\documentclass[11pt,a4paper]{article} \usepackage[utf8]{inputenc} \usepackage{amsmath,amssymb,amsthm} \usepackage{geometry} \usepackage{booktabs} \usepackage{graphicx} \usepackage{float} \usepackage{caption} \usepackage{hyperref} \geometry{margin=0.95in} \hypersetup{colorlinks=true, linkcolor=blue, citecolor=blue, urlcolor=blue} \newtheorem{theorem}{Theorem}[section] \newtheorem{proposition}[theorem]{Proposition} \newtheorem{corollary}[theorem]{Corollary} \renewcommand{\labelitemi}{$\bullet$} \title{\textbf{Plane Algebraic Curve Families from Normalized Zeta Series: Prime-Base Extensions and Composite Breakdown}} \author{Ryan Hackbarth \\ \texttt{\href{mailto:hackbarthr16@hanover.edu}{hackbarthr16@hanover.edu}}} \date{\today} \begin{document} \maketitle \begin{abstract} In a companion study of Euler's alternating series $\eta(s) = (1 - 2^{1-s})\zeta(s)$, distributing the prefactor across individual summands $t_n(s) = (-1)^{n-1}(1 - 2^{1-s})^{-1}n^{-s}$ exposed two distinct families of plane curves along vertical lines $s = a + ib$ in the critical strip. On the critical line $\operatorname{Re}(s) = \frac{1}{2}$, the pure powers $t_{2^k}(\frac{1}{2} + ib)$ trace an infinite hierarchy of closed, irreducible plane algebraic curves of degree $2k$ in $\mathbb{Z}[x, y]$ whose enclosed signed areas sum to zero, whereas weighting non-power terms by $(1 - 2^{-s})$ yields concentric circles centered at $(0, 0)$ uniquely at $a = \frac{1}{2}$ and open, precessing rosettes for $a \neq \frac{1}{2}$. This sequel investigates whether the emergence of these curve families is specific to base $2$ or extends to general integer normalizations $\eta_q(s) = (1 - q^{1-s})\zeta(s)$ for $q \ge 2$. We show that every prime base $p$ generates an analogous two-family structure: the pure-power terms $t_{p,p^k}(\frac{1}{2} + ib)$ form irreducible degree-$2k$ algebraic curves over $\mathbb{Z}$ governed by a two-step polynomial recurrence, a Blaschke-involution elliptic perimeter integral, and a zero-sum signed-area law; their seed quartics $t_{p,p^2}(\frac{1}{2} + ib)$ evolve geometrically from figure-eights ($p = 2, 3$) to teardrops and asymptotically circular ovals ($p \ge 5$) while retaining degree-$4$ integer structure; their rational equations in $\mathbb{Q}[x, y]$ uniquely single out $a = \frac{1}{2}$ among rational abscissas $a \in (0, 1) \cap \mathbb{Q}$; and all paired non-power curves ($n \neq p^k$) transition from concentric circles of radius $\mathcal{R}_{p,n} = |c_p(n)|/\sqrt{pn}$ on $a = \frac{1}{2}$ to two-dimensional rosettes off it. When $q$ is composite, however, this classification fractures. Proper prime powers $q = p^m$ ($m \ge 2$) produce rational algebraic curves on $2m - 1$ vertical lines $a = \frac{r}{2m}$ and create a hybrid third family of resonant sub-powers $p^j$ ($m \nmid j$) that remain closed algebraic loops off $\operatorname{Re}(s) = \frac{1}{2}$, while multi-prime composites fail to gather their prime divisors into algebraic towers and break the multiplicative structure of Euler's product. \end{abstract} \section{Introduction} For $\operatorname{Re}(s) > 1$, the Riemann zeta function is defined both additively and multiplicatively by \begin{equation} \zeta(s) = \sum_{n=1}^\infty \frac{1}{n^s} = \prod_{p\text{ prime}} \big(1 - p^{-s}\big)^{-1}. \end{equation} To continue $\zeta(s)$ into the critical strip $0 < \operatorname{Re}(s) < 1$, Euler introduced the alternating Dirichlet series $\eta(s) = (1 - 2^{1-s})\zeta(s) = \sum_{n=1}^\infty (-1)^{n-1}n^{-s}$, canceling the simple pole at $s = 1$ so that the series converges conditionally throughout $\operatorname{Re}(s) > 0$ \cite{Euler1749,Titchmarsh1986}. In our companion paper \cite{Hackbarth2026}, absorbing $(1 - 2^{1-s})^{-1}$ directly into each summand gave the termwise representation $\zeta(s) = \sum_{n=1}^\infty t_n(s)$ with $t_n(s) = (-1)^{n-1}(1 - 2^{1-s})^{-1}n^{-s}$. Tracking the trajectories $b \mapsto t_n(a + ib)$ along vertical lines $s = a + ib$ separated the positive integers into two contrasting curve families: \begin{itemize} \item \textbf{Pure-power algebraic curves ($n = 2^k$):} Sharing the single frequency $\ln 2$ in numerator and denominator, $t_{2^k}(\frac{1}{2} + ib)$ closes into an irreducible degree-$2k$ curve in $\mathbb{Z}[x, y]$ with complete elliptic perimeter and signed areas summing to zero across $k \ge 1$. \item \textbf{Paired circles and rosettes ($n \neq 2^k$):} Pairing odd-index neighbors $P_m(s) = t_m(s) + t_{2m}(s) = m^{-s}S_2(s)$ and multiples-of-$4$ differences $D_{4m}(s) = t_{4m}(s) - t_{8m}(s) = -4^{-s}P_m(s)$ via $(1 - 2^{-s})$ yields origin-centered circles of constant radius on $a = \frac{1}{2}$ and two-dimensional rosettes for $a \neq \frac{1}{2}$. \end{itemize} Because Euler's alternating signs rely on even and odd parity, a natural question is whether generating integer algebraic curves from $\zeta(s)$ requires base $2$, extends to every prime $p$, or works equally well for any integer $q \ge 2$. Here we show that the two-family curve classification extends intact to every prime base $p \ge 2$, examine how the non-circular seed quartics $t_{p,p^2}(\frac{1}{2} + ib)$ evolve geometrically across increasing primes, and determine how composite bases $q \ge 4$ fracture both the rational vertical-line pin and the circle-versus-rosette split. \section{\texorpdfstring{Base-$q$ Normalizations and the Prime Circle--Rosette Split}{Base-q Normalizations and the Prime Circle--Rosette Split}} For any integer base $q \ge 2$, subtracting $q\sum_{m=1}^\infty (qm)^{-s}$ from $\sum_{n=1}^\infty n^{-s}$ on $\operatorname{Re}(s) > 1$ cancels the pole at $s = 1$ by assigning weight $+1$ to integers indivisible by $q$ and weight $-(q - 1)$ to multiples of $q$ \cite{Apostol1976,HardyWright2008}. \medskip \noindent\textbf{Definition 2.1 (Base-$q$ Series and Normalized Curve Terms).} Fix an integer $q \ge 2$ and define the $q$-periodic weight \begin{equation} c_q(n) := \begin{cases} +1, & q \nmid n, \\ -(q - 1), & q \mid n. \end{cases} \end{equation} Set $\eta_q(s) := \sum_{n=1}^\infty c_q(n)n^{-s}$, and away from the prefactor poles $1 - q^{1-s} = 0$ on $\operatorname{Re}(s) = 1$, define the normalized curve terms \begin{equation} t_{q,n}(s) := \frac{c_q(n)}{\big(1 - q^{1-s}\big)n^s} \qquad (n \ge 1). \end{equation} Because every block of $q$ consecutive weights sums to $(q - 1)(+1) - (q - 1) = 0$, Dirichlet's test guarantees conditional convergence across the right half-plane. \begin{proposition}[Termwise Continuation Across $\operatorname{Re}(s) > 0$] For every integer $q \ge 2$, $\eta_q(s)$ converges conditionally on $\operatorname{Re}(s) > 0$ to $(1 - q^{1-s})\zeta(s)$, giving $\zeta(s) = \sum_{n=1}^\infty t_{q,n}(s)$ whenever $\operatorname{Re}(s) > 0$ and $1 - q^{1-s} \neq 0$. \end{proposition} To compare non-power terms across vertical lines, introduce the base-$q$ pairing factor $(1 - q^{-s})$, which converts $(1 - q^{1-s})^{-1}$ into the M\"obius ratio \cite{Needham1997} \begin{equation} S_q(s) := \frac{1 - q^{-s}}{1 - q^{1-s}} = t_{q,1}(s) + \frac{1}{q - 1}\,t_{q,q}(s), \qquad q_{q,n}(s) := \big(1 - q^{-s}\big)t_{q,n}(s) = c_q(n)n^{-s}S_q(s). \end{equation} At $q = 2$, $t_{2,n}(s) = t_n(s)$ and $q_{2,n}(s) = (-1)^{n-1}n^{-s}S_2(s)$ unify the companion paper's odd-index sums $q_{2,m}(s) = P_m(s)$ ($m$ odd, radius $\mathcal{R}_m = 1/\sqrt{2m}$) and multiples-of-$4$ differences $q_{2,4m}(s) = D_{4m}(s)$ \cite{Hackbarth2026}. On $\operatorname{Re}(s) > 1$, summing $q_{q,n}(s)$ across $n \ge 1$ also matches the index-sieved Dirichlet series $(1 - q^{-s})\zeta(s) = \sum_{q \nmid n} n^{-s}$ by canceling multiples of $q$. \begin{proposition}[Critical-Line Circle Criterion for $S_q$] For any integer $q \ge 2$ and any vertical line $s = a + ib$ ($a \neq 1$, $b \in \mathbb{R}$), $S_q(a + ib)$ traces an exact Euclidean circle of period $\frac{2\pi}{\ln q}$ with real center and radius \begin{equation} h_q(a) = \frac{1 - q^{1-2a}}{1 - q^{2-2a}}, \qquad R_q(a) = \frac{q^{-a}(q - 1)}{|1 - q^{2-2a}|}. \end{equation} This circle is centered at $(0, 0)$ if and only if $a = \frac{1}{2}$, where $|S_q(\frac{1}{2} + ib)| = \mathcal{R}_q := \frac{1}{\sqrt{q}}$ is constant. \end{proposition} When $q = p$ is prime, irrationality of $\frac{\ln n}{\ln p}$ for every $n \neq p^k$ produces a clean dichotomy across all non-power indices \cite{Weyl1916}. \begin{theorem}[Circle-Versus-Rosette Family for Prime Bases] Let $p$ be a prime and let $n \ge 2$ satisfy $n \neq p^k$ for all $k \ge 1$ (including $n = 2, 4, 8, \dots$ when $p$ is odd). \begin{enumerate} \item \textbf{Concentric Critical-Line Circles:} At $a = \frac{1}{2}$, every curve $b \mapsto q_{p,n}(\frac{1}{2} + ib)$ traces a circle centered at $(0, 0)$ of constant radius $\mathcal{R}_{p,n} := \frac{|c_p(n)|}{\sqrt{pn}}$. \item \textbf{Off-Line Precessing Rosettes:} For every $a \in (0, \infty) \setminus \{\frac{1}{2}, 1\}$, the curve $b \mapsto q_{p,n}(a + ib)$ unravels into an open, non-periodic rosette densely filling the annulus with inner and outer radii $|c_p(n)|n^{-a}\big||h_p(a)| - R_p(a)\big|$ and $|c_p(n)|n^{-a}\big(|h_p(a)| + R_p(a)\big)$. \end{enumerate} \end{theorem} \section{Pure-Power Algebraic Curve Towers for Prime Bases} Unlike the non-power terms, the unpaired pure powers $t_{p,p^k}(s) = -(p - 1)p^{-ks}(1 - p^{1-s})^{-1}$ ($k \ge 1$) oscillate exclusively at integer harmonics of $\ln p$, so $b \mapsto t_{p,p^k}(a + ib)$ closes into a periodic curve of period $\frac{2\pi}{\ln p}$ on every vertical line $a \neq 1$. Defining the epicyclic step circles $\mathcal{C}_{p,p^k}(s) := (p - 1)p^{-ks}S_p(s)$ (recovering $\mathcal{C}_{2^k}(s) = 2^{-ks}S_2(s)$ at $p = 2$ \cite{Hackbarth2026}), each curve in the tower is generated from its predecessor by phase-locked circle addition. \begin{proposition}[Epicyclic Construction of the Pure-Power Family] For every prime $p$ and every $s$ with $1 - p^{1-s} \neq 0$, the pure-power terms satisfy \begin{equation} t_{p,p}(s) = 1 - S_p(s), \qquad t_{p,p^{k+1}}(s) = t_{p,p^k}(s) + \mathcal{C}_{p,p^k}(s) \quad (k \ge 1). \end{equation} Along $\operatorname{Re}(s) = \frac{1}{2}$, $t_{p,p}$ is a circle of radius $\mathcal{R}_p = \frac{1}{\sqrt{p}}$ centered at $(+1, 0)$, and each step $k \mapsto k + 1$ attaches an origin-centered circle $\mathcal{C}_{p,p^k}$ of radius $\frac{p - 1}{p^{(k+1)/2}}$ rotating clockwise at frequency $(k + 1)\ln p$. \end{proposition} On the critical line $s = \frac{1}{2} + ib$, setting $\theta = b\ln p$ writes $t_{p,p^k}(\frac{1}{2} + ib) = -(p - 1)p^{-k/2}e^{-ik\theta}(1 - \sqrt{p}\,e^{-i\theta})^{-1}$. Despite the appearance of $\sqrt{p}$ in this parameterization, eliminating $\theta$ clears every radical simultaneously. Scaling out the integer factor $p - 1$ via $u_{p,k}(s) := \frac{1}{p - 1}t_{p,p^k}(s) = -p^{-ks}(1 - p^{1-s})^{-1}$ with coordinates $(u, v) = (\operatorname{Re} u_{p,k}, \operatorname{Im} u_{p,k})$---related to $(x, y) = (\operatorname{Re} t_{p,p^k}, \operatorname{Im} t_{p,p^k})$ by $(x, y) = (p - 1)(u, v)$, so $(u, v) = (x, y)$ directly at $p = 2$---places all prime bases under a single recurrence. \begin{theorem}[Integer Polynomial Hierarchy on $\operatorname{Re}(s) = \frac{1}{2}$] Let $p$ be a prime, let $k \ge 1$, and write $(u, v) = (\operatorname{Re} u_{p,k}(\frac{1}{2}+ib), \operatorname{Im} u_{p,k}(\frac{1}{2}+ib))$. Putting $W := p^k(u^2 + v^2)$, the locus of $u_{p,k}(\frac{1}{2} + ib)$ traces an irreducible plane algebraic curve of degree $2k$ in $\mathbb{Z}[u, v]$ defined by \begin{equation} Q_{p,k}(W) - 2p^k W^{k-1}u = 0, \end{equation} where $\{Q_{p,k}(W)\}_{k\ge 1} \subset \mathbb{Z}[W]$ is determined by the initial polynomials \begin{equation} Q_{p,1}(W) = (p - 1)W + 1, \qquad Q_{p,2}(W) = (p - 1)W^2 + (p + 2)W - 1, \end{equation} and the two-step recurrence \begin{equation} Q_{p,k+1}(W) = \big((p + 1)W - 1\big)Q_{p,k}(W) - pW^2 Q_{p,k-1}(W) \qquad (k \ge 2). \end{equation} Every $Q_{p,k}(W)$ has degree $k$, leading coefficient $p - 1$, and constant term $(-1)^{k-1}$; substituting $(u, v) = \frac{1}{p-1}(x, y)$ and clearing $(p - 1)^{2k}$ gives the irreducible degree-$2k$ equation of $t_{p,p^k}(\frac{1}{2} + ib)$ in $\mathbb{Z}[x, y]$ (matching $Q_k(W) - 2^{k+1}W^{k-1}x = 0$ at $p = 2$ \cite{Hackbarth2026}). \end{theorem} Across every prime tower, the positively oriented base circle $t_{p,p}$ and negatively oriented higher-order loops $t_{p,p^k}$ ($k \ge 2$) satisfy an exact signed-area conservation law, while the unit-circle Blaschke involution $w = \frac{\sqrt{p} - z}{1 - \sqrt{p}z}$ reduces every perimeter to a complete elliptic integral \cite{ByrdFriedman1971}. Notably, just as base $2$ produces the $(t_8, t_{16})$ isometric arc-length twin pair \cite{Hackbarth2026}, base $3$ yields a second and final prime isometric twin pair $(t_{3,9}, t_{3,27})$ before third-kind $\Pi$ terms persist across all higher primes $p \ge 5$. \begin{theorem}[Signed-Area Conservation and Elliptic Perimeter Across Prime Towers] Let $p$ be a prime and orient $t_{p,p^k}(\frac{1}{2} + ib)$ by increasing $b \in [0, \frac{2\pi}{\ln p}]$ ($k \ge 1$). \begin{enumerate} \item \textbf{Zero-Sum Signed-Area Law:} The enclosed signed area of $t_{p,p^k}(\frac{1}{2} + ib)$ is \begin{equation} \mathcal{A}_{\mathrm{signed}}\!\big(t_{p,p^k}\big) = \frac{p - k(p - 1)}{p^k}\,\pi, \qquad \sum_{k=1}^\infty \mathcal{A}_{\mathrm{signed}}\!\big(t_{p,p^k}\big) = 0, \end{equation} vanishing for an individual curve if and only if $k = \frac{p}{p-1} \in \mathbb{Z}$ (occurring uniquely for the Bernoulli lemniscate $t_4$ at $p = 2, k = 2$ \cite{Hackbarth2026}). \item \textbf{Blaschke-Involution Elliptic Perimeter:} Setting $\alpha_{p,k} := k - (k - 1)p$, the fundamental-period arc length of $t_{p,p^k}(\frac{1}{2} + ib)$ is given by \begin{equation} \mathcal{L}\!