Subject: [HM] Bolzano and continuity 1 (of 5)
From: Walter Felscher (walter.felscher@uni-tuebingen.de)
Date: Wed Jan 12 2000 - 15:58:59 EST
The purpose of this article, mailed out in five parts, is to serve as
reference to an upcoming article "Bolzano, Cauchy, epsilon and delta".
The content of this article is a reproduction and commented translation
into English of the major parts of
Bernhard Bolzano
Rein analytischer Beweis des Lehrsatzes, dass zwischen je zwey
Werthen, die ein entgegengesetzes Resultat gew„hren, wenigstens eine
reelle Wurzel der Gleichung liege.
Purely analytical proof of the theorem saying that between any
two arguments giving opposite results there is at least one real
root of the equation.
Prag 1817
The principal aim of the commentaries is to show that Bolzano's
notion of continuity, as presented and used in his above work, is
precisely that expressed by today's epsilon-delta definition.
Technical remark.
Quotations start in column 2 and contain as [p.nm] indications where
page nm in the original's pagination begins. Replacements of Greek
letters, not printable in ASCII, are indicated in the form [a: alpha].
Umlaute are reproduced as "a, "o, "u ; the German s-z is reproduced as sz .
Translations start in column 4 and may contain bracketed commentaries.
The German "Gr"osze" has always been translated as 'magnitude' (though
'quantity' would have done as well).
Section 1.
Bolzano begins the preface (Vorrede) of his book with a discussion of
familiar arguments purporting to prove the theorem in question. If
these are based on geometric observations, namely the fact that a
continuous line of simple curvature, first with positive and then
with negative ordinates, must pass through the abscissa in between,
then he points out that the argument is circular as that fact itself
can only be based on the theorem. Not more convincing are arguments
based on kinematical considerations which do rise above proofs from
examples. Discussing such circular argumentations in detail, Bolzano
formulates on p.11 what he considers a correct definition of
continuity:
Nach einer richtigen Erkl"arung n"ahmlich versteht man unter der
Redensart, dass eine Function f(x) f"ur alle Werthe von x , die
inner- oder auszerhalb gewisser Grenzen liegen, nach dem Gesetze
der Stetigkeit sich "andere, nur so viel, dass, wenn x irgend ein
solcher Werth ist, der Unterschied f(x+w)-f(x) [w: omega] kleiner
als jede gegebene Gr"osze gemacht werden k"onne, wenn man nur w so
klein, als man nur immer will, annehmen kann.
According to a correct explanation the phrase that a function
f(x) changes by the law of continuity, for all values of x inside
or outside certain bounds, is understood to mean only this: if x
is such a value then the difference f(x+w)-f(x) can be made
smaller than any given magnitude if ony w can be assumed as small
as one wishes.
So given e>0 , there will hold |f(x+w)-f(x)| < e provided w is 'as
small as one wishes'. In the following it will become clear from
Bolzano's text that this is intended to mean: given e>0 , there is
d>0 such that w<d implies |f(x+w)-f(x)| < e .
After some more discussions of what continuity does not mean, Bolzano
concludes his preface with an outline of the proof of the theorem in
the title, beginning on p.21 :
Folgendes ist eine kurze Uibersicht des des Ganges, den er [i.e.
the proof] nimmt.
Die zu beweisende Wahrheit, dasz zwischen den zwey Werthen a
[a: alpha] und b [b: beta], die ein entgegengesetztes Resultat
gew"ahren, jederzeit wenigstens eine reelle Wurzel liege, beruhet
offenbar auf jener allgemeineren, dasz, wenn zwey stetige Functionen
von x , f(x) und g(x) [g: phi], von solcher Beschaffenheit sind,
dasz f"ur x=a , f(a)<g(a), f"ur x=b aber f(b)>g(b) ausf"allt, allemahl
irgend ein zwischen a und b liegender Werth von x vorhanden seyn
m"usse, f"ur welchen f(x)=g(x) wird.
The following is a short outline of the road taken by the proof.
The fact to be shown rests on the more general statement: if f
and g are continuous such that f(a)<g(a) and f(b)>g(b) then
there is x between a and b such that f(x)=g(x)
Allein wenn f(a)<g(a) ist; so
ist verm"oge des Gesetzes der Stetigkeit auch noch f(a+i)<g(a+i),
wenn man nur i klein genug annimmt.
