'2-...t\-.~~ ..


.The C~assicsof

the ~t.~OhnIS Progr~


(1845 -.19l.8)
Three papers on transfinite numbers from the Mathematische Vol. XLVI (189S). PP. 481-512. Annalen



vol. XLIX (1$97). pp. 207-246.
~ •

Vol. XXI (1883) PP. 545-591. (Grundlagen)•

Translated jrom the.German

.'~ ,

George ~. Bingley

Je suis tel1ementpour 11infini actuel, qu'au'lieu d1admettre, que la nature l'abhorre, comme lIon dit vulgairement, je tiens qu1elle l'affecte par-tout, pour mieux marquer les perf'ectdons de son AuteUr. - Leibnitz

Copyright 1941by George A. Bingley


(Jlathematische Annalen, Vol. XXI, p. 545-91; '1883) Georg Cantor in Halle (Continuation of the article Also reprinted, in Vol. XXI, p. 51) "Grundlagen einer allgemeinen

Leipzig 1883, under the title


§ 1.
The presentation at last of m1 investigations up.to now in the·theor,y continuation depends upon

reached a point where further beyond its

'~''''''f''''''\''U of thE! ccncept of real integer furthElrroon" this extension lies ha.s proceeded.

limits as accepted today,

"in a direction

along l\1'l.ich to


This dependence, which brings me face to face with an extension ~umber conoept ~ is so great,. that. vd. thout such an extension it possible to take even the slightest may then, for this

YO uld


~tep forward in the theory of sets; or, it necessary, even

reason, be found justification" apparently unfamiliar

for introducing

ideas into my considerations. of the series of real inte-

it is a question of an extension and continuation into the infinite

(·,1ber das Unendliche); daring ae this may seem, I venture this extension will one. I make no

not only the hope but even·the firm cqnvictionthat

be recog~izedas

an entirely

simple I proper, and natural

to conceal the fact that in this

undertaking I submit a point of view

i~ considerable

opposition to widely accepted opinions on the mathernati-

cal infinite

and to frequently presented views on the nature of numerical quan-

As to the mathematical infinite, has been found a justifiable application

it seems to me that in so far for it in the sciences and it in the

has been employedfor their advancement, it has appeared principally

.role of a variable which grows [546] beyond all limits or di.mirishesto any • desired minuteness, but which always appears as a quantity remaining finite. I call the ideal-infinite (Uneigentlich-unendlich)*.

But in recent times and even at the present day, in geometry and in the theory of functions, another equally justifiable kind of concept of function

"'1nfini ty has arisen, 'in which for example, in the study of en analytic a complex variable,

it has becomethe geperal and necessary usage to think in the plane .of the complexvariable, that is, the

of a single point at infinity infinitely distant

but nevertheless definite

point" and to investigate

of the function in the neighborhood of this point; from which it turns out.that the behavior of the function in the neighborhood of the infinitely exactly the same properties as at any point in the finite

distant point exhibits

so that full justification

is assured to think of the infinite

in this

as lying at a completely determined point. Whenthe infinite actual-infinite appears in such a definite form I call it


, *Transllltoris Note: . The translator is not entirely satisfied with his renderings of "Uneigeritlich" and IlEigentlichU as ideal and actual, in that they fail to showthat the former is the negative of the later; "non-actualll and "actuulJt seemedtoo artificial. "Realtl for UEigentlich" is fraught with countless ambiguities. Bertrand Russell suggests "improper" for "Uneigentlich".

These two roles in which the mathematical infinite appears,
c.rI...,J .......

which, in both forms, imnense progress has been made in geometry" in

sis, and in mathematical physics, must be carefully distinguished from other if we are to understand what follows. In the first form, as ideal-infinite, it appears as finite and

in the other form, which I call the actual-infinite,

it is seen to be

The infinite real integers which I shall define and to whose study I had devoted a considerable period of years I became clearly aware that they were to be regarded as concrete numbers real meanin~ , have absolutely nothing in co~on , with the first of these .

, with the ideal-infinite, but possess rather the same cho.racter of deterss which is the property of the pqint at infinity of analytic functions; belong therefore to the forms and characters of the actual-infinite. reas the point nt infinity in the plane of the complex numbers stands alone distinot from all the points of the finite region we obtain not a single inte [547J integer but an infinite series of such, which are clearly differented from each other and· stand in orderly number-theory relations to each other well as to the finite integers. These relationships are not however such as

be referred in their unqerlying principles to the relations of finite n~ to each other; this kind of relationship (that of finite numbers to each
may appear, however, as it frequently does in the case of the various in-

...,,,, ,,,s ".. ... ,,, (St~rken) and forms of the ideal-infinite, ...

for example, for functions

which become infinitely small or infinitely large, in case they determinate finite order number-s' in their process of becoming infinite. relationships can only be looked upon Ln fnct as veiled mterrelations of

the finite or of such quantities as can be directly referred to the finite; the

'* Centor's Note:

Up to now I have called them IIde finitely defined symbols of vid. Math. Ann. Vol. XVII, p. 357; Vol. XX p. 113; Vdl. XXI p. 54.

·laws governing the actual-infinite integers (yet to be defined) are , on the · other hand, fundamentally different from the relationships which obtain for the finite, from which it is not unlikely that the finite real numbers the~ selves may gain certain new determinations with aid of the actual infinite numbers. The two erfnciples of gener~tion, by whose help, as will be shown, these new definite infinite numbers will be defined, are of such a natureJ that through their united operation ever,y bnrrier seemingly inherent in the , concept of the real integers can be broken through; fortunately, however, a :~ ~ principle appears as we shall see J which I cell the limiting or restrict-· principle, b,y which successive definite limits are imposed upon the endless

·process of generation, so that we obtain natural divisions in the absolutely infinite succession of the real tntegers, Which divisions I call number-classes ,(Zahlenclassen) •
The first number-class (I) is the set of the finite integers

2, 3,••••••••• , v,._ •••••, which is followed by


second number-class (II),

:consisting of certain infinite integers following each o~her in a determined succession; after defining the second number-class the third is reached, then fourth . &tc. The introduction of these new integers SGems to me, turthermore, to be of the grea.test importcnce for the development and clarification of .the concept of power (Uichtigkeitsbegriff) which I have introduced and fre-

quently applied in Ge.rlier portions of ntV -work. According to this the~ry a power is aasocd.at-ed with every well-defined set, in that the same

assigned to two sets if a reciprocal one-to-one correspondertce can be up between the two, element for element.

the case of finite sets power coincides with the number

(Anzahl) of elements since such sets, as is known, have the same number,


of elements in every ordering of them.
In the case

ot infinite sets, however, there has been no men-

tion up to now of a precisely defined enumeration (Anzahl) of their elements either in my works or else:where, although the concept of a power Which is en.tirelyindependent of any particular ordering has been discussed. The smallest power of infinite sets as is easily justified is .assigned to those sets which are related in a reciprocal one-to-one correspondence to the first number-class and aocordingly have the same power as that But as yet there has been lacking an equally simple, natural definihigher power s,

above mentioned number-classes of the definite-infinite

integers are now seen as the natural representatives, presented in unified 'form, of the orderly ascending succession of powers of well-defined sets. I

sert specifically that the power of the second number-class (II) differs not from the power ot the first number-class, but that it is in fact the next we may call it the second power or the power of the second class. the third number-class gives rise to the definition of the third

or the power of the third class, etc.

§ 2.

Another important gain which is attributable to these new is acquired, for me at least, in a new concept, now appearing for the first time, that of the enumeral (Anzahl) of a well-ordered (wohlgeordneten) ... infinite manifold (Mannichfaltigkeit); since this concept is always expressible

a completely determined number of our extended number domain so far as the order of the elements is determined in a manner immediately to be defined and since also the notion of enumeration representation (Anzahlbegiff) finds a directly objective

in our intuition (inneren Anschauung),

is the reality of the

demonstrated even for the cases in which they are definite-infinite,

through this connection between enumeral and number.

a well-ordered

set is Jll3:ant every well-defined

set \'lhose

elements are connected with each other by a definite prescribed succession, according to which there exists a first elament of the set and, further, every element of which is followed in case it is not the last in the succession by another ~nich is determinate, and also to eve~J finite or infinite subset of which there exists an element which is next in the succession to all the e1ements of that subset (unless it happens that there is no successor to the ps.rticular subset selected).


[549] Two well-ordered sets are said to be of

the same enumeral (With respect to the particular prescribed succession) when a one-to-one reCiprocal correspondence of such a kind is possible

that, if E

and F are any two elements of the one, El and PI the corresponding clements of the other, then the position of E and F in the succession of the first set is
in agreement with the position of El and Fl in the succession of the second set,

such that.t if E precedes F in the succession of the first set, then El precedes F1in the succession of the second set. This correspondence if at all possible,

is, as is easily seen, always a completely determined one and since in the extended number series :there exists one and on1yone nwnber 0(
, such that the

numbers which precede it (from 1 on) in the natural succession have the same enumera1, one is compelled to set the enumaral of both of these well-ordered sets as 0(



is an infinitely large number and as the number 0( -1

which immediately precedes 0( , if


is a finite integer.

The essential difference between finite and infinite sets is seen in this, that a finite set gives rise to the same enumera1 (Anzah1) of elements in every succession which one may impose upon its elements; while a sct composed of an infinity of elements will in general lead to different enumerals depending upon the succession which may be given its elements. The

power of a set, as we have seen, is a property independent of the ordering;


themumeral. is seen to be a property depending in general upon a given There




succession of elements, as soon as it is a question of infinite sets.

is, nevertheless, even in the case of infinite sets, a certain relationship between the power of the sct and the enumeral of its elements determined by a given succession. First, if we take a set which has the power of the first class, and give to its elements!!!.! determined succession so that it becomes a IIwell-ordered" set, its enumeral is always a definite number of the second number-class and can never be dlilterminedby the number of any other than the second number-class,
On the other hand, every set of the first power

may be. ordered in such a succession that its enurnerai, with respect to this

euceesedcn; becomes equal to any preassigned number of the second number-class.
We can express

these theorems as follows:

every set whose power is of the

first clae;.s countable (abzWbar) 'is

by numbers of the second number-class

and only by such, [550] and it is always possible to give to the set such a succession


its elements that it is countable in this succession by


arbitrarily chosen number of the second number-class, which number is given by the enumernl of the elements of the set with respect to thnt succession. Analogous laws hold for sets of higher powers.
Every well-

defined set whose power is of the second class is countable by numbers of the third number-class an~ only by such, and further it is always possible to give such a succession to its elements that it is countable* by any arbitrarily chosen number of the third number-olass, Which number is determined by the enumeral of the elements of the set with respect to that succession. *Tha.t which I·have called ",countable" (abz5hlbar) in the earlier numbers of this work, is, according to the sharper and more general definition which I have now introduced, nothing more than "countability" (or enumorability) (Abzahlbarkeit) by n~ers of the first class (finite se~s) or by numbers of the second 'cle.ss (sots of the first power).