\big(t_{p,p^k}\big) = \frac{1}{p^{k/2}}\int_0^{2\pi} \left|\frac{\alpha_{p,k} - \sqrt{p}\,e^{i\phi}}{1 - \sqrt{p}\,e^{i\phi}}\right| d\phi. \end{equation} \end{enumerate} \end{theorem} \section{Geometric Evolution of Seed Quartics Across Prime Bases} In base $p = 2$, the companion study \cite{Hackbarth2026} centered on the $k = 2$ term $t_4(\frac{1}{2} + ib)$, which serves as the first non-circular curve in the dyadic tower after the shifted circle $t_2 = 1 - S_2$ and anchors all higher inner-loop curves $t_{2^k}$ ($k \ge 3$) under circle addition. For a general prime $p$, the quadratic seed $t_{p,p^2}(\frac{1}{2} + ib) = t_{p,p}(\frac{1}{2} + ib) + \mathcal{C}_{p,p}(\frac{1}{2} + ib)$ occupies this exact structural position: it is the first folded degree-$4$ algebraic curve generated by $Q_{p,2}(W) = (p - 1)W^2 + (p + 2)W - 1$ and the second initial seed required to launch the two-step recurrence for $k \ge 3$. Plotting the curve $t_{p,p^2}(\frac{1}{2} + ib) = \frac{-(p - 1)e^{-2i\theta}}{p(1 - \sqrt{p}\,e^{-i\theta})}$ ($\theta = b\ln p$) across increasing primes shows how the denominator $1 - \sqrt{p}\,e^{-i\theta}$ governs the curve's silhouette. Rationalizing the denominator shows that $\operatorname{Im} t_{p,p^2}(\frac{1}{2} + ib)$ vanishes away from the apsides $\theta \in \{0, \pi\}$ if and only if $\cos\theta = \frac{\sqrt{p}}{2}$, while expanding $(1 - \sqrt{p}\,e^{-i\theta})^{-1}$ in powers of $p^{-1/2}e^{i\theta}$ gives the convergent Fourier representation \begin{equation} t_{p,p^2}\!\left(\frac{1}{2} + ib\right) = \frac{p - 1}{p^{3/2}}\,e^{-i\theta} + \frac{p - 1}{p^2} + \frac{p - 1}{p^{5/2}}\,e^{i\theta} + O\!\left(p^{-2}\right) \qquad (p \ge 2). \end{equation} As illustrated across the first nine primes ($p = 2$ through $p = 23$) in Figure \ref{fig:seed_quartics}, this produces a smooth three-stage geometric progression: \begin{itemize} \item \textbf{Self-intersecting figure-eights ($p = 2, 3$):} Only $p = 2$ and $p = 3$ satisfy $\frac{\sqrt{p}}{2} \le 1$, creating a real-axis double point at $x = \frac{p - 1}{p}$. At $p = 2$, $t_4$ forms the symmetric Bernoulli lemniscate crossing at $(\frac{1}{2}, 0)$ with $\mathcal{A}_{\mathrm{signed}} = 0$; at $p = 3$, $t_{3,9}$ crosses at $(\frac{2}{3}, 0)$ with a smaller right-hand lobe and net signed area $-\frac{\pi}{9}$. \item \textbf{Unfolded teardrops and ovals ($p = 5, 7, 11$):} Once $p \ge 5$, $\frac{\sqrt{p}}{2} > 1$ admits no real angle $\theta$, erasing the self-intersection node so that $t_{5,25}$ unfolds into a simple pointed teardrop that rounds progressively into a smooth convex oval across $p = 7$ and $p = 11$. \item \textbf{Asymptotic circularity ($p = 13, 17, 19, 23$):} As $p$ increases, the leading Fourier mode $\frac{p-1}{p^{3/2}}e^{-i\theta}$ dominates, rounding each seed curve toward a clockwise circle of radius $\frac{1}{\sqrt{p}}$ shifted right by $\frac{1}{p}$ with real-axis apsides at $-\frac{\sqrt{p}-1}{p}$ and $+\frac{\sqrt{p}+1}{p}$. \end{itemize} \begin{figure}[H] \centering \includegraphics[width=0.96\textwidth]{first_9_primes_seed_quartics.png} \caption{Geometric evolution of the degree-$4$ seed quartics $t_{p,p^2}(\frac{1}{2} + ib)$ across the first nine primes ($p = 2, 3, 5, 7, 11, 13, 17, 19, 23$), with the origin $(0, 0)$ marked in black and real-axis double points $(\frac{p-1}{p}, 0)$ marked for $p = 2, 3$. Although each panel is scaled to display its silhouette, the curves contract in true diameter as $O(p^{-1/2})$ and oscillate at $\mathbb{Q}$-linearly independent frequencies $\ln p$.