But if f(a)<g(a) then by continuity also f(a+i)<g(a+i) if i is
sufficiently small. [Commentary: no further proof is given here,
but appears later on p.52 . A proof in today's manner would say:
with e=2(g(a)-f(a)) let i be less than d1, d2 such that y<d1
implies |f(a+y)-f(a)|<e and y<d2 implies |g(a+y)-g(a)|<e , hence
|f(a+y)-g(a+y)| < |f(a+y)+(g(a)-f(a))-g(a+y)| ,
|f(a+y)-g(a+y)| < |f(a+y)-f(a)|+|g(a+y)-g(a)| = 2e = g(a)-f(a). ]
[p.22] Die Eigenschaft des
Kleinerseyns also k"ommt der Function von i , die der Ausdruck
f(a+i) darstellt, f"ur alle Werthe von i zu, die kleiner sind als
ein gewisser. Gleichwohl k"ommt diese Eigenschaft ihr nicht f"ur alle
Werthe von i ohne Einschr"ankung zu; nahmentlich nicht f"ur ein i ,
das =b-a w"are; indem f(b) schon >g(b) ist.
So the functions f(a+i), g(a+i) of i have the property f(a+i)<g(a+i)
for all i below some i0, but not for i=b-a .
Nun gilt der Lehrsatz,
dasz so oft eine gewisse Eigenschaft M allen Werthen einer
ver"anderlichen Gr"osze i , die kleiner als ein gegebener sind, und
doch nicht allen "uberhaupt zuk"ommt: so gibt es jederzeit irgend
einen gr"oszten Werth u , von dem behauptet werden kann, dasz alle i ,
die <u sind, die Eigenschaft M besitzen.
There holds the theorem that if M is a set of numbers i with i>0
and if j<i implies j in M , then there is a largest u such that
i<u implies i in M .
F"ur diesen Werth von i
selbst kann nun f(a+u) nicht < g(a+u) seyn; weil sonst nach dem
Gesetze der Stetigkeit f(a+u+w) < g(a+u+w) w"are, wenn man w nur
klein genug ann"ahme. Und folglich w"are es nicht war, dasz u der
gr"oszte von den Werthen ist, von welchen die Behauptung gilt, dasz
alle unter ihm stehenden Werthe von i , f(a+i) < g(a+i) machen;
sondern u+w w"are ein noch gr"oszerer Werth, von dem dasselbe gilt.
For this u there cannot hold f(a+u)<g(a+u). Because otherwise by
continuity [commentary: there is some d such that w<d implies]
f(a+u+w)<g(a+u+w) ; hence for such w also i<u+w would imply
f(a+i)<g(a+i) which contradics the maximality of u .
Noch weniger aber kann f(a+u) > g(a+u) seyn; indem sonst auch
f(a+u-w) > g(a+u-w) seyn m"uszte, wenn man w klein genug nimmt ; und
folglich w"are es nicht wahr, dasz f"ur alle Werthe von i , die <u
sind, f(a+i)<g(a+i) sey.
Even less f(a+u)>g(a+u) can hold, because otherwise otherwise
by continuity [commentary: there is some d such that w<d implies]
f(a+u-w)>g(a+u-w) ; hence it would not be the case that i<u
implies f(a+i)<g(a+i).
So musz denn also f(a+u) = g(a+u) seyn;
d.h. es gibt einen zwischen a und b liegenden Werth von x ,
n"ahmlich [p.23] a+u , f"ur welchen die Functionen f(x) und g(x)
einander gleich werden. Es handelt sich nur noch um den Beweis des
erw"ahnten Lehrsatzes.
Thus f(a+u)=g(a+u), and it remains to prove the theorem mentioned.
Diesen erweisen wir nun, indem wir zeigen,
dasz jene Werthe von i , von welchen behauptet werden kann, dasz alle
kleineren die Eigenschaft M besitzen, und jene, von denen sich diesz
nicht mehr behaupten l"aszt, einander so nahe gebracht werden k"onnen,
als man nur immer will; woraus sich f"ur Jeden, der einen richtigen
Begriff von Gr"osze hat, ergibt, dasz der Gedanke eines i , welches
das gr"oszte derjenigen ist, von denen gesagt werden mag, dasz alle
unter ihm stehende die Eigenschaft M besitzen, der Gedanke einer
reellen, d.h. wirklichen Gr"osze sey.
This we establish by showing that the i1 , for which i<i1 implies
i in M , can be brought as near as one wishes to the i2 , for
which i<i2 does not imply i in M . From this it follows for
everyone, who has the proper concept of magnitude, that the idea
of an i , being the largest such that all j with j<i are in M ,
is the idea of a real , i.e. of an actual magnitude.
So much for Bolzano's introduction.
This archive was generated by hypermail 2b28 : Wed Jan 12 2000 - 16:29:23 EST