This question is related to the [last question about van der Pol's identity for the sum of divisors][1]. In [Touchard (1953)][2] it is mentioned that the sum of divisors $\sigma(n)$ satisfies the following recurrence relation ($n>1$): $$n^2(n-1) = \frac{6}{\sigma(n)} \sum_{k=1}^{n-1}(3n^2-10k^2)\sigma(k)\sigma(n-k)$$ We can evaluate the convolution part with [Ramanujan's identity][3]: $$\sum_{k=0}^n\sigma(k)\sigma(n-k)=\tfrac5{12}\sigma_3(n)-\tfrac12n\sigma(n)$$ which for our case reads like this: $$\sum_{k=1}^{n-1}\sigma(k)\sigma(n-k)=\tfrac5{12}\sigma_3(n)-\tfrac12n\sigma(n)+\tfrac{\sigma(n)}{12}$$ Substituting in van der Pol's equation a perfect number $n = \sigma(n)/2$ and making use of Ramanujan's identity, we find that the perfect number $n$ satisfies the following quartic equation: $$ 8n^4-2n^3+3 \sigma_3(n)n^2+24A_2 =0 $$ where $$A_2 = \sum_{k=1}^{n-1}k^2 \sigma(k)\sigma(n-k)$$ I asked an expert of convolution identities for $\sigma(n)$ if $A_2$ can be evaluated and he said, that one could prove a similar formula, like the one of Ramanujan, "simply by considering the first and the second derivative of suitable identities between Eisenstein series". However I am not very confident with Eisenstein series, so **I am asking the experts for help to help evaluate $A_2$.** Thanks for your help! [1]: https://mathoverflow.net/questions/372476/van-der-pols-identity-for-the-sum-of-divisors-and-a-quartic-polynomial-equation [2]: https://oeis.org/A000385/a000385.pdf [3]: https://en.wikipedia.org/wiki/Eisenstein_series#Ramanujan_identities