You are on page 1of 12

25.05 - 06.06.

2019

The number at the end of an infinite series


or what exactly is wrong with 0.999... = 1
by Andrei P. G.

On the method of shifting infinite series

A couple of days ago I was watching a video1 of YouTube user Mathologer arguing
against the math in another YouTube video2 of user Numberphile. In the end it was made
clear that the solutions involved were indeed valid but only under special conditions in
specific cases. I still however was not satisfied with the explanation for why the original
methods presented were not generally valid although the math seemed to be consistent. It
was explained what the actual standard methods were and how the calculations should go
about, but I felt that it was not explained what exactly was wrong with the original method.
The video was about the infinite sum of “1+2+3+4+...” being equal to -1/12. In the proof there
were involved other sums:
S​1​ = 1-1+1-1+1-1+1-1+...
and
S​2​ = 1-2+3-4+5-6+7-8+...
Not knowing much about infinite sums I paused the video and started evaluating these
sums. So I reasoned that the sums have two solutions each, depending on whether they are
comprised of an even or odd number of ​summands. That ment that ​S​1​ was equal to 0 for an
even infinite sequence and 1 for an odd infinite sequence, while S​2​ was - ∞ for even and + ∞
for an odd infinity. But how can you have an odd or an even infinity? Shouldn’t that mean
that the infinity is bounded, that it has an end? How come that be? However, it has long
been accepted, since Georg Cantor, that we can have different sized infinities. That implies
that we can have two infinite sets that differ in size by just one element, so if we start with
two equal infinite sets and then add 1 to one of the sets, inevitably one of the infinite sets will
be odd and the other one will be even.
Also, it shouldn’t surprise us that we can have a known end of the infinity. After all, infinities
can be exterior/ outwards (since we can always add another element) or interior/ inwards
(since we can always divide something into smaller pieces). For instance, between the
numbers 1 and 2 we have an infinite set of rational numbers, between 1 and 4 we have an
infinite set twice as big, this is an interior infinity the bounds of which we know. The
Mandelbrot set has an infinite inwards pattern but is bounded externally. A small line is a
bounded infinite string of dots, a line twice as big is also comprised of infinite dots but clearly
this infinity is twice as big. A circle is also comprised of an infinity of dots, in this case the first
dot being right next to the last dot, the beginning of infinity right next to the end of infinity; if
this string of dots would be replaced by a sequence of +1 and -1 (or any other two distinct
elements), we would know if the number of dots in the circle is even or odd (granted that we
get to choose where the beginning is).
Now let us get back to the initial sums. The first sum, the infinite sequence of +1 and -1, was
evaluated to be equal to 1/2, a value that surprised me. The proof involved doubling the sum

1
​https://www.youtube.com/watch?v=YuIIjLr6vUA
2
​https://www.youtube.com/watch?v=w-I6XTVZXww

1
by adding each of its terms to a sequence of the same terms but shifted by one place as
follows:
S​1​ = 1 − 1 + 1 − 1 + 1 − 1 + 1 − 1…
S​1​ = 0 + 1 − 1 + 1 − 1 + 1 − 1 + 1…
2S​1​ = 1 + 0 + 0 + 0 + 0 + 0 + 0 + 0…
⟹ 2S​1​ = 1
⟹ S​1​ = 12
Now this is pretty convincing as it is astonishing, but something had to be wrong with this
method. On one hand ½ was none of the solutions I could logically explain using only one
series but on the other hand adding a zero to the sum didn’t seem to modify it in any
meaningful way. Except it did. While adding no matter how many zeros to an infinite sum
does not modify its solutions if you evaluate the sum individually, it does however modify it in
the aforementioned shifting method. This happens because if we are not taking into
consideration the end of the infinity, and if we are not bounding the infinity at both of its ends
making it an inward infinity, adding zeros at its only bounded end (the beginning of the
infinity) is equivalent to replacing its elements with gaps, decreasing the number of elements
in the infinity. Does that make any sense? Probably not yet. To illustrate this better I will use
the trick of turning the infinite sum into a finite one, which is achievable because it has a
repetitive pattern, analysing both its even and odd versions.
So the even infinite sum “1-1+1-1+1-1…” becomes just “1-1+1-1+1-1” while the odd version
becomes “1-1+1-1+1”.

