In [Tables of cellular automata][1], p.542, Wolfram defines the density $\delta$ of a rule to be the asymptotic density of nonzero sites when the initial configuration has density $1/2$. Wolfram quotes rational numbers for densities "whenever analytical
arguments yield exact results. In a few cases, the rigour of these arguments may be
subject to question." 

For rule 110 he quotes $\delta_{110} = 4/7$ (so seems to have an analytical argument). 

For rule 54 he quotes $\delta_{54} = 0.49 \pm 0.01$ (so seems to have no argument for $\delta_{54} = 1/2$). 

Note that rules 110 and 54 both belong to Wolfram's class IV.

My question is two-fold:

 1. I could not find Wolfram's analytical argument for $\delta_{110} = 4/7$. Can anyone give me a reference, please?

 2. The best rational approximation of $0.49$ with denominator $d \leq 32$ is $15/31 = (2^4 - 1)/(2^5 - 1)$. On the other hand we have $\delta_{110} = 4/7 = 2^2/(2^3 - 1)$. **Might there be an analytical argument that yields $\delta_{54} = 15/31$?** Or is the finding spurious and/or numerology?

  [1]: https://content.wolfram.com/uploads/sites/34/2020/07/cellular-automaton-properties.pdf