I am once again trying to understand Galois's original memoir.
#Galois was a very poor expositor, but his ideas are interesting. He does everything as computationally as possible. I'm following Edwards' account, because I honestly cannot understand Galois's original presentation (I don't know how Liouville ever managed to understand him):
http://www.galois.ihp.fr/wp-content/uploads/2011/12/H.-Edwards.pdf
https://vimeo.com/36957905
One of the first things he does is define is what we would now call a primitive element and he denotes by 𝑉, an integer linear combination of the roots of the polynomial. But the way he constructs this primitive element is interesting. He constructs a gigantic polynomial (for an equation of degree, this gigantic polynomial is of degree \(5!(5! - 1) = 14280\) in the coefficients of 𝑉 and as long as you pick a non-zero value of this polynomial, you have your primitive element.
The argument is very constructive, albeit clumsy by today's standards. A modern construction of a primitive element doesn't go to such lengths to show that you only need to avoid a linear space of smaller dimension to get your primitive element.