How it was done:


S​1​ = 1 − 1 + 1 − 1 + 1 − 1
S​1​ = 0 + 1 − 1 + 1 − 1 + 1
2S​1​ = 1 + 0 + 0 + 0 + 0 + 0
⟹ 2S​1​ = 1
⟹ S​1​ = 12
As we can see, the shifted sum is shrinked by one element and no longer equal to the one
above it, so what we are basically doing is adding the even version of the sum with its odd
version, and that is why the final solution is an average of the solutions of the two distinct
sums.

How it should have been done:


For the even version: For the odd version:
S​1​ = 1 − 1 + 1 − 1 + 1 − 1 + 0 S​1​’ = 1 − 1 + 1 − 1 + 1 + 0
S​1​ = 0 + 1 − 1 + 1 − 1 + 1 − 1 S​1​’ = 0 + 1 − 1 + 1 − 1 + 1
2S​1​ = 1 + 0 + 0 + 0 + 0 + 0 − 1 2S​1​’ = 1 + 0 + 0 + 0 + 0 + 1
⟹ 2S​1​ = 0 ⟹ 2S​1​’ = 2
⟹ S​1​ = 0 ⟹ S​1​’ = 1

And now we are back to the values that I first speculated to be the actual solutions. Moral of
the story? Shifting the front of your infinity should be symmetric with the shifting of its end,
that is if you want your method to remain valid.

2
Now let’s apply the same method to the infinite series that we applied to the finitized series:
For the even version: For the odd version:
S​1​ = 1 + (− 1 + 1 − 1 + 1 − 1…) + 0 S​1​’ = 1 + (− 1 + 1 − 1 + 1…) + 0
S​1​ = 0 + (+ 1 − 1 + 1 − 1 + 1…) − 1 S​1​’ = 0 + (+ 1 − 1 + 1 − 1…) + 1
2S​1​ = 1 + (+ 0 + 0 + 0 + 0 + 0…) − 1 2S​1​’ = 1 + (+ 0 + 0 + 0 + 0…) + 1
⟹ 2S​1​ = 0 ⟹ 2S​1​’ = 2
⟹ S​1​ = 0 ⟹ S​1​’ = 1
The essence of the problem is as follows:
We have two equal infinite strings of objects:
⚫+⚫+⚫+⚫+⚫+⚫+...
⚫+⚫+⚫+⚫+⚫+⚫+...
We shift one of them by a number of places, say one:
⚫+⚫+⚫+⚫+⚫+⚫+...
⚫+⚫+⚫+⚫+⚫+...
But now how does someone, who didn’t see what we did, know that the second string
doesn’t have a gap made by removing one object? He can’t know, the information is lost in
the representation above.
Instead, what he can see is that one of the infinite strings is larger than the other by one
object:
⚫+(⚫+⚫+⚫+⚫+⚫+...)
(⚫+⚫+⚫+⚫+⚫+...)
This means that by shifting the string we are shrinking the string.
So how do we preserve the information of the shifting? We turn the outwards oriented infinity
into an inwards oriented one by specifying its end:
⚫+(⚫+⚫+⚫+⚫+⚫+...)
(⚫+⚫+⚫+⚫+⚫+...)+⚫
Of course this will probably deem whatever we were trying to achieve by shifting the string
not worthy of pursuing in the first place.

However, I understand that these sums are used in quantum mechanics where the fraction
results make valid solutions. But quantum mechanics works with probabilities and
superpositions, in which case it is true that the probability for the sum above to evaluate to
one or the other solution is one in two. So ½ is a valid solution if we are speaking in terms of
probabilities.

