I don't know what you mean by "secondary source" here. Do you mean the accounts of historians? Or do you mean contemporaneous documents? I'm not a trained historian, but the people/colleagues I know who do historical work rely on the latter - they visit archives etc.
For instance, I've never heard of anyone trying to understand the causes of WWI by unearthing fragments in Belgium and France. They read documents. Likewise for trying to understand the war's social consequences. Looking at housing and urban design may be a part of this (eg who do houses after the war have dumb waiters?) but I think it is a relatively small part.
I was (perhaps incorrectly) drawing a distinction between sources in general and physical artifacts, which may not be actually something real historians do, but as that seems to be something you are reasonably comfortable with, I'll stick with treating "sources" as documents or expressions of some kind, rather than artifacts, relics, clothing, etc.
"Primary sources" are directly about the topic at hand. For example, a collection of letters from WWI soldiers to their families back home (perhaps recovered from a cache of undelivered letters) would be primary sources, because they are the personal accounts of people about their own experiences or the things they've personally witnessed or done. Likewise, actual propaganda posters would be a primary source, albeit one understood through the lens of "this is what the government wanted the people to see" and the like. Being a primary source doesn't mean the source needs no interpretation, but it does mean that the source is to at least some degree direct evidence of something. An actual work of literature is also a primary source.
"Secondary sources" are sources which collected, analyzed, interpreted, and/or discussed other, primary sources, often (but not exclusively) ones that we no longer have. So, for example, the poetry of Sappho is almost totally lost to us for a variety of reasons (primarily, she wrote in a dialect that was difficult for Greeks of later centuries to read, so people lost interest in her work and there was seemingly little to no appetite for translating it), so we have almost nothing in terms of primary sources about what she wrote. We do, however, have numerous commentaries on her work by other, later authors, who extol the virtues of her writing. By being second-hand in this way, not the primary speakers' thoughts, but a secondary speaker's personal selection from amongst those thoughts, we are kept at arms' length, and may not have access to critical details that would have been very relevant
to us but not very relevant to the secondary source's author(s). A book talking about events that happened a century before the author's birth, for example, would also be a secondary source even if it doesn't cite any primary sources directly.
Hence why I brought up Snorri Sturluson and his
Prose Edda. TL;DR: Snorri was HUGELY biased for a bunch of reasons, particularly political and personal-power ones, and because he was writing
centuries after Iceland had been fully Christianized, his accounts simply cannot be trusted as they are. The problem is...our only other sources are the Poetic Edda (better but still not good, because it's incomplete and again only written down
after Christianization had been the norm for over a century), tiny fragments/snippets of text, and artifacts like carved depictions of horned figures or stuff that matches how Loki was tortured after he revealed that he caused Baldur's death.
Good history is built on primary sources (as I've used the term here, meaning documents/expressions) and artifacts (which actual historians would also classify as primary sources), and only employs secondary sources as corroboration or extension unless there simply is no other option. When history has to depend on secondary sources alone, a good historian will be very cautious about how they present their conclusions, because we
know it's on much more shaky ground.
