The alloy is an alloy made by me for the occasion, it is virtually
impossible to have contaminated the mix since I was very careful
and clean in the alloying procedure.
Let's work with the assumption that there is nothing wrong with
alloy, that is chemically speaking.
How about pouring temperature?
Metal should be hot enough to completely fill mold before starting to
set. 3 ounces of gold would require 200 degree plus over alloy
liquidus, and that with pre-heated mold.
Lesser amounts would require even larger temperature buffer. If you
are melting only 0.5 ounce of metal, than better is simply melt it on
charcoal. So size of the melt matters.
Mold size is another variable to take a loot at.
Ideally, metal should completely fill the mold. If filled only
partially, unfilled part becomes heat drain and that results in ingot
stress, which would lead to cracking.
Since we live in the world far from ideal, we have to deal with such
stresses by forging ingot prior to rolling. On my website, there is a
video "Frugal Goldsmith", which demonstrates ingot forging.
When you start rolling, watch out for edge feathering. If started to
form, it has to be filed off, and deeper cracks should be filed in
with triangular file to stop cracks growing. Rolling must be done in
one direction only. If change of direction is required, ingot must
Ingot should not bent or become wavy during rolling. If it is, it
may indicate unevenness of ingot grain, or problems with mill itself.
Ingot is more likely candidate so anneal it and forge again. Forging
must achieve at least 30% reduction in thickness to be effective. If
problem persist than mill examination is in order.
If you do not know much about mills, I have a write up on my blog,
which will be useful. Especially pay attention to description of
non-homogenious deformation. While it almost an impossibility to
create conditions for non-homogenious deformation with hand-operated
mill, the theory assumes perfect ingot. Since ingots far from
perfect, the non-homogenious deformation could take place much
earlier that theory predicts.
If nothing of the above will help, that assumption of chemically
good alloy, should be reviewed.
If you describe alloy composition and methodology of smelting, we
can discuss it further.