3
Let us proceed to evaluating the second sum. The second sum is based on the validity of the
solution of the first sum.

How it should be:


If S​2​ has an even number of elements: If S​2​ has an odd number of elements:
S​2​ = 1 − 2 + 3 − 4 + 5 − 6 + 7 − 8… S​2​’ = 1 − 2 + 3 − 4 + 5 − 6 + 7 − 8…
S​2​ = (1 − 2) + (3 − 4) + (5 − 6)… + [n − (n + 1)] S​2​’ = 1 + (− 2 + 3) + (− 4 + 5) + (− 6 + 7)… + [− n + (n + 1)]
S​2​ = (− 1) + (− 1) + (− 1)… + (− 1) S​2​’ = 1 + (+ 1) + (+ 1) + (+ 1)… + (+ 1)
S​2​ = − 1 − 1 − 1 − 1 − 1 − 1… − 1 S​2​’ = 1 + 1 + 1 + 1 + 1 + 1 + … + 1
S​2​ = − ∞ S​2​’ = + ∞

How it was done:


S​2​ = 1 − 2 + 3 − 4 + 5 − 6 + 7 − 8…
S​2​ = 0 + 1 − 2 + 3 − 4 + 5 − 6 + 7…
2​S​2​ = 1 − 1 + 1 − 1 + 1 − 1 + 1 − 1…
⟹ ​2​S​2​ = S​1
1
S1 1 1 1
⟹ S​2​ =
2
2 =​ 2 =​ 2 · 2 ​= 4

How it should have been done:


For the even version: For the odd version:
S​1​ = 1 − 2 + 3 − 4 + 5 − 6 + … + (n − 1) − n S​1​’ = 1 − 2 + 3 − 4 + 5 − 6 + … − (n − 1) + n
S​1​ = 0 + 1 − 2 + 3 − 4 + 5 − … − (n − 2) + (n − 1) − n S​1​’ = 0 + 1 − 2 + 3 − 4 + 5 − … + (n − 2) − (n − 1) + n
2S​1​ = 1 − 1 + 1 − 1 + 1 − 1 + …… + 1 − 1 −n 2S​1​’ = 1 − 1 + 1 − 1 + 1 − 1 + …… − 1 + 1 +n
⟹ 2S​1​ = − n ⟹ 2S​1​’ = + n
n ∞ n ∞
⟹ S​1​ = − ​
2 = − 2 = −∞ ⟹ S​1​’ = 2 ​= 2 = +∞
Again, doing it consistently makes the shifting useless in this case.
But we only get two solutions, so in terms of probabilities the solution to this series should
also be ½. So what about ¼? Well, there is a small detail that was not relevant for the
solutions of the first series, the size of the infinities, which must be considered relative to
each other. If we consider that the solution is a superposition of two values, then if one of the
values is even the other will be odd, and it can be odd either by being greater by one or by
being smaller by one. The same applies if the first value is odd. This means that if S​2​ has an
even number of elements then S​2​’ has either S​2​-1 or S​2​+1 elements, and if S​2​’ has an odd
number of elements then S​2​ will have either S​2​’-1 or S​2​’+1 elements.
We must observe that if S​2​ has an even number of elements then the sum of all of them is
|n|
equal to −
2 (where ​𝒏​ is the last number of the series and ​𝒏​ is even) while if S​2​ has an
n+1
odd number of elements the same sum is equal to
2 (where ​𝒏​ is the last number of the
series and ​𝒏​ is odd).

4
Solution pairs for when the first series has even number Solution pairs for when the first series has odd number
of elements (e.n.e.) and ​𝒏​ (implicitly even) is the last of elements (o.n.e.) and ​𝒏​ (implicitly odd) is the last
number of the series: number of the series:

S2​ ​ with e.n.e.: S​2​ with o.n.e.:


S​2A​ =​ (1 − 2) + (3 − 4) + (5 − 6)… + [(n − 1) − n] ​ S​2D​ = (1 − 2) + (3 − 4) + (5 − 6) + … + n
|n| n+1
= − 2
= 2
Its lesser o.n.e. version: Its greater e.n.e. version:
S​2A + |n| = (1 − 2) + (3 − 4) + (5 − 6)… + [(n − 1) − 0] S​2D + (n + 1) = (1 − 2) + (3 − 4) + (5 − 6) + … + [n − (n + 1)]
n−1
=
|n| =​ − = S​2E
2 = S​2B
2
Their average:
Their average:
S 2A +S 2B − |2n| + |2n| S 2E +S 2D − n−1 n+1
2 + 2 1
= w = = 2 ​ ​= y
2 2 = ​0 ​= 2 2

S2​ ​ with e.n.e.: S​2​ with o.n.e.:


