= Solution
For $a$ prime to $p$, put
$$
b_a(X)=\frac{X^{-a}-X^a}{X^{-1}-X}
=X^{1-a}\frac{1-X^{2a}}{1-X^2},\qquad f_a(T)=b_a(1+T)^{p-1}.
$$
For positive $a$, $b_a(X)=X^{1-a}\sum_{j=0}^{a-1}X^{2j}$ is a <Laurent polynomial>. For negative $a$, use $b_a=-b_{-a}$. Hence it is regular at $X=1$ with value $a$, and $f_a\in\mathbb Z_p[[T]]^\times$ has constant coefficient $a^{p-1}\equiv1\pmod p$. Evaluation at $\zeta_n$ shows that $c_n(a)$ is a <unit> and reduces to one in the <residue field>, so $c_n(a)\in U_n$.
Because $p$ is odd, $\prod_{\xi\in\mu_p}\xi=1$. Both maps $\xi\mapsto\xi^2$ and $\xi\mapsto\xi^{2a}$ permute $\mu_p$, and therefore
$$
\begin{aligned}
\prod_{\xi^p=1}b_a(\xi X)
&=X^{p(1-a)}\frac{\prod_{\xi^p=1}(1-\xi^{2a}X^{2a})}
{\prod_{\xi^p=1}(1-\xi^2X^2)}\\
&=X^{p(1-a)}\frac{1-X^{2ap}}{1-X^{2p}}=b_a(X^p).
\end{aligned}
$$
This rational-function identity also holds for negative $a$ and extends across the removable point $X=1$. It implies $Nf_a=f_a$. Part 1(b) now proves all the required norm compatibilities, and \b[$c(a)\in U_\infty$], with Coleman series $f_a$.
Use the additive formal coordinate $z=\log(1+T)$, in which $D=d/dz$. Then
$$
f_a(e^z-1)=\left(\frac{\sinh(az)}{\sinh z}\right)^{p-1}.
$$
Its logarithm after normalization by $a^{p-1}$ is even, so all odd derivatives at zero vanish. For the even derivatives, the <hyperbolic cotangent generating function for Bernoulli numbers> gives
$$
\begin{aligned}
\frac{d}{dz}\log f_a(e^z-1)
&=(p-1)\bigl(a\coth(az)-\coth z\bigr)\\
&=(p-1)\sum_{j\geq1}\frac{2^{2j}B_{2j}(a^{2j}-1)}{(2j)!}z^{2j-1}.
\end{aligned}
$$
Here the <Bernoulli numbers> use $B_1=-1/2$. Differentiating $k-1$ more times yields the answer for every index:
$$
\boxed{\delta_k(c(a))=\begin{cases}
0,&k\text{ odd},\\
(p-1)2^k(a^k-1)\dfrac{B_k}{k},&k\geq2\text{ even}.
\end{cases}}
$$
These rational expressions lie in $\mathbb Z_p$ because they equal the integral logarithmic-derivative definition in part (a). In particular $\delta_1=0$; substituting $B_1$ into the even-index formula would give an incorrect answer. For $a=\pm1$ the sequence is one and every derivative vanishes, as the formula also shows.
Back to article page