If $b$ is a root (in some field of characteristic two) of a polynomial
defined over the prime field GF(2), then $b$, $b^2 = F(b)$, $b^4=F^2(b), b^8=F^3(b)$, and so on
are also roots. Here, $F$ is the Frobenius map.

So if $\beta^{23}$ is in the list, then $F(\beta^{23})=(\beta^{23})^2=\beta^{46}$ should be also on the list.
So i suspected some typo first. At some point, i decided to look for the cited data. The net gave me
an article with the following data:

<<<
To be more precise, the feedback polynomial of SNOW 2.0 is given by
$$\pi(x) = \alpha x^{16}+x^{14} + \alpha^{-1} x^5 +1\in F_{2^{32}}[x]\ ,$$
where $\alpha$ is a root of $x^4 + \beta^{23}x^3 + \beta^{245}x^2 +\beta^{48} x +\beta^{239}\in F_{2^8}[x]$,

and $\beta$ is a root of $x^8 + x^7 + x^5 + x^3 + 1\in\mathbb{F}_{2}[x]$ .>>>

In the wish from the post we do not have the information, that $\beta$ is a root of $P$.

But we have the other information, that $\beta^{23}, \beta^{245},\beta^{48},\beta^{239}$
are roots of $P$.

This is false, since with $\beta$ we have the other roots $\beta^n$ for n among

```
sage: print [ 23*2^k % 255 for k in [0..7] ]
[23, 46, 92, 184, 113, 226, 197, 139]
```

And we miss 245, 48, 239.