} \label{fig:seed_quartics} \end{figure} Figure \ref{fig:seed_quartics} also clarifies why primality is an arithmetic and whole-series property rather than a silhouette property: even as $t_{p,p^2}(\frac{1}{2} + ib)$ becomes visually indistinguishable from a circle at larger primes, each curve remains an irreducible degree-$4$ locus in $\mathbb{Z}[x, y]$ built from cancelling the quadratic radical $\sqrt{p} \notin \mathbb{Q}$ on its own independent clock $\theta = b\ln p$, from which higher powers $k \ge 3$ fold $(k - 1)$ nested loops around the origin. \section{Field-Theoretic Characterization of the Critical Line} When $a$ moves off $\frac{1}{2}$ inside $(0, 1)$, the pure-power curves $u_{p,k}(a + ib)$ remain closed real algebraic curves in $\mathbb{R}[u, v]$, but their defining coefficients leave $\mathbb{Q}$ at every other rational abscissa. \begin{theorem}[$\mathbb{Q}$-Rationality and Dual Integrality of Prime Curve Towers] Let $p$ be a prime, let $k \ge 1$, and fix $a \in (0, 1)$. Define the base parameter $w_p(a) := p^{2a}$ and write $V := w_p(a)^k(u^2 + v^2)$ for $(u, v) = (\operatorname{Re} u_{p,k}(a+ib), \operatorname{Im} u_{p,k}(a+ib))$. \begin{enumerate} \item \textbf{Off-Line Cartesian Equations:} Along $s = a + ib$, the curve $u_{p,k}(a + ib)$ satisfies the degree-$2k$ equation \begin{equation} P_{p,k}\big(V;\, w_p(a)\big) - 2\big(pw_p(a)\big)^k V^{k-1}u = 0, \end{equation} generated from $P_{p,1}(V; w) = (p^2 - w)V + w$ and $P_{p,2}(V; w) = w(p^2 - w)V^2 + w(p^2 + 2w)V - w^2$ by the recurrence \begin{equation} P_{p,k+1}(V; w) = \big((w + p^2)V - w\big)P_{p,k}(V; w) - p^2 w V^2 P_{p,k-1}(V; w) \qquad (k \ge 2), \end{equation} matching $P_k(V; w)$ of \cite{Hackbarth2026} identically at $p = 2$ and reducing to $p^k Q_{p,k}(W)$ at $w_p(\frac{1}{2}) = p$. \item \textbf{Uniqueness Among Rational Vertical Lines:} For $a \in (0, 1) \cap \mathbb{Q}$, the curve $t_{p,p^k}(a + ib)$ is defined over $\mathbb{Q}[x, y]$ if and only if $a = \frac{1}{2}$. For any $a = \frac{r}{d} \in (0, 1) \setminus \{\frac{1}{2}\}$ ($\gcd(r, d) = 1$), $w_p(a) = p^{2r/d} \notin \mathbb{Q}$ forces coefficients in the radical extension $\mathbb{Q}(p^{2r/d})$. \item \textbf{Dual Integrality Under $s \leftrightarrow 1 - s$:} Across all real $a \in (0, 1)$, $a = \frac{1}{2}$ is the unique abscissa at which both functional-equation partner parameters $w_p(a) = p^{2a}$ and $w_p(1 - a) = p^{2(1-a)}$ belong to $\mathbb{Z}$ (both equaling $p$). \end{enumerate} \end{theorem} \section{How the Curve Families Break Down for Composite Bases} Formally, the circle $S_q(s)$ and polynomial sequence $Q_{q,k}(W)$ remain well-defined for any composite integer $q \ge 4$. However, the clean two-family architecture---one tower of rational algebraic curves pinned to $a = \frac{1}{2}$, plus a complementary family of non-power curves that unravel into rosettes off $a = \frac{1}{2}$ while preserving Euler's prime product---breaks down in two distinct ways depending on the prime factorization of $q$. \subsection{\texorpdfstring{Proper Prime Powers ($q = p^m$, $m \ge 2$)}{Proper Prime Powers (q = p\^m, m >= 2)}} When $q = p^m$ is a higher power of a single prime (such as $q = 4, 8, 9, 16, 25$), fractional powers of $q$ exist as integers $p^j$ ($m \nmid j$) and alter both curve families simultaneously. \begin{theorem}[Extra Rational Lines and Leaked Algebraic Curves for $q = p^m$] Let $q = p^m$ for a prime $p$ and an integer $m \ge 2$. \begin{enumerate} \item \textbf{Proliferation of Rational and Integer Curve Lines:} Because $w_q(a) = q^{2a} = p^{2ma}$, the pure-power curves $t_{q,q^k}(a + ib)$ belong to $\mathbb{Q}[x, y]$ (with integer parameter $w_q(a) = p^r \in \mathbb{Z}$) on all $2m - 1$ rational vertical lines \begin{equation} a \in \left\{\frac{1}{2m},\ \frac{2}{2m},\ \dots,\ \frac{2m - 1}{2m}\right\} \subset (0, 1). \end{equation} Moreover, whenever $m$ is even, off-critical partner pairs $a = \frac{r}{m}$ and $1 - a = \frac{m - r}{m}$ ($1 \le r \le m - 1$) satisfy $w_q(a), w_q(1 - a) \in \mathbb{Z}$ simultaneously (for instance, $w_4(\frac{1}{4}) = 2$ and $w_4(\frac{3}{4}) = 8$ at $q = 4$). \item \textbf{Emergence of a Third Hybrid Family Off $\operatorname{Re}(s) = \frac{1}{2}$:} Every sub-power $n = p^j$ with $m \nmid j$ (such as $n = 2, 8, 32$ when $q = 4$) satisfies $n \neq q^k$ and therefore belongs to the paired non-power family $q_{q,n}(s)$. Because $\frac{\ln(p^j)}{\ln(p^m)} = \frac{j}{m} \in \mathbb{Q}$, $b \mapsto q_{q,p^j}(a + ib)$ fails to unravel into an area-filling rosette for $a \neq \frac{1}{2}$, tracing instead a closed, periodic one-dimensional algebraic curve on every vertical line $a \neq 1$. \end{enumerate} \end{theorem} \subsection{\texorpdfstring{Multi-Prime Composites ($q = 6, 10, 12, \dots$)}{Multi-Prime Composites (q = 6, 10, 12, ...)}} When $q$ has at least two distinct prime factors (such as $q = 6 = 2 \cdot 3$), and is not itself a perfect power, $q^{2a} \in \mathbb{Q}$ once again holds uniquely at $a = \frac{1}{2}$ for rational $a \in (0, 1)$, and no rational frequency ratios leak into $n \neq q^k$. Instead, the curve decomposition loses alignment with the prime factorization of the integers. For a prime base $p$, every prime power $p^k$ belongs to the closed algebraic curve family $\{t_{p,p^k}\}$, while the index-sieved series $(1 - p^{-s})\zeta(s) = \sum_{p \nmid n} n^{-s} = \prod_{\ell \neq p}(1 - \ell^{-s})^{-1}$ ($\operatorname{Re}(s) > 1$) is supported on integers indivisible by $p$---a set closed under multiplication by Euclid's lemma \cite{HardyWright2008}. Multi-prime bases forfeit both properties. \begin{theorem}[Prime-Tower Stranding and Multiplicative Breakdown for Multi-Prime $q$] Suppose $q \ge 6$ is divisible by at least two distinct primes $p_1, \dots, p_\omega$ ($\omega \ge 2$). \begin{enumerate} \item \textbf{Stranded Prime-Power Rosettes:} Only joint powers $n = q^k$ enter the closed algebraic curve tower $\{t_{q,q^k}\}$. For every constituent prime $p_i \mid q$ and every $k \ge 1$, the pure prime power $t_{q,p_i^k}(a + ib)$ has irrational frequency ratio $\frac{k\ln p_i}{\ln q} \notin \mathbb{Q}$ and unravels into an open two-dimensional rosette even on the critical line $a = \frac{1}{2}$. \item \textbf{Loss of Multiplicative Closure and Euler Factorization:} Writing $q = AB$ with $\gcd(A, B) = 1$ and $A, B > 1$, the indicator $f_q(n) := \mathbf{1}_{q \nmid n}$ satisfies $f_q(A) = f_q(B) = 1$ but $f_q(AB) = 0$. Consequently, the paired series $(1 - q^{-s})\zeta(s) = \sum_{n=1}^\infty f_q(n)n^{-s}$ lacks multiplicative coefficients and does not factor as a sub-product of Euler's prime product. \end{enumerate} \end{theorem} \section{Classification of Curve Families Across Integer Bases} Table \ref{tab:base_classification} summarizes how the algebraic and geometric curve families generated by $\eta_q(s) = (1 - q^{1-s})\zeta(s)$ depend on the arithmetic type of $q \ge 2$. \begin{table}[htbp] \centering \caption{Classification of algebraic and geometric curve families across integer bases $q \ge 2$.