# Is there some sense in which these two definitions of subobject are “the same”?

Let $X$ denote an object of a category.

Usually, we define that the subobjects of $X$ are equivalence classes of monomorphisms targeting $X$ such that for all $m,m' : M,M' \rightarrow X$ we have that $[m]=[m']$ iff there is an isomorphism $i : M \rightarrow M'$ making the obvious diagram commute.

Alternatively, we could define that the subobjects of $X$ are equivalence classes of morphisms targeting $X$ (not necessarily monic) such that for all $f,f' : F,F' \rightarrow X$ we have that $[f]=[f']$ iff there is a exist morphisms $i,i' : M \rightarrow M'$ (not necessarily inverses of each other) such that the two resultant diagrams commute.

Is there some sense in which these two definitions are "the same"?

-
It's not an "alternative" if you can't show it to be equivalent. Consider the category of abelian groups, and take $X = \mathbb{Z} / 2 \mathbb{Z}$; $\mathbb{Z} \to \mathbb{Z} / 2 \mathbb{Z}$ is not (strongly) connected to any subobject of $\mathbb{Z} / 2 \mathbb{Z}$. –  Zhen Lin Sep 16 '13 at 16:16
@ZhenLin, yes, of course you're right. If you have any thoughts about how to repair the "alternative" definition, please comment. –  goblin Sep 16 '13 at 16:31