There seems to be a pretty universally accepted theory that in doing a
design safety analysis, the protocol is that the hazards associated
with the product should first be designed out; Then those that cannot be
feasibly designed out should be guarded against; THEN, those that cannot be
feasibly guarded against should be managed by operator protective gear
and warnings.

There are variations on how this is expressed, but the basic principal
is that you shouldn't, for example, use a warning to manage a hazard
that can effectively be eliminated at a point higher in the protocol,
and instead only rely on warnings for
those hazards that survive other steps in the process.

What I am looking for are the most authoritative expressions of this
protocol possible. If there is a standard from the ASTM, or the SAE,
or a safety engineering society, or a military specification (I am
almost certain there is, but I cannot put my finger on it), that is
what I am looking for. Something that will help me establish this
principal without just saying that 'I hold these truths to be
self-evident'.