S​2A​ = (1 − 2) + (3 − 4) + (5 − 6)… + [(n − 1) − n] S​2D​ = (1 − 2) + (3 − 4) + (5 − 6) + …................................. + n
|n| n+1
= − 2
= 2
Its greater o.n.e. version: Its lesser e.n.e. version:
S​2A + (n + 1) = (1 − 2) + (3 − 4) + (5 − 6)… + [(n − 1) − n] + (n + 1) S​2D​ − n = (1 − 2) + (3 − 4) + (5 − 6) + … + [(n − 2) − (n − 1)]
|n|+2 n+1
= 2 = S​2C
= − 2
= S​2F
Their average: Their average:
n+1 n+1
S 2A +S 2C − |n2| + |n|2+2 S 2D +S 2F 2 − 2
=
1

= 2 = x =​ = ​0​ ​= z
2 2 2 2

Total average:
w+x+y+z 0+ 21 + 21 +0 2
2 1
4 =​ 4 =​ 4 =​ 4

I have rewritten the last pair of elements in a number of different ways for convenience, for
example [n − (n + 1)] was rewritten as [(n − 1) − n] but the value of ​𝑛​ is relative and case
dependent, so caution is advised as to not get lost in notation.
It seems that we only need either one of the columns above to obtain an average of ¼ while
the average of both columns being also ¼.
Anyways, I don’t really know how these sums are actually used in physics so I can’t tell if my
assumptions about them have anything to do with reality.

5
The problem with 0.9999… = 1

Another victim of nonsymmetrical shifting of infinities is the paradox of 0.9999… being equal
to 1. In the following, for denoting the repeating decimals after the decimal point, I will use
the parenthesis notation in the detriment of the vinculum (line above the repetend) as I find it
more convenient, so “0.9999…” becomes “0.(9)” instead
of “ 0.9 ”.
Now let us evaluate the classic proof of 0.(9) = 1.
x = 0.(9)
10x = 9.(9)
10x - x = 9.(9) - 0.(9)
9x = 9
x=1
Pretty convincing right? So where is the problem? To figure it out let us turn the infinite string
of nines into a finite one:
0.(9) → 0.9999
Multiply the finite value by 10:
0.9999 ​· 10 = 9.999
What do we notice? After multiplication, the string of decimals following the decimal point
decreases by one. This information is lost, when having an infinite string of decimals, in the
standard method. This means:
​ n
x = 0.​(9)​→​ there are more nines here (by one)...
n-1
10x = 9.​(9)​→​ than there are here
n-2
100x = 99.​(9)​→​ this infinite decimal string is 2 elements smaller than the first one
So we have a case of different size infinities. In order to add or subtract the periods
(repeating sequence of decimals) we can pull numbers out of them in order to bring them to
an equal size, but we have to keep track of this process:
n n-1 n-2 n-3 n-3
0.(9) = 0.(9)9 = 0.(9)99 = 0.999(9) = 0.99(9)9
n n-1 n+1
0.(9) ≠ 0.(9) ≠ 0.(9)
Apparently, the number of decimal places of a rational number is inverse proportional to the
number of digits in its integer part, so 1 is equivalent to 1.(0)​n-1​ and 11 to 11.(0)​n-2​, there are ​𝑛
decimal places in 0.(1) but ​𝑛​-1 decimal places in 1.(1) and 𝑛
​ ​-2 in 11.(1). The number of
decimal places only increases by multiplication.
Now let us repeat the same process of the proof using a finite number of decimals:

​𝑥​ = 0.9999
10​𝑥​ = 9.999
10x - ​𝑥​ = 9.999 - 0.9999
9​𝑥​ = 8.9991

6
8.9991
​𝑥​ =
9
​𝑥​ = 0.9999

We get what we started with, which is what we should be expecting. Let’s do the same thing
with the infinite string of decimals, but this time shifting the infinite string symmetrically. First
we have to bound the outwards oriented infinite string of decimals in 0.9999… by specifying
its end, which will be another 9 or another sequence of 9s, turning it into an inwards oriented
infinite string: 0.9999…9.
Using informal notation for recurring Using parenthesis notation for recurring
decimals: decimals:
𝑥​ ​ = 0.9999…99 𝑥​ ​ = 0.(9)9
10​𝑥​ = 9.9999…90 10​𝑥​ = 9.(9)0
10x - ​𝑥​ = 9.9999…90 - 0.9999…99 10x - ​𝑥​ = 9.(9)0 - 0.(9)9
9​𝑥​ = 8.9999…91 9​𝑥​ = 8.(9)1
8.9999…91 8.(9)1
​𝑥​ = ​𝑥​ =
9 9
​𝑥​ = 0.9999…99 ​𝑥​ = 0.(9)9 =0.(9)

Going further: Going further:


1 - 𝑥​ ​ = 1.0000…00 - 0.9999…99 1 - 𝑥​ ​ = 1.(0)0 - 0.(9)9
1 - ​𝑥​ = 0.0000…01 1 - ​𝑥​ = 0.(0)1
⟹ 1 = 0.9999…99 + 0.0000…01 ⟹ 1 = 0.(9)9 + 0.(0)1
The mismatch disappears if we don’t overlook the fact that infinities come in different sizes.
But the lengths of decimal strings have long been studied3, so why isn’t that accounted for in
the standard proof of 0.(9) = 1? That is one thing I can not figure out…

Currently, I am not aware of the concept of an irregularity following an infinitely recurring


decimal number or sequence of numbers existing in mathematics. So while we can have
values like 3.(3), 3.1(3), 3.12(3), 3.12(03) etc, there is no concept of values like 3.(3)1 or
3.1(3)1. Current mathematicians are satisfied with considering that 1 divided by 3 is equal to
0.(3), an irreversible operation if you ask me (that is if you don’t invoke convention), rather
than accepting the concept of an ​undistributable remainder​ equal to 0.(0)1.
How it is: How it should be:
1 ÷ 3 = 0.(3) 1 ÷ 3 = 0.(3)3 & 0.(0)1
3 × 0.(3) = 0.(9) 3 × 0.(3) = 0.(9)
0.(9) = 1 0.(9)9 + 0.(0)1 = 1
The same happens with 1/7:
How it is: How it should be:
1 ÷ 7 = 0.(​142857​) 1 ÷ 7 = 0.(​142857​) & 0.(0)1
7 × 0.(​142857​) = 0.(9) 7 × 0.(​142857​) = 0.(9)
0.(9) = 1 0.(9)9 + 0.(0)1 = 1

3
​https://en.wikipedia.org/wiki/Repeating_decimal#Other_properties_of_repetend_lengths

7
I used the ampersand symbol (&) as a remainder operator instead of the plus sign (which I
was using initially) in order to avoid the impression that the sum of the quotient and
remainder is equal to the division they resulted from.

Actually, all numbers having a recurring decimal expansion of a single digit from 0.(1) to
0.(9) are considered to be the exact quotients of digits 1 to 9 divided by 9 respectively (1/9,
2/9...8/9, 9/9), where in reality only 9 fully divides by itself while the rest of the digits produce
undistributable remainders when divided by it:

Considered
Actual solutions to column
products of
a Actual
Natural reversed
Considered products of Col. ​e
digits operation
quotients of reversed +
divided (col. ​b​ × 9)
column ​a operation col. ​f
by 9 and their Quotient Undistributable
(col. ​d​ × 9)
considered of col. ​a remainder
identities

a b c d e f g
1 0.(1) ? n n-1 n
1
9 0.(9) = 1 0.(1) 0.(0)1 0.(9)

