In your
possession
of these miraculous powers you are not far removed from the Almighty.
Gottlob-Frege-Posthumous-Writings
Then 'Some men are negroes' would mean 'Some-possibly all-but at least one man is a negro'.
If we are agreed on this, then a particular affirmative judgement such as ? some men are negroes' can be converted into 'Some negroes are men'. The resistance one naturally has to this is based on the fact that one involuntarily adds in thought 'but some negroes are not men'. This thought is excluded by our rider 'possibly all'.
Now you wanted to construe the expression 'Men exist' as having the same meaning as 'Something existing is a man'. The trouble with the latter is that, grammatically, it is not existence but being a man which occurs as predicate. But it is existence that is really meant to be asserted. Now we can adopt a grammatical form which brings this out by converting and saying 'Some men exist', taking this in the sense of 'Some-possibly all-but at least one man exists'. It will then follow that this means the same as 'Men exist'.
Now I have understood throughout that on your view you take the difference in meaning of 'exist' in 'Leo Sachse exists' and 'Men exist' to be of the same kind as the difference in meaning of'is a German' in 'Leo Sachse is u German' and 'Some men are German', so that 'exists' is related to 'exist' in the first two sentences as, in the last two, 'is a German' is related to 'are German'. I have deliberately made the subjects 'Leo Sachse' and 'some men' the same in both cases to create the impression that they do parallel one another. I believe the only reason we omit the 'some' in the sentence 'men exist' is to forestall the objection 'not all? '
Now you marshalled your argument, I believe, as follows.
First of all you wanted me to admit that 'There are men' means the same as 'Among that which has being is some man' or 'A part of what has being is a man', or 'Something that has being is a man'. As expressions with the same meaning as 'has being' you also used 'can be experienced', 'existing', 'that whose idea has been caused by something affecting the ego'. These
? ? 62 [Dialogue with Piinjer on Existence]
variants, I believe, make no essential difference. With them some secondary difficulties may crop up or disappear. The main difficulty remains the same and so does the general outline of the plan of attack. I now had further to be brought to admit that the verb 'to be' ('to exist') is used in the same sense as in the sentence 'Leo Sachse is' or 'exists'. ' With that you seemed to have won the day.
Now I can readily grant that the expression 'there are men' means the same as 'Something existing is a man'-only, however, on condition that 'exists' predicates something self-evident, so that it really has no content. The same goes for the other expressions which you use in place of'exist'.
But if the proposition 'Leo Sachse is' is self-evident then the 'is' cannot have the same content as the 'there are' of 'There are men', for the latter does not say something self-evident. Now if you express what is said by 'There are men' by 'Men exist' or 'Among that which has being is some man', that the content of the statement cannot lie in the 'exist' or 'has being' etc. And this is the npwrav l{leVOac; by which you could not help but be forced into making contradictory judgements-the error of thinking that the content of what is predicated in 'Some men exist' or 'Something existing is a man' or 'Men exist' is contained in the word 'exist'. This is not the case; this word only contains the form of a predicate as does the copula 'is' in the sentence 'The sky is blue'. Like the 'it' in 'it is raining', the 'exist' in 'Men exist' is to be understood as a mere auxiliary. As language, feeling at a loss for a grammatical subject, invented 'it', so here, feeling at a loss for a grammatical predicate, it has invented 'exist'.
I shall use the fact that instead of 'exists' one can also say 'is identical with itself' to show that the content of what is predicated does not lie in the word 'exists'. 'There are men' means the same as 'Some men are identical with themselves' or 'Something identical with itself is a man'. Neither in 'A is identical with itself' nor in 'A exists' does one learn anything new about A. Neither statement can be denied. In either you can put what you like for A, and it still remains true. They do not assign A to one of two classes in order to mark it off from some B which does not belong to that class. The point in saying 'A is identical with itself can only be to express the logical law of identity; the point cannot be to impart any further knowledge about A. Just as it could be maintained that 'exist' means the same in 'This table exists' and 'Tables exist', so we can say that the predicate 'identical with itself has the same sense in 'This table is identical with itself' and 'Tables are identical with themselves'. But in that case we have also to acknowledge that the judgements 'This table exists' and 'This table is identical with itself are completely self-evident, and that consequently in these judgements no real
1 If the reader is to follow the argument at this point, he should bear in mind that in German the expression 'there are', as in Frege's example 'There are men', is not rendered by a form of the verb 'to be' ('sein') but by 'es gibt' which comes from 'geben', meaning 'to give'. The corresponding English expression 'there is' is likewise so rendered (trans. ).
? [Dialogue with Piinjer on Existence] 63
content is being predicated of this table. Just as we call statements such as 'Men exist' existential judgements, thinking the content of what is predicated to lie in the word 'exist', so we might call the statement 'Some men are identical with themselves' an identity judgement, and 'There are men' would be an identity judgement. In general, in any attempt to prove that the content of what is predicated in 'There are men' was located in the 'exist' of 'Men exist', we could, without any falsification, switch 'are identical with themselves' throughout for 'exist'. I shall seek to show this.
But if the content of what is predicated in the judgement 'Men exist' does not lie in the 'exist', where then does it lie? I answer: in the form of the particular judgement. Every particular judgement is an existential judge- ment that can be converted into the 'there is' form. E. g. 'Some bodies are light' is the same as 'There are light bodies'. 'Some birds cannot fly' is the same as 'There are birds that cannot fly', and so on. It is more difficult to do the opposite and convert a judgement with 'there is' into a particular judgement. Out of context the word 'some' has no sense; it is an auxiliary like 'all', 'each', 'none' and so on, which, in the context of a sentence, has a logical function to perform. This function consists in putting two concepts into a certain logical relationship. In 'Some men are negroes' the concepts 'man' and 'negro' are put into this relationship. So we always need two concepts if we want to form a particular judgement. Now of course the sentence 'There are flying fish' can easily be converted into 'Some fish can fly', because we have two concepts 'fish' and 'being able to fly'. It becomes more difficult if we try to put 'There are men' into the form of a particular judgement. If we define man = rational living being, we may say 'Some living beings are rational' and, assuming the definition to be correct, this means the same as 'There are men'.
This recourse is only open to us when the concept can be analysed into two characteristic marks. There is another method closely connected with this one. E. g. if we have to convert 'There are negroes', we may say that negro = negro that is a man, because the concept 'negro' is subordinate to the concept 'man'. Here again we have two concepts and may say 'Some men are negroes' or 'Some negroes are men'. But this works only in the particular case of the concept 'negro'. For 'There are birches' we should have to select a different superordinate concept, such as 'tree'. If one wants to do the thing quite generally, one needs to look for a concept superordinate to all concepts. Such a concept, if one wishes to give it that name, can no longer have any content at all since its extension will be unlimited; for any content can only consist in a certain delimitation of the extension. As such a concept we might select that of 'being identical with itself, since we said that 'There are men' is the same as 'There are men identical with themselves' or 'Something identical with itself is a man'.
Language has availed itself of a different resource. The copula, i. e. the mere form of a predicate without content, was excellently suited for forming a concept without content. In the sentence 'The sky is blue' the predicate is
? 64 [Dialogue with Piinjer on Existence]
'is blue', but strictly the content of the predicate lies in the word 'blue'. Leave this out and what remains-'The sky is'-is a predicate without content. In this way we form a quasi-concept-'being'-without content, since its extension is unlimited. This makes it possible to say: men = men that have being; 'There are men' is the same as 'Some men are' or 'Something that has being is a man'. Thus here the real content of what is predicated does not lie in 'has being' but in the form of the particular judgement. Faced with an impasse, language has simply created the word 'being' in order to enable the form of the particular judgement to be employed. When philosophers speak of'absolute being', that is really an apotheosis ofthe copula.
But it is easy to see now how this came about. People felt that the sentence 'There is a centre of mass of the earth' is not self-evident, and that consequently there is a content to what is predicated. And it is now readily intelligible why, when they employed the form 'A centre of mass of the earth exists', people believed this content to reside in the word 'exist'. In this way a content was packed into the word 'exist', without anyone's being able to specify in what this content really consists.
