Orval Fairbairn wrote:

The admonition against pencils in aluminum shops also holds for engine
shops. The easiest way to fail a hot steel part is to mark it with a
lead pencil. The steel glows red hot and absorbs the carbon from the
mark, creating an instant stress concentration. An old friend related
how he saw a Lockheed Constellation exhaust manifold with a crack in the
shape of "OK", due to lead pencil.

The steel absorbs the carbon and becomes brittle, whereupon it
cracks after being flexed while in use. The carbon doesn't cause a
stress concentration; that's caused by scratches or nicks that
interfere with the lines of stress in a metal part. A nick in a metal
propeller is an excellent example of a stress riser, as is a scratch in
an aluminum skin.
For a steel to be hardenable, it needs a carbon content of at
least 0.4% carbon. That's not much. Few steels have as much as 1%. 0.6%
is commonly found in hand tools and spring steel, and a steel having 1%
might be found in ball or roller bearings. A pencil mark could raise
the local carbon content a lot, probably to much more that 1%
,especially on a hot item like an exhaust manifold where the heat
allows rapid and deep absorption of the carbon.
Aircraft tech schools will teach you to leave the pencils at
home when working on aluminum.

Dan