2 0.(2) ? n n-1 n-1


2
9 1.(9) = 2 0.(2) 0.(0)2 1.(9)8

3 0.(3) ? n n-1 n-1


3
9 2.(9) = 3 0.(3) 0.(0)3 2.(9)7

4 0.(4) ? n n-1 n-1


4
9 3.(9) = 4 0.(4) 0.(0)4 3.(9)6

5 0.(5) ? n n-1 n-1


5
9 4.(9) = 5 0.(5) 0.(0)5 4.(9)5

6 0.(6) ? n n-1 n-1


6
9 5.(9) = 6 0.(6) 0.(0)6 5.(9)4

7 0.(7) ? n n-1 n-1


7
9 6.(9) = 7 0.(7) 0.(0)7 6.(9)3

8 0.(8) ? n n-1 n-1


8
9 7.(9) = 8 0.(8) 0.(0)8 7.(9)2

0.(9) 8.(9) n n-1 n-1


9 =​? =​?
0.(9) 0.(0)9 8.(9)1
9
9 1 9
≠ ≠ ≠
1 0 9

8
Let us proceed to the next proof4 of 0.(9) = 1:


1
Since ∑ xk = 1−x for |x| < 1 , we have
k=0
2 3
0.9 = 9 (0.1 + 0.01 + 0.001 + · · ·) = 9 (0.1 + 0.1 + 0.1 + · · ·) =
∞ ∞
k k 1 0.9
= 9 ∑ 0.1 = 9 · 0.1 ∑ 0.1 = 0.9 1−0.1 = 0.9 = 1
k=1 k=0


1
First we must check the premise: ∑ xk = 1−x ​for |x| < 1
k=0
So for x = 0.1 we have:
∞ ∞ 1 1 1
∑x k ∑ = = = ​1.(1)​n-1​ & 0.(0)​n-1​1
= 0.1k = 1−x 1−0.1 0.9
k=0 k=0
= 0.1​ + 0.1​1​ + 0.1​2​ + 0.1​3​ +…+ 0.1​n-1​ + 0.1​n ​=
0​
Verification:
= 1 + 0.1 + 0.01 + 0.001 +…+ 0.(0)​n-2​1 + 0.(0)​n-1​1​ ​= 1 ÷ 0.9 = 1.(1)​n-1​ & 0.(0)​n-1​1
We observe that the exponents in the first series indicate
9 9.(9)n−1
the total number of decimal places in the corresponding 1.(1)​n-1​ × 0.9 = 1.(1)​n-1​ × 10 ​= 10 = 0.(9)​n
exponentiated values of the second series.
0.(9)​n​ + 0.(0)​n-1​1 = 0.(9)​n-1​9 + 0.(0)​n-1​1 = 1.(0)​n-1​ = 1
= ​1.(1)​n-1​ + 0.(0)​n-1​1​ = 1.(1)​n-1​1 = 1.(1)​n

So the premise is confirmed as being true. The next three identities are also correct, but then
∞ ∞
follows an identity that, again, needs checking: 9 ∑ 0.1k = 9 · 0.1 ∑ 0.1k . We will take
k=1 k=0
∞ ∞
the nines out from both sides and check that ∑ 0.1k = 0.1 × ∑ 0.1k :
k=1 k=0

∞ ∞
∑ 0.1k ​= 0.1 × ∑ 0.1k =
k=1 k=0
1​ 2​ 3​ n-1​ n​
= 0.1​ + 0.1​ + 0.1​ +…+ 0.1​ + 0.1​ = = 0.1 × [0.1​0​ + 0.1​1​ + 0.1​2​ + 0.1​3​ +…+ 0.1​n-1​ + 0.1​n​] ​ =

= 0.1 + 0.01 + 0.001 +…+ 0.(0)​n-2​1 + 0.(0)​n-1​1 = = 0.1 × [1 + 0.1 + 0.01 + 0.001 +…+ 0.(0)​n-2​1 + 0.(0)​n-1​1] ​ ​=

= ​0.(1)​n = 0.1 × [​1.(1)​n-1​ + 0.(0)​n-1​1​] ​ = 0.1 × ​1.(1)​n-1​1​ ​ ​=
= 0.1 × ​1.(1)​n​ = ​0.(1)​n+1
∞ ∞
k k 1
0.(1)​n​ ≠ 0.(1)​n+1​ ⇒ ∑ 0.1 ≠ 0.1 × ∑ 0.1 ≠ 1−0.1
k=1 k=0

We have a mismatch. What was the mistake?

4
​https://www.uio.no/studier/emner/matnat/math/MAT4010/v16/notater/decimal-expansion.pdf
(​Decimal Expansion of Rational Numbers,​ Helmer Aslaksen, Dept. of Teacher Education & Dept. of
Mathematics University of Oslo)

9
∞ ∞−1 ∞
If k = 1 ​becomes k = 0 ​then ∑ 0.1k becomes 0.1 × ∑ 0.1k and not 0.1 × ∑ 0.1k .
k=1 k=0 k=0
Again we have a case of overlooking the fact that infinities come in different sizes.
If we continue the initial proof from before it went wrong we get:

9 ∑ 0.1k = 9 × 0.(1)n = 0.(9)n = 0.9 …back where we started.
k=1
So 0.(9) is not equal to 1 this time either.