It may now be shown how Piinjer, as a result of the 7C{Jwruv 'lfevar; of seeing in the 'exist' the content of what is predicated in 'Men exist', was bound to be driven into making contradictory assertions. I was easily able to convince him that the denial of 'A can be experienced' is impossible where can be experienced = is = exists. He had also to concede that to say of a thing that it can be experienced does not characterize that thing in any way. On the other hand, however, he wanted to salvage a content for a statement saying of something that it can be experienced. Something was surely meant to be said in the sentence 'This table can be experienced', 'This table exists'; it was surely meant to state something that was not pleonastic or self-evident. So he was forced into the contradiction of taking the denial of 'This table can be experienced' as something that was not pleonastic and self-evident. 1 He had to deprive the expression 'can be experienced' of all content without making it devoid of content. Piinjer wanted to convey the content of the judgement 'This can be experienced' by 'The idea of the this is not an hallucination, it is not something which originates from myself alone, but the idea has been formed as a result of the ego's being affected by the this'. To this I had to object that one only has the right to form the expressions 'idea of the this', 'the ego's being affected by the this', after having made the judgement 'something corresponds to this idea of mine'. If nothing corresponds to this idea of mine, then the expression 'idea of the this' has no sense and thus the sentence as a whole has no sense. Piinjer then altered his account, though without conceding that it was incorrect: 'The object of the idea B can be experienced' means 'The idea B has been formed through something affecting the ego'. I was now able to infer from this that the negation of t11e sentence 'The object of the idea B can be experienced' has a
1 sic (trans. ).
? [Dialogue with Piinjer on Existence] 65
good sense. But Piinjer had said earlier that the negation of the sentence 'A can be experienced' is impossible. We now have to qualify this somewhat and say: If A is an object of experience, then the negation of 'A can be experienced' is impossible, but if A is object of an idea, then the negation of that sentence is possible. We see how this example confirms that it is impossible to assign to the predicate 'can be experienced' a sense that is not self-evident whilst making the general claim that it has no sense to deny of a thing that it can be experienced. At the same time we see that the concept of what can be experienced acquires a content only through its extension being delimited in some way. In fact all objects are divided into two classes: objects of experience and objects of ideas. The latter do not all fall under the concept 'what can be experienced'. From this it can be further inferred that not every concept is subordinate to the concept of what can be experienced-for the concept 'object of an idea' is not. And from this it follows that the concept of what can be experienced is not generally suited for the purpose of expressing a judgement with 'there is' in the form of a particular judgement. In order to justify the expression 'object of an idea' as being generally applicable, Piinjer had to maintain that every idea has an object, that there are ideas of objects which have not been formed through something affecting the ego. If we apply to this his definition of sentences with 'there is', a contradiction must emerge. In fact, according to this definition, the judgement 'there are objects of ideas that have not been formed through something affecting the ego' is synonymous with 'Among what can be experienced is something that falls under the concept "object of an idea that has not been formed through something affecting the ego"'. But now according to Piinjer's account, objects of ideas not formed by virtue of something affecting the ego cannot be experienced. So we arrive at the statement 'Among what can be experienced is something that cannot be experienced'.
We may also put it like this: from the two premisses
I. There are objects of ideas that are not formed through something affecting the ego;
2. Objects of ideas that are not formed through something affecting the ego cannot be experienced;
there follows the conclusion:
There are objects of ideas-which objects cannot be experienced. This is
a contradiction once it is allowed that the same kind of existence is expressed by 'there is' as is meant to be conveyed by 'can be experienced'.
In general one can lay down the following:
If you want to assign a content to the verb 'to be', so that the sentence 'A is' is not pleonastic and self-evident, you will have to allow circumstances under which the negation of 'A is' is possible; that is to say, that there are subjects of which being must be denied. But in that case the concept 'being'
? 66 [Dialogue with Piinjer on Existence]
will no longer be suitable for providing a general explanation of 'there are' under which 'there are B's' means the same as 'something that has being falls under the concept B'; for if we apply this explanation to 'There are subjects of which being must be denied', then we get 'Something that has being falls under the concept of not-being' or 'Something that has being is not'. There is no way of getting over this once a content of some kind-it doesn't matter what it is-is agreed to the concept of being. If the explanation of 'there are Bs' as meaning the same as 'Something that has being is B' is to work, we just have to understand by being something that goes entirely without saying.
For this reason the contradiction still remains if we say 'A exists' means 'The idea of the A has been caused by something affecting the ego'. However, still other difficulties crop up-. here, only some of which I want to mention.
When Leverrier put to himself the question whether there were planets beyond the orbit of Uranus, he was not asking whether his idea of a planet beyond the orbit of Uranus had been caused, or might have been caused, by something affecting the ego. When there is a dispute over whether there is a God, the dispute is not over whether our idea of a God has been caused, or may be caused, by something affecting the ego. Many who believe that there is a God will dispute that their idea of Him has been caused by God's immediately affecting their ego, for it can only be a question here of something affecting them immediately. However, this is only by the way. The upshot is as follows:
We can say that the meanings of the word 'exist' in the sentences 'Leo Sachse exists' and 'Some men exist' display no more difference than does the meanings of 'is a German' in the sentences 'Leo Sachse is a German' and 'Some men are Germans'. But then the sentence 'Some men exist' or 'Something existing is a man' only means the same as 'There are men' if the , concept 'existing thing' is superordinate to the concept man. So if such forms of expression are to have the same meaning in general, the concept 'existing thing' must be superordinate to every concept. This is only possible if the word 'exist' means something that goes entirely without saying, and if therefore nothing at all is predicated in the sentence 'Leo Sachse exists', and if in the sentence 'Some men exist' the content of what is predicated does not ' lie in the word 'exist'. The existence expressed by 'there is' is not contained in the word 'exist' but in the form of the particular judgement. 'Some men are Germans' is just as good an existential judgement as 'Some men exist'. But once the word 'exist' is given a content, which is predicated of an individual thing, this content can be made into the characteristic mark of a concept-a concept under which there falls the individual thing of which existence is being predicated. E. g. if one divides everything into two classes
1. What is in my mind, ideas, feelings etc. and
2. What is outside myself,
? [Dialogue with Piinjer on Existence] 67
and says of the latter that it exists, then one can construe existence as a characteristic mark of the concept 'centaur', although there are no centaurs. I would not acknowledge anything as a centaur that was not outside my mind; this means that I shall not call mere ideas or feelings centaurs.
The existence expressed by 'there is' cannot be a characteristic mark of a concept whose property it is, just because it is a property of it. In the sen- tence 'There are men' we seem to be speaking of individuals that fall under the concept 'man', whereas it is only the concept 'man' we are talking about. The content of the word 'exist' cannot well be taken as the characteristic mark of a concept, because 'exists', as it is used in the sentence 'Men exist', has no content.
We can see from all this how easily we can be led by language to see things in the wrong perspective, and what value it must therefore have for philosophy to free ourselves from the dominion of language. If one makes the attempt to construct a system of signs on quite other foundations and with quite other means, as I have tried to do in creating my concept-script, we shall have, so to speak, our very noses rubbed into the false analogies in language.
? ? [Draft towards a Review of Cantor's Gesammelte Abhandlungen zur Lehre vom Transfiniten] 1
[1890--1892]
Since these papers have been published in this journal, my concern will be less to expound than to critically evaluate their contents.
Their aim is to gain acceptance for the actual infinite. This is achieved in part negatively by refuting attempted disproofs, and in part positively by demonstrating its existence. Some of the considerations advanced belong more to theology or the philosophy of religion, some more to mathematics or logic. I may here be allowed to confine myself to an evaluation of the latter, which are closer to my interests and which of themselves provide a wealth of material for discussion.
