How was Mendeleev able to develop his table?

I asked my general chem instructor this and he didn't know:

Mendeleev was able to group elements and discovered a bunch of them and put them in the correct order in the mid-1800s. What equipment did he use to do this? Was he just weighing stuff and guessing?

Specifically, how did he figure out atomic weight?

• I don't know a lot of details, but I know part of it was noticing a pattern of similar reactivities, for example, lithium, sodium, and potassium did similar things, magnesium and calcium did similar things, fluorine, chlorine, bromine, and iodine did similar things. However, I don't really know how the concept of atomic mass or atomic number were arrived at, but you'd need some way to order the elements before their similar properties would have a recognizable periodicity. – user137 Sep 3 '14 at 18:50
• Mendeleev was not the first, he had predecessors that grouped elements by their similarity, though ignoring their weights. Relative element masses were somewhat known at the time, as $pV=nkT$ relates volume with number of molecules in gas, so it allows to guess relation of moles in two gas volumes. And yes, wiki is your friend en.wikipedia.org/wiki/Atomic_theory – permeakra Sep 3 '14 at 19:24

1 Answer

This quote from the wikipedia article on History of Chemistry might provide some insight into the determination of atomic mass:

Chemists in different countries developed several different incompatible atomistic systems. A paper that suggested a way out of this difficult situation was published as early as 1811 by the Italian physicist Amedeo Avogadro (1776-1856), who hypothesized that equal volumes of gases at the same temperature and pressure contain equal numbers of molecules, from which it followed that relative molecular weights of any two gases are the same as the ratio of the densities of the two gases under the same conditions of temperature and pressure.

So this implies that you could take equal volumes of different gases at the same temperature and pressure, then weigh them and take the ratios. This would allow you to get relative masses for $\ce{H2}$, $\ce{N2}$, $\ce{O2}$, $\ce{Cl2}$, $\ce{Br2}$, $\ce{I2}$. Other gases could be used, but they wouldn't be as useful here. You could then determine that $\ce{H2}$ is the lightest gas, and that $\ce{O2}$ is 16 times heavier, $\ce{N2}$ is 14 times heavier, and so on. You don't even need to know that these are diatomic gases in order to make these measurements. Once you had these determined you could move on to gases like $\ce{CO2}$, or $\ce{H2S}$. Once you knew the mass of the gas molecule and what elements were in it, you could calculate how much of each element must be there, and by subtracting the mass of the known element, the remaining mass must be the atomic mass for the other element. As you accumulated more data you could approach substances that don't easily form gases, such as $\ce{NaCl}$. If you know the mass for $\ce{Cl}$ you could determine the mass for $\ce{Na}$.

What I suggest might not be the way atomic masses were determined, but it would probably work. Looking at this image of Mendeleev's original table shows they already had pretty good estimates of atomic mass by 1869.