A terminal object $T$ in a category $\mathcal C$ is an object such that for every object $X$ there exists a unique morphism $X \to T$.

The pullback of two morphisms $f: X \to Z$ and $g: Y \to Z$ is the unique object $P$ with morphisms $p_1 : P \to X$ and $p_2: P \to Y$ such that for every object $Q$ and morphisms $q_1$, $q_2$ there exists a unique morphism $u: Q \to P$ such that the following diagram commutes:

enter image description here

If $Z$ is terminal then $P = X \times Y$. I think the way to see this is to apply a forgetful functor $F: \mathcal C \to \mathbf{Set}$. Then $Z$ maps to a one element set so that $f,g$ become the constant maps and then $P = \{(x,y) \mid f(x) = g(y) \} = X \times Y$.

Is there a different way to see that if $Z$ is terminal then $P = X \times Y$, not involving knowledge of what terminal objects in $\mathbf{Set}$ look like? Maybe not involving $\mathbf{Set}$ at all?

  • 3
    $\begingroup$ Just check that such an object has the universal property a product would have. That would mean it is isomorphic to it. Also, all (co)limits are defined up to isomorphism, thus your pullback isn't "unique" in a strict sense. $\endgroup$
    – Andy
    Aug 25, 2012 at 12:59
  • $\begingroup$ @Andy Right, of course. Thanks! $\endgroup$ Aug 25, 2012 at 13:01
  • 1
    $\begingroup$ @Andy What do you mean by your last sentence? What is not unique in a strict sense? The pullback is universal so it should be unique up to unique isomorphism, no? $\endgroup$ Aug 25, 2012 at 13:06
  • $\begingroup$ Well, the universal properties that define all (co)limits define objects which are "unique up to isomorphism", so when you said that the pullback P of you diagram is the "unique" object such that the diagram commutes, you were a little imprecise, nothing major. $\endgroup$
    – Andy
    Aug 25, 2012 at 13:08
  • 1
    $\begingroup$ This is getting too long to sort out in the comments, if you want you can email me at mecc_3 at hotmail dot com. Anyway, try to do that, it is an easy exercise: suppose that P and P' both satisfy the definition of pullback, this means they both have a unique arrow into each other such that all the diagrams commute. This makes those arrows isomorphisms. $\endgroup$
    – Andy
    Aug 25, 2012 at 14:20

2 Answers 2


Here is an easy way to "see it." The pullback is the limit of the diagram $X \stackrel{f}{\rightarrow} Z \stackrel{g}{\leftarrow} Y$. It is unique up to isomorphism. If $Z$ is the terminal object, there is exactly one $f$ and one $g$. So, such diagrams are one-to-one with pairs $(X,Y)$ of objects, or discrete diagrams with just $X$ and $Y$. Since the product is the limit of the discrete diagram, the pullback is a product (which is again unique up to isomorphism).

(By the way, categories in general do not have forgetful functors to Set, let alone the fact that "forgetful functor" is not a formally defined concept.)


(Using the OP drawing/symbols)

You can see it easily this way:

1-A pullback is a terminal object in the category of cones to the diagram $D:X \stackrel{f}{\rightarrow} Z \stackrel{g}{\leftarrow} Y$

2- A product is a terminal object in the category of cones to the diagram $D':X,Y$ (just the 2 objects).

We are going to show that if $Z$ is a terminal object in $C$, then a cone to $D$ is the same as a cone to $D'$.

Indeed: a cone from vertex $Q$ to $D$ consists of:

any 3 arrows $q_1:Q\to X, q_2:Q\to Y, q_3:Q\to Z$, such that:

$f \circ q_1 = q_3 $

$g \circ q_2 = q_3$

This simply means that it consists of

any 2 arrows $q_1$ and $q_2$ such that:

(1) $f \circ q_1 =g \circ q_2 $

Now if $Z$ is a terminal object, eq. 1 is automatically satisfied by any $q_1:Q\to X, q_2:Q\to Y$, since the lhs and the rhs of eq. 1 will both be equal to the one and only morphism $Q\to Z$.

So our cone from $Q$ to $D$ reduces to:

any 2 arrows $q_1:Q\to X, q_2:Q\to Y$.

(and no other condition)

This is the same as a cone from $Q$ to $D'$.

Conclusion: the cone categories to $D$ and to $D'$ are the same, so their terminals are the same and thus the pullback is the same as the product


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .