If you include the biosphere, it's carbon. If you include the oceans, it's hydrogen (the hydrogen atoms in water).
Even if you just include inorganic materials in the crust, I believe oxygen is more abundant than silicon (oxygen is part of silicate rocks).
Oxygen is #1, silicon is #2, aluminum is #3, and iron is #4 (though far less than O or Si). Carbon doesn't even rank in the top 9.
It says "in the Earth's crust". That indicates to me that the biosphere and the oceans are not being counted. Also, it's giving abundances by mass, which is not the only way to do it (I was actually thinking of abundance by atom count).
Later in the same Wikipedia article, the top eleven abundances by mass for the ocean are given: oxygen and hydrogen are the first two, carbon is #10, and silicon doesn't even make the list.
No figures are given in that article for the biosphere; my statement of carbon being the most abundant for that is based on the fact that it forms the "backbone" of all of the main types of molecules in living organisms: proteins, carbohydrates, lipids, and nucleic acids.
As for the "crust", the definition of the Earth's crust I'm pretty sure includes the oceans, the seafloor, and everything down to the mantle, so the biosphere and oceans should be counted there. Hydrogen doesn't rank highly because it has little mass compared to other elements. Atom count seems like a pretty pointless metric; we're talking about resources available for mining, in which case mass is what counts.
Of course silicon doesn't make the list for oceans because it's mostly water, and a lot of dissolved CO2. Count the seafloor and you'll find lots of silicon (and probably some iron, aluminum, titanium, etc.).
If the crust included the oceans, there wouldn't be different figures for the oceans in the same article.
You make a valid point about the seafloor being part of the crust; but I didn't intend to include the seafloor in "oceans".
> Atom count seems like a pretty pointless metric; we're talking about resources available for mining, in which case mass is what counts.
That depends on what we're mining the resource for. For example, if we're mining for metals to use in catalytic converters for vehicles, atom count is the relevant metric, since the catalytic effectiveness depends on the number of atoms, not on the total mass.
No, it isn't; each hemoglobin molecule has just four iron atoms in it (IIRC--each heme structure has one, and I think there are four heme structures in one hemoglobin molecule). Most of each such molecule, by either mass or atom count, is carbon. And each red blood cell is more than just hemoglobin molecules.