International Baccalaureate

Investigating the Painter’s Paradox of the

Gabriel’s Horn

Subject: Mathematics HL

Session: May 2017

Candidate Name: Andrej Perković

Session number: 003176-0024

Introduction

Many years ago, when I was still in my eighth grade, I came across an article in school

newspaper describing a magical trumpet that was infinitely long, a trumpet you could

never paint because you would need an infinite amount of paint to do so. But, what

was so magical about that trumpet is that if you were to pour paint into it you would

only need a certain amount to do so, not infinitely much.

Of course, the trumpet in question is Torricelli’s trumpet, introduced by Evangelista

Torricelli. Now, recently when I was deciding on my topic for the internal assessment,

I came across the magical trumpet again. This time, I was thrilled to finally be able to

analyse and understand the strange properties of this body. The curious child in me

galvanised me instantaneously into action.

Gabriel’s Horn or as Torricelli named it the acute hyperbolic solid in his book about it

- De solido hyperbolico acuto. The reference to the Gabriel was given later, due to the

fact that Gabriel is the archangel who uses a horn to announce news. Also because

the horn connects infinite (divine) nature with finite (mortal). This geometric figure has

an infinitely big surface area and a finite volume. It is quite an intriguing object. It

questions our perception of infinity and puzzles us with its paradoxical nature.

1
Gabriel’s horn
1
The way the horn is formed is by rotating the curve 𝑦 = 𝑥 for 𝑥 ≥ 1 around its vertical

asymptote (x-axis).

Graph 1 Gabriel’s horn

Upon knowing about the infinity of the solid, one would hurdle and instantly deduce

all the properties of the solid, primarily the volume and the surface area are logically

supposed to be infinite. But, that is not the case with this infinitely long figure and its
1
finite volume. What about similar curves? Will the solid of revolution of the function 𝑥2

have infinite surface area and finite volume? Turns out that, on the contrary, it has

both finite surface area and finite volume. But, more on this later.

Now, let’s analyse the methods for calculating the surface area and volume of a solid

of revolution

2
Area of solids of revolution

To get to the formula for the area, I first need to introduce the notion of arc length of

the function. To make an approximation, we need to use Riemann sum. Let there be

a continuous differentiable function in the interval 𝐼 = [𝑎, 𝑏] divided into 𝑛 line

segments, with endpoints 𝑃𝑘−1 and 𝑃𝑘 and where width of the interval 𝛥𝑥 = 𝑥𝑘 −

𝑥𝑘−1 is the same.

Graph 1: Continuous function divided into 𝑛 = 9 line segments

Now, the approximate length of the arc will be the sum of all these line segments
𝑛

𝐿 ≈ ∑|𝑃𝑘−1 𝑃𝑘 |
𝑘=1

where |𝑃𝑘−1 𝑃𝑘 | is the length of each segment.

For each subinterval let’s define 𝛥𝑦𝑘 = 𝑦𝑘 − 𝑦𝑘−1 = 𝑓(𝑥𝑘 ) − 𝑓(𝑥𝑘−1 ) (respectively)

and, as already mentioned above, 𝛥𝑥 = 𝑥𝑘 − 𝑥𝑘−1 (universally). Now by the

Pythagorean Theorem, the length of each segment is

|𝑃𝑘−1 𝑃𝑘 | = √𝛥𝑥 2 + 𝛥𝑦 2 = √(𝑥𝑘 − 𝑥𝑘−1 )2 + (𝑓(𝑥𝑘 ) − 𝑓(𝑥𝑘−1 ))2

3
The Mean Value Theorem states that for a function 𝑓(𝑥) differentiable on the open

interval (𝑎, 𝑏) and continuous on the closed interval [𝑎, 𝑏] there is at least one point

𝑐 in (𝑎, 𝑏) such that

𝑓(𝑏) − 𝑓(𝑎)
𝑓′(𝑐) =
𝑏−𝑎

If we rearrange, we get
𝑓(𝑥𝑘 ) − 𝑓(𝑥𝑘−1 )
𝑓′(𝑥𝑘∗ ) =
𝑥𝑘 − 𝑥𝑘−1

𝑓(𝑥𝑘 ) − 𝑓(𝑥𝑘−1 ) = 𝑓′(𝑥𝑘∗ )(𝑥𝑘 − 𝑥𝑘−1 )

𝛥𝑦𝑘 = 𝑓′(𝑥𝑘∗ )(𝑥𝑘 − 𝑥𝑘−1 )

When we put that back into the formula for the length of the segment, we get

|𝑃𝑘−1 𝑃𝑘 | = √(𝑥𝑘−1 − 𝑥𝑘 )2 + (𝑓(𝑥𝑘−1 ) − 𝑓(𝑥𝑘 ))2 =

= √𝛥𝑥 2 + (𝑓′(𝑥𝑘∗ ))2 𝛥𝑥 2 ⇒

|𝑃𝑘−1 𝑃𝑘 | = 𝛥𝑥√1 + 𝑓′(𝑥𝑘∗ )2

Now, if we can take a solid of revolution and divide it into segments, we get a figure

formed out of conical frustums, as seen on the graph 2. A conical frustum is

a frustum created by slicing the top off a cone (with the cut made parallel to the base).

