11:15 PM
A general relativity book I'm reading states that given a coordinate chart $\varphi=(x^1,...x^n)$, since the object $dx^1\wedge...\wedge dx^n$ under a coordinate transformation to $(y^1...y^n)$, we have $\mathrm{det}J dy^1\wedge...\wedge dy^n$ (where $J$ is the Jacobian matrix of the coordinate transformation), it is not a differential n-form. Instead it is a tensor density. Isn't that completely wrong?
Anytime we pull back a $n$-form between $n$-manifolds we get a jacobian determinant
Also, even thinking in terms of "transformation laws", that is the transformation law of a $n$-form, just written more compactly with the determinant (that contains the various $\partial x^k/\partial y^h$). A tensor density would require both the determinant and the $\partial x^k/\partial y^h$. I think my source is confused between the tensor densities and form invariance