Aleksandrov [A], proved a remarkable property of convex functions.

Theorem.If $f:\mathbb{R}^n\to\mathbb{R}$ is convex, then for almost every $x\in\mathbb{R}^n$ there is $Df(x)\in\mathbb{R}^n$ and a symmetric $(n\times n)$ matrix $D^2f(x)$ such that $$ \lim_{y\to x} \frac{|f(y)-f(x)-Df(x)(y-x)-\frac{1}{2}(y-x)^TD^2f(x)(y-x)|}{|y-x|^2}=0. $$

I know two proofs of this result. One based on the theory of maximal monotone functions and one based on the fact that the second order distributional derivatives of a convex function are Radon measure. Both proofs are mentioned in Second order differentiability of convex functions. Since these proofs use relatively modern techniques not available during Aleksandrov's time, his argument must have been very different.

Question 1.Can you briefly explain what was the idea of the original proof due to Aleksandrov?

My guess would be that his proof was based on methods of differnetial geometry. What else could he use in those days?

Question 2.In there any textbook where I can find the original proof due to Aleksandrov?

**[A] A. D. Alexandroff,**
Almost everywhere existence of the second differential of a convex function and some properties of convex surfaces connected with it. (Russian)
*Leningrad State Univ. Annals [Uchenye Zapiski] Math. Ser.* 6, (1939), 3–35.