The ~oncept of the well-ordered

" 3.

set turns out to be fundamental

for the whole theory of sets. Whether it is possible to reduce every welldefined set to the form 'of a well-ordered one, a fundamental and very useful


law of thought especially remarkable for its universal validity -- will be taken up in a Inter article*. the concept of the well-ordered Here I restrict myself to proving how out of set the fundamental operations for the integers can be derived in a very simple manner

whether finite or definite-infinite,

and how the laws governing these operations follow directly with apodictic certainty from intuition ( inneren Anschauung). If two well-Qrdered sets

M and Ml are given whose enumerals correspond to the numbers then M + Ml is also a weU-ordered



f3 '

set, which results if first the set M is and united with itj there then

considered and is followed by the set ~

corresponds to the set M + Ml a determinate number as enumeral in relation to the given succession of its elements; let this number be called the sum of


and ~


designated as, ~


; it is at once seen ..that, if differs in general from ~ + ~ It is now

and ~

are not both

finite ..C( + ~

The commutative law ceases to hold in general for addition.

simple to construct the concept of the sum of several given terms (SUmmanden) of given sequence, even to the extent of considering a succession which is itself a definite-infinite one, [551] that I need not go into further detail

and need only remark that the associative law is seen to be of universal vali*Translator's Note: The celebrated proof of Zermelo that such a reduction is always possible has given rise to heated contnJversies. The postulate, usually call~d the axiom of choice, upon which his proof rests, has not been .acepted by some as of universal validity.

dity. In particular

cc + (~

...we have:


r ) = (0< + ~ J + r .
elements equals determined by the num-

If one takes a sequence of sets all of which are alike and similarly 0( ber

ordered and for each one of Whichthe enumeral of its and if the succession of the sets is definitely

t' ' then one obtains
the definition

a new well-orgered set whose corresponding enumeral
O(~ ,

I Ii:


for t?e product

where ~

is the multiplier

and ex. from

the multiplicand;

here also, it turns out that in general the cOnmnltativelaw for multiplication law for multiplication



that is,

is not generally valid.

· But the associative

always holds, so that we have

O«(~r) =(D(~)('
Among these new numbers someare distinguished that they possess the prime number property, · be characterized in a very definite although this from others' in property mst now

way by saying that by a prime number is
0( ::

"meant a number for Whieh the factOring: is only possible When ".the multiplicand


(3 G '


where ~

is the nn.1l.tiplier,

1 or ~ =-

ex. ;

but in general for prime numbers 0(

will have a certain region of indeterminacy which in the
spite of this, as will be shown in a prime factors in (in

· nature of things can not be changed. · later



a number can be expressed in terms of its


only one way, and with respect to the sequence of its factors

so far as these are not neighboring finite duct) •

prime numbers appearing in the proprime numbers are found of to the finite prime numbers, different charac-

Accordingly two kinds of definite-infinite stands in a closer relationship

"Which the first

while the prime numbers of the second Idnd have an entirely

Further, knowledge to establish of my article,

it now becomes possible with the help of this new a theorem which appears at the conclusion



to the theory of ma.nifoldslt (Borchardt.t s Journal; linear infinite manifolds.

Vol. 84, p. 257) on the so-called

In the last issue (4) of this article (Vol. XXI, p. 54) I ved a_theorem for point-sets continuous region, P which are contained in an n-dimensional as follows by applying the new, well-

which can be stated

,defined terms (Ausdrucksweisc): (Ableitimg)

lilt P is a point-set whose deriva.tive Where

P (01)vanishes identicallY~


is an;y arbitrarily

chosen inp(l)

teger of the first or second number-class, ,therefore P itself, is a point-set

then the first derivative

, and

of. the pOwer of the



It seems

rather remarkable that this

theorem has a converse Whichcan be expressed as Whosefirst derivative of the ~ has the power of or second numberhaving

IIIf p [552] is a point-set

=--. class,
,this property,

then there exist integers


class, for which P (oe) identically there is a least".

vanishes ~ and of all the numbers 0(

I shall very shortly publish

the proof of this

theorem, at the

suggestion of my distinguished

friend, Prof. llittag-Leffler

in stock-

in the first volume of the new mathematical the conclusion of :which he will

journal which he will edit. will add an appendix in


Herr Mittag-Leffler

show, how on the basis of this theorem an important generaliza-

can be given his investigations and those of Prof. Weierstrass on the
of single-valued analytic

functions with given singular points.

The extended series of integers, if there is need to do so can be enlarged quite simply into a continuous number-set by merely adjoining to every integer C( , nIl real numbers x which are greater than zero and less than

, ,

:' . I ~i

Perhaps at this point the question will suggest itself, Whether, since in the manner described a definite extension of the region of real n~ bers into the infinitely great has been attained it would not be possible to
do the same with equal success for the infinitely small, or, what might turn





102. out to be the same, 'Whether it would be possible to define finite numbers (which would appear as limiting values of series of irrational numbers) which would not coincide with either the rational or irrational numbers but would insert themselves among the real numbers as the irrational numbers among the rational or the transcendental algebraic numbers. The question of the establishment of such interpolations

numbers into the texture of the

upon which some authors have expended considerable effort, can, in ion,and


as I shall show, be answered precisely and clearly first with the

aid of our neW numbers and particularly on the basis of the universal concept of enumeration of well-ordered ·sets; While the previous attempts, as it seems to me, rest partly upon a confusion between the ideal-infinite and

the .actual infinite, partly upon a thoroughly unsound and Wllvering foundation. The ideal-infinite has often been characterized by recent

philosophers as a IIbadll (schlcchtes) infinite, in my opinion quite unjustly since it has shown itself to be a very good and highly useful instrument in mathematics and the natural sciences. ti~ies, so far The [553] infinitely small quan-

as I know,

have until now been usefully devcloped only in the and are as such ca.pable of aU those variations,

form of the ideal-infinite moditiccDions

and interrelations

which are employed in infinitesimal analy-

sis as well as in function theory and ~nich necessarily arise if the rich harvest of analytic truths is to be gathered. this infinitely small into an actual-infinitely nbendoned as to no purpose. But all attempts to force small must finnIly be small quantities

If somehow actual-infinitely

did exist, i.e. were definable, they would certainly stand in no direct relationship to those familiar quantities which are in the process of becoming infinitely s~

In contrast

to these attempts to advance through the infinitethe two forms of the infinite there is frequently

ly small and to ~terchange

found a point of view on the nature and meaning of numerical quantities denies the actual ~s.tence
~ ..

of any other numbers than the integers

of our n~




(I) .. At most a cert~in reality is granted the rational But as for the irrationals, numbers which they are

proceed eo di~ctly

from the integers.

to be assigned l~:rely a formal meaning in pure nathematics, in that they serve ~ some ~erse as signs useful for calculation (Rechenmarken), for es-

tabl,ishing pr?pert:1.e$ of groups of integers uriifying ,laY._ exclusively

and describing these in a simple, is composed, in this opinion, and analy-

The actual JlBterial of analysis

of f:i,nite, real integers

and all truths in arithmetic

sis already discovered or still tionships of the finite integers

to be discovered muet be looked upon as relato each other; the infinitesimal analysis (legalisirt)

and with it the theory of functions ,are considered to be legalized only in so far as their the finite integers.;

theorems are demonstrable through laws holding for With this conception of pure mathematics, although I are undeniably associated certain advan-

am not at all in agreement with it,

tages, which I should here like to emphasize; for the importance of this point of view bespeaks the fact that amongits most meritorious mathematicians oftha followers belongs a group of the

present day. integers exist and

If, as is here assumed, only the finite all else is nothing other than forms of relationships then it for their can be expected that the proofs of analytical

(Beziehungsformen), theorems can be tested

"number-bheory content" and that every gap which appears in them out according to the basic prinCiples of arithmetic; of such a "filling-out" is to be seen the true criterion in the for

can be filled feasibility

the genuinesa and complete rigor of the proofs. that in this way the establishment

[554] It is not to be denied

of many theorems may be fully achieved and

104. improvements in method introduced in various fields of analysis i there is
in the strict observance of the principles Which flow from this point of

, a guarantee against every kind of nonsense or error. In this way a definite, if somellhat over-restrl'.ined and obprinciple is set up, that is recommended to all as a test; it should serve show any flight of mathematical speculation and fancy its true limits, with-

runs no danger of hurtling into the abyss of the "transcendental", it is said in tear and wholesolOO alarm that "anything is possible". knows, for it is an uncertain matter, if it was not merely the point of of usefulness which actuated the authors of this principle to recommend
t to those soaring forces Which so easily encounter danger through their excess

immoderation, as an effective regulation for their protection against all

although a fruitful prinCiple can not be found therein; for the assump-

that they themselves in finding new truths had proceeded from these princiis for me out of the question in that I, however good some pages of their
may be, Dillstregard them in the strictest sense as erroneous; we owe to

principles no true advances and if their recoomaaaat10Qshad

been fully

science would have been retarded or at least hemmed by the narrowest Fortunately matters are really not that bad and the observanoe of

rules as Vlell as obedienoe to them ..useful, under some circumstanoes and never been taken entirely literally; also, up

to now, so far

strikingly true that no one who has made the attempt has been formulate the proper rule better or more fully than is here attempted

If we look about in history it appears that similar views often been adva~ced qnd even as early as Aristotle, It is well known that coming from But if one con-

middle ages the principle of "infinitum actu non dctur1t, was accepted by all the scholastics as inviolable.

arguments vmich Aristotle ~ presented against the real existence

the infinite
t they


his Metaphysics, Book XI. Chap. 10) I it will be found -mich involves

reter back to an assumption,



a petitio




ILss,um;pvJ~on, namely, that there are only finit9 numbers, from which he concluded
to him only enumerations (Z'hlungen)

of finite

sets were recognizable.


that I have proven above and it will appear even more clearly in what
IU.l..L .....IVQ

in this

paper (555] that determinate enumerations of infinite

sets can


as well as



ones, assuming tha.t a definite

law is given

sets by which they become well-ordered.
~l;Jg ......

That without such an ordered euethis lies in

",u of the elements of a set

no enUmeration can be made -

nature of the concept enumeration; even for finite
made o~v in case there is a definite

sets an enumeration can

succession of the elements to be enurnera.sets is seen, namely, order-

except also that here a peculiar property of finite the enumeration ; While for infinite

the enumeral -

is independent of the particular

sets as we have seen, such an independence is in general

the case , but the enwneral of an infinite

set is given by the infinite


which deterndnes
the nature essential

the pa.rticular

law of enumeration; it is in thiS, embedded
never to be gotten rid of, that one finds

of things and therefore difference

be"tween the finite

and the infinite;

never a,ga.incan

existence ot the infinite of the finite

be denied because of this


but ra.ther

can now be firmly maintained;

if one of the two is abandoned,

other must go also; where should we be then? Another argument advanced by Aristotle against the reality of and nwnber



in the assumption that the finite it' it (the infinite) existed,

would be dissolved since the finite is in truth

by the infinite


destroyed by the "infinite; seen, tha.t to a infinite a finite

the situation

as follows as

llill be clearly

number if it is thought of as detemin·

~e nnd complete (vollendet) i

number can indeed be adjoined and united

)ut affecting

the dissolution

of the latter

(the finite


-- the innumber; only

tnite number is itself

modified by such an adjunction of a finite

106. reverse procedure, the adjunction of an infinite latter is given first, effects the dissolution
ting any modification .infinite, to analysis entirely of the former. number to a finite one if

of the latter Without introview of the finite and

This corrected

unknown to Aristotle,

should give new stimulation the natural sciences.

not only '

but to other sciences,


I have been logically
gards the infinitely

forced to the point of view which re~thout in



great not merely in the form of that Which increases
related form of convergent series first

·lilnit and in the closely


the seventeenth century, but also [556] which fixes (fixiren) it by numbers in
·the determinate form of the completed-infinite (vollendctunendlich) -- I have

been forced to this

point of view almost against my will since it is in opposi-

to traditions.which
on over long years,

I value, by scientific
and I believe


and investigations


that no arguments can be advanced against

.it which I shall not know how to ~et.

§ 5.
When I just now spoke of traditions, I Wo.s thinking rather of them to the

not in the narrower sense of that Which has long passed,' but refer founders of the newer philosophy and science.

As a means of passing judgment

. on the question which is being brought up here, I submit some of the most impor. ta.nt references.