Graph 2: Solid of revolution divided into 𝑛 = 4 frustums

4
The formula for the area of a conical frustum is

𝑟𝑟𝑖𝑔ℎ𝑡 + 𝑟𝑙𝑒𝑓𝑡
𝐴𝑓𝑟𝑢𝑠𝑡𝑢𝑚 = 2𝜋 ( ) |𝑃𝑘−1 𝑃𝑘 |
2
𝑓(𝑥𝑘 ) + 𝑓(𝑥𝑘−1 )
= 2𝜋 ( ) √1 + [𝑓′(𝑥𝑘∗ )]2 𝛥𝑥
2

When 𝛥𝑥 becomes very small (𝛥𝑥 → 0), both 𝑓(𝑥𝑘 ) and 𝑓(𝑥𝑘−1 ) approach the value

of𝑓(𝑥𝑘∗ ), so we get the formula for a frustum to be:

𝐴𝑓𝑟𝑢𝑠𝑡𝑢𝑚 = 2𝜋𝑓(𝑥𝑘∗ )√1 + [𝑓′(𝑥𝑘∗ )]2 𝛥𝑥

Now the total surface area of the solid is a Riemann sum of the surface areas of

individual conical frustums for 𝛥𝑥 → 0

𝑛

𝐴𝑠𝑜𝑙𝑖𝑑 = lim ∑ 2𝜋𝑓(𝑥𝑘∗ )√1 + [𝑓′(𝑥𝑘∗ )]2 𝛥𝑥

𝑛→∞
𝑘=1

If the function 𝑓(𝑥) is Riemann-integrable, then the limit of Riemann sum of the surface

areas of individual conical frustums as 𝛥𝑥 gets smaller (𝛥𝑥 → 0) and, hence, 𝑛 gets

bigger (𝑛 → ∞), is the Riemann integral

𝑏
𝐴𝑠𝑜𝑙𝑖𝑑 = ∫ 2𝜋𝑓(𝑥)√1 + [𝑓′(𝑥)]2 𝑑𝑥
𝑎

Inserting the function for Gabriel's Horn, we can prove that its surface area is infinite

𝑚
1 1 2 𝑚
1
𝐴ℎ𝑜𝑟𝑛 √
= lim ∫ 2𝜋 ( ) 1 + (− 2 ) 𝑑𝑥 ≥ lim ∫ 2𝜋 ( ) 𝑑𝑥 =
𝑚→∞ 1 𝑥 𝑥 𝑚→∞ 1 𝑥

= lim 2𝜋 [ln 𝑥]1𝑚 = 2𝜋 lim (ln 𝑚) = ∞

𝑚→∞ 𝑚→∞

5
Volume of solids of revolution

One of the methods by which we can calculate the volume, if it is the case of a solid

of revolution with the curve being rotated around the x-axis is:
𝑏
𝑉 = ∫ 𝜋(𝑓(𝑥))2
𝑎

Since the volume of a solid of revolution is being calculated with integration, inversely

it can be expressed as Riemann sum of volumes of parts of the figure. We can have

two outcomes when a limit of the integral is approaching infinity. It can diverge to

infinity or it can converge to a certain number.

The calculation for the volume of Gabriel's horn goes as follows:

𝑚
2
1 𝑚
1𝑚
𝑉 = lim ∫ 𝜋 𝑦 𝑑𝑥 = lim ∫ 𝜋 ( 2 ) 𝑑𝑥 = lim 𝜋 [− ]
𝑚→∞ 1 𝑚→∞ 1 𝑥 𝑚→∞ 𝑥1

1
= lim 𝜋 [1 − ] =𝜋
𝑚→∞ 𝑚

We can see that the integral converges to 𝜋.

We can conclude that, even though the horn is infinitely long, its surface area is

divergent but its volume is convergent

Similar curves

It is intuitive for us to assume that all the horns created by rotating an inverse of power

function have infinite surface area and volume because they are infinitely long. We’ve

already seen that it is not true for Gabriel’s horn. If we think that all horns have finite

volume and infinite are we would still be wrong. Let’s take a look at two examples.

