes.'He could only have associated the term 'Holmes' with the set we are now associating with 'Cholmes'.
Creationists claim correctly that a particular novel could have first been instantiated in a different format.Note that A Study in Scarlet -that very work -could have first been written on a different piece of paper.Parsons's view can go along with this.But that is not to allow that A Study in Scarlet could have been different.Certainly a story could have been written with different characteristics.And certainly it could have been called A Study in Scarlet.But the set now associated with that term could only ever have the members it currently has.This shows how Parsons's view avoids a cost that the Creationist must pay, which is to allow that once created, ficta exist forever.This is the same as saying that timeless objects can come into existence, which is a strange asymmetry.If something can be created, then surely it ought to be possible to destroy it as well.
It might be countered here that this is not so strange because of the asymmetries that arise in relation to facts about the past.Until a particular event occurred, there was no fact; but once it has, it is always a fact that it has, and nothing could then destroy that fact.However, there does not seem to be a close analogy between the actions involved in the two cases.If I create a fact now by acting in a certain way, then perhaps it will be a fact forever that I did so.On the Creationist view, if I create a character now by conducting whatever acts the Creationist specifies as sufficient, that character exists forever.In the former case, I act but I do not act in order to create a fact.If I create a character on the Creationist view, then I act specifically to create something which is then eternal.This seems quite a potent act of creation for a person to be able to perform: to be able to create deliberately the eternal.On balance, it is more useful to take Parsons's view, by which ficta are indeed timeless objects and are so at all times, as is appropriate.

Revision of abstract individuals objection
Creationism has also been defended by noting that we commonly talk about changes in abstracta.This would be a problem for Parsons's view in which no abstract items are created or changed.One purported example of change in abstracta is that the laws of cricket are revised from time to time.Yet this is better seen as meaning not that the abstract object associated with the term 'the Laws of Cricket' has changed any of its characteris-tics, but that the association of the term has been modified.The term is now associated with a set different from the original set in that the new set includes some new properties reflecting the rule changes.No abstract objects have changed.(It is no objection here that the term is an abstract object which has changed because it is now associated with a different set, because this relational property alteration is merely Cambridge change -i.e.relational only.) This situation is analogous in relation to ficta.No new properties are instantiated nor are any new abstracta created, when Conan Doyle introduces a new property for Holmes.What he has done is alter the set associated with the term 'Holmes' --as opposed to creating or modifying a set.If it is specified in a certain work that Holmes likes cricket, what this means is that the association of the term 'Holmes' has been shifted to a set slightly different set from the previous one.The new set is the one composed of all the properties in the previous set plus the property of likes cricket.
It may be objected here that Parsons's view cannot handle a situation where Conan Doyle subsequently revises this property because we could not say which set was associated with the term 'Holmes', and in particular, whether it contains the property of likes cricket.However, this would again be to confuse the timeless nature of the sets associated with and the changeable nature of the relations of the terms associated with them.Parsons's view can be defended in each of the situations mentioned.The process is simply explained by saying that one set is associated with before the change and a different one afterwards 5 .Parsons's view avoids some difficult questions for Creationists; viz., when and how are ficta created?After all, Brock (2010) founds his challenge to Creationism on the difficulty of these questions.What suffices to create a character?Would a character that was only named be created?These questions all have straightforward answers in Parsons's view.No abstracta are created at any point.An occurrence of a term associates it with a set.If no properties are specified, then the term is associated with an empty set.As properties are added, the term is associated with different sets.
Further benefits abound.There is no indeterminacy about the number of fictional characters, since the proponent of Parsons's view does not look to actual works of fiction to determine that number.There is no difficulty for Parsons's view in analysing claims such as 'the Odysseus of The Odyssey and the Ulysses of Tennyson's Ulysses are the same fictional character'.It will be unlikely that they are in fact identical.A necessary condition would be that both authors have arranged that their terms shall refer to the same set i.e. the fictional characters will have the same properties.However, Parsons's view has no difficulty accommodating the possible truth of claims such as 'Tennyson's character was based on the character in The Odyssey'.All such claims reduce to claims about the sets involved.They mean that the two sets associated with the terms by the two authors contain many identical properties.
Finally, Creationism may collapse into something like Parsons's view under some circumstances.Imagine a computer programme populated with all conceivable properties, and arranged to name an extremely large number of sets of combinations of those properties and print out the results.Presumably then the Creationist universe of abstracta would resemble the Meinongian.It would admittedly not be infinite, but only for contingent reasons relating to the time available to run the programme.That would scarcely suffice for the Creationist to charge the Meinongian with ontological profligacy.