Let’s get to the next proof (same source as the last one5):

Step Original proof with original notation Original proof with notation used in this Actual result
number document and highlighted mistakes for each step

step 0 a.a​1​a​2​ . . .(a​n​ − 1) 9 0.(9)​n-1​9 = 0.(9)​n



step 1 = a.a​1​a​2​ . . .(a​n​ − 1) + 9 ∑ 0.1​k = 0.(9)​n-1​ + 9 × ​0.(0)​n​1 0.(9)​n-1​09
k=n+1

step 2 = a.a​1​a​2​ . . .(a​n​ − 1) + 9 · 0.1​n+1 ∑ 0.1​k = 0.(9)​n-1​ + 9 × ​0.(0)​n​1 × 1.(1)​n 0.(9)​n-1​0(9)​n+1
k=0
n+1 1
step 3 = a.a​1​a​2​ . . .(a​n​ − 1) + 9 · 0.1​ 1 − 0.1 1 0.(9)​n-1​1
= 0.(9)​n-1​ + 9 × ​0.(0)​n​1 × 0.9
9
step 4 = a.a​1​a​2​ . . .(a​n​ − 1) + 0.1​n+1 0.9 9
= 0.(9)​n-1​ + ​0.(0)​n​1 × 0.9 0.(9)​n-1​1
step 5 = a.a​1​a​2​ . . .(a​n​ − 1) + 0.1​n+1​ · 10
= a.a​1​a​2​ . . .(a​n​ − 1) + 0.1​n = 0.(9)​n-1​ + ​0.(0)​n​1 ×10 0.(9)​n-1​1
step 6
​n - 1 = 0.(9)​n-1​ + ​0.(0)​n-1​1 0.(9)​n-1​1

step 7 = a.a​1​a​2​ . . .(a​n​ − 1) + 0.0 . . . 01


step 8 = a.a​1​a​2​ . . .(a​n​ − 1 + 1) = 0.(9)​n-1​ + ​0.(0)​n-1​1 0.(9)​n-1​1
step 9 = a.a​1​a​2​ . . . a​n = ​0.(9)​n-1+1 0.(9)​n
= 0.(9)​n 0.(9)​n

The first identity (steps 0 and 1) wants to say that 0.(9) = 0.(9)​n-1​9 = 0.(9)​n-1​ + 9 × 0.(0)​n-1​1, but

there is a problem, 0.(0)​n-1​1 is expressed as the sum ∑ 0.1​k​ which is weird and incorrect
k=n+1
because, as we saw before, when raising 0.1 to a certain power that power also indicates
the total number of decimal places in the resulting exponentiated number, that means that
​ ​+1 will have 𝑛
0.1 raised to the power 𝑛 ​ ​+1 decimal places of which 𝑛
​ ​ decimal places will be
zeros and the last decimal place (the ​𝑛​+1th) will be 1, which gives 0.(0)​n​1 which multiplied by
9 and added to 0.(9)​n-1​ give 0.(9)​n-1​09 which is not equal to the 0.(9)​n-1​9 we started with.
Another problem is that the aforementioned sum starts at 𝑘​ ​=𝑛 ​ ​+1 and ends at ∞ instead of
∞+1, but ​𝑛​ ​is​ infinity, it is the last element of an infinity, the “infinitieth” element, so the sum
appears to end before it starts. On the other hand, if ∞ is greater than ​𝑛​ it should be

5
​https://www.uio.no/studier/emner/matnat/math/MAT4010/v16/notater/decimal-expansion.pdf
(​Decimal Expansion of Rational Numbers,​ Helmer Aslaksen, Dept. of Teacher Education & Dept. of
Mathematics University of Oslo)

10
mentioned by what magnitude the two infinities differ relative to each other, however this will
result in a recurring 1 after the recurring 0 instead of a single ending 1, that is 0.(0)​n-1​(1)​?​.