The following should be consulted:
and XVII.' Descartes J Letters
Locke, Essay on the human understanding, Book II, Chaps. XVI

and comments on his meditations; also Principia I, 26. Spinoza,' Letter XXIX; cogitnta metaph. parts I and II. Le1bnitz, ErdmrumEdition, pageS'l3BJ 244, 436,· 744; Pertz

307, 322, ~;

Edition, II, I, page 209; III, 4, page 218; III, 5, pages III, 7, page 273*. Hobbes, de corpore cap, VII, 11. of human knowledgeJ CXCVIII - CXXXI.


following are also of interest: Berkeley J TrcC'.tise on the principles

Stronger arguments than these against the introduction integers of in-

can hardly be imagined even today; let these arguments be exA detailed and thorough discussion of and significant letter of

amined and compared in turn with mine.

these references as well as of the very pertinent

Spinoza to L. Meyerwill be reserved for another occasion, for the momentI have merely the following to say·. Howeverdifferent their judgmept of the finite the doctrines of these writers may be, in they all

and infinite

in the above references

.. essentially infinite

agree that finiteness

is a property of number and that the true As to the latter

or Absolute, which is God, permits no determination.

: of these two points I fully

agree as it cannot be otherwise since the theorem beside the point; but as as I have already pointed

"onmis determinatio est negatioll is tor me entirely . to the first point I see in it a petitio principii,

out in my discussion of Aristotle's

argument against the "infinitum actu'J, to be found in nearly all these The assumption that besides

which throws light on $1Y contradictions writers but particularly

in Spinoza and Leibnitz.

the Absolute which is not obtainable by any detennination

and the finite,


......... ... J_.~~

are no modifications which, although not finite,

nevertheless -

are deter-

numbers and are therefore

what I call the actual-infinite

this assumption

: I find to be thoroughly untenable and it stands, in my opinion in actual contrato certain theorems proposed by the Last, two. philosophers in the above have demonstrated in this work as there is a trahs-


What I declare and believe·to papers, is, -

well as in earlier finite

that following the finite


which might also be called

supra..,.finite (suprafinitum):

that is, there is an unlimited ascending ladder of nodes, Which in its nature
is not finite determinate, but infinite, well-defined but which can be deterndn~d as can the finite and distinguishable, quantities numbers. It is ~ by

conviction and

that the region of definable that the limits

is not exhausted with the finite

of our knowledge may be accordingly extended without doing our

I now modify the Aristotelian-scholastic sed in paragraph 4 as follows ~ theorem dis-

Omnia seu finita
deter.minari possunt.'l The finiteness the reason why only finite :assertion

seu infinita

definita et excepto Deo ab

of humanunderstanding is often advanced as

numbers are conceivable; but again I see in this It is tacitly assumed that

the vicious circle already mentioned.


of understanding" is meant that its

grasp with respect to num-

bers is limited to the finite.

But if it turns out that in a determinate infinite, that is supra-

. sense the understanding can also define and distinguish finite

(..iberendliche)numbers, then either the words llfinite understanding"
be given an enlarged meaning,

from Ylhich the aforementioned conclusion "infinitell must be permitted the

'can no longer be drawn; or the predicate

understanding with certain qualifications, this latter being in my opin10n the right procedure. The words "finite understanding" which one hears so may be,


.often in no way hit the mark; however limited in truth human nature

.it still

partakes greatly

of the infinite


! declare that if in many reas to

spects it were not itself the existence . In particular

the firm conviction and certainty

of the Absolute, to Which \'Ie. all agree, would be inexplicable • I submit the thought that the human understanding is an unbounded

abode for the step-b,y-step construction of the integer number-classes, which " stand in definite relationships to the infinite modes and whose powers are of ascending magnitude. The chief difficulties in the systems of the two last-named philosophers, which outwardly are so different but in an inner sense are so closely related, may be brought nearer to a solution, as I believe, if the point of view suggested by me be adopted, and some of the difficulties may now be entirely satisfactorily solved and explained. It is these very diffi-

culties which have given rise to later criticism, which, in turn, for all its


s has not assured a suitable substitute for the retarded development doctrines of Spinoza and Leibnitz. Alongside or in place of the mechani-

explanation of nature, Which, inside its spheres has all the aids and res of mathem9.tical analysis at its disposal but whose onesidedness and weaks have been strikingly pointed out by Kant, there has been up to now.not the beginning of an attempt to create an organic eXplanation of ~~ture which

equipped with the same nathematical rigor; ground can be broken for such

attempt, as I believe, only by taking up afresh the works of these philosophers continuing them.
An especially

difficult point in Spinoza' s system' is the rela;I I~


to the infinite or the infinite with respect to still higher infinities.

example already touched upon in paragraph 4 seems in its simple symbolism to way along which one can come nearer perhaps to a solution of this If'tJJ is the first number of the second number-class, 1 then 1 + tJ = GJ ,



+ 1), where (W + 1) is a number entirely distinct from (J •

depends as is here clearly seen, upon the position (Stellung) of the relative to the infinite; in the first case tho finite is absorbed into infinite and Vanishes, but if it modestly takes its plnca ~ the infinite

intact and unites with the infinite to form a new (since modified)

e 6.
If difficulties (~bgeschlossene) arise in imagining such infinitJl~ great, segreintegers, which nre comparable to each other and to the

.finite integers, and are bound to the finite numbers by fixed laws, it will be .found that these difficulties depend upon the realization that the new nwnbers

have in some aspects the character of the preceding ones but from several other


I: L



rio, s of view have an entirely

new and peculiar


in that

it my often

that disparate


(559] my be -united in one and the same number,
number in which they are always separated. given in the preceding pa.rabe even as well

te unlike the case of the finite observation

is made in one of the references

that an·infinite

number if it existed would necessarily cannot co-exist

and since the two properties such number exists. It is here tacitly have previously

in one and the same number,

assumed that properties

which for numbers

understood them are disjunct,

are equally so for the new of infinite numbers. or extenJ

IlWDnel's, and one. accordingly concluded the impossibility fails

to see this fallacy at a glance?


every generalization

of concepts associated with the abandonment of certain unthinkable without it? Weren't complex quantities,

special properties so important for the

lopment of analysis

and leading to such tremendous advancements but recently

:nt'roclueedwithout seeing any hindrance in the fact that they can neither


or negative?

And it ia only a similar

step that I amtaking

indeed it will probably be much easier

for general conviction

Bewusstsein) to follow me, than was possible in the case of complex for the new integers, characterized (the finite as they are numbers), Whereas the until their

emerging from the real;

a more intensive

determinacy than their

predecessors as these

enumerals the same kind of reality of complex quantities representation trouble. To return briefly us look agsdn at the number G.) those properties


was accompanied with difficulties

by' points or vectors in the plane was found after

to the consideration ,in

of oddness and eveness,

order to show, how in it may be found number. In were given and I

which are never united in a finite of addition and multiplication

ph 3 general definitions

lPU<<2.P ..

"~g~ the fact that in these operations the conunutative law had no validisaw an essential difference between finite and infinite numbers. ~ by

in this.1

I remind you that in a product the multiplicand.


by ~

I understand the mu1 tiplier

Twoforms are immediately found for

6J: eN ::: (,) • .t. and

• Accordingly (.J

be thought of as an even and as an odd numeven

Fromanother point of view it can also be said that GJ is neither odd, since, as can be easily proven, G.)

is neither [560] representable to its

form ~.o{ nor in the form
L~'.U·~ predecessors



The number 6J in contrast peculiar nature,

has therefore

an entirely

since all these

s and characters

are united in it.

Evert more strange are the succeedshow.

numbers of the second number class as I shall later

Although in paragraph 5 I have cited many passages in Leibniz t; in which he speeks against infinite
........ """, ... , 11 II

numbers, in that he says amongother ni de ligne ou autre quanti te infinia,

n I y a point de nombre Wini

les prend pour des Touts veritables. ation,c'est forme un finill 1labsolu; au contraire,



n ' est pas une

d~s qU'on modifie on se borne ou statement, but

(in the latter

quotation I agree with the first

,not with the second), nevertheless to cite diets

lam in the fortunute position of being able sense contrn-

passages of the arume philosopher in which he in a certain himself for the actual-infinite

himself and declares

(not the Absolute)

. in no ambig'llousway.

He says in the Erdm.'UlIl edition, pour l'infini

P. 118:

"Je suis tellement que la nature l'abhorre, partout,

actuaL, qu'au lieu d+admet.t.re
je tiens qu'elle l'affecte

comme lion dit vulgairement,

pour mieux nk~rquerles perfections

de son Auteur.

Ainsi je crois qulil mais

nly a aucune p~rtie de 10 mnti~re qui ne soit, aetuellament divise6;

je no dis pas divisible,

at p~r consequent la moindre partice1le de creatures

doit etre consider-

/ ee comme mondeplein dt une infinite un


';f'Lc' '.

The actual-infinite,. sets or in the constitution

as for example it is found 1nwell-defined of bodies from point-like atoms




don't mean

chemical-physical atoms of Democritus which I can not regard as existent ther conceptually or really, up .to a certain point by this although muchthat is useful hee been achieved fiction), has fO\Uldits.st critical defender


in a

very keen-minded philosopher and. mathematician of our century, Bernhard and valuable wo rk,

Balzano who has developed his ideas in his beautiful

·riparadoxesof the infinite that the contradicti~s

II ,

Leipzig, 1851, Whosepurpose it is to demonstrate of


':an times have investigated,
4·. 11'-;-

of the infinite which the sceptics and peripatetics
do not exist if one takes the trouble,

not always

to refiect

c()llscientiously upon the concepts of infinity
. •

in their true

ftthe ideal-infinite,
iJlr.'aelrie~s or

/k" "~nature and content.

In this book [561] is also to be found a discussion of pertinent in ma.tzy points, as it appears in the form of

of the first

and higher orders or in the swmnationof infinite processes. 'l'his infinite (called by some scholastics

other limiting

syncategorematic infinite) •to express relations,

is merely a concept of our thought to aid and included variability and of which

which in his definition

the "daturll can never be said in sense .f the actual. It is' rather remarkable that ~ ldth respect to this kind of no essential . philosophers, positivists infinite, differences of opinion prevail even amongpresent-day

except for the fact that certain modern schools of so-called or realists4); or materialists believe that in this sYpcategorematic

of which they themselves must admit that it has no actual being,

they see the highest concept (h~chsten Begriff). Even in Leibniz an essentially correct point of view is found to this ideal-infinite,

in many passages; the folloWing for example refers Erdmannedition p. 436:

Ego philosophice loquendo non magis sta~u. rnagnitudines infinite parvas quam infinite magnas seu non J1I!.gis , infinite simas quaminfinituplas.


enim per modum loquendi compendiosum pro mentis fictionibus habeo , ad aptis, quales etiam. sunt radices ilnaginariae':inAlgebra. Interim

vi, magnum. has expressiones usum habere ad. compendium cogitandi adeoque

inventionem; et in errorem ducere non posse, cum. .pro infinite parvo substituere ficiat tam parvum. quam quis valet, ut error sit in1nor dato, unde consequitur

Bolzano is perhaps the only one for whom there is a justification actual-infinite numbers, at least he says a great deal about them; but I

disagree with the nammer in which he operates with them without being to give a correct definition of th~m and'I regard, 'for example, paragraphs 33 of his book as errcneous and neaningless. The author, fails to give any

ble concept of definite-infinite
. •• $

nwnberfj:)or the general concept of power, It is true'that germs of both these

the precise concept of enumerability.

are found here and there in special instances but he fails to develop these to full clesrity and determinacy, thereby failing to explicate many inconstencies and even


errors in this worthy book.

Witholit these two concepts, in my conviction, one can not advance
.... ... ''''rn _

in the theory of sets and the same holds, I believe, of the fielcm Which

subsumed under it or have close a~sociation with it, [562] ,for example, rootheory on one side and logic and epistemology on the other. When

the infinite, as I have done in this an~ my earlier works there a genuine pleasure to which 1 thankfully yield, in seeing how the concept integer which,for the finite case has only the background of enumeration, when ascend to the infinite, separates into two concepts, one of power which is of the order which is given the elements of a set, and one of enumerais necessarily associated with a definite ordering of the set by virtue of Which the latter becomes well-ordered. And if I descend again from the infi"

the finite i see how equally clearly and beautifully, the two concepts and c'oalesce to form the concept of finite integer.




