The index-one value $2/\pi$ is evaluated with Arb ball arithmetic. Other entries use fresh central-series initialization and Taylor continuation in $t=\sqrt{\theta_n}$. Runs at 150 and 210 working decimal digits use Taylor orders 160 and 210, step caps $1/12$ and $1/18$, and matching radii $1/2$ and $2/3$. The largest relative difference over all nine coefficients, including the analytic control, is less than $1.82\times10^{-154}$. Only 100 significant digits are stored.
A separate radial-coordinate implementation accumulates the pressure integral and uses the polytropic virial identity to recover the mass normalization and then $N_n$. The analytic index-zero and index-one backgrounds are nonzero controls; displaced controls are rejected. Every selected coefficient agrees with this calculation to a relative tolerance of $10^{-40}$, and the six overlapping entries agree with the printed precision of [2], Table 4.1. Readback checks test stored intervals, the index-one anchor, the physical scale equations, and the radius, mass, concentration and pressure identities.
These comparisons do not rigorously bound ODE truncation, propagation or endpoint errors. The attached generator, checks, numerical evidence, source manifest and argument selection record the method and its limitations.