This is not the usual request for an intuitive proof, which has been asked already.

Having looked at various sources, I've basically concluded that Heron's formula relies on proving $$xyz = x+y+z$$ where $x$, $y$ and $z$ are the lengths between the meeting point of the incircle and the sides, and the vertices. If I take the diagram below, $x$ is $CX$, and $z$ is $XA$. $y$ would be the segment from $B$ to the incircle meeting the side, marked by a black dot, which is also $YC$. Hence $x+y+z=s$ where $s$ is the semiperimeter. From here it is easy to show that if the radius ($r$) of the incircle is $1$, then the area of the triangle is $x+y+z$. The radius does not really matter, because when it comes to proving the formula, you can just proportionally reduce the size of the triangle by a factor of $1/r^2$. So we can work with the $r=1$ case.

Proving $xyz = x+y+z$ visually is not difficult by looking at this diagram: http://jwilson.coe.uga.edu/emt725/Heron/Heron2/Heron2.html Heron's formula, University of Georgia page

When $r = 1$, then $EY = xy$ ($x = (s-c)$ and $y = (s-b)$ since $s = x+y+z$ as defined above) and $EY$ is also equal to $s/z$, so it is conceptually not difficult to show that $x+y+z = xyz$ and from this that $A^2 = (x+y+z)xyz = s(s-a)(s-b)(s-c)$.

But to my mind, it would be even better if we could show that $xyz$ corresponds to the area of the triangle directly, rather than messing about with equivalences. If we take the $\Delta AYE$ triangle in the above diagram and form a rectangle, whose new vertex we call $P$, and we extend a line from point $X$ to the side $EP$, calling this new vertex $Q$, then $PQXA$ is meant to be of the same area as the triangle, since $XA=z$ and $EY=xy$. Any ways to prove that rectangle $PQXA$ is equivalent to triangle $\Delta ABC$? Or perhaps there's a better way to prove $xyz$ corresponds to the area of the triangle $\Delta ABC$?


  • 1
    $\begingroup$ Please use mathjax to put the text of the question in a better shape. (As it is, it is not really friendly readable.) math.meta.stackexchange.com/questions/5020/… Mathjax is close to latex, and this is indispensable for any mathematician and/or any mathematical text. $\endgroup$ – dan_fulea Jul 10 '20 at 17:35
  • 1
    $\begingroup$ OK, i edited and inserted the picture... $\endgroup$ – dan_fulea Jul 10 '20 at 17:44
  • 1
    $\begingroup$ I don't know what kind of proof you would consider acceptable, but let me notice that $xyz$ has dimensions of volume, hence to become an area it must be divided by $r$. In other words: some kind of ratio must be involved. $\endgroup$ – Intelligenti pauca Jul 10 '20 at 19:32
  • 2
    $\begingroup$ First of all, you need seriously take care of the mess you did with the units. If you state, that "$x, y$ and $z$ are the lengths", then the $xyz$ has the units of volume, while $x+y+z$ is still in length units, and in this case expression $xyz=x+y+z$ make absolutely no sense, not talking about that in other place you state that " the area of the triangle is $x+y+z$". $\endgroup$ – g.kov Jul 10 '20 at 20:18
  • 1
    $\begingroup$ @g.kov You can have a segment length equivalent to the numerical value of the product of two other segments. You can even find a segment that is the square root of a given length: link This does not mean that units are half dimensional. $\endgroup$ – Derek Jul 10 '20 at 20:33

First, let's rectify the dimensionality issues by refraining from setting $r=1$.

With liberal re-labeling of the figure ...

enter image description here

... we have

$$\left.\begin{array}{cr} \triangle A'E'C\sim\triangle CEI \to & \dfrac{h}{s-b} =\dfrac{s-c}{r} \\[6pt] \triangle A'E'A\sim\triangle IEA \to & \dfrac{h}{s} = \dfrac{r}{s-a} \end{array}\right\}\to \frac{(s-b)(s-c)}{r}=h=\frac{rs}{s-a} \tag{1}$$ so that $$(s-a)(s-b)(s-c) = s r^2 \tag{2}$$

Now, one may read the right-hand side of $(2)$ as $|\triangle ABC|^2/s$, so that $|\triangle ABC|^2=s(s-a)(s-b)(s-c)$; that's Heron. OP prefers to interpret the right-hand side as $|\triangle ABC|\,r$, and thus seeks to establish directly that $$|\triangle ABC| = \frac{(s-a)(s-b)(s-c)}{r} \tag{3}$$

In particular, since $h=(s-b)(s-c)/r$ (via $(1)$), OP suggests showing $(3)$ via a demonstration that $|\triangle ABC|=h(s-a)$, perhaps by treating the product as twice the area of a triangle with base $s-a$ and height $h$. While OP considers introducing a rectangle, there's a more-natural option:

enter image description here

A little angle-chasing shows that the marked angles at $B$ and $F$ are congruent, as are those at $C$ and $E$. Thus,

$$\left.\begin{array}{r} \overline{A'B}\parallel\overline{DF}\;\to\;|\triangle A'DF|=|\triangle BDF| \\ \overline{A'C}\parallel\overline{DE}\;\to\;|\triangle A'DE|=|\triangle CDE| \end{array}\right\}\;\to\; \begin{align} \\ \\ |\triangle ABC| &= \phantom{2}\;|\square AFA'E| \\ &=2\;|\triangle AA'E| \\ &= \phantom{2}\;|AE|\;|A'E'|\end{align} \tag{$\star$}$$ as desired. $\square$

It's also worth noting that $|\triangle ABC|=h(s-a)=\frac12h(-a+b+c)$ follows immediately from $A'$'s role as an excenter (being equidistant from the side-lines of $\triangle ABC$): $$|\triangle ABC|+|\triangle A'BC| = |\square ABA'C| = |\triangle AA'C|+|\triangle AA'B|$$ $$\begin{align}\to\quad |\triangle ABC| &= -|\triangle A'BC|+|\triangle AA'C|+|\triangle AA'B| \\[4pt] &=-\tfrac12ha+\tfrac12hb+\tfrac12hc \\[4pt] &=\phantom{-}\tfrac12h(-a+b+c) \end{align}$$

  • 1
    $\begingroup$ Brilliant result! Thanks a lot. $\endgroup$ – Derek Jul 10 '20 at 22:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.