On the whole the objections to the infinite seem to me well and truly met. These objections arise because properties are ascribed to the infinite which do not belong to it: either properties of the finite are carried over as a matter of course to the infinite (p. 3), or a property that only belongs to the absolute infinite is extended indiscriminately to the infinite in general. It is a merit of these papers to have brought out this distinction within the infinite so forcibly (part equal to whole). All this relates only to the genuine, the 'actual' infinite. The opposition against acknowledging the actual infinite which mathematicians display is to be traced back in part to their confounding this with the potential infinite-an opposition which, properly speaking, holds only against construing the potential infinite as if it were the actual infinite. So, many mathematicians and philosophers only acknowledge the potential infinite. Cantor now succeeds in showing that this infinite
1 In the Zeitschrift fiir Philosophie und philosophische Kritik Cantor had published the following articles on the theory of the transfinite: Ober die verschiedenen Standpunkte in Bezug aufdas actuate Unendlichkeit (Vol 88 (1886) pp. 224-233), Mitteilungen zur Lehre vom Transfiniten (Vol. 91 (1887), pp. 81- 125, 252-270 and Vol. 92. (1888), pp. 240-265). In 1890 they were published together in a volume: Zur Lehre vom Transfiniten. Gesammelte Abhandlungen aus der Zeitschrift fiir Philosophie und philosophisehe Kritik. Erste Abteilung. They are also to be found in G. Cantor, Gesammelte Abhandlungen mathematischen und philosophischen Inhalts, ed. E. Zermelo (Berlin 1932). Frege reviewed the 1890 collection in the Zeitschrift fiir Philosophie und philosophisehe Kritik. See Vol. 100 (1892), pp. 269-272. Except for some small alterations, the first part of the piece from the NachlqjJ agrees with the first paragraph of this review. The agreement extends to the end of the sentence 'Cantor is less felicitous when he comes to giving definitions', but from then on the review diverges completely from what is printed here (ed. ).
? [Draft towards a Review o f Cantor's Lehre vom Transfiniten] 69
presupposes the actual infinite, that the 'receding limit' must have an infinite path, if it really is to recede ever and ever further (footnote to p. 30). 1
Cantor is less felicitous when he comes to giving definitions.
We may begin here by making a general observation. When negroes from 1he heart of Africa see a telescope or pocket watch for the first time, they are Hldined to credit these things with the most astounding magical properties. Many mathematicians react to philosophical expressions in a similar manner. I am thinking in particular here of the following: 'define' (Brahma), 'reflect' (Vishnu), 'abstract' (Shiva). The names of the Indian gods in hrackets are meant to indicate the kind of magical effects the expressions nre supposed to have. If, for instance, you find that some property of a thing bothers you, you abstract from it. But if you want to call a halt to this process of destruction so that properties you want to see retained should not he obliterated in the process, you reflect upon these properties. If, finally, you feel sorely the lack of certain properties in the thing, you bestow them on it by definition.
In your possession of these miraculous powers you are not far removed from the Almighty. The significance this would have is practically beyond measure. Think of how these powers could be put to use In the classroom: a teacher has a good-natured but lazy and stupid pupil. He will then abstract from the laziness and stupidity, reflecting all the while on the good-naturedness. Then by means of a definition he will confer on him the properties of keenness and intelligence. Of course so far people have confined themselves to mathematics. The following dialogue may serve as ltrt illustration:
Mathematician: The sign j=i has the property of yielding -1 when lljuared.
Layman: This pattern of printer's ink on paper? I can't see any trace of this property. Perhaps it has been discovered with the aid of a microscope or by some chemical means?
Mathematician: It can't be arrived at by any process of sense perception. And of course it isn't produced by the mere printer's ink either; a magic Incantation, called a definition, has first to be pronounced over it.
Layman: Ah, now I understand. You expressed yourself badly. You mean that a definition is used to stipulate that this pattern is a sign for aomething with those properties.
Mathematician: Not at all! It is a sign, but it doesn't designate or mean anything. It itself has these properties, precisely in virtue of the definition.
Layman: What extraordinary people you mathematicians are, and no mlHtakeI You don't bother at all about the properties a thing actually has, but imagine that in their stead you can bestow a property on it by a definition-a property that the thing in its innocence doesn't dream of-and
1The concept of a 'wandelnde Grenze' or 'wandelbare Grenze' [receding, oluanjling limit! was introduced by Herbart (ed. ).
? 70 [Draft towards a Review o f Cantor's Lehre vom Transfiniten]
now you investigate the property and believe in that way you can
accomplish the most extraordinary things!
This illustrates the might of the mathematical Brahma. In Cantor it is Shiva and Vishnu who receive the greater honour. Faced with a cage of mice, mathematicians react differently when the number [Anzahl] of them is in question. Some-and Biermann seems to be one of them-include in the number the mice just as they are, down to the last hair; others-and I may surely count Cantor amongst them-find it out of place that hairs should form part of the number and so abstract from them. They find in mice a whole host of other things besides which are out of place in number and are unworthy to be included in it. Nothing simpler: one abstracts from the whole lot. Indeed when you get down to it everything in the mice is out of place: the beadiness of their eyes no less than the length of their tails and the sharpness of their teeth. So one abstracts from the nature of the mice (p. 12, p. 23, p. 56). But from their nature as what is not said; so one abstracts presumably from all their properties, even from those in virtue of which we call them mice, even from those in virtue of which we call them animals, three-dimensional beings-properties which distinguish them, for instance, from the number 2.
Cantor demands even more: to arrive at cardinal numbers, we are required to abstract from the order in which they are given. What is to be understood by this? Well, if at a certain moment we compare the positions of the mice, we see that of any two one is further to the north than the other, or that both are the same distance to the north. The same applies to east and west and above and below. But this is not all: if we compare the mice in respect of their ages, we find likewise that of any two one is older than the other or that both have the same age. We can go on and compare them in respect of their length, both with and without their tails, in respect of the pitch of their squeaks, their weight, their muscular strength, and in many other respects besides. All these relations generate an order. We shall surely not go astray if we take it that this is what Cantor calls the order in which things are given. So we are meant to abstract from this order too. Now surely many people will say 'But we have already abstracted from their being in space; so ipso facto we have already abstracted from north and
south, from the difference in their lengths. We have already abstracted from the ages of the animals, and so ipso facto from one's being older than another. So why does special mention have also to be made of order? '
Well, Cantor also defines what he calls an ordinal type; and in order to arrive at this, we have, so he tells us, to stop short of abstracting from the order in which the things are given. So presumably this will be possible too, though only with Vishnu's help. We can hardly dispense with this in other cases too. For the moment let us stay with the cardinal numbers.
So let us get a number of men together and ask them to exert themselves to the utmost in abstracting from the nature of the pencil and the order in which its elem. ents are given. After we have allowed them sufficient time for
? ? [Draft towards a Review o f Cantor's Lehre vom Transfiniten] 71
this difficult task, we ask the first 'What general concept (p. 56) have you arrived at? ' Non-mathematician that he is, he answers 'Pure Being'. The second thinks rather 'Pure nothingness', the third-I suspect a pupil of Cantor's-'The cardinal number one'. A fourth is perhaps left with the woeful feeling that everything has evaporated, a fifth-surely a pupil of Cantor's-hears an inner voice whispering that graphite and wood, the wnstituents of the pencil, are 'constitutive elements', and so he arrives at the general concept called the cardinal number two. Now why shouldn't one man come out with the answer and another with another? Whether in fact Cantor's definitions have the sharpness and precision their author boasts of is accordingly doubtful to me. But perhaps we got such varying replies because it was a pencil we carried out our experiment with. It may be said 'But a pencil isn't a set'. Why not? Well then, let us look at the moon. 'The moon is not a set either! ' What a pity! The cardinal number one would be only too happy to come into existence at any place and at any time, and the moon seemed the very thing to assist at the birth. Well then, let us take a heap of sand. Oh dear, there's someone already trying to separate the grains. 'You are surely not going to try and count them all! That is strictly forbidden! You have to arrive at the number by a single act of abstraction' (footnote top. 15). 'But in order to be able to abstract from the nature of a grain of sand, I must surely first have looked at it, grasped it, come to know it! ' 'That's quite unnecessary. What would happen to the infinite cardinals in that case? By the time you had looked at the last grain, you would be bound to have forgotten the first ones. I must emphasize once more that you are meant to arrive at the number by a single act of abstraction. Of course for that you need the help of supernatural powers. Surely you don't imagine you can bring it off by ordinary abstraction. When you look at books, some in 4uarto, some in octavo, some thick, some thin, some in Gothic type and some in Roman and you abstract from these properties which distinguish them, and thus arrive at, say, the concept "book", this, when you come down to it, is no great feat. Allow me to clarify for you the difference between ordinary abstraction and the higher, supernatural, kind.
With ordinary abstraction we start out by comparing objects a, b, c, and lind that they agree in many properties but differ in others. We abstract from the latter and arrive at a concept tP under which a and b and c all fall. Now this concept has neither the properties abstracted from nor those common to a, band c. The concept "book", for instance, no more consists of printed sheets-although the individual books we started by comparing do consist of such-than the concept "female mammal" bears young or suckles them with milk secreted from its glands; for it has no glands. Things ure quite different with supernatural abstraction. Here we have, for instance, u heap of sand . . . 1
1 At this point the manuscript breaks ofT (cd. ).
? ? On the Concept of Number1 [1891/92]
[A Criticism ofBiermann]
In my Grundlagen (? 68) I called the concept F equal in number to the concept G if it is possible to correlate one-to-one the objects falling under F with those falling under G and then gave the following definition:
The number2 belonging to the concept F is the extension of the concept 'equal in number to the concept F'.
The following discussion will show that this definition gives the right results when applied, by deriving the basic properties of the numbers from it. But first we need to clarify a few points and meet some objections. The way the word number is used outside mathematics does not reveal the sense of the word with the clarity that is indispensable for scientific purposes. Things are, so to speak, dragged into the number, lock, stock and barrel; people imagine a number of trees as something rather like a group or row of trees, so that the trees themselves belong to the number. On this view, the number
of peas on the table would not only be changed, say, if I filed something off one of them or if I took away one and put another in its place; it would also be changed if I simply shuffled the peas about; for a change in spatial relations means that the whole is changed, as a heap of sand is changed if someone spreads it out without adding or taking away a grain of sand. If we ask what the number 2 is, likely enough the answer will be 'two things'. Now put an apple and a pear in front of someone who has given this answer and say 'Here you have your number 2'. Perhaps he will begin to hesitate at this point, but he will be even more unsure if we ask him to multiply this number 2 by the number 1 which we could give him, say, in the shape of a
1 According to notes on the transcripts on which this edition is based, both the following papers were found, unseparated, in a folder under the heading: 'On the Concept of Number'. They were probably written in the years 1891/92 since the section beginning on p. 87 is a preliminary draft of Frege's article Ober Begriff und Gegenstand, published in 1892. The section dealing with Biermann may have been written earlier. This is indicated by the use of the word 'Inhalt' (p. 85) where Frege's later practice would require 'Bedeutung' (ed. ).
2
Here and in the first five sentences of the paragraph following the word translated 'number' is 'Anzahf, which strictly means 'natural number'. Throughout most of the rest of the paper Frege uses 'Zahf, which has the same broad application as our 'number' (trans. ).
? ? On the Concept ofNumber 73
strawberry. It would, indeed, be very pleasant if we could conjure up a five- pound note by multiplying a cork by a nail. All this is pure childishness, of course, and that we have to bother with such things is a sufficient indictment of the times we live in. Yet even mathematicians will stoop to giving such definitions as:*
'The concept of number may be defined as the idea of a plurality composed of things of the same kind. When we use the term "one" for each of the elements of the same kind, counting the elements or units of the set consists in assigning the new terms two, three, etc. , to one and one-and one, etc. Number is the idea of the groups of elements designated by these terms. '
Expressions such as 'things of the same kind', 'elements', 'one', 'unit' are all jostled together here, as though the third section of my Grundlagen had never been written. Do these expressions have the same meaning, or not? If they do, why this plethora of terms? If they don't, what is the difference between them? ** At any rate, we may at least assume that 'elements' is intended to mean the same as 'things of the same kind' and 'group' the same as 'plurality'. Strange what importance is so often attached to a mere change in expression when presenting these rudiments. It seems almost as though the expressions 'one' and 'element of our group' are intended to have the same meaning, and yet the replacement of the latter by the former is stressed as an important step. Why does counting the elements or units of a set not consist in assigning new terms to element and element-and element? Or is that what it actually does? Let us assume, to clarify matters, that there is a lion standing beside a lioness lying on the ground and together they form a picturesque group of things of the same kind; both animals do, of course, belong to the species felis Ieo. At first we have nothing more than just this group. But now comes the highly significant act of designating both the lion und the lioness by the term one; then follows the no less significant act of assigning the new term two to one and one (and thus, indubitably, to our group); and we are very happy to have acquired the number two at this juncture. For surely what we need most of all is to have the terms and words; once we have them we can say: 'Number is the idea of groups of elements designated by these terms. ' Accordingly, in our case the number two would presumably be the idea of the group oflions. We might still ask what the word 'one' actually means in that case. Does it designate a number? According to Biermann's notion, one would then be a group of elements. Now we have just used the term one to designate the lioness. Since Ihe lioness is not an idea we must presumably accept the idea of the lioness
? Otto Biermann, Theorie der analytischen Funktionen [Leipzig 1887] ? I.
? ? I'll wager Biermann did not know this himself at the time and still does not know.
? ? 74
On the Concept ofNumber
as the meaning of the word 'one'. The lioness is made up of molecules, after all, and so may well be seen as a group.
Perhaps Biermann also takes the word 'group' in such a way that a single thing is also a group, no matter whether it is composite or not, although the contrary is indicated by the fact that Biermann speaks of elements, things of the same kind, and stresses the way a number is composed.
Admittedly, we also used one to designate the lion, and this could give us pause should we find the idea [image] 1 of the lion different from that of the lioness; on the other hand, the idea [image] we have can be so blurred that the idea [image] of the lion merges with that of the lioness. Hence, just as two is the idea [image] of our group of lions, one would be an idea [image] though a pretty blurred one, of a lion. What degree of vagueness we would have to assume admittedly remains uncertain. As a second example, let us take the Laocoon group, the well-known Greek sculpture found in 1506. We might well doubt whether in this case the components may be assumed to be of the same kind.
Biermann does not state to what extent the similarity he requires allows scope for differences. If the coincidence were exact, should we ever get beyond the number one? Presumably, then, a certain degree of latitude must be allowed. We also speak of two cats even if they are not the same colour, or two coins even if one is made of gold and the other of nickel. No doubt we shall always be able to discover some sort of similarity, even if it consists in nothing more than each of the elements having the property of being like itself in every respect. What then is the point of this condition of being of the same kind, when either it is always fulfilled, or we never know when it is fulfilled? Biermann does not know himself. He is onto something, but doesn't know what. He could have found out the answer from my Grundlagen, just as he could have learnt a great deal more that he does not know, but he must have thought in his heart of hearts: metaphysica sunt, non leguntur.
Let us return to the Laocoon group. Since the whole thing is made of marble and since, moreover, the parts of the group represent living beings, presumably there is nothing to prevent us from regarding the elements as being sufficiently alike. So what we do first-and this is very significant indeed-is to designate Laocoon, or more precisely his marble image, and each of his sons, and the serpent as well, by the term one. Then we again assign terms to one and one-and one . . . All right, what actually is one in this case? The idea [image] of a living being sculpted in marble which is so blurred we are not able to distinguish whether what is represented is the figure of a snake or a human being, an older or a younger man? That seems
1 The difficult word 'Vorstellung' we have rendered throughout as 'idea', sometimes enclosing the word 'image' in brackets afterwards where this sense seems more appropriate. Owing to the influence of Kant, the term was of course frequently used in German philosophical writings of Frege's time. The standard English translation of ? Vorstellung' as it occurs in Kant is 'representation' (trans. ).
? On the Concept ofNumber 75
a trifle implausible. In that case, 'one' would presumably have to have a different meaning from what it had in the previous example. Or must the idea [image] be taken to be so blurred that it is quite impossible to say what it is an idea of? Is there only one one? Or are there many ones? Could we say in the latter case 'Laocoon is a one' or 'The idea of Laocoon is a one'? Or can Biermann give us an example of something that is a one? What does it mean when someone says: this or that is a one? And if there are several ones, we must still be able to express the fact that something belongs to the species one. The phrase 'by designating each of the elements of the same kind by one' may lead someone to suppose that one is a title like 'Sir', for instance. We have the privilege of conferring this title, and everything and everyone receiving it is simply a one as a result. Then we could never go wrong in calling something one, just as a reigning monarch can never go wrong when he bestows a title on someone. Of course, this title 'One' would he pretty worthless; you cannot imagine it amounting to anything much. At 1he same time, conferring this title would be a source of revenue for us, in so far as it is through this that we come into possession of the numbers. If there were only one one, if, that is, 'one' were a proper name, what would 'one and one' mean? Clearly, one again; for what can 'Charlemagne and t 'harlemagne' mean, if not Charlemagne? Then there must be several ones after all, and our earlier conjecture that one is a vague idea is possibly false.