6
EXAMPLE 1.

1
𝑓(𝑥) = 1
𝑥2

The horn generated by rotating this curve around x-axis for 𝑥 ≥ 1 will have volume

𝑚
1
𝑉 = lim ∫ 𝜋 ( ) 𝑑𝑥 = lim 𝜋 (ln 𝑚) = ∞
𝑚→∞ 1 𝑥 𝑚→∞

The surface are is given by

2
𝑚 𝑚 1 𝑚
1 1 1
𝐴 = lim ∫ 2𝜋 ( √
1 ) 1 + (− 3 ) 𝑑𝑥 ≥ lim ∫ 2𝜋 ( 1 ) 𝑑𝑥 = lim 2𝜋 [2𝑥 ]
2
𝑚→∞ 1 𝑚→∞ 1 𝑚→∞ 1
𝑥2 2𝑥 2 𝑥2

1 𝑚
= lim 2𝜋 [2𝑚2 − 1] =∞
𝑚→∞ 1

This solid of revolution really does have infinite surface area and volume

EXAMPLE 2.

1
𝑓(𝑥) =
𝑥2

The horn generated by rotating this curve around x-axis for 𝑥 ≥ 1 will have volume
𝑚
1 1 𝑚 1 1 𝑚 𝜋
𝑉 = lim ∫ 𝜋 ( 4 ) 𝑑𝑥 = 𝜋 lim [− 3 ] = 𝜋 lim [ − ] =
𝑚→∞ 1 𝑥 𝑚→∞ 3𝑥 1 𝑚→∞ 3 3𝑚3 1 3

The surface are is given by

𝑚
1 2 2 𝑚
√5 1𝑚
𝐴 = lim ∫ 2𝜋 ( ) √1 + (− ) 𝑑𝑥 ≤ lim ∫ 2𝜋 ( ) 𝑑𝑥 = 2𝜋√5 lim [− ] =
𝑚→∞ 1 𝑥2 𝑥3 𝑚→∞ 1 𝑥2 𝑚→∞ 𝑥1

= 2𝜋√5

This solid of revolution has both finite surface area and finite volume

7
We can see that the intuitive, “logical” assumption are proved wrong by calculus.

Question is to what extent can we rely on our intuition and assumptions in general?

Let’s leave this question for another research and move on to the problem that is at

the focus of this exploration

Painter’s Paradox

The inner surface of the Gabriel’s horn is infinite; therefore an infinite amount of paint

is needed to paint the inner surface. But the volume of the horn is finite (𝜋), so the

inner surface can be painted by pouring a 𝜋 amount of paint into the horn and then

emptying it.

Definition of paradox in Oxford dictionary is that it is a seemingly absurd or self-

contradictory statement or proposition that when investigated or explained may prove

to be well founded or true. This paradox puzzles us with its juxtaposition of infinite and

finite concepts. Even though explanations, solutions and suggestions may be apparent

to us, they are likely emerging from presumed or learned misconceptions of concepts

mentioned in this paradox. Gabriel's horn is an abstract object. If we introduce it to a

real concept such as paint, it can lead to confusion. When faced with an abstract

concept or object, we tend to understand it using the concrete concepts we already

know, i.e. new knowledge is constructed based on existing knowledge. This is the

reducing abstraction phenomena underlined by Orit Hazzan, which happens

spontaneously and differently in every individual. Also the horn can be defined as a

platonic object characterised by its idealism – infinitely long, constructed out of a

curved plane with no thickness etc. On the other hand, paint is a contextualised object

8
characterised by its pragmatism – finite quantity, viscosity, constructed of molecules

etc. In relation to this, I will analyse three possible solutions to the paradox: in the case

where we contextualise the problem and in the case where we decontextualise it.

SOLUTIONS BASED ON CONTEXTUALISATION

Solution 1. The horn’s length approaches infinity, the diameter 𝑓(𝑥) becomes very,

very small (𝑓(𝑥) → 0) and after a certain point, we won’t even be able to see the horn.

Let’s say its visible part has a finite length, i.e. it would appear finitely long to the human

eye. This means that it can be painted with a finite amount of paint, because there is

only a part of the horn visible to the potential painter who will eventually consider it

painted.