} \label{tab:base_classification} \footnotesize \setlength{\tabcolsep}{3.5pt} \begin{tabular}{@{}lllll@{}} \toprule \textbf{Base Type} & \textbf{$\mathbb{Q}[x,y]$ Lines} & \textbf{$n \neq q^k$ ($a \neq \frac{1}{2}$)} & \textbf{Powers $p_i^k$ ($p_i \mid q$)} & \textbf{Set $q \nmid n$} \\ \midrule Prime ($2, 3, 5$) & Unique $a = \frac{1}{2}$ & All 2D rosettes & In $\mathbb{Z}[x,y]$ tower & Completely mult. \\ Prime power ($4, 8, 9$) & $2m - 1$ lines $\frac{r}{2m}$ & Rosettes + 1D loops & Split across families & Coprime mult.\ only \\ Multi-prime ($6, 10, 12$) & $a = \frac{1}{2}$ (if non-power) & All 2D rosettes & Stranded 2D rosettes & Non-multiplicative \\ \bottomrule \end{tabular} \end{table} In Table \ref{tab:base_classification}, ``$\mathbb{Q}[x,y]$ Lines'' lists rational abscissas $a \in (0, 1) \cap \mathbb{Q}$ on which $t_{q,q^k}(a + ib)$ satisfies a rational polynomial equation, ``$n \neq q^k$ ($a \neq \frac{1}{2}$)'' classifies the paired curves $q_{q,n}(a + ib)$ off the critical line, ``Powers $p_i^k$ ($p_i \mid q$)'' tracks the fate of individual prime-factor powers $t_{q,p_i^k}(\frac{1}{2}+ib)$, and ``Set $q \nmid n$'' records the multiplicativity of $f_q(n) = \mathbf{1}_{q \nmid n}$. This trichotomy clarifies what is special about prime normalizations of $\zeta(s)$: primes are the unique integer bases $q \ge 2$ for which $(1 - q^{1-s})^{-1}$ simultaneously gathers all powers of $q$'s prime divisors into an irreducible $\mathbb{Z}[x, y]$ algebraic curve tower on $\operatorname{Re}(s) = \frac{1}{2}$, confines rational curve equations exclusively to $a = \frac{1}{2}$, forces every remaining paired term off $a = \frac{1}{2}$ to unravel into a two-dimensional rosette, and preserves multiplicative closure across the surviving integers. \begin{thebibliography}{9} \bibitem{Apostol1976} T.~M.~Apostol, \emph{Introduction to Analytic Number Theory}, Undergraduate Texts in Mathematics, Springer-Verlag, New York, 1976. \bibitem{ByrdFriedman1971} P.~F.~Byrd and M.~D.~Friedman, \emph{Handbook of Elliptic Integrals for Engineers and Scientists}, 2nd ed., Grundlehren der mathematischen Wissenschaften, vol.~67, Springer-Verlag, Berlin, 1971. \bibitem{Euler1749} L.~Euler, ``Remarques sur un beau rapport entre les s\'eries des puissances tant directes que r\'eciproques,'' \emph{M\'emoires de l'Acad\'emie des Sciences et Belles-Lettres de Berlin} \textbf{17} (1768), 83--106 (read 1749). \bibitem{Hackbarth2026} R.~Hackbarth, ``Characterizing the Critical Line via Normalized Dirichlet Eta Terms: Concentric Circles, a Bernoulli Lemniscate, and Integer Algebraic Curves,'' preprint, 2026, \url{https://zenodo.org/records/24325291}. \bibitem{HardyWright2008} G.~H.~Hardy and E.~M.~Wright, \emph{An Introduction to the Theory of Numbers}, 6th ed., Oxford University Press, Oxford, 2008. \bibitem{Needham1997} T.~Needham, \emph{Visual Complex Analysis}, Clarendon Press, Oxford University Press, Oxford, 1997. \bibitem{Titchmarsh1986} E.~C.~Titchmarsh, \emph{The Theory of the Riemann Zeta-Function}, 2nd ed., Clarendon Press, Oxford, 1986. \bibitem{Weyl1916} H.~Weyl, ``\"Uber die Gleichverteilung von Zahlen mod.\ Eins,'' \emph{Mathematische Annalen} \textbf{77} (1916), 313--352. \end{thebibliography} \end{document}