We can speak of the reality or existence of the mtegers" finite or infinite, in two senses:
in fact they are the same two

ionships in which the reality of any concepts or ideas are to be oonsiFirstly, integers may be considered real in so far as they occupy an definite place in our understanding on the basis of definitions, can precisely differentiated from all other parts of our thought and stand in relationships to those parts" and accordingly modify the substance thought (Geistes) in a determinate fashion; I propose to call this kind reality of our numbers their intrasubjective· or imnanent reality 5) • Sec. i.

, reality can be ascribed to numbers In so far as they must be taken a.s

is exterior to the intellect, as, for instance, the various number·,classes (1) (II) (III) etc. are representatives of powers which are actually

.found in corporeal. and intellectual (geistigen) nature.
, "r,

This second species

of reality 1 call the transsubject:i..ver transient reality of the integers. o In lieu of the thoroughly realistic but at the same time none idealistic basis of mY considerations, there is no doubt in ~ mind that these two spheres of reality are always found together in the sense that a concept said to exist [5~3] in the first sense always also posse~ses in certain and even in an infinity of ways a transient reality ~" whose dcterrrd.na-

tion it must be gr~~ted becomes for the most part one of the most troublesome and profound problems of metaphysics nnd must frequently be left to times· in which the natural development of one of the other sciences eventually reveals the trnnsient meaning of the concept in question. The connection of both realities has its peculiar foundation in the unity (Einheit) of the !!l(Alls), to which we ourselves belons. A

mention of this connection has only the value here of enabling one to derive from it a result which seems of very great consequence for mathe~~tics, namely,

115. it· (mathematics) in the construction of its ideas has ~
" .;"

and ·solely

take account of the immanent reautY· of: its concepts and hae no obllgation'"'·
''''::IT ........,... ' .

to make test~ fo~·'their transient


On account of this


singular position which differentiate~ it from all of the other sciences and
an explanation

for the comparatively easy and unrestrained

manner With

which one may operate with it, it particularly deserves the name of free mathematics.(treien precedence Mathematik), a designati.on which if I had my way, would take

over the usual exprese:ion



Mathematics is ent1rely tree "in its deve.Lcpment,and is only


that its c.oncepts must be consistent


·~aoh.~ther and stand in dete~ate

(through definitions), orderly relationships

~~ th?se concepts which have preceded, these being already present and established



(mathe~ttc~) is obligated when new nwnbers are introduced to give
of them by·which such a determinacy and, under conditions such a


. rela.tions.hip to the older numbe%'si~ granted them, that they can in any given
. ,case be definitely distinguished fies real, all

from each other~

As soon ax a number satis-

these conditions it must be regarded as m&thematically existent and
. . I see the reason given in paragraph 4 why the rational,

It is in this that

irra.t.ionaland the complex numbers are to be considered finite poeit1ve integers.


much existent

as the

There is no need, as I believe, of fearing any danger to the

sciences in these principles, as many thi.nJC: the conditions imposed under which the freedom of constructing numbers can be excercised ~re of such a kind that they permit very little room for arbitrary action {Spiclraum); for every mnthem?tical concept carries within itself the necessary corrective; if it is
t ' "

fruitless ~nd purposeless, that is soon seen and it will [564] because of little success be abandoned. to mathematical

But I regard every superfluous restriction on the urge

investigation as creating a much grente'r danger and all the

greater since no justification can be advanced from the nature


the science;

nature (Wesan) of mathematics lies in its very freedom. If this property of mathematics had not been deducible from the

ve·Rmeln~ionedreasons, the very development of the science itself as it has would have inevitably led to the same conclusions. If Gauss, Cauchy, Abel, Jacobi, Dirichlet, Weierstrass, Hermite·

Riemann had been forced always to submit their new ideas to a metaphysical (Controlle), we should certainly not now enjoy the magnificent structure

new theory of functions which, although created and erected in full freewithout transient purpose nevertheless reveals its transient meaning in

to mechanics, astronomy, and mathematical physiCS, as was to be we should not be witnessing the great advance in the theory of dif~eI"1em:'1.au.

equations through Fuchs, Poincare and


others, if these great in-

forces had been restrained &Od restricted by outside influences; and Kummer had not allowed himself full freedom in his successful introduction 1I1d"al-J' numbers into number theory, we should today not be in a position of .the jmportant and excellent algebra.ic and arithematical WJrks of

How justified mathematics
'from. all metaphysical

is in moving with complete freedom

fetters, I do not, however, grant the same right to

.lIappliedtl mathematics as for examp'le analytic mechanics and mathem.~tticD.l physics; these di"iplines .are iri Jl\Y opinion in their founda'td.oneas well as in their if they strive to free themBelves from it as has been re2.

aims, metaphysical; cently suggested by

celebrated physicist, they take on

the form of a IIdes-

cription of nature", in which is lacking not only the fresh a.ir of free mathemetical thought but also the power of explaining (Erkl~rung) and laying the foundations (Ergr&ndting) of natural phenomena.


Because of the great


-mioh is attached


to 1'eal"

and irrational

nutbbers in the theory of manifolds,

I 4d not Wish I am not going

omit to say some important things about their definitions.
speak of the introduction rigorous of the rational

numbers in any detail


arith uet.1c' presentations

have frequently

been made; among

se with Which 1 ammost familiar



. Berlin, 1861) and J. H. T • Mtill.er(Lehrbuch der allgemeinen ArithI do" however" though briefly .. "Wish to speak in more (with Which I am familiar and which

I single out H. Grassmann (Lehrbuch der

, Hille, 1855).

of the three lending definitions identical)

·probably are essentially

which have been proposod for int~oducL~g arithmetical

the general real number in a rigorous is the form which Prof'. Weierstrass lectures on analytic functions


The first

of these

has made use of for many years in his of which are to be found in

and some traces

the Prograrmnabha.ndlung of Herr. Berlin,

E. Kossak. (Die Elemente der Arithmetikj


The second i{3 an unusual form. published by Herr Dedekind
Continuity and irrational numbers, Brunswick" 1872 and the (Math. Annalen,

in his work; third

is a form of definition given by me in the year Ian has a certain ~

Vol. V, p. 123);1 which superficially

similarity to Weierstrass' Mathematik und Physik"

form, so much so that HerrH. Weber (Zeitschrift

27 Jahrg., historisch liter. Abth., p. 163) confused the two; in my opinion
this third, later developed by Herr Lipschitz (Grundlagen der Analysis, of all and one finds in it the Calculus.

Bonn, 1877), is the simplestand JOOstnatural advantage that it

is most directly applicable

to the analytic

For the definition of an irrational ~lways associated rational numbers; a well-defined infinite

real number there is of

set of the first power,

this is commonto all forms of definition,

the difference


lies in the manner of generation by which the set is connected with the

to be defined, and in the conditions

which the set has to .fulfill to definition.

it as ba.sis for" the particular

In case of the first det~tion a set of positive" rational

era By is assumed, designated by (ay) and ful1'i~ing

the condition,


any given number of numbers, s~lected in any way, but finite,

is chosen which can be shown

summed, the sum. so formed always remains less than the limit
If two such aggregates three

"' i
I ~;',". Ii',

(av) and Catv) are given, it is rigorously

cases can ari~e; either every part ~ of unity is contained equally"

l~ )

orten in both aggregates, so tar as one ~
: (Anzahl) of t~rms as large as one pleases;

the elements in a finite number or, from a given n on

1. is n

contained . "

always more otten in the first aggregate than in the second, or thirdly,

.! , n

from a given p. on is contained [566] in the second aggregate more often than Corresponding to these cases, if band b' are the numbers to be defined by the aggregates (ay) and (a'v)' respectively, in the first case we
b I, in the second b



and in the third

case b


b! ,

If the

two aggregates are combined to torm a new one (ay" atv), this
for the definition of b + b'; the new aggregate

forms the basis

if from each of the aggregates (ay) and (atv) of

(a .at






is formed in mich ,the elements are the products

and all a'vt this new aggregate forms the basis for the definition of
It ts seen that the motivating principle of generation Which

the product bb! ,

associates the set with the numbers to be defined by it, lies in the formation of sums; but it must be emphasized as essential that only the summation of an always finite number (Anzahl) of rationaJ..elements is performed and that the number b to be defined is not set at the start as equal to the sum ~ the infinite series
By a.



if this were done, the procedure would contain of the sum

error since the definition

;£ av is not reached


until it is set equal to the number b already defined.

I believe that this

logical error first pointed out by Herr Weierstrass was in earlier times in

119. versal practice and was not noticed because it belongs to those rare cases mich actual erroz-s.of reasoning do no appreciable harm to calculations.

in my opinion all those difficulties which are found in the concept

,irrational depend on this error, Which if avoided, leads to a concept of
""'U ... V',la. .......

ty which is as determinately and as clearly lodged in our minds as

t of r~tional number. The definition of Herr Dedekind considers the totality of all rational numbers, divides these into two groups in such a way, that, if the numoer-e of the first group are designated by Ay, those of the second group by then always c~ls



By; such a partition of the rational numbers Herr Dede-

a cut* (Schnitt), desi~ates

it bY. (Av/

. I\r)

and assccd.at.es with

. t a number b.

If two such cuts are compared, one finds as for the first de-

tinition, three possibilities of which the two numbers b and bl corresponding two cuts are either equal or b


b' or b

< ~'.'

The first case occurs,

tor certain easily regulated modifications which must be made if the numbers to be defined are rationnl,if the two cuts are completely identical and here is seen the undeniable ~dvantage of this definition over the two [56,7] that every number b corresponds to a single cut, which circumhowever, has the great disadvantage that the numbers of analysis never appear in the form of "cuts" J into which form they must be brought with great art and ingenuit:v:. Then follow definitions of the sum b + b' and the product bb' from new cuts derived from the given ones. The disadvantage which is inherent in the ~ and ~ de-

finitions is that here the srume, i.e. equal, numbers can be represented in an *Translatorls Note: ·Compare·]mclid's definitions V and VII, Book V, of magnitudes with this !lcut"of Dedekdnd,

, ,



of ways and that a.ccordingly no general. unified view ot the entire
of real numbers is directly




This disadvantage can be easily .

by specializing single-valued

the defining set in that one employs one of the well-

systems as for instance the decimal system or a simple devel-

continued fractions.
I come nowto the third


~et of rational

numbers (~)



of real numberB. Here also

of the ~irst power is assumed, of which



demandanother property as in the Wiererstrass

definition; , a finite

I require,


any o.rbitrarily

small rational


number (Anza.hl) of

set can be removed, so that the remaining ones taken in pnirs have which in' absolute value is less than which also can be characterized


Every set (ay) of

by the requirement:

L i",

( 0. Y +)A -

a.V) =


(for any desired




:.I call a fundamental series . to be defined,

(Fundamentalreihe) and associate

with it a numberb as is done at (Compare

for which one can well use the symbol (ny) itself,

the suggestion of Herr Heine after Borchardt's

many oral discussions with hin4

Journal, Vol. 74, p. 172).

Such a fundamental series presents three definitions; either the terms of

Cases as can be rigorously deduced from their the series a


for sufficiently

large values of yare


less in absolute value than v on they are greater than

any ar~itra.rily

assigned number; or, from a certain number

a deterndnable positive rational

a determinable negative quantity -


p ; or,

from a certain v on less than case I say that b equals in the third,

In the first

zero, in the second, that b is greater that b is less than zero or negative.

than zero or is positive,



Then comethe elementary operations. two [568J fundamental s:ries that (~

If (ay)




determining the nwnbers band bt , it can be shown are also fundamental series, which determine tbree b t. bl aad the


atv) and (ay.aty)

new numbers serving as def:initions product bb I •

of the sum and difference

121. If b is different from zero, a definition of which has just it can be shown that


is also a fundamental series, lIhose
bl b

number furnishes the definition for the quotient - • The elementary operations between a number b given by a tunda-

(ay~ and a direct~
definitions, by setting al

given rational number a are included in the


equals a, bl equals a.