Solution 2. We know that paint is made of molecules like all physical objects. Even

though the size of the molecule is very small, there is a part of the horn that is

unreachable by the molecules of the paint. The diameter of the molecules of the paint

cannot be smaller than that of hydrogen. The Van der Waals radius of hydrogen atoms

is 𝑑𝑚𝑖𝑛 = 120 𝑝𝑚 = 1.2 × 10−10 𝑚. Also, the diameter cannot be greater than that of

PG5, the largest stable synthetic molecule ever made, whose diameter is 𝑑𝑚𝑎𝑥 =

10 𝑛𝑚 = 10−8 𝑚. Now we can calculate the minimum and maximum hypothetical

surface that could be covered by the paint. Let’s denote the length of the horn that can

be reached with 𝑚. We get the value of 𝑚 from the equation:

2 2
𝑑 = 2 × 𝑓(𝑥) = ⇒𝑚=
𝑚 𝑑
2
𝑚1 = = 1.67 × 1010
1.2 × 10−10

9
2
𝑚2 = = 2 × 108
10−8

The formula for the surface area that is accessible in a horn of initial radius

𝑟𝑖 = 1𝑚 is:

1𝑚
1 2
√
A = ∫ 2𝜋 ( ) 1 + (− 2 ) 𝑑𝑥
1 𝑥 𝑥

When calculated with the help of GDC and rounded up to 3 significant figures the

minimum and maximum accessible surface area in respect to the size of molecules is:

1.67×1010
1 1 2
𝐴𝑚𝑎𝑥 = ∫ √
2𝜋 ( ) 1 + (− 2 ) 𝑑𝑥 ≈ 149𝑚2
1 𝑥 𝑥

2×108
1 1 2
𝐴𝑚𝑖𝑛 = ∫ √
2𝜋 ( ) 1 + (− 2 ) 𝑑𝑥 ≈ 121𝑚2
1 𝑥 𝑥

We can see that the area that can be covered is somewhere between 121𝑚2 and

149𝑚2 . When we augment solution 2 with solution one, we can say that the part of the

horn that cannot be reached by the molecules is anyways invisible to the human eye,

so the horn can be considered “painted”.

SOLUTION BASED ON DECONTEXTUALISED PAINT

Solution 3. Another approach to the problem can be achieved if we decontextualise

the paint and make it platonic, like the horn itself. We ought to use a mathematical

paint (Gethner, 2005), a decontextualised platonic paint of thickness greater than 0 or

10
more precisely with the thickness of the paint layer at 𝑥 being smaller than or equal to
1
(Wijeratne, Zazkis 2015). As I prefer to elaborate it like this and as I’ve visually
𝑥

represented it on the Graph 3, one way we could paint the horn is if we were to observe

the layer of paint (orange) as a section of the space (orange and light grey) inside the

horn and the required amount of paint as a part of the volume of the horn. Then we

would require 𝜋 or less cubic units of paint.

Graph 3: A cross-section of the horn

Conclusion

It is very hard and counterintuitive to imagine an infinitely long trumpet that has volume

equal to 𝜋 cubic units. Not to mention trying to paint that same horn. In this

Mathematical Exploration, the Painter’s Paradox is analysed and three possible

solutions are presented. The analysis proves the paradoxical nature of the horn using

Riemann’s integrals. In this exploration, I remained within the boundaries of Euclidean

geometry. The horn can also be analysed using the hyperbolic (non-Euclidean)

11
geometry. This is just another method that would lead to same results and have no

significant improvements in the analysis.

The problem occurs when we try to combine objects of completely different natures,

such as paint from our everyday life and this horn existing only it the realms of

hypothetical geometry. These differences force us to use different perspectives when

looking at this problem – hypothetical, platonic one and pragmatic, contextual one.

Now I come to think that paradoxes are not necessarily unsolvable problem, but just

problems that can have solutions when observed from various angles, not just one.

To conclude, I think that this paradox is exactly as it defined by the Oxford dictionary

– a contradiction that we can, after examining it from different angles, prove to be quite

possible.

12
Bibliography

Gethner, R.M. 2005, ‘Can you paint a can of paint?’, The College Mathematics
Journal vol. 36, no. 5, pp.400-402. [20 August 2016]

Weisstein, E W n.d., Mean-Value Theorem Available from

http://mathworld.wolfram.com/Mean-ValueTheorem.html [6 November 2016]

Weisstein, E W n.d., Conical Frustum Available from

http://mathworld.wolfram.com/Mean-ValueTheorem.html [6 November 2016]

Continuous function divided into line segments n.d., image. Available from
http://tutorial.math.lamar.edu/Classes/CalcII/ArcLength_files/image001.gif
[20.8.2016]

Solid of revolution divided into conical frustums n.d., image. Available from
http://tutorial.math.lamar.edu/Classes/CalcII/SurfaceArea_files/image004.gif [20
August 2016]

Hazzan, O 1999, ‘Reducing abstraction level when learning abstract algebra

concepts’, Educational Studies in Mathematics vol. 40, no. 1, pp. 76-76. Available
from http://centroedumatematica.com/ [9 November 2016]

Wijeratne, C & Zazkis, R, 2015, ‘On Painter’s Paradox: Contextual and Mathematical
Approaches to Infinity’, International Journal of Research in Undergraduate
Mathematics Education, vol. 1, no. 2, pp. 163-186 [18 August 2016]

13