At this stage the most obvious thing seems to be to regard 'one' as a title. We have conferred this title on Laocoon, his sons and the serpent, and we uow assign a new term to the Laocoon group. Very well, but which?
If we are agreed on this, then a particular affirmative judgement such as ? some men are negroes' can be converted into 'Some negroes are men'. The resistance one naturally has to this is based on the fact that one involuntarily adds in thought 'but some negroes are not men'. This thought is excluded by our rider 'possibly all'.
Now you wanted to construe the expression 'Men exist' as having the same meaning as 'Something existing is a man'. The trouble with the latter is that, grammatically, it is not existence but being a man which occurs as predicate. But it is existence that is really meant to be asserted. Now we can adopt a grammatical form which brings this out by converting and saying 'Some men exist', taking this in the sense of 'Some-possibly all-but at least one man exists'. It will then follow that this means the same as 'Men exist'.
Now I have understood throughout that on your view you take the difference in meaning of 'exist' in 'Leo Sachse exists' and 'Men exist' to be of the same kind as the difference in meaning of'is a German' in 'Leo Sachse is u German' and 'Some men are German', so that 'exists' is related to 'exist' in the first two sentences as, in the last two, 'is a German' is related to 'are German'. I have deliberately made the subjects 'Leo Sachse' and 'some men' the same in both cases to create the impression that they do parallel one another. I believe the only reason we omit the 'some' in the sentence 'men exist' is to forestall the objection 'not all? '
Now you marshalled your argument, I believe, as follows.
First of all you wanted me to admit that 'There are men' means the same as 'Among that which has being is some man' or 'A part of what has being is a man', or 'Something that has being is a man'. As expressions with the same meaning as 'has being' you also used 'can be experienced', 'existing', 'that whose idea has been caused by something affecting the ego'. These
? ? 62 [Dialogue with Piinjer on Existence]
variants, I believe, make no essential difference. With them some secondary difficulties may crop up or disappear. The main difficulty remains the same and so does the general outline of the plan of attack. I now had further to be brought to admit that the verb 'to be' ('to exist') is used in the same sense as in the sentence 'Leo Sachse is' or 'exists'. ' With that you seemed to have won the day.
Now I can readily grant that the expression 'there are men' means the same as 'Something existing is a man'-only, however, on condition that 'exists' predicates something self-evident, so that it really has no content. The same goes for the other expressions which you use in place of'exist'.
But if the proposition 'Leo Sachse is' is self-evident then the 'is' cannot have the same content as the 'there are' of 'There are men', for the latter does not say something self-evident. Now if you express what is said by 'There are men' by 'Men exist' or 'Among that which has being is some man', that the content of the statement cannot lie in the 'exist' or 'has being' etc. And this is the npwrav l{leVOac; by which you could not help but be forced into making contradictory judgements-the error of thinking that the content of what is predicated in 'Some men exist' or 'Something existing is a man' or 'Men exist' is contained in the word 'exist'. This is not the case; this word only contains the form of a predicate as does the copula 'is' in the sentence 'The sky is blue'. Like the 'it' in 'it is raining', the 'exist' in 'Men exist' is to be understood as a mere auxiliary. As language, feeling at a loss for a grammatical subject, invented 'it', so here, feeling at a loss for a grammatical predicate, it has invented 'exist'.
I shall use the fact that instead of 'exists' one can also say 'is identical with itself' to show that the content of what is predicated does not lie in the word 'exists'. 'There are men' means the same as 'Some men are identical with themselves' or 'Something identical with itself is a man'. Neither in 'A is identical with itself' nor in 'A exists' does one learn anything new about A. Neither statement can be denied. In either you can put what you like for A, and it still remains true. They do not assign A to one of two classes in order to mark it off from some B which does not belong to that class. The point in saying 'A is identical with itself can only be to express the logical law of identity; the point cannot be to impart any further knowledge about A. Just as it could be maintained that 'exist' means the same in 'This table exists' and 'Tables exist', so we can say that the predicate 'identical with itself has the same sense in 'This table is identical with itself' and 'Tables are identical with themselves'. But in that case we have also to acknowledge that the judgements 'This table exists' and 'This table is identical with itself are completely self-evident, and that consequently in these judgements no real
1 If the reader is to follow the argument at this point, he should bear in mind that in German the expression 'there are', as in Frege's example 'There are men', is not rendered by a form of the verb 'to be' ('sein') but by 'es gibt' which comes from 'geben', meaning 'to give'. The corresponding English expression 'there is' is likewise so rendered (trans. ).
? [Dialogue with Piinjer on Existence] 63
content is being predicated of this table. Just as we call statements such as 'Men exist' existential judgements, thinking the content of what is predicated to lie in the word 'exist', so we might call the statement 'Some men are identical with themselves' an identity judgement, and 'There are men' would be an identity judgement. In general, in any attempt to prove that the content of what is predicated in 'There are men' was located in the 'exist' of 'Men exist', we could, without any falsification, switch 'are identical with themselves' throughout for 'exist'. I shall seek to show this.
But if the content of what is predicated in the judgement 'Men exist' does not lie in the 'exist', where then does it lie? I answer: in the form of the particular judgement. Every particular judgement is an existential judge- ment that can be converted into the 'there is' form. E. g. 'Some bodies are light' is the same as 'There are light bodies'. 'Some birds cannot fly' is the same as 'There are birds that cannot fly', and so on. It is more difficult to do the opposite and convert a judgement with 'there is' into a particular judgement. Out of context the word 'some' has no sense; it is an auxiliary like 'all', 'each', 'none' and so on, which, in the context of a sentence, has a logical function to perform. This function consists in putting two concepts into a certain logical relationship. In 'Some men are negroes' the concepts 'man' and 'negro' are put into this relationship. So we always need two concepts if we want to form a particular judgement. Now of course the sentence 'There are flying fish' can easily be converted into 'Some fish can fly', because we have two concepts 'fish' and 'being able to fly'. It becomes more difficult if we try to put 'There are men' into the form of a particular judgement. If we define man = rational living being, we may say 'Some living beings are rational' and, assuming the definition to be correct, this means the same as 'There are men'.
This recourse is only open to us when the concept can be analysed into two characteristic marks. There is another method closely connected with this one. E. g. if we have to convert 'There are negroes', we may say that negro = negro that is a man, because the concept 'negro' is subordinate to the concept 'man'. Here again we have two concepts and may say 'Some men are negroes' or 'Some negroes are men'. But this works only in the particular case of the concept 'negro'. For 'There are birches' we should have to select a different superordinate concept, such as 'tree'. If one wants to do the thing quite generally, one needs to look for a concept superordinate to all concepts. Such a concept, if one wishes to give it that name, can no longer have any content at all since its extension will be unlimited; for any content can only consist in a certain delimitation of the extension. As such a concept we might select that of 'being identical with itself, since we said that 'There are men' is the same as 'There are men identical with themselves' or 'Something identical with itself is a man'.
Language has availed itself of a different resource. The copula, i. e. the mere form of a predicate without content, was excellently suited for forming a concept without content. In the sentence 'The sky is blue' the predicate is
? 64 [Dialogue with Piinjer on Existence]
'is blue', but strictly the content of the predicate lies in the word 'blue'. Leave this out and what remains-'The sky is'-is a predicate without content. In this way we form a quasi-concept-'being'-without content, since its extension is unlimited. This makes it possible to say: men = men that have being; 'There are men' is the same as 'Some men are' or 'Something that has being is a man'. Thus here the real content of what is predicated does not lie in 'has being' but in the form of the particular judgement. Faced with an impasse, language has simply created the word 'being' in order to enable the form of the particular judgement to be employed. When philosophers speak of'absolute being', that is really an apotheosis ofthe copula.
But it is easy to see now how this came about. People felt that the sentence 'There is a centre of mass of the earth' is not self-evident, and that consequently there is a content to what is predicated. And it is now readily intelligible why, when they employed the form 'A centre of mass of the earth exists', people believed this content to reside in the word 'exist'. In this way a content was packed into the word 'exist', without anyone's being able to specify in what this content really consists.
