Arguments that Infinity is Incomprehensible
The Soupdish
Galileo identifies a stucture formed of two interpentrating
solids with the following property: a plane perpendicular to
their common axis will cut two solids (a cone and a strangely
shaped ring) with the same volume. Here is a reconstruction of
his argument:
- a. We can reason about infinity.
b. There are actual infinities.
(The text is a little unclear about which of these
is the assumption for reductio.)
- There exists a figure with the features of the
soupdish.
- Any plane cut perpendicular to the axis of the
soupdish yields two solids of the same volume (a ring
and cone).
- There is a cut of a plane through the soupdish that
yields a point and a circle.
- A point and circle have the same volume (by 2 and 3).
- By definition a point has one point.
- By definition a circle has infinitely many points.
- A circle has more points than a point (by 5
and 6 and the observation that infinitely many is
more than one).
- A volume X is larger
than a volume Y if and only if X has more points than Y.
- The volume of the circle is greater than the volume
of the point (by 7 and 8).
- A point and circle do not have the same volume
(by 9).
4 and 10 contradict. Galileo is unclear about which of two
conclusions we should draw. (1) We might think that an implicit
premise of our argument is that we assumed we can reason about
infinity. Right after this proof, Salviati says, "the infinite
is inherently incomprehensible to us, as indivisibles are
likewise" (38 [77]). (2) We might conclude (knowing what he says
later) that there is no actual infinity of points in a volume
(instead he might say there is only a potential infinity of
points in a volume).
Composing Indivisible: two indivisibles together cannot make a divisible
- Assumption for reductio: we can reason about
indivisibles (or infinitesimals).
- If we could reason about indivisibles, then
we can see that any number of points makes a line.
- Any number of points makes a line (by 1 and 2).
- Five five points will make a line (by 3).
- Any line can be divided evenly in half.
- The line made of five points can be evenly
divided in half (by 4 and 5).
- Then the middle point (the third point) of the
line made of five points can be divided in half.
- No point can be divided in half, including
the middle point (the third point) of the line made
of five points.
7 and 8 contradict. We conclude that we cannot reason about
indivisibles.
Comparing Lines
Galileo compares two lines, one longer than the other.
The proof is more elegant if we assume a line and a segment
of that line.
- Assumption for reductio: there are an actual
infinity of points in any line.
- Suppose line AC is twice the lenght of its
segment AB.
- AC has infinitely many points.
- AB has infinitely many points.
- Any infinity of points is the same size as
any other infinity of points.
- AC and AB have the same number of points
(by 3, 4, and 5).
- If one line is twice as long as the other,
then that line has twice as many points.
- AC has twice as many points as AB (by 7).
- AC does not have the same number of points
as AB (by 8).
5 and 9 contradict.
We conclude that there is no actual infinity of points in any
line.
(NOTE: up to this point, Galileo has made a very important
assumption about the relation between length and the number of points
in line or circle; and about volume and the number of points in a
volume. This is required for the soupdish and the lengths arguments
to have contradictory conclusions. Can you identify this assumption?)
Squares and other functions
- Assumption for reductio: there is an actual infinity
of natural numbers and an actual infinity of their squares.
- If there is a one-to-one correspondence between
all the natural numbers and all the squares, they
are the same quantity.
- There is a one-to-one correspondence between the
natural numbers and the squares, because each
natural number is the root of a square
(x2 is on the naturals and onto the
squares), and each square has a natural number as
root (the inverse of x2, namely the
square root of y, is on the squares and onto the
naturals).
- The natural numbers and the squares are the same quantity
(by 2 and 3).
- If one group is a proper subset of the other, that group
has a smaller quantity.
- The squares are a proper subset of the naturals, because
every square is a natural number but not every natural number
is a square.
- The squares have a smaller quantity than the naturals
(by 5 and 6).
- The natural numbers and the squares are not the same
quantity (by 7).
4 and 8 contradict. We conclude we were wrong to suppose that there
is an actual infinity of natural numbers or of their squares.
Galileo also makes an argument about how the squares become fewer and fewer.
I'm skipping that here. But his point is that it is even more absurd to find
that the squares and the naturals are the same size (as we did in step 4 above)
because as we count through the squares we see we missed more and more of the
natural numbers.
Based on these arguments, Galileo concludes that we cannot quantify
infinities, and so should avoid talking about infinities and also
infinitesimals. He proposes an approach that might seem similar to
Aristotle's notion of potential infinities. Instead of infinity, we
should say that a line, for example, has any number of points; or that
we can identify or make use of each natural number.
Guiding questions as we look ahead
- Refering to infinity is useful! For example, Calculus
appears to refer to infinitesimals. Is it enough to say
just that "we can always find another" or that "there are
at least as many points as any number we choose"? Also,
is it reasonable to ask how many there are when we can
always find another or when there are at least as many as
any number we choose?
- Have we discovered here truly a limit to human reason,
or can we find some way to talk about infinite quantities?
[Revised 13 September 2018.]