There is nothing magical about 'original
sources' that makes them inherently correct.

What are you defining as an "original" source?
I would define such as things such as minutes of meetings, letters, orders,
weather reports, diaries, logbooks. By their nature they are as "correct" as
we are likely ever going to get and are very useful at determining the truth.
For example, S.L.A. Marshall claimed to have participated in the 1918 campaigns
at St. Mihiel, Soissons, and the Meuse-Argonne and won a battlefield
commission. However, military records state that he was a sergeant in the
315th Engineers constructing roads behind the lines. Anyone relying on his
autobiography, "Bringing Up the Rear," to write about him would be using a
secondary source, and would succumb to the errors and distortions it contains.
Had he, however, going directly to the "original source," unit records and
Marshall's military records, he would have avoided that and what he wrote would
be as close to accurate as you can get.

(I hope Art has his
tranquillizers at hand.)

Oral history is not the same as "original source" history. Although it can be
useful, memories can be faulty and individual selection and emphasis of detail
(even if unconscious) can distort. Of course, the first-person account
provides an immediacy to the past that documents rarely can.

Often they were written by people who
had some policy or interest to defend;

What type of document do you have in mind? It may well be that the researcher
is looking for just such proofs of policy interest and is quite pleased to
discover them. (Here I am thinking of things like position papers, memos,
letters, transcripts.)

official, contemporary
reports can still be biased and unreliable.

Well, they are reliable in that they _are_ the contemporary reports and as such
they are valuable for revealing what was conveyed, whether it was, after the
fact, accurate and truthful or not, and may explain why otherwise puzzling
decisions were made.

And 'being there'
doesn't always protect people from being misinformed or poor
observers.

True in general but not necessarily in particular. Having original sources at
hand helps you sort out the accurate memories from the inaccurate ones.

A good researcher with plentiful background knowledge
and a diverse range of sources at hand can provide a very useful
analysis.

Provided he is unbiased or is honest (especially with himself) about his own
biases.

Whether a source is original should IMHO only be one
element in evaluating its credibility, although an important one.

If it is an original source, it _is_ credible, within the confines of what it
is.
But what do you mean by "credible" and "original source"? Could you provide an
example of an "original source" that is not "credible"?

But given a single secondary source and a single
primary source that contradict each other, it would be overhasty
to reject the secondary as wrong
maybe the secondary is based
on five primaries that all contradict your single primary.

What do you mean by "primary" source?
In any case "maybe" shouldn't be part of the equation.
Original historical sources are not fungible, like scientific tests repeated by
various laboratories, with the anomalous one rejected and condemned to a
footnote in report of results.
That doesn't mean every original source carries the same weight. History is
more like criminal prosecution than research science. You interview
eyewitnesses, you examine physical evidence at the "crime" scene, you examine
documents, letters, diaries, computer disks, you take statements from
interested parties or those who know the parties involved, you talk to experts
to understand the significance various pieces of evidence. You try to fit it
all together to detect patterns, motives. In doing this, some things you have
collected will prove invaluable, others inconsequential. But you collect them
all, study them all until you do understand their degree of significance. And
you are always skeptical, assuming nothing without corroboration.
Secondary sources, in this comparison, are like hearsay. It can be useful
during the investigation (although it can also be misleading, so always treat
it with caution). It is never allowed "in court" and if you base your case on
it, you will lose.


Chris Mark