Then come definitions of equality, greater than and less than numbers b and bl (of which bl can also or b /' b I or b


:: a) in that one can say that

,:1 :'i


according as b - b f is equal to zero or great-

zero or less than zero. After these preliminaties it turns out as a first rigorously .. emonstrable theorem, that, if b is a definite number determined by a fundad .mental series (By)

then b - By with increasing values of v becomes less in

absolute value than every imaginable rational number, or what is the same thing, -,that: Lim 'f -=




Care must be taken on this cardinal point (Cardinalpunct), \\hose meaning might be easily overlooked: in the third definition the number b is not defined as

the limit of the terms ay of a fundamentru..series (ay); since this VIOuld be a logical error similar to the one discussed for the first definition and indeed it is an error for the reason that the existence of the lL~t


would be

presumed (pr~sumirt); the situation is rather the reverse, that through our previous definitions the concept b has been thought of with such properties and relationships to the rational numbers, that out of it the conclusion can I
" .. cP


be dravm vd th logical evidence that: L.im Q., exists and is equal to b.

hope to be pardoned the emphasis which I place on the following warning, that most persons pass over this seemingly smull detail and consequently land in perplexity and contradictions in relation to the irrational from which by fully observing this warning they might have spared themselves; they vrould then

cl8a~ly seen that the irrational the definitions rational attains numberthrough the properties given in just as definite

[$69) a reality in our tninds as the
process is needed

or even the integer and that not even a limiting intelligibility



but rather on the contrary through possession of it and evidence of limiting processes in

one i6 convinced of the feasibility

generalS} ; for the theorem just stated can nowbe easily extended as follows:

if (by) is a set of rational

or irrational

numbers with the property that a numberb







be), then there exists such that:

determined by atundamental

series (By),

,",'M b;
'i ;: 00

numbers b, which are defined by them fundamental series of the of ~, are ,also representable in

It also happens that the ~ fundamental series of this kind (av) (l.call ~ order) such that they appear as limits of ways as limits of series order

a variety

damental series. of the first ,

(cf;1 ),

(bv) where every by is defined by a funwith fixed v.

I accordingly call

such a set bv if it has the property, that chosen JA ), I a fundamental series of

!..i... (Oy......,_by) v=CO


(for any arbitrarily

the second order. Simila~ order may be constructed, fundamental series of the third, fourth •••••••••• nth

or even of the 0( - order, where


is any number

chosen from tQe second number-class. All these fundamental series achieve for the determination of a real number b exactly that which is achieved by fundamental series of the first order and the difference lies only in the more complicated, more extenNevertheless it seems to me to fix this

ded form of what is presented (des Gegebenseins).

highly proper, in so far as one accepts the third definition, tinction

in a Y~y similar to that adopted by me in another place (Math. Ann. I accordingly make use of the expression: the numerical

Vol. V, p. 123)..

quantity b is given by a fundamental series of the nth or even the {)(- order.

.tf ;~:'
It this

is accepted, one gain'

thereby an extraordinarily

easy and intelli-

gible phrase with mich to describe in the simplest.and JDlst succinct way the mass (FUne) of many-formed and orten complicated webs of analysis,


at the same time achieving a not to be belittled lucidity.

improvement in clarity and

I oppose the doubt expressed by Herr Dedeldnd in the introducNumbers"against these dismotive of introducing new
,', !

tion to his paper "Continuity and Irrational tinctions; I was not actuated by an ulterior

numbers [570] by these fundamental series

of the second, third

order etc. order




1 J


Whichcould not be represented by fundamental series of the first but rather

I had in mind only the conceptually different
has appeared clearly

forms of that al-

ready given; this

in several passages in my writing. to the remarkable fact

I wish at this point to call attention that through these fundamen~al series and second number-classes, series in the ordinary

Whoseorders are numbers of the first

all thinkable forms of analysis appearing as
already foWld or still to be found, \'bose order

sense, either

are completely exhausted, in the sense that fundamental series . numbers are numbers of the third rigorously prove at some tuture number-class do not exist,

as I shall

the usefulness

I shall attempt brieflY to explicate
(Zweckmlssigkeit) of the third definition.

As a symbol that a number b is given by a fundamental series (ey) of any order n or

,I employ the following: .



with the general

Suppose, for exampleI we have a convergent series term cv' then the necessary and sufficient

condition for ~onvergence is J

knownto he, that series

kim (<:1+,"'C""'l



C ..... !")s-O for any


Accordingly the

is swmmed,y the formula: b

L. en n.o


rv (~ Ctl).


for example, all Cv are defined on the basis of fundamental aeries of the order J a similar relation holds for

2 C..



and the sum ~

bi a fundamentaJ. series of the (k+l) order.

It-., C.,

is de-



(gedankliche Inhalt) of the theorem SJl1(~


For example, suppose the is to be deecrdbed ,

its powers can be thought of as being given by the formulas:

~ ~ l~~ J where, for brevity, we set
)"" L-

r: . /. )





rv ( Q¥




we have


J.n ..., :Q v .



order, and through that relationship the equality of the rational number 1 and the number ,sin' is given!, S~larly the meaning of more complicated formulas as for instance
expre Baed by a fundamental series of the second order,

f (f -)



""(i!-,r (Z:.,)!

( T) JoWl



is defined on the basis of a fundamental series of the second

those in the theor,y of theta-functions can be precisely and relatively simply described" - whereas the reduction of infinite series to those comprised solely of rational-terma with like signs and converging unconditionally, is bound up with the greatest deviousness, which can be avoided, however, by using the'~definition rather than the first,

long as it is not a ques-

tion of numerical approximation of infinite sums by rational numbers but one of finding an entirely sharp definition. The ~ definition certainly seems

not to be a useful one with which to define precisely the sums of infinite series which do not converge non-conditionally and in which the order of the terms, positive and negative, is definitely given. Even for non-conditionally

convergent series the determination of the sum even if it is independent of the order of the terms, is only actually possible for some definite order; for such cases, also, one is tempted to prefer the ~ definition to the first.

125 ..

Finally, the possibility is seen of generalizing (Verallgemeinerungsfahigkeit) "
.the third definition to apply to super-finite numbers, while such an extension of thetirst definition ie entirely imgossible; this difference lies simply in numbers the commutative law even for addition

the fact that for supe~finite

is.no longer of genera11Talidity; the first definition is inextricably bound

this law, it stands or falls With it.

For all species of numbers for
~ .; •I

Which the commutative law of addition holds the first definition, however, except for the cases here pointed out, is entirely adequate ..

[572] The concept of the "continuum" has not only everywhere played animpottant role in the development of the sciences but has always evoked This

the greatest varieties of opinion and has led to tremendous quarrels.

lies perhaps in the tact that the basic idea of this phenomenon has taken on

different content for the various combatants, perhaps because a precise cOmPl$\~ definition



this concept had not been bequeathed them, perhaps also,

and thts seems the most probable, the idea had not been thought out by the Greeks whornay have been the first to give attention to it, with the clarity and completeness which would have been necessary to preclude the possibUity. , . of the various opinions of their successors. We see, then, that Leucippus,

De~~citus and Aristotle regarded the continuum as a Compositum, consisting of "partdbue sine fine divisibilibus" but that Epicurus and Lucretius thought of it as constituted of finite things, atoms, out of which grew a great quarrel among the philosophers of whom some followed Aristotle, others Epicurus·; others J remaining aloof from the quarrel claimed With Thomas Aquinas 9) , that the continuum consisted neither of infinitely many nor of a finite number of parts) but of !:2 parts; this last opinion seems to contain less an explanation of the situation than a tacit recognition of the fact that one had not got tQ the

bottom of the matter and that one had better withdraw with dignity from the

Here we see the origin of the medieval-scholastic
,.1'1&1 ..... ""

point of view, of

defenders are found even today, in which the continuum is thought to

an irreducible concept or, as others express it, a pure a priori intuition which is scarc~ly determinable by concepts; evcr.i arithmetical attempt at determination of this ~stery is regarded as a forbidden venture and rejected

with appropriate vigor; timid souls gain the impression that for the IIcontinuumll it is not a question of a mathematically logical concept, hut rather a question of rel~&ious dosma. It· is far from ~ purpose to conjure up anew these controversial

questdons , and, in any case, the narrow frame of my present WJrk offers too

little spa~efor

a detailed discussion of them; I feel merely obliged to devethe continuum with the calm logic (logisch-nuchtern) which 1

... i
I. Ii II,


lop the' id~aof

n~ed and make Use of in the theory of manifolds, briefly and with consideration for the ma~pematical theory of sets.
an easy

This effort has been fundamentally not

orte~; '573] since, among those mB:~hema.ticiansto whose authority I C

should gladly appeal there is not a single one who has studied the continuum in the precise Way Which I here find necessary.


assuming as

basic one or several real or complex continuous

quantities (or~ t~ employ what I believe is a better expression, sets of continuous quantitie~ -- continuirlicher GrOssenmengen --) there has been con-

structed in what seems the best way the concept of a single-valued or manyvalued continuum dependent upon them, that is, the concept of function extended in a variety of directions, and it is in this ~er that the theory

of the so-called analry~c functions arose as well as that of even more general functions with highly remarkable properties (non-differentiability and the

like); but the independent continuum has been assumed by mathematical writers only in its simplest fonn and no more thorough investigations have been attempted.

~ext J I have to declare that in my opinion to enlist of time {Zeitbegriff)

the conoept

or the perception of time (Zeitanschauung) in any er-

planation of the muchmore basic and more general concept of the continuum is not the proper procedurej time is in lI\Y opinion an idea (Vorstellung) clear explication "~4uires the concept of continuity whose

upon which it depends and objectively as a

even with whose aid it (time) cannot be determined either stlbstance or subjectively as the form of an a priori


it (time) is (H&fs-und

merely a concept employable as an aid and t.o express relat.ionships Bezieh\Ulgbegriff),
by vmich the relation

between various motions occuring in Such a thing as an objective or

natu~e and perceived by us are expressible. absolute "t~e' riever occurs in nature therefore

time can not be looked upon as

6£-Inot.ionbut rather

this latter

as a measure of time, were it not appeari,ng as it does in intuition pre-

for the" fact that in the latter the modest"role as a subjective,

(motion) time itself

necessary form of an a priori

vents the complete and unassailable

success of this procedure, although there

bas been plenty "of occasions since Kant when this has been attempted. In the same way it is my conviction that one can not begin
"dth the so-called

perception form (Anschauungsform) of ~

to gain knowto it can

ledge of the continuum, since space and the structure only attain


with the help of an already conceptually existent

continuum that observa-

meaning necessary 'for it to become an object not merely of aestbetic tion or philosphic scrutiny

or inexact comparisons, but of careful precise

mathematical investigation. There is therefore nothing left for me to do, except, with

help of the concept of real number as developed in § 9, to try to find a concept of point-continuum, purely arithmetical and as general as possible. I space

now make fundamental use [574] of the l'loo<iimens1.Qnal. plane arithmetical Gnt that is, the" assemblage (:in thought) of all

( XI

I Xtl.. .. I X n),

systems of values:

in 'Whioh overy x is kept independent of the other all real values from - 00

to + ~

Every special system of v~lues of that kind I callan


" .,'

point of Gn• pression:

The distance between two such points may be defined by the ex-

+ Y(x:"


t()(~ - X1)l + - • - _ + (x'"




" I,~I

-t .


and by an arithmetical point-set P contained in Gn is meant every orderly, .

! ;


given assemblage of ,points of the space Gn-

The problem then becomes to dis-

:1'; ,

i, l


a sh~p

and at the same time most general definition possible of the eon-

l:' :

dit10n~ under which P is to be called a continuum.. ,I proved in Borchardt's Journal, Vol. ?4, p. 242, that all
,~paCEl$ GIl' however great .the so-called dimension number n rNlybe

have the

sam~ power:

and Ore therefore of equal power with the linear continuum as well The problem

as the t\iu!o~l8.geof aU real numbers of the interval (0••.• _.:1). . .' . . -

qf .,,'




determining the power of, On accordingly reduces to the

~.~. ..