Conclusion
The two statements in the title may be reconciled by understanding them as follows.'Sherlock Holmes doesn't exist' means that there are no existent objects with all of the properties in the set of properties associated with the term 'Sherlock Holmes'.This is true.'Sherlock Holmes was created by Conan Doyle' means that Conan Doyle through his work specified the set of properties associated with the term 'Sherlock Holmes'.This is also true.The disambiguation we need of 'Sherlock Holmes' is between Sherlock Holmes (1) the fictional man in the story, and Sherlock Holmes (2) the set of properties associated with 'Sherlock Holmes'.It is true that Sherlock Holmes (1) does not exist because there is no such man in the real world.It is also true though that Sherlock Holmes (2) does exist because Sherlock Holmes ( 2) is a set which has had its elements specified; the set is real.Sherlock Holmes (2) was associated with a set by Conan Doyle.So there is no conflict between the statements because they refer to different entities: Sherlock Holmes (1) is not identical to Sherlock Holmes (2).
We are also now in a position to deal with problems mentioned in the literature in relation to true negative existential statements.Note that one commentator writes that she will ignore true negative existential statements such as 'Iago does not exist' because they are 'problematic on every theory' (Friend 2007: 143).The fact that Parsons's view can handle true negative existentials very easily is therefore a major point in its favour.This advantage of course has carried through from the more widely applicable benefit of the Meinongian ontology -as Reicher (2010: 3.1) puts it: '[t]he appeal to nonexistent objects thus supplies an elegant solution to the problem of negative singular existentials' -but is none the less valuable for that.
The reason for the difficulties is that we seem to be referring to something when we discuss the purportedly ' empty terms' of ficta.Moreover, we seem to be referring to distinct objects when we say 'Zeus is not identical to Pegasus'.The solution is that we are indeed referring to distinct non-existent objects that are associated with different existent sets.'Zeus' is associated with one set and 'Pegasus' with another.These are different ficta for many reasons including that the first set includes the property is divine and the second set does not.So in this view, we can retain the truth of the negative existential statements.It is true that Zeus is not real, although the set associated with 'Zeus' is real -and also the distinction between different unreal objects 6 .
Parsons's view is the correct view of ficta and it explains how the two statements in the title question can both be true.

Acknowledgement
I would like to acknowledge valuable comments from Opticon1826 staff, which considerably improved the set-up and motivation of the argument.

Notes
1 We might allow that authors 'identify' sets rather than specifying them if we accept that authors associate terms with exactly one set.Alternatively, they specify a group of sets if we include sets with properties not listed by the author as also associated.For example, we might allow the property renate to Holmes even if Conan Doyle makes no mention of this. 2 While it is indeed true that Conan Doyle's act of specification takes place in the domain of concreta, we are not interested in that but only in what effects his acts have in terms of associating a term with a set in the domain of abstracta. 3Even if this runs counter to appearances, that would not count against the view.As Thomasson ( 2003: 205) notes, 'since there are apparent inconsistencies, any consistent theory must give up appearances somewhere'.Some might also object here that it is undesirable to regard ficta as associated with sets because the view is committed to statements like 'the null set is a subset of Sherlock Holmes'.This does indeed appear undesirable, but only because of our habit of regarding Holmes as a man, who cannot have subsets in any useful sense.When we regard the term 'Holmes' as associated with a set, there is no problem.Indeed, if an author names a character but gives him no properties, he associates a term with the null set. 4 As Brock (2010: 343) points out, some varieties of Creationism -those defended by Deutsch which hold that specification of a pre-existing character suffices for its creation -will be compatible with Parsons's view.I will not consider this further because I agree with Brock that the absence of a new entity means no creation has taken place.Some might deny that Deutsch qualifies as a Creationist because his view is too similar to the one I defend here. 5A further objection here to Parsons's view is that if characters are to be associated with sets, and 'Holmes' is associated with a different set at different times, how do we know that the two Holmes's are the same character?For lack of space, I cannot go into detail here but authorial intention would play a role in the solution. 6It might be objected here that an author could write a story about a number that does not exist --for example, an even prime not identical to two.What are we saying when we say this does not exist?This is simply dealt with by noting that there is no concrete or abstract item which is an even prime not identical to two.What there is, in Parsons's view, is a non-existent object which is associated with the set containing the properties is even, is prime and is identical to two.