In step 2, the sum ∑ 0.1​k​ which is equal to 0.1​n+1​ or 0.(0)​n​1 is somehow also equal to
k=n+1

0.1​n+1​ × ∑ 0.1 or 0.(0)​n​1 × 1.(1)​n​ which is 0.(0)​n​(1)​n+1​ and doesn’t seem to lead anywhere.
k=0
∞ ∞ ∞
In step 3, based on the wrong assumption that ∑ 0.1​k​ = 0.1​n+1 ∑ 0.1​k​ the sum ∑ 0.1​k​ is
k=n+1 k=0 k=0
1
replaced with its equivalent fraction 1 − 0.1 which actually slightly turns the calculations in the
right direction in steps 4 to 7.
In steps 7 and 8 it is concluded that 0.(9)​n-1​ + 0.(0)​n-1​1 is the same as 0.(9)​n-1+1​, which is
obviously false, and from here was drawn the final result which in the original proof does not
specify an actual value it just indicates that we finished with a number that has the same
number of decimal places as the one we started with (a.a​1​a​2​...a​n-1​a​n​), namely 0.(9)​n​. So this
proof is rather a disproof of 0.(9) = 1, but its math is flawed. I don’t really understand why it
was listed as a proof...

A curious infinite decimal string is the product of 9.(9) multiplied by itself. First we make use
of the “finitization” of the decimal string, and then increase it progressively:
9.9 × 9.9 = 98.01 0.9 × 0.9 = 0.81
9.99 × 9.99 = 99.8001 0.99 × 0.99 = 0.9801
9.999 × 9.999 = 99.980001 0.999 × 0.999 = 0.998001
9.9999 × 9.9999 = 99.99800001 0.9999 × 0.9999 = 0.99980001
9.99999 × 9.99999 = 99.9998000001 0.99999 × 0.99999 = 0.9999800001
9.999999 × 9.999999 = 99.999980000001 0.999999 × 0.999999 = 0.999998000001
… …
9.(9) × 9.(9) = 99.(9)8(0)1 0.(9) × 0.(9) = 0.(9)8(0)1

n n n-2 n n n n-1 n-1


​9.(9) × 9.(9) = 99.(9)8(0)1 ​0.(9) × 0.(9) = 0.(9)8(0)1

2n 2n
Where ​n​ is the number of digits (an infinity of a certain size) in the associated decimal string.

We obtain a very odd case of irregular even infinite string of decimal places. The total
number of decimal places of the product is equal to the sum of the decimal places of the
factors. The total number of nines as well as the total number of zeros in the product is equal
to the number of decimal places in one of the factors, although the number of decimal nines
is smaller by two compared to the number of decimal zeros. So the decimal string of this
product contains two infinities of different sizes separated by a single distinct digit and
ending also in a single distinct digit.

11
Let’s do another fun calculation - try dividing the undistributable remainder 0.(0)1:

1 ÷ 3 = 0.(3)3 & 0.(0)1


0.1 ÷ 3 = 0.0(3)3 & 0.0(0)1
0.01 ÷ 3 = 0.00(3)3 & 0.00(0)1
0.001 ÷ 3 = 0.000(3)3 & 0.000(0)1
0.0001 ÷ 3 = 0.0000(3)3 & 0.0000(0)1

n n m-1 n m-1 n+1 m n+1 m-1 n+1 m n+m
0.(0)1 ÷ 3 = 0.0(0)(3)3 & 0.0(0)(0)1 = 0.(0)(3) & 0.(0)(0)1 = 0.(0)(3) & 0.(0)1

As a final remark, all this probably doesn’t have much relevance, mainly because usually
division operations with non-integer quotients are left in the form of common fraction notation
with integer numerators and denominators, such that lossy irreversible operations are
avoided, at least until the final result, any loss of information being negligible as it is not
accumulating in earlier steps of the calculation. On a physical level, whether or not things
like an undistributable remainder matter at all comes down to whether the phenomena
involved are continuous or discrete in their fundamental nature. For a continuous nature, an
undistributable remainder should not exist, while for a discrete nature it could maybe create
some butterfly effect that can turn things in one direction rather than another.
But let’s be honest, an infinite string of zeros followed by the number one, is the smallest
most insignificant error one could possibly make, if there’s one thing that will ever be
negligible, this is the one!

12

You might also like