It may now be shown how Piinjer, as a result of the 7C{Jwruv 'lfevar; of seeing in the 'exist' the content of what is predicated in 'Men exist', was bound to be driven into making contradictory assertions. I was easily able to convince him that the denial of 'A can be experienced' is impossible where can be experienced = is = exists. He had also to concede that to say of a thing that it can be experienced does not characterize that thing in any way. On the other hand, however, he wanted to salvage a content for a statement saying of something that it can be experienced. Something was surely meant to be said in the sentence 'This table can be experienced', 'This table exists'; it was surely meant to state something that was not pleonastic or self-evident. So he was forced into the contradiction of taking the denial of 'This table can be experienced' as something that was not pleonastic and self-evident. 1 He had to deprive the expression 'can be experienced' of all content without making it devoid of content. Piinjer wanted to convey the content of the judgement 'This can be experienced' by 'The idea of the this is not an hallucination, it is not something which originates from myself alone, but the idea has been formed as a result of the ego's being affected by the this'. To this I had to object that one only has the right to form the expressions 'idea of the this', 'the ego's being affected by the this', after having made the judgement 'something corresponds to this idea of mine'. If nothing corresponds to this idea of mine, then the expression 'idea of the this' has no sense and thus the sentence as a whole has no sense. Piinjer then altered his account, though without conceding that it was incorrect: 'The object of the idea B can be experienced' means 'The idea B has been formed through something affecting the ego'. I was now able to infer from this that the negation of t11e sentence 'The object of the idea B can be experienced' has a
1 sic (trans. ).
? [Dialogue with Piinjer on Existence] 65
good sense. But Piinjer had said earlier that the negation of the sentence 'A can be experienced' is impossible. We now have to qualify this somewhat and say: If A is an object of experience, then the negation of 'A can be experienced' is impossible, but if A is object of an idea, then the negation of that sentence is possible. We see how this example confirms that it is impossible to assign to the predicate 'can be experienced' a sense that is not self-evident whilst making the general claim that it has no sense to deny of a thing that it can be experienced. At the same time we see that the concept of what can be experienced acquires a content only through its extension being delimited in some way. In fact all objects are divided into two classes: objects of experience and objects of ideas. The latter do not all fall under the concept 'what can be experienced'. From this it can be further inferred that not every concept is subordinate to the concept of what can be experienced-for the concept 'object of an idea' is not. And from this it follows that the concept of what can be experienced is not generally suited for the purpose of expressing a judgement with 'there is' in the form of a particular judgement. In order to justify the expression 'object of an idea' as being generally applicable, Piinjer had to maintain that every idea has an object, that there are ideas of objects which have not been formed through something affecting the ego. If we apply to this his definition of sentences with 'there is', a contradiction must emerge. In fact, according to this definition, the judgement 'there are objects of ideas that have not been formed through something affecting the ego' is synonymous with 'Among what can be experienced is something that falls under the concept "object of an idea that has not been formed through something affecting the ego"'. But now according to Piinjer's account, objects of ideas not formed by virtue of something affecting the ego cannot be experienced. So we arrive at the statement 'Among what can be experienced is something that cannot be experienced'.
We may also put it like this: from the two premisses
I. There are objects of ideas that are not formed through something affecting the ego;
2. Objects of ideas that are not formed through something affecting the ego cannot be experienced;
there follows the conclusion:
There are objects of ideas-which objects cannot be experienced. This is
a contradiction once it is allowed that the same kind of existence is expressed by 'there is' as is meant to be conveyed by 'can be experienced'.
In general one can lay down the following:
If you want to assign a content to the verb 'to be', so that the sentence 'A is' is not pleonastic and self-evident, you will have to allow circumstances under which the negation of 'A is' is possible; that is to say, that there are subjects of which being must be denied. But in that case the concept 'being'
? 66 [Dialogue with Piinjer on Existence]
will no longer be suitable for providing a general explanation of 'there are' under which 'there are B's' means the same as 'something that has being falls under the concept B'; for if we apply this explanation to 'There are subjects of which being must be denied', then we get 'Something that has being falls under the concept of not-being' or 'Something that has being is not'. There is no way of getting over this once a content of some kind-it doesn't matter what it is-is agreed to the concept of being. If the explanation of 'there are Bs' as meaning the same as 'Something that has being is B' is to work, we just have to understand by being something that goes entirely without saying.
For this reason the contradiction still remains if we say 'A exists' means 'The idea of the A has been caused by something affecting the ego'. However, still other difficulties crop up-. here, only some of which I want to mention.
When Leverrier put to himself the question whether there were planets beyond the orbit of Uranus, he was not asking whether his idea of a planet beyond the orbit of Uranus had been caused, or might have been caused, by something affecting the ego. When there is a dispute over whether there is a God, the dispute is not over whether our idea of a God has been caused, or may be caused, by something affecting the ego. Many who believe that there is a God will dispute that their idea of Him has been caused by God's immediately affecting their ego, for it can only be a question here of something affecting them immediately. However, this is only by the way. The upshot is as follows:
We can say that the meanings of the word 'exist' in the sentences 'Leo Sachse exists' and 'Some men exist' display no more difference than does the meanings of 'is a German' in the sentences 'Leo Sachse is a German' and 'Some men are Germans'. But then the sentence 'Some men exist' or 'Something existing is a man' only means the same as 'There are men' if the , concept 'existing thing' is superordinate to the concept man. So if such forms of expression are to have the same meaning in general, the concept 'existing thing' must be superordinate to every concept. This is only possible if the word 'exist' means something that goes entirely without saying, and if therefore nothing at all is predicated in the sentence 'Leo Sachse exists', and if in the sentence 'Some men exist' the content of what is predicated does not ' lie in the word 'exist'. The existence expressed by 'there is' is not contained in the word 'exist' but in the form of the particular judgement. 'Some men are Germans' is just as good an existential judgement as 'Some men exist'. But once the word 'exist' is given a content, which is predicated of an individual thing, this content can be made into the characteristic mark of a concept-a concept under which there falls the individual thing of which existence is being predicated. E. g. if one divides everything into two classes
1. What is in my mind, ideas, feelings etc. and
2. What is outside myself,
? [Dialogue with Piinjer on Existence] 67
and says of the latter that it exists, then one can construe existence as a characteristic mark of the concept 'centaur', although there are no centaurs. I would not acknowledge anything as a centaur that was not outside my mind; this means that I shall not call mere ideas or feelings centaurs.
The existence expressed by 'there is' cannot be a characteristic mark of a concept whose property it is, just because it is a property of it. In the sen- tence 'There are men' we seem to be speaking of individuals that fall under the concept 'man', whereas it is only the concept 'man' we are talking about. The content of the word 'exist' cannot well be taken as the characteristic mark of a concept, because 'exists', as it is used in the sentence 'Men exist', has no content.
We can see from all this how easily we can be led by language to see things in the wrong perspective, and what value it must therefore have for philosophy to free ourselves from the dominion of language. If one makes the attempt to construct a system of signs on quite other foundations and with quite other means, as I have tried to do in creating my concept-script, we shall have, so to speak, our very noses rubbed into the false analogies in language.
? ? [Draft towards a Review of Cantor's Gesammelte Abhandlungen zur Lehre vom Transfiniten] 1
[1890--1892]
Since these papers have been published in this journal, my concern will be less to expound than to critically evaluate their contents.
Their aim is to gain acceptance for the actual infinite. This is achieved in part negatively by refuting attempted disproofs, and in part positively by demonstrating its existence. Some of the considerations advanced belong more to theology or the philosophy of religion, some more to mathematics or logic. I may here be allowed to confine myself to an evaluation of the latter, which are closer to my interests and which of themselves provide a wealth of material for discussion.
On the whole the objections to the infinite seem to me well and truly met. These objections arise because properties are ascribed to the infinite which do not belong to it: either properties of the finite are carried over as a matter of course to the infinite (p. 3), or a property that only belongs to the absolute infinite is extended indiscriminately to the infinite in general. It is a merit of these papers to have brought out this distinction within the infinite so forcibly (part equal to whole). All this relates only to the genuine, the 'actual' infinite. The opposition against acknowledging the actual infinite which mathematicians display is to be traced back in part to their confounding this with the potential infinite-an opposition which, properly speaking, holds only against construing the potential infinite as if it were the actual infinite. So, many mathematicians and philosophers only acknowledge the potential infinite. Cantor now succeeds in showing that this infinite
1 In the Zeitschrift fiir Philosophie und philosophische Kritik Cantor had published the following articles on the theory of the transfinite: Ober die verschiedenen Standpunkte in Bezug aufdas actuate Unendlichkeit (Vol 88 (1886) pp. 224-233), Mitteilungen zur Lehre vom Transfiniten (Vol. 91 (1887), pp. 81- 125, 252-270 and Vol. 92. (1888), pp. 240-265). In 1890 they were published together in a volume: Zur Lehre vom Transfiniten. Gesammelte Abhandlungen aus der Zeitschrift fiir Philosophie und philosophisehe Kritik. Erste Abteilung. They are also to be found in G. Cantor, Gesammelte Abhandlungen mathematischen und philosophischen Inhalts, ed. E. Zermelo (Berlin 1932). Frege reviewed the 1890 collection in the Zeitschrift fiir Philosophie und philosophisehe Kritik. See Vol. 100 (1892), pp. 269-272. Except for some small alterations, the first part of the piece from the NachlqjJ agrees with the first paragraph of this review. The agreement extends to the end of the sentence 'Cantor is less felicitous when he comes to giving definitions', but from then on the review diverges completely from what is printed here (ed. ).
