P. Leopardi and the author recently investigated, among other things, the validity of the inequality
![$n\theta_n^{(\alpha,\beta)}\!<\! (n\!+\!1)\theta_{n+1}^{(\alpha,\beta)}$](/math_tex/9f6a472492fb25f38bc5b0899763ac6482.gif)
between the largest zero
![$x_n\!=\!\cos\theta_n^{(\alpha,\beta)}$](/math_tex/58d8339759fb58f3d2f53b29b5562dce82.gif)
and
![$x_{n+1}= \cos\theta_{n+1}^{(\alpha,\beta)}$](/math_tex/3484d772f66d581fa3af5ecdf6ee26a182.gif)
of the Jacobi polynomial
![$P_n^{(\alpha,\beta)}(x)$](/math_tex/e8f2b72ba89999dc6c1d0c88e0f0ce6182.gif)
resp.
![$P_{n+1}^{( \alpha,\beta)}(x)$](/math_tex/f4d01ac40fa9169564455cd4482f32d282.gif)
, α > − 1, β > − 1. The domain in the parameter space (α, β) in which the inequality holds for all n ≥ 1, conjectured by us, is shown here to require a small adjustment—the deletion of a very narrow lens-shaped region in the square { − 1 < α < − 1/2, − 1/2 < β < 0}.