In "Introduction to functional differential equations" by Hale and Lunel (1993), I found the following definition of principal term.
Let $P(z,w)$ be a polynomial of the form
\begin{equation} P(z,w)=\sum_{m=0}^{r}\sum_{n=0}^{s} a_{mn} z^m w^n \; . \end{equation}
We call $a_{rs}z^rw^s$ the principal term of the polynomial if $a_{rs}\not=0$ and, if, for each other term $a_{mn}z^mw^n$ with $a_{mn}\not=0$, we have either $ r > m$, $s > n$ or $r = m$, $s>n$, or $r > m$, $s = n$.
I honestly do not understand this definition; if we write the polynomial as above with the assumption that $a_{rs}\not=0$, doesn't this means the principal term is exactly $a_{rs}$?
The polynomial $P(z,w)=z+w$ does not have a principal term; to write it in the form \begin{equation} P(z,w)=\sum_{m=0}^{r}\sum_{n=0}^{s} a_{mn} z^m w^n \;, \end{equation} we must have $r,s\geq1$. The only nonzero coefficients are $a_{10}=a_{01}=1$, so in particular $a_{rs}=0$. This means the polynomial has no principal term.
And indeed the definition is needlessly roundabout. Given a polynomial \begin{equation}\tag{1} P(z,w)=\sum_{m=0}^{r}\sum_{n=0}^{s} a_{mn} z^m w^n \; , \end{equation} it is immediate that for every coefficient $a_{mn}$ of $P(z,w)$ we have $0\leq m\leq r$ and $0\leq n\leq s$. So for each coefficient other than $a_{rs}$ indeed the condition
is satisfied. Hence the latter part of the definition is redundant.
What the definition seems to aim at, is the fact that the representation of the form $(1)$ is not unique. Indeed one might as well write $$P(z,w)=\sum_{m=0}^{r+1}\sum_{n=0}^{s+1} a_{mn} z^m w^n \; ,$$ and define $a_{mn}=0$ if $m=r+1$ or $n=s+1$ or both. The definition fails, however, by incorrect double usage of the variables $r$ and $s$. This can be fixed as follows:
Alternatively, and slightly less concretely, one might define it as follows: