Inherited arcs in Andr\'e planes of even order
Extracted source text · arXiv 2606.14345
\documentclass[12pt]{article}%%%%
\usepackage{graphicx,amsmath,amssymb,amsthm,rotating,stmaryrd,datetime,pifont,booktabs,enumitem}
\usepackage[mathscr]{eucal}%Euler script, use \mathscr
\usepackage{setspace}%this stops lists having so much gap in them
\usepackage{enumitem}
\setlist{nolistsep}
\setlist{nosep}
% colours
\usepackage[usenames,dvipsnames]{color}
\newcommand\red[1]{{\color{red} #1}}
\newtheorem{theorem}{Theorem}
\newtheorem{thm}[theorem]{Theorem}
\newtheorem{result}[theorem]{Result}
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{corollary}[theorem]{Corollary}
\newtheorem{construction}[theorem]{Construction}
\newtheorem{conjecture}[theorem]{Conjecture}
\newtheorem{defn}[theorem]{Definition}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{question}[theorem]{\red{Question:}}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{note}[theorem]{Note}
\newtheorem{example}{Example}
\newtheorem{exercise}{Exercise}
%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%
%Page size
\setlength{\topmargin}{-1cm}
\setlength{\textheight}{21.8cm}
\setlength{\oddsidemargin}{-.5cm}
\setlength{\evensidemargin}{-.5cm}
\setlength{\textwidth}{17cm}
\setlength{\parindent}{0pt}
\setlength{\parskip}{.25cm}
\renewcommand\baselinestretch{1.1}
\newcommand{\new}{\nu}
\newcommand{\f}{f}
\newcommand{\g}{g}
\renewcommand{\O}{\mathcal O}
\newcommand{\A}{\mathcal A}
%\newcommand{\B}{\mathbf B}
\newcommand{\X}{\mathbb X}
\newcommand{\Y}{\mathbb Y}
\newcommand{\Cplus}{{\C^\textup{+}}}
\newcommand{\Q}{\mathcal{Q}}
\newcommand{\T}{\mathcal{T}}
\renewcommand{\H}{\mathcal{H}}
\renewcommand{\P}{\mathcal{P}}
\renewcommand{\S}{\mathcal{S}}
\renewcommand{\L}{\mathcal{L}}
\newcommand{\K}{\mathcal{K}}
\newcommand{\N}{\mathcal{N}}
\newcommand{\E}{\mathcal{E}}
\newcommand{\R}{\mathcal{R}}
\newcommand{\C}{\mathcal{C}}
\newcommand{\D}{\mathcal{D}}
\newcommand{\HH}{\mathcal{H}}
\newcommand{\li}{\ell_\infty}
\newcommand{\si}{\Sigma_\infty}
\newcommand{\B}{{\mathbf{B}}}
\newcommand{\F}{\mathcal{F}}
\newcommand{\PG}{{\textup{PG}}}
\newcommand{\AG}{{\textup{AG}}}
\newcommand{\PGL}{{\textup{PGL}}}
\newcommand{\PGammaL}{\textup{P}\Gamma{\textup {L}}}
\newcommand{\Fqq}{\mathbb{F}_{q^2}}
\newcommand{\Fqqq}{\mathbb{F}_{\hspace*{-1mm}{q^{3}}}}
\newcommand{\Fqn}{\mathbb{F}_{\hspace*{-1mm}{q^{n}}}}
\newcommand{\Fqt}{\mathbb{F}_{\hspace*{-1mm}{q^{t}}}}
\newcommand{\Fq}{\mathbb{F}_{\hspace*{-.5mm}q}}
\newcommand{\Fqts}{\mathbb{F}^*_{\hspace*{-0.5mm}{q^{t}}}}
\newcommand{\hyp}{\frac{q^t-1}{q-1}}
\DeclareMathOperator{\Nm}{N}
\begin{document}
\title{Inherited arcs in Andr\'e planes of even order
}
\author{S.G. Barwick, Alice M.W. Hui and Wen-Ai Jackson}
\date{}
\maketitle
AMS code: 51E20
Keywords: projective geometry, conics, inherited arcs, Andr\'e replacements, Andr\'e planes
%\begin{spacing}{0} \tableofcontents \end{spacing}
\begin{abstract} Andr\'e planes of order $q^t$ can be constructed from the Desarguesian plane $\PG(2,q^t)$ using a process that involves replacing Andr\'e nets. This article constructs ovals in Andr\'e planes of order $q^t$, $q$ even and $t$ prime that are inherited from conics of the Desarguesian plane $\PG(2,q^t)$. Further, it is shown that certain conics do not inherit to arcs under the process of Andr\'e replacement.
\end{abstract}
\section{Introduction}
The process of deriving the Desarguesian plane $\PG(2,q^2)$ replaces certain lines with Baer subplanes, and constructs the Hall plane. A well studied problem involves taking a conic in $\PG(2,q^2)$ and using the process of derivation to construct inherited arcs in the Hall plane. The effect of derivation on conics of $\PG(2,q^2)$ has been completely determined, with contributions by a number of authors, see \cite{barmar}, \cite{BKNS}, \cite{Cher}, \cite{GlynnStein}, \cite{Korch1}, \cite{Korch2}, \cite{OKeefePasPen}, \cite{OKeefePas}, \cite{Szon}.
In this article we are interested in the case when $q$ is even, and the complete results when $q$ is even are as follows.
\begin{result}\label{120e}
Let $\C$ be a non-degenerate conic in $\PG(2,q^2)$, $q$ even, $q>2$, and $\D$ a Baer subline of $\li$. Derive $\PG(2,q^2)$ using $\D$ to get the Hall plane $\H(\D)$. The affine points of $\C$ correspond to a set of affine points of $\H(\D)$ denoted by $\D(\C)$.
\begin{enumerate}
\item \cite{Cher,OKeefePas} If $\C$ is secant to $\li$, then $\D(\C)$ is not an arc of the Hall plane.
\item \cite{Cher} If $\C$ is exterior to $\li$, then $\D(\C)$ is not an arc of the Hall plane.
\item Suppose $\C$ is tangent to $\li$ at the point $T$ and $\C$ has nucleus $N$.
\begin{enumerate}
\item \cite{OKeefePas} If $T,N\in\D$, then $\D(\C)$ is not an arc of the Hall plane.
\item \cite{OKeefePasPen} If exactly one of $T,N$ lies in $\D$ then $\D(\C)$ is an arc of the Hall plane.
\item \cite{GlynnStein} If $T,N\notin\D$, then $\D(\C)$ is an arc of the Hall plane if and only if $q$ is a square and $T,N$ are
conjugate with respect to $\D$.
\end{enumerate}
\end{enumerate}
\end{result}
The process of deriving $\PG(2,q^2)$ using a Baer subline of $
\li$ to get the Hall plane can be generalised to performing an Andr\'e replacement of $\PG(2,q^t)$, $t\geq3$, using an Andr\'e set of $\li$ to get an Andr\'e plane of order $q^t$.
This process of Andr\'e replacement is described in Section \ref{sec2}. Note that given a Baer subline of $\li$ in $\PG(2,q^2)$, there is a unique way to derive $\PG(2,q^2)$ to get the Hall plane. In $\PG(2,q^t)$, $t\geq3$, given an Andr\'e set of $\li$, there are $t-1$ different Andr\'e replacement nets, leading to $t-1$ different ways to perform an Andr\'e replacement and construct an associated Andr\'e plane. This means that when studying whether conics inherit to arcs when $t\geq3$, we need to consider all of the $t-1$ associated Andr\'e planes.
The main results of this article are Theorems~\ref{thm000}, \ref{201}, and \ref{014} which generalise Result~\ref{120e} to partially answer when conics in $\PG(2,q^t)$, $q$ even, $t\geq3$ prime, are inherited to arcs in Andr\'e planes.
In particular, Theorem~\ref{thm000} generalises parts 1 and 2 of Result \ref{120e} to show that conics of $\PG(2,q^t)$, $q$ even, $t$ a prime, that are secant or exterior to $\li$ never inherit to arcs of an Andr\'e plane. When $\C$ is tangent to $\li$ there are several cases to consider. Theorem \ref{201} generalises part 3a) of Result \ref{120e} to $\PG(2,q^t)$, $q$ even, $t$ a prime, and we never get an inherited arc in this case. Theorem \ref{014} generalises part 3c) of Result \ref{120e} to give an example of an inherited arc -- the result for this case is more complex when $t\geq3$ than the $t=2$ case, and the generalisation is not straightforward.
This leaves several open cases which are discussed in Section~\ref{sec-con}, in particular the case of an inherited arc given in Result \ref{120e}(3b) when $t=2$ is conjectured to not give an inherited arc when $t\geq3$.
\section{Background}\label{sec2}
The finite field of prime power order $q$ is denoted $\Fq$, and we let $\Fq^*=\Fq\setminus\{0\}$. The \emph{norm mapping} from $\Fqt$ to $\Fq$ is denoted $\Nm (k)=k^{q^{t-1}+\cdots+q+1}$ for $k\in\Fqt$.
The Bruck-Bose representation \cite{andr54,bruc69,bruc64,segre} of $\PG(2,q^t)$ in $\PG(2t,q)$ is well known. Let $\si$ be the hyperplane at infinity in $\PG(2t,q)$ and let $\S$ be a $(t-1)$-spread in $\si$. The incidence structure $\A(\S)$ with points the points of $\PG(2t,q)\setminus\si$; lines the $t$-spaces of $\PG(2t,q)\setminus\si$ that contain an element of $\S$; and incidence being inclusion, is an affine plane. We can complete this to a projective plane denoted $\P(\S)$: points on the line at infinity $\li$ correspond to the elements of $\S$. Moreover, $\P(\S)\cong \PG(2,q^t)$ iff $\S$ is a regular (Desarguesian) spread.
If $\mathcal X$ is a set of points of $\PG(2,q^t)$, then we denote the corresponding set of points in $\PG(2t,q)$ by $[\mathcal X]$.
The representation of $\li\cong\PG(1,q^t)$ as a regular $(t-1)$-spread in $\si\cong\PG(2t-1,q)$ is often referred to as field reduction, and denoted by $F_{2,t,q}$. Coordinates for this will be denoted as follows. If $P=(x,y)\in\PG(1,q^t)$, then the corresponding $(t-1)$-space in $\PG(2t-1,q)$ is denoted $[P]=F_{2,t,q}(P)$ and has coordinates written as $[P]=\{\langle (sx,sy)\rangle_q\,|\,s\in\Fqt^*\}$.
An Andr\'e set of $\PG(1,q^t)$, $t\geq3$, can be constructed as follows.
Let $E,F$ be two distinct points of $\PG(1,q^t)$ and let $G$ be the collineation group acting on points of $\PG(1,q^t)$ that fixes $E$ and $F$ pointwise. As $G$ is Singer group of order $q^t-1$, $G$ has a unique Singer subgroup $H$ of order $q^{t-1}+\cdots+q+1$. The orbits of $H$ partition the points of $\PG(1,q^t)\setminus\{E,F\}$ into $q-1$ sets of size $q^{t-1}+\cdots+q+1$,
these sets are called \emph{Andr\'e sets} and $E,F$ are called
the \emph{transversal points} of the Andr\'e set.
Andr\'e sets are defined in multiple settings, and a short discussion of the different settings and different nomenclature is given in \cite{qoddpaper}. In particular, in the linear set literature, Andr\'e sets are scattered $\Fq$-linear sets of $\PG(1,q^t)$ of pseudoregulus type.
We note that an Andr\'e set of $\PG(1,q^t)$, $t\geq3$, is projectively equivalent to $$
\A_1=\{(1,\theta)\,|\,k\in\Fqt^*,\Nm(\theta)=1\}$$ which has transversal points $(1,0)$, $(0,1)$.
Moreover the $q-1$ Andr\'e sets with transversal points $(1,0)$, $(0,1)$ are the sets $\A_\delta$, $\delta\in\Fq^*$ with coordinates given by
$$
\A_\delta=\{(1,\theta)\,|\,k\in\Fqt^*,\Nm (\theta)=\delta\}.
$$
Under field reduction, the Andr\'e set $
\A_\delta=\{(1,\theta)\,|\,k\in\Fqt^*,\Nm (\theta)=\delta\}
$ of $\PG(1,q^t)$ corresponds to a set $[\A_\delta]$ of $\hyp$ elements of the regular (Desarguesian) $(t-1)$-spread $\S$ in $\PG(2t-1,q)$.
This set is called an \emph{Andr\'e net} and has coordinates:
$$ [\A_\delta] =\{[P_k]=\{\langle(s,sk)\rangle_q\,|\, s\in\Fqt^*\}\ \,|\,k\in\Fqt^*, \Nm(k)=\delta\}.$$
In $\PG(2t-1,q)$, there are $t-1$ \emph{Andr\'e replacement nets} for $[\A_\delta]$, denoted $\A_\delta^m$, $m=1,\ldots,{t-1}$. Coordinates for these nets are given in multiple places, see \cite{qoddpaper} for a short survey and for a discussion on why we only consider the case $t$ prime.
The coordinates for the Andr\'e replacement nets are given by the following:
\begin{equation}
\label{e2}
\A_\delta^m=\{X_{m,u}=\{\langle(s,s^{q^m}u)\rangle_q\,|\, s\in\Fqt^*\}\ \,|\,u\in\Fqt^*, \Nm(u)=\delta\}.
\end{equation}
We can form a new spread of $\PG(2t-1,q)$ from the regular $(t-1)$-spread $\S$, namely the spread $(\S\setminus [\A_\delta])\cup\A_\delta^m$ for some $m\in\{1,\ldots,t-1\}$.
Recall that as $\S$ is regular, the Bruck-Bose plane $\P(\S)\cong \PG(2,q^t)$ is Desarguesian. The plane $\P\big((\S\setminus [\A_\delta])\cup\A_\delta^m\big)$ is an Andr\'e plane, we abbreviate the notation and write $\P(\A_\delta^m)=\P\big((\S\setminus [\A_\delta])\cup\A_\delta^m\big)$. The process of constructing $\P(\A_\delta^m)$ from $\P(\S)\cong \PG(2,q^t)$ is called \emph{Andr\'e replacement}, and the next statement carefully describes it (using a generic notation for the Andr\'e set).
\begin{construction}\label{cons1} {\bfseries \emph{(Andr\'e replacement)}} In $\PG(2,q^t)$, $t$ prime, $t\geq 3$, let $\D$ be an Andr\'e set of $\li$. In the $\PG(2t,q)$ Bruck-Bose setting, $[\D]$ is an Andr\'e net of the regular spread $\S\subset\si$, denote the $t-1$ Andr\'e replacement nets by $\D^m$, $m=1,\ldots,t-1$.
For $m\in\{1,\ldots,t-1\}$, define the incidence structure $\mathcal T_m$ as follows.
\begin{itemize}
\item The points of $\mathcal T_m$ are the
points of $\AG(2,q^t)$
\item The lines of $\mathcal T_m$ consist of
\begin{itemize}
\item the lines of $\AG(2,q^t)$ that meet $\li$ in a point not in $\D$,
\item the \emph{$\D^m$-blocks} (a set of affine points $\mathcal X$ in $\PG(2,q^t)$ is called an \emph{$\D^m$-block} if in $\PG(2t,q)$, the corresponding set $[\mathcal X]$ is an affine $t$-space whose projective completion meets $\si$ in a $(t-1)$-space that lies in the replacement net $\D^m$).
\end{itemize}
\item Incidence is inclusion
\end{itemize}
Then $\mathcal T_m$ is an affine plane of order $q^t$ whose projective completion is an Andr\'e plane denoted $\P(\D^m)$. This process is called an \emph{Andr\'e replacement with respect to the Andr\'e replacement net $\D^m$. }
\end{construction}
Recall that a $(t-1)$-regulus in $\PG(2t-1,q)$ is a set $\R$ of $q+1$ pairwise disjoint $(t-1)$-spaces, with the property that if a line meets three elements of $\R$, then it meets all elements of $\R$. Hence a $(t-1)$-regulus is ruled by a set of $q^{t-1}+\cdots+q+1$ (pairwise disjoint) lines.
It is well known that if $b$ is an $\Fq$-line contained in $\li$ in $\PG(2,q^t)$, then in the $\PG(2t,q)$ Bruck-Bose representation, $[b]$ is a $(t-1)$-regulus contained in the regular spread $\S$. Conversely, every $(t-1)$-regulus contained in $\S$ corresponds to an $\Fq$-line of $\li$.
When $t$ is prime, the $\Fq$-lines contained in an Andr\'e set $\D$ can be partitioned into $t-1$ families. We need the following relationship between the $t-1$ families of $\Fq$-lines and the $t-1$ Andr\'e replacement nets for $\D$; this key result motivates why we only consider the case when $t$ is prime.
\begin{result}\cite{qoddpaper} \label{M110}
Let $\D$ be an Andr\'e set of $\PG(1,q^t)$, $t\geq3$ prime, $q\geq t$.
\begin{enumerate}
\item\label{M110a} The $\Fq$-lines contained in $\D$ can be partitioned into $t-1$ families of size $(q^{t-1}+\ldots+q+1)(q^{t-1}-1)/(q^2-1)$. We can label the families by $\F_1,\ldots,\F_{t-1}$ so that for $i\in\{1,\ldots,t-1\}$, family $\F_i$ corresponds to
the Andr\'e replacement net $\D^i$ in the following way.
\begin{enumerate}
\item If $\ell$ is a line contained in one of the $(t-1)$-spaces in $\D^i$, then the set $\{P\in\li\,|\,[P]\cap\ell\neq\emptyset\}$ is an $\Fq$-line belonging to family $\F_i$.
\item If
$b$ is an $\Fq$-line in family $\F_i$, then every ruling line of the $(t-1)$-regulus $[b]$ is contained in a unique
$(t-1)$-space of $\D^i$.
\end{enumerate}
\item\label{M110b} Any two distinct points of $\D$ lie in exactly $t-1$ $\Fq$-lines of $\D$, one from each family.
\end{enumerate}
\end{result}
These families of $\Fq$-lines are critical to studying inherited arcs due to the following relationship with the $\D^m$-blocks defined in Construction~\ref{cons1}.
\begin{result}\cite{qoddpaper}\label{001}
In $\PG(2,q^t)$, $q\geq t$, $t\geq3$ prime, let $\D$ be an Andr\'e set of $\li$, and denote the families of $\Fq$-lines contained in $\D$ by $\F_1,\ldots,\F_{t-1}$. Three non-collinear affine points $A,B,C$ lie in a common $\D^m$-block for some $m\in\{1,\ldots,t-1\}$ if and only if the $\Fq$-line determined by the three points $AB\cap\li$, $AC\cap\li$, $BC\cap\li$ belongs to family $\F_m$.
\end{result}
%%%%%%%%%%%%
%%%%%%
%%%%%%
%%%%%%
%%%%%%
\section{Secant and exterior conics}
In this section we look at a conic which is either secant or exterior to $\li$ in $\PG(2,q^t)$, $q$ even, and show that an Andr\'e replacement never results in an inherited arc.
The article
\cite{BKNS} showed how powerful the following result is in the case when $t=2$, that is, when deriving $\PG(2,q^2)$ and looking at which conics inherit to arcs of the Hall plane. It is also powerful when $t\geq3$ and can be used to look at the cases when a conic is secant or exterior to $\li$.
\begin{result}\label{result000} \cite{segrekorch}
%(See also Proposition 3.2 of \cite{BKNS})
In $\PG(2,q)$, $q$ even, let $\C$ be a conic and $\ell$ a line that is not tangent to $\C$. For any triple $\{P_1 , P_2 , P_3 \}$ of points contained in $\ell\setminus\C$, there is exactly one triangle $\{A_1,A_2,A_3\}$ inscribed in $\C \setminus \ell$ such that $A_i A_j \cap \ell = P_k$, where $i, j, k$ is a permutation of $1, 2, 3$.
\end{result}
\begin{theorem}\label{thm000} In $\PG(2,q^t)$, $q$ even, $q\geq t$, $t\geq3$ prime, let $\C$ be a conic and $\D$ be an Andr\'e set of $\li$. Perform an Andr\'e replacement using the Andr\'e replacement net $\D^m$, and let $\D_m(\C)$ denote the set of points in $\P(\D^m)$ corresponding to the affine points of $\C$. If $\C$ is a conic that is either exterior or secant to $\li$, then $\D_m(\C)$ is not an arc in the Andr\'e plane $\P(\D^m)$ for any $m\in\{1,\dots,t-1\}$.
\end{theorem}
\begin{proof} Let $\C$ be a conic that is exterior of secant to $\li$.
Let $m\in\{1,\ldots,t-1\}$ and let $b$ be an $\Fq$-line in $\D$ belonging to family $\F_m$. Let $P_1,P_2,P_3$ be three points in $b$ which do not lie in $\C$. By Result \ref{result000}, there are points $A_1,A_2,A_3 \in \C$ such that $A_iA_j \cap l_\infty = P_j$ for any $\{i,j,k\}=\{1,2,3\}$. By Result~\ref{001}, the points $A_1,A_2,A_3$ lie in a $\D^m$-block. This $\D^m$-block corresponds to a line in $\P(\D^m)$ that meets $\D_m(\C)$ in three points, so $\D_m(\C)$ is not an arc.
\end{proof}
\section{Tangent conics}
In this section we look at a conic in $\PG(2,q^t)$ which is tangent to $\li$ and try to determine whether an Andr\'e replacement leads to an inherited arc.
Let $\C$ be a conic in $\PG(2,q^t)$ tangent to $\li$ at the point $T$ with nucleus $N$, and let
$\D$ be an Andr\'e set with transversal points $E,F$. There are a number of cases to consider, depending on the intersection of $\C\cup N$ and the set $\D\cup \{E,F\}$. In this section we consider the following two cases.
Firstly, Section \ref{tgt-sec1} looks at the case when both $T,N\in\D$ and shows that in this case we never get an inherited arc -- this directly generalises Result \ref{120e}(3a).
Section \ref{tgt-sec2} looks at the case when $\{T,N\}=\{E,F\}$, (that is, where $\{T,N\}$ are the transversal points of $\D$) and shows that in most cases we get an inherited arc. The main result is stated in Theorem~\ref{014} -- this result generalises Result \ref{120e}(3c), although the generalisation is not an obvious one.
\subsection{Tangent conics - the case where $T,N\in\D$}\label{tgt-sec1}
\begin{theorem}\label{201} In $\PG(2,q^t)$, $q$ even, $q\geq t$, $t\geq3$ prime, let $\C$ be a conic and $\D$ be an Andr\'e set of $\li$. Perform an Andr\'e replacement using the Andr\'e replacement net $\D^m$, and let $\D_m(\C)$ denote the set of points in $\P(\D^m)$ corresponding to the affine points of $\C$. If $\C$ is a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$, with $T,N\in\D$, then $\D_m(\C)$ is not an arc in the Andr\'e plane $\P(\D^m)$ for any $m\in\{1,\dots,t-1\}$.
\end{theorem}
\begin{proof}
Let $m\in\{1,\ldots,t-1\}$.
By Result~\ref{M110},
there is a unique $\Fq$-line $b$ contained in family $\F_m$ that contains the two points $T,N$. Let $X$ be a point of $b$ distinct from $T,N$ and let $\ell$ be a $2$-secant of $\C$ through the point $X$, with $\ell\cap\C=\{A,B\}$.
The term $\Fq$-conic contained in $\C$ refers to a set of $q+1$ points of $\C$ that are projectively equivalent to a non-degenerate conic contained in $\PG(2,q)$. As $\C$ is projectively equivalent to the set $\{(\theta^2,\theta,1)\,|\,\theta\in\Fqt\cup\{\infty\}\}$, it follows that any three points of $\C$ are contained in a unique $\Fq$-conic contained in $\C$.
Let $\O$ be the unique $\Fq$-conic contained in $\C$ that contains the
three points $T,A,B$.
As $q\geq4$, $\O$ lies in a unique $\Fq$-plane which we denote by $\alpha$.
As $T\in\O$, $\O$ has nucleus $N$, so $N\in\alpha$. Hence $\alpha$ is secant to $\li$ and so $\alpha\cap\li=b$. As $q\geq4$, $\O$ contains a point $C$ distinct from $T,A,B$. The three points $AB\cap\li$, $AC\cap\li$, $BC\cap\li$ lie in $b$. It follows from Result~\ref{001} that $A,B,C$ lie in a common $\D^m$-block, and so they are collinear in $\P(\D^m)$. Hence $\D_m(\C)$ is not an arc in $\P(\D^m)$.
\end{proof}
\subsection{Tangent conic - the case where $\{T,N\}$ are the transversal points of $\D$}\label{tgt-sec2}
\subsubsection{Preliminaries}
In what follows, let $\tau$ be a multiplicative generator for $\Fqt^*$.
Recall from Section~\ref{sec2} that the $q-1$ Andr\'e sets
with transversal points $(1,0),(0,1)$ are projectively equivalent to
$\A_\delta=\{(1,\theta)\,|\,\theta\in\Fqts,\Nm(\theta)=\delta\}$, $\delta\in\Fq^*$. We rewrite this in a different form.
If $\theta\in\Fqts$, with $ \Nm(\theta)=\delta$, then $\theta=\new \tau^{(q-1)i}$ for some $\nu\in\Fqts$ with $\Nm(\nu)=\delta$ and $i\in I$ where
$$I=\{0,1,\ldots,q^{t-1}+\cdots+q\}$$ is an index set.
So the $q-1$ Andr\'e sets
with transversal points $(1,0),(0,1)$ are projectively equivalent to sets $\A_\new$, $\new\in\Fqt^*$ where
$\A_{\new}=\{(1,\new\tau^{(q-1)i})\,|\,i\in I \}.$
For the remainder of this section, we let $\D$ be an Andr\'e set with transversal points $T,N$, and write
\begin{equation}\label{e5}\D=\A_{\new}=\{(1,\new\tau^{(q-1)i})\,|\,i\in I \} \quad\textup{for some }\new\in\Fqt^*.\end{equation}
The $t-1$ Andr\'e replacement nets for $\D=\A_\new$ are denoted $\D^m$, $m=1,\ldots, t-1$; the coordinates can be calculated using (\ref{e2}), giving
\begin{equation}\label{e6}
\D^m=\{ X_{m,u}=\{\langle(s,s^{q^m}u)\rangle_q\,|\, s\in\Fqt^*\}\ \,|\,u\in\Fqt^*, \Nm(u)=\Nm(\new)\}.
\end{equation}
%\gray{OR can write as
%$$\A^m_\nu=\{ X_{m,k}=\{\langle(s,s^{q^m}\nu k )\rangle_q\,|\, s\in\Fqt^*\}\ \,|\,k\in\Fqt^*, \Nm(k)=1\}$$}
%}
%
%
%\gray{
%OR there exists unique $d\in\Fq$ with $\delta=\tau^d$, so if $\theta\in\Fqts$ with $ \Nm(\theta)=\delta$, then
%$\theta=\tau^{(q-1)i+d}$ for some $d\in\{1,\ldots,q-1\}$, $i\in I $
% So we can write $\A_\delta$ as $\A_d$ where
%$$\A_d=\{(1,\tau^{(q-1)i +d},0)\,|\,i\in I \}.$$
%}
\begin{lemma}\label{002}
In $\PG(2,q^t)$, $q$ even, $t\geq3$ prime, let $\C$ be a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$ with $\{T,N\}=\{(1,0,0),(0,1,0)\}$. Let
$\D$ be an Andr\'e set with transversal points $T,N$.
\begin{enumerate}
\item If there exists three affine points $A,B,C$ in $\C$ with $AB\cap\li$, $AC\cap\li$, $BC\cap\li$ in $\D$, that is $AB\cap \li=(1,\new \tau^{(q-1)i},0)$, $AC\cap \li=(1,\new \tau^{(q-1)j},0)$, $BC\cap \li=(1,\new \tau^{(q-1)k},0)$ for distinct $i,j,k\in I $, then
$
\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0$.
\item If there exists distinct $i,j,k\in I $ satisfying $
\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0$, then there exists $A,B,C\in\C$ with
$AB\cap \li=(1,\new \tau^{(q-1)i},0)$, $AC\cap \li=(1,\new \tau^{(q-1)j},0)$, $BC\cap \li=(1,\new \tau^{(q-1)k},0)$ which lie in $\D$.
\end{enumerate}
\end{lemma}
\begin{proof} First suppose that $T=(0,1,0)$ and $N=(1,0,0)$, so $\C$ has equation $\f x^2+\g z^2+yz=0$ for some $\f,\g\in\Fqt$, $\f\neq0$.
By (\ref{e5}), if $\D$ is an Andr\'e set with transversal points $(1,0,0),(0,1,0)$, then
$\D=\{(1,\new\tau^{(q-1)i},0)\,|\,i\in I \}$ for some $\new\in\Fqts$.
For part 1, suppose there exists three affine points $A,B,C$ in $\C$ such that the three points $AB\cap\li$, $AC\cap\li$, $BC\cap\li$ lie in $\D$. The points $A,B,C$ have coordinates $A=(a, a^2 \f+\g,1)$, $B=(b, b^2 \f +\g,1)$, $C=(c, c^2 \f +\g,1)$ for some distinct $a,b,c\in\Fqt$. We compute the following coordinates:
$P_1=AB\cap\li=(1,(a+b) \f,0)$,
$P_2=AC\cap\li=(1,(a+c) \f,0)$,
$P_3=BC\cap\li=(1,(b+c) \f,0)$.
So if the three distinct points $P_1,P_2,P_3$ lie in $\D$, then there exists distinct $i,j,k\in I $ with $(a+b)\f=\new \tau^{(q-1)i}$,
$(a+c)\f= \new\tau^{(q-1)j}$,
$(b+c)\f= \new \tau^{(q-1)k}$. As $q$ is even, summing these three equations gives $\new\tau^{(q-1)i}+\new\tau^{(q-1)j}+\new\tau^{(q-1)k}=0$ and so $\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0$ as required.
For part 2, suppose there exists distinct $i,j,k\in I $ satisfying $\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0.$ Let $c\in\Fqt$ and let
\begin{equation}\label{e07}
a=\new\tau^{(q-1)j} \f^{-1}-c,\quad
b=\new\tau^{(q-1)k} f^{-1} -c.
\end{equation}
Then $a,b,c$ are distinct, so the points with coordinates $C=(c, c^2 \f+\g,1)$, $A=(a, a^2 \f +g,1)$, $B=(b, b^2 \f +g,1)$ are distinct affine points of $\C$. Computing coordinates and using (\ref{e07}) we have $AC\cap\li=(1,(a+c)\f,0)=(1,\new\tau^{(q-1)j},0)$ and $BC\cap\li=(1,(b+c)\f,0)=(1,\new\tau^{(q-1)k},0)$. Further $AB\cap\li=(1,(a+b)\f,0)=(1,\new \tau^{(q-1)j}+\new\tau^{(q-1)k}-2c,0)=(1,\new\tau^{(q-1)i},0)$ as $q$ even and $\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0.$
The case where $T=(1,0,0)$, $N=(0,1,0)$ is similar. In this case the conic $\C$ has equation $\f y^2+\g z^2+xz=0$, for some $\f,\g\in\Fqt$, $\f\neq0$, so for example, $A=( y^2 \f+\g,y,1)$. If $\D$ is an Andr\'e set with transversal points $(1,0,0),(0,1,0)$, then we can arrange coordinates to write $\D$ as
$\D=\{(\new\tau^{(q-1)i},1,0)\,|\,i\in I \}$ for some $\new\in\Fqts$. The computations are now similar to the above case.
\end{proof}
\begin{lemma}\label{003} In $\PG(2,q^t)$, $q$ even, $q\geq t$, $t\geq3$ prime, let $\D$ be an Andr\'e set with transversal points $(1,0,0)$, $(0,1,0)$. Denote the families of $\Fq$-lines contained in $\D$ by $\F_1,\ldots,\F_{t-1}$ and let $m\in\{1,\ldots,t-1\}$.
Suppose $P_1=(1,\new \tau^{(q-1)i},0)$, $P_2=(1,\new \tau^{(q-1)j},0)$, $P_3=(1,\new \tau^{(q-1)k},0)$ are three distinct points in $\D$ with $i,j,k\in I $ satisfying
$
\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0.
$
The $\Fq$-line determined by $P_1,P_2,P_3$ belongs to family $\F_m$ if and only if
$q^t-1$ is a factor of both $(q-1)(q^m-2)(j-i)$ and $(q-1)(q^m-2)(k-i)$.
\end{lemma}
\begin{proof} As in (\ref{e5}), we can write $\D=
\{(1,\new\tau^{(q-1)i},0)\,|\,i\in I \}$ for some $\nu\in\Fqts$.
Let
$P_1=(1,\new\tau^{(q-1)i},0)$, $P_2=(1,\new\tau^{(q-1)j},0)$, $P_3=(1,\new\tau^{(q-1)k},0)$ be three distinct points in $\D$ with (distinct) $i,j,k\in I$ satisfying
$
\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0$.
Let $\C$ be a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$ with $\{T,N\}=\{(1,0,0),(0,1,0)\}$.
Then
by Lemma~\ref{002}, there exists points $A$, $B$, $C$ in $\C$ with $P_1=AB\cap \li$,
$P_2=AC\cap \li$ and $P_3=BC\cap \li$.
We prove the result for the case where $T=(0,1,0)$ and $N=(1,0,0)$.
The case where $T=(1,0,0)$, $N=(0,1,0)$ is discussed in the last paragraph of this proof. As $T=(0,1,0)$ and $N=(1,0,0)$, $\C$ has equation $\f x^2+\g z^2+yz=0$ for some $\f,\g\in\Fqt$, $\f\neq0$. So $A=(a,a^2 \f +\g,1)$, $B=(b, b^2 \f+\g,1)$, $C=(c, c^2 \f+\g,1)$ for some distinct $a,b,c\in\Fqt$.
We compute $P_1=AB\cap\li=(1, (a+b)\f,0)$, and equate with $P_1=(1,\new\tau^{(q-1)i},0)$ to get
\begin{equation}\label{e08}
a+b=\new\tau^{(q-1)i} \f^{-1}.
\end{equation}
Similarly $a+c= \new\tau^{(q-1)j}\f^{-1}$ ,
$b+c=\new \tau^{(q-1)k}\f^{-1}$.
Before continuing, we introduce some coordinate notation. Section \ref{sec2} introduced the coordinate notation where a point $P=(x,y)\in\PG(1,q^t)$ corresponds to the $(t-1)$-space $[P]$ with coordinates written as $[P]=\{\langle (sx,sy)\rangle_q\,|\,s\in\Fqt^*\}$. We generalise this to a coordinate notation for affine points in the Bruck-Bose setting of $\PG(2,q^t)$ in $\PG(2t,q)$. An affine point in $\PG(2,q^t)$ has coordinates that can be written as $Q=(x,y,1)$, the corresponding affine point in $\PG(2t,q)$ has coordinates denoted by $[Q]=(\langle x
\rangle_q,\langle y\rangle_q,1)$.
In $\PG(2t,q)$, $A,B,C$ correspond to the affine points with coordinates $[A]=(\langle a\rangle_q,\langle a^2\f +\g\rangle_q ,1)$, $[B]=(\langle b\rangle_q,\langle b^2 \f+\g \rangle_q ,1)$, $[C]=(\langle c\rangle_q,\langle c^2 \f+\g\rangle_q ,1)$.
The point $Q_1=[A][B]\cap\si$ has coordinates $Q_1=(\langle a\rangle_q+\langle b\rangle_q,\langle a^2 \f +\g\rangle_q+\langle b^2 \f+\g \rangle_q,0)=(\langle a+ b\rangle_q,\langle (a+ b)^2 \f\rangle_q,0)$ (as $q$ even). Similarly
$Q_2=[A][C]\cap\si=(\langle a+ c\rangle_q,\langle (a+ c)^2 \f\rangle_q,0)$,
$Q_3=[B][C]\cap\si=(\langle b+ c\rangle_q,\langle (b+ c)^2 \f\rangle_q,0)$.
Let $m\in\{1,\ldots,t-1\}$ and consider the Andr\'e replacement net $\D^m$.
The point $Q_1=(\langle a+ b\rangle_q,\langle (a+ b)^2 \f \rangle_q,0)$ lies in the spread element corresponding to the point $P_1=AB\cap\li$. As $P_1\in\D$, the point $Q_1$ lies in a unique $(t-1)$-space of replacement net $\D^m$. That is, recalling
coordinates from (\ref{e6}), there exists $u_1\in\Fqt^*$ with $ \Nm(u_1)=\Nm(\new)$
such that
$Q_1\in X_{m,u_1}=\{(\langle s\rangle_q,\langle s^{q^m}u_1\rangle_q,0)\,|\, s\in\Fqt^*\}$.
So for some $s\in\Fqts$ and $d\in\Fq^*$, we have
$d(\langle a+ b\rangle_q,\langle (a+ b)^2 \f\rangle_q,0)=(\langle s\rangle_q,\langle s^{q^m}u_1\rangle_q,0)$.
Equating the first coordinate we have
$d\langle a+ b\rangle_q=\langle s\rangle_q$, and so as $d\in\Fq^*$, we have $s=d(a+b)$.
Equating the second coordinate gives $d\langle (a+ b)^2 \f \rangle_q=\langle s^{q^m}u_1\rangle_q$ and so $d(a+ b)^2 \f=s^{q^m}u_1$. Substituting $s=d(a+b)$ and rearranging gives $u_1^{-1}=d^{q^m-1}(a+b)^{q^m-2} \f^{-1}$. As $d\in\Fq^*$, we have $d^{q^m-1} =1$ and so $u_1^{-1}=(a+b)^{q^m-2}\f^{-1}$.
Using (\ref{e08}) gives $u_1^{-1}=\new^{q^m-2}\tau^{i(q-1)(q^m-2)} \f^{1-q^m}$.
Similarly, $Q_2$ lies in a unique element of $\D^m$, so for some $u_2\in\Fqts$ with $\Nm(u_2)=\Nm(\new)$, we have $Q_2\in X_{m,u_2}$ and similar calculations give
\begin{equation}\label{e031}
u_2^{-1}=\new^{q^m-2}\tau^{j(q-1)(q^m-2)} \f^{1-q^m}.
\end{equation}
Similarly, $Q_3$ lies in a unique element of $\D^m$, so for some $u_3\in\Fqts$ with $\Nm(u_3)=\Nm(\new)$, we have $u_3^{-1}=\new^{q^m-2}\tau^{k(q-1)(q^m-2)}\f^{1-q^m}$.
We now prove the forward direction of the result. Suppose
the unique $\Fq$-line containing $P_1,P_2,P_3$ (denoted by $b$) lies in family $\F_m$. Let $\pi$ be the unique $\Fq$-plane that is secant to $\li$ and contains $A,B,C$, then $\pi$ meets $\li$ in the $\Fq$-line $b$. In $\PG(2t,q)$, $[\pi]$ is a plane that meets $\si$ in a line $\ell$. Further, $\ell$ meets every element of the $(t-1)$-regulus $[b]$, so $\ell$ is a ruling line of this $(t-1)$-regulus. As $b$ belongs to family $\F_m$, by Result~\ref{M110}, the line $\ell$ belongs to one $(t-1)$-space of the replacement net $\D^m$, that is, $\ell$ is contained $X_{m,u}$ for some $u\in\Fqts$ with $\Nm(u)=\Nm(\new)$.
Further the points $[A], [B], [C]$ lie in $[\pi]$, so $Q_1,Q_2,Q_3\in\ell$. That is, $Q_1,Q_2,Q_3$ lie in the same $(t-1)$-space of $\D^m$, and so $u=u_1=u_2=u_3$.
As $u_1=u_2$, we have $u_1^{-1}=u_2^{-1}$ and so
$\new^{q^m-2}\tau^{i(q-1)(q^m-2)}\f^{1-q^m}=\new^{q^m-2}\tau^{j(q-1)(q^m-2)}\f^{1-q^m}$
(see (\ref{e031})). Hence
$\tau^{(j-i)(q-1)(q^m-2)}=1$ and so $q^t-1$ is a factor of $(j-i)(q-1)(q^m-2)$. Similarly $u_1=u_3$ implies $q^t-1$ is a factor of $(k-i)(q-1)(q^m-2)$ as required.
We now prove the converse. Suppose that $q^t-1$ is a factor of both $(j-i)(q-1)(q^m-2)$ and $(k-i)(q-1)(q^m-2)$.
As $q^t-1$ is a factor of $(j-i)(q-1)(q^m-2)$, there exists $e\in\mathbb Z$ with $e(q^t-1)=(j-i)(q-1)(q^m-2)$ and so $j(q-1)(q^m-2)=i(q-1)(q^m-2)+e(q^t-1)$. From (\ref{e031}), we have
$u_2^{-1}=\new^{q^m-2}\tau^{j(q-1)(q^m-2)} \f^{1-q^m}$ and so
$u_2^{-1}=\new^{q^m-2} \tau^{i(q-1)(q^m-2)\f^{1-q^m} +e(q^t-1)}=\new^{q^m-2}\tau^{i(q-1)(q^m-2)}\f^{1-q^m}=u_1^{-1}$.
Hence $u_1=u_2$. Similarly we have $u_1=u_3$. That is, $Q_1,Q_2,Q_3$ lie in a common $(t-1)$-space $X_{m,u_1}$ of $\D^m$.
The plane $\alpha$ spanned by $[A], [B],[C]$ meets $\si$ in a line $\ell$ which contains the three points $Q_1,Q_2,Q_3$, so $\ell\subset X_{m,u_1}$.
In $\PG(2,q^t)$, the corresponding set of points is denoted $\B(\alpha)$, that is
$\B(\alpha)=\{P\in\PG(2,q^t)\,|\, [P]\cap\alpha\neq\emptyset\}$.
So $\B(\alpha)$ is the unique $\Fq$-plane that is secant to $\li$ and contains $A,B,C$.
Further $\B(\alpha)\cap\li$ is the $\Fq$-line denoted $\B(\ell)$ where $\B(\ell)=\{P\in\PG(2,q^t)\,|\, [P]\cap\ell\neq\emptyset\}$.
The points $P_1=AB\cap \li$, $P_2=AC\cap \li$, $P_3=BC\cap \li$ lie in $\B(\alpha)$, so $\B(\ell)$ is the unique $\Fq$-line containing the three points $P_1,P_2,P_3$. Finally, as $\ell$ is contained in the element $X_{m,u_1}$ of $\D^m$, it follows from Result~\ref{M110} that $\B(\ell)$ belongs to family $\F_m$ as required.
The other case where $T=(1,0,0)$, $N=(0,1,0)$ is similar. In this case the conic $\C$ has equation $\f y^2+\g z^2+xz=0$, for some $\f,\g\in\Fqt$, $\f\neq0$, so for example, $A=( y^2 \f +\g,y,1)$ If $\D$ is an Andr\'e set with transversal points $(1,0,0),(0,1,0)$, then we can arrange coordinates so
$\D=\{(\new\tau^{(q-1)i},1,0)\,|\,i\in I \}$ for some $\new\in\Fqts$. The replacement sets can be written as
$\D^m=\{ X_{m,u}=\{(\langle s^{q^m}u\rangle_q,\langle s\rangle_q,0)\,|\, s\in\Fqt^*\}\ \,|\,u\in\Fqt^*, \Nm(u)=\Nm(\new)\}.$
The computations are now similar to the above case.
\end{proof}
%%%
The next two results are used to show that we can replace the condition that $q^t-1$ is a factor of both $(q-1)(q^m-2)(j-i)$ and $(q-1)(q^m-2)(k-i)$ with the condition that $t\bigm| hm-1$.
\begin{lemma}\label{008}
Suppose $q$ even, $t\geq3$ prime and $m\in\{1,\ldots,t-1\}$, then
$\gcd( q^m-2,\hyp)$ is either $1$ or $2^t-1$. Further, $\gcd( q^m-2,\hyp)=2^t-1$ if and only if
$ t\bigm| hm-1.$
\end{lemma}
\begin{proof}
For $m,n\in\mathbb Z$, it is straightforward to check that
\begin{equation}\label{e007} \gcd(2^m-1,2^n-1)=2^{\gcd( m,n)}-1. \end{equation}
Using this we have
\begin{equation}\label{e007b}
\gcd( 2^{hm-1}-1,2^{h}-1)=2^{\gcd( hm-1,h)}-1=1.
\end{equation}
Hence
\[
\begin{split}
\gcd( q^m-2,\hyp)
=&\gcd( 2(2^{hm-1}-1),\frac{2^{ht}-1}{2^h-1})\\
=&\gcd( 2^{hm-1}-1,\frac{2^{ht}-1}{2^h-1})
\qquad \mbox{ as $\gcd\left( 2,\frac{2^{ht}-1}{2^h-1}\right)=1$}\\
=&\gcd( 2^{hm-1}-1,2^{ht}-1) % \qquad \mbox{ as $\gcd( 2,2^{ht}-1)=1$}\\
\qquad \mbox{ by (\ref{e007b})}\\
=&2^{\gcd( hm-1,ht)}-1\qquad\qquad \mbox{by (\ref{e007})}\\
=&2^{\gcd( hm-1,t)}-1\qquad \qquad\mbox{ as $\gcd( hm-1,h)=1$.}\\
\end{split}
\]
As $t$ is a prime, either $\gcd( hm-1,t)=1$ or $\gcd( hm-1,t)=t$. So $\gcd( q^m-2,\hyp)$ is 1 or $2^t-1$, the latter case occurs if and only if $t\bigm|hm-1$.
\end{proof}
\begin{lemma}\label{011}
Suppose $q$ even, $t\geq3$ prime and $m\in\{1,\ldots,t-1\}$.
\begin{enumerate}
\item If $q^t-1$ is a factor of $(q-1)(q^m-2)(j-i)$ for some distinct $i,j\in I $, then $ t\bigm| hm-1$.
\item If $ t\bigm| hm-1$, then there exists
distinct $i,j,k\in I $ satisfying $\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0$ and $q^t-1$ is a factor of both $(q-1)(q^m-2)(j-i)$ and $(q-1)(q^m-2)(k-i)$.
\end{enumerate}
\end{lemma}
\begin{proof} To prove part 1, suppose $q^t-1$ is a factor of $(q-1)(q^m-2)(j-i)$ for some $i,j\in I $. To show that $ t\bigm| hm-1$, it follows from by Lemma \ref{008} that we only need show that $\gcd( q^m-2,\hyp) \neq 1$. Suppose $\gcd( q^m-2,\hyp)= 1$. As $q^t-1$ is a factor of $(q-1)(q^m-2)(j-i)$, it follows that $\hyp$ is a factor of $j-i$, which contradicts $0\le i,j<\hyp$.
To prove part 2, suppose that $ t\bigm| hm-1$. As
$\tau$ is a multiplicative generator for $\Fqt^*$, the subfield $\mathbb F_{2^t}^*$ has multiplicative generator $\omega=\tau^{\frac{q^t-1}{2^t-1}}$.
Now $1+\omega\in \mathbb F_{2^t}^*$, so there exists unique
$a\in \{2,3,\ldots,2^t-2\}$ with $\omega^a=1+\omega$.
As $ t\bigm| hm-1$, it follows from Lemma~\ref{008} that
$2^t-1\bigm| \hyp$, so setting
\begin{equation}\label{e09}
i=0,\quad j=\frac 1{2^t-1}\left(\hyp\right),\quad k=\frac a{2^t-1}\left(\hyp\right),
\end{equation}
gives distinct $i,j,k\in I $.
Using (\ref{e09}) we have
\begin{eqnarray*}
\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}
&=& 1+\tau^{(q-1)\frac 1{2^t-1}\hyp}+\tau^{(q-1)\frac a{2^t-1}\hyp}\\
&=&\omega^0+\omega^1+\omega^a.
\end{eqnarray*}
This equals $0$ by the definition of $a$.
Further, as $t\bigm| hm-1$, by Lemma~\ref{008} we have
$2^t-1\bigm| q^m-2$. By (\ref{e09}), $i,j,k$ are all multiples of $\frac 1{2^t-1}\left(\hyp\right)$, so it follows that both $(q-1)(q^m-2)(j-i)$ and $(q-1)(q^m-2)(k-i)$
are divisible by
\[
(q-1)\times (2^t-1)\times \frac 1{2^t-1}\left(\hyp\right)=q^t-1,
\]
as required.
\end{proof}
\subsubsection{Tangent conics - inherited arcs}
We now use these technical lemmas to prove the existence of inherited arcs
in the Andr\'e plane.
\begin{lemma}\label{012}
In $\PG(2,q^t)$, $q$ even, $q\geq t$, $t\geq3$ prime, let $\C$ be a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$ with $\{T,N\}=\{(1,0,0),(0,1,0)\}$ and let
$\D$ be an Andr\'e set with transversal points $T,N$ (so coordinates for $\D$ and $\D^m$ are given in equations \textup{(\ref{e5})} and \textup{(\ref{e6})}).
For each $m\in\{1,\ldots,t-1\}$, the set $\D_m(\C)$ is an arc in the Andr\'e plane $\P(\D^m)$ if and only if
$t$ does not divide $h m -1$.
\end{lemma}
\begin{proof} We prove the contrapositive, namely we show that $\D_m(\C)$ is not an arc of $\P(\D^m)$ if and only if $t\bigm| hm-1$.
Suppose $\D_m(\C)$ is not an arc of $\P(\D^m)$. Then there are three affine points $A,B,C\in\C$ with $A,B,C$ collinear in $\P(\D^m)$, so $A,B,C$ lie in a common $\D^m$-block. It follows from Result~\ref{001} that the $\Fq$-line determined by the three points $P_1=AB\cap\li$, $P_2=AC\cap\li$, $P_3=BC\cap\li$ belongs to family $\F_m$. In particular this means that $P_1,P_2,P_3\in\D$, and so $P_1=(1,\tau^{(q-1)i},0)$,
$P_2=(1,\tau^{(q-1)j},0)$,
$P_3=(1,\tau^{(q-1)k},0)$ for distinct $i,j,k\in I $.
By Lemma~\ref{002}, $i,j,k$ satisfy
$\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0$. It follows from Lemma~\ref{003} that $q^t-1$ is a factor of both $(q-1)(q^m-2)(j-i)$ and $(q-1)(q^m-2)(k-i)$. Hence by Lemma~\ref{011}, we have $t\bigm| hm-1$.
Conversely suppose $t\bigm| hm-1$. By Lemma~\ref{011}, there exists distinct $i,j,k\in I $ satisfying
$\tau^{(q-1)i}+\tau^{(q-1)j}+\tau^{(q-1)k}=0$ and $q^t-1$ is a factor of both $(q-1)(q^m-2)(j-i)$ and $(q-1)(q^m-2)(k-i)$.
By Lemma~\ref{003}, the
$\Fq$-line determined by the three points $P_1=(1,\tau^{(q-1)i},0)$,
$P_2=(1,\tau^{(q-1)j},0)$,
$P_3=(1,\tau^{(q-1)k},0)$ belongs to family $\F_m$.
By Lemma~\ref{002}, there exists three points $A,B,C\in\C$ with $AB\cap\li=P_1$, $AC\cap\li=P_2$, $BC\cap\li=P_3$. By Result~\ref{001}, the three points $A,B,C$ lie in a common $\D^m$-block. Hence $A,B,C$ are collinear in $\P(\D^m)$ and so
$\D_m(\C)$ is not an arc.
\end{proof}
\begin{theorem}\label{013}
In $\PG(2,q^t)$, $q$ even, $q\geq t$, $t\geq3$ prime, let $\C$ be a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$ with $\{T,N\}=\{(1,0,0),(0,1,0)\}$ and let
$\D=\{(1,\new\tau^{(q-1)i})\,|\,i\in I \}$ (for some $\new\in\Fqt^*$) be an Andr\'e set with transversal points $T,N$. Perform an Andr\'e replacement using the Andr\'e replacement net $\D^m=\{ X_{m,u}=\{\langle(s,s^{q^m}u)\rangle_q\,|\, s\in\Fqt^*\}\ \,|\,u\in\Fqt^*, \Nm(u)=\Nm(\new)\}$; and let $\D_m(\C)$ denote the set of points in $\P(\D^m)$ corresponding to the affine points of $\C$.
\begin{enumerate}
\item If $h\equiv 0\pmod{t}$, then for each $m\in\{1,\ldots,t-1\}$, $\D_m(\C)$ is an arc which can be completed to an oval in the Andr\'e plane $\P(\D^m)$.
\item If $h\not\equiv 0\pmod{t}$, then there is a unique $n\in\{1,\ldots,t-1\}$ for which $t$ divides $h n -1 $. For this $n$ the following holds:
\begin{enumerate}
\item
$\D_n(\C)$ is not an arc in the Andr\'e plane $\P(\D^n)$;
\item for each $m\in\{1,\ldots,t-1\}\setminus\{n\}$,
$\D_m(\C)$ is an arc which can be completed to an oval in the Andr\'e plane $\P(\D^m)$.
\end{enumerate}
\end{enumerate}
\end{theorem}
\begin{proof}
Suppose $h\equiv 0\pmod{t}$, that is,
$t\bigm| h$. Then $t$ does not divide $hm-1$ for any $m\in\{1,\ldots,t-1\}$. So by Lemma~\ref{012}, $\D_m(\C)$ is an arc in the Andr\'e plane $\P(\D^m)$ for each $m\in\{1,\ldots,t-1\}$. As the point $T$ does not lie in $\D$, all the lines of $\P(\D^m)$ through $T$ are also lines of $\PG(2,q^t)$, so $T$ does not lie on a $3$-secant of $\D_m(\C)$. Hence the set of points $\D_m(\C)\cup T$ form an oval of $\P(\D^m)$.
Suppose $h\not\equiv 0\pmod{t}$,
then $\gcd( h,t)=1$ as $t$ is a prime.
Hence for $m\in\{1,\ldots,t-1\}$, we have $\gcd(h m,t)=1$, and so $h m\not\equiv0\pmod{t}$.
Next suppose there exists distinct $m_1,m_2\in\{1,\ldots,t-1\}$ with $h m_1\equiv h m_2 \pmod t$. Then
$ t \bigm|h m_1-h m_2 $, (since
$\gcd( t,h)=1$).
As $0<m_1,m_2<t$, it follows that $m_1=m_2$. That is, we have shown that the set
$\{h m\pmod t\,|\,m=1,\ldots,t-1\}$ equals the set $\{1,\ldots,t-1\}$. Hence there is a unique $n\in\{1,\ldots,t-1\}$ with $h n\equiv 1\pmod{t}$, that is, there is a unique $n\in\{1,\ldots,t-1\}$ for which $t$ divides $h n-1$.
It follows from Lemma~\ref{012} that $\D_m(\C)$ is an arc in the Andr\'e plane $\P(\D^m)$ for each $m\in\{1,\ldots,t-1\}\setminus\{n\}$, and $\D_n(\C)$ is not an arc in $\P(\D^n)$. Moreover we can complete the arcs to ovals by adding the point $T$.
\end{proof}
As all Andr\'e sets are projectively equivalent, we have the following general result.
\begin{theorem}\label{014}
In $\PG(2,q^t)$, $q$ even, $q\geq t$, $t\geq3$ prime, let $\C$ be a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$ and let
$\D$ be an Andr\'e set with transversal points $T,N$. Perform an Andr\'e replacement using the Andr\'e replacement net $\D^m$; and let $\D_m(\C)$ denote the set of points in $\P(\D^m)$ corresponding to the affine points of $\C$.
\begin{enumerate}
\item If $h\equiv 0\pmod{t}$, then for each $m\in\{1,\ldots,t-1\}$, $\D_m(\C)$ is an arc which can be completed to an oval in the Andr\'e plane $\P(\D^m)$.
\item If $h\not\equiv 0\pmod{t}$, then there is a unique $n\in\{1,\ldots,t-1\}$ for which $\D_n(\C)$ is not an arc in the Andr\'e plane $\P(\D^n)$; for the remaining $t-2$ values $m\in\{1,\ldots,t-1\}\setminus\{n\}$,
$\D_m(\C)$ is an arc which can be completed to an oval in the Andr\'e plane $\P(\D^m)$.
\end{enumerate}
\end{theorem}
\section{Conclusion}\label{sec-con}
Theorem~\ref{thm000} showed that a conic of $\PG(2,q^t)$, $q$ even, which is secant or exterior to $\li$ never inherits to an arc of an Andr\'e plane. The tangent conic case is more complex. If a conic $\C$ of $\PG(2,q^t)$, $q$ even, is tangent to $\li$ at point $T$ with nucleus $N$, there are a number of cases to consider. Theorem~\ref{201}
shows that when $T,N\in\D$, $\C$ does not inherit to an arc. Theorem~\ref{014} looked at a case where $\C$ does inherit to an arc (namely when $T,N$ are the transversal points of the Andr\'e set $\D$).
The remaining cases to consider include: the cases where exactly one of $T,N$ lies in $\D$, and the other either is, or is not, a transversal point of $\D$; and the case where $T,N\notin\D$, with either zero or one of $T,N$ being a transversal point of $\D$.
Computer searches when $t=3$ for small $q$ using Magma~\cite{magma} tested all of these cases and did not find any examples where the conic inherited to an arc.
In particular, computer results suggest that the inherited arc described in Result~\ref{120e}(3b) does not generalise to the case $t= 3$.
It seems likely that there are no other cases which lead to inherited arcs and the following conjecture encompasses this in the case $t=3$, that is, in $\PG(2,q^3)$.
\begin{conjecture}
In $\PG(2,q^3)$, $q$ even, $q>2$, let $\C$ be a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$ and let
$\D$ be an Andr\'e set of $\li$ with transversal points $E,F$.
If $|\{T,N\}\cap\{E,F\}|<2$, then
$\D_m(\C)$ is not an arc in the Andr\'e plane $\P(\D^m)$ for any $m\in\{1,2\}$.
\end{conjecture}
%\begin{conjecture}
%In $\PG(2,q^t)$, $q$ even, $q\geq t$, $t\geq3$ prime, let $\C$ be a conic with nucleus $N$ and tangent to the line $\li$ at the point $T$ and let
%$\D$ be an Andr\'e set of $\li$ with transversal points $E,F$.
%If $|\{T,N\}\cap\{E,F\}|<2$, then
%$\D_m(\C)$ is not an arc in the Andr\'e plane $\P(\D^m)$ for any $m\in\{1,\ldots,t-1\}$.
%\end{conjecture}
Another interesting question arises from a conjecture in \cite[p556]{GlynnStein} that considers the arc constructed in Result~\ref{120e}(3c); they conjecture that all ovals through two conjugate points in the Hall plane of order $q=2^{2k}$ derive from conics in the Desarguesian plane. An interesting generalisation
is to an Andr\'e plane $\P(\D^m)$ of order $q^t$, $q$ even, $t\geq3$ prime that is obtained from the Desarguesian plane $\PG(2,q^t)$ using one Andr\'e plane net replacement. We can ask whether an oval in $\P(\D^m)$ that contains the two transversal points on $\li$ is inherited from a conic of $\PG(2,q^t)$.
\begin{thebibliography}{99}
\bibitem{andr54}
J. Andr\'e.
\"Uber nicht-Desarguessche Ebenen mit transitiver Translationgruppe.
{\em Math. Z.},
60 (1954)
156--186.
%
%
\bibitem{qoddpaper}
S.G. Barwick, A.M.W. Hui and W.-A. Jackson.
Enumerating arcs inherited from conics in Andr\'e planes of odd order. arXiv:2606.11763
%\red{ BHJ Our q odd paper - check title. add arxiv citation}
%\bibitem{BJ2012}
%S.G. Barwick and W.-A. Jackson.
%Sublines and subplanes of PG$(2,q^3)$ in the Bruck-Bose representation in PG$(6,q)$.
%{\em Finite Fields Appl.},
%18 (2012)
%93--107.
%\red{unused}
\bibitem{barmar}
S.G. Barwick and D.J. Marshall.
Conics and multiple derivation.
{\em Discrete Math.},
312 (2012)
1623--1632.
%
%
\bibitem{BKNS}
A. Blokhuis, I. Kov\'{a}cs, G.P. Nagy and T. Sz\"{o}nyi.
Inherited conics in Hall planes.
\emph{Discrete Math.},
342 (2019)
1098--1107.
%
\bibitem{magma}
W. Bosma, J. Cannon and C. Playoust.
The Magma algebra system. I. The user language,
{\em J. Symbolic Comput.},
24 (1997)
235--265.
%
\bibitem{bruc69}
R.H. Bruck.
Construction problems of finiteprojective planes.
{\em Conference on Combinatorial Mathematics and its Applications},
University of North Carolina Press, (1969)
426--514.
%
%\bibitem{bruck70} R.H. Bruck. Some relatively unknown ruled surfaces in projective space. Arch. Inst. Grand-Ducal Luxembourg Sect. \emph{Sci. Nat. Phys. Math. N.S.} (1970) 361--376.
%%
%\bibitem{bruc73a} R.H. Bruck. Circle geometry in higher dimensions. {\em A Survey of Combinatorial Theory}, eds. J. N. Srivastava et al, Amsterdam (1973) 69--77.
%%
%\bibitem{bruc73b} R.H. Bruck. Circle geometry in higher dimensions. II. {\em Geom. Dedicata}, 2 (1973) 133--188.
%%
%%
\bibitem{bruc64}
R.H. Bruck and R.C. Bose.
The construction of translation planes from projective spaces.
{\em J.\ Algebra},
1 (1964)
85--102.
%
%\bibitem{bruc66} R.H.\ Bruck and R.C.\ Bose. Linear
%representations of projective planes in projective spaces. {\em
%J.\ Algebra}, {\bf 4} (1966) 117--172.
%
%%
%\bibitem{csaj} B. Csajb\'ok, G. Marino and F. Zullo. New maximum scattered linear sets of the projective line. \emph{Finite Fields App.,} 54 (2018) 133--150.
%%
%\bibitem{csaj16} B. Csajb\'ok and C. Zanella. On scattered linear sets of pseudoregulus type in $\PG(1,q^t)$. \emph{Finite Fields App.,} 41 (2016) 34--54.
%%
%\bibitem{casa22} V. Casarino, G. Longobardi and C. Zanella.
%Scattered linear sets in a finite projective line and translation planes.
%\emph{Linear Algebra Appl.,} 650 (2022) 286--298.
%
%%
\bibitem{Cher}
W. Cherowitzo.
The classification of inherited hyperconics in Hall planes of even order.
\emph{European J. Combin.},
31 (2010)
81--86.
%
%
%\bibitem{dona14}
%G. Donati and N. Durante.
%Scattered linear sets generated by collineations between pencils of lines.
%{\em J. Algebraic. Comb.},
%40 (2014)
%1121--1134.
%\red{unused}
%
%\bibitem{ferr03} S. Ferret and L. Storme. Results on maximal partial spreads in $\PG(3,p^3)$ and on related minihypers. {\em Des. Codes Cryptogr.},
% 29 (2003) 105--122.
%%
%%
%%
%\bibitem{grim} G.G. Grimaldi, S. Gupta, G. Longobardi, R. Trombetti.
%A geometric characterization of known maximum scattered linear sets of $\PG(1, q^n)$.
%http://arxiv.org/abs/2405.01374v1
%
\bibitem{GlynnStein}
D.G. Glynn and G.F. Steinke.
On conics that are ovals in a Hall plane.
\emph{Europ. J. Combin.},
14 (1993), 521--528.
%
%\bibitem{Hir}
%J.W.P. Hirschfeld.
%\emph{Projective Geometries over Finite Fields}, second edition,
%Oxford University Press, Oxford (1998).
%
%
%
%%
% \bibitem{JJ2008} V.~Jha and N.L.~Johnson. A new class of translation planes constructed by hyper-regulus replacement. \emph{J. Geom.}, 90 (2008) 83--99.
%%
%%
%\bibitem{johnson2003} N.L. Johnson. Hyper-reguli and non-Andr\'e quasi subgeometry partitions of projective spaces. \emph{J. Geom.,} 78 (2003) 59--82.
%%
%%
%\bibitem{john07} N.L. Johnson, V. Jha and M. Biliotti. {\em Handbook of Finite
% Translation Planes.} Chapman and Hall/CRC 2007.
%%%
%%
%%
%
%
\bibitem{Korch1}
G. Korchm\'{a}ros.
Ovali nei piani di hall di ordine dispari.
\emph{Atti Accad. Naz. Lincei, Rend.},
56 (1974)
315--317.
\bibitem{Korch2}
G. Korchm\'aros.
Inherited arcs in finite affine planes.
\emph{J. Combin. Theory Ser. A},
42 (1986)
140--143.
%
%
%%
%%
%\bibitem{Lav}
%M. Lavrauw.
%Sublines of prime order contained in the set of internal points of a conic.
%\emph{Des. Codes Cryptogr.}, 38 (2006) 113--123.
%%
%
%%
%%
%\bibitem{lavr10}
%M. Lavrauw and G. Van de Voorde.
%On linear sets on a projective line.
%{\em Des. Codes Cryptogr.},
%56 (2010)
%89--104.
%\red{unused}
%
%\bibitem{lavr14a} M. Lavrauw and G. Van de Voorde. Field Reduction and linear sets in finite geometry. \emph{Topics in Finite Fields,} AMS Contemporary Math., ed by G. Kyureghyan, G. L. Mullen, and A. Pott (2015)
%%%
%%
%%
%\bibitem{LavR} M. Lavrauw and M. Rodgers. Classification of 8-dim rank 2 commutative semifields. \emph{Adv. Geom.,} 19 (2019) 57--64.
%%
%%
%%
%%
%\bibitem{lavr15} M. Lavrauw, J, Sheekey
%and C. Zanella. On embeddings of minimum distance of $\PG(n,q)\times\PG(n,q)$. \emph{Des. Codes Cryptogr.,} 74 (2015) 427--440.
%%
%\bibitem{lavr14} M. Lavrauw
%and C. Zanella. Subgeometries and linear sets on a projective line. {\em Finite Fields Appl.}, to appear.
%
%
%
%\bibitem{LMPT14}
%G. Lunardon, G. Marino, O. Polverino and R. Trombetti.
%Maximum scattered linear sets of pseudoregulus type
%and the Segre variety ${\cal S}_{n,n}$.
%{\em J. Algebraic. Comb.},
%39 (2014)
%807--831.
%\red{unused}
%
%
%
%\bibitem{luna01} G. Lunardon and O. Polverino. Blocking Sets and Derivable Partial Spreads. \emph{J. Algebr. Comb}., 14 (2001), 49–56.
%%
%
%
\bibitem{OKeefePas}
C.M. O'Keefe and A.A. Pascasio.
Images of conics under derivation.
\emph{Discr. Math.},
151 (1996)
189--199.
%
\bibitem{OKeefePasPen}
C.M. O'Keefe, A.A. Pascasio, and T. Pentilla.
Hyperovals in Hall planes. \emph{Eur. J. Combin.},
13 (1992)
195--199.
%
%%
%\bibitem{ostrom70} T.G. Ostrom. Finite translation planes. Lecture Notes in Mathematics. 158, Springer-Verlag, New York, 1970.
%%
%\bibitem{ostrom}
%T.G. Ostrom.
%Hyper-reguli.
%\emph{J. Geom.},
%48 (1993)
%157--166.
%\red{unused}
%
%
\bibitem{segre}
B. Segre.
Teoria di Galois, fibrazioni proiettive e geometrie non desarguesiane.
\emph{Ann. Mat. Pura Appl.},
64 (1964)
1--64.
\bibitem{segrekorch}
B. Segre and G. Korchm\'{a}ros.
Una proprieta degli insiemi di punti di un piano di Galois caratterizzante quelli formati dai punti delle singole rette esterne ad una conica.
\emph{Atti Accad. Naz. Lincei. Rend.},
62 (1977)
613--619.
%\bibitem{sheek2018} J. Sheekey. A new family of linear maximum rank distance codes. \emph{Adv. Math. Commun.,} 10 (2016) 475--488.
%%
%
%
\bibitem{Szon}
T. Sz\"{o}nyi.
Arcs and $k$-sets with large nucleus set in Hall planes.
\emph{J. Geom.},
34 (1989)
187--194.
\end{thebibliography}
\end{document}