? [Draft towards a Review o f Cantor's Lehre vom Transfiniten] 69
presupposes the actual infinite, that the 'receding limit' must have an infinite path, if it really is to recede ever and ever further (footnote to p. 30). 1
Cantor is less felicitous when he comes to giving definitions.
We may begin here by making a general observation. When negroes from 1he heart of Africa see a telescope or pocket watch for the first time, they are Hldined to credit these things with the most astounding magical properties. Many mathematicians react to philosophical expressions in a similar manner. I am thinking in particular here of the following: 'define' (Brahma), 'reflect' (Vishnu), 'abstract' (Shiva). The names of the Indian gods in hrackets are meant to indicate the kind of magical effects the expressions nre supposed to have. If, for instance, you find that some property of a thing bothers you, you abstract from it. But if you want to call a halt to this process of destruction so that properties you want to see retained should not he obliterated in the process, you reflect upon these properties. If, finally, you feel sorely the lack of certain properties in the thing, you bestow them on it by definition.
In your possession of these miraculous powers you are not far removed from the Almighty. The significance this would have is practically beyond measure. Think of how these powers could be put to use In the classroom: a teacher has a good-natured but lazy and stupid pupil. He will then abstract from the laziness and stupidity, reflecting all the while on the good-naturedness. Then by means of a definition he will confer on him the properties of keenness and intelligence. Of course so far people have confined themselves to mathematics. The following dialogue may serve as ltrt illustration:
Mathematician: The sign j=i has the property of yielding -1 when lljuared.
Layman: This pattern of printer's ink on paper? I can't see any trace of this property. Perhaps it has been discovered with the aid of a microscope or by some chemical means?
Mathematician: It can't be arrived at by any process of sense perception. And of course it isn't produced by the mere printer's ink either; a magic Incantation, called a definition, has first to be pronounced over it.
Layman: Ah, now I understand. You expressed yourself badly. You mean that a definition is used to stipulate that this pattern is a sign for aomething with those properties.
Mathematician: Not at all! It is a sign, but it doesn't designate or mean anything. It itself has these properties, precisely in virtue of the definition.
Layman: What extraordinary people you mathematicians are, and no mlHtakeI You don't bother at all about the properties a thing actually has, but imagine that in their stead you can bestow a property on it by a definition-a property that the thing in its innocence doesn't dream of-and
1The concept of a 'wandelnde Grenze' or 'wandelbare Grenze' [receding, oluanjling limit! was introduced by Herbart (ed. ).
? 70 [Draft towards a Review o f Cantor's Lehre vom Transfiniten]
now you investigate the property and believe in that way you can
accomplish the most extraordinary things!
This illustrates the might of the mathematical Brahma. In Cantor it is Shiva and Vishnu who receive the greater honour. Faced with a cage of mice, mathematicians react differently when the number [Anzahl] of them is in question. Some-and Biermann seems to be one of them-include in the number the mice just as they are, down to the last hair; others-and I may surely count Cantor amongst them-find it out of place that hairs should form part of the number and so abstract from them. They find in mice a whole host of other things besides which are out of place in number and are unworthy to be included in it. Nothing simpler: one abstracts from the whole lot. Indeed when you get down to it everything in the mice is out of place: the beadiness of their eyes no less than the length of their tails and the sharpness of their teeth. So one abstracts from the nature of the mice (p. 12, p. 23, p. 56). But from their nature as what is not said; so one abstracts presumably from all their properties, even from those in virtue of which we call them mice, even from those in virtue of which we call them animals, three-dimensional beings-properties which distinguish them, for instance, from the number 2.
Cantor demands even more: to arrive at cardinal numbers, we are required to abstract from the order in which they are given. What is to be understood by this? Well, if at a certain moment we compare the positions of the mice, we see that of any two one is further to the north than the other, or that both are the same distance to the north. The same applies to east and west and above and below. But this is not all: if we compare the mice in respect of their ages, we find likewise that of any two one is older than the other or that both have the same age. We can go on and compare them in respect of their length, both with and without their tails, in respect of the pitch of their squeaks, their weight, their muscular strength, and in many other respects besides. All these relations generate an order. We shall surely not go astray if we take it that this is what Cantor calls the order in which things are given. So we are meant to abstract from this order too. Now surely many people will say 'But we have already abstracted from their being in space; so ipso facto we have already abstracted from north and
south, from the difference in their lengths. We have already abstracted from the ages of the animals, and so ipso facto from one's being older than another. So why does special mention have also to be made of order? '
Well, Cantor also defines what he calls an ordinal type; and in order to arrive at this, we have, so he tells us, to stop short of abstracting from the order in which the things are given. So presumably this will be possible too, though only with Vishnu's help. We can hardly dispense with this in other cases too. For the moment let us stay with the cardinal numbers.
So let us get a number of men together and ask them to exert themselves to the utmost in abstracting from the nature of the pencil and the order in which its elem. ents are given. After we have allowed them sufficient time for
? ? [Draft towards a Review o f Cantor's Lehre vom Transfiniten] 71
this difficult task, we ask the first 'What general concept (p. 56) have you arrived at? ' Non-mathematician that he is, he answers 'Pure Being'. The second thinks rather 'Pure nothingness', the third-I suspect a pupil of Cantor's-'The cardinal number one'. A fourth is perhaps left with the woeful feeling that everything has evaporated, a fifth-surely a pupil of Cantor's-hears an inner voice whispering that graphite and wood, the wnstituents of the pencil, are 'constitutive elements', and so he arrives at the general concept called the cardinal number two. Now why shouldn't one man come out with the answer and another with another? Whether in fact Cantor's definitions have the sharpness and precision their author boasts of is accordingly doubtful to me. But perhaps we got such varying replies because it was a pencil we carried out our experiment with. It may be said 'But a pencil isn't a set'. Why not? Well then, let us look at the moon. 'The moon is not a set either! ' What a pity! The cardinal number one would be only too happy to come into existence at any place and at any time, and the moon seemed the very thing to assist at the birth. Well then, let us take a heap of sand. Oh dear, there's someone already trying to separate the grains. 'You are surely not going to try and count them all! That is strictly forbidden! You have to arrive at the number by a single act of abstraction' (footnote top. 15). 'But in order to be able to abstract from the nature of a grain of sand, I must surely first have looked at it, grasped it, come to know it! ' 'That's quite unnecessary. What would happen to the infinite cardinals in that case? By the time you had looked at the last grain, you would be bound to have forgotten the first ones. I must emphasize once more that you are meant to arrive at the number by a single act of abstraction. Of course for that you need the help of supernatural powers. Surely you don't imagine you can bring it off by ordinary abstraction. When you look at books, some in 4uarto, some in octavo, some thick, some thin, some in Gothic type and some in Roman and you abstract from these properties which distinguish them, and thus arrive at, say, the concept "book", this, when you come down to it, is no great feat. Allow me to clarify for you the difference between ordinary abstraction and the higher, supernatural, kind.
With ordinary abstraction we start out by comparing objects a, b, c, and lind that they agree in many properties but differ in others. We abstract from the latter and arrive at a concept tP under which a and b and c all fall. Now this concept has neither the properties abstracted from nor those common to a, band c. The concept "book", for instance, no more consists of printed sheets-although the individual books we started by comparing do consist of such-than the concept "female mammal" bears young or suckles them with milk secreted from its glands; for it has no glands. Things ure quite different with supernatural abstraction. Here we have, for instance, u heap of sand . . . 1
1 At this point the manuscript breaks ofT (cd. ).
? ? On the Concept of Number1 [1891/92]
[A Criticism ofBiermann]
In my Grundlagen (? 68) I called the concept F equal in number to the concept G if it is possible to correlate one-to-one the objects falling under F with those falling under G and then gave the following definition:
The number2 belonging to the concept F is the extension of the concept 'equal in number to the concept F'.
The following discussion will show that this definition gives the right results when applied, by deriving the basic properties of the numbers from it. But first we need to clarify a few points and meet some objections. The way the word number is used outside mathematics does not reveal the sense of the word with the clarity that is indispensable for scientific purposes. Things are, so to speak, dragged into the number, lock, stock and barrel; people imagine a number of trees as something rather like a group or row of trees, so that the trees themselves belong to the number. On this view, the number
of peas on the table would not only be changed, say, if I filed something off one of them or if I took away one and put another in its place; it would also be changed if I simply shuffled the peas about; for a change in spatial relations means that the whole is changed, as a heap of sand is changed if someone spreads it out without adding or taking away a grain of sand. If we ask what the number 2 is, likely enough the answer will be 'two things'. Now put an apple and a pear in front of someone who has given this answer and say 'Here you have your number 2'. Perhaps he will begin to hesitate at this point, but he will be even more unsure if we ask him to multiply this number 2 by the number 1 which we could give him, say, in the shape of a
1 According to notes on the transcripts on which this edition is based, both the following papers were found, unseparated, in a folder under the heading: 'On the Concept of Number'. They were probably written in the years 1891/92 since the section beginning on p. 87 is a preliminary draft of Frege's article Ober Begriff und Gegenstand, published in 1892. The section dealing with Biermann may have been written earlier. This is indicated by the use of the word 'Inhalt' (p. 85) where Frege's later practice would require 'Bedeutung' (ed. ).
2
Here and in the first five sentences of the paragraph following the word translated 'number' is 'Anzahf, which strictly means 'natural number'. Throughout most of the rest of the paper Frege uses 'Zahf, which has the same broad application as our 'number' (trans. ).
? ? On the Concept ofNumber 73
strawberry. It would, indeed, be very pleasant if we could conjure up a five- pound note by multiplying a cork by a nail. All this is pure childishness, of course, and that we have to bother with such things is a sufficient indictment of the times we live in. Yet even mathematicians will stoop to giving such definitions as:*
'The concept of number may be defined as the idea of a plurality composed of things of the same kind. When we use the term "one" for each of the elements of the same kind, counting the elements or units of the set consists in assigning the new terms two, three, etc. , to one and one-and one, etc. Number is the idea of the groups of elements designated by these terms. '
Expressions such as 'things of the same kind', 'elements', 'one', 'unit' are all jostled together here, as though the third section of my Grundlagen had never been written. Do these expressions have the same meaning, or not? If they do, why this plethora of terms? If they don't, what is the difference between them? ** At any rate, we may at least assume that 'elements' is intended to mean the same as 'things of the same kind' and 'group' the same as 'plurality'. Strange what importance is so often attached to a mere change in expression when presenting these rudiments. It seems almost as though the expressions 'one' and 'element of our group' are intended to have the same meaning, and yet the replacement of the latter by the former is stressed as an important step. Why does counting the elements or units of a set not consist in assigning new terms to element and element-and element? Or is that what it actually does? Let us assume, to clarify matters, that there is a lion standing beside a lioness lying on the ground and together they form a picturesque group of things of the same kind; both animals do, of course, belong to the species felis Ieo. At first we have nothing more than just this group. But now comes the highly significant act of designating both the lion und the lioness by the term one; then follows the no less significant act of assigning the new term two to one and one (and thus, indubitably, to our group); and we are very happy to have acquired the number two at this juncture. For surely what we need most of all is to have the terms and words; once we have them we can say: 'Number is the idea of groups of elements designated by these terms. ' Accordingly, in our case the number two would presumably be the idea of the group oflions. We might still ask what the word 'one' actually means in that case. Does it designate a number? According to Biermann's notion, one would then be a group of elements. Now we have just used the term one to designate the lioness. Since Ihe lioness is not an idea we must presumably accept the idea of the lioness
? Otto Biermann, Theorie der analytischen Funktionen [Leipzig 1887] ? I.
? ? I'll wager Biermann did not know this himself at the time and still does not know.
? ? 74
On the Concept ofNumber
as the meaning of the word 'one'. The lioness is made up of molecules, after all, and so may well be seen as a group.
Perhaps Biermann also takes the word 'group' in such a way that a single thing is also a group, no matter whether it is composite or not, although the contrary is indicated by the fact that Biermann speaks of elements, things of the same kind, and stresses the way a number is composed.
Admittedly, we also used one to designate the lion, and this could give us pause should we find the idea [image] 1 of the lion different from that of the lioness; on the other hand, the idea [image] we have can be so blurred that the idea [image] of the lion merges with that of the lioness. Hence, just as two is the idea [image] of our group of lions, one would be an idea [image] though a pretty blurred one, of a lion. What degree of vagueness we would have to assume admittedly remains uncertain. As a second example, let us take the Laocoon group, the well-known Greek sculpture found in 1506. We might well doubt whether in this case the components may be assumed to be of the same kind.
Biermann does not state to what extent the similarity he requires allows scope for differences. If the coincidence were exact, should we ever get beyond the number one? Presumably, then, a certain degree of latitude must be allowed. We also speak of two cats even if they are not the same colour, or two coins even if one is made of gold and the other of nickel. No doubt we shall always be able to discover some sort of similarity, even if it consists in nothing more than each of the elements having the property of being like itself in every respect. What then is the point of this condition of being of the same kind, when either it is always fulfilled, or we never know when it is fulfilled? Biermann does not know himself. He is onto something, but doesn't know what. He could have found out the answer from my Grundlagen, just as he could have learnt a great deal more that he does not know, but he must have thought in his heart of hearts: metaphysica sunt, non leguntur.
Let us return to the Laocoon group. Since the whole thing is made of marble and since, moreover, the parts of the group represent living beings, presumably there is nothing to prevent us from regarding the elements as being sufficiently alike. So what we do first-and this is very significant indeed-is to designate Laocoon, or more precisely his marble image, and each of his sons, and the serpent as well, by the term one. Then we again assign terms to one and one-and one . . . All right, what actually is one in this case? The idea [image] of a living being sculpted in marble which is so blurred we are not able to distinguish whether what is represented is the figure of a snake or a human being, an older or a younger man? That seems
1 The difficult word 'Vorstellung' we have rendered throughout as 'idea', sometimes enclosing the word 'image' in brackets afterwards where this sense seems more appropriate. Owing to the influence of Kant, the term was of course frequently used in German philosophical writings of Frege's time. The standard English translation of ? Vorstellung' as it occurs in Kant is 'representation' (trans. ).
? On the Concept ofNumber 75
a trifle implausible. In that case, 'one' would presumably have to have a different meaning from what it had in the previous example. Or must the idea [image] be taken to be so blurred that it is quite impossible to say what it is an idea of? Is there only one one? Or are there many ones? Could we say in the latter case 'Laocoon is a one' or 'The idea of Laocoon is a one'? Or can Biermann give us an example of something that is a one? What does it mean when someone says: this or that is a one? And if there are several ones, we must still be able to express the fact that something belongs to the species one. The phrase 'by designating each of the elements of the same kind by one' may lead someone to suppose that one is a title like 'Sir', for instance. We have the privilege of conferring this title, and everything and everyone receiving it is simply a one as a result. Then we could never go wrong in calling something one, just as a reigning monarch can never go wrong when he bestows a title on someone. Of course, this title 'One' would he pretty worthless; you cannot imagine it amounting to anything much. At 1he same time, conferring this title would be a source of revenue for us, in so far as it is through this that we come into possession of the numbers. If there were only one one, if, that is, 'one' were a proper name, what would 'one and one' mean? Clearly, one again; for what can 'Charlemagne and t 'harlemagne' mean, if not Charlemagne? Then there must be several ones after all, and our earlier conjecture that one is a vague idea is possibly false.
At this stage the most obvious thing seems to be to regard 'one' as a title. We have conferred this title on Laocoon, his sons and the serpent, and we uow assign a new term to the Laocoon group. Very well, but which?