Q,W~ . . .' l}l' regard stion


the interval (0••••••• ,1) .and I hope .to be able ' ..... ~

soon by. a r~goro~~ proof* to show that the power sought is no other than that " ~ .of our second ~finite number-class (II). Then will follow that all of the

infinite point~sets hav~ either the. power of the fir~t nUlIiber-class(I) or that of the second number-class (II)~ The further conclusion can be drawn

that the totality of 0.+1 functions of one or several, variables which are representable

a prescribed infinite series, in however many different ways, (II) and is therefore

has also only the power of the second number-class

countable (abzahlbar) by numbers of the third nuIDber-cla~s (III)


• This



theorem can also be extended, for example, to the assemblage of all lIanalyticll

*Translator's Note: C;;:,ntorever' succeeded' in doing this and the problem still n romains unsolved.


i.~~f o

functions arising

from the analytic


of conve~

gent power-series of one or several variables of one or several real variables

or to the set of all functions
by trigonometric

which are representable


In order now to examine more closely the general concept of a contiI?-uum lying inside G I recall n given point-set the concept of the deriva.tive P (ll developed in D\V Wl'itings: of any

P J as it



Math. Ann. Vol.

V, then extended

in Vols.. XV, XVII, XX ana XXI

to the concept of a derrJ'at&,':e



t.· is

any integer of the number-classes

(1) (II) (III)



P may now be divided into two classes according to the

power of their



If P (., has the power of (I). it turns a fir5~ in-

out as I have alr~ady stated in § 3 of this paper_that there exists teger 0( of t~e first But it p(tJ
be separated

or second number-class (II) J for which pfXJ vanishes ..

tne power of the second nutobe:r-class (II) then pC"~ can always

into tWo sets R and 5, and only p(J):

R +S

in one way, so that:


Where Rand Shave extraordinarly

different propertiest

R is of such a nature that b.Y repeatedly taking successive derivatives



it may be eventually annihilated,

so that there always exists

a first in-

of the number-claas

(I) or (II) for which

R(f.) == 0 ;
such ppint-sets R I call reducible. S is of such a nature that continued differentiation produces


no change, in that:
, I ;.:,

and accordingly

We can now say: If p~J is of the power
the~ p('J can be separated into a definite




such sets I call Eerfect (perfecta). . is.,


the .second number-class (II)

reduoible and a definite

perfect point-set.

Although the two predicates, taneously applicable


and "perfect" nevertheless

are DOt irreduci-

to one and the same point-set,

is not the same as perfect nor is imperfect precisely ,as one with a little care may see. point-sets

the same as reducible

The mrfect

S are in no sense in their

inner structure

always that which I have called in Il\V previous works, mentioned above, lIeverywhere-dense" )1) (~berall dicht); they consequently are in themselves not adeof a point-continuum, even if one J'I!lstgrant at .) must always be perfect sets.

quate for a complete definition once that the latter


Another concept is muchneeded" in order that,
above concept,

combinedwith the

the continuum may be defined" naJD3ly the concept of a connected


if, for every two of its points,

Wecall T a connected point-set t and tl, for any arbitrarily

small. £

,[576) there ,always exists


finite of

number (Anzahl) of points tl,


t2' •••••••••

ways" such that the distances than

. t2 t3' ••••••••••••" tytl are all less


of T, appearing in a variety

All the geometric podnb-contdnua, knownto us fall, as is easily I believe that in these ~

seen, under this concept of connected point-sets; predicates

"perfect!! and IIconnected" I have found the necessary and sufficient (Merkmale) of a point-continuum and define accordingly the nPerfect" and "con-


point-continnum inside Gn as a FE:rfect-connected set. 12) nect.ed" are not merely words but are predicates most sharply by the preceding definitions. The Balzano definition tainly incorrectj

of the continuum, characterized

of the continuum (Paradoxes, § 38) is cer-





it expresses in a one-sided fashion just one property of for sets which can. be formed from Gn by point-set (vid. Math. Ann. Vol. XXI,

the continuum which is fulfilled thinking of an "isolated!!



p. 51) as removed; also for sets Whichconsist of several separated continua,' obviously such cases are not cont1nPre;. although for Bolsano they w>u1dbe. We see here a violation of the theorem: 'lad ossentiam alieujus rei pertinet tollitur;

id" quo dato res necessario ponitur et quo sublato res necessario vel id,

aU. quo res et vice versa quod sine re nee esse nec concipi potest,!! Equally so, in the writings of Herr Dedekind (Continuity and


Numbers)a single other property of the continuum seems'to have

been emphasized in a one-sided way~that property, namely, which it has in com.on with ~ all nperfectllsets. ~ .

§ ll ..
Wenow proceed to showhow one is lead to the definitions of,the new numbers and howthe natural succession of integers, this explication partitions of the absolutely unending For

which I call the number-classes, come about.

I shall make use only of previously mentioned theorems on relation to the first. The series (I) has

the second-number class and its its basis

of generation in the repented setting-down (Setzung) of fundamental and their finite urrlonj the number v is aooord.ingly number (Anzahl) of such successunii.;
'I i, I,
I ~


regarded as always alike,

the expression not only tor a definite ive settings-down ties into a whole.

(setzungen), but also tor the union of the established The formation of the finite real integers


rests then upon


the principle

of adjoining a unity to an already present number; I call this which, as we shall soon see, plays,an essential

method (MOment) generation, of

role in the building up of the higher integers, generation.

[577] the first



The number (Anzahl) of the numbers v of class emongth6m. Howevercontradictory

(I) is infinite it might be to nothing offen-

and there is no greatest speak of n greatest

number of cless

(I)~ there is nevertheless

in thinking
Dl·...i:lg."uu '

of a. new number Which we shall ca11* 6J and Which will be the natural,

for the idea that the "entire assemblage (I) is given in its



(Just as v is an expression for the idea that a certain is united to form a Whole). It is accordingly

te number (Anzahl) of unities
ICIrIJlI.Li:lD.I..,1 .......

to think of the newly created number CrJ as the limit to mich num(iJ

v approach, if by it nothing else is understood than that . .
ever.y one of the numbers v.

is the ~

which succeeds all numbers v, that is, it is to be regarded as greater

Then if further settings-down (Setzungen) of

be pennitted to follow this established of generation we have VJ T " W +~ J


with aid of the



- - - , (' ..)-1-

since by this proce-


no grea.test number is reached we then think of a new number which we can 2 tiJ and which shall be the first to succeed all of the al ready imagined principle of genera.tlonto

nUlIIIDE~rs v and UJ + V ; if again we apply the ~

number 26J

, we obtain the sequence:

2.w +1, J.(,J +.11 • - - - J. CiJ

+ ~ ... " .

'!'he logioal

function (logische Funktion) which has produced ..he t different from the first principle of genera-

two numbers W· and 2GJ is clearly tion, and I shall call

it the second principle as follows -

of gener~tion of real integers succession of definite

.'and define it more precisely real integers

if 3lV definite exists,

is given for which no greatest second principle


new number is crea.ted on

the basis of this the limit

of generation. is.

which is to be thought of as

(Grenze) of those numbers, that

is defined as the next greater

number to all of them.
By continued application

of both principles

of generation,


obtains successively

the following further

extensions of our number system:

*Cantor's Note: The symbol CP WhichI used in Number2 of this series of articles (Vol. XVII> p. 357) is hereby replaced from nowon by GV • since the sign ~ has been frequently used as a symbol of indefinite infinities.

3w~ .3w


+l~ - - - _.._,

36> ~YI ~


}AW, ,p..GJ "'1, - ...- ...'.. . ,

-~~--- .. -~ .... -- - ---






v. - ~.



No end is. reached in this process, ,si~ce of the


no number is the greatest. The second ,princiPle of generation per,mits us then to intro,d_,~suco~ssor
to all the numbers

f-W ...Y




which can be called



. I

Which in ~urn new numbers definitelY


Aw~ T}'-W'"


and one clea~ly reaches~ by applying the two principles of generation, numbers ~,

of the follcpd,ng torm:
... , \ 1



.1 r. , '" 1', "'"'.

_ _ __

+ Y u -, r: I



V ....JA" .

but ~

are forced now b,y the second principle of generation to set up a new. B:ll these and itlich we can appro-

number which will. be the next greater after priately denote as:



This construction of new numbers as one sees, has no end; by
applying the

two princtples ot generation one continually obtains new numbers
At this

and sequences of numbers, Which have a completely determined succession,
point the likelihood is suggested that by this manner

of constructing is endless

new definite infinite

integers we mst lose ourselves in what

(ins Grenzenlose) and that. we,,~


position to g:i,.ve this


less process a definite mpm~n~arltopping point in order to gain a limitas
tion similar to the one that

in a certain

sense actually


for the earlier

numbers of class (I); there use was made only of the first principle of generation and any emergence from class (I) accordingly impossible. The second number region


of generation must lead us not only beyond the earlier

but it is seen also as an instrument which in union with the ~


of generation gives us the possibility

of bre~

through evety barrier in the

formation of the concept of real integers. If we observe that the earlier follow, fulfill nwnbers and those '¥bich imnediately if it is imposed as a restricthird

a certain


this condition,

tion upon all the numbers still principle, ates,

to be formed, appears as a new additional or lim:i ting .principle

Yihich I shall call a restrictips

and '\'bich oper-





as I shall show so that the second numbe~lass

(II) defined ~ its applica!: ,

tion will not only attain a higher power than that of class (I), but indeed the next higher or second power. The suggested condition, so far defined, fulfills, as is easily which everyone of the infinite seen, is numbers



that [579] the set of numbers


which are predecessors to this number 0\. first number-claSs (I).

in the sequence is of the power of the



If we take, for example, the number 6J

, then its pre-



decessors are given b.Y the formula:
\(, (ij'to

V, W,...·f

T •• _ •


VJ ...

Vr )


fA I Yo) Vtr-·-;Vr

are all finite, positive

integers including zero but

excludi.ng the combination Vo: "I. As is !mown, this infinite

= - ..- - - . - - - =

V]A ::: O.

set can be expressed in the form of a simple,

series and has accordingly the power of (I).

Since, further,

every sequence of sets,

each of which has the




power, even if the succession itself

is of the first power, becomes itself

a set of the power of (I),

it is clear that by the extension of our sequence of reached as have the same power of (I),

numbers, only such numbers are actually

so that the suggested condition is in fact fulfilled. Wetherefore define the second number-class (II) as the assemblago of generation, and

of all numbers fonned by the aid of the two principles UdVo.nc~ / in a ~inite ordered succession: G),W+I,----}v'oW 14








tAl t----!




j !' ;.

upon. Which the condition lon,


is imposeda that all the predecessors




set of the power of the number-class {I).

WenoWproceed to demonstrate the theorem that the new numberclass

(II) has a power, which differs


from that of the first

number-class' (I).

This theorem follows from the following theorem:

"If 0<.. )0(.2.'_ •• ,OJ.'I,

• is any set of various

numbers of the

second number-class and having the power of (I)

(so that we are justified

regarding the nwnbers as a:ranged in. a simple sequence(o<v) ), then either one of the .set is greatest, a definit~ set number say ~ , or, if there is no greatest, there exists


of the second number-class (II) not a memberof the is greater than all but that every integer

oc, ,

such that'~

(3' <. ~



VP"'l?"""'- . "..........,(.

will be surpassed in size (Grriss~) by certain ·.,bers ~ and p,

numbers of the sequence; the num-,.03 be called the upper limit


of the set

(0<") ."
The proof of this theorem is simply the following: let O(K~ be than number appearing in the series


, respectively,


can appropriately



the first o<.K3


, etc.




the first

which is greater




than o{K,t

<. ex.( 2. <



< Kli <. <<
0( ~"

[580] We then have





0( #<.). }

as soon as

V< KA'
Nowit can happen, that from a certain terms in the series number D<Kf

on, all the

{O<v} Which follow

are smaller than this number; then this

number is obviously the largest

of all the terms and we have:

r" 0(,<) •


136. r, if we think at the set of all integers
and adjoin


from lon,

which are smaller

to this set the set of

all -

integers, which ~


then the set of all numbers. which ~

ex K l
two numberpower and

obtain a definite classes and further

part of the successive numbers of our first this set of numbers is obviously of the ~ (from the definition

. there exists


of (II» a determinate number ~ to all these numbers.

, of the assemblage (II), Which is the next greatest




and also:


, since K" can always be chosen ~ large,

that it will be greater

than a preassigned v ~d


One also easily

sees tha.t every number


< O(K}Io

will be surpassed of the theorem

in size by certain are now proven.

numbers C(Kv

; with Which remark all parts

The theorem now follows, second number-class (II)

that the totality

of all numbers of the for otherwise we

does not have the power of (I);

should be able to think of the entire series:

assemblage (II) in the form of a simple

oc, ) o<..1} exists

- - • _) oC. Y




which according to the theorem just proven either there

has a greatest




a certain



from (II) by Which all the terms <:Xy case the number

be surpassed in size J in the first ber-class number ~ (II)



belonging to the num-

can not appear among the terms 0( V , in the second case the

belonging to the number-class both cases contradict therefore

(II) does not appear in the series

(C(~) ;

the assumption that the sets (II) and



the number-class

(II) has a power different

from the power

of the number-class

(I) and (II), the the immediate successor

That of the two powers of the number-classes second is actually exists

of the first,

that is, that no power

between these two powers, definitely follows from a theorem which I shall

give at once and demonstrate'.


j ~; :'

, ~~ ,iii'

1.37. [581]
If we now look back

:~ i, 'I:'

and review the means Which have led not




only to an extension of the concept of real integers but also to the concept


of a new power of well-defined

sets" differing from the first power" we see that emerged as distinct principles.

i '~'



i,,\" 1

there were three logical moments (Uomente) ~i~h

They were the two previously defined principles of generation and besides these a restricting or limiting principle ~ich consisted in the requirement that a

new number could be created with help of one of the principles of generation only when the totality of all the preceding numbers had the power of a defined number-class \'bieh already existed and was complete in its entire range.

this way, taking due regard to these three principles, one arrived with .-ery great ease and evidence (Evidenz) at ever new number-classes and one attained all the various, successive" ascending powers appearing in the worl.d of body
Or mind.. The new numbers acquired in this way were throughout of the same

concrete determinacy and objective reality as the earlier ones; I saw no reason

we should hesitate in the construction of such new numbersl as soon aait

became evident that for the advancement of the "sciences the introduction of new number-classes indispensable from the infinity of possible ones, bec~ desirable or even

for our reflections on nature.

I now come to the promised proof, that the powers of (I)
and (II)

follow each other directly. so that no other powers lie between. If out of the assemblage (II), one selects according to any

law a set (0(
( 0(


) of different numbers, 0<



is" if one considers any eubse


contained in (II), then such a subset displays certain peculiarities '
may be expressed in the following theorems:
UAmong the



numbers of the set (,,() there is always a ~.II if one has a succession of numbers of the which continually diminish in size so that

!lIn particular



ex, ) 0\ 2 J --


,CX{3 •• "",

0(" >'\",91

if (3' >(3 ~
one. 11

then this series necessarily breaks off with a finite the numbers; the series in-

.dex*' (Gliederzahl)

and ends with the smallest



be an infinite

It is remarkable that this when the numbers case of infinite which easily
0( ~

theorem ~ich

follows immediately

are finite


can also be demonstrated for the




Actually according to the preceding theorem of the number-class (II) [582] there is"

follows from the definition "if

amongthe numbers 0< v finite"

one considers only those for which the index v is is


a least;

if this,

for instance"



, it is seen that,

since must consist

, the series

of eX2.ctly



and accordingly the entire finite.

terms and is therefore

Wenow have the following fundamental theorem: "If three cases arise: finite


I )

is any number-set contained ('


Oe J

iri the

assemblage (II)"

~s a finite

assemblage, that is,


of a class,

number (Anzahl) of numbers, or it


has the power of the first

or t.hirdly has the power of (II);

quartum non datur.tt let

The proof proceeds simply as follows: ber of the ~


be the first


number-class (III);

then all

of the numbers <X

of the set


smaller than


since they are contained in (II).

If now we think of the numbers 0(1 ordered according to size; with r:;I..", as the smallest,O(W+1 the n~ set larger where ~ etc.,

we obtain the set



the form of a. "well-orderedtl naturally

O<f3 '

runs through numbers of our

extended number series

from Won;



always remains sm;l.ller The number ~
0(. )0(2,--

or aqual to o(~ and sinceCXj:!o *Trcnslator's Note:


then also

(.3 <.0 .

The index is the subscript

of the 0<.

in the series

cannot th.erefore adYance beyond the number-class (II), region; three easea can then arise: number of the series either but remains m.thin this


remains less than a determinable is





in Which caee(o..')



set; or ~


all values of the series GJ T V
the series thirdly, through!!! the set

but remains less than a determinable number of

(II), in Which case
{'" assumes arbitrarily


is obviously ~ set of t~e first

power; or,

great values (II), in which case case the assemblage q.e.d.


runs is,

numbers of (II); in this last

(o<~).~ that

(0(') ,

has obviously the power of (II).

As :imIntediate consequence of the theorem just proven, we have the folloWing: lilt there is given any well-defined class set M of the power of the numberthen the assemblage series, or it

(II) and if any infinite

subset Y' of M is considered,

M' can be thought of either in the form of a simple infinite
and M'


that a one-to-one,


correspondence can be set up between


I i I .

"If there is given any well-defined further

set M of the second power, and

a subset},{1 of M and a subset


of Mr, and it it is knownthat the set correspondence, then M' is relaI.

M'I is related to 14 in a one-to-one, reciprocal
ted in the same way to both M and 14'I ."

1 announce this last theorem at this point because of its [583]
connection with the preceding, under the assumption, that M has the power of (II); obviously it is also true if Y has the power of (I); it seems to be highly ~:

remar-kabl.e and I expressly emphaaize the fact that this theorem has general:: validity, whatever power the set 11mnyhave. in a later I shall comeback to this point interest Which

in more detail

paper and there point out the peculiar

is associated

with this general theorem.



i ~;ii




In conclusion I shall make some further

remarks on the numbers

second number-class and some possible operations with them limi.ing self for the present, however, to ~diate


and obvious considerations,

since I

reserve more thorough investigations The operations in gland

for later

of addition and multiplication integers

have been defined law holds also

I have shown that for the infinite

the associative

but that in general the commutative law does for the numbers of the second number-class. holds in general only in the following fonn:



holds in particular law.. it

As to the distributive

(0( i" f3 )'
{Where0( + (3, 0(,

:0( ~ of'



appear as DIll tipliers)


as is seen directly

and intuitively.

If 0( and ~ are

any two integers .. 0(

can be considered in tm way.

< ~ .. one easily


sees that the equation:


permits one and only one solution of

(II) ..


S ..where, if 0(

and :(3

are numbers of

is a numberof (I) or (II).

This nUIDber § is set equal to



But if on the other hand we consider the following equation:

it turns out that this can often not be solved at all for is seen in the equation:



for example,






Even in those case s for which the equation: solvable of l: this for


S + 0<. = (3

is of value~ For

,it often happens that it is satisfied

by an infinity


; there is alwaysl however, a least number of all these solutions.
root of the equation


S-rcX (31





[584] in case the latter
therefore if

is at all

solvable,. we choose the symbol_fo< , Vlhich

is in general different

fromp-o< ,

vtlich latter

number always exists

Iill il

I' !p


\3 •
~fJ further, the following equation holds for three integers




! is
0<. .

the multiplier),


one easily

sees, that the equation:

has no other solution with respect to
by ~:



5 :~ and this

is indicated

But, on the other hand, it turns out that the equation

.(3 ~ 0( 5

(where ~

is the multiplicand),

if it is solvable at all with respect to ~
of solutions


often has several and even an infinity least; ·this smallest

solution of the equation: ~:


S ,.

one of v.hich is always the in case the latter has

a solution

at all, we shall.designate


The numbers 0( of the second number-class are of two kinds: 1) ,such

ex J

which have an ·immediate predecessor .in the sequence vtlich is




such I call numbers of the first predeoessars, for which therefore


kind. ~

2) SUch <X does not exist,

, which have no inmediate

the second kind. kind, but

The nwnbers GJI J.W1 W -+- W, Ware,


these numbers I call of

for example, of the second kind.

+ '.1

+ W +.2,. fA)



of the first

Prime numbers of the second number-class which I have defined in g 1, correspondingly first kind. Prime numbers of the second kind are in the order of their appearance amongthe numbers of the second number-class (II), the following: fall into two classes~ primes of the second and of the













so that among all numbers of the form.:

t = VO 6JJA ... VI W ,..-1



only.2!!! prime number GJ of the second kind exists; one must not conclude, . . however from this relatively sparse distribution of primes of the second kind, that the assemblage of all of them has a lower power than the number-class (II) itself; it'turns out that this assemblage has the same power as (II). kind are:


The prime numbers of the first


2. + I, (,.) + r t






These are the only primes of the first
just indica.ted by

kind which appear among the numbers
of all primes of the first kind in (II)



the totality

has also the power (II). The primes of the second kind have a property Which gives them

an entirely



if ~ is such a prime (of the second kind), than ., ; from Yttich

then always ~ot. ~ it follows, that,

tl '

where 0( is any number ~er


and ~

are two numbers, both smaller than ~ , then •

the product O(~ must always be smaller than ~

If we restrict

ourselves next to those numbers of the second

which have the form hold.


then the following rules for addition

and multiplication

., = \'0 (,.) .. V, W




... - • - - - -


Where we assume that





Vo and







+- - - - - - _ ~)t. + from zero.

are different

1) If

.fA <


2) If'ft

"> A


we have:



+- ~



Vo w}'\ ____ + Y

r -~ -,



V~ -).. +~"



-r ~, W


+ 2.. W



+ - - . - 1"

f" .

3) For)A::


iT cVoW
In case


A. ,

we have:


T -:(Yo-t-f4»GJ


,\ .. ,







1) It' V". differs from zero we have: tA+)' 14+>--1 )'''' A· +'1,W .... -- +Vt'-,W -tYrrow "'fafIJ

1- ---


then the last. term on the right hand side is

oi-f'" Vr,)o •

'f T ': Va W
The factoring


2) It'

V,.. = 0 ,

we have:

+- v,(iJ"" .... - -

of a number

1 'r =CoW J4 ...C,W "'", T



'Y'tt -l W
'" 1






into its prime factors

is as follows :



+ -- -- ... CeriA)

!4 f;

and Co, C,) ----C(j
from zero, we have:
-- -

[586] are positive finite



I)C, (w 11-.-/4 '+ I)e J

numbers different

C~_, (6).4I..G.'·~:I) C6
prime factors

WJA.6 ;
according to then we have a

if the Co C,) - - - Cfi_, are factored J the rules for such factoring of

into their

completed factoring


of numbers of the first class, into prime factors.

This factoring of numbers

is uniquely deter.mined even ~th abstracts from the commutability
C and if it is agreed, that

respect to the order of the factors,

if one

of the prime factors within the individual
the last

factor shall be a power of



equal to unity and that


may only appear in the last

As for a


of this factoring of any number

of the second number-class

into prime factors, I shall consider i~ at a later


144. [587]
.To § 1 1) Theo;:,,?f ma,a\fo1ds (Jlannichfaltigkeitslehre).
By this


phrase I mean a very comprehensive concept which up to now I ha~e attempted to investigate Qf sets. only in the special form of an arithmetic or geometric theory {jedes

By a manifold or set I mean in general



Viele) 'which can be thought·of

as a unity (als Eines), that is,

every asse~

blage of determinate elements which can be united into a whole by somelaw and I believe that I have accordingly defined something Which is related the flatonic logue"Philebus the to
'1 )

e' &os
~ that

or ..OED( ., a.s well as With that Vlhich Plato in his dia-




on the highest good" calls is, unlimited,

fA' 1(10 ~'.

He opposes this to



as well as to the

tr2fOl S ~ that


indeterminate w'nich I ,call the idea1is, the limit, and explains it as

an ordered "mixture" of. these two.

That these concepts are of Pythagorean

Plato himself polnts out; compare A. Boeckh, Teachings of Philolaos 1819,. To 2) Aristotle,

the Pythagorean, Berlin,

of Zeller in his great Second Part Second is

Comparethe presentation


"The Philosophy of the Greeks, Third Edition, Plato's

Abtheilung, pp. 393-403. entirely different

point of view in regard to the infinite CompareZeller Second Pert,

from that of Aristotle;


Abthei1ung., pp. 62S-646.

Also I find certain points of contact with my ideas Compare: P. Zimmennann,Cardinal d. Wiener

in the philosophy of Nicholas of Cuea,

Nicholas of Cusa as the predecessor Akadamie d, Wiss. Jahrg, 1852).

of Leibnitz (Sitzungsberichte

I say the same.In relation

to Giordano Bruno

as the successor of Nicholas of Cusa. Philosophy (Weltanschauung) and destiny
An essential

CompareBrumhofer, Giordano Bruno1s (Verhangniss). however lies Leipzig 1882.


in the fact that I fix

145. conceptually once and for all the various steps (Abstuiungen) of the ideal-


infinite by the number-classes to be ~

(I), (II), (III) and etc. and then consider it
these super-finite numbers

problem to investigate the relationso!

not only nathematically

but to study them and point out their appearance in There is no doubt in Jl6T mind but that in this

possible forms in nature.

~way we mount ever higher, never arriving at any impassible barrier but at the

l same time never

reaching any approximate comprehension of the Absolute.


Absolute can only be recognized (anerkannt), never known (erkannt), not even For, as-inside the first numbe~class (1) for any finite number,

however large, there exists beyond it


set of the same power, just so there

follows every super-finite number, however large, another assemblage of numbers of the higher number-classes (II) or (III) etc. which has lost nothing in power

in respect to that totality of numbers beginning w.t th unity and continuing in an absolutely

infinite succession.

The situation is similar to that Which

Albrecht yon Haller said of eternity, [SSa) "I subtract it (the enormous number) and thou (eternity) liest yet before mell• numbers The absolutely infinite sequenoe of

seems in a certain sense to be·an appropriate symbol of the Absolute; (I), which up to now has alone served

the infinity of the first number-class

that purpose (symbolizing the Absolute), since I held it as a conceivable idea (~ not Vorstellung), now becomes an utterly vanishing nothing (Nichts) in It seems remarkable to ~ that each of the number-

comparison with the others.

classes, therefore each of the powers, corresponds to a completely determined number of the absolutely infinite assemblage also to every superfinite number the

of numbers and in such away,



there is a power which is to


; the



powers therefore/an


called This

absolutely infinite series.

is the more remarkable since the number (in case the number that number-class

which gives the order of a power


has an immediate predecessor) stands to the numbers of

which has this power, in a relationship of size whose minutethis the more the larger ( is taken.

ness mocks all attempt at description·and

To § 5 .3) cieterminari possunt. I can not ascribe
in whatever


any being (Sein) to

the incieter.minate, variable, ideal-infinite since it is presented

for.m it may appear, concepts or pure subbut

either in the for.m of relational or intuitions idea.

jective ideas (Vorstellungen)
in no case as an




If merely the ideal-infinite,

in the theorem


actu non datur" were meant, I could subscribe to it, although it The meaning of this theorem in the sources disof a con-

then becomes a pure tautology.

c~ssed seems rather to be that there is expressed the impossibility ceptual determination theorem as false.

of a definite infinite and in this sense I regard the

§7 and realist standpoint ~th respect

4) Realists.

The positivist

to the infinite is expounded for example in:

nUhring., Natural dialectic, Berlin, 124-130. . d. Prdne , der uber

1965, pp, 109-135 and in von Kirchmanil, Catechism of philosophy Also compare Ueberwegf s Remarks on Berkeley, Abhondlungen menschlichen Erkenntniss in von Kirchmann's philosophical


I can only

repeat that in their evaluation

of the ideal-infinite

I am in essential agree-

ment with all these authors; the point of difference lies only in the fact that for them this syncategorematic infinite is regarded as the only infinite attaina-

ble by "tl\ckingsll (Wendungen) or concepts and here especially only through re,. lational concepts. The proofs of Duhring against the actual-infinite could be demonstrated with fewer words and seems to run in either the direction of saying

that a definite finite number, however large one may think of it as being, can never be an infinite ono, vmich follows directly from its concept, or in the direction of sa.ying thnt the variable ..unlimited, large finite number can hot be

thought of as associated with the predicate determinate and therefore not with the predicate ity. of being, which again follows directly from the nature of varlabil-

That in this not the lea.st argument is ndvanced against the conceptual

possibility of definite superfinite numbers is not to be doubted, and yet these proofs are held to be proofs against the reality of superfinit~ numbers. This

way of arguing is similar to the one in which it is said that since there are countless many intensities of green, one concludes (589) that there can be no red. It is certainly remarkable that nUhring on pagel26 of his Work himself

admits that for the explanation of "the possibility of unlimited synthesis" a basis must exist which he calls I1fully unknowable in a conceptual wayll (,rbegreiflicher Weise v;Uig tradiction. Equally so we find those thinkers who are close to idealism or who even pay full homage to it, denying any justification for definite-infinite numbers. Chr. S'igwart in his outstanding work: Die Methodenlehre Logic


In this, it seems to me lies a con-

Second. Volume


• • (Tubingen 1872) argues as does Duhring and says on page 47:
Similar ideas are found in Kant and J", F. Fries; compare the

"an infinite number is a contradictio in adjecto."

latter's System der Metapbysik philosophers of the Hegelian

(Heidelberg 1824) in S 51 and II 52.

Also the

school vdll not allow the actual-infinite; .1 refer

to the valuable work of K. Fischer~ System der logik und Metaphysik oder Wissenschaftslehre, Second Edition (Heidelberg 1865) page 275. To § 8
5) What I here call intrasubjective

or immanent reality of con-

cepts or ideas may be designated as lIadequatell using the word in the sense of Spinoza when he says (Ethics, Part II, Definition IV): "Per ideam adaequatam

intelligo ideam, quae, quatenus in se sine relatione ad objectum consideratur, omnes verae ideae proprietates sive denominationes intrinsecas habetl1.
6) This conviction is in essential agreement not only with the

Platonic system, but also with a fundamental trend of Spinoza1s system; in the first relationship I recommend Zeller, Philosophy of the Greeks, Third Edition,

Second Part, First Abtheilung pp. 541-602. the section: He says at the very beg1nningof

"Only conceptual knowledge (Begrifrliche Wissen) assures, accordIn

ing to Plato, a true cognition (Erkenntiss).

so far as truth comes to our

ideas (Vorstellungen) -- this assumption Plato shares with others (Parmenides)-just so much does renlity come to their objects, and c·onversely. What can be known, ii, what can't be known, is not" and in the same measure as anything is, it is lmowa.ble". As for Spinoza I need only remind you of this passage in the Ethics, Part II, prop. VII: rerum". Also in the Philosophy of Leibnitz the same epistemological principle may be seen. "ordo et connexio idearum idem est ac ordo et connexio

Only since the newer empiricism, sensualism, and scepticism, as

well as the Knntian criticism which accompanied it, has one begun to believe that the source of knowledge and certainty (Gewissheit) is to be found in mind (Sinne) or in the so-called pure fo~ of intuition in the world of ideas (Vorstellungs-

welt) and must be limited to these; in D\Y opinion these elements completely fail to afford any such certo.in knowledge since the latter can only be attained by concepts and ideas, and, although stimulated to a high degree by external experienoe in the main is formed through inner induotion and deduction into a some-

thing, which in some way was already there and was nerely aroused and brought .into consciousness. ~ § 87, and 98 The procedure for the correct construction of concepts is in my opinion [590J everywhere the same; one sets up a thing (Ding) having no properties, which at first is nothing more than a name or sign A and gives this according to some law, different and even an infinite number of predicates, Whose

meaning is known through ideas already existent which may not contradict one another; in this way the relations of A to concepts already existent is determined; if the process is complete, then all the conditions for awakening the con-

cept A (zur Weckung des Begriffes and it emerges into being (Dasein)
is all that

A) Which was slumbering within us; are present

equipped with intrasubjective its transient





can be demanded or concepts; to esta.blish of metaphysics. To § 10



then the business

9) Thomas Aquinas, Opuscufa, XLII, de natura

generd.a, chapters

19 and - 20; Ltl, de natura loci; .'

XXXII, de natura materiae et de demensionibus

intcrminat.is, Compare: C. Jourdin, la Philo sophie de Saint Thomas dtAquin,
p. ,303-329; K, Werner, der Heilige Thomas von Aquino (Regensburg, 1859, Volume


177-201. 10) Even the assemblege of nll continuous functions or even



aU· integrable

functions of one or several variables can have only the
(II); if all the limitntions are dropped and one

power of the second number-class considers

tl1~ asscmbl.llge of

all continuous or discontinuous

junctions of one

or a! several variables

set has the power of the .third number-class (III).

11) The th~orem'can be proven of perfect sets that they can . never have the power of (I). As an example of n perfect
in no interval whatever, however Small, point-set

which is everywhere dense

consider the assemblage of ill real numCv +_.,-.

bers Which are given by the formula:

-::t _£.: +~. .... .. ... tJ - '3 ~t
may consist of a finite
12) It is free


....', ..

where the coefficients Cv may be assigned either the volues 0 or 2 and the series as well as an infinite number of terms.

should be observed that this definition of a continuum

from every reference to th~t which is called the dimension of a continualso such continu;. as are solids,

ous. mn.nifold (Gebildes); the definition includes

composed of connected pieces of different dimensions

ns lines, surfaces,

On a l~ter occassion I shall show how it is possible to proceed in nn

orderly f~shion from this general continuum to the specinl continu~ with definite



only ~

well that the word tlo.ontinuumn has previously not

had a .precise



fq 'definition will bo judged by some to be too narrow, .

by others as too .__. I trust tha.t I have succee,ded in fiilcl1ng a . proper mean broad; .


.between the two. In lfW definition, by a continuum is meant only a perfect and
. .

connected manifold.


to this, for example, a straight line are lacking, or a circular


one or both of whose end-points

lamina whose bound-

ary is excluded, are not completed oontinuua; I call such point-sets tinuua..
In general I understand


by a somi-continuum. an impertect,' con-

nected point-set

belonging to the second-class Which is so constituted that every
by a complete continuum Which is itself a part

two points of it can be connecbed

(Bestnndtheil) of the point-set.

So, for instance, is [591] the space designat-

ed by me in VolumeXX, p. 119 of the Math. Annalem as
.. Gn by the removal of any point-set

at ..which arises


of the first power-, a semi-continuum.

The deriv~tive ot
in which case it is of no importance

connected point-set is always a continuum,

"Whether t~e connected set has the t:1,.rst or

second power. If a oonnected point-set either is' of the first power it can be called


continuum or a semi-continuum, Through the concepts Which


have set at the head of the theory all the manifolds of algebraic

of JIlB.n1folds obligate mYself to investigate I as well as transcendental

geometry for all possibilities,

so that the univer-

sality and sharpness of the results can not be surpassed by other methods,

Halle, October, 1682