Don't use armor estimates from random websites

In many internet discussions about modern tanks, one will find people posting armor estimates (in form as RHA equivalency) being posted to compare tanks. These values should not be taken serious, as most of them are based on extremely limited and outdated sources.

One prime example are the armor estimations (called Armor Basics) from Paul Lakowski from SteelBeasts, which have found their way to Scribd. These values were reposted on J. Collin's table-top war gaming reference website (link death, use link with web archive to visit the site) and from their have found their way to Fabian Prado's The Armor Site! and even some of the worse written Wikipedia articles (yes, this is the Abrams' article).

Why are these values wrong and outdated? 

First of all, they are based on a faulty conception of modern armor design. The author believed that all Western composite armor was based on a relatively simple layout, consisting of steel plates, between which layers of ceramic and kevlar were sandwiched. In the same way Chobham armor was described during the 1980s.
We know however from literature, declassified documents from the development of Chobham armor, and from photographs of damaged tanks, that this is not the case. So the basic idea of his armor estimations is already wrong.

Damaged Merkava IV tank - note that this is not "kevlar + ceramic"

The second problem arrives when we take a look at the thickness values. The best example for this is the Leopard 2. According to Armor Basics the Leopard 2 has an armor thickness of 1000 milimetres at the turret corners and 1300 milimetres at the turret center, with a 700 mm thick mantlet!
However we do know that this is not the case, because on different Leopard 2 tanks people have used simple measure bands to measure the armor thickness.
It happens that the Leopard 2 has only about 860 milimetres of frontal armor, with an 420 milimetres thick gun shield (behind this is still a lot of steel from the gun trunion and the gun mount). So the thickness values to calculate the armor protection levels are also false.

Leopard 2 frontal turret armor - 800 mm without backplate!

Wen have seen that the idea for the armor composition is wrong and that the thickness values are wrong. As a result the conclusion of these two factors is also wrong: The armor density.
Calculating the density is a good way to estimate the armor protection of a tank, when you have a basic idea about how modern armor works. But when you already have wrong values for armor thickness, it is obvious that you will end up with either a higher or lower density than the real value.
In case of the Leopard 2, which has approximately 20% - 40% thinner armor than estimated, the density will be a lot lower than it should be. Hence in Armor Basics the Leopard 2 turret armor consists of "2/3 aluminum + SHS + AD-85 + Rubber", i.e. it is made mostly out of aluminium.
In reality however Paul-Werner Krapke, project manager of the Leopard 2 development, describes the armor to utilize different steel alloys of high hardness and ductility in combination with non-metallic and elastic materials, which by my understanding suggests NERA.